Grammalecte  Diff

#### GRAMMAR CHECKING ENGINE PLUGIN: Suggestion mechanisms

from . import conj
from . import mfsp
from . import phonet


## Verbs

def splitVerb (sVerb):
    "renvoie le verbe et les pronoms séparément"
    iRight = sVerb.rfind("-")
    sSuffix = sVerb[iRight:]
    sVerb = sVerb[:iRight]
    if sVerb.endswith(("-t", "-le", "-la", "-les")):
        iRight = sVerb.rfind("-")
        sSuffix = sVerb[iRight:] + sSuffix
        sVerb = sVerb[:iRight]
    return sVerb, sSuffix


def suggVerb (sFlex, sWho, funcSugg2=None):
def suggVerb (sFlex, sWho, funcSugg2=None, bVC=False):
    "change <sFlex> conjugation according to <sWho>"
    if bVC:
        sFlex, sSfx = splitVerb(sFlex)
    aSugg = set()
    for sStem in stem(sFlex):
    for sStem in _oSpellChecker.getLemma(sFlex):
        tTags = conj._getTags(sStem)
        if tTags:
            # we get the tense
            aTense = set()
            for sMorph in _dAnalyses.get(sFlex, []): # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
                for m in re.finditer(">"+sStem+" .*?(:(?:Y|I[pqsf]|S[pq]|K|P))", sMorph):
            for sMorph in _oSpellChecker.getMorph(sFlex):
                for m in re.finditer(">"+sStem+"/.*?(:(?:Y|I[pqsf]|S[pq]|K|P))", sMorph):
                    # stem must be used in regex to prevent confusion between different verbs (e.g. sauras has 2 stems: savoir and saurer)
                    if m:
                        if m.group(1) == ":Y":
                            aTense.add(":Ip")
                            aTense.add(":Iq")
                            aTense.add(":Is")
                        elif m.group(1) == ":P":
                            aTense.add(":Ip")
                        else:
                            aTense.add(m.group(1))
            for sTense in aTense:
                if sWho == ":1ś" and not conj._hasConjWithTags(tTags, sTense, ":1ś"):
                    sWho = ":1s"
                if conj._hasConjWithTags(tTags, sTense, sWho):
                    aSugg.add(conj._getConjWithTags(sStem, tTags, sTense, sWho))
    if funcSugg2:
        aSugg2 = funcSugg2(sFlex)
        if aSugg2:
            aSugg.add(aSugg2)
    if aSugg:
        if bVC:
            aSugg = list(map(lambda sSug: sSug + sSfx, aSugg))
        return "|".join(aSugg)
    return ""


def suggVerbPpas (sFlex, sWhat=None):
def suggVerbPpas (sFlex, sPattern=None):
    "suggest past participles for <sFlex>"
    aSugg = set()
    for sStem in stem(sFlex):
    for sStem in _oSpellChecker.getLemma(sFlex):
        tTags = conj._getTags(sStem)
        if tTags:
            if not sWhat:
            if not sPattern:
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q2"))
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q3"))
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q4"))
                aSugg.discard("")
            elif sWhat == ":m:s":
            elif sPattern == ":m:s":
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
            elif sWhat == ":m:p":
            elif sPattern == ":m:p":
                if conj._hasConjWithTags(tTags, ":PQ", ":Q2"):
                    aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q2"))
                else:
                    aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
            elif sWhat == ":f:s":
            elif sPattern == ":f:s":
                if conj._hasConjWithTags(tTags, ":PQ", ":Q3"):
                    aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q3"))
                else:
                    aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
            elif sWhat == ":f:p":
            elif sPattern == ":f:p":
                if conj._hasConjWithTags(tTags, ":PQ", ":Q4"):
                    aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q4"))
                else:
                    aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
            elif sWhat == ":s":
            elif sPattern == ":s":
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q3"))
                aSugg.discard("")
            elif sWhat == ":p":
            elif sPattern == ":p":
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q2"))
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q4"))
                aSugg.discard("")
            else:
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
    if aSugg:
        return "|".join(aSugg)
    return ""


