#include "HTMLTag.h"

StringMapCI<HTMLTag*>   HTMLTag::m_tags; //HTML tags lookup
vector<HTMLTag*>        HTMLTag::m_ids;  //id lookup

//static tag initialisation (enters into a map)
//For inline/block we distinguish between tags that are:
// 1. used always in text in-line just for in-line formatting (converted to formats in the Document class)
// 2. used almost always in text block for layout (converted to \n)
// 3. unknown usage (may require CSS analysis to be sure)
// 4. everything else: commonly used in unwanted markup

//processing instructions
//don't jump these because they don't have end tags
HTMLTag HTMLTag::tag_comment("--", PI, JUMP, BLOCK, NCSS, NORM, "-->"); //does not require a space after name <!--comment ...
HTMLTag HTMLTag::tag_DOCTYPE("DOCTYPE", PI, PARSE, BLOCK, NCSS);        //no end tag so don't jump

//jump tags
HTMLTag HTMLTag::tag_select("select", NORM, JUMP, BLOCK, NCSS);
HTMLTag HTMLTag::tag_noscript("noscript", NORM, JUMP, BLOCK, NCSS);
HTMLTag HTMLTag::tag_object("object", NORM, JUMP, BLOCK, NCSS);
HTMLTag HTMLTag::tag_textarea("textarea", NORM, JUMP, BLOCK, NCSS);
HTMLTag HTMLTag::tag_style("style", NORM, JUMP, BLOCK, NCSS);
HTMLTag HTMLTag::tag_script("script", NORM, JUMP, BLOCK, NCSS);

//block tags
//TREATASUNKNOWN are for performance:
//the default is (..., NORM, PARSE, BLOCK, CSS) and NCSS/CSS is not so important
HTMLTag HTMLTag::tag_head("head", NORM, PARSE, BLOCK, NCSS);
HTMLTag HTMLTag::tag_html("html", NORM, PARSE, BLOCK, NCSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_hr("hr", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_base("base", NORM, PARSE, BLOCK, NCSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_body("body", NORM, PARSE, BLOCK, CSS);
HTMLTag HTMLTag::tag_button("button", NORM, PARSE, BLOCK, NCSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_col("col", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_colgroup("colgroup", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_dd("dd", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_div("div", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_dl("dl", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_dt("dt", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_fieldset("fieldset", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_form("form", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN); //don't jump forms because .NET puts them around everything
HTMLTag HTMLTag::tag_legend("legend", NORM, PARSE, BLOCK, CSS);
HTMLTag HTMLTag::tag_input("input", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_label("label", NORM, PARSE, BLOCK, CSS);
HTMLTag HTMLTag::tag_li("li", NORM, PARSE, BLOCK, CSS);
HTMLTag HTMLTag::tag_map("map", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_meta("meta", NORM, PARSE, BLOCK, NCSS);
HTMLTag HTMLTag::tag_ol("ol", NORM, PARSE, BLOCK, CSS);
HTMLTag HTMLTag::tag_optgroup("optgroup", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_option("option", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_param("param", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_table("table", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_tbody("tbody", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_td("td", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_tfoot("tfoot", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_th("th", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_thead("thead", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_title("title", NORM, PARSE, BLOCK, CSS);
HTMLTag HTMLTag::tag_tr("tr", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_tt("tt", NORM, PARSE, BLOCK, CSS, NORM, 0, TREATASUNKNOWN);
HTMLTag HTMLTag::tag_ul("ul", NORM, PARSE, BLOCK, CSS);

//headings
HTMLTag HTMLTag::tag_h1("h1", NORM, PARSE, BLOCK, CSS, HEADING);
HTMLTag HTMLTag::tag_h2("h2", NORM, PARSE, BLOCK, CSS, HEADING);
HTMLTag HTMLTag::tag_h3("h3", NORM, PARSE, BLOCK, CSS, HEADING);
HTMLTag HTMLTag::tag_h4("h4", NORM, PARSE, BLOCK, CSS, HEADING);
HTMLTag HTMLTag::tag_h5("h5", NORM, PARSE, BLOCK, CSS, HEADING);
HTMLTag HTMLTag::tag_h6("h6", NORM, PARSE, BLOCK, CSS, HEADING);

