/* tags.c -- recognize HTML tags

  (c) 1998 (W3C) MIT, INRIA, Keio University
  See tidy.c for the copyright notice.


  The HTML tags are stored as 8 bit ASCII strings.
  Use lookupw() to find a tag given a wide char string.
*/


#include "platform.h"   /* platform independent stuff */
#include "html.h"       /* to pull in definition of nodes */

#define HASHSIZE 101

extern Bool XmlTags;

Dict *tag_html;
Dict *tag_head;
Dict *tag_title;
Dict *tag_body;
Dict *tag_frameset;
Dict *tag_noframes;
Dict *tag_hr;
Dict *tag_h1;
Dict *tag_h2;
Dict *tag_pre;
Dict *tag_listing;
Dict *tag_p;
Dict *tag_ul;
Dict *tag_ol;
Dict *tag_dir;
Dict *tag_li;
Dict *tag_dt;
Dict *tag_dd;
Dict *tag_td;
Dict *tag_th;
Dict *tag_tr;
Dict *tag_col;
Dict *tag_br;
Dict *tag_a;
Dict *tag_link;
Dict *tag_b;
Dict *tag_i;
Dict *tag_param;
Dict *tag_option;
Dict *tag_optgroup;
Dict *tag_map;
Dict *tag_area;
Dict *tag_nobr;
Dict *tag_wbr;
Dict *tag_font;
Dict *tag_layer;
Dict *tag_spacer;
Dict *tag_center;
Dict *tag_style;
Dict *tag_script;
Dict *tag_noscript;
Dict *tag_table;
Dict *tag_caption;
Dict *tag_form;
Dict *tag_blockquote;
Dict *tag_div;
Dict *tag_span;

Dict *xml_tags;  /* dummy for xml tags */

static Dict *hashtab[HASHSIZE];

static struct tag
{
    char *name;
    unsigned versions;
    unsigned model;
    Parser *parser;
    CheckAttribs *chkattrs;
} tags[] =
{
    "HTML",       (VERS_ALL|VERS_FRAMES),     CM_HTML,  ParseHTML, CheckHTML,

    "HEAD",       (VERS_ALL|VERS_FRAMES),     (CM_HTML|CM_OPT), ParseHead, null,

    "TITLE",      (VERS_ALL|VERS_FRAMES),     CM_HEAD, ParseTitle, null,
    "BASE",       (VERS_ALL|VERS_FRAMES),     (CM_HEAD|CM_EMPTY), null, null,
    "LINK",       (VERS_ALL|VERS_FRAMES),     (CM_HEAD|CM_EMPTY), null, null,
    "META",       (VERS_ALL|VERS_FRAMES),     (CM_HEAD|CM_EMPTY), null, null,
    "STYLE",      (VERS_FROM32|VERS_FRAMES),  CM_HEAD, ParseScript, null,
    "SCRIPT",     (VERS_FROM32|VERS_FRAMES),  (CM_HEAD|CM_MIXED|CM_BLOCK|CM_INLINE), ParseScript, CheckSCRIPT,

    "BODY",       VERS_ALL,     (CM_HTML|CM_OPT), ParseBody, null,
    "FRAMESET",   VERS_FRAMES,  (CM_HTML|CM_FRAMES), ParseFrameSet, null,

