Grammalecte  Diff

    def __init__ (self, sText):
        self.sText = sText
        self.sText0 = sText
        self.sSentence = ""
        self.sSentence0 = ""
        self.nOffsetWithinParagraph = 0
        self.lTokens = []
        self.dTokenPos = {}         # {position: token}
        self.dTags = {}             # {position: tags}
        self.dError = {}            # {position: error}
        self.dSentenceError = {}    # {position: error} (for the current sentence only)
        self.dErrorPriority = {}    # {position: priority of the current error}

    def __str__ (self):
        s = "===== TEXT =====\n"
        s += "sentence: " + self.sSentence0 + "\n"
        s += "now:      " + self.sSentence  + "\n"
        for dToken in self.lTokens:
            s += '#{i}\t{nStart}:{nEnd}\t{sValue}\t{sType}'.format(**dToken)
            if "lMorph" in dToken:
                s += "\t" + str(dToken["lMorph"])
            if "aTags" in dToken:
                s += "\t" + str(dToken["aTags"])
            s += "\n"
        #for nPos, dToken in self.dTokenPos.items():
        #    s += "{}\t{}\n".format(nPos, dToken)
        return s

    def parse (self, sCountry="${country_default}", bDebug=False, dOptions=None, bContext=False, bFullInfo=False):
        "analyses <sText> and returns an iterable of errors or (with option <bFullInfo>) paragraphs errors and sentences with tokens and errors"
        #sText = unicodedata.normalize("NFC", sText)
        dOpt = dOptions or gc_options.dOptions
        bShowRuleId = gc_options.dOptions.get('idrule', False)
        # parse paragraph
        try:
            self.parseText(self.sText, self.sText0, True, 0, sCountry, dOpt, bShowRuleId, bDebug, bContext)
        except:
            raise
        self.lTokens = None
        self.lTokens0 = None
        if bFullInfo:
            lParagraphErrors = list(self.dError.values())
            lSentences = []
            self.dSentenceError.clear()
        # parse sentences
        sText = self._getCleanText()
        for iStart, iEnd in text.getSentenceBoundaries(sText):
            if 4 < (iEnd - iStart) < 2000:
                try:
                    self.sSentence = sText[iStart:iEnd]
                    self.sSentence0 = self.sText0[iStart:iEnd]
                    self.nOffsetWithinParagraph = iStart
                    self.lTokens = list(_oTokenizer.genTokens(self.sSentence, True))
                    self.dTokenPos = { dToken["nStart"]: dToken  for dToken in self.lTokens  if dToken["sType"] != "INFO" }
                    if bFullInfo:
                        self.lTokens0 = list(self.lTokens)  # the list of tokens is duplicated, to keep tokens from being deleted when analysis
                    self.parseText(self.sSentence, self.sSentence0, False, iStart, sCountry, dOpt, bShowRuleId, bDebug, bContext)
                    if bFullInfo:
                        for dToken in self.lTokens0:
                            if dToken["sType"] == "WORD":
                                dToken["bValidToken"] = _oSpellChecker.isValidToken(dToken["sValue"])


                            if "lMorph" not in dToken:
                                dToken["lMorph"] = _oSpellChecker.getMorph(dToken["sValue"])
                            _oSpellChecker.setLabelsOnToken(dToken)
                        lSentences.append({
                            "nStart": iStart,
                            "nEnd": iEnd,
                            "sSentence": self.sSentence0,
                            "lTokens": self.lTokens0,
                            "lGrammarErrors": list(self.dSentenceError.values())
                        })
                        self.dSentenceError.clear()
                except:
                    raise
        if bFullInfo:
            # Grammar checking and sentence analysis
            return lParagraphErrors, lSentences
        else:

                self.sText = sText
            else:
                self.sSentence = sText

    def update (self, sSentence, bDebug=False):
        "update <sSentence> and retokenize"
        self.sSentence = sSentence
        lNewTokens = list(_oTokenizer.genTokens(sSentence, True))
        for dToken in lNewTokens:
            if "lMorph" in self.dTokenPos.get(dToken["nStart"], {}):
                dToken["lMorph"] = self.dTokenPos[dToken["nStart"]]["lMorph"]
            if "aTags" in self.dTokenPos.get(dToken["nStart"], {}):
                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 _getNextPointers (self, dToken, dGraph, dPointer, bDebug=False):
        "generator: return nodes where <dToken> “values” match <dNode> arcs"
        dNode = dGraph[dPointer["iNode"]]

            dPointer2 = { "iToken1": iToken1, "iNode": dNode["<>"], "bKeep": True }
            yield from self._getNextPointers(dToken, dGraph, dPointer2, bDebug)