def suggVerbTense (sFlex, sTense, sWho):
    "change <sFlex> to a verb according to <sTense> and <sWho>"
    aSugg = set()
    for sStem in stem(sFlex):
    for sStem in _oSpellChecker.getLemma(sFlex):
        if conj.hasConj(sStem, sTense, sWho):
            aSugg.add(conj.getConj(sStem, sTense, sWho))
    if aSugg:
        return "|".join(aSugg)
    return ""


def suggVerbImpe (sFlex):
    "change <sFlex> to a verb at imperative form"
    aSugg = set()
    for sStem in stem(sFlex):
    for sStem in _oSpellChecker.getLemma(sFlex):
        tTags = conj._getTags(sStem)
        if tTags:
            if conj._hasConjWithTags(tTags, ":E", ":2s"):
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":E", ":2s"))
            if conj._hasConjWithTags(tTags, ":E", ":1p"):
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":E", ":1p"))
            if conj._hasConjWithTags(tTags, ":E", ":2p"):
                aSugg.add(conj._getConjWithTags(sStem, tTags, ":E", ":2p"))
    if aSugg:
        return "|".join(aSugg)
    return ""


def suggVerbInfi (sFlex):
    "returns infinitive forms of <sFlex>"
    return "|".join([ sStem  for sStem in stem(sFlex)  if conj.isVerb(sStem) ])
    return "|".join([ sStem  for sStem in _oSpellChecker.getLemma(sFlex)  if conj.isVerb(sStem) ])


_dQuiEst = { "je": ":1s", "j’": ":1s", "j’en": ":1s", "j’y": ":1s", \
             "tu": ":2s", "il": ":3s", "on": ":3s", "elle": ":3s", "nous": ":1p", "vous": ":2p", "ils": ":3p", "elles": ":3p" }
_lIndicatif = [":Ip", ":Iq", ":Is", ":If"]
_lSubjonctif = [":Sp", ":Sq"]

def suggVerbMode (sFlex, cMode, sSuj):
    "returns other conjugations of <sFlex> acconding to <cMode> and <sSuj>"
    if cMode == ":I":
        lMode = _lIndicatif
    elif cMode == ":S":
        lMode = _lSubjonctif
    elif cMode.startswith((":I", ":S")):
        lMode = [cMode]
    else:
        return ""
    sWho = _dQuiEst.get(sSuj.lower(), None)
    if not sWho:
        if sSuj[0:1].islower(): # pas un pronom, ni un nom propre
            return ""
        sWho = ":3s"
    aSugg = set()
    for sStem in stem(sFlex):
    for sStem in _oSpellChecker.getLemma(sFlex):
        tTags = conj._getTags(sStem)
        if tTags:
            for sTense in lMode:
                if conj._hasConjWithTags(tTags, sTense, sWho):
                    aSugg.add(conj._getConjWithTags(sStem, tTags, sTense, sWho))
    if aSugg:
        return "|".join(aSugg)
    return ""


## Nouns and adjectives

def suggPlur (sFlex, sWordToAgree=None):
    "returns plural forms assuming sFlex is singular"
    if sWordToAgree:
        lMorph = _oSpellChecker.getMorph(sFlex)
        if sWordToAgree not in _dAnalyses and not _storeMorphFromFSA(sWordToAgree):
        if not lMorph:
            return ""
        sGender = cr.getGender(_dAnalyses.get(sWordToAgree, []))
        sGender = cr.getGender(lMorph)
        if sGender == ":m":
            return suggMasPlur(sFlex)
        elif sGender == ":f":
            return suggFemPlur(sFlex)
    aSugg = set()
    if "-" not in sFlex:
        if sFlex.endswith("l"):

    if aSugg:
        return "|".join(aSugg)
    return ""


def suggMasSing (sFlex, bSuggSimil=False):
    "returns masculine singular forms"
    # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
    aSugg = set()
    for sMorph in _dAnalyses.get(sFlex, []):
    for sMorph in _oSpellChecker.getMorph(sFlex):
        if not ":V" in sMorph:
            # not a verb
            if ":m" in sMorph or ":e" in sMorph:
                aSugg.add(suggSing(sFlex))
            else:
                sStem = cr.getLemmaOfMorph(sMorph)
                if mfsp.isFemForm(sStem):

