%PDF- %PDF-
Mini Shell

Mini Shell

Direktori : /usr/lib/python3/dist-packages/orca/scripts/web/
Upload File :
Create Path :
Current File : //usr/lib/python3/dist-packages/orca/scripts/web/script_utilities.py

# Orca
#
# Copyright 2010 Joanmarie Diggs.
# Copyright 2014-2015 Igalia, S.L.
#
# This library is free software; you can redistribute it and/or
# modify it under the terms of the GNU Lesser General Public
# License as published by the Free Software Foundation; either
# version 2.1 of the License, or (at your option) any later version.
#
# This library is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
# Lesser General Public License for more details.
#
# You should have received a copy of the GNU Lesser General Public
# License along with this library; if not, write to the
# Free Software Foundation, Inc., Franklin Street, Fifth Floor,
# Boston MA  02110-1301 USA.

__id__        = "$Id$"
__version__   = "$Revision$"
__date__      = "$Date$"
__copyright__ = "Copyright (c) 2010 Joanmarie Diggs." \
                "Copyright (c) 2014-2015 Igalia, S.L."
__license__   = "LGPL"

import gi
gi.require_version("Atspi", "2.0")
from gi.repository import Atspi

import functools
import re
import time
import urllib

from orca import debug
from orca import focus_manager
from orca import input_event
from orca import messages
from orca import orca_state
from orca import script_utilities
from orca import script_manager
from orca import settings_manager
from orca.ax_component import AXComponent
from orca.ax_document import AXDocument
from orca.ax_hypertext import AXHypertext
from orca.ax_object import AXObject
from orca.ax_table import AXTable
from orca.ax_text import AXText
from orca.ax_utilities import AXUtilities


class Utilities(script_utilities.Utilities):

    def __init__(self, script):
        super().__init__(script)

        self._currentTextAttrs = {}
        self._caretContexts = {}
        self._priorContexts = {}
        self._canHaveCaretContextDecision = {}
        self._contextPathsRolesAndNames = {}
        self._paths = {}
        self._inDocumentContent = {}
        self._inTopLevelWebApp = {}
        self._isTextBlockElement = {}
        self._isContentEditableWithEmbeddedObjects = {}
        self._isCodeDescendant = {}
        self._isEntryDescendant = {}
        self._hasGridDescendant = {}
        self._isGridDescendant = {}
        self._isLabelDescendant = {}
        self._isModalDialogDescendant = {}
        self._isMenuDescendant = {}
        self._isNavigableToolTipDescendant = {}
        self._isToolBarDescendant = {}
        self._isWebAppDescendant = {}
        self._isLayoutOnly = {}
        self._isFocusableWithMathChild = {}
        self._mathNestingLevel = {}
        self._isOffScreenLabel = {}
        self._labelIsAncestorOfLabelled = {}
        self._elementLinesAreSingleChars= {}
        self._elementLinesAreSingleWords= {}
        self._hasLongDesc = {}
        self._hasVisibleCaption = {}
        self._hasDetails = {}
        self._isDetails = {}
        self._isNonInteractiveDescendantOfControl = {}
        self._isClickableElement = {}
        self._isAnchor = {}
        self._isEditableComboBox = {}
        self._isErrorMessage = {}
        self._isInlineIframeDescendant = {}
        self._isInlineListItem = {}
        self._isInlineListDescendant = {}
        self._isLandmark = {}
        self._isLink = {}
        self._isListDescendant = {}
        self._isNonNavigablePopup = {}
        self._isNonEntryTextWidget = {}
        self._isCustomImage = {}
        self._isUselessImage = {}
        self._isRedundantSVG = {}
        self._isUselessEmptyElement = {}
        self._hasNameAndActionAndNoUsefulChildren = {}
        self._isNonNavigableEmbeddedDocument = {}
        self._isParentOfNullChild = {}
        self._inferredLabels = {}
        self._labelsForObject = {}
        self._labelTargets = {}
        self._descriptionListTerms = {}
        self._valuesForTerm = {}
        self._displayedLabelText = {}
        self._preferDescriptionOverName = {}
        self._shouldFilter = {}
        self._shouldInferLabelFor = {}
        self._treatAsTextObject = {}
        self._treatAsDiv = {}
        self._currentObjectContents = None
        self._currentSentenceContents = None
        self._currentLineContents = None
        self._currentWordContents = None
        self._currentCharacterContents = None
        self._lastQueuedLiveRegionEvent = None
        self._findContainer = None
        self._validChildRoles = {Atspi.Role.LIST: [Atspi.Role.LIST_ITEM]}

    def _cleanupContexts(self):
        toRemove = []
        for key, [obj, offset] in self._caretContexts.items():
            if not AXObject.is_valid(obj):
                toRemove.append(key)

        for key in toRemove:
            self._caretContexts.pop(key, None)

    def dumpCache(self, documentFrame=None, preserveContext=False):
        if not AXObject.is_valid(documentFrame):
            documentFrame = self.documentFrame()

        documentFrameParent = AXObject.get_parent(documentFrame)
        context = self._caretContexts.get(hash(documentFrameParent))

        tokens = ["WEB: Clearing all cached info for", documentFrame,
                  "Preserving context:", preserveContext, "Context:", context[0], ",", context[1]]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        self._script.structuralNavigation.clearCache(documentFrame)
        self.clearCaretContext(documentFrame)
        self.clearCachedObjects()

        if preserveContext and context:
            tokens = ["WEB: Preserving context of", context[0], ",", context[1]]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            self._caretContexts[hash(documentFrameParent)] = context

    def clearCachedObjects(self):
        debug.printMessage(debug.LEVEL_INFO, "WEB: cleaning up cached objects", True)
        self._inDocumentContent = {}
        self._inTopLevelWebApp = {}
        self._isTextBlockElement = {}
        self._isContentEditableWithEmbeddedObjects = {}
        self._isCodeDescendant = {}
        self._isEntryDescendant = {}
        self._hasGridDescendant = {}
        self._isGridDescendant = {}
        self._isLabelDescendant = {}
        self._isMenuDescendant = {}
        self._isModalDialogDescendant = {}
        self._isNavigableToolTipDescendant = {}
        self._isToolBarDescendant = {}
        self._isWebAppDescendant = {}
        self._isLayoutOnly = {}
        self._isFocusableWithMathChild = {}
        self._mathNestingLevel = {}
        self._isOffScreenLabel = {}
        self._labelIsAncestorOfLabelled = {}
        self._elementLinesAreSingleChars= {}
        self._elementLinesAreSingleWords= {}
        self._hasLongDesc = {}
        self._hasVisibleCaption = {}
        self._hasDetails = {}
        self._isDetails = {}
        self._isNonInteractiveDescendantOfControl = {}
        self._isClickableElement = {}
        self._isAnchor = {}
        self._isEditableComboBox = {}
        self._isErrorMessage = {}
        self._isInlineIframeDescendant = {}
        self._isInlineListItem = {}
        self._isInlineListDescendant = {}
        self._isLandmark = {}
        self._isLink = {}
        self._isListDescendant = {}
        self._isNonNavigablePopup = {}
        self._isNonEntryTextWidget = {}
        self._isCustomImage = {}
        self._isUselessImage = {}
        self._isRedundantSVG = {}
        self._isUselessEmptyElement = {}
        self._hasNameAndActionAndNoUsefulChildren = {}
        self._isNonNavigableEmbeddedDocument = {}
        self._isParentOfNullChild = {}
        self._inferredLabels = {}
        self._labelsForObject = {}
        self._labelTargets = {}
        self._descriptionListTerms = {}
        self._valuesForTerm = {}
        self._displayedLabelText = {}
        self._preferDescriptionOverName = {}
        self._shouldFilter = {}
        self._shouldInferLabelFor = {}
        self._treatAsTextObject = {}
        self._treatAsDiv = {}
        self._paths = {}
        self._contextPathsRolesAndNames = {}
        self._canHaveCaretContextDecision = {}
        self._cleanupContexts()
        self._priorContexts = {}
        self._lastQueuedLiveRegionEvent = None
        self._findContainer = None

    def clearContentCache(self):
        self._currentObjectContents = None
        self._currentSentenceContents = None
        self._currentLineContents = None
        self._currentWordContents = None
        self._currentCharacterContents = None
        self._currentTextAttrs = {}

    def isDocument(self, obj, excludeDocumentFrame=True):
        if AXUtilities.is_document_web(obj) or AXUtilities.is_embedded(obj):
            return True

        if not excludeDocumentFrame:
            return AXUtilities.is_document_frame(obj)

        return False

    def inDocumentContent(self, obj=None):
        if not obj:
            obj = focus_manager.getManager().get_locus_of_focus()


        if self.isDocument(obj):
            return True

        rv = self._inDocumentContent.get(hash(obj))
        if rv is not None:
            return rv

        document = self.getDocumentForObject(obj)
        rv = document is not None
        self._inDocumentContent[hash(obj)] = rv
        return rv

    def _getDocumentsEmbeddedBy(self, frame):
        return AXObject.get_relation_targets(frame, Atspi.RelationType.EMBEDS, self.isDocument)

    def sanityCheckActiveWindow(self):
        app = self._script.app
        window = focus_manager.getManager().get_active_window()
        if AXObject.get_parent(window) == app:
            return True

        tokens = ["WARNING:", window, "is not child of", app]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        # TODO - JD: Is this exception handling still needed?
        try:
            script = script_manager.getManager().getScript(app, window)
            tokens = ["WEB: Script for active Window is", script]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
        except Exception:
            msg = "ERROR: Exception getting script for active window"
            debug.printMessage(debug.LEVEL_INFO, msg, True)
        else:
            if isinstance(script, type(self._script)):
                attrs = script.getTransferableAttributes()
                for attr, value in attrs.items():
                    tokens = ["WEB: Setting", attr, "to", value]
                    debug.printTokens(debug.LEVEL_INFO, tokens, True)
                    setattr(self._script, attr, value)

        window = focus_manager.getManager().find_active_window(app)
        self._script.app = AXObject.get_application(window)
        tokens = ["WEB: updating script's app to", self._script.app]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        focus_manager.getManager().set_active_window(window)
        return True

    def activeDocument(self, window=None):
        window = window or focus_manager.getManager().get_active_window()
        documents = self._getDocumentsEmbeddedBy(window)
        documents = list(filter(AXUtilities.is_showing, documents))
        if len(documents) == 1:
            return documents[0]
        return None

    def documentFrame(self, obj=None):
        if not obj and self.sanityCheckActiveWindow():
            document = self.activeDocument()
            if document:
                return document

        return self.getDocumentForObject(obj or focus_manager.getManager().get_locus_of_focus())

    def grabFocusWhenSettingCaret(self, obj):
        # To avoid triggering popup lists.
        if AXUtilities.is_entry(obj):
            return False

        if AXUtilities.is_image(obj):
            return AXObject.find_ancestor(obj, AXUtilities.is_link) is not None

        if AXUtilities.is_heading(obj) and AXObject.get_child_count(obj) == 1:
            return self.isLink(AXObject.get_child(obj, 0))

        return AXUtilities.is_focusable(obj)

    def setCaretPosition(self, obj, offset, documentFrame=None):
        if self._script.flatReviewPresenter.is_active():
            self._script.flatReviewPresenter.quit()
        grabFocus = self.grabFocusWhenSettingCaret(obj)

        obj, offset = self.findFirstCaretContext(obj, offset)
        self.setCaretContext(obj, offset, documentFrame)
        if self._script.focusModeIsSticky():
            return

        oldFocus = focus_manager.getManager().get_locus_of_focus()
        AXText.clear_all_selected_text(oldFocus)
        focus_manager.getManager().set_locus_of_focus(None, obj, notify_script=False)
        if grabFocus:
            AXObject.grab_focus(obj)

        AXText.set_caret_offset(obj, offset)
        if self._script.useFocusMode(obj, oldFocus) != self._script.inFocusMode():
            self._script.togglePresentationMode(None)

        # TODO - JD: Can we remove this?
        if obj:
            AXObject.clear_cache(obj, False, "Set caret in object.")

        # TODO - JD: This is private.
        self._script._saveFocusedObjectInfo(obj)

    def getNextObjectInDocument(self, obj, documentFrame):
        if not obj:
            return None

        relation = AXObject.get_relation(obj, Atspi.RelationType.FLOWS_TO)
        if relation:
            return relation.get_target(0)

        if obj == documentFrame:
            obj, offset = self.getCaretContext(documentFrame)
            for child in AXObject.iter_children(documentFrame):
                if AXHypertext.get_character_offset_in_parent(child) > offset:
                    return child

        if AXObject.get_child_count(obj):
            return AXObject.get_child(obj, 0)

        while obj and obj != documentFrame:
            nextObj = AXObject.get_next_sibling(obj)
            if nextObj:
                return nextObj
            obj = AXObject.get_parent(obj)

        return None

    def getLastObjectInDocument(self, documentFrame):
        return AXObject.find_deepest_descendant(documentFrame)

    def getRoleDescription(self, obj, isBraille=False):
        attrs = AXObject.get_attributes_dict(obj)
        rv = attrs.get('roledescription', '')
        if isBraille:
            rv = attrs.get('brailleroledescription', rv)

        return rv

    def nodeLevel(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().nodeLevel(obj)

        rv = -1
        if not (self.inMenu(obj) or AXUtilities.is_heading(obj)):
            attrs = AXObject.get_attributes_dict(obj)
            # ARIA levels are 1-based; non-web content is 0-based. Be consistent.
            rv = int(attrs.get('level', 0)) -1

        return rv

    def _shouldCalculatePositionAndSetSize(self, obj):
        return True

    def getPositionAndSetSize(self, obj, **args):
        posinset = self.getPositionInSet(obj)
        setsize = self.getSetSize(obj)
        if posinset is not None and setsize is not None:
            # ARIA posinset is 1-based
            return posinset - 1, setsize

        if self._shouldCalculatePositionAndSetSize(obj):
            return super().getPositionAndSetSize(obj, **args)

        return -1, -1

    def getPositionInSet(self, obj):
        attrs = AXObject.get_attributes_dict(obj, False)
        position = attrs.get('posinset')
        if position is not None:
            return int(position)

        if AXUtilities.is_table_row(obj):
            rowindex = attrs.get('rowindex')
            if rowindex is None and AXObject.get_child_count(obj):
                cell = AXObject.find_descendant(obj, AXUtilities.is_table_cell_or_header)
                rowindex = AXObject.get_attributes_dict(cell, False).get('rowindex')

            if rowindex is not None:
                return int(rowindex)

        return None

    def getSetSize(self, obj):
        attrs = AXObject.get_attributes_dict(obj, False)
        setsize = attrs.get('setsize')
        if setsize is not None:
            return int(setsize)

        if AXUtilities.is_table_row(obj):
            rows = AXTable.get_row_count(AXTable.get_table(obj))
            if rows != -1:
                return rows

        return None

    def _getID(self, obj):
        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('id')

    def _getDisplayStyle(self, obj):
        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('display', '')

    def _getTag(self, obj):
        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('tag')

    def _getXMLRoles(self, obj):
        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('xml-roles', '').split()

    def inFindContainer(self, obj=None):
        if not obj:
            obj = focus_manager.getManager().get_locus_of_focus()

        if self.inDocumentContent(obj):
            return False

        return super().inFindContainer(obj)

    def isEmpty(self, obj):
        if not self.isTextBlockElement(obj):
            return False

        if AXObject.get_name(obj):
            return False

        return not self.treatAsTextObject(obj, False)

    def isHidden(self, obj):
        attrs = AXObject.get_attributes_dict(obj, False)
        return attrs.get('hidden', False)

    def _isOrIsIn(self, child, parent):
        if not (child and parent):
            return False

        if child == parent:
            return True

        return AXObject.find_ancestor(child, lambda x: x == parent)

    def isTextArea(self, obj):
        if not self.inDocumentContent(obj):
            return super().isTextArea(obj)

        if self.isLink(obj):
            return False

        if AXUtilities.is_combo_box(obj) \
           and AXUtilities.is_editable(obj) \
           and not AXObject.get_child_count(obj):
            return True

        if AXObject.get_role(obj) in self._textBlockElementRoles():
            document = self.getDocumentForObject(obj)
            if AXUtilities.is_editable(document):
                return True

        return super().isTextArea(obj)

    def isReadOnlyTextArea(self, obj):
        # NOTE: This method is deliberately more conservative than isTextArea.
        if not AXUtilities.is_entry(obj):
            return False

        if AXUtilities.is_read_only(obj):
            return True

        return AXUtilities.is_focusable(obj) and not AXUtilities.is_editable(obj)

    def setCaretOffset(self, obj, characterOffset):
        self.setCaretPosition(obj, characterOffset)
        self._script.updateBraille(obj)

    def nextContext(self, obj=None, offset=-1, skipSpace=False):
        if not obj:
            obj, offset = self.getCaretContext()

        nextobj, nextoffset = self.findNextCaretInOrder(obj, offset)
        if skipSpace:
            while AXText.get_character_at_offset(nextobj, nextoffset)[0].isspace():
                nextobj, nextoffset = self.findNextCaretInOrder(nextobj, nextoffset)

        return nextobj, nextoffset

    def previousContext(self, obj=None, offset=-1, skipSpace=False):
        if not obj:
            obj, offset = self.getCaretContext()

        prevobj, prevoffset = self.findPreviousCaretInOrder(obj, offset)
        if skipSpace:
            while AXText.get_character_at_offset(prevobj, prevoffset)[0].isspace():
                prevobj, prevoffset = self.findPreviousCaretInOrder(prevobj, prevoffset)

        return prevobj, prevoffset

    def lastContext(self, root):
        offset = 0
        if self.treatAsTextObject(root):
            offset = AXText.get_character_count(root) - 1

        def _isInRoot(o):
            return o == root or AXObject.find_ancestor(o, lambda x: x == root)

        obj = root
        while obj:
            lastobj, lastoffset = self.nextContext(obj, offset)
            if not (lastobj and _isInRoot(lastobj)):
                break
            obj, offset = lastobj, lastoffset

        return obj, offset

    def contextsAreOnSameLine(self, a, b):
        if a == b:
            return True

        aObj, aOffset = a
        bObj, bOffset = b
        aExtents = self.getExtents(aObj, aOffset, aOffset + 1)
        bExtents = self.getExtents(bObj, bOffset, bOffset + 1)
        return self.extentsAreOnSameLine(aExtents, bExtents)

