...

[iramuteq] / import_txm.py

diff --git a/import_txm.py b/import_txm.py
index d9eea31..0dfd1eb 100644 (file)
--- a/import_txm.py
+++ b/import_txm.py
@@ -9,12 +9,6 @@ import os
 import xml.sax
 import glob
 
 import xml.sax
 import glob
 

-
-
-#infiledir = '/home/pierre/TXM/corpora/voeux-bin/txm/VOEUX/'
-#fileout = 'VOEUXExportfromTXM.txt'
-
-

 class TXMParser(xml.sax.ContentHandler) :
     def __init__(self, fileout, encodage_out) :
         self.fileout = fileout
 class TXMParser(xml.sax.ContentHandler) :
     def __init__(self, fileout, encodage_out) :
         self.fileout = fileout


@@ -35,7 +29,7 @@ class TXMParser(xml.sax.ContentHandler) :
             pass
 
     def endElement(self, name) :
             pass
 
     def endElement(self, name) :

-        if name == 's' :
+        if name == 's' or name == 'w' :

             self.printsent()
         if name == 'p' :
             self.printsent()
             self.printsent()
         if name == 'p' :
             self.printsent()


@@ -48,7 +42,7 @@ class TXMParser(xml.sax.ContentHandler) :
             #self.fileout.write(content.encode('utf8'))
 
     def text2stars(self, attrs) :
             #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))
         stars = [''.join([u'*', val]) for val in stars]
         stars = u'**** ' + ' '.join(stars)
         self.fileout.write(stars.encode(self.encodage_out))


@@ -63,16 +57,14 @@ class TXMParser(xml.sax.ContentHandler) :
             self.fileout.write(sent.encode(self.encodage_out))
         self.sent = []
 
             self.fileout.write(sent.encode(self.encodage_out))
         self.sent = []
 

-
-
-class TXM2IRA :
-    def __init__(self, pathin, fileout, encodage_in, encodage_out) :
+def TXM2IRA(pathin, fileout, encodage_in, encodage_out) :

         parser = xml.sax.make_parser()
         files = glob.glob(os.path.join(pathin,'*.xml'))
         parser = xml.sax.make_parser()
         files = glob.glob(os.path.join(pathin,'*.xml'))

+        if len(files) == 0 :
+            return 'nofile'

         with open(fileout, 'w') as fout :
             parser.setContentHandler(TXMParser(fout, encodage_out))
             for f in files :
                 parser.parse(open(f, 'r'))
                 fout.write('\n\n')
         with open(fileout, 'w') as fout :
             parser.setContentHandler(TXMParser(fout, encodage_out))
             for f in files :
                 parser.parse(open(f, 'r'))
                 fout.write('\n\n')

-
-#TXM2IRA(infiledir, fileout)
+        return None
\ No newline at end of file