import locale
import sys
from time import time
-from functions import decoupercharact, ReadDicoAsDico, DoConf, ReadLexique
+from functions import decoupercharact, ReadDicoAsDico, DoConf, ReadLexique, progressbar
import re
import sqlite3
import itertools
from operator import itemgetter
from uuid import uuid4
from chemins import PathOut
-from dialog import CorpusPref
+from dialog import CorpusPref, SubTextFromMetaDial
+from copy import copy
from colors import colors
import datetime
copy_corpus.conn_all()
return copy_corpus
+def CopyUce(uce) :
+ return Uce(uce.ident, uce.para, uce.uci)
+
+
+def CopyUci(uci):
+ nuci = Uci(uci.ident, '')
+ nuci.etoiles = copy(uci.etoiles)
+ nuci.uces = [CopyUce(uce) for uce in uci.uces]
+ return nuci
+
class Corpus :
lem = word
self.formes[word] = Word(word, gramtype, len(self.formes), lem)
self.idformesuces[self.formes[word].ident] = {self.ucis[-1].uces[-1].ident : 1}
+
+ def add_word_from_forme(self, word, stident):
+ if word.forme in self.formes :
+ self.formes[word.forme].freq += 1
+ if self.formes[word.forme].ident in self.idformesuces :
+ if stident in self.idformesuces[self.formes[word.forme].ident] :
+ self.idformesuces[self.formes[word.forme].ident][stident] += 1
+ else :
+ self.idformesuces[self.formes[word.forme].ident][stident] = 1
+ else :
+ self.idformesuces[self.formes[word.forme].ident] = {stident: 1}
+ else :
+ self.formes[word.forme] = Word(word.forme, word.gram, len(self.formes), word.lem)
+ self.idformesuces[self.formes[word.forme].ident] = {stident : 1}
def conn_all(self):
"""connect corpus to db"""
query = 'SELECT uces FROM uces where id IN (%s) ORDER BY id' % formesid
res = self.cformes.execute(query)
return list(set(list(itertools.chain(*[[int(val) for val in row[0].split()] if not isinstance(row[0], int) else [row[0]] for row in res]))))
+
+ def gettgenst(self, tgen):
+ formesid = []
+ for lem in tgen :
+ if lem in self.lems :
+ formesid += self.lems[lem].formes
+ else :
+ print 'abscent : %s' % lem
+ query = 'SELECT uces FROM uces where id IN %s ORDER BY id' % str(tuple(formesid))
+ res = self.cformes.execute(query)
+ return list(set(list(itertools.chain(*[[int(val) for val in row[0].split()] if not isinstance(row[0], int) else [row[0]] for row in res]))))
+
+ def gettgenstprof(self, tgen, classe, i, clnb):
+ tgenst = []
+ for lem in tgen :
+ if lem in self.lems :
+ lemst = self.getlemuces(lem)
+ tgenst += lemst
+ if not lem in self.tgenlem :
+ self.tgenlem[lem] = [0] * clnb
+ self.tgenlem[lem][i] = len(set(lemst).intersection(classe))
+ else :
+ print 'abscent: ',lem
+ return list(set(tgenst))
+
+ def gettgentxt(self, tgen):
+ sts = self.gettgenst(tgen)
+ return list(set([self.getucefromid(val).uci for val in sts]))
def getlemucis(self, lem) :
uces = self.getlemuces(lem)
return [len(uce[1].split()) for uce in res]
def getconcorde(self, uces) :
- return self.cuces.execute('select * from uces where id IN (%s);' % ', '.join([`i` for i in uces]))
