Tried to allow the parser to work under Python 2.X and SAX2

svn: r23

gramps/src/GrampsParser.py

@@ -26,6 +26,19 @@ import os
 
 from xml.sax import handler
 
+#-------------------------------------------------------------------------
+#
+# Try to abstract SAX1 from SAX2
+#
+#-------------------------------------------------------------------------
+try:
+    import xml.sax.saxexts
+    sax = 1
+except:
+    sax = 2
+
+from latin_utf8 import utf8_to_latin
+
 #-------------------------------------------------------------------------
 #
 # Remove extraneous spaces
@@ -201,7 +214,7 @@ class GrampsParser(handler.ContentHandler):
     #---------------------------------------------------------------------
     def start_bmark(self,attrs):
         if self.is_import:
-            person = self.database.findPerson("x" + attrs["ref"],self.pmap)
+            person = self.database.findPerson("x%s" % attrs["ref"],self.pmap)
         else:
             person = self.database.findPersonNoMap(attrs["ref"])
         self.database.bookmarks.append(person)
@@ -216,7 +229,7 @@ class GrampsParser(handler.ContentHandler):
             self.callback(float(self.count)/float(self.entries))
         self.count = self.count + 1
         if self.is_import:
-            self.person = self.database.findPerson("x" + attrs["id"],self.pmap)
+            self.person = self.database.findPerson("x%s" % attrs["id"],self.pmap)
         else:
             self.person = self.database.findPersonNoMap(attrs["id"])
 
@@ -239,7 +252,7 @@ class GrampsParser(handler.ContentHandler):
     #---------------------------------------------------------------------
     def start_father(self,attrs):
         if self.is_import:
-            father = self.database.findPerson("x" + attrs["ref"],self.pmap)
+            father = self.database.findPerson("x%s" % attrs["ref"],self.pmap)
         else:
             father = self.database.findPersonNoMap(attrs["ref"])
         self.family.setFather(father)
@@ -251,7 +264,7 @@ class GrampsParser(handler.ContentHandler):
     #---------------------------------------------------------------------
     def start_mother(self,attrs):
         if self.is_import:
-            mother = self.database.findPerson("x" + attrs["ref"],self.pmap)
+            mother = self.database.findPerson("x%s" % attrs["ref"],self.pmap)
         else:
             mother = self.database.findPersonNoMap(attrs["ref"])
         self.family.setMother(mother)
@@ -263,7 +276,7 @@ class GrampsParser(handler.ContentHandler):
     #---------------------------------------------------------------------
     def start_child(self,attrs):
         if self.is_import:
-            child = self.database.findPerson("x" + attrs["ref"],self.pmap)
+            child = self.database.findPerson("x%s" % attrs["ref"],self.pmap)
         else:
             child = self.database.findPersonNoMap(attrs["ref"])
         self.family.addChild(child)
@@ -277,14 +290,16 @@ class GrampsParser(handler.ContentHandler):
 
         if not attrs.has_key("href"):
             return
-        if attrs.has_key("description"):
+        try:
             desc = attrs["description"]
-        else:
+        except KeyError:
             desc = ""
 
+        try:
             url = Url(attrs["href"],desc)
-        
             self.person.addUrl(url)
+        except KeyError:
+            return
 
     #---------------------------------------------------------------------
     #
@@ -753,12 +768,10 @@ class GrampsParser(handler.ContentHandler):
     #---------------------------------------------------------------------
     def startElement(self,tag,attrs):
 
-        self.data[tag] = ""
         self.active = tag
-
+        self.data[tag] = ""
         if GrampsParser.start.has_key(tag):
             GrampsParser.start[tag](self,attrs)
-            return
 
     #---------------------------------------------------------------------
     #
@@ -766,16 +779,23 @@ class GrampsParser(handler.ContentHandler):
     #
     #---------------------------------------------------------------------
     def endElement(self,tag):
-        import latin_utf8
 
         if GrampsParser.stop.has_key(tag):
-            data = latin_utf8.utf8_to_latin(self.data[tag])
+            if sax == 1:
+                data = utf8_to_latin(self.data[tag])
+            else:
+                data = self.data[tag]
             GrampsParser.stop[tag](self,data)
 
+    if sax == 1:
+        def characters(self, data, offset, length):
+            self.data[self.active] = self.data[self.active] + data
+    else:
+        def characters(self, data):
+            self.data[self.active] = self.data[self.active] + data
+
     #---------------------------------------------------------------------
     #
     #
     #
     #---------------------------------------------------------------------
-    def characters(self, data, offset, length):
-        self.data[self.active] = self.data[self.active] + data

gramps/src/ReadXML.py

@@ -21,6 +21,8 @@
 from RelLib import *
 from GrampsParser import *
 import intl
+_ = intl.gettext
+
 
 import string
 import time
@@ -29,11 +31,19 @@ import os
 from gnome.ui import *
 
 import xml.sax
-import xml.sax.saxexts
 import xml.sax.saxutils
-import xml.parsers.expat
 
