1 # -*- coding: utf-8 -*-
3 #------------------------------------
4 # import des modules python
5 #------------------------------------
13 #------------------------------------
15 #definition de fonction,
16 #encore execution directe
18 #------------------------------------
20 corpus_out = 'corpus.txt'
22 with codecs.open(corpus_out ,'r', 'utf8') as f:
25 ucis_paras_uces = [[[uce for uce in para.splitlines()] for para in uci.split('$$$')] for uci in content.split(sep)]
26 print(ucis_paras_uces[0])
28 #conn = sqlite3.connect(db)
30 #conn.text_factory = str
32 #c.execute('''CREATE TABLE if not exists uce (id INTEGER PRIMARY KEY, iduci INTEGER, idpara INTEGER, content TEXT)''')
40 def addforme(word, formes, iduce) :
43 if iduce in formes[word][1] :
44 formes[word][1][iduce] += 1
46 formes[word][1][iduce] = 1
48 formes[word] = [1, {iduce:1}]
50 for i, uci in enumerate(ucis_paras_uces) :
56 uce_uci_para[iduce] = [i, idpara]
57 fileout = os.path.join('uce', '%i.txt' % iduce)
58 with open(fileout, 'w') as f :
62 addforme(word, formes, iduce)
63 t1 = time() #chronométrage
64 d = shelve.open('shelves.db')
67 print(time() - t1) #chronométrage
68 t2 = time() #chronométrage
69 d = shelve.open('shelves.db')
72 print(time() - t2) #chronométrage
73 t3 = time() #chronométrage
75 ucis = [uce_uci_para[iduce][0] for iduce in word[1]]
77 print(time() - t3) #chronométrage
79 #c.execute('INSERT INTO uce values (?, ?, ?, ?)', (iduce, i, idpara, uce))