//maybe tags
HTMLTag HTMLTag::tag_br("br", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_span("span", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_img("img", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_p("p", NORM, PARSE, INLINE, CSS);

//definite inline tags
HTMLTag HTMLTag::tag_u("u", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_s("s", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_strike("strike", NORM, PARSE, INLINE, CSS);

HTMLTag HTMLTag::tag_a("a", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_abbr("abbr", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_acronym("acronym", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_address("address", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_area("area", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_b("b", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_bdo("bdo", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_big("big", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_blink("blink", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_blockquote("blockquote", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_caption("caption", NORM, PARSE, INLINE, CSS); //table summaries
HTMLTag HTMLTag::tag_cite("cite", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_code("code", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_del("del", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_dfn("dfn", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_em("em", NORM, PARSE, INLINE, CSS); //emphasis
HTMLTag HTMLTag::tag_font("font", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_i("i", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_ins("ins", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_kbd("kbd", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_link("link", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_pre("pre", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_q("q", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_samp("samp", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_small("small", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_strong("strong", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_sub("sub", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_sup("sup", NORM, PARSE, INLINE, CSS);
HTMLTag HTMLTag::tag_var("var", NORM, PARSE, INLINE, CSS);

HTMLTag::HTMLTag(const char *_text, 
	bool _processingInstruction, bool _jumpTag, bool _inlineTag, bool _cssTag, bool _headingTag, 
	const char *_endtag, const bool treatAsUnknown):
	m_id(m_ids.size()),
	m_text(_text), 
	m_processingInstruction(_processingInstruction), m_jumpTag(_jumpTag), m_inlineTag(_inlineTag), m_cssTag(_cssTag), 
	m_headingTag(_headingTag), m_endtag(_endtag), 
	m_dynamicEndTag(0)
{
	if (!m_endtag) {
		m_dynamicEndTag = (char*)malloc(strlen(m_text)+4);
		sprintf(m_dynamicEndTag, "</%s>", m_text);
	}
	if (!treatAsUnknown) m_tags.insert(make_pair<const char*, HTMLTag*>(m_text, this));
	m_ids.push_back(this);
}
HTMLTag::~HTMLTag() {
	if (m_dynamicEndTag) free(m_dynamicEndTag);
}

HTMLTag *HTMLTag::gettag(const char *tagStart) {
	const char *tagEnd;
	char c;
	HTMLTag *thisTag = 0;

	//move to standard place at beginning of name
	while (is((c = *tagStart), "</!")) tagStart++;  //after < or </ or <! to EOF

	//tag end
	if (tagStart[0] == '-' && tagStart[1] == '-') tagEnd = tagStart + 2;   //comment oes not require a space after the -- tag name e.g. <!--comment-->
	else {
		//normal tagname resolution
		tagEnd = tagStart;
		while ((c = *tagEnd) && isAlphaNumeric(c)) tagEnd++; //nearest space or > or / (<br/>) or EOF
	}
	if (c) thisTag = tag(tagStart, tagEnd);                           //EOF check, we have the tag name now :)

	return thisTag;
}

HTMLTag *HTMLTag::tag(const char *text) {
	StringMapCI<HTMLTag*>::const_iterator i;
	i = m_tags.find(text);
	return (i == m_tags.end() ? 0 : i->second);
}
HTMLTag *HTMLTag::tag(const char *textstart, const char *textfinish) {
	char *endc = (char*) textfinish;
	char c = *endc; *endc = 0;
	HTMLTag *t = tag(textstart);
	*endc = c;
	return t;
}