-_ = intl.gettext
+#-------------------------------------------------------------------------
+#
+# Try to abstract SAX1 from SAX2
+#
+#-------------------------------------------------------------------------
+try:
+    import xml.sax.saxexts
+    sax = 1
+except:
+    from codecs import *
+    sax = 2
 
 #-------------------------------------------------------------------------
 #
@@ -43,15 +53,23 @@ _ = intl.gettext
 #-------------------------------------------------------------------------
 def importData(database, filename, callback):
 
-    parser = xml.sax.saxexts.make_parser()
     basefile = os.path.dirname(filename)
     database.smap = {}
     database.pmap = {}
     database.fmap = {}
-    parser.setDocumentHandler(GrampsParser(database,callback,basefile,1))
-    xml_file = gzip.open(filename,"rb")
 
+    if sax == 1:
+        parser = xml.sax.saxexts.make_parser()
+        parser.setDocumentHandler(GrampsParser(database,callback,basefile,1))
+        parser.setErrorHandler(xml.sax.saxutils.ErrorRaiser())
+        xml_file = gzip.open(filename,"rb")
         parser.parseFile(xml_file)
+    else:
+        parser = xml.sax.make_parser()
+        parser.setContentHandler(GrampsParser(database,callback,basefile,1))
+        xml_file = EncodedFile(gzip.open(filename,"rb"),'utf-8','latin-1')
+        parser.parse(xml_file)
+        
     xml_file.close()
 
 #-------------------------------------------------------------------------
@@ -62,19 +80,26 @@ def importData(database, filename, callback):
 #-------------------------------------------------------------------------
 def loadData(database, filename, callback):
 
-    parser = xml.sax.saxexts.make_parser()
-
     basefile = os.path.dirname(filename)
     database.smap = {}
     database.pmap = {}
     database.fmap = {}
-    parser.setErrorHandler(xml.sax.saxutils.ErrorRaiser())
-    parser.setDocumentHandler(GrampsParser(database,callback,basefile,0))
 
     filename = os.path.normpath(filename)
 
+    if sax == 1:
+        parser = xml.sax.saxexts.make_parser()
+        parser.setDocumentHandler(GrampsParser(database,callback,basefile,0))
+        parser.setErrorHandler(xml.sax.saxutils.ErrorRaiser())
+    else:
+        parser = xml.sax.make_parser()
+        parser.setContentHandler(GrampsParser(database,callback,basefile,0))
+
     try:
+        if sax == 1:
             xml_file = gzip.open(filename,"rb")
+        else:
+            xml_file = EncodedFile(gzip.open(filename,"rb"),'utf-8','latin-1')
     except IOError,msg:
         GnomeErrorDialog(filename + _(" could not be opened\n") + str(msg))
         return 0
@@ -83,13 +108,13 @@ def loadData(database, filename, callback):
         return 0
         
     try:
+        if sax == 1:
             parser.parseFile(xml_file)
+        else:
+            parser.parse(xml_file)
     except xml.sax.SAXParseException:
         GnomeErrorDialog(filename + _(" is a corrupt file"))
         return 0
-    except xml.parsers.expat.ExpatError:
-        GnomeErrorDialog(filename + _(" is a corrupt file"))
-        return 0
     except IOError,msg:
         GnomeErrorDialog(filename + _(" is not a valid gramps file\n") + \
                          str(msg))
@@ -98,11 +123,28 @@ def loadData(database, filename, callback):
         GnomeErrorDialog(_("Could not read ") + filename)
         return 0
 
+
     xml_file.close()
     return 1
 
 
-
+if __name__ == "__main__":
+
+    import sys
+    import time
+    import profile
+    
+    def lcb(val):
+        pass
+    
+    db = RelDataBase()
+    file = sys.argv[1]
+
+    t1 = time.time()
+
+    loadData(db,file,lcb)
+    t2 = time.time()
+    print t2 - t1
 
 
 

gramps/src/latin_utf8.py

@@ -1,25 +1,15 @@
 
-
-from xml.unicode.utf8_iso import utf8_to_code, code_to_utf8
-
 import cStringIO
 
-def utf8_to_latin(s):
+try:
-    buff = cStringIO.StringIO()
+    from xml.unicode.utf8_iso import code_to_utf8
-    while s:
+    from xml.unicode.iso8859 import UTF8String
-        try:
-            head,s = utf8_to_code(1,s)
-        except Exception,e:
-            from traceback import print_exc
-            print_exc()
-            head = ''
-            s = s[1:0]
-        buff.write(head)
-    ans = buff.getvalue()
-    buff.close()
-    return ans
 
-def latin_to_utf8(s):
+    def utf8_to_latin(s):
+        y = UTF8String(s)
+        return y.encode("iso-8859-1")
+
+    def latin_to_utf8(s):
         buff = cStringIO.StringIO()
         for c in s:
             try:
@@ -32,3 +22,13 @@ def latin_to_utf8(s):
         ans = buff.getvalue()
         buff.close()
         return ans
+
+except:
+    
+    def utf8_to_latin(s):
+        return s
+
+    def latin_to_utf8(s):
+        return s
+
+