Grammalecte  Diff

                        sText = self.parseGraph(_rules_graph.dAllGraph[sGraphName], sCountry, dOptions, bShowRuleId, bDebug, bContext)
            elif not sOption or dOptions.get(sOption, False):
                # regex rules
                for zRegex, bUppercase, sLineId, sRuleId, nPriority, lActions in lRuleGroup:
                    if sRuleId not in _aIgnoredRules:
                        for m in zRegex.finditer(sText):
                            bCondMemo = None
                            for sFuncCond, cActionType, sAction, *eAct in lActions:
                                # action in lActions: [ condition, action type, replacement/suggestion/action[, iGroup[, message, URL]] ]
                                try:
                                    bCondMemo = not sFuncCond or getattr(gc_functions, sFuncCond)(sText, sText0, m, self.dTokenPos, sCountry, bCondMemo)
                                    if bCondMemo:
                                        if bDebug:
                                            echo("RULE: " + sLineId)
                                        if cActionType == "-":
                                            # grammar error
                                            nErrorStart = nOffset + m.start(eAct[0])
                                            if nErrorStart not in self.dError or nPriority > self.dErrorPriority.get(nErrorStart, -1):
                                                self.dError[nErrorStart] = self._createErrorFromRegex(sText, sText0, sAction, nOffset, m, eAct[0], sLineId, sRuleId, bUppercase, eAct[1], eAct[2], bShowRuleId, sOption, bContext)
                                                self.dErrorPriority[nErrorStart] = nPriority
                                                self.dSentenceError[nErrorStart] = self.dError[nErrorStart]
                                        elif cActionType == "~":
                                            # text processor
                                            sText = self.rewriteText(sText, sAction, eAct[0], m, bUppercase)
                                            bChange = True
                                            if bDebug:
                                                echo("~ " + sText + "  -- " + m.group(eAct[0]) + "  # " + sLineId)
                                        elif cActionType == "=":
                                            # disambiguation
                                            if not bParagraph:
                                                getattr(gc_functions, sAction)(sText, m, self.dTokenPos)
                                                if bDebug:
                                                    echo("= " + m.group(0) + "  # " + sLineId)
                                        elif cActionType == ">":
                                            # we do nothing, this test is just a condition to apply all following actions
                                            pass
                                        else:
                                            echo("# error: unknown action at " + sLineId)

                dToken["aTags"] = self.dTokenPos[dToken["nStart"]]["aTags"]
        self.lTokens = lNewTokens
        self.dTokenPos = { dToken["nStart"]: dToken  for dToken in self.lTokens  if dToken["sType"] != "INFO" }
        if bDebug:
            echo("UPDATE:")
            echo(self)

    def _getNextNodes (self, dGraph, dToken, dNode, bKeep=False):
        "generator: return matches where <dToken> “values” match <dNode> arcs"
        bTokenFound = False
        # token value
        if dToken["sValue"] in dNode:
            yield (" ", dToken["sValue"], dNode[dToken["sValue"]])
            bTokenFound = True
        if dToken["sValue"][0:2].istitle(): # we test only 2 first chars, to match words such as "Laissez-les", "Crève-cœur".

                                        bTokenFound = True
                            else:
                                if sNegPattern and any(re.search(sNegPattern, sMorph)  for sMorph in lMorph):
                                    continue
                                if not sPattern or any(re.search(sPattern, sMorph)  for sMorph in lMorph):
                                    yield ("@", sRegex, dNode["<re_morph>"][sRegex])
                                    bTokenFound = True
            # regex multi morph arcs
            if "<re_mmorph>" in dNode:
                if "nMultiStartTo" in dToken:
                    lMorph = dToken["dMultiToken"]["lMorph"]
                    for sRegex in dNode["<re_mmorph>"]:
                        if "¬" not in sRegex:
                            # no anti-pattern
                            if any(re.search(sRegex, sMorph)  for sMorph in lMorph):
                                yield ("&", sRegex, dNode["<re_mmorph>"][sRegex])
                                bTokenFound = True
                        else:
                            # there is an anti-pattern
                            sPattern, sNegPattern = sRegex.split("¬", 1)
                            if sNegPattern == "*":
                                # all morphologies must match with <sPattern>
                                if sPattern:
                                    if all(re.search(sPattern, sMorph)  for sMorph in lMorph):
                                        yield ("&", sRegex, dNode["<re_mmorph>"][sRegex])
                                        bTokenFound = True
                            else:
                                if sNegPattern and any(re.search(sNegPattern, sMorph)  for sMorph in lMorph):
                                    continue
                                if not sPattern or any(re.search(sPattern, sMorph)  for sMorph in lMorph):
                                    yield ("&", sRegex, dNode["<re_mmorph>"][sRegex])
                                    bTokenFound = True
        # token tags
        if "aTags" in dToken and "<tags>" in dNode:
            for sTag in dToken["aTags"]:
                if sTag in dNode["<tags>"]:
                    yield ("/", sTag, dNode["<tags>"][sTag])
                    bTokenFound = True
        # meta arc (for token type)