    if aSugg:
        return "|".join(aSugg)
    return ""


def suggMasPlur (sFlex, bSuggSimil=False):
    "returns masculine plural forms"
    # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
    aSugg = set()
    for sMorph in _dAnalyses.get(sFlex, []):
    for sMorph in _oSpellChecker.getMorph(sFlex):
        if not ":V" in sMorph:
            # not a verb
            if ":m" in sMorph or ":e" in sMorph:
                aSugg.add(suggPlur(sFlex))
            else:
                sStem = cr.getLemmaOfMorph(sMorph)
                if mfsp.isFemForm(sStem):

    if aSugg:
        return "|".join(aSugg)
    return ""


def suggFemSing (sFlex, bSuggSimil=False):
    "returns feminine singular forms"
    # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
    aSugg = set()
    for sMorph in _dAnalyses.get(sFlex, []):
    for sMorph in _oSpellChecker.getMorph(sFlex):
        if not ":V" in sMorph:
            # not a verb
            if ":f" in sMorph or ":e" in sMorph:
                aSugg.add(suggSing(sFlex))
            else:
                sStem = cr.getLemmaOfMorph(sMorph)
                if mfsp.isFemForm(sStem):

    if aSugg:
        return "|".join(aSugg)
    return ""


def suggFemPlur (sFlex, bSuggSimil=False):
    "returns feminine plural forms"
    # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
    aSugg = set()
    for sMorph in _dAnalyses.get(sFlex, []):
    for sMorph in _oSpellChecker.getMorph(sFlex):
        if not ":V" in sMorph:
            # not a verb
            if ":f" in sMorph or ":e" in sMorph:
                aSugg.add(suggPlur(sFlex))
            else:
                sStem = cr.getLemmaOfMorph(sMorph)
                if mfsp.isFemForm(sStem):

            aSugg.add(e)
    if aSugg:
        return "|".join(aSugg)
    return ""


def hasFemForm (sFlex):
    "return True if there is a feminine form of <sFlex>"
    for sStem in stem(sFlex):
    for sStem in _oSpellChecker.getLemma(sFlex):
        if mfsp.isFemForm(sStem) or conj.hasConj(sStem, ":PQ", ":Q3"):
            return True
    if phonet.hasSimil(sFlex, ":f"):
        return True
    return False


def hasMasForm (sFlex):
    "return True if there is a masculine form of <sFlex>"
    for sStem in stem(sFlex):
    for sStem in _oSpellChecker.getLemma(sFlex):
        if mfsp.isFemForm(sStem) or conj.hasConj(sStem, ":PQ", ":Q1"):
            # what has a feminine form also has a masculine form
            return True
    if phonet.hasSimil(sFlex, ":m"):
        return True
    return False


def switchGender (sFlex, bPlur=None):
    # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
    "return feminine or masculine form(s) of <sFlex>"
    aSugg = set()
    if bPlur == None:
        for sMorph in _dAnalyses.get(sFlex, []):
        for sMorph in _oSpellChecker.getMorph(sFlex):
            if ":f" in sMorph:
                if ":s" in sMorph:
                    aSugg.add(suggMasSing(sFlex))
                elif ":p" in sMorph:
                    aSugg.add(suggMasPlur(sFlex))
            elif ":m" in sMorph:
                if ":s" in sMorph:
                    aSugg.add(suggFemSing(sFlex))
                elif ":p" in sMorph:
                    aSugg.add(suggFemPlur(sFlex))
                else:
                    aSugg.add(suggFemSing(sFlex))
                    aSugg.add(suggFemPlur(sFlex))
    elif bPlur:
        for sMorph in _dAnalyses.get(sFlex, []):
        for sMorph in _oSpellChecker.getMorph(sFlex):
            if ":f" in sMorph:
                aSugg.add(suggMasPlur(sFlex))
            elif ":m" in sMorph:
                aSugg.add(suggFemPlur(sFlex))
    else:
        for sMorph in _dAnalyses.get(sFlex, []):
        for sMorph in _oSpellChecker.getMorph(sFlex):
            if ":f" in sMorph:
                aSugg.add(suggMasSing(sFlex))
            elif ":m" in sMorph:
                aSugg.add(suggFemSing(sFlex))
    if aSugg:
        return "|".join(aSugg)
    return ""


