From 81edce2eb857306082d3a93695a70ce70a9a8b28 Mon Sep 17 00:00:00 2001
From: Yorick Peterse <yorickpeterse@gmail.com>
Date: Fri, 15 Aug 2014 20:36:40 +0200
Subject: [PATCH] Fixed lexing of XML comments.

The previous setup would consume too much. For example the following HTML:

    <a><!--foo--><b><!--bar--></b></a>

would result in the following T_COMMENT token:

    "foo--><b><!--bar"

The new setup requires the marking of a start position. I'm not a huge fan of
this but there doesn't appear to be a way around this.
---
 ext/c/lexer.rl                      | 11 ++++++-----
 ext/java/org/liboga/xml/Lexer.rl    | 11 ++++++-----
 ext/ragel/base_lexer.rl             | 27 ++++++++++++++++++++++-----
 spec/oga/xml/lexer/comments_spec.rb | 13 +++++++++++++
 4 files changed, 47 insertions(+), 15 deletions(-)

diff --git a/ext/c/lexer.rl b/ext/c/lexer.rl
index cada319..6505a7e 100644
--- a/ext/c/lexer.rl
+++ b/ext/c/lexer.rl
@@ -77,11 +77,12 @@ VALUE oga_xml_lexer_advance(VALUE self, VALUE data_block)
 
     char *data_str_val = StringValuePtr(data_block);
 
-    const char *p   = data_str_val;
-    const char *pe  = data_str_val + strlen(data_str_val);
-    const char *eof = pe;
-    const char *ts  = 0;
-    const char *te  = 0;
+    const char *p    = data_str_val;
+    const char *pe   = data_str_val + strlen(data_str_val);
+    const char *eof  = pe;
+    const char *ts   = 0;
+    const char *te   = 0;
+    const char *mark = 0;
 
     int act = NUM2INT(oga_ivar_get(self, "@act"));
     int cs  = NUM2INT(oga_ivar_get(self, "@cs"));
diff --git a/ext/java/org/liboga/xml/Lexer.rl b/ext/java/org/liboga/xml/Lexer.rl
index 8300972..9449a34 100644
--- a/ext/java/org/liboga/xml/Lexer.rl
+++ b/ext/java/org/liboga/xml/Lexer.rl
@@ -90,11 +90,12 @@ public class Lexer extends RubyObject
 
         byte[] data = rb_str.getBytes();
 
-        int ts  = 0;
-        int te  = 0;
-        int p   = 0;
-        int pe  = data.length;
-        int eof = data.length;
+        int ts   = 0;
+        int te   = 0;
+        int p    = 0;
+        int mark = 0;
+        int pe   = data.length;
+        int eof  = data.length;
 
         %% write exec;
 
diff --git a/ext/ragel/base_lexer.rl b/ext/ragel/base_lexer.rl
index aa49c5b..c095a00 100644
--- a/ext/ragel/base_lexer.rl
+++ b/ext/ragel/base_lexer.rl
@@ -47,7 +47,27 @@
     # such as `--` and `->`. Putting extra checks in for these sequences would
     # actually make the rules/actions more complex.
     #
-    comment = '<!--' any* '-->';
+
+    comment_start = '<!--';
+    comment_end   = '-->';
+
+    action start_comment {
+        mark = ts + 4;
+
+        fnext comment_body;
+    }
+
+    comment_body := |*
+        comment_end => {
+            callback("on_comment", data, encoding, mark, te - 3);
+
+            mark = 0;
+
+            fnext main;
+        };
+
+        any;
+    *|;
 
     # CDATA
     #
@@ -215,10 +235,7 @@
     main := |*
         doctype_start  => start_doctype;
         xml_decl_start => start_xml_decl;
-
-        comment => {
-            callback("on_comment", data, encoding, ts + 4, te - 3);
-        };
+        comment_start  => start_comment;
 
         cdata => {
             callback("on_cdata", data, encoding, ts + 9, te - 3);
diff --git a/spec/oga/xml/lexer/comments_spec.rb b/spec/oga/xml/lexer/comments_spec.rb
index 0c3176e..34cb54b 100644
--- a/spec/oga/xml/lexer/comments_spec.rb
+++ b/spec/oga/xml/lexer/comments_spec.rb
@@ -36,5 +36,18 @@ describe Oga::XML::Lexer do
         [:T_COMMENT, '', 1]
       ]
     end
+
+    example 'lex two comments following each other' do
+      lex('<a><!--foo--><b><!--bar--></b></a>').should == [
+        [:T_ELEM_START, nil, 1],
+        [:T_ELEM_NAME, 'a', 1],
+        [:T_COMMENT, 'foo', 1],
+        [:T_ELEM_START, nil, 1],
+        [:T_ELEM_NAME, 'b', 1],
+        [:T_COMMENT, 'bar', 1],
+        [:T_ELEM_END, nil, 1],
+        [:T_ELEM_END, nil, 1]
+      ]
+    end
   end
 end