Merge remote branch 'remotes/origin/pootle-import'
[vuplus_dvbapp] / po / xml2po.py
index 382f2d8..cfbeaf4 100755 (executable)
@@ -1,35 +1,64 @@
 #!/usr/bin/python
 import sys
 import os
+import string
 from xml.sax import make_parser
-from xml.sax.handler import ContentHandler
+from xml.sax.handler import ContentHandler, property_lexical_handler
+try:
+       from _xmlplus.sax.saxlib import LexicalHandler
+       no_comments = False
+except ImportError:
+       class LexicalHandler:
+               pass
+       no_comments = True
 
-class parseXML(ContentHandler):
+class parseXML(ContentHandler, LexicalHandler):
        def __init__(self, attrlist):
                self.isPointsElement, self.isReboundsElement = 0, 0
                self.attrlist = attrlist
+               self.last_comment = None
+
+       def comment(self, comment):
+               if comment.find("TRANSLATORS:") != -1:
+                       self.last_comment = comment
 
        def startElement(self, name, attrs):
-               if (attrs.has_key('text')):
-                       attrlist[attrs.get('text', "")] = "foo"
+               for x in ["text", "title", "value", "caption"]:
+                       try:
+                               attrlist.add((attrs[x], self.last_comment))
+                               self.last_comment = None
+                       except KeyError:
+                               pass
 
 parser = make_parser()
 
-attrlist = {}          
+attrlist = set()
 
 contentHandler = parseXML(attrlist)
 parser.setContentHandler(contentHandler)
+if not no_comments:
+       parser.setProperty(property_lexical_handler, contentHandler)
 
-dir = os.listdir(sys.argv[1])
-for x in dir:
-       if (str(x[-4:]) == ".xml"):
-               parser.parse(sys.argv[1] + str(x))
+for arg in sys.argv[1:]:
+       if os.path.isdir(arg):
+               for file in os.listdir(arg):
+                       if (file.endswith(".xml")):
+                               parser.parse(os.path.join(arg, file))
+       else:
+               parser.parse(arg)
 
-#parser.parse(sys.argv[1])
+       attrlist = list(attrlist)
+       attrlist.sort(key=lambda a: a[0])
 
-for k, v in attrlist.items():
-       print
-       print '#: ' + sys.argv[1]
-       print 'msgid "' + str(k) + '"'
-       print 'msgstr ""'
+       for (k,c) in attrlist:
+               print
+               print '#: ' + arg
+               string.replace(k, "\\n", "\"\n\"")
+               if c:
+                       for l in c.split('\n'):
+                               print "#. ", l
+               if str(k).strip() != "":
+                       print 'msgid "' + str(k) + '"'
+                       print 'msgstr ""'
 
+       attrlist = set()