search using regular expression

README.md

@@ -1,13 +1,14 @@
-# EnhancedFindDialog for NVDA 1.6.1
+# EnhancedFindDialog for NVDA 1.7.0
 Enhanced find dialog addon for NVDA, implementing search improvements:
 
 * search history
+* Search by Regular Expressions
 * search wrapping, configured per profile
 * case sensitivity, configured per profile
 * contextual information on searches
 
 ## Download
-Download the [Enhanced Find Dialog 1.6.1 addon](https://github.com/marlon-sousa/EnhancedFindDialog/releases/download/1.6.1/EnhancedFindDialog-1.6.1.nvda-addon)
+Download the [Enhanced Find Dialog 1.7.0 addon](https://github.com/marlon-sousa/EnhancedFindDialog/releases/download/1.7.0/EnhancedFindDialog-1.7.0.nvda-addon)
 
 ## Note about secure mode
 
@@ -36,6 +37,27 @@ Simply install the addon. When it is activated, pressing down and up arrows on t
 
 You can at any time type a new term as usual.
 
+### Search by Regular Expressions
+
+In addition to the normal search supported by NVDA, this addon supports searching by regular expressions. For more information on how to get started with Regular Expressions, you can see for example [the Regular Expression HowTo tutorial for Python](https://docs.python.org/3/howto/regex.html), however many other tutorials are available on the Internet.
+
+Regular Expressions are specially useful to find some text that can vary on a web page.
+
+This option is profile specific, meaning that you can have a profile where it is active and other where it isn't.
+
+Please note that due to underlying technological implementation differences, this feature is not available on all programs (for example Microsoft Office Word documents) that support NVDA's find dialog.
+
+#### How it works?
+
+Simply install the addon. When it is activated, the find dialog will offer a `Search Type` radio box with two search options:
+
+* **Normal search** will perform NVDA's default search functionality.
+
+* **Regular Expression Search** will perform the search by Regular Expressions. Simply enter the Regular Expression on the text box and NVDA will place the cursor on the next match.
+
+
+Changing this radio box and performing a search will save the new state (normal or Regular Expression search for the active profile. Canceling the search won't change its state on the active profile, even if you changed it before canceling the search.
+
 ### Search wrapping
 
 Search Wrapping is a feature that, if configured, doesn't consider the current position you are on a text when performing searches.

README.tpl.md

@@ -2,6 +2,7 @@
 Enhanced find dialog addon for NVDA, implementing search improvements:
 
 * search history
+* Search by Regular Expressions
 * search wrapping, configured per profile
 * case sensitivity, configured per profile
 * contextual information on searches
@@ -36,6 +37,27 @@ Simply install the addon. When it is activated, pressing down and up arrows on t
 
 You can at any time type a new term as usual.
 
+### Search by Regular Expressions
+
+In addition to the normal search supported by NVDA, this addon supports searching by regular expressions. For more information on how to get started with Regular Expressions, you can see for example [the Regular Expression HowTo tutorial for Python](https://docs.python.org/3/howto/regex.html), however many other tutorials are available on the Internet.
+
+Regular Expressions are specially useful to find some text that can vary on a web page.
+
+This option is profile specific, meaning that you can have a profile where it is active and other where it isn't.
+
+Please note that due to underlying technological implementation differences, this feature is not available on all programs (for example Microsoft Office Word documents) that support NVDA's find dialog.
+
+#### How it works?
+
+Simply install the addon. When it is activated, the find dialog will offer a `Search Type` radio box with two search options:
+
+* **Normal search** will perform NVDA's default search functionality.
+
+* **Regular Expression Search** will perform the search by Regular Expressions. Simply enter the Regular Expression on the text box and NVDA will place the cursor on the next match.
+
+
+Changing this radio box and performing a search will save the new state (normal or Regular Expression search for the active profile. Canceling the search won't change its state on the active profile, even if you changed it before canceling the search.
+
 ### Search wrapping
 
 Search Wrapping is a feature that, if configured, doesn't consider the current position you are on a text when performing searches.

addon/globalPlugins/EnhancedFindDialog/__init__.py

@@ -18,6 +18,7 @@ def initConfiguration():
 	confspec = {
 		"searchCaseSensitivity": "boolean( default=False)",
 		"searchWrap": "boolean( default=False)",
+		"searchType": "string( default='NORMAL')",
 	}
 	config.conf.spec[module] = confspec
 
@@ -51,3 +52,5 @@ def injectProcessing(self):
 
 		# add methods to CursorManager class
 		cursorManagerHelper.patchCursorManager()
+		# add methods to offsetsTextInfo class
+		cursorManagerHelper.patchOffsetsTextInfo()

addon/globalPlugins/EnhancedFindDialog/cursorManagerHelper.py

@@ -4,14 +4,20 @@
 # This file is covered by the GNU General Public License.
 # See the file COPYING.txt for more details.
 