+ return self.cuces.execute('select * from uces where id IN (%s) ORDER BY id;' % ', '.join([`i` for i in uces]))
def getuciconcorde(self, ucis) :
uces = [[val,[uce.ident for uce in self.ucis[val].uces]] for val in ucis]
self.lems[self.formes[forme].lem] = Lem(self, self.formes[forme])
else :
self.lems = dict([[forme, Lem(self, self.formes[forme])] for forme in self.formes])
+
+ def make_lems_from_dict(self, dictionnaire, dolem = True) :
+ log.info('make lems from dict')
+ self.lems = {}
+ for forme in self.formes :
+ if self.formes[forme].forme in dictionnaire :
+ lem = dictionnaire[forme][0]
+ gram = dictionnaire[forme][1]
+ elif forme.isdigit() :
+ gram = u'num'
+ lem = forme
+ else :
+ gram = u'nr'
+ lem = forme
+ self.formes[forme].lem = lem
+ self.formes[forme].gram = gram
+ if dolem :
+ if self.formes[forme].lem in self.lems :
+ if self.formes[forme].ident not in self.lems[self.formes[forme].lem].formes :
+ self.lems[self.formes[forme].lem].add_forme(self.formes[forme])
+ else :
+ self.lems[self.formes[forme].lem] = Lem(self, self.formes[forme])
+ else :
+ self.lems[forme] = Lem(self, self.formes[forme])
def make_idformes(self) :
self.idformes = dict([[self.formes[forme].ident, self.formes[forme]] for forme in self.formes])
for lem in tokeep :
deff = self.getlemuceseff(lem)
ucesk = deff.keys()
- tab.append([lem] + [sum([deff[uce] for uce in et.intersection(ucesk)]) for et in etuces])
+ line = [lem] + [sum([deff[uce] for uce in et.intersection(ucesk)]) for et in etuces]
+ if sum(line[1:]) >= mineff :
+ tab.append(line)
tab.insert(0, [''] + etoiles)
return tab
+ def make_tgen_table(self, tgen, etoiles, tot = None):
+ lclasses = [self.getucesfrometoile(etoile) for etoile in etoiles]
+ sets = [set(cl) for cl in lclasses]
+ totoccurrences = dict([[val, 0] for val in etoiles])
+ if tot is None :
+ for forme in self.formes :
+ formeuceeff = self.getformeuceseff(forme)
+ for i, classe in enumerate(lclasses) :
+ concern = sets[i].intersection(formeuceeff.keys())
+ if len(concern) :
+ totoccurrences[etoiles[i]] += sum([formeuceeff[uce] for uce in concern])
+ #tgenoccurrences = dict([[val, 0] for val in etoiles])
+ tgenoccurrences = {}
+ for t in tgen.tgen :
+ tgenoccurrences[t] = dict([[val, 0] for val in etoiles])
+ for lem in tgen[t] :
+ lemuceeff = self.getlemuceseff(lem)
+ for i, classe in enumerate(lclasses) :
+ concern = sets[i].intersection(lemuceeff.keys())
+ if len(concern) :
+ tgenoccurrences[t][etoiles[i]] += sum([lemuceeff[uce] for uce in concern])
+ return tgenoccurrences, totoccurrences
+
+ def make_tgen_profile(self, tgen, ucecl, uci = False) :
+ log.info('tgen/classes')
+ self.tgenlem = {}
+ clnb = len(ucecl)
+ if uci :
+ #FIXME : NE MARCHE PLUS CHANGER CA
+ tab = [[lem] + [len(set(self.gettgentxt(tgen[lem])).intersection(classe)) for i, classe in enumerate(ucecl)] for lem in tgen]
+ else :
+ tab = [[lem] + [len(set(self.gettgenstprof(tgen[lem], classe, i, clnb)).intersection(classe)) for i, classe in enumerate(ucecl)] for lem in tgen]