                        yield ("*", sMeta, dNode["<meta>"][sMeta])
                        bTokenFound = True
        if not bTokenFound and bKeep:
            yield (None, "", -1)
        # JUMP
        # Warning! Recursion!
        if "<>" in dNode:
            yield from self._getNextNodes(dGraph, dToken, dGraph[dNode["<>"]], bKeep=True)

    def parseGraph (self, dGraph, sCountry="${country_default}", dOptions=None, bShowRuleId=False, bDebug=False, bContext=False):
        "parse graph with tokens from the text and execute actions encountered"
        lPointers = []
        bTagAndRewrite = False
        for iToken, dToken in enumerate(self.lTokens):
            if bDebug:
                echo("TOKEN: " + dToken["sValue"])
            # check arcs for each existing pointer
            lNextPointers = []
            for dPointer in lPointers:
                if dPointer["nMultiEnd"] != -1:
                    if dToken["i"] <= dPointer["nMultiEnd"]:
                        lNextPointers.append(dPointer)
                    if dToken["i"] != dPointer["nMultiEnd"]:
                        continue
                for cNodeType, sMatch, iNode in self._getNextNodes(dGraph, dToken, dGraph[dPointer["iNode"]]):
                    if cNodeType is None:
                        lNextPointers.append(dPointer)
                        continue
                    if bDebug:
                        echo("  MATCH: " + cNodeType + sMatch)
                    nMultiEnd = -1  if cNodeType != "&"  else dToken["nMultiStartTo"]
                    lNextPointers.append({ "iToken1": dPointer["iToken1"], "iNode": iNode, "nMultiEnd": nMultiEnd })
            lPointers = lNextPointers
            # check arcs of first nodes
            for cNodeType, sMatch, iNode in self._getNextNodes(dGraph, dToken, dGraph[0]):
                if cNodeType is None:
                    continue
                if bDebug:
                    echo("  MATCH: " + cNodeType + sMatch)
                nMultiEnd = -1  if cNodeType != "&"  else dToken["nMultiStartTo"]
                lPointers.append({ "iToken1": iToken, "iNode": iNode, "nMultiEnd": nMultiEnd })
            # check if there is rules to check for each pointer
            for dPointer in lPointers:
                if dPointer["nMultiEnd"] != -1:
                    if dToken["i"] < dPointer["nMultiEnd"]:
                        continue
                    if dToken["i"] == dPointer["nMultiEnd"]:
                        dPointer["nMultiEnd"] = -1
                if "<rules>" in dGraph[dPointer["iNode"]]:
                    bChange = self._executeActions(dGraph, dGraph[dPointer["iNode"]]["<rules>"], dPointer["iToken1"]-1, iToken, dOptions, sCountry, bShowRuleId, bDebug, bContext)
                    if bChange:
                        bTagAndRewrite = True
        if bTagAndRewrite:
            self.rewriteFromTags(bDebug)
        if bDebug:
            echo(self)
        return self.sSentence