+
 import addonHandler
 import buildVersion
 import config
 import controlTypes
 from cursorManager import CursorManager
 import gui
+import textInfos.offsets
+import textUtils
 from . import guiHelper
+from .searchHistory import SearchHistory
+from .searchType import SearchType
 import inspect
+from logHandler import log
 import re
 import speech
 from scriptHandler import willSayAllResume, script
@@ -64,6 +70,12 @@ def patchCursorManager():
 	CursorManager.script_find = script_enhancedFind
 	CursorManager.script_findNext = script_enhancedFindNext
 	CursorManager.script_findPrevious = script_EnhancedFindPrevious
+	CursorManager.supportsRegexpSearch = supportsRegexpSearch
+	CursorManager.getSearchEntries = getSearchEntries
+
+
+def patchOffsetsTextInfo():
+	setattr(textInfos.offsets.OffsetsTextInfo, "findRegexp", findRegexp)
 
 
 def script_enhancedFind(self, gesture, reverse=False):
@@ -75,68 +87,97 @@ def run():
 		profile = config.conf.getActiveProfile()
 
 		gui.mainFrame.prePopup()
-		d = guiHelper.EnhancedFindDialog(gui.mainFrame, self, profile, self._searchEntries, reverse)
+		d = guiHelper.EnhancedFindDialog(gui.mainFrame, self, profile, reverse)
 		d.ShowModal()
 		gui.mainFrame.postPopup()
 	wx.CallAfter(run)
 
 
 def script_enhancedFindNext(self, gesture):
-	if not self._searchEntries:
+	mostRecentSearchTerm = getMostRecentSearchTerm()
+	if (mostRecentSearchTerm is None
+				or (mostRecentSearchTerm.searchType == SearchType.REGULAR_EXPRESSION.name
+					and not self.supportsRegexpSearch())):
 		self.script_find(gesture)
 		return
 	doFindText(
 		self,
-		self._searchEntries[SEARCH_HISTORY_MOST_RECENT_INDEX],
+		mostRecentSearchTerm,
 		caseSensitive=self._lastCaseSensitivity,
 		searchWrap=self._searchWrap,
 		willSayAllResume=willSayAllResume(gesture),
 	)
 
 
 def script_EnhancedFindPrevious(self, gesture):
-	if not self._searchEntries:
+	mostRecentSearchTerm = getMostRecentSearchTerm()
+	if (mostRecentSearchTerm is None
+				or (mostRecentSearchTerm.searchType == SearchType.REGULAR_EXPRESSION.name
+					and not self.supportsRegexpSearch())):
 		self.script_find(gesture, reverse=True)
 		return
 	doFindText(
 		self,
-		self._searchEntries[SEARCH_HISTORY_MOST_RECENT_INDEX],
+		mostRecentSearchTerm,
 		reverse=True,
 		caseSensitive=self._lastCaseSensitivity,
 		searchWrap=self._searchWrap,
 		willSayAllResume=willSayAllResume(gesture)
 	)
 
 
-def doFindText(cursorManagerInstance, text,
+def supportsRegexpSearch(self):
+	info = self.makeTextInfo(textInfos.POSITION_CARET)
+	return isinstance(info, textInfos.offsets.OffsetsTextInfo)
+
+
+def getSearchEntries(self):
+	searchHistory = SearchHistory.get()
+	if self.supportsRegexpSearch():
+		return searchHistory.getItems()
+	return searchHistory.getItems(searchType="normal")
+
+
+def getMostRecentSearchTerm():
+	searchHistory = SearchHistory.get()
+	return searchHistory.getMostRecent()
+
+
+def doFindText(cursorManagerInstance, searchTerm,
                reverse=False, caseSensitive=False, searchWrap=False, willSayAllResume=False):  # noqa: E101
-	if not text:
+	log.debug("doFindText, reverse=%s, caseSensitive=%s, searchWrap=%s", reverse, caseSensitive, searchWrap)
+	if not searchTerm:
 		return
 
