Grammalecte  Diff

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

import metagraphe
import metaphone2
import thes_build


# Dictionnaire des caractères pour le tri naturel.
# Ordre souhaitable, mais pose problème pour la recherche, car engendre des égalités de lemmes différents.
# Il faut donc travailler sur un dictionnaire trié *numériquement* et le sauvegarder selon le tri *naturel*
CHARMAP = str.maketrans({ 'à': 'a',  'À': 'A',  'â': 'a',  'Â': 'A',  'ä': 'a',  'Ä': 'A',  'å': 'a',  'Å': 'A',  'ā': 'a',  'Ā': 'A',
                          'ç': 'c',  'Ç': 'C',

        #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')

        with open(sPathFile, 'w', encoding='utf-8', newline="\n") as hDst:
            for t in self.lLex:
                hDst.write(str(t)+"\n")
            for e in self.dFlexions.items():
                hDst.write("{} - {}\n".format(e[0], e[1]))


def createThesaurusPackage (spBuild, sVersion, spCopy=""):
    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
        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)


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")
    xParser.add_argument("-s", "--simplify", help="no virtual lemmas", action="store_true")

    oStatsLex.addLexFromFile('lexique/corpus_data/stats_google_ngram_1.txt', 'G', 'Google 1-grams')
    oStatsLex.addLexFromFile('lexique/corpus_data/stats_frwiki.txt', 'W', 'Wikipédia')
    oStatsLex.addLexFromFile('lexique/corpus_data/stats_frwikisource.txt', 'S', 'Wikisource')
    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...")

    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, xArgs.verdic, spLibreOfficeExtDestGL)
    oFrenchDict.createLibreOfficeExtension(spBuild, dMOZEXT, [dTOUTESVAR, dCLASSIQUE, dREFORME1990], spLibreOfficeExtDestGL)
    oFrenchDict.createMozillaExtensions(spBuild, dMOZEXT, [dTOUTESVAR, dCLASSIQUE, dREFORME1990], spMozillaExtDestGL)
    oFrenchDict.createDictConj(spBuild, spDataDestGL)
    oFrenchDict.createDictDecl(spBuild, spDataDestGL)



if __name__ == '__main__':
    main()