Sentence navigation

Author: mltony

source/NVDAObjects/UIA/__init__.py

@@ -65,6 +65,7 @@
 import ui
 import winVersion
 import NVDAObjects
+from documentNavigation import sentenceNavigation
 
 
 paragraphIndentIDs = {
@@ -899,26 +900,66 @@ def _get_boundingRects(self):
 		return self._getBoundingRectsFromUIARange(self._rangeObj)
 
 	def expand(self, unit: str) -> None:
-		UIAUnit=UIAHandler.NVDAUnitsToUIAUnits[unit]
-		self._rangeObj.ExpandToEnclosingUnit(UIAUnit)
+		if unit == textInfos.UNIT_SENTENCE:
+			context = sentenceNavigation.SentenceContext(self)
+			sentence = context.moveSentence(0)
+			self._rangeObj = sentence._rangeObj
+		else:
+			UIAUnit = UIAHandler.NVDAUnitsToUIAUnits[unit]
+			self._rangeObj.ExpandToEnclosingUnit(UIAUnit)
 
 	def move(
 			self,
 			unit: str,
 			direction: int,
 			endPoint: Optional[str] = None,
 	):
-		UIAUnit=UIAHandler.NVDAUnitsToUIAUnits[unit]
-		if endPoint=="start":
-			res=self._rangeObj.MoveEndpointByUnit(UIAHandler.TextPatternRangeEndpoint_Start,UIAUnit,direction)
-		elif endPoint=="end":
-			res=self._rangeObj.MoveEndpointByUnit(UIAHandler.TextPatternRangeEndpoint_End,UIAUnit,direction)
+		if unit == textInfos.UNIT_SENTENCE:
+			if direction == 0:
+				return 0
+			endPointInfo = self.copy()
+			if endPoint == "start":
+				endPointInfo.collapse()
+			elif endPoint == "end":
+				endPointInfo.collapse(end=True)
+			if direction > 0:
+				iteration = range(direction)
+				direction = 1
+			else:
+				iteration = range(-direction)
+				direction = -1
+			result = 0
+			for i in iteration:
+				context = sentenceNavigation.SentenceContext(endPointInfo)
+				sentence = context.moveSentence(direction)
+				if sentence is not None:
+					result += direction
+					endPointInfo = sentence
+				else:
+					break
+			if result == 0:
+				return 0
+			endPointInfo.collapse()
+			if endPoint == "start":
+				self.setEndPoint(endPointInfo, "startToStart")
+			elif endPoint == "end":
+				self.setEndPoint(endPointInfo, "endToEnd")
+			else:
+				self._rangeObj = endPointInfo._rangeObj
+			return result
 		else:
-			res=self._rangeObj.Move(UIAUnit,direction)
-		#Some Implementations of Move and moveEndpointByUnit return a positive number even if the direction is negative
-		if direction<0 and res>0:
-			res=0-res
-		return res
+			UIAUnit = UIAHandler.NVDAUnitsToUIAUnits[unit]
+			if endPoint == "start":
+				res = self._rangeObj.MoveEndpointByUnit(UIAHandler.TextPatternRangeEndpoint_Start, UIAUnit, direction)
+			elif endPoint == "end":
+				res = self._rangeObj.MoveEndpointByUnit(UIAHandler.TextPatternRangeEndpoint_End, UIAUnit, direction)
+			else:
+				res = self._rangeObj.Move(UIAUnit, direction)
+			# Some Implementations of Move and moveEndpointByUnit return a positive number even if the direction is
+			# negative
+			if direction < 0 and res > 0:
+				res = 0 - res
+			return res
 
 	def copy(self):
 		return self.__class__(self.obj,None,_rangeObj=self._rangeObj)

source/browseMode.py

@@ -56,6 +56,7 @@
 from abc import ABCMeta, abstractmethod
 import globalVars
 from typing import Optional
+from documentNavigation.sentenceNavigation import getSentenceStopRegex
 
 
 def reportPassThrough(treeInterceptor,onlyIfChanged=True):
@@ -473,12 +474,14 @@ def _quickNavScript(self,gesture, itemType, direction, errorMessage, readUnit):
 		if itemType=="notLinkBlock":
 			iterFactory=self._iterNotLinkBlock
 		elif itemType == "textParagraph":
-			punctuationMarksRegex = re.compile(
-				config.conf["virtualBuffers"]["textParagraphRegex"],
-			)
+			punctuationMarksRegex = getSentenceStopRegex()
 
 			def paragraphFunc(info: textInfos.TextInfo) -> bool:
-				return punctuationMarksRegex.search(info.text) is not None
+				# sentence regex always matches beginning and end of string. We add some words at the end to test
+				# whether there is third match in the middle, that would be suggestive of a complete sentence.
+				text = info.text + " traiiling word"
+				matches = punctuationMarksRegex.findall(text)
+				return len(matches) >= 3
 
 			def iterFactory(direction: str, pos: textInfos.TextInfo) -> Generator[TextInfoQuickNavItem, None, None]:
 				return self._iterSimilarParagraph(
@@ -1680,6 +1683,10 @@ def event_caretMovementFailed(self, obj, nextHandler, gesture=None):
 	currentExpandedControl=None #: an NVDAObject representing the control that has just been expanded with the collapseOrExpandControl script.
 
