Fix parsing of leading zeros in named HTML entities (fixes #288)

earwig · Jun 30, 2022 · f963df7 · f963df7
1 parent 6665505
commit f963df7
Show file tree

Hide file tree

Showing 4 changed files with 11 additions and 3 deletions.
diff --git a/CHANGELOG b/CHANGELOG
@@ -1,6 +1,6 @@
 v0.7 (unreleased):
 
-- ...
+- Fixed parsing of leading zeros in named HTML entities. (#288)
 
 v0.6.4 (released February 14, 2022):
 

diff --git a/docs/changelog.rst b/docs/changelog.rst
@@ -7,7 +7,8 @@ v0.7
 Unreleased
 (`changes <https://github.com/earwig/mwparserfromhell/compare/v0.6.4...main>`__):
 
-- ...
+- Fixed parsing of leading zeros in named HTML entities.
+  (`#288 <https://github.com/earwig/mwparserfromhell/issues/288>`_)
 
 v0.6.4
 ------

diff --git a/src/mwparserfromhell/parser/ctokenizer/tok_parse.c b/src/mwparserfromhell/parser/ctokenizer/tok_parse.c
@@ -1099,7 +1099,7 @@ Tokenizer_really_parse_entity(Tokenizer *self)
             }
             break;
         }
-        if (i == 0 && this == '0') {
+        if (i == 0 && numeric && this == '0') {
             zeroes++;
             self->head++;
             continue;

diff --git a/tests/tokenizer/html_entities.mwtest b/tests/tokenizer/html_entities.mwtest
@@ -156,3 +156,10 @@ name:   invalid_partial_amp_pound_x_semicolon
 label:  invalid entities: an ampersand, pound sign, and x
 input:  "&#x"
 output: [Text(text="&#x")]
+
+---
+
+name:   invalid_zeros_before_named
+label:  invalid entities: zeros before a valid named entity
+input:  "&000nbsp;"
+output: [Text(text="&000nbsp;")]