Grammalecte  Diff

import sys
import re
import collections
import zipfile
import math
import argparse
import tags

from enum import Enum

from distutils import dir_util
from distutils import file_util
from string import Template

import tags

import metagraphe
import metaphone2
import thes_build

    if os.path.isfile(spf):
        with open(spf, "r", encoding="utf-8") as hSrc:
            for sLine in hSrc:
                yield sLine
    else:
        print("# Error: file not found.")








class Dictionnaire:
    def __init__ (self, version, name):
        # Dictionary
        self.sName = name
        self.lEntry = []

            hDst.write(Flexion.simpleHeader())
            for oFlex in self.lFlexions:
                hDst.write(oFlex.getGrammarCheckerRepr())

    def createFiles (self, spDst, lDictVars, nMode, bSimplified):
        sDicName = PREFIX_DICT_PATH + self.sVersion
        spDic = spDst + '/' + sDicName
        dir_util.mkpath(spDic)
        for dVars in lDictVars:
            # template vars
            dVars['version'] = self.sVersion
            # Dictionaries files (.dic) (.aff)
            self.writeAffixes(spDic, dVars, nMode, bSimplified)
            self.writeDictionary(spDic, dVars, nMode, bSimplified)
        copyTemplate('orthographe', spDic, 'README_dict_fr.txt', dVars)
        createZipFiles(spDic, spDst, sDicName + '.zip')

    def createLibreOfficeExtension (self, spBuild, dTplVars, lDictVars, sThesVer, spDestGL=""):
        # LibreOffice extension
        echo(" * Dictionnaire >> extension pour LibreOffice")
        dTplVars['version'] = self.sVersion
        sExtensionName = EXT_PREFIX_OOO + self.sVersion
        spExt = spBuild + '/' + sExtensionName
        dir_util.mkpath(spExt+'/META-INF')
        dir_util.mkpath(spExt+'/ui')
        dir_util.mkpath(spExt+'/dictionaries')
        dir_util.mkpath(spExt+'/pythonpath')
        file_util.copy_file('_templates/ooo/manifest.xml', spExt+'/META-INF')
        file_util.copy_file('_templates/ooo/DictionarySwitcher.py', spExt)
        file_util.copy_file('_templates/ooo/ds_strings.py', spExt+'/pythonpath')
        file_util.copy_file('_templates/ooo/addons.xcu', spExt+'/ui')
        file_util.copy_file('_templates/ooo/french_flag.png', spExt)
        file_util.copy_file('_templates/ooo/french_flag_16.bmp', spExt+'/ui')
        copyTemplate('_templates/ooo', spExt, 'description.xml', dTplVars)
        copyTemplate('_templates/ooo', spExt, 'dictionaries.xcu', dTplVars)
        #file_util.copy_file('_templates/ooo/dictionaries.xcu.tpl.xml', spExt)
        copyTemplate('_templates/ooo', spExt, 'package-description.txt', dTplVars)
        for dVars in lDictVars:
            dicPath = spBuild + '/' + PREFIX_DICT_PATH + self.sVersion
            file_util.copy_file(dicPath+'/'+dVars['asciiName']+'.dic', spExt+'/dictionaries/'+dVars['asciiName']+'.dic')
            file_util.copy_file(dicPath+'/'+dVars['asciiName']+'.aff', spExt+'/dictionaries/'+dVars['asciiName']+'.aff')
        copyTemplate('orthographe', spExt+'/dictionaries', 'README_dict_fr.txt', dTplVars)
        # hyphenation
        file_util.copy_file('césures/hyph_fr.dic', spExt+'/dictionaries')
        file_util.copy_file('césures/hyph_fr.iso8859-1.dic', spExt+'/dictionaries')
        file_util.copy_file('césures/frhyph.tex', spExt+'/dictionaries')
        file_util.copy_file('césures/hyph-fr.tex', spExt+'/dictionaries')
        file_util.copy_file('césures/README_hyph_fr-3.0.txt', spExt+'/dictionaries')
        file_util.copy_file('césures/README_hyph_fr-2.9.txt', spExt+'/dictionaries')
        # thesaurus
        file_util.copy_file(spBuild+"/thesaurus-v"+sThesVer+'/thes_fr.dat', spExt+"/dictionaries")
        file_util.copy_file(spBuild+"/thesaurus-v"+sThesVer+'/thes_fr.idx', spExt+"/dictionaries")
        file_util.copy_file(spBuild+"/thesaurus-v"+sThesVer+'/README_thes_fr.txt', spExt+"/dictionaries")
        # zip
        createZipFiles(spExt, spBuild, sExtensionName + '.oxt')
        # copy to Grammalecte Project
        if spDestGL:
            echo("   Dictionnaires Hunspell copiés dans Grammalecte pour LibreOffice...")
            dir_util.copy_tree(spExt+'/dictionaries', spDestGL)