    @staticmethod
    def extentsAreOnSameLine(a, b, pixelDelta=5):
        if a == b:
            return True

        aX, aY, aWidth, aHeight = a
        bX, bY, bWidth, bHeight = b

        if aWidth == 0 and aHeight == 0:
            return bY <= aY <= bY + bHeight
        if bWidth == 0 and bHeight == 0:
            return aY <= bY <= aY + aHeight

        highestBottom = min(aY + aHeight, bY + bHeight)
        lowestTop = max(aY, bY)
        if lowestTop >= highestBottom:
            return False

        aMiddle = aY + aHeight / 2
        bMiddle = bY + bHeight / 2
        if abs(aMiddle - bMiddle) > pixelDelta:
            return False

        return True

    def getExtents(self, obj, startOffset, endOffset):
        if not obj:
            return [0, 0, 0, 0]

        result = [0, 0, 0, 0]
        if self.treatAsTextObject(obj) and 0 <= startOffset < endOffset:
            rect = AXText.get_range_rect(obj, startOffset, endOffset)
            result = [rect.x, rect.y, rect.width, rect.height]
            if result[0] and result[1] and result[2] == 0 and result[3] == 0 \
               and AXText.get_substring(obj, startOffset, endOffset).strip():
                tokens = ["WEB: Suspected bogus range extents for",
                          obj, "(chars:", startOffset, ",", endOffset, "):", result]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
            else:
                return result

        parent = AXObject.get_parent(obj)
        if (AXUtilities.is_menu(obj) or AXUtilities.is_list_item(obj)) \
            and (AXUtilities.is_combo_box(parent) or AXUtilities.is_list_box(parent)):
            ext = AXComponent.get_rect(parent)
        else:
            ext = AXComponent.get_rect(obj)

        return [ext.x, ext.y, ext.width, ext.height]

    def _preserveTree(self, obj):
        if not (obj and AXObject.get_child_count(obj)):
            return False

        if self.isMathTopLevel(obj):
            return True

        return False

    def expandEOCs(self, obj, startOffset=0, endOffset=-1):
        if not self.inDocumentContent(obj):
            return super().expandEOCs(obj, startOffset, endOffset)

        if self.hasGridDescendant(obj):
            tokens = ["WEB: not expanding EOCs:", obj, "has grid descendant"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return ""

        if not self.treatAsTextObject(obj):
            return ""

        if self._preserveTree(obj):
            utterances = self._script.speechGenerator.generateSpeech(obj)
            return self._script.speechGenerator.utterancesToString(utterances)

        return super().expandEOCs(obj, startOffset, endOffset).strip()

    def substring(self, obj, startOffset, endOffset):
        if not self.inDocumentContent(obj):
            return super().substring(obj, startOffset, endOffset)

        if self.treatAsTextObject(obj):
            return AXText.get_substring(obj, startOffset, endOffset)

        return ""

    def textAttributes(self, acc, offset=None, get_defaults=False):
        attrsForObj = self._currentTextAttrs.get(hash(acc)) or {}
        if offset in attrsForObj:
            return attrsForObj.get(offset)

        attrs = super().textAttributes(acc, offset, get_defaults)
        objAttributes = AXObject.get_attributes_dict(acc, False)
        for key in self._script.attributeNamesDict.keys():
            value = objAttributes.get(key)
            if value is not None:
                attrs[0][key] = value

        self._currentTextAttrs[hash(acc)] = {offset:attrs}
        return attrs

    def localizeTextAttribute(self, key, value):
        if key == "justification" and value == "justify":
            value = "fill"

        return super().localizeTextAttribute(key, value)

    def adjustContentsForLanguage(self, contents):
        rv = []
        for content in contents:
            split = self.splitSubstringByLanguage(*content[0:3])
            for start, end, string, language, dialect in split:
                rv.append([content[0], start, end, string])

        return rv

    def getLanguageAndDialectFromTextAttributes(self, obj, startOffset=0, endOffset=-1):
        rv = super().getLanguageAndDialectFromTextAttributes(obj, startOffset, endOffset)
        if rv or obj is None:
            return rv

        # Embedded objects such as images and certain widgets won't implement the text interface
        # and thus won't expose text attributes. Therefore try to get the info from the parent.
        parent = AXObject.get_parent(obj)
        if parent is None or not self.inDocumentContent(parent):
            return rv

        start = AXHypertext.get_link_start_offset(obj)
        end = AXHypertext.get_link_end_offset(obj)
        language, dialect = self.getLanguageAndDialectForSubstring(parent, start, end)
        rv.append((0, 1, language, dialect))

        return rv

    def findObjectInContents(self, obj, offset, contents, usingCache=False):
        if not obj or not contents:
            return -1

        offset = max(0, offset)
        matches = [x for x in contents if x[0] == obj]
        match = [x for x in matches if x[1] <= offset < x[2]]
        if match and match[0] and match[0] in contents:
            return contents.index(match[0])
        if not usingCache:
            match = [x for x in matches if offset == x[2]]
            if match and match[0] and match[0] in contents:
                return contents.index(match[0])

        if not self.isTextBlockElement(obj):
            return -1

        child = AXHypertext.get_child_at_offset(obj, offset)
        if child and not self.isTextBlockElement(child):
            matches = [x for x in contents if x[0] == child]
            if len(matches) == 1:
                return contents.index(matches[0])

        return -1

    def findPreviousObject(self, obj):
        result = super().findPreviousObject(obj)
        if not (obj and self.inDocumentContent(obj)):
            return result

        if not (result and self.inDocumentContent(result)):
            return None

        if self.getTopLevelDocumentForObject(result) != self.getTopLevelDocumentForObject(obj):
            return None

        tokens = ["WEB: Previous object for", obj, "is", result, "."]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        return result

    def findNextObject(self, obj):
        result = super().findNextObject(obj)
        if not (obj and self.inDocumentContent(obj)):
            return result

        if not (result and self.inDocumentContent(result)):
            return None

        if self.getTopLevelDocumentForObject(result) != self.getTopLevelDocumentForObject(obj):
            return None

        tokens = ["WEB: Next object for", obj, "is", result, "."]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        return result

    def isNonEntryTextWidget(self, obj):
        rv = self._isNonEntryTextWidget.get(hash(obj))
        if rv is not None:
            return rv

        roles = [Atspi.Role.CHECK_BOX,
                 Atspi.Role.CHECK_MENU_ITEM,
                 Atspi.Role.MENU,
                 Atspi.Role.MENU_ITEM,
                 Atspi.Role.PAGE_TAB,
                 Atspi.Role.RADIO_MENU_ITEM,
                 Atspi.Role.RADIO_BUTTON,
                 Atspi.Role.PUSH_BUTTON,
                 Atspi.Role.TOGGLE_BUTTON]

        role = AXObject.get_role(obj)
        if role in roles:
            rv = True
        elif role == Atspi.Role.LIST_ITEM:
            rv = not AXUtilities.is_list(AXObject.get_parent(obj))
        elif role == Atspi.Role.TABLE_CELL:
            if AXUtilities.is_editable(obj):
                rv = False
            else:
                rv = not self.isTextBlockElement(obj)

        self._isNonEntryTextWidget[hash(obj)] = rv
        return rv

    def treatAsTextObject(self, obj, excludeNonEntryTextWidgets=True):
        if not obj or AXObject.is_dead(obj):
            return False

        rv = self._treatAsTextObject.get(hash(obj))
        if rv is not None:
            return rv

        if not AXObject.supports_text(obj):
            return False

        if not self.inDocumentContent(obj) or self._script.browseModeIsSticky():
            return True

        rv = AXText.get_character_count(obj) > 0 or AXUtilities.is_editable(obj)
        if rv and self._treatObjectAsWhole(obj, -1) and AXObject.get_name(obj) \
            and not self.isCellWithNameFromHeader(obj):
            tokens = ["WEB: Treating", obj, "as non-text: named object treated as whole."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False

        elif rv and not self.isLiveRegion(obj):
            doNotQuery = [Atspi.Role.LIST_BOX]
            role = AXObject.get_role(obj)
            if rv and role in doNotQuery:
                tokens = ["WEB: Treating", obj, "as non-text due to role."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                rv = False
            if rv and excludeNonEntryTextWidgets and self.isNonEntryTextWidget(obj):
                tokens = ["WEB: Treating", obj, "as non-text: is non-entry text widget."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                rv = False
            if rv and (self.isHidden(obj) or self.isOffScreenLabel(obj)):
                tokens = ["WEB: Treating", obj, "as non-text: is hidden or off-screen label."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                rv = False
            if rv and self.isNonNavigableEmbeddedDocument(obj):
                tokens = ["WEB: Treating", obj, "as non-text: is non-navigable embedded document."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                rv = False
            if rv and self.isFakePlaceholderForEntry(obj):
                tokens = ["WEB: Treating", obj, "as non-text: is fake placeholder for entry."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                rv = False

        self._treatAsTextObject[hash(obj)] = rv
        return rv

    def hasNameAndActionAndNoUsefulChildren(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._hasNameAndActionAndNoUsefulChildren.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        if self.hasExplicitName(obj) and AXObject.supports_action(obj):
            for child in AXObject.iter_children(obj):
                if not self.isUselessEmptyElement(child) or self.isUselessImage(child):
                    break
            else:
                rv = True

        if rv:
            tokens = ["WEB:", obj, "has name and action and no useful children"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)

        self._hasNameAndActionAndNoUsefulChildren[hash(obj)] = rv
        return rv

    def isNonInteractiveDescendantOfControl(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isNonInteractiveDescendantOfControl.get(hash(obj))
        if rv is not None:
            return rv

        role = AXObject.get_role(obj)
        rv = False
        roles = self._textBlockElementRoles()
        roles.extend([Atspi.Role.IMAGE, Atspi.Role.CANVAS])
        if role in roles and not AXUtilities.is_focusable(obj):
            controls = [Atspi.Role.CHECK_BOX,
                        Atspi.Role.CHECK_MENU_ITEM,
                        Atspi.Role.LIST_BOX,
                        Atspi.Role.MENU_ITEM,
                        Atspi.Role.RADIO_MENU_ITEM,
                        Atspi.Role.RADIO_BUTTON,
                        Atspi.Role.PUSH_BUTTON,
                        Atspi.Role.TOGGLE_BUTTON,
                        Atspi.Role.TREE_ITEM]
            rv = AXObject.find_ancestor(obj, lambda x: AXObject.get_role(x) in controls)

        self._isNonInteractiveDescendantOfControl[hash(obj)] = rv
        return rv

    def _treatObjectAsWhole(self, obj, offset=None):
        always = [Atspi.Role.CHECK_BOX,
                  Atspi.Role.CHECK_MENU_ITEM,
                  Atspi.Role.LIST_BOX,
                  Atspi.Role.MENU_ITEM,
                  Atspi.Role.PAGE_TAB,
                  Atspi.Role.RADIO_MENU_ITEM,
                  Atspi.Role.RADIO_BUTTON,
                  Atspi.Role.PUSH_BUTTON,
                  Atspi.Role.TOGGLE_BUTTON]

        descendable = [Atspi.Role.MENU,
                       Atspi.Role.MENU_BAR,
                       Atspi.Role.TOOL_BAR,
                       Atspi.Role.TREE_ITEM]

        role = AXObject.get_role(obj)
        if role in always:
            return True

        if role in descendable:
            if self._script.inFocusMode():
                return True

            # This should cause us to initially stop at the large containers before
            # allowing the user to drill down into them in browse mode.
            return offset == -1

        if role == Atspi.Role.ENTRY:
            if AXObject.get_child_count(obj) == 1 \
              and self.isFakePlaceholderForEntry(AXObject.get_child(obj, 0)):
                return True
            return False

        if AXUtilities.is_editable(obj):
            return False

        if role == Atspi.Role.TABLE_CELL:
            if self.isFocusModeWidget(obj):
                return not self._script.browseModeIsSticky()
            if self.hasNameAndActionAndNoUsefulChildren(obj):
                return True

        if role in [Atspi.Role.COLUMN_HEADER, Atspi.Role.ROW_HEADER] \
           and self.hasExplicitName(obj):
            return True

        if role == Atspi.Role.COMBO_BOX:
            return True

        if role in [Atspi.Role.EMBEDDED, Atspi.Role.TREE, Atspi.Role.TREE_TABLE]:
            return not self._script.browseModeIsSticky()

        if role == Atspi.Role.LINK:
            return self.hasExplicitName(obj) or self.hasUselessCanvasDescendant(obj)

        if self.isNonNavigableEmbeddedDocument(obj):
            return True

        if self.isFakePlaceholderForEntry(obj):
            return True

        if self.isCustomImage(obj):
            return True

        # Example: Some StackExchange instances have a focusable "note"/comment role
        # with a name (e.g. "Accepted"), and a single child div which is empty.
        if role in self._textBlockElementRoles() and AXUtilities.is_focusable(obj) \
           and self.hasExplicitName(obj):
            for child in AXObject.iter_children(obj):
                if not self.isUselessEmptyElement(child):
                    return False
            return True

        return False

    def __findSentence(self, obj, offset):
        # TODO - JD: Move this sad hack to AXText.
        text = AXText.get_all_text(obj)
        spans = [m.span() for m in re.finditer(r"\S*[^\.\?\!]+((?<!\w)[\.\?\!]+(?!\w)|\S*)", text)]
        rangeStart, rangeEnd = 0, len(text)
        for span in spans:
            if span[0] <= offset <= span[1]:
                rangeStart, rangeEnd = span[0], span[1] + 1
                break
        return text[rangeStart:rangeEnd], rangeStart, rangeEnd

    def _getTextAtOffset(self, obj, offset, boundary):
        def stringForDebug(x):
            return x.replace(self.EMBEDDED_OBJECT_CHARACTER, "[OBJ]").replace("\n", "\\n")

        if not obj:
            tokens = [f"WEB: Text at offset {offset} for", obj, "using", boundary, ":",
                      "'', Start: 0, End: 0. (obj is None)"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return '', 0, 0

        if not self.treatAsTextObject(obj):
            tokens = [f"WEB: Text at offset {offset} for", obj, "using", boundary, ":",
                      "'', Start: 0, End: 1. (treatAsTextObject() returned False)"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return '', 0, 1

        allText = AXText.get_all_text(obj)
        if boundary is None:
            string, start, end = allText, 0, len(allText)
            s = stringForDebug(string)
            tokens = [f"WEB: Text at offset {offset} for", obj, "using", boundary, ":",
                      f"'{s}', Start: {start}, End: {end}."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return string, start, end

        if boundary == Atspi.TextBoundaryType.SENTENCE_START and not AXUtilities.is_editable(obj):
            if AXObject.get_role(obj) in [Atspi.Role.LIST_ITEM, Atspi.Role.HEADING] \
               or not (re.search(r"\w", allText) and self.isTextBlockElement(obj)):
                string, start, end = allText, 0, len(allText)
                s = stringForDebug(string)
                tokens = [f"WEB: Text at offset {offset} for", obj, "using", boundary, ":",
                          f"'{s}', Start: {start}, End: {end}."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                return string, start, end

        if boundary == Atspi.TextBoundaryType.LINE_START and self.treatAsEndOfLine(obj, offset):
            offset -= 1
            tokens = ["WEB: Line sought for", obj, "at end of text. Adjusting offset to",
                      offset, "."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)

        offset = max(0, offset)

        # TODO - JD: Audit callers so we don't have to use boundaries.
        # Also, can the logic be entirely moved to AXText?
        if boundary == Atspi.TextBoundaryType.LINE_START:
            string, start, end = AXText.get_line_at_offset(obj, offset)
        elif boundary == Atspi.TextBoundaryType.SENTENCE_START:
            string, start, end = AXText.get_sentence_at_offset(obj, offset)
        elif boundary == Atspi.TextBoundaryType.WORD_START:
            string, start, end = AXText.get_word_at_offset(obj, offset)
        elif boundary == Atspi.TextBoundaryType.CHAR:
            string, start, end = AXText.get_character_at_offset(obj, offset)
        else:
            string, start, end = AXText.get_line_at_offset(obj, offset)

        s = stringForDebug(string)
        tokens = [f"WEB: Text at offset {offset} for", obj, "using", boundary, ":",
                  f"'{s}', Start: {start}, End: {end}."]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        # https://bugzilla.mozilla.org/show_bug.cgi?id=1141181
        needSadHack = boundary == Atspi.TextBoundaryType.SENTENCE_START and allText \
           and (string, start, end) == ("", 0, 0)

        if needSadHack:
            sadString, sadStart, sadEnd = self.__findSentence(obj, offset)
            s = stringForDebug(sadString)
            tokens = ["HACK: Attempting to recover from above failure. Result:",
                      f"'{s}', Start: {sadStart}, End: {sadEnd}."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return sadString, sadStart, sadEnd

        return string, start, end

    def _getContentsForObj(self, obj, offset, boundary):
        tokens = ["WEB: Attempting to get contents for", obj, boundary]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        if not obj:
            return []

        if boundary == Atspi.TextBoundaryType.SENTENCE_START and self.isTime(obj):
            string = AXText.get_all_text(obj)
            if string:
                return [[obj, 0, len(string), string]]

        if boundary == Atspi.TextBoundaryType.LINE_START:
            if self.isMath(obj):
                if self.isMathTopLevel(obj):
                    math = obj
                else:
                    math = self.getMathAncestor(obj)
                return [[math, 0, 1, '']]

            treatAsText = self.treatAsTextObject(obj)
            if self.elementLinesAreSingleChars(obj):
                if AXObject.get_name(obj) and treatAsText:
                    tokens = ["WEB: Returning name as contents for", obj, "(single-char lines)"]
                    debug.printTokens(debug.LEVEL_INFO, tokens, True)
                    return [[obj, 0, AXText.get_character_count(obj), AXObject.get_name(obj)]]

                tokens = ["WEB: Returning all text as contents for", obj, "(single-char lines)"]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                boundary = None

            if self.elementLinesAreSingleWords(obj):
                if AXObject.get_name(obj) and treatAsText:
                    tokens = ["WEB: Returning name as contents for", obj, "(single-word lines)"]
                    debug.printTokens(debug.LEVEL_INFO, tokens, True)
                    return [[obj, 0, AXText.get_character_count(obj), AXObject.get_name(obj)]]

                tokens = ["WEB: Returning all text as contents for", obj, "(single-word lines)"]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                boundary = None

        if AXUtilities.is_internal_frame(obj) and AXObject.get_child_count(obj) == 1:
            return self._getContentsForObj(AXObject.get_child(obj, 0), 0, boundary)

        string, start, end = self._getTextAtOffset(obj, offset, boundary)
        if not string:
            return [[obj, start, end, string]]

        stringOffset = offset - start
        try:
            char = string[stringOffset]
        except Exception as error:
            msg = f"WEB: Could not get char {stringOffset} for '{string}': {error}"
            debug.printMessage(debug.LEVEL_INFO, msg, True)
        else:
            if char == self.EMBEDDED_OBJECT_CHARACTER:
                child = AXHypertext.get_child_at_offset(obj, offset)
                if child:
                    return self._getContentsForObj(child, 0, boundary)

        ranges = [m.span() for m in re.finditer("[^\ufffc]+", string)]
        strings = list(filter(lambda x: x[0] <= stringOffset <= x[1], ranges))
        if len(strings) == 1:
            rangeStart, rangeEnd = strings[0]
            start += rangeStart
            string = string[rangeStart:rangeEnd]
            end = start + len(string)

        if boundary in [Atspi.TextBoundaryType.WORD_START, Atspi.TextBoundaryType.CHAR]:
            return [[obj, start, end, string]]

        return self.adjustContentsForLanguage([[obj, start, end, string]])

    def getSentenceContentsAtOffset(self, obj, offset, useCache=True):
        self._canHaveCaretContextDecision = {}
        rv = self._getSentenceContentsAtOffset(obj, offset, useCache)
        self._canHaveCaretContextDecision = {}
        return rv

    def _getSentenceContentsAtOffset(self, obj, offset, useCache=True):
        if not obj:
            return []

        offset = max(0, offset)

        if useCache:
            if self.findObjectInContents(
                    obj, offset, self._currentSentenceContents, usingCache=True) != -1:
                return self._currentSentenceContents

        boundary = Atspi.TextBoundaryType.SENTENCE_START
        objects = self._getContentsForObj(obj, offset, boundary)
        if AXUtilities.is_editable(obj):
            if AXUtilities.is_focused(obj):
                return objects
            if self.isContentEditableWithEmbeddedObjects(obj):
                return objects

        def _treatAsSentenceEnd(x):
            xObj, xStart, xEnd, xString = x
            if not self.isTextBlockElement(xObj):
                return False

            if self.treatAsTextObject(xObj) and 0 < AXText.get_character_count(xObj) <= xEnd:
                return True

            if 0 <= xStart <= 5:
                xString = " ".join(xString.split()[1:])

            match = re.search(r"\S[\.\!\?]+(\s|\Z)", xString)
            return match is not None

