From 22e869b1429dabd30976e4bdb8b819ed240c3f29 Mon Sep 17 00:00:00 2001
From: Ben Kurtovic <ben.kurtovic@verizon.net>
Date: Sun, 19 May 2013 01:45:09 -0400
Subject: [PATCH] Fix a failing HTML entity test in the C tokenizer.

Remove some extraneous whitespace in string_mixin.py.
---
 mwparserfromhell/parser/tokenizer.c | 19 +++++++++++++++++--
 mwparserfromhell/string_mixin.py    |  1 -
 2 files changed, 17 insertions(+), 3 deletions(-)

diff --git a/mwparserfromhell/parser/tokenizer.c b/mwparserfromhell/parser/tokenizer.c
index df0882e..939f30c 100644
--- a/mwparserfromhell/parser/tokenizer.c
+++ b/mwparserfromhell/parser/tokenizer.c
@@ -911,8 +911,8 @@ Tokenizer_really_parse_entity(Tokenizer* self)
 {
     PyObject *token, *kwargs, *textobj;
     Py_UNICODE this;
-    int numeric, hexadecimal, i, j, test;
-    char *valid, *text, *def;
+    int numeric, hexadecimal, i, j, zeroes, test;
+    char *valid, *text, *buffer, *def;
 
     #define FAIL_ROUTE_AND_EXIT() { \
         Tokenizer_fail_route(self); \
@@ -984,6 +984,7 @@ Tokenizer_really_parse_entity(Tokenizer* self)
         return -1;
     }
     i = 0;
+    zeroes = 0;
     while (1) {
         this = Tokenizer_READ(self, 0);
         if (this == *";") {
@@ -992,6 +993,7 @@ Tokenizer_really_parse_entity(Tokenizer* self)
             break;
         }
         if (i == 0 && this == *"0") {
+            zeroes++;
             self->head++;
             continue;
         }
@@ -1029,6 +1031,19 @@ Tokenizer_really_parse_entity(Tokenizer* self)
             i++;
         }
     }
+    if (zeroes) {
+        buffer = calloc(strlen(text) + zeroes + 1, sizeof(char));
+        if (!buffer) {
+            free(text);
+            PyErr_NoMemory();
+            return -1;
+        }
+        for (i = 0; i < zeroes; i++)
+            strcat(buffer, "0");
+        strcat(buffer, text);
+        free(text);
+        text = buffer;
+    }
     textobj = PyUnicode_FromString(text);
     if (!textobj) {
         free(text);
diff --git a/mwparserfromhell/string_mixin.py b/mwparserfromhell/string_mixin.py
index 6bee9c4..89c1bc0 100644
--- a/mwparserfromhell/string_mixin.py
+++ b/mwparserfromhell/string_mixin.py
@@ -40,7 +40,6 @@ def inheritdoc(method):
     method.__doc__ = getattr(str, method.__name__).__doc__
     return method
 
-
 class StringMixIn(object):
     """Implement the interface for ``unicode``/``str`` in a dynamic manner.