 	def script_collapseOrExpandControl(self, gesture: inputCore.InputGesture):
+		states = self.currentNVDAObject.states
+		if controlTypes.State.COLLAPSED not in states and controlTypes.State.EXPANDED not in states:
+			direction = 1 if gesture.mainKeyName == "downArrow" else -1
+			return self._caretMovementScriptHelper(gesture, textInfos.UNIT_SENTENCE, direction)
 		if not config.conf["virtualBuffers"]["autoFocusFocusableElements"]:
 			self._focusLastFocusableObject()
 			# Give the application time to focus the control.

source/config/configDefaults.py

@@ -2,32 +2,46 @@
 # Copyright (C) 2024 NV Access Limited
 # This file is covered by the GNU General Public License.
 # See the file COPYING for more details.
+import json
+
 
 DEFAULT_TEXT_PARAGRAPH_REGEX = (
-	r"({lookBehind}{optQuote}{punc}{optQuote}{optWiki}{lookAhead}|{punc2}|{cjk})".format(
+	r"^|({plb}{nlb}{optQuote}{dot}{optQuote}|{punc2}{optQuote}|{cjk}){optWiki}{spaces}|{n2}|\Z".format(
 		# Look behind clause ensures that we have a text character before text punctuation mark.
 		# We have a positive lookBehind \w that resolves to a text character in any language,
 		# coupled with negative lookBehind \d that excludes digits.
-		lookBehind=r'(?<=\w)(?<!\d)',
+		plb=r"(?<=\w)(?<!\d)",
+		# Language-specific exceptions: characters suggesting that the following dot is not indicative
+		# of a sentence stop.
+		# This is a negative look-behind and will be inserted later when language is specified.
+		nlb="{nonBreakingRegex}",
 		# In some cases quote or closing parenthesis might appear right before or right after text punctuation.
 		# For example:
 		# > He replied, "That's wonderful."
 		optQuote=r'["”»)]?',
 		# Actual punctuation marks that suggest end of sentence.
 		# We don't include symbols like comma and colon, because of too many false positives.
 		# We include question mark and exclamation mark below in punc2.
-		punc=r'[.…]{1,3}',
+		dot=r"[.]{{1,3}}",
 		# On Wikipedia references appear right after period in sentences, the following clause takes this
 		# into account. For example:
 		# > On his father's side, he was a direct descendant of John Churchill.[3]
-		optWiki=r'(\[\d+\])*',
-		# LookAhead clause checks that punctuation mark must be followed by either space,
+		optWiki=r"(\[\d+\])*",
+		# spaces clause checks that punctuation mark must be followed by either space,
 		# or newLine symbol or end of string.
-		lookAhead=r'(?=[\r\n  ]|$)',
+		spaces=r"([  ]+|([  \t]*\n)+|$)",
 		# Include question mark and exclamation mark with no extra conditions,
 		# since they don't trigger as many false positives.
-		punc2=r'[?!]',
+		punc2=r"[?!…]",
 		# We also check for CJK full-width punctuation marks without any extra rules.
-		cjk=r'[．！？：；]',
+		cjk=r"[．！？：；]",
+		# Double newline means end of sentence too.
+		n2=r"([  \t]*\n){{2,}}",
 	)
 )
+
+
+nonBreakingPrefix = json.dumps({
+	"en": r"\b[A-Z]|\bMr|\bMs|\bMrs|\bDr|\bProf|\bSt|\be.g|\bi.e",
+	"ru": r"\b[A-ZА-Я]|\b[Тт]ов|\b[Уу]л|\bт.[ке]|\bт. [ке]",
+})

source/config/configSpec.py

@@ -244,6 +244,8 @@
 
 [documentNavigation]
 	paragraphStyle = featureFlag(optionsEnum="ParagraphNavigationFlag", behaviorOfDefault="application")
+	sentenceReconstruction = featureFlag(optionsEnum="SentenceReconstructionFlag", behaviorOfDefault="same_style_paragraphs")  # noqa: E501 Breaking this line makes invalid config
+	nonBreakingPrefixRegex = string(default='{configDefaults.nonBreakingPrefix}')
 
 [reviewCursor]
 	simpleReviewMode = boolean(default=True)

source/config/featureFlagEnums.py

@@ -100,6 +100,25 @@ def _displayStringLabels(self):
 	ALWAYS = enum.auto()
 
 
+class SentenceReconstructionFlag(DisplayStringEnum):
+	@property
+	def _displayStringLabels(self):
+		return {
+			# Translators: Sentence reconstruction mode
+			SentenceReconstructionFlag.NEVER: _("Never reconstruct sentences across paragraphs"),
+			# Translators: Sentence reconstruction mode
+			SentenceReconstructionFlag.SAME_STYLE_PARAGRAPHS: _("Reconstruct sentences across same style paragraphs"),
+			# Translators: Sentence reconstruction mode
+			SentenceReconstructionFlag.ANY_PARAGRAPHS: _("Reconstruct sentences across any paragraphs"),
+		}
+
+	DEFAULT = enum.auto()
+	NEVER = enum.auto()
+	SAME_STYLE_PARAGRAPHS = enum.auto()
+	ANY_PARAGRAPHS = enum.auto()
+
+
+
 class WindowsTerminalStrategyFlag(DisplayStringEnum):
 	"""
 	A feature flag for defining how new text is calculated in Windows Terminal