def switchPlural (sFlex):
    # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
    "return plural or singular form(s) of <sFlex>"
    aSugg = set()
    for sMorph in _dAnalyses.get(sFlex, []):
    for sMorph in _oSpellChecker.getMorph(sFlex):
        if ":s" in sMorph:
            aSugg.add(suggPlur(sFlex))
        elif ":p" in sMorph:
            aSugg.add(suggSing(sFlex))
    if aSugg:
        return "|".join(aSugg)
    return ""


def hasSimil (sWord, sPattern=None):
    "return True if there is words phonetically similar to <sWord> (according to <sPattern> if required)"
    return phonet.hasSimil(sWord, sPattern)


def suggSimil (sWord, sPattern=None, bSubst=False):
def suggSimil (sWord, sPattern=None, bSubst=False, bVC=False):
    "return list of words phonetically similar to sWord and whom POS is matching sPattern"
    # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
    if bVC:
        sWord, sSfx = splitVerb(sWord)
    aSugg = phonet.selectSimil(sWord, sPattern)
    for sMorph in _dAnalyses.get(sWord, []):
    for sMorph in _oSpellChecker.getMorph(sWord):
        aSugg.update(conj.getSimil(sWord, sMorph, bSubst))
        break
    if aSugg:
        if bVC:
            aSugg = list(map(lambda sSug: sSug + sSfx, aSugg))
        return "|".join(aSugg)
    return ""


def suggCeOrCet (sWord):
    "suggest “ce” or “cet” or both according to the first letter of <sWord>"
    if re.match("(?i)[aeéèêiouyâîï]", sWord):
        return "cet"
    if sWord[0:1] == "h" or sWord[0:1] == "H":
        return "ce|cet"
    return "ce"


def suggLesLa (sWord):
    # we don’t check if word exists in _dAnalyses, for it is assumed it has been done before
    if any( ":p" in sMorph  for sMorph in _dAnalyses.get(sWord, []) ):
    "suggest “les” or “la” according to <sWord>"
    if any( ":p" in sMorph  for sMorph in _oSpellChecker.getMorph(sWord) ):
        return "les|la"
    return "la"


_zBinary = re.compile("^[01]+$")

def formatNumber (s):
    "add spaces or hyphens to big numbers"
    nLen = len(s)
    if nLen < 4:
        return s
    sRes = ""
    # nombre ordinaire
    nEnd = nLen
    while nEnd > 0:

    elif nLen == 9 and s.startswith("0"):
        sRes += "|" + s[0:3] + " " + s[3:5] + " " + s[5:7] + " " + s[7:9]                   # fixe belge 1
        sRes += "|" + s[0:2] + " " + s[2:5] + " " + s[5:7] + " " + s[7:9]                   # fixe belge 2
    return sRes


def formatNF (s):
    "typography: format NF reference (norme française)"
    try:
        m = re.match("NF[  -]?(C|E|P|Q|S|X|Z|EN(?:[  -]ISO|))[  -]?([0-9]+(?:[/‑-][0-9]+|))", s)
        if not m:
            return ""
        return "NF " + m.group(1).upper().replace(" ", " ").replace("-", " ") + " " + m.group(2).replace("/", "‑").replace("-", "‑")
    except:
        traceback.print_exc()
        return "# erreur #"


def undoLigature (c):
    "typography: split ligature character <c> in several chars"
    if c == "fi":
        return "fi"
    elif c == "fl":
        return "fl"
    elif c == "ff":
        return "ff"
    elif c == "ffi":

_xNormalizedCharsForInclusiveWriting = str.maketrans({
    '(': '_',  ')': '_',
    '.': '_',  '·': '_',
    '–': '_',  '—': '_',
    '/': '_'
 })
})


def normalizeInclusiveWriting (sToken):
    "typography: replace word separators used in inclusive writing by underscore (_)"
    return sToken.translate(_xNormalizedCharsForInclusiveWriting)