X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=import_txm.py;h=bc55fc2a378ff6e18424a9ac622567d5dae798a4;hp=6ef0c2b439d824caf03bf65c06c93abe91685362;hb=94db7716520d397ab87a1a7730d2d0fd9f873da0;hpb=970a30c06f7399edfb9d0a6823776c611f066af1

diff --git a/import_txm.py b/import_txm.py
index 6ef0c2b..bc55fc2 100644
--- a/import_txm.py
+++ b/import_txm.py
@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 #Author: Pierre Ratinaud
 #Copyright (c) 2013 Pierre Ratinaud
-#Lisense: GNU/GPL
+#License: GNU/GPL
 
 
 import os
@@ -35,7 +35,7 @@ class TXMParser(xml.sax.ContentHandler) :
             pass
 
     def endElement(self, name) :
-        if name == 's' :
+        if name == 's' or name == 'w' :
             self.printsent()
         if name == 'p' :
             self.printsent()
@@ -48,7 +48,7 @@ class TXMParser(xml.sax.ContentHandler) :
             #self.fileout.write(content.encode('utf8'))
 
     def text2stars(self, attrs) :
-        stars = ['_'.join(val).replace(' ', '_').replace("'", '_') for val in attrs.items()]
+        stars = ['_'.join(val).replace(' ', '_').replace("'", '_').replace('/','').replace('.','').replace(';', '').replace(':', '').replace(u'·','') for val in attrs.items()]
         stars = [''.join([u'*', val]) for val in stars]
         stars = u'**** ' + ' '.join(stars)
         self.fileout.write(stars.encode(self.encodage_out))
@@ -74,5 +74,6 @@ class TXM2IRA :
             for f in files :
                 parser.parse(open(f, 'r'))
                 fout.write('\n\n')
+        print 'done'
 
 #TXM2IRA(infiledir, fileout)