        # Check for things in the same sentence before this object.
        firstObj, firstStart, firstEnd, firstString = objects[0]
        while firstObj and firstString:
            if self.isTextBlockElement(firstObj):
                if firstStart == 0:
                    break
            elif self.isTextBlockElement(AXObject.get_parent(firstObj)):
                if AXHypertext.get_character_offset_in_parent(firstObj) == 0:
                    break

            prevObj, pOffset = self.findPreviousCaretInOrder(firstObj, firstStart)
            onLeft = self._getContentsForObj(prevObj, pOffset, boundary)
            onLeft = list(filter(lambda x: x not in objects, onLeft))
            endsOnLeft = list(filter(_treatAsSentenceEnd, onLeft))
            if endsOnLeft:
                i = onLeft.index(endsOnLeft[-1])
                onLeft = onLeft[i+1:]

            if not onLeft:
                break

            objects[0:0] = onLeft
            firstObj, firstStart, firstEnd, firstString = objects[0]

        # Check for things in the same sentence after this object.
        while not _treatAsSentenceEnd(objects[-1]):
            lastObj, lastStart, lastEnd, lastString = objects[-1]
            nextObj, nOffset = self.findNextCaretInOrder(lastObj, lastEnd - 1)
            onRight = self._getContentsForObj(nextObj, nOffset, boundary)
            onRight = list(filter(lambda x: x not in objects, onRight))
            if not onRight:
                break

            objects.extend(onRight)

        if useCache:
            self._currentSentenceContents = objects

        return objects

    def getCharacterContentsAtOffset(self, obj, offset, useCache=True):
        self._canHaveCaretContextDecision = {}
        rv = self._getCharacterContentsAtOffset(obj, offset, useCache)
        self._canHaveCaretContextDecision = {}
        return rv

    def _getCharacterContentsAtOffset(self, obj, offset, useCache=True):
        if not obj:
            return []

        offset = max(0, offset)

        if useCache:
            if self.findObjectInContents(
                    obj, offset, self._currentCharacterContents, usingCache=True) != -1:
                return self._currentCharacterContents

        boundary = Atspi.TextBoundaryType.CHAR
        objects = self._getContentsForObj(obj, offset, boundary)
        if useCache:
            self._currentCharacterContents = objects

        return objects

    def getWordContentsAtOffset(self, obj, offset, useCache=True):
        self._canHaveCaretContextDecision = {}
        rv = self._getWordContentsAtOffset(obj, offset, useCache)
        self._canHaveCaretContextDecision = {}
        return rv

    def _getWordContentsAtOffset(self, obj, offset, useCache=True):
        if not obj:
            return []

        offset = max(0, offset)

        if useCache:
            if self.findObjectInContents(
                    obj, offset, self._currentWordContents, usingCache=True) != -1:
                self._debugContentsInfo(obj, offset, self._currentWordContents, "Word (cached)")
                return self._currentWordContents

        boundary = Atspi.TextBoundaryType.WORD_START
        objects = self._getContentsForObj(obj, offset, boundary)
        extents = self.getExtents(obj, offset, offset + 1)

        def _include(x):
            if x in objects:
                return False

            xObj, xStart, xEnd, xString = x
            if xStart == xEnd or not xString:
                return False

            xExtents = self.getExtents(xObj, xStart, xStart + 1)
            return self.extentsAreOnSameLine(extents, xExtents)

        # Check for things in the same word to the left of this object.
        firstObj, firstStart, firstEnd, firstString = objects[0]
        prevObj, pOffset = self.findPreviousCaretInOrder(firstObj, firstStart)
        while prevObj and firstString and prevObj != firstObj:
            char = AXText.get_character_at_offset(prevObj, pOffset)[0]
            if not char or char.isspace():
                break

            onLeft = self._getContentsForObj(prevObj, pOffset, boundary)
            onLeft = list(filter(_include, onLeft))
            if not onLeft:
                break

            if self._contentIsSubsetOf(objects[0], onLeft[-1]):
                objects.pop(0)

            objects[0:0] = onLeft
            firstObj, firstStart, firstEnd, firstString = objects[0]
            prevObj, pOffset = self.findPreviousCaretInOrder(firstObj, firstStart)

        # Check for things in the same word to the right of this object.
        lastObj, lastStart, lastEnd, lastString = objects[-1]
        while lastObj and lastString and not lastString[-1].isspace():
            nextObj, nOffset = self.findNextCaretInOrder(lastObj, lastEnd - 1)
            if nextObj == lastObj:
                break

            onRight = self._getContentsForObj(nextObj, nOffset, boundary)
            if onRight and self._contentIsSubsetOf(objects[0], onRight[-1]):
                onRight = onRight[0:-1]

            onRight = list(filter(_include, onRight))
            if not onRight:
                break

            objects.extend(onRight)
            lastObj, lastStart, lastEnd, lastString = objects[-1]

        # We want to treat the list item marker as its own word.
        firstObj, firstStart, firstEnd, firstString = objects[0]
        if firstStart == 0 and AXUtilities.is_list_item(firstObj):
            objects = [objects[0]]

        if useCache:
            self._currentWordContents = objects

        self._debugContentsInfo(obj, offset, objects, "Word (not cached)")
        return objects

    def getObjectContentsAtOffset(self, obj, offset=0, useCache=True):
        self._canHaveCaretContextDecision = {}
        rv = self._getObjectContentsAtOffset(obj, offset, useCache)
        self._canHaveCaretContextDecision = {}
        return rv

    def _getObjectContentsAtOffset(self, obj, offset=0, useCache=True):
        if not obj:
            return []

        if AXObject.is_dead(obj):
            msg = "ERROR: Cannot get object contents at offset for dead object."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return []

        offset = max(0, offset)

        if useCache:
            if self.findObjectInContents(
                    obj, offset, self._currentObjectContents, usingCache=True) != -1:
                self._debugContentsInfo(
                    obj, offset, self._currentObjectContents, "Object (cached)")
                return self._currentObjectContents

        objIsLandmark = self.isLandmark(obj)

        def _isInObject(x):
            if not x:
                return False
            if x == obj:
                return True
            return _isInObject(AXObject.get_parent(x))

        def _include(x):
            if x in objects:
                return False

            xObj, xStart, xEnd, xString = x
            if xStart == xEnd:
                return False

            if objIsLandmark and self.isLandmark(xObj) and obj != xObj:
                return False

            return _isInObject(xObj)

        objects = self._getContentsForObj(obj, offset, None)
        if not objects:
            tokens = ["ERROR: Cannot get object contents for", obj, f"at offset {offset}"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return []

        lastObj, lastStart, lastEnd, lastString = objects[-1]
        nextObj, nOffset = self.findNextCaretInOrder(lastObj, lastEnd - 1)
        while nextObj:
            onRight = self._getContentsForObj(nextObj, nOffset, None)
            onRight = list(filter(_include, onRight))
            if not onRight:
                break

            objects.extend(onRight)
            lastObj, lastEnd = objects[-1][0], objects[-1][2]
            nextObj, nOffset = self.findNextCaretInOrder(lastObj, lastEnd - 1)

        if useCache:
            self._currentObjectContents = objects

        self._debugContentsInfo(obj, offset, objects, "Object (not cached)")
        return objects

    def _contentIsSubsetOf(self, contentA, contentB):
        objA, startA, endA, stringA = contentA
        objB, startB, endB, stringB = contentB
        if objA == objB:
            setA = set(range(startA, endA))
            setB = set(range(startB, endB))
            return setA.issubset(setB)

        return False

    def _debugContentsInfo(self, obj, offset, contents, contentsMsg=""):
        if debug.LEVEL_INFO < debug.debugLevel:
            return

        tokens = ["WEB: ", contentsMsg, "for", obj, "at offset", offset, ":"]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        indent = " " * 8
        for i, (acc, start, end, string) in enumerate(contents):
            try:
                extents = self.getExtents(acc, start, end)
            except Exception as error:
                extents = f"(exception: {error})"
            msg = f"     {i}. chars: {start}-{end}: '{string}' extents={extents}\n"
            msg += debug.getAccessibleDetails(debug.LEVEL_INFO, acc, indent)
            debug.printMessage(debug.LEVEL_INFO, msg, True)

    def treatAsEndOfLine(self, obj, offset):
        if not self.isContentEditableWithEmbeddedObjects(obj):
            return False

        if not AXObject.supports_text(obj):
            return False

        if self.isDocument(obj):
            return False

        if offset == AXText.get_character_count(obj):
            tokens = ["WEB: ", obj, "offset", offset, "is end of line: offset is characterCount"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        # Do not treat a literal newline char as the end of line. When there is an
        # actual newline character present, user agents should give us the right value
        # for the line at that offset. Here we are trying to figure out where asking
        # for the line at offset will give us the next line rather than the line where
        # the cursor is physically blinking.
        char = AXText.get_character_at_offset(obj, offset)[0]
        if char == self.EMBEDDED_OBJECT_CHARACTER:
            prevExtents = self.getExtents(obj, offset - 1, offset)
            thisExtents = self.getExtents(obj, offset, offset + 1)
            sameLine = self.extentsAreOnSameLine(prevExtents, thisExtents)
            tokens = ["WEB: ", obj, "offset", offset, "is [obj]. Same line: ",
                      sameLine, "Is end of line: ", not sameLine]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return not sameLine

        return False

    def getLineContentsAtOffset(self, obj, offset, layoutMode=None, useCache=True):
        self._canHaveCaretContextDecision = {}
        rv = self._getLineContentsAtOffset(obj, offset, layoutMode, useCache)
        self._canHaveCaretContextDecision = {}
        return rv

    def _getLineContentsAtOffset(self, obj, offset, layoutMode=None, useCache=True):
        startTime = time.time()
        if not obj:
            return []

        if AXObject.is_dead(obj):
            msg = "ERROR: Cannot get line contents at offset for dead object."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return []

        offset = max(0, offset)
        if (AXUtilities.is_tool_bar(obj) or AXUtilities.is_menu_bar(obj)) \
                and not self._treatObjectAsWhole(obj):
            child = AXHypertext.get_child_at_offset(obj, offset)
            if child:
                obj = child
                offset = 0

        if useCache:
            if self.findObjectInContents(
                    obj, offset, self._currentLineContents, usingCache=True) != -1:
                self._debugContentsInfo(
                    obj, offset, self._currentLineContents, "Line (cached)")
                return self._currentLineContents

        if layoutMode is None:
            layoutMode = settings_manager.getManager().getSetting('layoutMode') \
                or self._script.inFocusMode()

        objects = []
        if offset > 0 and self.treatAsEndOfLine(obj, offset):
            extents = self.getExtents(obj, offset - 1, offset)
        else:
            extents = self.getExtents(obj, offset, offset + 1)

        if self.isInlineListDescendant(obj):
            container = self.listForInlineListDescendant(obj)
            if container:
                extents = self.getExtents(container, 0, 1)

        objBanner = AXObject.find_ancestor(obj, self.isLandmarkBanner)

        def _include(x):
            if x in objects:
                return False

            xObj, xStart, xEnd, xString = x
            if xStart == xEnd:
                return False

            xExtents = self.getExtents(xObj, xStart, xStart + 1)

            if obj != xObj:
                if self.isLandmark(obj) and self.isLandmark(xObj):
                    return False
                if self.isLink(obj) and self.isLink(xObj):
                    xObjBanner = AXObject.find_ancestor(xObj, self.isLandmarkBanner)
                    if (objBanner or xObjBanner) and objBanner != xObjBanner:
                        return False
                    if abs(extents[0] - xExtents[0]) <= 1 and abs(extents[1] - xExtents[1]) <= 1:
                        # This happens with dynamic skip links such as found on Wikipedia.
                        return False
                elif self.isBlockListDescendant(obj) != self.isBlockListDescendant(xObj):
                    return False
                elif AXUtilities.is_tree_related(obj) and AXUtilities.is_tree_related(xObj):
                    return False
                elif AXUtilities.is_heading(obj) and AXComponent.has_no_size(obj):
                    return False
                elif AXUtilities.is_heading(xObj) and AXComponent.has_no_size(xObj):
                    return False

            if self.isMathTopLevel(xObj) or self.isMath(obj):
                onSameLine = self.extentsAreOnSameLine(extents, xExtents, extents[3])
            elif self.isTextSubscriptOrSuperscript(xObj):
                onSameLine = self.extentsAreOnSameLine(extents, xExtents, xExtents[3])
            else:
                onSameLine = self.extentsAreOnSameLine(extents, xExtents)
            return onSameLine

        boundary = Atspi.TextBoundaryType.LINE_START
        objects = self._getContentsForObj(obj, offset, boundary)
        if not layoutMode:
            if useCache:
                self._currentLineContents = objects

            self._debugContentsInfo(obj, offset, objects, "Line (not layout mode)")
            return objects

        if not (objects and objects[0]):
            tokens = ["WEB: Error. No objects found for", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return []

        firstObj, firstStart, firstEnd, firstString = objects[0]
        if (extents[2] == 0 and extents[3] == 0) or self.isMath(firstObj):
            extents = self.getExtents(firstObj, firstStart, firstEnd)

        lastObj, lastStart, lastEnd, lastString = objects[-1]
        if self.isMathTopLevel(lastObj):
            lastObj, lastEnd = self.lastContext(lastObj)
            lastEnd += 1

        document = self.getDocumentForObject(obj)
        prevObj, pOffset = self.findPreviousCaretInOrder(firstObj, firstStart)
        nextObj, nOffset = self.findNextCaretInOrder(lastObj, lastEnd - 1)

        # Check for things on the same line to the left of this object.
        prevStartTime = time.time()
        while prevObj and self.getDocumentForObject(prevObj) == document:
            char = AXText.get_character_at_offset(prevObj, pOffset)[0]
            if char.isspace():
                prevObj, pOffset = self.findPreviousCaretInOrder(prevObj, pOffset)

            char = AXText.get_character_at_offset(prevObj, pOffset)[0]
            if char == "\n" and firstObj == prevObj:
                break

            onLeft = self._getContentsForObj(prevObj, pOffset, boundary)
            onLeft = list(filter(_include, onLeft))
            if not onLeft:
                break

            if self._contentIsSubsetOf(objects[0], onLeft[-1]):
                objects.pop(0)

            objects[0:0] = onLeft
            firstObj, firstStart = objects[0][0], objects[0][1]
            prevObj, pOffset = self.findPreviousCaretInOrder(firstObj, firstStart)

        prevEndTime = time.time()
        msg = f"INFO: Time to get line contents on left: {prevEndTime - prevStartTime:.4f}s"
        debug.printMessage(debug.LEVEL_INFO, msg, True)

        # Check for things on the same line to the right of this object.
        nextStartTime = time.time()
        while nextObj and self.getDocumentForObject(nextObj) == document:
            char = AXText.get_character_at_offset(nextObj, nOffset)[0]
            if char.isspace():
                nextObj, nOffset = self.findNextCaretInOrder(nextObj, nOffset)

            char = AXText.get_character_at_offset(nextObj, nOffset)[0]
            if char == "\n" and lastObj == nextObj:
                break

            onRight = self._getContentsForObj(nextObj, nOffset, boundary)
            if onRight and self._contentIsSubsetOf(objects[0], onRight[-1]):
                onRight = onRight[0:-1]

            onRight = list(filter(_include, onRight))
            if not onRight:
                break

            objects.extend(onRight)
            lastObj, lastEnd = objects[-1][0], objects[-1][2]
            if self.isMathTopLevel(lastObj):
                lastObj, lastEnd = self.lastContext(lastObj)
                lastEnd += 1

            nextObj, nOffset = self.findNextCaretInOrder(lastObj, lastEnd - 1)

        nextEndTime = time.time()
        msg = f"INFO: Time to get line contents on right: {nextEndTime - nextStartTime:.4f}s"
        debug.printMessage(debug.LEVEL_INFO, msg, True)

        firstObj, firstStart, firstEnd, firstString = objects[0]
        if firstString == "\n" and len(objects) > 1:
            objects.pop(0)

        if useCache:
            self._currentLineContents = objects

        msg = f"INFO: Time to get line contents: {time.time() - startTime:.4f}s"
        debug.printMessage(debug.LEVEL_INFO, msg, True)

        self._debugContentsInfo(obj, offset, objects, "Line (layout mode)")

        self._canHaveCaretContextDecision = {}
        return objects

    def getPreviousLineContents(self, obj=None, offset=-1, layoutMode=None, useCache=True):
        if obj is None:
            obj, offset = self.getCaretContext()

        tokens = ["WEB: Current context is: ", obj, ", ", offset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        if not AXObject.is_valid(obj):
            tokens = ["WEB: Current context obj", obj, "is not valid. Clearing cache."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            self.clearCachedObjects()

            obj, offset = self.getCaretContext()
            tokens = ["WEB: Now Current context is: ", obj, ", ", offset]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)

        line = self.getLineContentsAtOffset(obj, offset, layoutMode, useCache)
        if not (line and line[0]):
            return []

        firstObj, firstOffset = line[0][0], line[0][1]
        tokens = ["WEB: First context on line is: ", firstObj, ", ", firstOffset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        skipSpace = not self.elementIsPreformattedText(firstObj)
        obj, offset = self.previousContext(firstObj, firstOffset, skipSpace)
        if not obj and firstObj:
            tokens = ["WEB: Previous context is: ", obj, ", ", offset, ". Trying again."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            self.clearCachedObjects()
            obj, offset = self.previousContext(firstObj, firstOffset, skipSpace)

        tokens = ["WEB: Previous context is: ", obj, ", ", offset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        contents = self.getLineContentsAtOffset(obj, offset, layoutMode, useCache)
        if not contents:
            tokens = ["WEB: Could not get line contents for ", obj, ", ", offset]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return []

        if line == contents:
            obj, offset = self.previousContext(obj, offset, True)
            tokens = ["WEB: Got same line. Trying again with ", obj, ", ", offset]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            contents = self.getLineContentsAtOffset(obj, offset, layoutMode, useCache)

        if line == contents:
            start = AXHypertext.get_link_start_offset(obj)
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            if start >= 0:
                parent = AXObject.get_parent(obj)
                obj, offset = self.previousContext(parent, start, True)
                tokens = ["WEB: Trying again with", obj, ", ", offset]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                contents = self.getLineContentsAtOffset(obj, offset, layoutMode, useCache)

        return contents

    def getNextLineContents(self, obj=None, offset=-1, layoutMode=None, useCache=True):
        if obj is None:
            obj, offset = self.getCaretContext()

        tokens = ["WEB: Current context is: ", obj, ", ", offset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        if not AXObject.is_valid(obj):
            tokens = ["WEB: Current context obj", obj, "is not valid. Clearing cache."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            self.clearCachedObjects()

            obj, offset = self.getCaretContext()
            tokens = ["WEB: Now Current context is: ", obj, ", ", offset]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)

        line = self.getLineContentsAtOffset(obj, offset, layoutMode, useCache)
        if not (line and line[0]):
            return []

        lastObj, lastOffset = line[-1][0], line[-1][2] - 1
        math = self.getMathAncestor(lastObj)
        if math:
            lastObj, lastOffset = self.lastContext(math)

        tokens = ["WEB: Last context on line is: ", lastObj, ", ", lastOffset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        skipSpace = not self.elementIsPreformattedText(lastObj)
        obj, offset = self.nextContext(lastObj, lastOffset, skipSpace)
        if not obj and lastObj:
            tokens = ["WEB: Next context is: ", obj, ", ", offset, ". Trying again."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            self.clearCachedObjects()
            obj, offset = self.nextContext(lastObj, lastOffset, skipSpace)

