Fix two parser bugs involving wikitable error handling.

earwig · Dec 5, 2015 · 61b6b98 · 61b6b98
1 parent 2ec35e2
commit 61b6b98
Show file tree

Hide file tree

Showing 5 changed files with 29 additions and 12 deletions.
diff --git a/CHANGELOG b/CHANGELOG
@@ -1,6 +1,8 @@
 v0.5 (unreleased):
 
--
+- Fixed parsing bugs involving:
+  - wikitables nested in templates;
+  - wikitable error recovery when unable to recurse.
 
 v0.4.3 (released October 29, 2015):
 

diff --git a/docs/changelog.rst b/docs/changelog.rst
@@ -7,7 +7,10 @@ v0.5
 Unreleased
 (`changes <https://github.com/earwig/mwparserfromhell/compare/v0.4.3...develop>`__):
 
--
+- Fixed parsing bugs involving:
+
+  - wikitables nested in templates;
+  - wikitable error recovery when unable to recurse.
 
 v0.4.3
 ------

diff --git a/mwparserfromhell/parser/ctokenizer/tok_parse.c b/mwparserfromhell/parser/ctokenizer/tok_parse.c
@@ -2190,7 +2190,7 @@ static PyObject* Tokenizer_handle_table_style(Tokenizer* self, Unicode end_token
 */
 static int Tokenizer_parse_table(Tokenizer* self)
 {
-    Py_ssize_t reset = self->head + 1;
+    Py_ssize_t reset = self->head;
     PyObject *style, *padding;
     PyObject *table = NULL;
     self->head += 2;
@@ -2201,7 +2201,7 @@ static int Tokenizer_parse_table(Tokenizer* self)
     if (BAD_ROUTE) {
         RESET_ROUTE();
         self->head = reset;
-        if (Tokenizer_emit_text(self, "{|"))
+        if (Tokenizer_emit_char(self, '{'))
             return -1;
         return 0;
     }
@@ -2220,7 +2220,7 @@ static int Tokenizer_parse_table(Tokenizer* self)
         Py_DECREF(padding);
         Py_DECREF(style);
         self->head = reset;
-        if (Tokenizer_emit_text(self, "{|"))
+        if (Tokenizer_emit_char(self, '{'))
             return -1;
         return 0;
     }
@@ -2689,10 +2689,8 @@ PyObject* Tokenizer_parse(Tokenizer* self, uint64_t context, int push)
                 if (Tokenizer_parse_table(self))
                     return NULL;
             }
-            else if (Tokenizer_emit_char(self, this) || Tokenizer_emit_char(self, next))
+            else if (Tokenizer_emit_char(self, this))
                 return NULL;
-            else
-                self->head++;
         }
         else if (this_context & LC_TABLE_OPEN) {
             if (this == '|' && next == '|' && this_context & LC_TABLE_TD_LINE) {

diff --git a/mwparserfromhell/parser/tokenizer.py b/mwparserfromhell/parser/tokenizer.py
@@ -1074,14 +1074,14 @@ def _handle_table_style(self, end_token):
 
     def _parse_table(self):
         """Parse a wikicode table by starting with the first line."""
-        reset = self._head + 1
+        reset = self._head
         self._head += 2
         self._push(contexts.TABLE_OPEN)
         try:
             padding = self._handle_table_style("\n")
         except BadRoute:
             self._head = reset
-            self._emit_text("{|")
+            self._emit_text("{")
             return
         style = self._pop()
 
@@ -1090,7 +1090,7 @@ def _parse_table(self):
             table = self._parse(contexts.TABLE_OPEN)
         except BadRoute:
             self._head = reset
-            self._emit_text("{|")
+            self._emit_text("{")
             return
 
         self._emit_table_tag("{|", "table", style, padding, None, table, "|}")
@@ -1352,7 +1352,7 @@ def _parse(self, context=0, push=True):
                 if self._can_recurse():
                     self._parse_table()
                 else:
-                    self._emit_text("{|")
+                    self._emit_text("{")
             elif self._context & contexts.TABLE_OPEN:
                 if this == next == "|" and self._context & contexts.TABLE_TD_LINE:
                     if self._context & contexts.TABLE_CELL_OPEN:

diff --git a/tests/tokenizer/integration.mwtest b/tests/tokenizer/integration.mwtest
@@ -332,3 +332,17 @@ name:   wikilink_to_external_link_fallback_2
 label:  an external link enclosed in an extra pair of brackets (see issue #120)
 input:  "[[http://example.com]]"
 output: [Text(text="["), ExternalLinkOpen(brackets=True), Text(text="http://example.com"), ExternalLinkClose(), Text(text="]")]
+
+---
+
+name:   tables_in_templates
+label:  catch error handling mistakes when wikitables are inside templates
+input:  "{{hello|test\n{|\n|} }}"
+output: [TemplateOpen(), Text(text="hello"), TemplateParamSeparator(), Text(text="test\n"), TagOpenOpen(wiki_markup="{|"), Text(text="table"), TagCloseOpen(padding="\n"), TagOpenClose(wiki_markup="|}"), Text(text="table"), TagCloseClose(), Text(text=" "), TemplateClose()]
+
+---
+
+name:   tables_in_templates_2
+label:  catch error handling mistakes when wikitables are inside templates
+input:  "{{hello|test\n{|\n| }}"
+output: [TemplateOpen(), Text(text="hello"), TemplateParamSeparator(), Text(text="test\n{"), TemplateParamSeparator(), Text(text="\n"), TemplateParamSeparator(), Text(text=" "), TemplateClose()]