 	info = cursorManagerInstance.makeTextInfo(textInfos.POSITION_CARET)
-	res = performSearch(cursorManagerInstance, text, info, reverse, caseSensitive, searchWrap)
+	res = performSearch(cursorManagerInstance, searchTerm, info, reverse, caseSensitive, searchWrap)
 	if res:
 		cursorManagerInstance.selection = info
 		speech.cancelSpeech()
-		info.move(textInfos.UNIT_LINE, 1, endPoint="start")
-		info.expand(textInfos.UNIT_LINE)
+
+		if searchTerm.searchType == SearchType.REGULAR_EXPRESSION.name:
+			info.move(textInfos.UNIT_LINE, 1, endPoint="end")
+		else:
+			info.move(textInfos.UNIT_LINE, 1, endPoint="start")
+			info.expand(textInfos.UNIT_LINE)
+
 		if not willSayAllResume:
 			speech.speakTextInfo(info, reason=controlTypes.OutputReason.CARET)
 	else:
-		wx.CallAfter(gui.messageBox, __('text "%s" not found') % text,
+		wx.CallAfter(gui.messageBox, __('text "%s" not found') % searchTerm.text,
 		             FIND_ERROR_DIALOG_TITLE, wx.OK | wx.ICON_ERROR)  # Noqa E101
-	CursorManager._lastFindText = text
+	CursorManager._lastFindText = searchTerm.text
 	CursorManager._lastCaseSensitivity = caseSensitive
 	CursorManager._searchWrap = searchWrap
 
 
-def performSearch(cursorManager, text, info, reverse, caseSensitive, wrapSearch):
-	res = info.find(text, reverse=reverse, caseSensitive=caseSensitive)
+def performSearch(cursorManager, searchTerm, info, reverse, caseSensitive, wrapSearch):
+	res = find(cursorManager, searchTerm, info, reverse=reverse, caseSensitive=caseSensitive)
 	# if either not interested in search wrapping or we have found a result then we are done here
 	if not wrapSearch or res:
 		return res
 	found = False
-	while info.find(text, reverse=(not reverse), caseSensitive=caseSensitive):
+	while find(cursorManager, searchTerm, info, reverse=(not reverse), caseSensitive=caseSensitive):
 		found = True
 	if found:
 		beep(440, 30)
@@ -152,12 +193,56 @@ def performSearch(cursorManager, text, info, reverse, caseSensitive, wrapSearch)
 	info = cursorManager.makeTextInfo(textInfos.POSITION_CARET).copy()
 	info.expand(textInfos.UNIT_STORY)
 	inText = info._get_text()
-	found = re.search(re.escape(text), inText, (0 if caseSensitive else re.IGNORECASE) | re.UNICODE)
+
+	if searchTerm.searchType == SearchType.REGULAR_EXPRESSION.name:
+		found = re.search(searchTerm.text, inText, re.UNICODE)
+	else:
+		found = re.search(re.escape(searchTerm.text), inText, (0 if caseSensitive else re.IGNORECASE) | re.UNICODE)
+
 	if found:
 		beep(440, 30)
 	return found
 
 
+def find(cursorManager, searchTerm, info, reverse, caseSensitive):
+	if searchTerm.searchType == SearchType.REGULAR_EXPRESSION.name:
+		if not cursorManager.supportsRegexpSearch():
+			wx.CallAfter(
+				# Translators: Message shown when an invalid regular expression is entered.
+				_("current textInfo backend does not support regular expression searches"),
+				FIND_ERROR_DIALOG_TITLE, wx.OK | wx.ICON_ERROR
+			)
+			return None
+		res = info.findRegexp(searchTerm.text, reverse=reverse)
+	else:
+		res = info.find(searchTerm.text, reverse=reverse, caseSensitive=caseSensitive)
+	return res
+
+
+def findRegexp(self, text, reverse=False):
+	if reverse:
+		inText = self._getTextRange(0, self._startOffset)
+		matches = list(re.finditer(text, inText, re.UNICODE))
+		if not matches:
+			return False
+		m = matches[-1]
+	else:
+		inText = self._getTextRange(self._startOffset + 1, self._getStoryLength())
+		m = re.search(text, inText, re.UNICODE)
+	if not m:
+		return False
+
+	converter = textUtils.getOffsetConverter(self.encoding)(inText)
+
+	if reverse:
+		offset = converter.strToEncodedOffsets(m.start())
+	else:
+		offset = self._startOffset + 1 + converter.strToEncodedOffsets(m.start())
+
+	self._startOffset = self._endOffset = offset
+	return True
+
+
 # Patch say all functionality.
 # Fix the patched script losing the reference to the decorated say all,
 # documentation and other necessary attributes for the gesture