[Zope-Checkins] CVS: Zope/lib/python/Products/PluginIndexes/TextIndex/Splitter/ZopeSplitter/src - ZopeSplitter.c:1.2.10.6

[Andreas Jung]

Update of /cvs-repository/Zope/lib/python/Products/PluginIndexes/TextIndex/Splitter/ZopeSplitter/src
In directory cvs.zope.org:/tmp/cvs-serv4474/src

Modified Files:
      Tag: ajung-unicode
	ZopeSplitter.c 
Log Message:
fixed problem with string encoded != ascii


=== Zope/lib/python/Products/PluginIndexes/TextIndex/Splitter/ZopeSplitter/src/ZopeSplitter.c 1.2.10.5 => 1.2.10.6 ===
         int start=0;
 
-#ifdef DEBUG
-        puts("before List_New");
-        fflush(stdout);
-#endif
-      
         self->list = PyList_New(0);
 
-#ifdef DEBUG
-        puts("after List_New");
-        fflush(stdout);
-#endif
-
-
         do {
                 register Py_UNICODE ch;
 
@@ -276,7 +265,7 @@
 }
 
 static PyObject *
-get_Splitter(PyObject *modinfo, PyObject *args)
+get_Splitter(PyObject *modinfo, PyObject *args,PyObject *keywds)
 {
         Splitter *self;
         PyObject *doc, *unicodedoc,*synstop=NULL;
@@ -291,19 +280,14 @@
 #endif
 
         if (PyString_Check(doc)) {
+                // This sux a bit. The default encoding should be ascii or latin1.
+                // But there must be better support to pass an optional encoding parameter
+                
+                unicodedoc = PyUnicode_FromEncodedObject(doc,"latin1","strict");
+                if (! unicodedoc) goto err;
 
-#ifdef DEBUG
-                puts("got a string object");
-                fflush(stdout);
-#endif
-                unicodedoc = PyUnicode_FromObject(doc);
 
         } else if( PyUnicode_Check(doc)) {
-#ifdef DEBUG
-                puts("got a unicode object");
-                fflush(stdout);
-
-#endif
                 unicodedoc = doc;
         } else {
                 PyErr_SetString(PyExc_TypeError, "first argument is neither string nor unicode.");
@@ -313,20 +297,9 @@
         if (synstop) {
                 self->synstop = synstop;
                 Py_INCREF(synstop);
-        } else {
-                self->synstop=NULL;
-        }
-
-#ifdef DEBUG
-        puts("before splitUnicodeString");
-        PyObject_Print(unicodedoc,stdout,0);
-#endif
+        } else  self->synstop=NULL;
 
         splitUnicodeString(self,(PyUnicodeObject *)unicodedoc);
-
-#ifdef DEBUG
-        puts("after splitUnicodeString");
-#endif
 
         return (PyObject*)self;