+ tab = [[line[0]] + [val for val in line[1:]] for line in tab if sum(line[1:]) >= 3]
+ return tab
+ #i = 0
+ #nam = 'total'
+ #while nam + `i` in tgen :
+ # i += 1
+ #nam = nam + `i`
+ #last = [nam] + [`len(classe)` for classe in ucecl]
+ #tab += [last]
+ #line0 = ['tgen'] + ['_'.join(['cluster', `i+1`]) for i in range(len(ucecl))]
+ #tab = [line0] + tab
+ #with open(fileout, 'w') as f :
+ # f.write('\n'.join(['\t'.join(line) for line in tab]).encode(self.parametres['syscoding']))
+
def make_efftype_from_etoiles(self, etoiles) :
dtype = {}
etuces = [[] for et in etoiles]
ident += 1
f.write('\n'.join([self.ucis[self.iduces[uce[0]].uci].paras[ident].encode(self.parametres['syscoding']), uce[1].encode(self.parametres['syscoding'])]) + '\n')
+ def export_meta_table(self, outf) :
+ metas = [[`i`] + text.etoiles[1:] for i, text in enumerate(self.ucis)]
+ longueur_max = max([len(val) for val in metas])
+ first = ['column_%i' % i for i in range(longueur_max)]
+ metas.insert(0, first)
+ with open(outf, 'w') as f :
+ f.write('\n'.join(['\t'.join(line) for line in metas]).encode(self.parametres['syscoding']))
+
def export_corpus_classes(self, outf, alc = True, lem = False, uci = False) :
ucecl = {}
for i, lc in enumerate(self.lc) :
if lem :
guce = ' '.join([self.formes[forme].lem for forme in guce.split()])
f.write(guce.encode(self.parametres['syscoding']) + '\n\n')
+
+ def export_owledge(self, rep, classe, lem = False, uci = False) :
+ sts = self.lc[classe - 1]
+ if not uci :
+ res = self.getconcorde(sts)
+ self.make_iduces()
+ else :
+ res = self.getuciconcorde(sts)
+ for uce in res :
+ ident = uce[0]
+ guce = uce[1]
+ outf = '.'.join([`ident`, 'txt'])
+ outf = os.path.join(rep, outf)
+ if lem :
+ guce = ' '.join([self.formes[forme].lem for forme in guce.split()])
+ with open(outf, 'w') as f :
+ f.write(guce.encode('cp1252', errors = 'replace'))
+
+ def export_tropes(self, fileout, classe, lem = False, uci = False) :
+ sts = self.lc[classe - 1]
+ if not uci :
+ res = self.getconcorde(sts)
+ self.make_iduces()
+ else :
+ res = self.getuciconcorde(sts)
+ with open(fileout, 'w') as f :
+ for uce in res :
+ guce = uce[1]
+ if lem :
+ guce = ' '.join([self.formes[forme].lem for forme in guce.split()])
+ f.write(guce.encode('cp1252', errors = 'replace'))
+ f.write('\n')
def make_and_write_sparse_matrix_from_uces(self, actives, outfile, listuce = False) :
log.info('make_and_write_sparse_matrix_from_uces %s' % outfile)
f.write(''.join([' '.join([`duces[uce]+1`,`i+1`,`1`]),'\n']))
f.seek(0)
with open(outfile, 'w') as ffin :
- ffin.write("%%%%MatrixMarket matrix coordinate integer general\n%i %i %i\n" % (self.getucenb(), len(actives), nbl))
+ ffin.write("%%%%MatrixMarket matrix coordinate integer general\n%i %i %i\n" % (len(uces), len(actives), nbl))
for line in f :
ffin.write(line)
os.remove(outfile + '~')
for uci in self.ucis :
etoiles.update(uci.etoiles[1:])
return list(etoiles)
+
