From 6954480263b537c775c960f9b64e3a9cd4706481 Mon Sep 17 00:00:00 2001
From: Ben Kurtovic <ben.kurtovic@gmail.com>
Date: Thu, 10 Jul 2014 20:17:45 -0400
Subject: [PATCH] Fix template parsing when comments are inside the name (fixes
 #59).

---
 mwparserfromhell/parser/tokenizer.c  | 17 ++++++++++++++---
 mwparserfromhell/parser/tokenizer.py |  8 ++++++++
 tests/tokenizer/integration.mwtest   | 14 ++++++++++++++
 3 files changed, 36 insertions(+), 3 deletions(-)

diff --git a/mwparserfromhell/parser/tokenizer.c b/mwparserfromhell/parser/tokenizer.c
index d62b965..814ad50 100644
--- a/mwparserfromhell/parser/tokenizer.c
+++ b/mwparserfromhell/parser/tokenizer.c
@@ -1553,6 +1553,12 @@ static int Tokenizer_parse_comment(Tokenizer* self)
                 return -1;
             Py_DECREF(comment);
             self->head += 2;
+            if (self->topstack->context & LC_FAIL_NEXT) {
+                /* _verify_safe() sets this flag while parsing a template name
+                   when it encounters what might be a comment -- we must unset
+                   it to let _verify_safe() know it was correct: */
+                self->topstack->context ^= LC_FAIL_NEXT;
+            }
             return 0;
         }
         if (Tokenizer_emit_char(self, this))
@@ -2478,8 +2484,13 @@ static int Tokenizer_verify_safe(Tokenizer* self, int context, Py_UNICODE data)
             return 0;
         if (context & LC_HAS_TEXT) {
             if (context & LC_FAIL_ON_TEXT) {
-                if (!Py_UNICODE_ISSPACE(data))
+                if (!Py_UNICODE_ISSPACE(data)) {
+                    if (data == '<' && Tokenizer_READ(self, 1) == '!') {
+                        self->topstack->context |= LC_FAIL_NEXT;
+                        return 0;
+                    }
                     return -1;
+                }
             }
             else {
                 if (data == '\n')
@@ -2496,8 +2507,8 @@ static int Tokenizer_verify_safe(Tokenizer* self, int context, Py_UNICODE data)
             }
         }
         else if (context & LC_FAIL_ON_LBRACE) {
-            if (data == '{' || (Tokenizer_READ(self, -1) == '{' &&
-                                 Tokenizer_READ(self, -2) == '{')) {
+            if (data == '{' || (Tokenizer_READ_BACKWARDS(self, 1) == '{' &&
+                                 Tokenizer_READ_BACKWARDS(self, 2) == '{')) {
                 if (context & LC_TEMPLATE)
                     self->topstack->context |= LC_FAIL_ON_EQUALS;
                 else
diff --git a/mwparserfromhell/parser/tokenizer.py b/mwparserfromhell/parser/tokenizer.py
index d867234..44f0d60 100644
--- a/mwparserfromhell/parser/tokenizer.py
+++ b/mwparserfromhell/parser/tokenizer.py
@@ -608,6 +608,11 @@ class Tokenizer(object):
                 self._emit(tokens.CommentEnd())
                 self._emit_all(self._pop())
                 self._head += 2
+                if self._context & contexts.FAIL_NEXT:
+                    # _verify_safe() sets this flag while parsing a template
+                    # name when it encounters what might be a comment -- we
+                    # must unset it to let _verify_safe() know it was correct:
+                    self._context ^= contexts.FAIL_NEXT
                 return
             self._emit_text(this)
             self._head += 1
@@ -1021,6 +1026,9 @@ class Tokenizer(object):
             if context & contexts.HAS_TEXT:
                 if context & contexts.FAIL_ON_TEXT:
                     if this is self.END or not this.isspace():
+                        if this == "<" and self._read(1) == "!":
+                            self._context |= contexts.FAIL_NEXT
+                            return True
                         return False
                 else:
                     if this == "\n":
diff --git a/tests/tokenizer/integration.mwtest b/tests/tokenizer/integration.mwtest
index ef6d5c5..1019175 100644
--- a/tests/tokenizer/integration.mwtest
+++ b/tests/tokenizer/integration.mwtest
@@ -227,3 +227,17 @@ name:   newline_and_comment_in_template_name_5
 label:  a template name containing a newline followed by a comment
 input:  "{{foobar\n<!-- comment -->\ninvalid|key=value}}"
 output: [Text(text="{{foobar\n"), CommentStart(), Text(text=" comment "), CommentEnd(), Text(text="\ninvalid|key=value}}")]
+
+---
+
+name:   newline_and_comment_in_template_name_6
+label:  a template name containing a newline followed by a comment
+input:  "{{foobar\n<!--|key=value}}"
+output: [Text(text="{{foobar\n<!--|key=value}}")]
+
+---
+
+name:   newline_and_comment_in_template_name_7
+label:  a template name containing a newline followed by a comment
+input:  "{{foobar\n<!|key=value}}"
+output: [Text(text="{{foobar\n<!|key=value}}")]