Grammalecte  Diff

            if nDist <= self.nDistLimit:
                if nDist not in self.dSugg:
                    self.dSugg[nDist] = []
                self.dSugg[nDist].append(sSugg)
                self.aSugg.add(sSugg)
                if nDist < self.nMinDist:
                    self.nMinDist = nDist
                self.nDistLimit = min(self.nDistLimit, self.nMinDist+1)

    def getSuggestions (self, nSuggLimit=10):
        "return a list of suggestions"
        if self.dSugg[0]:
            # we sort the better results with the original word
            self.dSugg[0].sort(key=lambda sSugg: st.distanceDamerauLevenshtein(self.sWord, sSugg))
        lRes = self.dSugg.pop(0)

    def _initBinary (self):
        "initialize with binary structure file"
        if self.by[0:17] != b"/grammalecte-fsa/":
            raise TypeError("# Error. Not a grammalecte-fsa binary dictionary. Header: {}".format(self.by[0:9]))
        if not(self.by[17:18] == b"1" or self.by[17:18] == b"2" or self.by[17:18] == b"3"):
            raise ValueError("# Error. Unknown dictionary version: {}".format(self.by[17:18]))
        try:
            byHeader, byInfo, byValues, by2grams, byDic = self.by.split(b"\0\0\0\0", 4)
        except Exception:
            raise Exception

        self.nCompressionMethod = int(self.by[17:18].decode("utf-8"))
        self.sHeader = byHeader.decode("utf-8")
        self.lArcVal = byValues.decode("utf-8").split("\t")
        self.nArcVal = len(self.lArcVal)
        self.byDic = byDic
        self.a2grams = set(by2grams.decode("utf-8").split("\t"))

        l = byInfo.decode("utf-8").split("//")
        self.sLangCode = l.pop(0)
        self.sLangName = l.pop(0)
        self.sDicName = l.pop(0)
        self.sDate = l.pop(0)
        self.nChar = int(l.pop(0))
        self.nBytesArc = int(l.pop(0))
        self.nBytesNodeAddress = int(l.pop(0))

        self.nBytesOffset = 1 # version 3

    def _initJSON (self, oJSON):
        "initialize with a JSON text file"
        self.__dict__.update(oJSON)
        self.byDic = binascii.unhexlify(self.sByDic)
        self.dCharVal = { v: k  for k, v in self.dChar.items() }
        self.a2grams = set(self.l2grams)

    def getInfo (self):
        "return string about the IBDAWG"
        return  "  Language: {0.sLangName}   Lang code: {0.sLangCode}   Dictionary name: {0.sDicName}" \
                "  Compression method: {0.nCompressionMethod:>2}   Date: {0.sDate}   Stemming: {0.cStemming}FX\n" \
                "  Arcs values:  {0.nArcVal:>10,} = {0.nChar:>5,} characters,  {0.nAff:>6,} affixes,  {0.nTag:>6,} tags\n" \
                "  Dictionary: {0.nEntry:>12,} entries,    {0.nNode:>11,} nodes,   {0.nArc:>11,} arcs\n" \

                "nCompressionMethod": self.nCompressionMethod,
                "nBytesArc": self.nBytesArc,
                "nBytesNodeAddress": self.nBytesNodeAddress,
                "nBytesOffset": self.nBytesOffset,
                # JavaScript is a pile of shit, so Mozilla’s JS parser don’t like file bigger than 4 Mb!
                # So, if necessary, we use an hexadecimal string, that we will convert later in Firefox’s extension.
                # https://github.com/mozilla/addons-linter/issues/1361
                "sByDic": self.byDic.hex()  if bBinaryDictAsHexString  else [ e  for e in self.byDic ],
                "l2grams": list(self.a2grams)
            }, ensure_ascii=False))
            if bInJSModule:
                hDst.write(";\n\nexports.dictionary = dictionary;\n")

    def isValidToken (self, sToken):
        "checks if <sToken> is valid (if there is hyphens in <sToken>, <sToken> is split, each part is checked)"
        sToken = cp.spellingNormalization(sToken)

        if nDist > oSuggResult.nDistLimit:
            return
        cCurrent = sRemain[0:1]
        for cChar, jAddr in self._getCharArcs(iAddr):
            if cChar in cp.d1to1.get(cCurrent, cCurrent):
                self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, jAddr, sNewWord+cChar)
            elif not bAvoidLoop:
                if nMaxHardRepl and self.isNgramsOK(cChar+sRemain[1:2]):
                    self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl-1, nMaxJump, nDist+1, nDeep+1, jAddr, sNewWord+cChar, True)
                if nMaxJump:
                    self._suggest(oSuggResult, sRemain, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump-1, nDist+1, nDeep+1, jAddr, sNewWord+cChar, True)
        if not bAvoidLoop: # avoid infinite loop
            if len(sRemain) > 1:
                if cCurrent == sRemain[1:2]:
                    # same char, we remove 1 char without adding 1 to <sNewWord>
                    self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord)
                else:
                    # switching chars
                    if nMaxSwitch and self.isNgramsOK(sNewWord[-1:]+sRemain[1:2]) and self.isNgramsOK(sRemain[1:2]+sRemain[0:1]):
                        self._suggest(oSuggResult, sRemain[1:2]+sRemain[0:1]+sRemain[2:], nMaxSwitch-1, nMaxDel, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, True)
                    # delete char
                    if nMaxDel and self.isNgramsOK(sNewWord[-1:]+sRemain[1:2]):
                        self._suggest(oSuggResult, sRemain[1:], nMaxSwitch, nMaxDel-1, nMaxHardRepl, nMaxJump, nDist+1, nDeep+1, iAddr, sNewWord, True)
                # Phonetic replacements
                for sRepl in cp.get1toXReplacement(sNewWord[-1:], cCurrent, sRemain[1:2]):
                    self._suggest(oSuggResult, sRepl + sRemain[1:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
                for sRepl in cp.d2toX.get(sRemain[0:2], ()):
                    self._suggest(oSuggResult, sRepl + sRemain[2:], nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
            # end of word
            if len(sRemain) == 2:
                for sRepl in cp.dFinal2.get(sRemain, ()):
                    self._suggest(oSuggResult, sRepl, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)
            elif len(sRemain) == 1:
                self._suggest(oSuggResult, "", nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True) # remove last char and go on
                for sRepl in cp.dFinal1.get(sRemain, ()):
                    self._suggest(oSuggResult, sRepl, nMaxSwitch, nMaxDel, nMaxHardRepl, nMaxJump, nDist, nDeep+1, iAddr, sNewWord, True)

    def isNgramsOK (self, sChars):
        if len(sChars) != 2:
            return True
        return sChars in self.a2grams

    #@timethis
    def suggest2 (self, sWord, nSuggLimit=10):
        "returns a set of suggestions for <sWord>"
        sWord = cp.spellingNormalization(sWord)
        sPfx, sWord, sSfx = cp.cut(sWord)
        oSuggResult = SuggResult(sWord)
        self._suggest2(oSuggResult)