X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=corpus.py;h=d4357cf97559ce4c4b4cdcd8886494e469d06603;hp=7cc9306d23558018df967abad308dd1258a492af;hb=ef45aa7e5e55a37956ce86dc4ce86471f11b018d;hpb=7fb5b2b86f6c9a0617208ee85211177c23d12f47 diff --git a/corpus.py b/corpus.py index 7cc9306..d4357cf 100644 --- a/corpus.py +++ b/corpus.py @@ -210,6 +210,18 @@ class Corpus : def getucesfrometoile(self, etoile) : return [uce.ident for uci in self.ucis for uce in uci.uces if etoile in uci.etoiles] + def getetoileuces(self) : + log.info('get uces etoiles') + etoileuces = {} + for uci in self.ucis : + etoiles = uci.etoiles[1:] + uci.paras + for et in etoiles : + if et in etoileuces : + etoileuces[et] += [uce.ident for uce in uci.uces] + else : + etoileuces[et] = [uce.ident for uce in uci.uces] + return etoileuces + def getucefromid(self, uceid) : if self.iduces is None : self.make_iduces() return self.iduces[uceid] @@ -564,13 +576,16 @@ class Corpus : elif get != [] : etuces[listet.index(get[0])] += [uce.ident for uce in uci.uces] return etuces - def make_and_write_profile_et(self, ucecl, fileout) : log.info('etoiles/classes') - etoiles = self.make_etoiles() + etoileuces = self.getetoileuces() + etoileuces = dict([[et, etoileuces[et]] for et in etoileuces if len(etoileuces[et]) > 1]) with open(fileout, 'w') as f : - f.write('\n'.join([';'.join([etoile] + [`len(set(self.getucesfrometoile(etoile)).intersection(classe))` for classe in ucecl]) for etoile in etoiles]).encode(self.parametres['syscoding'])) + f.write('\n'.join([';'.join([et] + [`len(set(etoileuces[et]).intersection(classe))` for classe in ucecl]) for et in etoileuces]).encode(self.parametres['syscoding'])) + #etoiles = self.make_etoiles() + #with open(fileout, 'w') as f : + # f.write('\n'.join([';'.join([etoile] + [`len(set(self.getucesfrometoile(etoile)).intersection(classe))` for classe in ucecl]) for etoile in etoiles]).encode(self.parametres['syscoding'])) def make_colored_corpus(self) : ucecl = {}