+ def make_themes(self):
+ themes = set([])
+ for uci in self.ucis :
+ themes.update(uci.paras)
+ return list(themes)
def make_etoiles_dict(self) :
etoiles = [et for uci in self.ucis for et in uci.etoiles[1:]]
except IndexError :
det[et[0]] = 1
return det
+
+ def make_theme_dict(self):
+ themes = [val for uci in self.ucis for val in uci.paras]
+ det = {}
+ for theme in themes :
+ th = theme.split('_')
+ if th[0] in det :
+ try :
+ endth = '_'.join(th[1:])
+ if theme in det[th[0]] :
+ det[th[0]][theme] += 1
+ else :
+ det[th[0]][theme] = 1
+ except IndexError :
+ det[th[0]] += 1
+ else :
+ try :
+ endth = '_'.join(th[1:])
+ det[th[0]] = {theme:1}
+ except IndexError :
+ det[th[0]] = 1
+ return det
def make_etline(self, listet) :
etuces = [[] for et in listet]
listlem.sort()
with open(fileout, 'w') as f :
f.write('\n'.join(['\t'.join(lem) for lem in listlem]).encode(syscoding))
-
-
+
class MakeUciStat :
self.corpus.pathout = PathOut(filename = parametres_corpus['originalpath'], dirout = parametres_corpus['pathout'])
self.corpus.pathout.createdir(parametres_corpus['pathout'])
self.corpus.parametres['uuid'] = str(uuid4())
- self.corpus.parametres['corpus_name'] = os.path.split(self.corpus.parametres['pathout'])[1]
+ self.corpus.parametres['corpus_name'] = parametres_corpus['corpus_name']#os.path.split(self.corpus.parametres['pathout'])[1]
self.corpus.parametres['type'] = 'corpus'
if self.corpus.parametres['keep_ponct'] :
self.ponctuation_espace = [' ', '']
self.cf.execute('CREATE INDEX ideff ON eff (id);')
self.c.close()
self.cf.close()
- #backup corpora
+ #backup corpus
self.conn_corpus = sqlite3.connect(self.corpus.pathout['corpus.db'])
self.ccorpus = self.conn_corpus.cursor()
self.ccorpus.execute('CREATE TABLE IF NOT EXISTS etoiles (uci INTEGER, et TEXT, paras TEXT);')
self.cleans.append(self.dotiret)
def make_expression(self,txt) :
- for expression in self.expressions:
+ exp = self.expressions.keys()
+ exp.sort(reverse=True)
+ for expression in exp :
if expression in txt :
txt = txt.replace(expression, self.expressions[expression][0])
return txt
pourhapaxocc = (float(hapaxnb) / self.corpus.parametres['occurrences']) * 100
self.corpus.parametres['hapax'] = '%i - %.2f %% des formes - %.2f %% des occurrences' % (hapaxnb, pourhapaxf, pourhapaxocc)
+class BuildSubCorpus(BuildCorpus):
+ def __init__(self, corpus, parametres, dlg = None) :
+ log.info('begin subcorpus...')
+ self.dlg = dlg
+ self.ori = corpus
+ self.infile = None
+ self.corpus = Corpus(self, {'type' : 'corpus', 'originalpath' : corpus.parametres['originalpath'], 'encoding' : corpus.parametres['encoding']})
+ self.last = 0
+ self.parametres = parametres
+ self.encoding = corpus.parametres['encoding']
+ self.corpus.parametres['corpus_name'] = parametres['corpus_name']
+ self.corpus.pathout = PathOut(filename = corpus.parametres['originalpath'], dirout = parametres['pathout'])
+ self.corpus.pathout.createdir(parametres['pathout'])