    def createMozillaExtensions (self, spBuild, dTplVars, lDictVars, spDestGL=""):
        # Mozilla extension 1
        echo(" * Dictionnaire >> extension pour Mozilla")
        dTplVars['version'] = self.sVersion
        sExtensionName = EXT_PREFIX_MOZ + self.sVersion
        spExt = spBuild + '/' + sExtensionName
        dir_util.mkpath(spExt+'/dictionaries')
        copyTemplate('_templates/moz', spExt, 'manifest.json', dTplVars)
        spDict = spBuild + '/' + PREFIX_DICT_PATH + self.sVersion
        file_util.copy_file(spDict+'/fr-classique.dic', spExt+'/dictionaries/fr-classic.dic')
        file_util.copy_file(spDict+'/fr-classique.aff', spExt+'/dictionaries/fr-classic.aff')
        copyTemplate('orthographe', spExt, 'README_dict_fr.txt', dTplVars)
        createZipFiles(spExt, spBuild, sExtensionName + '.xpi')
        # Grammalecte
        if spDestGL:
            echo("   Dictionnaires Hunspell copiés dans Grammalecte pour Mozilla")
            for dVars in lDictVars:
                file_util.copy_file(spDict+'/'+dVars['asciiName']+'.dic', spDestGL+'/'+dVars['mozAsciiName']+"/"+dVars['mozAsciiName']+'.dic')
                file_util.copy_file(spDict+'/'+dVars['asciiName']+'.aff', spDestGL+'/'+dVars['mozAsciiName']+"/"+dVars['mozAsciiName']+'.aff')

    def createFileIfqForDB (self, spBuild):
        echo(" * Dictionnaire >> indices de fréquence pour la DB...")
        with open(spBuild+'/dictIdxIfq-'+self.sVersion+'.diff.txt', 'w', encoding='utf-8', newline="\n") as hDiff, \
             open(spBuild+'/dictIdxIfq-'+self.sVersion+'.notes.txt', 'w', encoding='utf-8', newline="\n") as hNotes:
            for oEntry in self.lEntry:
                if oEntry.fq != oEntry.oldFq:
                    hDiff.write("{0.iD}\t{0.fq}\n".format(oEntry))
                    hNotes.write("{0.lemma}/{0.flags}\t{0.oldFq} > {0.fq}\n".format(oEntry))

    def createLexiconPackages (self, spBuild, version, oStatsLex, spDestGL=""):
        sLexName = LEX_PREFIX + version
        spLex = spBuild + '/' + sLexName
        dir_util.mkpath(spLex)
        # write lexicon
        self.sortLexiconByFreq()
        self.writeLexicon(spLex + '/' + sLexName + '.txt', version, oStatsLex)
        self.writeGrammarCheckerLexicon(spBuild + '/' + sLexName + '.lex', version)
        copyTemplate('lexique', spLex, 'README_lexique.txt', {'version': version})
        # zip
        createZipFiles(spLex, spBuild, sLexName + '.zip')
        # copy GC lexicon to Grammalecte
        if spDestGL:
            file_util.copy_file(spBuild + '/' + sLexName + '.lex', spDestGL + '/French.lex')
            file_util.copy_file('lexique/French.tagset.txt', spDestGL)

    def createDictConj (self, spBuild, spDestGL=""):
        echo(" * Dictionnaire >> fichier de conjugaison...")
        with open(spBuild+'/dictConj.txt', 'w', encoding='utf-8', newline="\n") as hDst:
            for oEntry in self.lEntry:
                if oEntry.po.startswith("v"):
                    hDst.write(oEntry.getConjugation())
        if spDestGL:
            echo("   Fichier de conjugaison copié dans Grammalecte...")
            file_util.copy_file(spBuild+'/dictConj.txt', spDestGL)

    def createDictDecl (self, spBuild, spDestGL=""):
        echo(" * Dictionnaire >> fichier de déclinaison...")
        with open(spBuild+'/dictDecl.txt', 'w', encoding='utf-8', newline="\n") as hDst:
            for oEntry in self.lEntry:
                if re.match("[SXFWIA]", oEntry.flags) and (oEntry.po.startswith("nom") or oEntry.po.startswith("adj")):
                    hDst.write(oEntry.getDeclination())
        if spDestGL:
            echo("   Fichier de déclinaison copié dans Grammalecte...")
            file_util.copy_file(spBuild+'/dictDecl.txt', spDestGL)