        tokens = ["WEB: Next context is: ", obj, ", ", offset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        contents = self.getLineContentsAtOffset(obj, offset, layoutMode, useCache)
        if line == contents:
            obj, offset = self.nextContext(obj, offset, True)
            tokens = ["WEB: Got same line. Trying again with ", obj, ", ", offset]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            contents = self.getLineContentsAtOffset(obj, offset, layoutMode, useCache)

        if line == contents:
            end = AXHypertext.get_link_end_offset(obj)
            if end >= 0:
                parent = AXObject.get_parent(obj)
                obj, offset = self.nextContext(parent, end, True)
                tokens = ["WEB: Trying again with", obj, ", ", offset]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                contents = self.getLineContentsAtOffset(obj, offset, layoutMode, useCache)

        if not contents:
            tokens = ["WEB: Could not get line contents for ", obj, ", ", offset]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return []

        return contents

    def updateCachedTextSelection(self, obj):
        if not self.inDocumentContent(obj):
            super().updateCachedTextSelection(obj)
            return

        if self.hasPresentableText(obj):
            super().updateCachedTextSelection(obj)

    def _findSelectionBoundaryObject(self, root, findStart=True):
        string = AXText.get_selected_text(root)[0]
        if not string:
            return None

        if findStart and not string.startswith(self.EMBEDDED_OBJECT_CHARACTER):
            return root

        if not findStart and not string.endswith(self.EMBEDDED_OBJECT_CHARACTER):
            return root

        indices = list(range(AXObject.get_child_count(root)))
        if not findStart:
            indices.reverse()

        for i in indices:
            result = self._findSelectionBoundaryObject(root[i], findStart)
            if result:
                return result

        return None

    def _getSelectionAnchorAndFocus(self, root):
        obj1 = self._findSelectionBoundaryObject(root, True)
        obj2 = self._findSelectionBoundaryObject(root, False)
        return obj1, obj2

    def _getSubtree(self, startObj, endObj):
        if not (startObj and endObj):
            return []

        if AXObject.is_dead(startObj):
            msg = "INFO: Cannot get subtree: Start object is dead."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return []

        def _include(x):
            return x is not None

        def _exclude(x):
            return self.isStaticTextLeaf(x)

        subtree = []
        startObjParent = AXObject.get_parent(startObj)
        for i in range(AXObject.get_index_in_parent(startObj),
                        AXObject.get_child_count(startObjParent)):
            child = AXObject.get_child(startObjParent, i)
            if self.isStaticTextLeaf(child):
                continue
            subtree.append(child)
            subtree.extend(self.findAllDescendants(child, _include, _exclude))
            if endObj in subtree:
                break

        if endObj == startObj:
            return subtree

        if endObj not in subtree:
            subtree.append(endObj)
            subtree.extend(self.findAllDescendants(endObj, _include, _exclude))

        endObjParent = AXObject.get_parent(endObj)
        endObjIndex = AXObject.get_index_in_parent(endObj)
        lastObj = AXObject.get_child(endObjParent, endObjIndex + 1) or endObj

        try:
            endIndex = subtree.index(lastObj)
        except ValueError:
            pass
        else:
            if lastObj == endObj:
                endIndex += 1
            subtree = subtree[:endIndex]

        return subtree

    def handleTextSelectionChange(self, obj, speakMessage=True):
        if not self.inDocumentContent(obj):
            return super().handleTextSelectionChange(obj)

        oldStart, oldEnd = \
            self._script.pointOfReference.get('selectionAnchorAndFocus', (None, None))
        start, end = self._getSelectionAnchorAndFocus(obj)
        self._script.pointOfReference['selectionAnchorAndFocus'] = (start, end)

        def _cmp(obj1, obj2):
            return self.pathComparison(AXObject.get_path(obj1), AXObject.get_path(obj2))

        oldSubtree = self._getSubtree(oldStart, oldEnd)
        if start == oldStart and end == oldEnd:
            descendants = oldSubtree
        else:
            newSubtree = self._getSubtree(start, end)
            descendants = sorted(set(oldSubtree).union(newSubtree), key=functools.cmp_to_key(_cmp))

        if not descendants:
            return False

        for descendant in descendants:
            if descendant not in (oldStart, oldEnd, start, end) \
               and AXObject.find_ancestor(descendant, lambda x: x in descendants):
                super().updateCachedTextSelection(descendant)
            else:
                super().handleTextSelectionChange(descendant, speakMessage)

        return True

    def inTopLevelWebApp(self, obj=None):
        if not obj:
            obj = focus_manager.getManager().get_locus_of_focus()

        rv = self._inTopLevelWebApp.get(hash(obj))
        if rv is not None:
            return rv

        document = self.getDocumentForObject(obj)
        if not document and self.isDocument(obj):
            document = obj

        rv = self.isTopLevelWebApp(document)
        self._inTopLevelWebApp[hash(obj)] = rv
        return rv

    def isTopLevelWebApp(self, obj):
        if AXUtilities.is_embedded(obj) \
           and not self.getDocumentForObject(AXObject.get_parent(obj)):
            uri = AXDocument.get_uri(obj)
            rv = bool(uri and uri.startswith("http"))
            tokens = ["WEB:", obj, "is top-level web application:", rv, "(URI:", uri, ")"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return rv

        return False

    def forceBrowseModeForWebAppDescendant(self, obj):
        if not self.isWebAppDescendant(obj):
            return False

        if AXUtilities.is_tool_tip(obj):
            return AXUtilities.is_focused(obj)

        if AXUtilities.is_document_web(obj):
            return not self.isFocusModeWidget(obj)

        return False

    def isFocusModeWidget(self, obj):
        if AXUtilities.is_editable(obj):
            tokens = ["WEB:", obj, "is focus mode widget because it's editable"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        if AXUtilities.is_expandable(obj) and AXUtilities.is_focusable(obj) \
           and not AXUtilities.is_link(obj):
            tokens = ["WEB:", obj, "is focus mode widget because it's expandable and focusable"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        alwaysFocusModeRoles = [Atspi.Role.COMBO_BOX,
                                Atspi.Role.ENTRY,
                                Atspi.Role.LIST_BOX,
                                Atspi.Role.MENU,
                                Atspi.Role.MENU_ITEM,
                                Atspi.Role.CHECK_MENU_ITEM,
                                Atspi.Role.RADIO_MENU_ITEM,
                                Atspi.Role.PAGE_TAB,
                                Atspi.Role.PASSWORD_TEXT,
                                Atspi.Role.PROGRESS_BAR,
                                Atspi.Role.SLIDER,
                                Atspi.Role.SPIN_BUTTON,
                                Atspi.Role.TOOL_BAR,
                                Atspi.Role.TREE_ITEM,
                                Atspi.Role.TREE_TABLE,
                                Atspi.Role.TREE]

        role = AXObject.get_role(obj)
        if role in alwaysFocusModeRoles:
            tokens = ["WEB:", obj, "is focus mode widget due to its role"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        if role in [Atspi.Role.TABLE_CELL, Atspi.Role.TABLE] \
           and AXTable.is_layout_table(AXTable.get_table(obj)):
            tokens = ["WEB:", obj, "is not focus mode widget because it's layout only"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return False

        if AXUtilities.is_list_item(obj):
            rv = AXObject.find_ancestor(obj, AXUtilities.is_list_box)
            if rv:
                tokens = ["WEB:", obj, "is focus mode widget because it's a listbox descendant"]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return rv

        if self.isButtonWithPopup(obj):
            tokens = ["WEB:", obj, "is focus mode widget because it's a button with popup"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        focusModeRoles = [Atspi.Role.EMBEDDED,
                          Atspi.Role.TABLE_CELL,
                          Atspi.Role.TABLE]

        if role in focusModeRoles \
           and not self.isTextBlockElement(obj) \
           and not self.hasNameAndActionAndNoUsefulChildren(obj) \
           and not AXDocument.is_pdf(self.documentFrame()):
            tokens = ["WEB:", obj, "is focus mode widget based on presumed functionality"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        if self.isGridDescendant(obj):
            tokens = ["WEB:", obj, "is focus mode widget because it's a grid descendant"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        if self.isMenuDescendant(obj):
            tokens = ["WEB:", obj, "is focus mode widget because it's a menu descendant"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        if self.isToolBarDescendant(obj):
            tokens = ["WEB:", obj, "is focus mode widget because it's a toolbar descendant"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        if self.isContentEditableWithEmbeddedObjects(obj):
            tokens = ["WEB:", obj, "is focus mode widget because it's content editable"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        return False

    def _textBlockElementRoles(self):
        roles = [Atspi.Role.ARTICLE,
                 Atspi.Role.CAPTION,
                 Atspi.Role.COLUMN_HEADER,
                 Atspi.Role.COMMENT,
                 Atspi.Role.DEFINITION,
                 Atspi.Role.DESCRIPTION_LIST,
                 Atspi.Role.DESCRIPTION_TERM,
                 Atspi.Role.DESCRIPTION_VALUE,
                 Atspi.Role.DOCUMENT_FRAME,
                 Atspi.Role.DOCUMENT_WEB,
                 Atspi.Role.FOOTER,
                 Atspi.Role.FORM,
                 Atspi.Role.HEADING,
                 Atspi.Role.LIST,
                 Atspi.Role.LIST_ITEM,
                 Atspi.Role.PARAGRAPH,
                 Atspi.Role.ROW_HEADER,
                 Atspi.Role.SECTION,
                 Atspi.Role.STATIC,
                 Atspi.Role.TEXT,
                 Atspi.Role.TABLE_CELL]

        # Remove this check when we bump dependencies to 2.34
        try:
            roles.append(Atspi.Role.CONTENT_DELETION)
            roles.append(Atspi.Role.CONTENT_INSERTION)
        except Exception:
            pass

        # Remove this check when we bump dependencies to 2.36
        try:
            roles.append(Atspi.Role.MARK)
            roles.append(Atspi.Role.SUGGESTION)
        except Exception:
            pass

        return roles

    def mnemonicShortcutAccelerator(self, obj):
        attrs = AXObject.get_attributes_dict(obj)
        keys = map(lambda x: x.replace("+", " "), attrs.get("keyshortcuts", "").split(" "))
        keys = map(lambda x: x.replace(" ", "+"), map(self.labelFromKeySequence, keys))
        rv = ["", " ".join(keys), ""]
        if list(filter(lambda x: x, rv)):
            return rv

        return super().mnemonicShortcutAccelerator(obj)

    def unrelatedLabels(self, root, onlyShowing=True, minimumWords=3):
        if not (root and self.inDocumentContent(root)):
            return super().unrelatedLabels(root, onlyShowing, minimumWords)

        return []

    def isFocusableWithMathChild(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isFocusableWithMathChild.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        if AXUtilities.is_focusable(obj) \
            and not self.isDocument(obj):
            for child in AXObject.iter_children(obj, self.isMathTopLevel):
                rv = True
                break

        self._isFocusableWithMathChild[hash(obj)] = rv
        return rv

    def isFocusedWithMathChild(self, obj):
        if not self.isFocusableWithMathChild(obj):
            return False
        return AXUtilities.is_focused(obj)

    def isTextBlockElement(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isTextBlockElement.get(hash(obj))
        if rv is not None:
            return rv

        role = AXObject.get_role(obj)
        textBlockElements = self._textBlockElementRoles()
        if role not in textBlockElements:
            rv = False
        elif not AXObject.supports_text(obj):
            rv = False
        elif AXUtilities.is_editable(obj):
            rv = False
        elif self.isGridCell(obj):
            rv = False
        elif AXUtilities.is_document(obj):
            rv = True
        elif self.isCustomImage(obj):
            rv = False
        elif not AXUtilities.is_focusable(obj):
            rv = not self.hasNameAndActionAndNoUsefulChildren(obj)
        else:
            rv = False

        self._isTextBlockElement[hash(obj)] = rv
        return rv

    def _advanceCaretInEmptyObject(self, obj):
        if AXUtilities.is_table_cell(obj) and not self.treatAsTextObject(obj):
            return not self._script.caretNavigation.last_input_event_was_navigation_command()

        return True

    def textAtPoint(self, obj, x, y, boundary=None):
        if boundary is None:
            boundary = Atspi.TextBoundaryType.LINE_START

        string, start, end = super().textAtPoint(obj, x, y, boundary)
        if string == self.EMBEDDED_OBJECT_CHARACTER:
            child = AXHypertext.get_child_at_offset(obj, start)
            if child:
                return self.textAtPoint(child, x, y, boundary)

        return string, start, end

    def _treatAlertsAsDialogs(self):
        return False

    def treatAsDiv(self, obj, offset=None):
        if not (obj and self.inDocumentContent(obj)):
            return False

        if self.isDescriptionList(obj):
            return False

        if AXUtilities.is_list(obj) and offset is not None:
            string = self.substring(obj, offset, offset + 1)
            if string and string != self.EMBEDDED_OBJECT_CHARACTER:
                return True

        childCount = AXObject.get_child_count(obj)
        if AXUtilities.is_panel(obj) and not childCount:
            return True

        rv = self._treatAsDiv.get(hash(obj))
        if rv is not None:
            return rv

        validRoles = self._validChildRoles.get(AXObject.get_role(obj))
        if validRoles:
            if not childCount:
                rv = True
            else:
                def pred1(x):
                    return x is not None and AXObject.get_role(x) not in validRoles

                rv = bool([x for x in AXObject.iter_children(obj, pred1)])

        if not rv:
            parent = AXObject.get_parent(obj)
            validRoles = self._validChildRoles.get(parent)
            if validRoles:
                def pred2(x):
                    return x is not None and AXObject.get_role(x) not in validRoles

                rv = bool([x for x in AXObject.iter_children(parent, pred2)])

        self._treatAsDiv[hash(obj)] = rv
        return rv

    def isAriaAlert(self, obj):
        return 'alert' in self._getXMLRoles(obj)

    def isBlockquote(self, obj):
        if super().isBlockquote(obj):
            return True

        return self._getTag(obj) == 'blockquote'

    def isComment(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isComment(obj)

        if AXUtilities.is_comment(obj):
            return True

        return 'comment' in self._getXMLRoles(obj)

    def isContentDeletion(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isContentDeletion(obj)

        if AXUtilities.is_content_deletion(obj):
            return True

        return 'deletion' in self._getXMLRoles(obj) or 'del' == self._getTag(obj)

    def isContentError(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isContentError(obj)

        if AXObject.get_role(obj) not in self._textBlockElementRoles():
            return False

        return AXUtilities.is_invalid_entry(obj)

    def isContentInsertion(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isContentInsertion(obj)

        if AXUtilities.is_content_insertion(obj):
            return True

        return 'insertion' in self._getXMLRoles(obj) or 'ins' == self._getTag(obj)

    def isContentMarked(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isContentMarked(obj)

        if AXUtilities.is_mark(obj):
            return True

        return 'mark' in self._getXMLRoles(obj) or 'mark' == self._getTag(obj)

    def isContentSuggestion(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isContentSuggestion(obj)

        if AXUtilities.is_suggestion(obj):
            return True

        return 'suggestion' in self._getXMLRoles(obj)

    def isCustomElement(self, obj):
        tag = self._getTag(obj)
        return tag and '-' in tag

    def isInlineIframe(self, obj):
        if not AXUtilities.is_internal_frame(obj):
            return False

        displayStyle = self._getDisplayStyle(obj)
        if "inline" not in displayStyle:
            return False

        return self.getDocumentForObject(obj) is not None

    def isInlineIframeDescendant(self, obj):
        if not obj:
            return False

        rv = self._isInlineIframeDescendant.get(hash(obj))
        if rv is not None:
            return rv

        ancestor = AXObject.find_ancestor(obj, self.isInlineIframe)
        rv = ancestor is not None
        self._isInlineIframeDescendant[hash(obj)] = rv
        return rv

    def isInlineSuggestion(self, obj):
        if not self.isContentSuggestion(obj):
            return False

        displayStyle = self._getDisplayStyle(obj)
        return "inline" in displayStyle

    def isSVG(self, obj):
        return 'svg' == self._getTag(obj)

    def isTextField(self, obj):
        if AXUtilities.is_text_input(obj):
            return True

        if AXUtilities.is_combo_box(obj):
            return self.isEditableComboBox(obj)

        return False

    def isFirstItemInInlineContentSuggestion(self, obj):
        suggestion = AXObject.find_ancestor(obj, self.isInlineSuggestion)
        if not (suggestion and AXObject.get_child_count(suggestion)):
            return False

        return suggestion[0] == obj

    def isLastItemInInlineContentSuggestion(self, obj):
        suggestion = AXObject.find_ancestor(obj, self.isInlineSuggestion)
        if not (suggestion and AXObject.get_child_count(suggestion)):
            return False

        return suggestion[-1] == obj

    def speakMathSymbolNames(self, obj=None):
        obj = obj or focus_manager.getManager().get_locus_of_focus()
        return self.isMath(obj)

    def isInMath(self):
        return self.isMath(focus_manager.getManager().get_locus_of_focus())

    def isMath(self, obj):
        tag = self._getTag(obj)
        rv = tag in ['math',
                     'maction',
                     'maligngroup',
                     'malignmark',
                     'menclose',
                     'merror',
                     'mfenced',
                     'mfrac',
                     'mglyph',
                     'mi',
                     'mlabeledtr',
                     'mlongdiv',
                     'mmultiscripts',
                     'mn',
                     'mo',
                     'mover',
                     'mpadded',
                     'mphantom',
                     'mprescripts',
                     'mroot',
                     'mrow',
                     'ms',
                     'mscarries',
                     'mscarry',
                     'msgroup',
                     'msline',
                     'mspace',
                     'msqrt',
                     'msrow',
                     'mstack',
                     'mstyle',
                     'msub',
                     'msup',
                     'msubsup',
                     'mtable',
                     'mtd',
                     'mtext',
                     'mtr',
                     'munder',
                     'munderover']

        return rv

    def isNoneElement(self, obj):
        return self._getTag(obj) == 'none'

    def isMathLayoutOnly(self, obj):
        return self._getTag(obj) in ['mrow', 'mstyle', 'merror', 'mpadded']

    def isMathMultiline(self, obj):
        return self._getTag(obj) in ['mtable', 'mstack', 'mlongdiv']

    def isMathEnclose(self, obj):
        return self._getTag(obj) == 'menclose'

    def isMathFenced(self, obj):
        return self._getTag(obj) == 'mfenced'

    def isMathFractionWithoutBar(self, obj):
        if not AXUtilities.is_math_fraction(obj):
            return False

        attrs = AXObject.get_attributes_dict(obj)
        linethickness = attrs.get('linethickness')
        if not linethickness:
            return False

        for char in linethickness:
            if char.isnumeric() and char != '0':
                return False

        return True

    def isMathPhantom(self, obj):
        return self._getTag(obj) == 'mphantom'

    def isMathMultiScript(self, obj):
        return self._getTag(obj) == 'mmultiscripts'

    def _isMathPrePostScriptSeparator(self, obj):
        return self._getTag(obj) == 'mprescripts'

    def isMathSubOrSuperScript(self, obj):
        return self._getTag(obj) in ['msub', 'msup', 'msubsup']

    def isMathTable(self, obj):
        return self._getTag(obj) == 'mtable'

    def isMathTableRow(self, obj):
        return self._getTag(obj) in ['mtr', 'mlabeledtr']

    def isMathTableCell(self, obj):
        return self._getTag(obj) == 'mtd'

    def isMathUnderOrOverScript(self, obj):
        return self._getTag(obj) in ['mover', 'munder', 'munderover']

    def _isMathSubElement(self, obj):
        return self._getTag(obj) == 'msub'

    def _isMathSupElement(self, obj):
        return self._getTag(obj) == 'msup'