    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"
        lPointer = []
        bTagAndRewrite = False
        for iToken, dToken in enumerate(self.lTokens):
            if bDebug:
                echo("TOKEN: " + dToken["sValue"])
            # check arcs for each existing pointer
            lNextPointer = []
            for dPointer in lPointer:
                lNextPointer.extend(self._getNextPointers(dToken, dGraph, dPointer, bDebug))
            lPointer = lNextPointer

                    # 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, "!", "",                            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 "bImmune" not in self.lTokens[nTokenErrorStart]:
                                    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(sWhat, 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(sWhat, nTokenStart, nTokenEnd, nTokenOffset, nLastToken, eAct[2], bDebug)
                                bChange = True
                                if bDebug:
                                    echo("    TEXT_PROCESSOR: [{}:{}]  > {}".format(self.lTokens[nTokenStart]["sValue"], self.lTokens[nTokenEnd]["sValue"], sWhat))
                            elif cActionType == "=":
                                # disambiguation
                                getattr(gc_functions, sWhat)(self.lTokens, nTokenOffset, nLastToken)
                                if bDebug:
                                    echo("    DISAMBIGUATOR: ({})  [{}:{}]".format(sWhat, 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(sWhat.split("|"))
                                    else:
                                        self.lTokens[i]["aTags"] = set(sWhat.split("|"))
                                if bDebug:
                                    echo("    TAG: {} >  [{}:{}]".format(sWhat, self.lTokens[nTokenStart]["sValue"], self.lTokens[nTokenEnd]["sValue"]))
                                for sTag in sWhat.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]
                                if nTokenEnd - nTokenStart == 0:
                                    self.lTokens[nTokenStart]["bImmune"] = True
                                    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]["bImmune"] = True
                                        nErrorStart = self.nOffsetWithinParagraph + self.lTokens[i]["nStart"]
                                        if nErrorStart in self.dError:
                                            del self.dError[nErrorStart]
                            else:
                                echo("# error: unknown action at " + sLineId)
                        elif cActionType == ">":
                            if bDebug:
                                echo("    COND_BREAK")

        if _bWriterError:
            return self._createErrorForWriter(nStart, nEnd - nStart, sRuleId, sOption, sMessage, lSugg, sURL)
        return self._createErrorAsDict(nStart, nEnd, sLineId, sRuleId, sOption, sMessage, lSugg, sURL, bContext)

    def _createErrorFromTokens (self, sSugg, nTokenOffset, nLastToken, iFirstToken, nStart, nEnd, sLineId, sRuleId, bCaseSvty, sMsg, sURL, bShowRuleId, sOption, bContext):
        # suggestions
        if sSugg[0:1] == "=":
            sSugg = getattr(gc_functions, sSugg[1:])(self.lTokens, nTokenOffset, nLastToken)
            lSugg = sSugg.split("|")  if sSugg  else []
        elif sSugg == "_":
            lSugg = []
        else:
            lSugg = self._expand(sSugg, nTokenOffset, nLastToken).split("|")
        if bCaseSvty and lSugg and self.lTokens[iFirstToken]["sValue"][0:1].isupper():
            lSugg = list(map(lambda s: s.upper(), lSugg))  if self.sSentence[nStart:nEnd].isupper()  else list(map(lambda s: s[0:1].upper()+s[1:], lSugg))
        # Message
        sMessage = getattr(gc_functions, sMsg[1:])(self.lTokens, nTokenOffset, nLastToken)  if sMsg[0:1] == "="  else self._expand(sMsg, nTokenOffset, nLastToken)
        if bShowRuleId:
            sMessage += "  #" + sLineId + " / " + sRuleId
        #
        if _bWriterError:
            return self._createErrorForWriter(nStart, nEnd - nStart, sRuleId, sOption, sMessage, lSugg, sURL)
        return self._createErrorAsDict(nStart, nEnd, sLineId, sRuleId, sOption, sMessage, lSugg, sURL, bContext)

            dErr['sBefore'] = self.sText0[max(0,nStart-80):nStart]
            dErr['sAfter'] = self.sText0[nEnd:nEnd+80]
        return dErr

    def _expand (self, sText, nTokenOffset, nLastToken):
        for m in re.finditer(r"\\(-?[0-9]+)", sText):
            if m.group(1)[0:1] == "-":
                sText = sText.replace(m.group(0), self.lTokens[nLastToken+int(m.group(1))+1]["sValue"])
            else:
                sText = sText.replace(m.group(0), self.lTokens[nTokenOffset+int(m.group(1))]["sValue"])
        return sText

