ola5doc/content/leo_martin/sources/pattern__premier-essai.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-

from pattern.fr import parse, split

texte_brut = u"Le petit chat rugit alors que l'esclave humain tarde a accomplir sa tache de nourrissage du maitre et seigneur.".encode('utf-8')
texte_traite = parse( texte_brut )

def imprimer_erreur ( erreur ):
    print '💩 erreur : ' + str( erreur )

def imprimer_structure_mot ( mot ):
    '''imprime la structure de la phrase'''
    print 'index : ' + str( mot.index )
    print 'string : ' + str( mot.string )
    print '_custom_tags : ' + str( mot._custom_tags )
    print 'sentence : ' + str( mot.sentence )
    print 'pnp : ' + str( mot.pnp )
    print 'chunk : ' + str( mot.chunk )
    print 'lemma : ' + str( mot.lemma )
    print 'type : ' + str( mot.type )

for phrase in split( texte_traite ):
    print "——————————"
    print "mots :"
    try:
        print phrase.words
    except Exception as erreur:
        imprimer_erreur( erreur )
    print "——————————"
    print "groupes de mots :"
    try:
        print str(phrase.chunks)
    except Exception as erreur:
        imprimer_erreur( erreur )
    print "——————————"
    print "structure, phrase :"

    try:
        print phrase.__dict__.keys()
    except Exception as erreur:
        imprimer_erreur( erreur )
    print "——————————"
    print "structure, mot :"

    try:
        print phrase.words[0].__dict__.keys()
    except Exception as erreur:
        imprimer_erreur( erreur )
    print "——————————"
    print "structure, chunk (groupe de mots) :"
    print phrase.chunks[0].__dict__.keys()