    def _isMathSubsupElement(self, obj):
        return self._getTag(obj) == 'msubsup'

    def _isMathUnderElement(self, obj):
        return self._getTag(obj) == 'munder'

    def _isMathOverElement(self, obj):
        return self._getTag(obj) == 'mover'

    def _isMathUnderOverElement(self, obj):
        return self._getTag(obj) == 'munderover'

    def isMathSquareRoot(self, obj):
        return self._getTag(obj) == 'msqrt'

    def isMathToken(self, obj):
        return self._getTag(obj) in ['mi', 'mn', 'mo', 'mtext', 'ms', 'mspace']

    def isMathTopLevel(self, obj):
        return AXUtilities.is_math(obj)

    def getMathAncestor(self, obj):
        if not self.isMath(obj):
            return None

        if self.isMathTopLevel(obj):
            return obj

        return AXObject.find_ancestor(obj, self.isMathTopLevel)

    def getMathDenominator(self, obj):
        return AXObject.get_child(obj, 1)

    def getMathNumerator(self, obj):
        return AXObject.get_child(obj, 0)

    def getMathRootBase(self, obj):
        if self.isMathSquareRoot(obj):
            return obj

        return AXObject.get_child(obj, 0)

    def getMathRootIndex(self, obj):
        return AXObject.get_child(obj, 1)

    def getMathScriptBase(self, obj):
        if self.isMathSubOrSuperScript(obj) \
           or self.isMathUnderOrOverScript(obj) \
           or self.isMathMultiScript(obj):
            return AXObject.get_child(obj, 0)

        return None

    def getMathScriptSubscript(self, obj):
        if self._isMathSubElement(obj) or self._isMathSubsupElement(obj):
            return AXObject.get_child(obj, 1)

        return None

    def getMathScriptSuperscript(self, obj):
        if self._isMathSupElement(obj):
            return AXObject.get_child(obj, 1)

        if self._isMathSubsupElement(obj):
            return AXObject.get_child(obj, 2)

        return None

    def getMathScriptUnderscript(self, obj):
        if self._isMathUnderElement(obj) or self._isMathUnderOverElement(obj):
            return AXObject.get_child(obj, 1)

        return None

    def getMathScriptOverscript(self, obj):
        if self._isMathOverElement(obj):
            return AXObject.get_child(obj, 1)

        if self._isMathUnderOverElement(obj):
            return AXObject.get_child(obj, 2)

        return None

    def _getMathPrePostScriptSeparator(self, obj):
        for child in AXObject.iter_children(obj):
            if self._isMathPrePostScriptSeparator(child):
                return child

        return None

    def getMathPrescripts(self, obj):
        separator = self._getMathPrePostScriptSeparator(obj)
        if not separator:
            return []

        children = []
        child = AXObject.get_next_sibling(separator)
        while child:
            children.append(child)
            child = AXObject.get_next_sibling(child)

        return children

    def getMathPostscripts(self, obj):
        separator = self._getMathPrePostScriptSeparator(obj)
        children = []
        child = AXObject.get_child(obj, 1)
        while child and child != separator:
            children.append(child)
            child = AXObject.get_next_sibling(child)

        return children

    def getMathEnclosures(self, obj):
        if not self.isMathEnclose(obj):
            return []

        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('notation', 'longdiv').split()

    def getMathFencedSeparators(self, obj):
        if not self.isMathFenced(obj):
            return ['']

        attrs = AXObject.get_attributes_dict(obj)
        return list(attrs.get('separators', ','))

    def getMathFences(self, obj):
        if not self.isMathFenced(obj):
            return ['', '']

        attrs = AXObject.get_attributes_dict(obj)
        return [attrs.get('open', '('), attrs.get('close', ')')]

    def getMathNestingLevel(self, obj, test=None):
        rv = self._mathNestingLevel.get(hash(obj))
        if rv is not None:
            return rv

        def pred(x):
            if test is not None:
                return test(x)
            return self._getTag(x) == self._getTag(obj)

        rv = -1
        ancestor = obj
        while ancestor:
            ancestor = AXObject.find_ancestor(ancestor, pred)
            rv += 1

        self._mathNestingLevel[hash(obj)] = rv
        return rv

    def filterContentsForPresentation(self, contents, inferLabels=False):
        def _include(x):
            obj, start, end, string = x
            if not obj or AXObject.is_dead(obj):
                return False

            rv = self._shouldFilter.get(hash(obj))
            if rv is not None:
                return rv

            displayedText = string or AXObject.get_name(obj)
            rv = True
            if ((self.isTextBlockElement(obj) or self.isLink(obj)) and not displayedText) \
               or (self.isContentEditableWithEmbeddedObjects(obj) and not string.strip()) \
               or self.isEmptyAnchor(obj) \
               or (AXComponent.has_no_size(obj) and not displayedText) \
               or self.isHidden(obj) \
               or self.isOffScreenLabel(obj) \
               or self.isUselessImage(obj) \
               or self.isErrorForContents(obj, contents) \
               or self.isLabellingContents(obj, contents):
                rv = False
            elif AXUtilities.is_table_row(obj):
                rv = self.hasExplicitName(obj)
            else:
                widget = self.isInferredLabelForContents(x, contents)
                alwaysFilter = [Atspi.Role.RADIO_BUTTON, Atspi.Role.CHECK_BOX]
                if widget and (inferLabels or AXObject.get_role(widget) in alwaysFilter):
                    rv = False

            self._shouldFilter[hash(obj)] = rv
            return rv

        if len(contents) == 1:
            return contents

        rv = list(filter(_include, contents))
        self._shouldFilter = {}
        return rv

    def needsSeparator(self, lastChar, nextChar):
        if lastChar.isspace() or nextChar.isspace():
            return False

        openingPunctuation = ["(", "[", "{", "<"]
        closingPunctuation = [".", "?", "!", ":", ",", ";", ")", "]", "}", ">"]
        if lastChar in closingPunctuation or nextChar in openingPunctuation:
            return True
        if lastChar in openingPunctuation or nextChar in closingPunctuation:
            return False

        return lastChar.isalnum()

    def supportsSelectionAndTable(self, obj):
        return AXObject.supports_table(obj) and AXObject.supports_selection(obj)

    def hasGridDescendant(self, obj):
        if not obj:
            return False

        rv = self._hasGridDescendant.get(hash(obj))
        if rv is not None:
            return rv

        if not AXObject.get_child_count(obj):
            rv = False
        else:
            document = self.documentFrame(obj)
            if obj != document:
                document_has_grids = self.hasGridDescendant(document)
                if not document_has_grids:
                    rv = False

        if rv is None:
            grids = AXUtilities.find_all_grids(obj)
            rv = bool(grids)

        self._hasGridDescendant[hash(obj)] = rv
        return rv

    def isGridDescendant(self, obj):
        if not obj:
            return False

        rv = self._isGridDescendant.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.find_ancestor(obj, self.supportsSelectionAndTable) is not None
        self._isGridDescendant[hash(obj)] = rv
        return rv

    def isSorted(self, obj):
        attrs = AXObject.get_attributes_dict(obj, False)
        return attrs.get("sort") not in ("none", None)

    def isAscending(self, obj):
        attrs = AXObject.get_attributes_dict(obj, False)
        return attrs.get("sort") == "ascending"

    def isDescending(self, obj):
        attrs = AXObject.get_attributes_dict(obj, False)
        return attrs.get("sort") == "descending"

    def isCellWithNameFromHeader(self, obj):
        if not AXUtilities.is_table_cell(obj):
            return False

        name = AXObject.get_name(obj)
        if not name:
            return False

        headers = AXTable.get_column_headers(obj)
        for header in headers:
            if AXObject.get_name(header) == name:
                return True

        headers = AXTable.get_row_headers(obj)
        for header in headers:
            if AXObject.get_name(header) == name:
                return True

        return False

    def setSizeUnknown(self, obj):
        if super().setSizeUnknown(obj):
            return True

        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('setsize') == '-1'

    def rowOrColumnCountUnknown(self, obj):
        if super().rowOrColumnCountUnknown(obj):
            return True

        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('rowcount') == '-1' or attrs.get('colcount') == '-1'

    def shouldReadFullRow(self, obj, prevObj=None):
        if not (obj and self.inDocumentContent(obj)):
            return super().shouldReadFullRow(obj, prevObj)

        if not super().shouldReadFullRow(obj, prevObj):
            return False

        if self.isGridDescendant(obj):
            return not self._script.inFocusMode()

        if self.lastInputEventWasLineNav():
            return False

        if self.lastInputEventWasMouseButton():
            return False

        return True

    def isEntryDescendant(self, obj):
        if not obj:
            return False

        rv = self._isEntryDescendant.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.find_ancestor(obj, AXUtilities.is_entry) is not None
        self._isEntryDescendant[hash(obj)] = rv
        return rv

    def isLabelDescendant(self, obj):
        if not obj:
            return False

        rv = self._isLabelDescendant.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.find_ancestor(obj, AXUtilities.is_label_or_caption) is not None
        self._isLabelDescendant[hash(obj)] = rv
        return rv

    def isMenuInCollapsedSelectElement(self, obj):
        return False

    def isMenuDescendant(self, obj):
        if not obj:
            return False

        rv = self._isMenuDescendant.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.find_ancestor(obj, AXUtilities.is_menu) is not None
        self._isMenuDescendant[hash(obj)] = rv
        return rv

    def isModalDialogDescendant(self, obj):
        if not obj:
            return False

        rv = self._isModalDialogDescendant.get(hash(obj))
        if rv is not None:
            return rv

        rv = super().isModalDialogDescendant(obj)
        self._isModalDialogDescendant[hash(obj)] = rv
        return rv

    def isNavigableToolTipDescendant(self, obj):
        if not obj:
            return False

        rv = self._isNavigableToolTipDescendant.get(hash(obj))
        if rv is not None:
            return rv

        if AXUtilities.is_tool_tip(obj):
            ancestor = obj
        else:
            ancestor = AXObject.find_ancestor(obj, AXUtilities.is_tool_tip)
        rv = ancestor and not self.isNonNavigablePopup(ancestor)
        self._isNavigableToolTipDescendant[hash(obj)] = rv
        return rv

    def isTime(self, obj):
        return 'time' in self._getXMLRoles(obj) or 'time' == self._getTag(obj)

    def isToolBarDescendant(self, obj):
        if not obj:
            return False

        rv = self._isToolBarDescendant.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.find_ancestor(obj, AXUtilities.is_tool_bar) is not None
        self._isToolBarDescendant[hash(obj)] = rv
        return rv

    def isWebAppDescendant(self, obj):
        if not obj:
            return False

        rv = self._isWebAppDescendant.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.find_ancestor(obj, AXUtilities.is_embedded) is not None
        self._isWebAppDescendant[hash(obj)] = rv
        return rv

    def isLayoutOnly(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isLayoutOnly(obj)

        rv = self._isLayoutOnly.get(hash(obj))
        if rv is not None:
            if rv:
                tokens = ["WEB:", obj, "is deemed to be layout only"]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return rv

        if AXUtilities.is_list(obj):
            rv = self.treatAsDiv(obj)
        elif self.isDescriptionList(obj):
            rv = False
        elif self.isDescriptionListTerm(obj):
            rv = False
        elif self.isDescriptionListDescription(obj):
            rv = False
        elif self.isMath(obj):
            rv = False
        elif self.isLandmark(obj):
            rv = False
        elif self.isContentDeletion(obj):
            rv = False
        elif self.isContentInsertion(obj):
            rv = False
        elif self.isContentMarked(obj):
            rv = False
        elif self.isContentSuggestion(obj):
            rv = False
        elif self.isDPub(obj):
            rv = False
        elif self.isFeed(obj):
            rv = False
        elif self.isFigure(obj):
            rv = False
        elif self.isGrid(obj):
            rv = False
        elif self.isInlineIframe(obj):
            rv = not self.hasExplicitName(obj)
        elif AXUtilities.is_table_header(obj):
            rv = False
        elif AXUtilities.is_separator(obj):
            rv = False
        elif AXUtilities.is_panel(obj):
            rv = not self.hasExplicitName(obj)
        elif AXUtilities.is_table_row(obj) and not AXUtilities.is_expandable(obj):
            rv = not self.hasExplicitName(obj)
        elif self.isCustomImage(obj):
            rv = False
        else:
            rv = super().isLayoutOnly(obj)

        if rv:
            tokens = ["WEB:", obj, "is deemed to be layout only"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)

        self._isLayoutOnly[hash(obj)] = rv
        return rv

    def elementIsPreformattedText(self, obj):
        if self._getTag(obj) in ["pre", "code"]:
            return True

        if "code" in self._getXMLRoles(obj):
            return True

        return False

    def elementLinesAreSingleWords(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        if self.elementIsPreformattedText(obj):
            return False

        rv = self._elementLinesAreSingleWords.get(hash(obj))
        if rv is not None:
            return rv

        nChars = AXText.get_character_count(obj)
        if not nChars:
            return False

        if not self.treatAsTextObject(obj):
            return False

        # If we have a series of embedded object characters, there's a reasonable chance
        # they'll look like the one-word-per-line CSSified text we're trying to detect.
        # We don't want that false positive. By the same token, the one-word-per-line
        # CSSified text we're trying to detect can have embedded object characters. So
        # if we have more than 30% EOCs, don't use this workaround. (The 30% is based on
        # testing with problematic text.)
        string = AXText.get_all_text(obj)
        eocs = re.findall("\ufffc", string)
        if len(eocs)/nChars > 0.3:
            return False

        # TODO - JD: Can we remove this?
        AXObject.clear_cache(obj, False, "Checking if element lines are single words.")
        tokens = list(filter(lambda x: x, re.split(r"[\s\ufffc]", string)))

        # Note: We cannot check for the editable-text interface, because Gecko
        # seems to be exposing that for non-editable things. Thanks Gecko.
        rv = not AXUtilities.is_editable(obj) and len(tokens) > 1
        if rv:
            i = 0
            while i < nChars:
                string, start, end = AXText.get_line_at_offset(obj, i)
                if len(string.split()) != 1:
                    rv = False
                    break
                i = max(i+1, end)

        self._elementLinesAreSingleWords[hash(obj)] = rv
        return rv

    def elementLinesAreSingleChars(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._elementLinesAreSingleChars.get(hash(obj))
        if rv is not None:
            return rv

        nChars = AXText.get_character_count(obj)
        if not nChars:
            return False

        if not self.treatAsTextObject(obj):
            return False

        # If we have a series of embedded object characters, there's a reasonable chance
        # they'll look like the one-char-per-line CSSified text we're trying to detect.
        # We don't want that false positive. By the same token, the one-char-per-line
        # CSSified text we're trying to detect can have embedded object characters. So
        # if we have more than 30% EOCs, don't use this workaround. (The 30% is based on
        # testing with problematic text.)
        string = AXText.get_all_text(obj)
        eocs = re.findall("\ufffc", string)
        if len(eocs)/nChars > 0.3:
            return False

        # TODO - JD: Can we remove this?
        AXObject.clear_cache(obj, False, "Checking if element lines are single chars.")

        # Note: We cannot check for the editable-text interface, because Gecko
        # seems to be exposing that for non-editable things. Thanks Gecko.
        rv = not AXUtilities.is_editable(obj)
        if rv:
            for i in range(nChars):
                char = AXText.get_character_at_offset(obj, i)[0]
                if char.isspace() or char in ["\ufffc", "\ufffd"]:
                    continue

                string = AXText.get_line_at_offset(obj, i)[0]
                if len(string.strip()) > 1:
                    rv = False
                    break

        self._elementLinesAreSingleChars[hash(obj)] = rv
        return rv

    def labelIsAncestorOfLabelled(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._labelIsAncestorOfLabelled.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        for target in self.targetsForLabel(obj):
            if AXObject.find_ancestor(target, lambda x: x == obj):
                rv = True
                break

        self._labelIsAncestorOfLabelled[hash(obj)] = rv
        return rv

    def isOffScreenLabel(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isOffScreenLabel.get(hash(obj))
        if rv is not None:
            return rv

        if self.labelIsAncestorOfLabelled(obj):
            return False

        rv = False
        targets = self.labelTargets(obj)
        if targets:
            end = max(1, AXText.get_character_count(obj))
            rect = AXText.get_range_rect(obj, 0, end)
            if rect.x < 0 or rect.y < 0:
                rv = True

        self._isOffScreenLabel[hash(obj)] = rv
        return rv

    def isDetachedDocument(self, obj):
        if AXUtilities.is_document(obj) and not AXObject.is_valid(AXObject.get_parent(obj)):
            tokens = ["WEB:", obj, "is a detached document"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        return False

    def iframeForDetachedDocument(self, obj, root=None):
        root = root or self.documentFrame()
        for iframe in AXUtilities.find_all_internal_frames(root):
            if AXObject.get_parent(obj) == iframe:
                tokens = ["WEB: Returning", iframe, "as iframe parent of detached", obj]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                return iframe

        return None

    def targetsForLabel(self, obj):
        return AXObject.get_relation_targets(obj, Atspi.RelationType.LABEL_FOR)

    def labelTargets(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return []

        rv = self._labelTargets.get(hash(obj))
        if rv is not None:
            return rv

        rv = [hash(t) for t in self.targetsForLabel(obj)]
        self._labelTargets[hash(obj)] = rv
        return rv

    def isLinkAncestorOfImageInContents(self, link, contents):
        if not self.isLink(link):
            return False

        for obj, start, end, string in contents:
            if not AXUtilities.is_image(obj):
                continue
            if AXObject.find_ancestor(obj, lambda x: x == link):
                return True

        return False

    def isInferredLabelForContents(self, content, contents):
        obj, start, end, string = content
        objs = list(filter(self.shouldInferLabelFor, [x[0] for x in contents]))
        if not objs:
            return None

        for o in objs:
            label, sources = self.inferLabelFor(o)
            if obj in sources and label.strip() == string.strip():
                return o

        return None

    def isLabellingInteractiveElement(self, obj):
        if self._labelTargets.get(hash(obj)) == []:
            return False

        targets = self.targetsForLabel(obj)
        for target in targets:
            if AXUtilities.is_focusable(target):
                return True

        return False

    def isLabellingContents(self, obj, contents=[]):
        if self.isFocusModeWidget(obj):
            return False

        targets = self.labelTargets(obj)
        if not contents:
            return bool(targets) or self.isLabelDescendant(obj)

        for acc, start, end, string in contents:
            if hash(acc) in targets:
                return True

        if not self.isTextBlockElement(obj):
            return False

        if not self.isLabelDescendant(obj):
            return False

        for acc, start, end, string in contents:
            if not self.isLabelDescendant(acc) or self.isTextBlockElement(acc):
                continue

            if AXUtilities.is_label_or_caption(self.commonAncestor(acc, obj)):
                return True

        return False

    def isAnchor(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isAnchor.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        if AXUtilities.is_link(obj) \
           and not AXUtilities.is_focusable(obj) \
           and not AXObject.has_action(obj, "jump") \
           and not self._getXMLRoles(obj):
            rv = True

        self._isAnchor[hash(obj)] = rv
        return rv

    def isEmptyAnchor(self, obj):
        return self.isAnchor(obj) and not self.treatAsTextObject(obj)

    def isEmptyToolTip(self, obj):
        return AXUtilities.is_tool_tip(obj) and not self.treatAsTextObject(obj)

    def isBrowserUIAlert(self, obj):
        if not AXUtilities.is_alert(obj):
            return False

        if self.inDocumentContent(obj):
            return False

        return True

    def isTopLevelBrowserUIAlert(self, obj):
        if not self.isBrowserUIAlert(obj):
            return False