    def _executeActions (self, dGraph, dNode, nTokenOffset, nLastToken, dOptions, sCountry, bShowRuleId, bDebug, bContext):
        "execute actions found in the DARG"
        bChange = False
        for sLineId, nextNodeKey in dNode.items():
            bCondMemo = None
            for sRuleId in dGraph[nextNodeKey]:
                try:
                    if bDebug:
                        echo("   >TRY: " + sRuleId + " " + sLineId)
                    _, sOption, sFuncCond, cActionType, sAction, *eAct = _rules_graph.dRule[sRuleId]
                    # Suggestion    [ option, condition, "-", replacement/suggestion/action, iTokenStart, iTokenEnd, cStartLimit, cEndLimit, bCaseSvty, nPriority, sMessage, iURL ]
                    # TextProcessor [ option, condition, "~", replacement/suggestion/action, iTokenStart, iTokenEnd, bCaseSvty ]
                    # Disambiguator [ option, condition, "=", replacement/suggestion/action ]
                    # Tag           [ option, condition, "/", replacement/suggestion/action, iTokenStart, iTokenEnd ]
                    # Immunity      [ option, condition, "!", option,                        iTokenStart, iTokenEnd ]
                    # Multi-token   [ option, condition, "&", morphologies,                  iTokenStart, iTokenEnd ]
                    # Test          [ option, condition, ">", "" ]
                    if not sOption or dOptions.get(sOption, False):
                        bCondMemo = not sFuncCond or getattr(gc_functions, sFuncCond)(self.lTokens, nTokenOffset, nLastToken, sCountry, bCondMemo, self.dTags, self.sSentence, self.sSentence0)
                        if bCondMemo:
                            if cActionType == "-":
                                # grammar error
                                iTokenStart, iTokenEnd, cStartLimit, cEndLimit, bCaseSvty, nPriority, sMessage, iURL = eAct
                                nTokenErrorStart = nTokenOffset + iTokenStart  if iTokenStart > 0  else nLastToken + iTokenStart
                                if "sImmunity" not in self.lTokens[nTokenErrorStart] or (self.lTokens[nTokenErrorStart]["sImmunity"] != "*" and sOption not in self.lTokens[nTokenErrorStart]["sImmunity"]):
                                    nTokenErrorEnd = nTokenOffset + iTokenEnd  if iTokenEnd > 0  else nLastToken + iTokenEnd
                                    nErrorStart = self.nOffsetWithinParagraph + (self.lTokens[nTokenErrorStart]["nStart"] if cStartLimit == "<"  else self.lTokens[nTokenErrorStart]["nEnd"])
                                    nErrorEnd = self.nOffsetWithinParagraph + (self.lTokens[nTokenErrorEnd]["nEnd"] if cEndLimit == ">"  else self.lTokens[nTokenErrorEnd]["nStart"])
                                    if nErrorStart not in self.dError or nPriority > self.dErrorPriority.get(nErrorStart, -1):
                                        self.dError[nErrorStart] = self._createErrorFromTokens(sAction, nTokenOffset, nLastToken, nTokenErrorStart, nErrorStart, nErrorEnd, sLineId, sRuleId, bCaseSvty, \
                                                                                               sMessage, _rules_graph.dURL.get(iURL, ""), bShowRuleId, sOption, bContext)
                                        self.dErrorPriority[nErrorStart] = nPriority
                                        self.dSentenceError[nErrorStart] = self.dError[nErrorStart]
                                        if bDebug:
                                            echo("    NEW_ERROR: {}".format(self.dError[nErrorStart]))
                            elif cActionType == "~":
                                # text processor
                                nTokenStart = nTokenOffset + eAct[0]  if eAct[0] > 0  else nLastToken + eAct[0]
                                nTokenEnd = nTokenOffset + eAct[1]  if eAct[1] > 0  else nLastToken + eAct[1]
                                self._tagAndPrepareTokenForRewriting(sAction, nTokenStart, nTokenEnd, nTokenOffset, nLastToken, eAct[2], bDebug)
                                bChange = True
                                if bDebug:
                                    echo("    TEXT_PROCESSOR: [{}:{}]  > {}".format(self.lTokens[nTokenStart]["sValue"], self.lTokens[nTokenEnd]["sValue"], sAction))
                            elif cActionType == "=":
                                # disambiguation
                                getattr(gc_functions, sAction)(self.lTokens, nTokenOffset, nLastToken)
                                if bDebug:
                                    echo("    DISAMBIGUATOR: ({})  [{}:{}]".format(sAction, self.lTokens[nTokenOffset+1]["sValue"], self.lTokens[nLastToken]["sValue"]))
                            elif cActionType == ">":
                                # we do nothing, this test is just a condition to apply all following actions
                                if bDebug:
                                    echo("    COND_OK")
                            elif cActionType == "/":
                                # Tag
                                nTokenStart = nTokenOffset + eAct[0]  if eAct[0] > 0  else nLastToken + eAct[0]
                                nTokenEnd = nTokenOffset + eAct[1]  if eAct[1] > 0  else nLastToken + eAct[1]
                                for i in range(nTokenStart, nTokenEnd+1):
                                    if "aTags" in self.lTokens[i]:
                                        self.lTokens[i]["aTags"].update(sAction.split("|"))
                                    else:
                                        self.lTokens[i]["aTags"] = set(sAction.split("|"))
                                if bDebug:
                                    echo("    TAG: {} >  [{}:{}]".format(sAction, self.lTokens[nTokenStart]["sValue"], self.lTokens[nTokenEnd]["sValue"]))
                                for sTag in sAction.split("|"):
                                    if sTag not in self.dTags:
                                        self.dTags[sTag] = [nTokenStart, nTokenEnd]
                                    else:
                                        self.dTags[sTag][0] = min(nTokenStart, self.dTags[sTag][0])
                                        self.dTags[sTag][1] = max(nTokenEnd, self.dTags[sTag][1])
                            elif cActionType == "!":
                                # immunity
                                if bDebug:
                                    echo("    IMMUNITY: " + sLineId + " / " + sRuleId)
                                nTokenStart = nTokenOffset + eAct[0]  if eAct[0] > 0  else nLastToken + eAct[0]
                                nTokenEnd = nTokenOffset + eAct[1]  if eAct[1] > 0  else nLastToken + eAct[1]
                                sImmunity = sAction or "*"
                                if nTokenEnd - nTokenStart == 0:
                                    self.lTokens[nTokenStart]["sImmunity"] = sImmunity
                                    nErrorStart = self.nOffsetWithinParagraph + self.lTokens[nTokenStart]["nStart"]
                                    if nErrorStart in self.dError:
                                        del self.dError[nErrorStart]
                                else:
                                    for i in range(nTokenStart, nTokenEnd+1):
                                        self.lTokens[i]["sImmunity"] = sImmunity
                                        nErrorStart = self.nOffsetWithinParagraph + self.lTokens[i]["nStart"]
                                        if nErrorStart in self.dError:
                                            del self.dError[nErrorStart]
                            elif cActionType == "&":
                                # multi-tokens
                                nTokenStart = nTokenOffset + eAct[0]  if eAct[0] > 0  else nLastToken + eAct[0]
                                nTokenEnd = nTokenOffset + eAct[1]  if eAct[1] > 0  else nLastToken + eAct[1]
                                dMultiToken = {
                                    "nTokenStart": nTokenStart,
                                    "nTokenEnd": nTokenEnd,
                                    "lTokens": self.lTokens[nTokenStart:nTokenEnd+1],
                                    "lMorph": sAction.split("|")  if sAction else  [":HM"]
                                }
                                self.lTokens[nTokenStart]["nMultiStartTo"] = nTokenEnd
                                self.lTokens[nTokenEnd]["nMultiEndFrom"] = nTokenStart
                                self.lTokens[nTokenStart]["dMultiToken"] = dMultiToken
                                self.lTokens[nTokenEnd]["dMultiToken"] = dMultiToken
                                print(dMultiToken)
                            else:
                                echo("# error: unknown action at " + sLineId)
                        elif cActionType == ">":
                            if bDebug:
                                echo("    COND_BREAK")
                            break
                except Exception as e:

