From 667a9e8f466f5c97b962b6885d575e6b01405542 Mon Sep 17 00:00:00 2001 From: DEC05EBA Date: Wed, 1 Jan 2020 18:14:32 +0100 Subject: Do not skip whitespaces in text in pre tags --- src/HtmlParser.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) (limited to 'src/HtmlParser.c') diff --git a/src/HtmlParser.c b/src/HtmlParser.c index 7c91a77..90c1149 100644 --- a/src/HtmlParser.c +++ b/src/HtmlParser.c @@ -93,6 +93,7 @@ static void html_parser_reset(HtmlParser *self) { self->text.size = 0; self->is_tag_void = 0; self->inside_script_tag = 0; + self->pre_tag_depth = 0; self->unclosed_tags_offset = 0; } @@ -104,7 +105,7 @@ void html_parser_init(HtmlParser *self, const char *html_source, size_t len, Htm } void html_parser_deinit(HtmlParser *self) { - + (void)self; } static char html_parser_next_char(HtmlParser *self) { @@ -354,6 +355,8 @@ static void html_parser_parse_tag_start(HtmlParser *self) { if(self->tag_name.size == 3 && memcmp(self->tag_name.data, "!--", 3) == 0) { html_parser_goto_comment_end(self); return; + } else if(self->tag_name.size == 3 && memcmp(self->tag_name.data, "pre", 3) == 0) { + ++self->pre_tag_depth; } self->is_tag_void = is_void_tag(&self->tag_name); if(!self->is_tag_void) { @@ -409,6 +412,10 @@ static void html_parser_parse_tag_end(HtmlParser *self) { self->tag_name = self->unclosed_tags[self->unclosed_tags_offset - 1]; self->parse_callback(self, HTML_PARSE_TAG_END, self->callback_userdata); } + + if(self->tag_name.size == 3 && memcmp(self->tag_name.data, "pre", 3) == 0) { + --self->pre_tag_depth; + } } else { fprintf(stderr, "Warning: start tag not found for end tag '%.*s'\n", (int)tag_end_name.size, tag_end_name.data); } @@ -443,7 +450,8 @@ void html_parser_parse(HtmlParser *self) { html_parser_advance_char(self); } 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); + if(self->pre_tag_depth == 0) + strip(self->text.data, self->text.size, &self->text.data, &self->text.size, is_whitespace); if(self->text.size > 0) self->parse_callback(self, HTML_PARSE_TEXT, self->callback_userdata); } -- cgit v1.2.3