]> granicus.if.org Git - python/commitdiff
Issue #2892: preserve iterparse events in case of SyntaxError
authorFlorent Xicluna <florent.xicluna@gmail.com>
Tue, 1 Nov 2011 22:34:41 +0000 (23:34 +0100)
committerFlorent Xicluna <florent.xicluna@gmail.com>
Tue, 1 Nov 2011 22:34:41 +0000 (23:34 +0100)
Lib/test/test_xml_etree.py
Lib/xml/etree/ElementTree.py
Modules/_elementtree.c

index 26e8eab796f17926d42ebd7c2e44e71fc84f60dd..1b66a8988f4d4b7312455b029abf42a515d95001 100644 (file)
@@ -739,6 +739,7 @@ def iterparse():
     ... except ET.ParseError, v:
     ...   print v
     junk after document element: line 1, column 12
+    end document
     """
 
 def writefile():
index 19862e2d34b2872da6264d04c2dc5732a226e483..2b0cf0c783398e152d6291e474875e1fd64a2c09 100644 (file)
@@ -1210,6 +1210,7 @@ class _IterParseIterator(object):
         self._close_file = close_source
         self._events = []
         self._index = 0
+        self._error = None
         self.root = self._root = None
         self._parser = parser
         # wire up the parser for event reporting
@@ -1255,24 +1256,31 @@ class _IterParseIterator(object):
         while 1:
             try:
                 item = self._events[self._index]
+                self._index += 1
+                return item
             except IndexError:
-                if self._parser is None:
-                    self.root = self._root
-                    if self._close_file:
-                        self._file.close()
-                    raise StopIteration
-                # load event buffer
-                del self._events[:]
-                self._index = 0
-                data = self._file.read(16384)
-                if data:
+                pass
+            if self._error:
+                e = self._error
+                self._error = None
+                raise e
+            if self._parser is None:
+                self.root = self._root
+                if self._close_file:
+                    self._file.close()
+                raise StopIteration
+            # load event buffer
+            del self._events[:]
+            self._index = 0
+            data = self._file.read(16384)
+            if data:
+                try:
                     self._parser.feed(data)
-                else:
-                    self._root = self._parser.close()
-                    self._parser = None
+                except SyntaxError as exc:
+                    self._error = exc
             else:
-                self._index = self._index + 1
-                return item
+                self._root = self._parser.close()
+                self._parser = None
 
     def __iter__(self):
         return self
index c368e1456bda10a84e0055d608c649034c62138b..379aa01d0a57c92f474457a9fc51949122b4d04f 100644 (file)
@@ -2970,6 +2970,7 @@ init_elementtree(void)
         "  self._file = file\n"
         "  self._events = []\n"
         "  self._index = 0\n"
+        "  self._error = None\n"
         "  self.root = self._root = None\n"
         "  b = cElementTree.TreeBuilder()\n"
         "  self._parser = cElementTree.XMLParser(b)\n"
@@ -2978,24 +2979,31 @@ init_elementtree(void)
         "  while 1:\n"
         "    try:\n"
         "      item = self._events[self._index]\n"
+        "      self._index += 1\n"
+        "      return item\n"
         "    except IndexError:\n"
-        "      if self._parser is None:\n"
-        "        self.root = self._root\n"
-        "        if self._close_file:\n"
-        "          self._file.close()\n"
-        "        raise StopIteration\n"
-        "      # load event buffer\n"
-        "      del self._events[:]\n"
-        "      self._index = 0\n"
-        "      data = self._file.read(16384)\n"
-        "      if data:\n"
+        "      pass\n"
+        "    if self._error:\n"
+        "      e = self._error\n"
+        "      self._error = None\n"
+        "      raise e\n"
+        "    if self._parser is None:\n"
+        "      self.root = self._root\n"
+        "      if self._close_file:\n"
+        "        self._file.close()\n"
+        "      raise StopIteration\n"
+        "    # load event buffer\n"
+        "    del self._events[:]\n"
+        "    self._index = 0\n"
+        "    data = self._file.read(16384)\n"
+        "    if data:\n"
+        "      try:\n"
         "        self._parser.feed(data)\n"
-        "      else:\n"
-        "        self._root = self._parser.close()\n"
-        "        self._parser = None\n"
+        "      except SyntaxError as exc:\n"
+        "        self._error = exc\n"
         "    else:\n"
-        "      self._index = self._index + 1\n"
-        "      return item\n"
+        "      self._root = self._parser.close()\n"
+        "      self._parser = None\n"
         " def __iter__(self):\n"
         "  return self\n"
         "cElementTree.iterparse = iterparse\n"