#### 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, bVC=False):
"change <sFlex> conjugation according to <sWho>"
if bVC:
sFlex, sSfx = splitVerb(sFlex)
aSugg = set()
for sStem in _oSpellChecker.getLemma(sFlex):
tTags = conj._getTags(sStem)
if tTags:
# we get the tense
aTense = set()
for sMorph in _oSpellChecker.getMorph(sFlex):
for m in re.finditer(">"+sStem+"/.*?(:(?:Y|I[pqsf]|S[pq]|K|P|Q))", 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" or m.group(1) == ":Q":
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, sPattern=None):
"suggest past participles for <sFlex>"
aSugg = set()
for sStem in _oSpellChecker.getLemma(sFlex):
tTags = conj._getTags(sStem)
if tTags:
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 sPattern == ":m:s":
aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
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 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 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 sPattern == ":s":
aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q1"))
aSugg.add(conj._getConjWithTags(sStem, tTags, ":PQ", ":Q3"))
aSugg.discard("")
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 _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, bVC=False):
"change <sFlex> to a verb at imperative form"
if bVC:
sFlex, sSfx = splitVerb(sFlex)
aSugg = set()
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:
if bVC:
aSugg = list(map(lambda sSug: sSug + sSfx, aSugg))
return "|".join(aSugg)
return ""
def suggVerbInfi (sFlex):
"returns infinitive forms of <sFlex>"
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 _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 not lMorph:
return ""
sGender = cr.getGender(lMorph)
if sGender == ":m":
return suggMasPlur(sFlex)
if sGender == ":f":
return suggFemPlur(sFlex)
aSugg = set()
if "-" not in sFlex:
if sFlex.endswith("l"):
if sFlex.endswith("al") and len(sFlex) > 2 and _oSpellChecker.isValid(sFlex[:-1]+"ux"):
aSugg.add(sFlex[:-1]+"ux")
if sFlex.endswith("ail") and len(sFlex) > 3 and _oSpellChecker.isValid(sFlex[:-2]+"ux"):
aSugg.add(sFlex[:-2]+"ux")
if _oSpellChecker.isValid(sFlex+"s"):
aSugg.add(sFlex+"s")
if _oSpellChecker.isValid(sFlex+"x"):
aSugg.add(sFlex+"x")
if mfsp.hasMiscPlural(sFlex):
aSugg.update(mfsp.getMiscPlural(sFlex))
if aSugg:
return "|".join(aSugg)
return ""
def suggSing (sFlex):
"returns singular forms assuming sFlex is plural"
if "-" in sFlex:
return ""
aSugg = set()
if sFlex.endswith("ux"):
if _oSpellChecker.isValid(sFlex[:-2]+"l"):
aSugg.add(sFlex[:-2]+"l")
if _oSpellChecker.isValid(sFlex[:-2]+"il"):
aSugg.add(sFlex[:-2]+"il")
if _oSpellChecker.isValid(sFlex[:-1]):
aSugg.add(sFlex[:-1])
if aSugg:
return "|".join(aSugg)
return ""
def suggMasSing (sFlex, bSuggSimil=False):
"returns masculine singular forms"
aSugg = set()
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):
aSugg.update(mfsp.getMasForm(sStem, False))
else:
# a verb
sVerb = cr.getLemmaOfMorph(sMorph)
if conj.hasConj(sVerb, ":PQ", ":Q1") and conj.hasConj(sVerb, ":PQ", ":Q3"):
# We also check if the verb has a feminine form.
# If not, we consider it’s better to not suggest the masculine one, as it can be considered invariable.
aSugg.add(conj.getConj(sVerb, ":PQ", ":Q1"))
if bSuggSimil:
for e in phonet.selectSimil(sFlex, ":m:[si]"):
aSugg.add(e)
if aSugg:
return "|".join(aSugg)
return ""
def suggMasPlur (sFlex, bSuggSimil=False):
"returns masculine plural forms"
aSugg = set()
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):
aSugg.update(mfsp.getMasForm(sStem, True))
else:
# a verb
sVerb = cr.getLemmaOfMorph(sMorph)
if conj.hasConj(sVerb, ":PQ", ":Q2"):
aSugg.add(conj.getConj(sVerb, ":PQ", ":Q2"))
elif conj.hasConj(sVerb, ":PQ", ":Q1"):
sSugg = conj.getConj(sVerb, ":PQ", ":Q1")