            if bUppercase and m.group(iGroup)[0:1].isupper():
                sNew = sNew.capitalize()
        else:
            sNew = m.expand(sRepl)
            sNew = sNew + " " * (nLen-len(sNew))
        return sText[0:m.start(iGroup)] + sNew + sText[m.end(iGroup):]

    def _tagAndPrepareTokenForRewriting (self, sAction, nTokenRewriteStart, nTokenRewriteEnd, nTokenOffset, nLastToken, bCaseSvty, bDebug):
        "text processor: rewrite tokens between <nTokenRewriteStart> and <nTokenRewriteEnd> position"
        if sAction == "*":
            # purge text
            if nTokenRewriteEnd - nTokenRewriteStart == 0:
                self.lTokens[nTokenRewriteStart]["bToRemove"] = True
            else:
                for i in range(nTokenRewriteStart, nTokenRewriteEnd+1):
                    self.lTokens[i]["bToRemove"] = True
        elif sAction == "␣":
            # merge tokens
            self.lTokens[nTokenRewriteStart]["nMergeUntil"] = nTokenRewriteEnd
        elif sAction.startswith("␣"):
            sAction = self._expand(sAction, nTokenOffset, nLastToken)
            self.lTokens[nTokenRewriteStart]["nMergeUntil"] = nTokenRewriteEnd
            self.lTokens[nTokenRewriteStart]["sMergedValue"] = sAction[1:]
        elif sAction == "_":
            # neutralized token
            if nTokenRewriteEnd - nTokenRewriteStart == 0:
                self.lTokens[nTokenRewriteStart]["sNewValue"] = "_"
            else:
                for i in range(nTokenRewriteStart, nTokenRewriteEnd+1):
                    self.lTokens[i]["sNewValue"] = "_"
        else:
            if sAction.startswith("="):
                sAction = getattr(gc_functions, sAction[1:])(self.lTokens, nTokenOffset, nLastToken)
            else:
                sAction = self._expand(sAction, nTokenOffset, nLastToken)
            bUppercase = bCaseSvty and self.lTokens[nTokenRewriteStart]["sValue"][0:1].isupper()
            if nTokenRewriteEnd - nTokenRewriteStart == 0:
                # one token
                if bUppercase:
                    sAction = sAction[0:1].upper() + sAction[1:]
                self.lTokens[nTokenRewriteStart]["sNewValue"] = sAction
            else:
                # several tokens
                lTokenValue = sAction.split("|")
                if len(lTokenValue) != (nTokenRewriteEnd - nTokenRewriteStart + 1):
                    if bDebug:
                        echo("Error. Text processor: number of replacements != number of tokens.")
                    return
                for i, sValue in zip(range(nTokenRewriteStart, nTokenRewriteEnd+1), lTokenValue):
                    if not sValue or sValue == "*":
                        self.lTokens[i]["bToRemove"] = True