class Entree:
    def __init__ (self, sLine):
        self.lemma = ''
        self.flags = ''
        # champs morphologiques Hunspell

            for e in self.dFlexions.items():
                hDst.write("{} - {}\n".format(e[0], e[1]))


def createThesaurusPackage (spBuild, sVersion, spCopy="", spDataDestGL=""):
    print(" * Création du thésaurus")
    spThesaurus = spBuild+"/thesaurus-v"+sVersion
    dir_util.mkpath(spThesaurus)
    thes_build.build("thesaurus/thes_fr.dat", "thesaurus/synsets_fr.dat", spThesaurus)
    file_util.copy_file('thesaurus/README_thes_fr.txt', spThesaurus)
    if spCopy:
        # copy in libreoffice extension package
        print("   Copie du thésaurus dans:", spCopy)
        file_util.copy_file(spThesaurus+'/thes_fr.dat', spCopy)
        file_util.copy_file(spThesaurus+'/thes_fr.idx', spCopy)
        file_util.copy_file(spThesaurus+'/README_thes_fr.txt', spCopy)
    if spModulesDestGL:
        # copy in data source folder of Grammalecte
        file_util.copy_file(spThesaurus+'/thes_fr.json', spDataDestGL)


def main ():
    xParser = argparse.ArgumentParser()
    xParser.add_argument("-v", "--verdic", help="set dictionary version, i.e. 5.4", type=str, default="X.Y.z")
    xParser.add_argument("-m", "--mode", help="0: no tags,  1: Hunspell tags (default),  2: All tags", type=int, choices=[0, 1, 2], default=1)
    xParser.add_argument("-u", "--uncompress", help="do not use Hunspell compression", action="store_true")

    echo("Version: " + xArgs.verdic)
    echo("Simplify: " + str(xArgs.simplify))
    echo("Mode: " + str(xArgs.mode))
    echo("Compression: " + str(not(xArgs.uncompress)))

    ### création du répertoire
    spBuild = BUILD_PATH + '/' + xArgs.verdic
    dir_util.mkpath(spBuild)

    ### Lecture des fichiers et création du dictionnaire
    oFrenchDict = Dictionnaire(xArgs.verdic, "French dictionary")
    for sFile in ['orthographe/FRANCAIS.dic']:
        oFrenchDict.readDictionary(sFile)
    oFrenchDict.readAffixes('orthographe/FRANCAIS_7.aff')

    oStatsLex.addLexFromFile('lexique/corpus_data/stats_litterature.txt', 'L', 'Littérature')
    oStatsLex.write(spBuild+'/test_lex.txt')
    oFrenchDict.calculateStats(oStatsLex, spfStats)

    ### Écriture des paquets
    echo("Création des paquets...")

    nThesaurusVersion = 2.4
    spLexiconDestGL = "../../../lexicons"  if xArgs.grammalecte  else ""
    spLibreOfficeExtDestGL = "../oxt/Dictionnaires/dictionaries"  if xArgs.grammalecte  else ""
    spMozillaExtDestGL = ""  if xArgs.grammalecte  else "" # no more Hunspell dictionaries in Mozilla extensions for now
    spDataDestGL = "../data"  if xArgs.grammalecte  else ""

    ### dictionnaires
    if not xArgs.uncompress:
        oFrenchDict.defineAbreviatedTags(xArgs.mode, spfStats)
    oFrenchDict.createFiles(spBuild, [dTOUTESVAR, dCLASSIQUE, dREFORME1990], xArgs.mode, xArgs.simplify)
    oFrenchDict.createLexiconPackages(spBuild, xArgs.verdic, oStatsLex, spLexiconDestGL)
    oFrenchDict.createFileIfqForDB(spBuild)
    createThesaurusPackage(spBuild, nThesaurusVersion, spLibreOfficeExtDestGL, spDataDestGL)
    oFrenchDict.createLibreOfficeExtension(spBuild, dMOZEXT, [dTOUTESVAR, dCLASSIQUE, dREFORME1990], nThesaurusVersion, spLibreOfficeExtDestGL)
    oFrenchDict.createMozillaExtensions(spBuild, dMOZEXT, [dTOUTESVAR, dCLASSIQUE, dREFORME1990], spMozillaExtDestGL)
    oFrenchDict.createDictConj(spBuild, spDataDestGL)
    oFrenchDict.createDictDecl(spBuild, spDataDestGL)


if __name__ == '__main__':
    main()