X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=corpus.py;h=51061fbb2006bb58385173f235a35d4cbdc5f147;hp=38648123b307062f2ea6b94c8cf85ad84f66d460;hb=f89443607fdb078b159267a8eaaececaab76ba08;hpb=ea81472844df73875a9b2ba9122d487361fd88ac diff --git a/corpus.py b/corpus.py index 3864812..51061fb 100644 --- a/corpus.py +++ b/corpus.py @@ -786,6 +786,22 @@ class Corpus : with open('/tmp/testhapxuce.html','w') as f : f.write(txt) + def export_dictionary(self, fileout, syscoding) : + listformes = [[self.formes[forme].freq, forme, self.formes[forme].lem, self.formes[forme].gram] for forme in self.formes] + listformes.sort(reverse = True) + listformes = [forme[1:] + [`forme[0]`] for forme in listformes] + with open(fileout, 'w') as f : + f.write('\n'.join(['\t'.join(forme) for forme in listformes]).encode(syscoding)) + + def export_lems(self, fileout, syscoding) : + self.make_idformes() + listlem = [[lem, '\t'.join(['\t'.join([self.idformes[forme].forme, `self.lems[lem].formes[forme]`]) for forme in self.lems[lem].formes])] for lem in self.lems] + listlem.sort() + with open(fileout, 'w') as f : + f.write('\n'.join(['\t'.join(lem) for lem in listlem]).encode(syscoding)) + + + class MakeUciStat : def __init__(self, corpus) :