# it is necessary to filter these flexions, like “succédé” or “agi” that are not masculine plural.
if sSugg.endswith("s"):
aSugg.add(sSugg)
if bSuggSimil:
for e in phonet.selectSimil(sFlex, ":m:[pi]"):
aSugg.add(e)
if aSugg:
return "|".join(aSugg)
return ""
def suggFemSing (sFlex, bSuggSimil=False):
"returns feminine singular forms"
aSugg = set()
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):
aSugg.add(sStem)
else:
# a verb
sVerb = cr.getLemmaOfMorph(sMorph)
if conj.hasConj(sVerb, ":PQ", ":Q3"):
aSugg.add(conj.getConj(sVerb, ":PQ", ":Q3"))
if bSuggSimil:
for e in phonet.selectSimil(sFlex, ":f:[si]"):
aSugg.add(e)
if aSugg:
return "|".join(aSugg)
return ""
def suggFemPlur (sFlex, bSuggSimil=False):
"returns feminine plural forms"
aSugg = set()
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(sStem+"s")
else:
# a verb
sVerb = cr.getLemmaOfMorph(sMorph)
if conj.hasConj(sVerb, ":PQ", ":Q4"):
aSugg.add(conj.getConj(sVerb, ":PQ", ":Q4"))
if bSuggSimil:
for e in phonet.selectSimil(sFlex, ":f:[pi]"):
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 _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 _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):
"return feminine or masculine form(s) of <sFlex>"
aSugg = set()
if bPlur is None:
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 _oSpellChecker.getMorph(sFlex):
if ":f" in sMorph:
aSugg.add(suggMasPlur(sFlex))
elif ":m" in sMorph:
aSugg.add(suggFemPlur(sFlex))
else:
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):
"return plural or singular form(s) of <sFlex>"
aSugg = set()
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, bVC=False):
"return list of words phonetically similar to sWord and whom POS is matching sPattern"
if bVC:
sWord, sSfx = splitVerb(sWord)
aSugg = phonet.selectSimil(sWord, sPattern)
if not aSugg or not bSubst:
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] in "hH":
return "ce|cet"
return "ce"
def suggLesLa (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 (sNumber):
"add spaces or hyphens to big numbers"
nLen = len(sNumber)
if nLen < 4:
return sNumber
sRes = ""
if "," not in sNumber:
# nombre entier
sRes = _formatNumber(sNumber, 3)
# binaire
if _zBinary.search(sNumber):
sRes += "|" + _formatNumber(sNumber, 4)
# numéros de téléphone
if nLen == 10:
if sNumber.startswith("0"):
sRes += "|" + _formatNumber(sNumber, 2) # téléphone français
if sNumber[1] == "4" and (sNumber[2]=="7" or sNumber[2]=="8" or sNumber[2]=="9"):
sRes += "|" + sNumber[0:4] + " " + sNumber[4:6] + " " + sNumber[6:8] + " " + sNumber[8:] # mobile belge
sRes += "|" + sNumber[0:3] + " " + sNumber[3:6] + " " + sNumber[6:8] + " " + sNumber[8:] # téléphone suisse
sRes += "|" + sNumber[0:4] + " " + sNumber[4:7] + "-" + sNumber[7:] # téléphone canadien ou américain
elif nLen == 9 and sNumber.startswith("0"):
sRes += "|" + sNumber[0:3] + " " + sNumber[3:5] + " " + sNumber[5:7] + " " + sNumber[7:9] # fixe belge 1
sRes += "|" + sNumber[0:2] + " " + sNumber[2:5] + " " + sNumber[5:7] + " " + sNumber[7:9] # fixe belge 2
else:
# Nombre réel
sInt, sFloat = sNumber.split(",", 1)
sRes = _formatNumber(sInt, 3) + "," + sFloat
return sRes
def _formatNumber (sNumber, nGroup=3):
sRes = ""
nEnd = len(sNumber)
while nEnd > 0:
nStart = max(nEnd-nGroup, 0)
sRes = sNumber[nStart:nEnd] + " " + sRes if sRes else sNumber[nStart:nEnd]
nEnd = nEnd - nGroup
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 (re.error, IndexError):
traceback.print_exc()
return "# erreur #"
def undoLigature (c):
"typography: split ligature character <c> in several chars"
if c == "fi":
return "fi"
if c == "fl":
return "fl"
if c == "ff":
return "ff"
if c == "ffi":
return "ffi"
if c == "ffl":
return "ffl"
if c == "ſt":
return "ft"
if c == "st":
return "st"
return "_"
_xNormalizedCharsForInclusiveWriting = str.maketrans({
'(': '_', ')': '_',
'.': '_', '·': '_', '•': '_',
'–': '_', '—': '_',
'/': '_'
})
def normalizeInclusiveWriting (sToken):
"typography: replace word separators used in inclusive writing by underscore (_)"
return sToken.translate(_xNormalizedCharsForInclusiveWriting)