+ self.corpus.parametres['pathout'] = parametres['pathout']
+ self.corpus.parametres['meta'] = parametres.get('meta', False)
+ self.corpus.parametres['uuid'] = str(uuid4())
+ if parametres.get('frommeta', False) :
+ print 'make subtexts'
+ self.corpus.ucis = [CopyUci(uci) for uci in self.ori.ucis if set(parametres['meta']).intersection(uci.etoiles) != set()]
+ elif parametres.get('fromtheme', False) :
+ print 'make subtexts from theme'
+ idpara = 0
+ for uci in self.ori.ucis :
+ if uci.paras != [] :
+ newuce = []
+ newpara = []
+ for et in uci.paras :
+ if et in parametres['meta'] :
+ newuce += [CopyUce(uce) for uce in uci.uces if uce.para == idpara]
+ newpara.append(et)
+ idpara += 1
+ if newuce != [] :
+ nuci = CopyUci(uci)
+ nuci.uces = newuce
+ nuci.paras = newpara
+ self.corpus.ucis.append(nuci)
+ else :
+ idpara += 1
+ elif parametres.get('fromclusters', False) :
+ self.parametres['uceids'] = [st for i in self.parametres['meta'] for st in self.parametres['lc'][i]]
+ self.fromuceids()
+ elif parametres.get('fromuceids', False) :
+ self.fromuceids()
+ #create database
+ self.connect()
+ self.dobuild()
+
+ def fromuceids(self):
+ print 'fromuceids'
+ dictucekeep = dict(zip(self.parametres['uceids'], self.parametres['uceids']))
+ idpara = 0
+ for uci in self.ori.ucis :
+ if uci.paras == [] :
+ keepuces = [CopyUce(uce) for uce in uci.uces if uce.ident in dictucekeep]
+ if keepuces != [] :
+ nuci = CopyUci(uci)
+ nuci.uces = keepuces
+ self.corpus.ucis.append(nuci)
+ idpara += 1
+ else :
+ newuces = []
+ newpara = []
+ for et in uci.paras :
+ keepuces = [CopyUce(uce) for uce in uci.uces if uce.ident in dictucekeep]
+ idpara += 1
+ if keepuces != [] :
+ newuces += keepuces
+ newpara.append(et)
+ if newuces != [] :
+ nuci = CopyUci(uci)
+ nuci.uces = newuces
+ nuci.paras = newpara
+ self.corpus.ucis.append(nuci)
+
+ def read_corpus(self, infile = None):
+ self.olduceid = [uce.ident for uci in self.corpus.ucis for uce in uci.uces]
+ ident_uci = 0
+ ident_uce = 0
+ ident_para = -1
+ lastpara = -1
+ newuceident = {}
+ print 'redo text, para and st ident'
+ for uci in self.corpus.ucis :
+ uci.ident = ident_uci
+ ident_uci += 1
+ for uce in uci.uces :
+ uce.uci = uci.ident
+ if uce.para != lastpara :
+ ident_para += 1
+ lastpara = uce.para
+ uce.para = ident_para
+ else :
+ uce.para = ident_para
+ newuceident[uce.ident] = ident_uce
+ uce.ident = ident_uce
+ ident_uce += 1
+ print 'backup st text and forms'
+ for row in self.ori.getconcorde(self.olduceid) :
+ self.c.execute('INSERT INTO uces VALUES(?,?);', (`newuceident[row[0]]`, row[1]))
+ for word in row[1].split() :
+ self.corpus.add_word_from_forme(self.ori.formes[word], newuceident[row[0]])
+ self.backup_uce()
+ print 'done'
class BuildFromAlceste(BuildCorpus) :
def read_corpus(self, infile) :
if iduci != -1 and iduce != -1:
self.backup_uce()
else :
- log.info(_(u"No Text in corpora. Are you sure of the formatting ?"))
+ log.info(_(u"No Text in corpus. Are you sure of the formatting ?"))