        parent = AXObject.get_parent(obj)
        while parent and self.isLayoutOnly(parent):
            parent = AXObject.get_parent(parent)

        return AXUtilities.is_frame(parent)

    def isClickableElement(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isClickableElement.get(hash(obj))
        if rv is not None:
            return rv

        if self.labelIsAncestorOfLabelled(obj):
            return False

        if self.hasGridDescendant(obj):
            tokens = ["WEB:", obj, "is not clickable: has grid descendant"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return False

        rv = False
        if not self.isFocusModeWidget(obj):
            if not AXUtilities.is_focusable(obj):
                rv = AXObject.has_action(obj, "click")
            else:
                rv = AXObject.has_action(obj, "click-ancestor")

        if rv and not AXObject.get_name(obj) and AXObject.supports_text(obj):
            text = AXText.get_all_text(obj)
            if not text.replace("\ufffc", ""):
                tokens = ["WEB:", obj, "is not clickable: its text is just EOCs"]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                rv = False
            elif not text.strip():
                rv = not (AXUtilities.is_static(obj) or AXUtilities.is_link(obj))

        self._isClickableElement[hash(obj)] = rv
        return rv

    def isCodeDescendant(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isCodeDescendant(obj)

        rv = self._isCodeDescendant.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.find_ancestor(obj, self.isCode) is not None
        self._isCodeDescendant[hash(obj)] = rv
        return rv

    def isCode(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isCode(obj)

        return self._getTag(obj) == "code" or "code" in self._getXMLRoles(obj)

    def isDescriptionList(self, obj):
        if super().isDescriptionList(obj):
            return True

        return self._getTag(obj) == "dl"

    def isDescriptionListTerm(self, obj):
        if super().isDescriptionListTerm(obj):
            return True

        return self._getTag(obj) == "dt"

    def isDescriptionListDescription(self, obj):
        if super().isDescriptionListDescription(obj):
            return True

        return self._getTag(obj) == "dd"

    def descriptionListTerms(self, obj):
        if not obj:
            return []

        rv = self._descriptionListTerms.get(hash(obj))
        if rv is not None:
            return rv

        rv = super().descriptionListTerms(obj)
        if not self.inDocumentContent(obj):
            return rv

        self._descriptionListTerms[hash(obj)] = rv
        return rv

    def valuesForTerm(self, obj):
        if not obj:
            return []

        rv = self._valuesForTerm.get(hash(obj))
        if rv is not None:
            return rv

        rv = super().valuesForTerm(obj)
        if not self.inDocumentContent(obj):
            return rv

        self._valuesForTerm[hash(obj)] = rv
        return rv

    def getComboBoxValue(self, obj):
        attrs = AXObject.get_attributes_dict(obj, False)
        return attrs.get("valuetext", super().getComboBoxValue(obj))

    def isEditableComboBox(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isEditableComboBox(obj)

        rv = self._isEditableComboBox.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        if AXUtilities.is_combo_box(obj):
            rv = AXUtilities.is_editable(obj)

        self._isEditableComboBox[hash(obj)] = rv
        return rv

    def getEditableComboBoxForItem(self, item):
        if not AXUtilities.is_list_item(item):
            return None

        listbox = AXObject.find_ancestor(item, AXUtilities.is_list_box)
        if listbox is None:
            return None

        targets = AXObject.get_relation_targets(listbox,
                                                Atspi.RelationType.CONTROLLED_BY,
                                                self.isEditableComboBox)
        if len(targets) == 1:
            return targets[0]

        return AXObject.find_ancestor(listbox, self.isEditableComboBox)

    def isItemForEditableComboBox(self, item, comboBox):
        if not AXUtilities.is_list_item(item):
            return False
        if not self.isEditableComboBox(comboBox):
            return False

        rv = self.getEditableComboBoxForItem(item) == comboBox
        tokens = ["WEB:", item, "is item of", comboBox, ":", rv]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        return rv

    def isDPub(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        roles = self._getXMLRoles(obj)
        rv = bool(list(filter(lambda x: x.startswith("doc-"), roles)))
        return rv

    def isDPubAbstract(self, obj):
        return 'doc-abstract' in self._getXMLRoles(obj)

    def isDPubAcknowledgments(self, obj):
        return 'doc-acknowledgments' in self._getXMLRoles(obj)

    def isDPubAfterword(self, obj):
        return 'doc-afterword' in self._getXMLRoles(obj)

    def isDPubAppendix(self, obj):
        return 'doc-appendix' in self._getXMLRoles(obj)

    def isDPubBacklink(self, obj):
        return 'doc-backlink' in self._getXMLRoles(obj)

    def isDPubBiblioref(self, obj):
        return 'doc-biblioref' in self._getXMLRoles(obj)

    def isDPubBibliography(self, obj):
        return 'doc-bibliography' in self._getXMLRoles(obj)

    def isDPubChapter(self, obj):
        return 'doc-chapter' in self._getXMLRoles(obj)

    def isDPubColophon(self, obj):
        return 'doc-colophon' in self._getXMLRoles(obj)

    def isDPubConclusion(self, obj):
        return 'doc-conclusion' in self._getXMLRoles(obj)

    def isDPubCover(self, obj):
        return 'doc-cover' in self._getXMLRoles(obj)

    def isDPubCredit(self, obj):
        return 'doc-credit' in self._getXMLRoles(obj)

    def isDPubCredits(self, obj):
        return 'doc-credits' in self._getXMLRoles(obj)

    def isDPubDedication(self, obj):
        return 'doc-dedication' in self._getXMLRoles(obj)

    def isDPubEndnote(self, obj):
        return 'doc-endnote' in self._getXMLRoles(obj)

    def isDPubEndnotes(self, obj):
        return 'doc-endnotes' in self._getXMLRoles(obj)

    def isDPubEpigraph(self, obj):
        return 'doc-epigraph' in self._getXMLRoles(obj)

    def isDPubEpilogue(self, obj):
        return 'doc-epilogue' in self._getXMLRoles(obj)

    def isDPubErrata(self, obj):
        return 'doc-errata' in self._getXMLRoles(obj)

    def isDPubExample(self, obj):
        return 'doc-example' in self._getXMLRoles(obj)

    def isDPubFootnote(self, obj):
        return 'doc-footnote' in self._getXMLRoles(obj)

    def isDPubForeword(self, obj):
        return 'doc-foreword' in self._getXMLRoles(obj)

    def isDPubGlossary(self, obj):
        return 'doc-glossary' in self._getXMLRoles(obj)

    def isDPubGlossref(self, obj):
        return 'doc-glossref' in self._getXMLRoles(obj)

    def isDPubIndex(self, obj):
        return 'doc-index' in self._getXMLRoles(obj)

    def isDPubIntroduction(self, obj):
        return 'doc-introduction' in self._getXMLRoles(obj)

    def isDPubNoteref(self, obj):
        return 'doc-noteref' in self._getXMLRoles(obj)

    def isDPubPagelist(self, obj):
        return 'doc-pagelist' in self._getXMLRoles(obj)

    def isDPubPagebreak(self, obj):
        return 'doc-pagebreak' in self._getXMLRoles(obj)

    def isDPubPart(self, obj):
        return 'doc-part' in self._getXMLRoles(obj)

    def isDPubPreface(self, obj):
        return 'doc-preface' in self._getXMLRoles(obj)

    def isDPubPrologue(self, obj):
        return 'doc-prologue' in self._getXMLRoles(obj)

    def isDPubPullquote(self, obj):
        return 'doc-pullquote' in self._getXMLRoles(obj)

    def isDPubQna(self, obj):
        return 'doc-qna' in self._getXMLRoles(obj)

    def isDPubSubtitle(self, obj):
        return 'doc-subtitle' in self._getXMLRoles(obj)

    def isDPubToc(self, obj):
        return 'doc-toc' in self._getXMLRoles(obj)

    def isErrorMessage(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isErrorMessage(obj)

        rv = self._isErrorMessage.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.has_relation(obj, Atspi.RelationType.ERROR_FOR)
        self._isErrorMessage[hash(obj)] = rv
        return rv

    def isFakePlaceholderForEntry(self, obj):
        if not (obj and self.inDocumentContent(obj) and AXObject.get_parent(obj)):
            return False

        if AXUtilities.is_editable(obj):
            return False

        entryName = AXObject.get_name(AXObject.find_ancestor(obj, AXUtilities.is_entry))
        if not entryName:
            return False

        def _isMatch(x):
            string = AXText.get_all_text(x).strip()
            if entryName != string:
                return False
            return AXUtilities.is_section(x) or AXUtilities.is_static(x)

        if _isMatch(obj):
            return True

        return AXObject.find_descendant(obj, _isMatch) is not None

    def isGrid(self, obj):
        return 'grid' in self._getXMLRoles(obj)

    def isGridCell(self, obj):
        return 'gridcell' in self._getXMLRoles(obj)

    def isInlineListItem(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isInlineListItem.get(hash(obj))
        if rv is not None:
            return rv

        if not AXUtilities.is_list_item(obj):
            rv = False
        else:
            displayStyle = self._getDisplayStyle(obj)
            rv = displayStyle and "inline" in displayStyle

        self._isInlineListItem[hash(obj)] = rv
        return rv

    def isBlockListDescendant(self, obj):
        if not self.isListDescendant(obj):
            return False

        return not self.isInlineListDescendant(obj)

    def isListDescendant(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isListDescendant.get(hash(obj))
        if rv is not None:
            return rv

        ancestor = AXObject.find_ancestor(obj, AXUtilities.is_list)
        rv = ancestor is not None
        self._isListDescendant[hash(obj)] = rv
        return rv

    def isInlineListDescendant(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isInlineListDescendant.get(hash(obj))
        if rv is not None:
            return rv

        if self.isInlineListItem(obj):
            rv = True
        else:
            ancestor = AXObject.find_ancestor(obj, self.isInlineListItem)
            rv = ancestor is not None

        self._isInlineListDescendant[hash(obj)] = rv
        return rv

    def listForInlineListDescendant(self, obj):
        if not self.isInlineListDescendant(obj):
            return None

        return AXObject.find_ancestor(obj, AXUtilities.is_list)

    def isFeed(self, obj):
        return 'feed' in self._getXMLRoles(obj)

    def isFeedArticle(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        if not AXUtilities.is_article(obj):
            return False

        return AXObject.find_ancestor(obj, self.isFeed) is not None

    def isFigure(self, obj):
        return 'figure' in self._getXMLRoles(obj) or self._getTag(obj) == 'figure'

    def isLandmark(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isLandmark.get(hash(obj))
        if rv is not None:
            return rv

        if AXUtilities.is_landmark(obj):
            rv = True
        elif self.isLandmarkRegion(obj):
            rv = bool(AXObject.get_name(obj))
        else:
            roles = self._getXMLRoles(obj)
            rv = bool(list(filter(lambda x: x in self.getLandmarkTypes(), roles)))

        self._isLandmark[hash(obj)] = rv
        return rv

    def isLandmarkWithoutType(self, obj):
        roles = self._getXMLRoles(obj)
        return not roles

    def isLandmarkBanner(self, obj):
        return 'banner' in self._getXMLRoles(obj)

    def isLandmarkComplementary(self, obj):
        return 'complementary' in self._getXMLRoles(obj)

    def isLandmarkContentInfo(self, obj):
        return 'contentinfo' in self._getXMLRoles(obj)

    def isLandmarkForm(self, obj):
        return 'form' in self._getXMLRoles(obj)

    def isLandmarkMain(self, obj):
        return 'main' in self._getXMLRoles(obj)

    def isLandmarkNavigation(self, obj):
        return 'navigation' in self._getXMLRoles(obj)

    def isLandmarkRegion(self, obj):
        return 'region' in self._getXMLRoles(obj)

    def isLandmarkSearch(self, obj):
        return 'search' in self._getXMLRoles(obj)

    def isLiveRegion(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        attrs = AXObject.get_attributes_dict(obj)
        return 'container-live' in attrs

    def isLink(self, obj):
        if not obj:
            return False

        rv = self._isLink.get(hash(obj))
        if rv is not None:
            return rv

        if AXUtilities.is_link(obj) and not self.isAnchor(obj):
            rv = True
        elif AXUtilities.is_static(obj) \
           and AXUtilities.is_link(AXObject.get_parent(obj)) \
           and AXObject.has_same_non_empty_name(obj, AXObject.get_parent(obj)):
            rv = True
        else:
            rv = False

        self._isLink[hash(obj)] = rv
        return rv

    def isNonNavigablePopup(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isNonNavigablePopup.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXUtilities.is_tool_tip(obj) \
            and not AXUtilities.is_focusable(obj)

        self._isNonNavigablePopup[hash(obj)] = rv
        return rv

    def hasUselessCanvasDescendant(self, obj):
        return len(AXUtilities.find_all_canvases(obj, self.isUselessImage)) > 0

    def isTextSubscriptOrSuperscript(self, obj):
        if self.isMath(obj):
            return False

        return AXUtilities.is_subscript_or_superscript(obj)

    def isSwitch(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isSwitch(obj)

        return 'switch' in self._getXMLRoles(obj)

    def isNonNavigableEmbeddedDocument(self, obj):
        rv = self._isNonNavigableEmbeddedDocument.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        if self.isDocument(obj) and self.getDocumentForObject(obj):
            try:
                name = AXObject.get_name(obj)
            except Exception:
                rv = True
            else:
                rv = "doubleclick" in name

        self._isNonNavigableEmbeddedDocument[hash(obj)] = rv
        return rv

    def isRedundantSVG(self, obj):
        if not self.isSVG(obj) or AXObject.get_child_count(AXObject.get_parent(obj)) == 1:
            return False

        rv = self._isRedundantSVG.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        parent = AXObject.get_parent(obj)
        children = [x for x in AXObject.iter_children(parent, self.isSVG)]
        if len(children) == AXObject.get_child_count(parent):
            sortedChildren = AXComponent.sort_objects_by_size(children)
            if obj != sortedChildren[-1]:
                objExtents = AXComponent.get_rect(obj)
                largestExtents = AXComponent.get_rect(sortedChildren[-1])
                intersection = AXComponent.get_rect_intersection(objExtents, largestExtents)
                rv = intersection == objExtents

        self._isRedundantSVG[hash(obj)] = rv
        return rv

    def isCustomImage(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isCustomImage.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        if self.isCustomElement(obj) and self.hasExplicitName(obj) \
           and AXUtilities.is_section(obj) \
           and AXObject.supports_text(obj) \
           and not re.search(r'[^\s\ufffc]', AXText.get_all_text(obj)):
            for child in AXObject.iter_children(obj):
                if not (AXUtilities.is_image_or_canvas(child) or self.isSVG(child)):
                    break
            else:
                rv = True

        self._isCustomImage[hash(obj)] = rv
        return rv

    def isUselessImage(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isUselessImage.get(hash(obj))
        if rv is not None:
            return rv

        rv = True
        if not (AXUtilities.is_image_or_canvas(obj) or self.isSVG(obj)):
            rv = False
        if rv and (AXObject.get_name(obj) \
                   or AXObject.get_description(obj) \
                   or self.hasLongDesc(obj)):
            rv = False
        if rv and (self.isClickableElement(obj) and not self.hasExplicitName(obj)):
            rv = False
        if rv and AXUtilities.is_focusable(obj):
            rv = False
        if rv and AXUtilities.is_link(AXObject.get_parent(obj)) and not self.hasExplicitName(obj):
            uri = AXHypertext.get_link_uri(AXObject.get_parent(obj))
            if uri and not uri.startswith('javascript'):
                rv = False
        if rv and AXObject.supports_image(obj):
            if AXObject.get_image_description(obj):
                rv = False
            elif not self.hasExplicitName(obj) and not self.isRedundantSVG(obj):
                width, height = AXObject.get_image_size(obj)
                if width > 25 and height > 25:
                    rv = False
        if rv and AXObject.supports_text(obj):
            rv = not self.treatAsTextObject(obj)
        if rv and AXObject.get_child_count(obj):
            for i in range(min(AXObject.get_child_count(obj), 50)):
                if not self.isUselessImage(AXObject.get_child(obj, i)):
                    rv = False
                    break

        self._isUselessImage[hash(obj)] = rv
        return rv

    def hasValidName(self, obj):
        name = AXObject.get_name(obj)
        if not name:
            return False

        if len(name.split()) > 1:
            return True

        parsed = urllib.parse.parse_qs(name)
        if len(parsed) > 2:
            tokens = ["WEB: name of", obj, "is suspected query string"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return False

        if len(name) == 1 and ord(name) in range(0xe000, 0xf8ff):
            tokens = ["WEB: name of", obj, "is in unicode private use area"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return False

        return True

    def isUselessEmptyElement(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isUselessEmptyElement.get(hash(obj))
        if rv is not None:
            return rv

        roles = [Atspi.Role.PARAGRAPH,
                 Atspi.Role.SECTION,
                 Atspi.Role.STATIC,
                 Atspi.Role.TABLE_ROW]
        role = AXObject.get_role(obj)
        if role not in roles and not self.isAriaAlert(obj):
            rv = False
        elif AXUtilities.is_focusable(obj):
            rv = False
        elif AXUtilities.is_editable(obj):
            rv = False
        elif self.hasValidName(obj) \
                or AXObject.get_description(obj) or AXObject.get_child_count(obj):
            rv = False
        elif AXText.get_character_count(obj) and AXText.get_all_text(obj) != AXObject.get_name(obj):
            rv = False
        elif AXObject.supports_action(obj):
            names = AXObject.get_action_names(obj)
            ignore = ["click-ancestor", "show-context-menu", "do-default"]
            names = list(filter(lambda x: x not in ignore, names))
            rv = not names
        else:
            rv = True

        self._isUselessEmptyElement[hash(obj)] = rv
        return rv

    def isParentOfNullChild(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isParentOfNullChild.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        childCount = AXObject.get_child_count(obj)
        if childCount and AXObject.get_child(obj, 0) is None:
            tokens = ["ERROR: ", obj, "reports", childCount,
                      "children, but AXObject.get_child(obj, 0) is None"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = True

        self._isParentOfNullChild[hash(obj)] = rv
        return rv

    def hasExplicitName(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('explicit-name') == 'true'

    def hasLongDesc(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._hasLongDesc.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.has_action(obj, "showlongdesc")
        self._hasLongDesc[hash(obj)] = rv
        return rv

    def hasVisibleCaption(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().hasVisibleCaption(obj)

        if not (self.isFigure(obj) or AXObject.supports_table(obj)):
            return False

        rv = self._hasVisibleCaption.get(hash(obj))
        if rv is not None:
            return rv

        labels = self.labelsForObject(obj)

        def isVisibleCaption(x):
            return AXUtilities.is_caption(x) \
                and AXUtilities.is_showing(x) and AXUtilities.is_visible(x)

        rv = bool(list(filter(isVisibleCaption, labels)))
        self._hasVisibleCaption[hash(obj)] = rv
        return rv

    def hasDetails(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().hasDetails(obj)

        rv = self._hasDetails.get(hash(obj))
        if rv is not None:
            return rv

        relation = AXObject.get_relation(obj, Atspi.RelationType.DETAILS)
        rv = relation and relation.get_n_targets() > 0
        self._hasDetails[hash(obj)] = rv
        return rv

    def detailsIn(self, obj):
        if not self.hasDetails(obj):
            return []

        return AXObject.get_relation_targets(obj, Atspi.RelationType.DETAILS)

    def isDetails(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().isDetails(obj)

        rv = self._isDetails.get(hash(obj))
        if rv is not None:
            return rv

        relation = AXObject.get_relation(obj, Atspi.RelationType.DETAILS_FOR)
        rv = relation and relation.get_n_targets() > 0
        self._isDetails[hash(obj)] = rv
        return rv