    def rewriteText (self, sText, sRepl, iGroup, m, bUppercase):
        "text processor: write <sRepl> in <sText> at <iGroup> position"
        nLen = m.end(iGroup) - m.start(iGroup)
        if sRepl == "*":
            sNew = " " * nLen

        return sText[0:m.start(iGroup)] + sNew + sText[m.end(iGroup):]

    def _tagAndPrepareTokenForRewriting (self, sWhat, nTokenRewriteStart, nTokenRewriteEnd, nTokenOffset, nLastToken, bCaseSvty, bDebug):
        "text processor: rewrite tokens between <nTokenRewriteStart> and <nTokenRewriteEnd> position"
        if sWhat == "*":
            # 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 sWhat == "␣":
            # merge tokens
            self.lTokens[nTokenRewriteStart]["nMergeUntil"] = nTokenRewriteEnd
        elif sWhat == "_":
            # neutralized token
            if nTokenRewriteEnd - nTokenRewriteStart == 0:
                self.lTokens[nTokenRewriteStart]["sNewValue"] = "_"
            else:
                for i in range(nTokenRewriteStart, nTokenRewriteEnd+1):
                    self.lTokens[i]["sNewValue"] = "_"
        else:
            if sWhat.startswith("="):
                sWhat = getattr(gc_functions, sWhat[1:])(self.lTokens, nTokenOffset, nLastToken)
            else:
                sWhat = self._expand(sWhat, nTokenOffset, nLastToken)
            bUppercase = bCaseSvty and self.lTokens[nTokenRewriteStart]["sValue"][0:1].isupper()
            if nTokenRewriteEnd - nTokenRewriteStart == 0:
                # one token
                if bUppercase:
                    sWhat = sWhat[0:1].upper() + sWhat[1:]
                self.lTokens[nTokenRewriteStart]["sNewValue"] = sWhat
            else:
                # several tokens
                lTokenValue = sWhat.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
                    else:
                        if bUppercase:
                            sValue = sValue[0:1].upper() + sValue[1:]
                        self.lTokens[i]["sNewValue"] = sValue

    def rewriteFromTags (self, bDebug=False):
        "rewrite the sentence, modify tokens, purge the token list"
        if bDebug:
            echo("REWRITE")
        lNewTokens = []
        lNewTokens0 = []
        nMergeUntil = 0
        dTokenMerger = {}
        for iToken, dToken in enumerate(self.lTokens):
            bKeepToken = True
            if dToken["sType"] != "INFO":
                if nMergeUntil and iToken <= nMergeUntil:
                    # token to merge
                    dTokenMerger["sValue"] += " " * (dToken["nStart"] - dTokenMerger["nEnd"]) + dToken["sValue"]
                    dTokenMerger["nEnd"] = dToken["nEnd"]
                    if bDebug:
                        echo("  MERGED TOKEN: " + dTokenMerger["sValue"])
                    dToken["bMerged"] = True
                    bKeepToken = False
                if "nMergeUntil" in dToken:
                    # first token to be merge with
                    if iToken > nMergeUntil: # this token is not to be merged with a previous token
                        dTokenMerger = dToken
                    if dToken["nMergeUntil"] > nMergeUntil:
                        nMergeUntil = dToken["nMergeUntil"]
                    del dToken["nMergeUntil"]
                elif "bToRemove" in dToken:
                    # deletion required
                    if bDebug:
                        echo("  REMOVED: " + dToken["sValue"])
                    self.sSentence = self.sSentence[:dToken["nStart"]] + " " * (dToken["nEnd"] - dToken["nStart"]) + self.sSentence[dToken["nEnd"]:]
                    bKeepToken = False
            #
            if bKeepToken:
                lNewTokens.append(dToken)
                if "sNewValue" in dToken:
                    # rewrite token and sentence
                    if bDebug:
                        echo(dToken["sValue"] + " -> " + dToken["sNewValue"])
                    dToken["sRealValue"] = dToken["sValue"]
                    dToken["sValue"] = dToken["sNewValue"]
                    nDiffLen = len(dToken["sRealValue"]) - len(dToken["sNewValue"])
                    sNewRepl = (dToken["sNewValue"] + " " * nDiffLen)  if nDiffLen >= 0  else dToken["sNewValue"][:len(dToken["sRealValue"])]
                    self.sSentence = self.sSentence[:dToken["nStart"]] + sNewRepl + self.sSentence[dToken["nEnd"]:]
                    del dToken["sNewValue"]
            else:
                try:
                    del self.dTokenPos[dToken["nStart"]]
                except KeyError:
                    echo(self)
                    echo(dToken)
        if bDebug:
            echo("  TEXT REWRITED: " + self.sSentence)
        self.lTokens.clear()
        self.lTokens = lNewTokens