From 8d8e23320e48f1d8fd98c3c914696f6fe0f7161e Mon Sep 17 00:00:00 2001
From: DEC05EBA <dec05eba@protonmail.com>
Date: Wed, 1 Jan 2020 09:13:18 +0100
Subject: Ignore comments, ignore end tags without a start tag.

Fixes tags closing too soon
---
 src/HtmlParser.c | 42 +++++++++++++++++++++++++++++++++---------
 1 file changed, 33 insertions(+), 9 deletions(-)

(limited to 'src')

diff --git a/src/HtmlParser.c b/src/HtmlParser.c
index 81104b7..7c91a77 100644
--- a/src/HtmlParser.c
+++ b/src/HtmlParser.c
@@ -125,7 +125,8 @@ static char html_parser_peek_char(HtmlParser *self) {
 }
 
 static void html_parser_advance_char(HtmlParser *self) {
-    ++self->offset;
+    if(self->offset < self->source_len)
+        ++self->offset;
 }
 
 static int is_alpha(char c) {
@@ -248,7 +249,7 @@ static void html_parser_goto_script_end_tag(HtmlParser *self) {
         if(c == '"' || c == '\'') {
             html_parser_advance_char(self);
             html_parser_goto_end_of_js_string(self, c);
-        } else if(c == '<' && self->offset + 7 < self->source_len && strncmp(self->source + self->offset + 1, "/script", 7) == 0) {
+        } else if(c == '<' && self->offset + 7 < self->source_len && memcmp(self->source + self->offset + 1, "/script", 7) == 0) {
             self->text.size = (self->source + self->offset) - self->text.data;
             strip(self->text.data, self->text.size, &self->text.data, &self->text.size, is_whitespace);
             self->offset += 7;
@@ -276,6 +277,16 @@ static void html_parser_goto_script_end_tag(HtmlParser *self) {
         self->parse_callback(self, HTML_PARSE_JAVASCRIPT_CODE, self->callback_userdata);
 }
 
+static void html_parser_goto_comment_end(HtmlParser *self) {
+    for(;;) {
+        if(self->source_len - self->offset >= 3 && memcmp(self->source + self->offset, "-->", 3) == 0) {
+            self->offset += 3;
+            break;
+        }
+        html_parser_advance_char(self);
+    }
+}
+
 static void html_parser_parse_tag_start(HtmlParser *self) {
     int tag_name_found = 0;
     for(;;) {
@@ -340,6 +351,10 @@ static void html_parser_parse_tag_start(HtmlParser *self) {
                 /* tag name */
                 self->tag_name = identifier;
                 tag_name_found = 1;
+                if(self->tag_name.size == 3 && memcmp(self->tag_name.data, "!--", 3) == 0) {
+                    html_parser_goto_comment_end(self);
+                    return;
+                }
                 self->is_tag_void = is_void_tag(&self->tag_name);
                 if(!self->is_tag_void) {
                     html_parser_try_append_unclosed_tag(self, self->tag_name.data, self->tag_name.size);
@@ -373,20 +388,29 @@ static void html_parser_parse_tag_end(HtmlParser *self) {
                 }
             }
             tag_end_name.size = (self->source + self->offset) - tag_end_name.data;
+            tag_name_found = 1;
 
             /* void tags close themselves, this is probably invalid html but we choose to ignore it silently */
             if(is_void_tag(&tag_end_name)) {
-                fprintf(stderr, "Warning: got end tag for void tag '%.*s'\n", tag_end_name.size, tag_end_name.data);
+                fprintf(stderr, "Warning: got end tag for void tag '%.*s'\n", (int)tag_end_name.size, tag_end_name.data);
                 continue;
             }
 
-            HtmlStringView top_unclosed_tag;
-            while(html_parser_try_get_top_unclosed_tag(self, &top_unclosed_tag)) {
-                self->tag_name = top_unclosed_tag;
-                self->parse_callback(self, HTML_PARSE_TAG_END, self->callback_userdata);
-                html_parser_pop_unclosed_tag(self);
-                if(string_view_equals(&top_unclosed_tag, &tag_end_name))
+            ssize_t found_start_tag_index = -1;
+            for(ssize_t i = self->unclosed_tags_offset - 1; i >= 0; --i) {
+                if(string_view_equals(&self->unclosed_tags[i], &tag_end_name)) {
+                    found_start_tag_index = i;
                     break;
+                }
+            }
+
+            if(found_start_tag_index != -1) {
+                for(; self->unclosed_tags_offset > (size_t)found_start_tag_index; --self->unclosed_tags_offset) {
+                    self->tag_name = self->unclosed_tags[self->unclosed_tags_offset - 1];
+                    self->parse_callback(self, HTML_PARSE_TAG_END, self->callback_userdata);
+                }
+            } else {
+                fprintf(stderr, "Warning: start tag not found for end tag '%.*s'\n", (int)tag_end_name.size, tag_end_name.data);
             }
         } else if(c == '\0') {
             return;
-- 
cgit v1.2.3