    def detailsFor(self, obj):
        if not self.isDetails(obj):
            return []

        return AXObject.get_relation_targets(obj, Atspi.RelationType.DETAILS_FOR)

    def popupType(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return 'false'

        attrs = AXObject.get_attributes_dict(obj)
        return attrs.get('haspopup', 'false').lower()

    def inferLabelFor(self, obj):
        if not self.shouldInferLabelFor(obj):
            return None, []

        rv = self._inferredLabels.get(hash(obj))
        if rv is not None:
            return rv

        rv = self._script.labelInference.infer(obj, False)
        self._inferredLabels[hash(obj)] = rv
        return rv

    def shouldInferLabelFor(self, obj):
        if not self.inDocumentContent() or self.isWebAppDescendant(obj):
            return False

        rv = self._shouldInferLabelFor.get(hash(obj))
        if rv and not self._script.caretNavigation.last_input_event_was_navigation_command():
            return not self._script.inSayAll()
        if rv is False:
            return rv

        role = AXObject.get_role(obj)
        name = AXObject.get_name(obj)
        if name:
            rv = False
        elif self._getXMLRoles(obj):
            rv = False
        elif not rv:
            roles = [Atspi.Role.CHECK_BOX,
                     Atspi.Role.COMBO_BOX,
                     Atspi.Role.ENTRY,
                     Atspi.Role.LIST_BOX,
                     Atspi.Role.PASSWORD_TEXT,
                     Atspi.Role.RADIO_BUTTON]
            rv = role in roles and not self.displayedLabel(obj)

        self._shouldInferLabelFor[hash(obj)] = rv

        if self._script.caretNavigation.last_input_event_was_navigation_command() \
           and role not in [Atspi.Role.RADIO_BUTTON, Atspi.Role.CHECK_BOX]:
            return False

        return rv

    def displayedLabel(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().displayedLabel(obj)

        rv = self._displayedLabelText.get(hash(obj))
        if rv is not None:
            return rv

        labels = self.labelsForObject(obj)
        strings = [AXObject.get_name(label)
                   or self.displayedText(label) for label in labels if label is not None]
        rv = " ".join(strings)

        self._displayedLabelText[hash(obj)] = rv
        return rv

    def labelsForObject(self, obj):
        if not obj:
            return []

        rv = self._labelsForObject.get(hash(obj))
        if rv is not None:
            return rv

        rv = super().labelsForObject(obj)
        if not self.inDocumentContent(obj):
            return rv

        self._labelsForObject[hash(obj)] = rv
        return rv

    def isSpinnerEntry(self, obj):
        if not self.inDocumentContent(obj):
            return False

        if not AXUtilities.is_editable(obj):
            return False

        if AXUtilities.is_spin_button(obj) or AXUtilities.is_spin_button(AXObject.get_parent(obj)):
            return True

        return False

    def eventIsSpinnerNoise(self, event):
        if not self.isSpinnerEntry(event.source):
            return False

        if event.type.startswith("object:text-changed") \
           or event.type.startswith("object:text-selection-changed"):
            lastKey, mods = self.lastKeyAndModifiers()
            if lastKey in ["Down", "Up"]:
                return True

        return False

    def treatEventAsSpinnerValueChange(self, event):
        if event.type.startswith("object:text-caret-moved") and self.isSpinnerEntry(event.source):
            lastKey, mods = self.lastKeyAndModifiers()
            if lastKey in ["Down", "Up"]:
                obj, offset = self.getCaretContext()
                return event.source == obj

        return False

    def eventIsBrowserUINoise(self, event):
        if self.inDocumentContent(event.source):
            return False

        if event.type.startswith("object:text-") \
           and self.isSingleLineAutocompleteEntry(event.source):
            lastKey, mods = self.lastKeyAndModifiers()
            return lastKey == "Return"
        if event.type.startswith("object:text-") or event.type.endswith("accessible-name"):
            return AXUtilities.is_status_bar(event.source) or AXUtilities.is_label(event.source) \
                or AXUtilities.is_frame(event.source)
        if event.type.startswith("object:children-changed"):
            return True

        return False

    def eventIsAutocompleteNoise(self, event, documentFrame=None):
        inContent = documentFrame or self.inDocumentContent(event.source)
        if not inContent:
            return False

        def isListBoxItem(x):
            return AXUtilities.is_list_box(AXObject.get_parent(x))

        def isMenuItem(x):
            return AXUtilities.is_menu(AXObject.get_parent(x))

        def isComboBoxItem(x):
            return AXUtilities.is_combo_box(AXObject.get_parent(x))

        if AXUtilities.is_editable(event.source) \
           and event.type.startswith("object:text-"):
            obj, offset = self.getCaretContext(documentFrame)
            if isListBoxItem(obj) or isMenuItem(obj):
                return True

            if obj == event.source and isComboBoxItem(obj):
                lastKey, mods = self.lastKeyAndModifiers()
                if lastKey in ["Down", "Up"]:
                    return True

        return False

    def eventIsBrowserUIAutocompleteNoise(self, event):
        if self.inDocumentContent(event.source):
            return False

        if self._eventIsBrowserUIAutocompleteTextNoise(event):
            return True

        return self._eventIsBrowserUIAutocompleteSelectionNoise(event)

    def _eventIsBrowserUIAutocompleteSelectionNoise(self, event):
        selection = ["object:selection-changed", "object:state-changed:selected"]
        if event.type not in selection:
            return False

        if not AXUtilities.is_menu_related(event.source):
            return False

        focus = focus_manager.getManager().get_locus_of_focus()
        if AXUtilities.is_entry(focus) and AXUtilities.is_focused(focus):
            lastKey, mods = self.lastKeyAndModifiers()
            if lastKey not in ["Down", "Up"]:
                return True

        return False

    def _eventIsBrowserUIAutocompleteTextNoise(self, event):
        if not event.type.startswith("object:text-") \
           or not self.isSingleLineAutocompleteEntry(event.source):
            return False

        focus = focus_manager.getManager().get_locus_of_focus()
        if not AXUtilities.is_selectable(focus):
            return False

        if AXUtilities.is_menu_item_of_any_kind(focus) \
           or AXUtilities.is_list_item(focus):
            lastKey, mods = self.lastKeyAndModifiers()
            return lastKey in ["Down", "Up"]

        return False

    def eventIsBrowserUIPageSwitch(self, event):
        selection = ["object:selection-changed", "object:state-changed:selected"]
        if event.type not in selection:
            return False

        if not AXUtilities.is_page_tab_list_related(event.source):
            return False

        if self.inDocumentContent(event.source):
            return False

        if not self.inDocumentContent(focus_manager.getManager().get_locus_of_focus()):
            return False

        return True

    def eventIsFromLocusOfFocusDocument(self, event):
        if focus_manager.getManager().focus_is_active_window():
            focus = self.activeDocument()
            source = self.getTopLevelDocumentForObject(event.source)
        else:
            focus = self.getDocumentForObject(focus_manager.getManager().get_locus_of_focus())
            source = self.getDocumentForObject(event.source)

        tokens = ["WEB: Event doc:", source, ". Focus doc:", focus, "."]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        if not (source and focus):
            return False

        if source == focus:
            return True

        if not AXObject.is_valid(focus) and AXObject.is_valid(source):
            if self.activeDocument() == source:
                msg = "WEB: Treating active doc as locusOfFocus doc"
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                return True

        return False

    def eventIsIrrelevantSelectionChangedEvent(self, event):
        if event.type != "object:selection-changed":
            return False

        focus = focus_manager.getManager().get_locus_of_focus()
        if not focus:
            msg = "WEB: Selection changed event is relevant (no locusOfFocus)"
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False
        if event.source == focus:
            msg = "WEB: Selection changed event is relevant (is locusOfFocus)"
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False
        if AXObject.find_ancestor(focus, lambda x: x == event.source):
            msg = "WEB: Selection changed event is relevant (ancestor of locusOfFocus)"
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False

        # There may be other roles where we need to do this. For now, solve the known one.
        if AXUtilities.is_page_tab_list(event.source):
            tokens = ["WEB: Selection changed event is irrelevant (unrelated",
                      AXObject.get_role_name(event.source), ")"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return True

        msg = "WEB: Selection changed event is relevant (no reason found to ignore it)"
        debug.printMessage(debug.LEVEL_INFO, msg, True)
        return False

    def textEventIsDueToDeletion(self, event):
        if not self.inDocumentContent(event.source) \
           or not AXUtilities.is_editable(event.source):
            return False

        if self.isDeleteCommandTextDeletionEvent(event) \
           or self.isBackSpaceCommandTextDeletionEvent(event):
            return True

        return False

    def textEventIsDueToInsertion(self, event):
        if not event.type.startswith("object:text-"):
            return False

        if not self.inDocumentContent(event.source) \
           or not AXUtilities.is_editable(event.source):
            return False

        if event.source != focus_manager.getManager().get_locus_of_focus():
            return False

        if isinstance(orca_state.lastInputEvent, input_event.KeyboardEvent):
            inputEvent = orca_state.lastNonModifierKeyEvent
            return inputEvent and inputEvent.isPrintableKey() and not inputEvent.modifiers

        return False

    def textEventIsForNonNavigableTextObject(self, event):
        if not event.type.startswith("object:text-"):
            return False

        return self._treatObjectAsWhole(event.source)

    def eventIsEOCAdded(self, event):
        if not self.inDocumentContent(event.source):
            return False

        if event.type.startswith("object:text-changed:insert") \
           and self.EMBEDDED_OBJECT_CHARACTER in event.any_data:
            return not re.match(r"[^\s\ufffc]", event.any_data)

        return False

    def caretMovedOutsideActiveGrid(self, event, oldFocus=None):
        if not (event and event.type.startswith("object:text-caret-moved")):
            return False

        oldFocus = oldFocus or focus_manager.getManager().get_locus_of_focus()
        if not self.isGridDescendant(oldFocus):
            return False

        return not self.isGridDescendant(event.source)

    def caretMovedToSamePageFragment(self, event, oldFocus=None):
        if not (event and event.type.startswith("object:text-caret-moved")):
            return False

        if AXUtilities.is_editable(event.source):
            return False

        fragment = AXDocument.get_document_uri_fragment(self.documentFrame())
        if not fragment:
            return False

        sourceID = self._getID(event.source)
        if sourceID and fragment == sourceID:
            return True

        oldFocus = oldFocus or focus_manager.getManager().get_locus_of_focus()
        if self.isLink(oldFocus):
            link = oldFocus
        else:
            link = AXObject.find_ancestor(oldFocus, self.isLink)

        return link and AXHypertext.get_link_uri(link) == AXDocument.get_uri(self.documentFrame())

    def isChildOfCurrentFragment(self, obj):
        fragment = AXDocument.get_document_uri_fragment(self.documentFrame(obj))
        if not fragment:
            return False

        def isSameFragment(x):
            return self._getID(x) == fragment

        return AXObject.find_ancestor(obj, isSameFragment) is not None

    def isContentEditableWithEmbeddedObjects(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return False

        rv = self._isContentEditableWithEmbeddedObjects.get(hash(obj))
        if rv is not None:
            return rv

        rv = False
        def hasTextBlockRole(x):
            return AXObject.get_role(x) in self._textBlockElementRoles() \
                and not self.isFakePlaceholderForEntry(x) and not self.isStaticTextLeaf(x)

        if self._getTag(obj) in ["input", "textarea"]:
            rv = False
        elif AXUtilities.is_multi_line_entry(obj):
            rv = AXObject.find_descendant(obj, hasTextBlockRole)
        elif AXUtilities.is_editable(obj):
            rv = hasTextBlockRole(obj) or self.isLink(obj)
        elif not self.isDocument(obj):
            document = self.getDocumentForObject(obj)
            rv = self.isContentEditableWithEmbeddedObjects(document)

        self._isContentEditableWithEmbeddedObjects[hash(obj)] = rv
        return rv

    def _rangeInParentWithLength(self, obj):
        parent = AXObject.get_parent(obj)
        if not self.treatAsTextObject(parent):
            return -1, -1, 0

        start = AXHypertext.get_link_start_offset(obj)
        end = AXHypertext.get_link_end_offset(obj)
        return start, end, AXText.get_character_count(parent)

    def getError(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return super().getError(obj)

        if not AXUtilities.is_invalid_entry(obj):
            return False

        try:
            self._currentTextAttrs.pop(hash(obj))
        except Exception:
            pass

        attrs, start, end = self.textAttributes(obj, 0, True)
        error = attrs.get("invalid")
        if error == "false":
            return False
        if error not in ["spelling", "grammar"]:
            return True

        return error

    def _getErrorMessageContainer(self, obj):
        if not (obj and self.inDocumentContent(obj)):
            return None

        if not self.getError(obj):
            return None

        relation = AXObject.get_relation(obj, Atspi.RelationType.ERROR_MESSAGE)
        if relation:
            return relation.get_target(0)

        return None

    def getErrorMessage(self, obj):
        return self.expandEOCs(self._getErrorMessageContainer(obj))

    def isErrorForContents(self, obj, contents=[]):
        if not self.isErrorMessage(obj):
            return False

        for acc, start, end, string in contents:
            if self._getErrorMessageContainer(acc) == obj:
                return True

        return False

    def _canHaveCaretContext(self, obj):
        rv = self._canHaveCaretContextDecision.get(hash(obj))
        if rv is not None:
            return rv

        if obj is None:
            return False
        if AXObject.is_dead(obj):
            msg = "WEB: Dead object cannot have caret context"
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False
        if not AXObject.is_valid(obj):
            tokens = ["WEB: Invalid object cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return False

        startTime = time.time()
        rv = None
        if AXUtilities.is_focusable(obj):
            tokens = ["WEB: Focusable object can have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = True
        elif AXUtilities.is_editable(obj):
            tokens = ["WEB: Editable object can have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = True
        elif AXUtilities.is_landmark(obj):
            tokens = ["WEB: Landmark can have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = True
        elif self.isStaticTextLeaf(obj):
            tokens = ["WEB: Static text leaf cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isUselessEmptyElement(obj):
            tokens = ["WEB: Useless empty element cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isOffScreenLabel(obj):
            tokens = ["WEB: Off-screen label cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isNonNavigablePopup(obj):
            tokens = ["WEB: Non-navigable popup cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isUselessImage(obj):
            tokens = ["WEB: Useless image cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isEmptyAnchor(obj):
            tokens = ["WEB: Empty anchor cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isEmptyToolTip(obj):
            tokens = ["WEB: Empty tool tip cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isParentOfNullChild(obj):
            tokens = ["WEB: Parent of null child cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isPseudoElement(obj):
            tokens = ["WEB: Pseudo element cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isFakePlaceholderForEntry(obj):
            tokens = ["WEB: Fake placeholder for entry cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isNonInteractiveDescendantOfControl(obj):
            tokens = ["WEB: Non interactive descendant of control cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif self.isHidden(obj):
            # We try to do this check only if needed because getting object attributes is
            # not as performant, and we cannot use the cached attribute because aria-hidden
            # can change frequently depending on the app.
            tokens = ["WEB: Hidden object cannot have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = False
        elif AXComponent.has_no_size(obj):
            tokens = ["WEB: Allowing sizeless object to have caret context", obj]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = True
        else:
            tokens = ["WEB: ", obj, f"can have caret context. ({time.time() - startTime:.4f}s)"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = True

        self._canHaveCaretContextDecision[hash(obj)] = rv
        msg = f"INFO: _canHaveCaretContext took {time.time() - startTime:.4f}s"
        debug.printMessage(debug.LEVEL_INFO, msg, True)
        return rv

    def isPseudoElement(self, obj):
        return False

    def searchForCaretContext(self, obj):
        tokens = ["WEB: Searching for caret context in", obj]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        container = obj
        contextObj, contextOffset = None, -1
        while obj:
            offset = AXText.get_caret_offset(obj)
            if offset < 0:
                obj = None
            else:
                contextObj, contextOffset = obj, offset
                child = AXHypertext.get_child_at_offset(obj, offset)
                if child:
                    obj = child
                else:
                    break

        if contextObj and not self.isHidden(contextObj):
            return self.findNextCaretInOrder(contextObj, max(-1, contextOffset - 1))

        if self.isDocument(container):
            return container, 0

        return None, -1

    def _getCaretContextViaLocusOfFocus(self):
        obj = focus_manager.getManager().get_locus_of_focus()
        msg = "WEB: Getting caret context via locusOfFocus"
        debug.printMessage(debug.LEVEL_INFO, msg, True)
        if not self.inDocumentContent(obj):
            return None, -1

        if not AXObject.supports_text(obj):
            return obj, 0

        return obj, AXText.get_caret_offset(obj)

    def getCaretContext(self, documentFrame=None, getReplicant=False, searchIfNeeded=True):
        tokens = ["WEB: Getting caret context for", documentFrame]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        if not AXObject.is_valid(documentFrame):
            documentFrame = self.documentFrame()
            tokens = ["WEB: Now getting caret context for", documentFrame]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)

        if not documentFrame:
            if not searchIfNeeded:
                msg = "WEB: Returning None, -1: No document and no search requested."
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                return None, -1

            obj, offset = self._getCaretContextViaLocusOfFocus()
            tokens = ["WEB: Returning", obj, ", ", offset, "(from locusOfFocus)"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return obj, offset

        context = self._caretContexts.get(hash(AXObject.get_parent(documentFrame)))
        if context is not None:
            tokens = ["WEB: Cached context of", documentFrame, "is", context[0], ", ", context[1]]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
        else:
            tokens = ["WEB: No cached context for", documentFrame, "."]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            obj, offset = None, -1

        if not context or not self.isTopLevelDocument(documentFrame):
            if not searchIfNeeded:
                msg = "WEB: Returning None, -1: No top-level document with context " \
                      "and no search requested."
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                return None, -1
            obj, offset = self.searchForCaretContext(documentFrame)
        elif not getReplicant:
            obj, offset = context
        elif not AXObject.is_valid(context[0]):
            msg = "WEB: Context is not valid. Searching for replicant."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            obj, offset = self.findContextReplicant()
            if obj:
                caretObj, caretOffset = self.searchForCaretContext(AXObject.get_parent(obj))
                if caretObj and AXObject.is_valid(caretObj):
                    obj, offset = caretObj, caretOffset
        else:
            obj, offset = context

        tokens = ["WEB: Result context of", documentFrame, "is", obj, ", ", offset, "."]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        self.setCaretContext(obj, offset, documentFrame)
        return obj, offset

    def getCaretContextPathRoleAndName(self, documentFrame=None):
        documentFrame = documentFrame or self.documentFrame()
        if not documentFrame:
            return [-1], None, None

        rv = self._contextPathsRolesAndNames.get(hash(AXObject.get_parent(documentFrame)))
        if not rv:
            return [-1], None, None

        return rv

    def clearCaretContext(self, documentFrame=None):
        self.clearContentCache()
        documentFrame = documentFrame or self.documentFrame()
        if not documentFrame:
            return

        parent = AXObject.get_parent(documentFrame)
        self._caretContexts.pop(hash(parent), None)
        self._priorContexts.pop(hash(parent), None)

    def handleEventFromContextReplicant(self, event, replicant):
        if AXObject.is_dead(replicant):
            msg = "WEB: Context replicant is dead."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False

        if not focus_manager.getManager().focus_is_dead():
            msg = "WEB: Not event from context replicant, locus of focus is not dead."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False

        path, role, name = self.getCaretContextPathRoleAndName()
        replicantPath = AXObject.get_path(replicant)
        if path != replicantPath:
            tokens = ["WEB: Not event from context replicant. Path", path,
                      " != replicant path", replicantPath]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return False