raise Exception('TextBeforeTextMark %i' % linenb)
except UnicodeDecodeError :
raise Exception("CorpusEncoding")
def __init__(self, parent, dlg = None) :
self.parent = parent
self.dlg = dlg
+
parametres = DoConf(os.path.join(self.parent.UserConfigPath,'corpus.cfg')).getoptions('corpus')
parametres['pathout'] = PathOut(parent.filename, 'corpus').mkdirout()
+ parametres['corpus_name'] = os.path.split(parametres['pathout'])[1]
dial = CorpusPref(parent, parametres)
dial.CenterOnParent()
dial.txtpath.SetLabel(parent.filename)
#dial.repout_choices.SetValue(parametres['pathout'])
self.res = dial.ShowModal()
+ if self.dlg is not None :
+ self.dlg = progressbar(self.parent, self.dlg)
if self.res == 5100 :
parametres = dial.doparametres()
parametres['originalpath'] = parent.filename
PathOut().createdir(parametres['pathout'])
- ReadLexique(self.parent, lang = parametres['lang'])
+ if parametres.get('dictionary', False) :
+ filein = parametres['dictionary']
+ else :
+ filein = None
+ if dial.corpusname.GetValue() != '' :
+ parametres['corpus_name'] = dial.corpusname.GetValue()
+ dial.Destroy()
+ ReadLexique(self.parent, lang = parametres['lang'], filein = filein)
if parametres['lang'] != 'other' and os.path.exists(self.parent.DictPath.get(parametres['lang']+'_exp', 'french_exp')):
self.parent.expressions = ReadDicoAsDico(self.parent.DictPath.get(parametres['lang']+'_exp', 'french_exp'))
else :
self.parent.expressions = {}
self.parametres = parametres
else :
+ dial.Destroy()
if self.dlg is not None :
self.dlg.Destroy()
- dial.Destroy()
def doanalyse(self) :
return BuildFromAlceste(self.parent.filename, self.parametres, self.parent.lexique, self.parent.expressions, dlg = self.dlg).corpus
-
-if __name__ == '__main__' :
- t1 = time()
- parametres = {'formesdb':'formes.db', 'ucesdb': 'uces.db', 'corpusdb' : 'corpus.db', 'syscoding' : 'utf-8', 'encoding' : 'utf8'}
- intro = BuildCorpus(infile, parametres)#, tar_in, tar_infouce)#, tar_formes)
- print time() - t1
+class SubBuilder :
+ def __init__(self, parent, corpus, parametres = None, dlg = None):
+ self.parent = parent
+ self.ori = corpus
+ self.dlg = dlg
+ corpus_name = 'Sub' + corpus.parametres['corpus_name']
+ if dlg is not None :
+ busy = wx.BusyInfo(_("Please wait...").decode('utf8'), self)
+ wx.SafeYield()
+ parametres['corpus_name'] = corpus_name
+ if parametres.get('frommeta', False) :
+ parametres['meta'] = corpus.make_etoiles()
+ elif parametres.get('fromtheme', False) :
+ parametres['meta'] = corpus.make_themes()
+ elif parametres.get('fromclusters', False) :
+ parametres['meta'] = [' '.join(['classe', `i`]) for i in range(1,parametres['clnb'] + 1)]
+ else :
+ parametres['meta'] = []
+ if 'fromclusters' not in parametres :
+ parametres['meta'].sort()
+ if dlg is not None :
+ del busy
+ dial = SubTextFromMetaDial(parent, parametres)
+ self.res = dial.ShowModal()
+ if self.res == 5100 :
+ if dial.subcorpusname.GetValue() != '' :
+ corpus_name = ''.join([l for l in dial.subcorpusname.GetValue() if l.isalnum() or l in ['_']])
+ if corpus_name != '' :
+ parametres['corpus_name'] = corpus_name
+ else :
+ parametres['corpus_name'] = 'Sub' + corpus.parametres['corpus_name']
+ pathout = os.path.join(corpus.parametres['pathout'], parametres['corpus_name'])
+ i = 1
+ while os.path.exists(pathout + '_%i' % i) :
+ i += 1
+ parametres['pathout'] = pathout + '_%i' % i
+ meta = dial.m_listBox1.GetSelections()
+ if not 'fromclusters' in parametres :
+ parametres['meta'] = [parametres['meta'][val] for val in meta]
+ else :
+ parametres['meta'] = meta
+ self.parametres = parametres
+ dial.Destroy()
+ else :
+ dial.Destroy()
+
+ def doanalyse(self):
+ return BuildSubCorpus(self.ori, parametres = self.parametres, dlg = self.dlg).corpus
projects / iramuteq / blobdiff