        replicantRole = AXObject.get_role(replicant)
        if role != replicantRole:
            tokens = ["WEB: Not event from context replicant. Role", role,
                      " != replicant role", replicantRole]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return False

        notify = AXObject.get_name(replicant) != name
        documentFrame = self.documentFrame()
        obj, offset = self._caretContexts.get(hash(AXObject.get_parent(documentFrame)))

        tokens = ["WEB: Is event from context replicant. Notify:", notify]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        focus_manager.getManager().set_locus_of_focus(event, replicant, notify)
        self.setCaretContext(replicant, offset, documentFrame)
        return True

    def _handleEventForRemovedSelectableChild(self, event):
        container = None
        if AXUtilities.is_list_box(event.source):
            container = event.source
        elif AXUtilities.is_tree(event.source):
            container = event.source
        else:
            container = AXObject.find_ancestor(event.source, AXUtilities.is_list_box) \
                or AXObject.find_ancestor(event.source, AXUtilities.is_tree)
        if container is None:
            msg = "WEB: Could not find listbox or tree to recover from removed child."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False

        tokens = ["WEB: Checking", container, "for focused child."]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        # TODO - JD: Can we remove this? If it's needed, should it be recursive?
        AXObject.clear_cache(container, False, "Handling event for removed selectable child.")
        item = AXUtilities.get_focused_object(container)
        if not (AXUtilities.is_list_item(item) or AXUtilities.is_tree_item):
            msg = "WEB: Could not find focused item to recover from removed child."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False

        names = self._script.pointOfReference.get('names', {})
        oldName = names.get(hash(focus_manager.getManager().get_locus_of_focus()))
        notify = AXObject.get_name(item) != oldName

        tokens = ["WEB: Recovered from removed child. New focus is: ", item, "0"]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        focus_manager.getManager().set_locus_of_focus(event, item, notify)
        self.setCaretContext(item, 0)
        return True

    def handleEventForRemovedChild(self, event):
        focus = focus_manager.getManager().get_locus_of_focus()
        if event.any_data == focus:
            msg = "WEB: Removed child is locus of focus."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
        elif AXObject.find_ancestor(focus, lambda x: x == event.any_data):
            msg = "WEB: Removed child is ancestor of locus of focus."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
        elif focus_manager.getManager().focus_is_dead() \
           and self.isSameObject(event.any_data, focus, True, True):
            msg = "WEB: Removed child appears to be replicant of locus of focus."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
        else:
            msg = "WEB: Removed child is not locus of focus nor ancestor of locus of focus."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False

        if event.detail1 == -1:
            msg = "WEB: Event detail1 is useless."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False

        if self._handleEventForRemovedSelectableChild(event):
            return True

        obj, offset = None, -1
        notify = True
        keyString, mods = self.lastKeyAndModifiers()
        childCount = AXObject.get_child_count(event.source)
        if keyString == "Up":
            if event.detail1 >= childCount:
                msg = "WEB: Last child removed. Getting new location from end of parent."
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                obj, offset = self.previousContext(event.source, -1)
            elif 0 <= event.detail1 - 1 < childCount:
                child = AXObject.get_child(event.source, event.detail1 - 1)
                tokens = ["WEB: Getting new location from end of previous child", child, "."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                obj, offset = self.previousContext(child, -1)
            else:
                prevObj = self.findPreviousObject(event.source)
                tokens = ["WEB: Getting new location from end of source's previous object",
                          prevObj, "."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                obj, offset = self.previousContext(prevObj, -1)

        elif keyString == "Down":
            if event.detail1 == 0:
                msg = "WEB: First child removed. Getting new location from start of parent."
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                obj, offset = self.nextContext(event.source, -1)
            elif 0 < event.detail1 < childCount:
                child = AXObject.get_child(event.source, event.detail1)
                tokens = ["WEB: Getting new location from start of child", event.detail1,
                          child, "."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                obj, offset = self.nextContext(child, -1)
            else:
                nextObj = self.findNextObject(event.source)
                tokens = ["WEB: Getting new location from start of source's next object",
                          nextObj, "."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                obj, offset = self.nextContext(nextObj, -1)

        else:
            notify = False
            # TODO - JD: Can we remove this? Even if it is needed, we now also clear the
            # cache in _handleEventForRemovedSelectableChild. Also, if it is needed, should
            # it be recursive?
            AXObject.clear_cache(event.source, False, "Handling event for removed child.")
            obj, offset = self.searchForCaretContext(event.source)
            if obj is None:
                obj = AXUtilities.get_focused_object(event.source)

            # Risk "chattiness" if the locusOfFocus is dead and the object we've found is
            # focused and has a different name than the last known focused object.
            if obj and focus_manager.getManager().focus_is_dead() and AXUtilities.is_focused(obj):
                names = self._script.pointOfReference.get('names', {})
                oldName = names.get(hash(focus_manager.getManager().get_locus_of_focus()))
                notify = AXObject.get_name(obj) != oldName

        if obj:
            msg = "WEB: Setting locusOfFocus and context to: %s, %i" % (obj, offset)
            focus_manager.getManager().set_locus_of_focus(event, obj, notify)
            self.setCaretContext(obj, offset)
            return True

        tokens = ["WEB: Unable to find context for child removed from", event.source]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        return False

    def findContextReplicant(self, documentFrame=None, matchRole=True, matchName=True):
        path, oldRole, oldName = self.getCaretContextPathRoleAndName(documentFrame)
        obj = self.getObjectFromPath(path)
        if obj and matchRole:
            if AXObject.get_role(obj) != oldRole:
                obj = None
        if obj and matchName:
            if AXObject.get_name(obj) != oldName:
                obj = None
        if not obj:
            return None, -1

        obj, offset = self.findFirstCaretContext(obj, 0)
        tokens = ["WEB: Context replicant is", obj, ", ", offset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        return obj, offset

    def getPriorContext(self, documentFrame=None):
        if not AXObject.is_valid(documentFrame):
            documentFrame = self.documentFrame()

        if documentFrame:
            context = self._priorContexts.get(hash(AXObject.get_parent(documentFrame)))
            if context:
                return context

        return None, -1

    def _getPath(self, obj):
        rv = self._paths.get(hash(obj))
        if rv is not None:
            return rv

        rv = AXObject.get_path(obj) or [-1]
        self._paths[hash(obj)] = rv
        return rv

    def setCaretContext(self, obj=None, offset=-1, documentFrame=None):
        documentFrame = documentFrame or self.documentFrame()
        if not documentFrame:
            return

        parent = AXObject.get_parent(documentFrame)
        oldObj, oldOffset = self._caretContexts.get(hash(parent), (obj, offset))
        self._priorContexts[hash(parent)] = oldObj, oldOffset
        self._caretContexts[hash(parent)] = obj, offset

        path = self._getPath(obj)
        role = AXObject.get_role(obj)
        name = AXObject.get_name(obj)
        self._contextPathsRolesAndNames[hash(parent)] = path, role, name

    def findFirstCaretContext(self, obj, offset):
        self._canHaveCaretContextDecision = {}
        rv = self._findFirstCaretContext(obj, offset)
        self._canHaveCaretContextDecision = {}
        return rv

    def _findFirstCaretContext(self, obj, offset):
        tokens = ["WEB: Looking for first caret context for", obj, ", ", offset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)

        role = AXObject.get_role(obj)
        lookInChild = [Atspi.Role.LIST,
                       Atspi.Role.INTERNAL_FRAME,
                       Atspi.Role.TABLE,
                       Atspi.Role.TABLE_ROW]
        if role in lookInChild \
           and AXObject.get_child_count(obj) and not self.treatAsDiv(obj, offset):
            firstChild = AXObject.get_child(obj, 0)
            tokens = ["WEB: Will look in child", firstChild, "for first caret context"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return self._findFirstCaretContext(firstChild, 0)

        treatAsText = self.treatAsTextObject(obj)
        if not treatAsText and self._canHaveCaretContext(obj):
            tokens = ["WEB: First caret context for non-text context is", obj, "0"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return obj, 0

        length = AXText.get_character_count(obj)
        if treatAsText and offset >= length:
            if self.isContentEditableWithEmbeddedObjects(obj) and self.lastInputEventWasCharNav():
                nextObj, nextOffset = self.nextContext(obj, length)
                if not nextObj:
                    tokens = ["WEB: No next object found at end of contenteditable", obj]
                    debug.printTokens(debug.LEVEL_INFO, tokens, True)
                elif not self.isContentEditableWithEmbeddedObjects(nextObj):
                    tokens = ["WEB: Next object", nextObj,
                              "found at end of contenteditable", obj, "is not editable"]
                    debug.printTokens(debug.LEVEL_INFO, tokens, True)
                else:
                    tokens = ["WEB: First caret context at end of contenteditable", obj,
                              "is next context", nextObj, ", ", nextOffset]
                    debug.printTokens(debug.LEVEL_INFO, tokens, True)
                    return nextObj, nextOffset

            tokens = ["WEB: First caret context at end of", obj, ", ", offset, "is",
                      obj, ", ", length]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return obj, length

        offset = max(0, offset)
        if treatAsText:
            allText = AXText.get_all_text(obj)
            if allText[offset] != self.EMBEDDED_OBJECT_CHARACTER or role == Atspi.Role.ENTRY:
                msg = "WEB: First caret context is unchanged"
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                return obj, offset

            # Descending an element that we're treating as whole can lead to looping/getting stuck.
            if self.elementLinesAreSingleChars(obj):
                msg = "WEB: EOC in single-char-lines element. Returning context unchanged."
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                return obj, offset

        child = AXHypertext.get_child_at_offset(obj, offset)
        if not child:
            msg = "WEB: Child at offset is null. Returning context unchanged."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return obj, offset

        if self.isDocument(obj):
            while self.isUselessEmptyElement(child):
                tokens = ["WEB: Child", child, "of", obj, "at offset", offset, "cannot be context."]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                offset += 1
                child = AXHypertext.get_child_at_offset(obj, offset)

        if self.isListItemMarker(child):
            tokens = ["WEB: First caret context is next offset in", obj, ":",
                      offset + 1, "(skipping list item marker child)"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return obj, offset + 1

        if self.isEmptyAnchor(child):
            nextObj, nextOffset = self.nextContext(obj, offset)
            if nextObj:
                tokens = ["WEB: First caret context at end of empty anchor", obj,
                          "is next context", nextObj, ", ", nextOffset]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                return nextObj, nextOffset

        if not self._canHaveCaretContext(child):
            tokens = ["WEB: Child", child, "cannot be context. Returning", obj, ", ", offset]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            return obj, offset

        tokens = ["WEB: Looking in child", child, "for first caret context for", obj, ", ", offset]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        return self._findFirstCaretContext(child, 0)

    def findNextCaretInOrder(self, obj=None, offset=-1):
        startTime = time.time()
        rv = self._findNextCaretInOrder(obj, offset)
        tokens = ["WEB: Next caret in order for", obj, ", ", offset, ":",
                  rv[0], ", ", rv[1], f"({time.time() - startTime:.4f}s)"]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        return rv

    def _findNextCaretInOrder(self, obj=None, offset=-1):
        if not obj:
            obj, offset = self.getCaretContext()

        if not obj or not self.inDocumentContent(obj):
            return None, -1

        if self._canHaveCaretContext(obj):
            if self.treatAsTextObject(obj):
                allText = AXText.get_all_text(obj)
                for i in range(offset + 1, len(allText)):
                    child = AXHypertext.get_child_at_offset(obj, i)
                    if child and allText[i] != self.EMBEDDED_OBJECT_CHARACTER:
                        tokens = ["ERROR: Child", child, "found at offset with char '",
                                  allText[i].replace("\n", "\\n"), "'"]
                        debug.printTokens(debug.LEVEL_INFO, tokens, True)
                    if self._canHaveCaretContext(child):
                        if self._treatObjectAsWhole(child, -1):
                            return child, 0
                        return self._findNextCaretInOrder(child, -1)
                    if allText[i] not in (
                            self.EMBEDDED_OBJECT_CHARACTER, self.ZERO_WIDTH_NO_BREAK_SPACE):
                        return obj, i
            elif AXObject.get_child_count(obj) and not self._treatObjectAsWhole(obj, offset):
                return self._findNextCaretInOrder(AXObject.get_child(obj, 0), -1)
            elif offset < 0 and not self.isTextBlockElement(obj):
                return obj, 0

        # If we're here, start looking up the tree, up to the document.
        if self.isTopLevelDocument(obj):
            return None, -1

        while obj and AXObject.get_parent(obj):
            if self.isDetachedDocument(AXObject.get_parent(obj)):
                obj = self.iframeForDetachedDocument(AXObject.get_parent(obj))
                continue

            parent = AXObject.get_parent(obj)
            if not AXObject.is_valid(parent):
                msg = "WEB: Finding next caret in order. Parent is not valid."
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                replicant = self.findReplicant(self.documentFrame(), parent)
                if AXObject.is_valid(replicant):
                    parent = replicant
                elif AXObject.get_parent(parent):
                    obj = parent
                    continue
                else:
                    break

            start, end, length = self._rangeInParentWithLength(obj)
            if start + 1 == end and 0 <= start < end <= length:
                return self._findNextCaretInOrder(parent, start)

            child = AXObject.get_next_sibling(obj)
            if child:
                return self._findNextCaretInOrder(child, -1)
            obj = parent

        return None, -1

    def findPreviousCaretInOrder(self, obj=None, offset=-1):
        startTime = time.time()
        rv = self._findPreviousCaretInOrder(obj, offset)
        tokens = ["WEB: Previous caret in order for", obj, ", ", offset, ":",
                  rv[0], ", ", rv[1], f"({time.time() - startTime:.4f}s)"]
        debug.printTokens(debug.LEVEL_INFO, tokens, True)
        return rv

    def _findPreviousCaretInOrder(self, obj=None, offset=-1):
        if not obj:
            obj, offset = self.getCaretContext()

        if not obj or not self.inDocumentContent(obj):
            return None, -1

        if self._canHaveCaretContext(obj):
            if self.treatAsTextObject(obj):
                allText = AXText.get_all_text(obj)
                if offset == -1 or offset > len(allText):
                    offset = len(allText)
                for i in range(offset - 1, -1, -1):
                    child = AXHypertext.get_child_at_offset(obj, i)
                    if child and allText[i] != self.EMBEDDED_OBJECT_CHARACTER:
                        tokens = ["ERROR: Child", child, "found at offset with char '",
                                  allText[i].replace("\n", "\\n"), "'"]
                        debug.printTokens(debug.LEVEL_INFO, tokens, True)
                    if self._canHaveCaretContext(child):
                        if self._treatObjectAsWhole(child, -1):
                            return child, 0
                        return self._findPreviousCaretInOrder(child, -1)
                    if allText[i] not in (
                            self.EMBEDDED_OBJECT_CHARACTER, self.ZERO_WIDTH_NO_BREAK_SPACE):
                        return obj, i
            elif AXObject.get_child_count(obj) and not self._treatObjectAsWhole(obj, offset):
                return self._findPreviousCaretInOrder(
                    AXObject.get_child(obj, AXObject.get_child_count(obj) - 1), -1)
            elif offset < 0 and not self.isTextBlockElement(obj):
                return obj, 0

        # If we're here, start looking up the tree, up to the document.
        if self.isTopLevelDocument(obj):
            return None, -1

        while obj and AXObject.get_parent(obj):
            if self.isDetachedDocument(AXObject.get_parent(obj)):
                obj = self.iframeForDetachedDocument(AXObject.get_parent(obj))
                continue

            parent = AXObject.get_parent(obj)
            if not AXObject.is_valid(parent):
                msg = "WEB: Finding previous caret in order. Parent is not valid."
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                replicant = self.findReplicant(self.documentFrame(), parent)
                if AXObject.is_valid(replicant):
                    parent = replicant
                elif AXObject.get_parent(parent):
                    obj = parent
                    continue
                else:
                    break

            start, end, length = self._rangeInParentWithLength(obj)
            if start + 1 == end and 0 <= start < end <= length:
                return self._findPreviousCaretInOrder(parent, start)

            child = AXObject.get_previous_sibling(obj)
            if child:
                return self._findPreviousCaretInOrder(child, -1)
            obj = parent

        return None, -1

    def lastQueuedLiveRegion(self):
        if self._lastQueuedLiveRegionEvent is None:
            return None

        if self._lastQueuedLiveRegionEvent.type.startswith("object:text-changed:insert"):
            return self._lastQueuedLiveRegionEvent.source

        if self._lastQueuedLiveRegionEvent.type.startswith("object:children-changed:add"):
            return self._lastQueuedLiveRegionEvent.any_data

        return None

    def handleAsLiveRegion(self, event):
        if not settings_manager.getManager().getSetting('inferLiveRegions'):
            return False

        if not self.isLiveRegion(event.source):
            return False

        if not settings_manager.getManager().getSetting('presentLiveRegionFromInactiveTab') \
           and self.getTopLevelDocumentForObject(event.source) != self.activeDocument():
            msg = "WEB: Live region source is not in active tab."
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return False

        if event.type.startswith("object:text-changed:insert"):
            alert = AXObject.find_ancestor(event.source, self.isAriaAlert)
            if alert and AXUtilities.get_focused_object(alert) == event.source:
                msg = "WEB: Focused source will be presented as part of alert"
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                return False

            if self._lastQueuedLiveRegionEvent \
               and self._lastQueuedLiveRegionEvent.type == event.type \
               and self._lastQueuedLiveRegionEvent.any_data == event.any_data:
                msg = "WEB: Event is believed to be duplicate message"
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                return False

        if isinstance(event.any_data, Atspi.Accessible):
            if AXUtilities.is_unknown_or_redundant(event.any_data) \
               and self._getTag(event.any_data) in ["", None, "br"]:
                tokens = ["WEB: Child has unknown role and no tag", event.any_data]
                debug.printTokens(debug.LEVEL_INFO, tokens, True)
                return False

            if self.lastQueuedLiveRegion() == event.any_data \
               and self._lastQueuedLiveRegionEvent.type != event.type:
                msg = "WEB: Event is believed to be redundant live region notification"
                debug.printMessage(debug.LEVEL_INFO, msg, True)
                return False

        self._lastQueuedLiveRegionEvent = event
        return True

    def preferDescriptionOverName(self, obj):
        if not self.inDocumentContent(obj):
            return super().preferDescriptionOverName(obj)

        rv = self._preferDescriptionOverName.get(hash(obj))
        if rv is not None:
            return rv

        name = AXObject.get_name(obj)
        if len(name) == 1 and ord(name) in range(0xe000, 0xf8ff):
            tokens = ["WEB: name of", obj, "is in unicode private use area"]
            debug.printTokens(debug.LEVEL_INFO, tokens, True)
            rv = True
        elif AXObject.get_description(obj):
            rv = AXUtilities.is_push_button(obj) and len(name) == 1
        else:
            rv = False

        self._preferDescriptionOverName[hash(obj)] = rv
        return rv

    def _getCtrlShiftSelectionsStrings(self):
        """Hacky and to-be-obsoleted method."""
        return [messages.LINE_SELECTED_DOWN,
                messages.LINE_UNSELECTED_DOWN,
                messages.LINE_SELECTED_UP,
                messages.LINE_UNSELECTED_UP]

    def lastInputEventWasCopy(self):
        if super().lastInputEventWasCopy():
            return True

        if not self.inDocumentContent():
            return False

        if not self.topLevelObjectIsActiveAndCurrent():
            return False

        if AXObject.supports_action(focus_manager.getManager().get_locus_of_focus()):
            msg = "WEB: Treating locus of focus as source of copy"
            debug.printMessage(debug.LEVEL_INFO, msg, True)
            return True

        return False

Zerion Mini Shell 1.0