    "P",          VERS_ALL,     (CM_BLOCK|CM_OPT), ParseInline, null,
    "H1",         VERS_ALL,     (CM_BLOCK|CM_HEADING), ParseInline, null,
    "H2",         VERS_ALL,     (CM_BLOCK|CM_HEADING), ParseInline, null,
    "H3",         VERS_ALL,     (CM_BLOCK|CM_HEADING), ParseInline, null,
    "H4",         VERS_ALL,     (CM_BLOCK|CM_HEADING), ParseInline, null,
    "H5",         VERS_ALL,     (CM_BLOCK|CM_HEADING), ParseInline, null,
    "H6",         VERS_ALL,     (CM_BLOCK|CM_HEADING), ParseInline, null,
    "UL",         VERS_ALL,     CM_BLOCK, ParseList, null,
    "OL",         VERS_ALL,     CM_BLOCK, ParseList, null,
    "DL",         VERS_ALL,     CM_BLOCK, ParseDefList, null,
    "DIR",        VERS_LOOSE,   (CM_BLOCK|CM_OBSOLETE), ParseList, null,
    "MENU",       VERS_LOOSE,   (CM_BLOCK|CM_OBSOLETE), ParseList, null,
    "PRE",        VERS_ALL,     CM_BLOCK, ParsePre, null,
    "LISTING",    VERS_ALL,     (CM_BLOCK|CM_OBSOLETE), ParsePre, null,
    "XMP",        VERS_ALL,     (CM_BLOCK|CM_OBSOLETE), ParsePre, null,
    "PLAINTEXT",  VERS_ALL,     (CM_BLOCK|CM_OBSOLETE), ParsePre, null,
    "ADDRESS",    VERS_ALL,     CM_BLOCK, ParseBlock, null,
    "BLOCKQUOTE", VERS_ALL,     CM_BLOCK, ParseBlock, null,
    "FORM",       VERS_ALL,     CM_BLOCK, ParseBlock, null,
    "ISINDEX",    VERS_ALL,     (CM_BLOCK|CM_EMPTY), null, null,
    "FIELDSET",   VERS_HTML40,  CM_BLOCK, ParseBlock, null,
    "TABLE",      VERS_FROM32,  CM_BLOCK, ParseTableTag, CheckTABLE,
    "HR",         VERS_ALL,     (CM_BLOCK|CM_EMPTY),  null, null,
    "DIV",        VERS_FROM32,  CM_BLOCK, ParseBlock, null,
    "NOSAVE",     VERS_NETSCAPE, CM_BLOCK, ParseBlock, null,
    "LAYER",      VERS_NETSCAPE, CM_BLOCK, ParseBlock, null,
    "NOLAYER",    VERS_NETSCAPE, CM_BLOCK, ParseBlock, null,
    "ALIGN",      VERS_NETSCAPE, CM_BLOCK, ParseBlock, null,
    "CENTER",     VERS_LOOSE,   CM_BLOCK, ParseBlock, null,
    "INS",        VERS_HTML40,  (CM_INLINE|CM_BLOCK|CM_MIXED), ParseInline, null,
    "DEL",        VERS_HTML40,  (CM_INLINE|CM_BLOCK|CM_MIXED), ParseInline, null,

    "LI",         VERS_ALL,     (CM_LIST|CM_OPT|CM_NO_INDENT), ParseBlock, null,
    "DT",         VERS_ALL,     (CM_DEFLIST|CM_OPT|CM_NO_INDENT), ParseInline, null,
    "DD",         VERS_ALL,     (CM_DEFLIST|CM_OPT|CM_NO_INDENT), ParseBlock, null,

    "CAPTION",    VERS_FROM32,  CM_TABLE, ParseInline, CheckCaption,
    "COLGROUP",   VERS_HTML40,  (CM_TABLE|CM_OPT), ParseColGroup, null,
    "COL",        VERS_HTML40,  (CM_TABLE|CM_EMPTY),  null, null,
    "THEAD",      VERS_HTML40,  (CM_TABLE|CM_ROWGRP|CM_OPT), ParseRowGroup, null,
    "TFOOT",      VERS_HTML40,  (CM_TABLE|CM_ROWGRP|CM_OPT), ParseRowGroup, null,
    "TBODY",      VERS_HTML40,  (CM_TABLE|CM_ROWGRP|CM_OPT), ParseRowGroup, null,
    "TR",         VERS_FROM32,  (CM_TABLE|CM_OPT), ParseRow, null,
    "TD",         VERS_FROM32,  (CM_ROW|CM_OPT|CM_NO_INDENT), ParseBlock, null,
    "TH",         VERS_FROM32,  (CM_ROW|CM_OPT|CM_NO_INDENT), ParseBlock, null,

    "Q",          VERS_HTML40,  CM_INLINE, ParseInline, null,
    "A",          VERS_ALL,     CM_INLINE, ParseInline, null,
    "BR",         VERS_ALL,     (CM_INLINE|CM_EMPTY), null, null,
    "IMG",        VERS_ALL,     (CM_INLINE|CM_IMG|CM_EMPTY), null, CheckIMG,
    "OBJECT",     VERS_HTML40,  (CM_OBJECT|CM_IMG|CM_INLINE|CM_PARAM), ParseBlock, null,
    "APPLET",     VERS_LOOSE,   (CM_OBJECT|CM_IMG|CM_INLINE|CM_PARAM), ParseBlock, null,
    "SERVLET",    VERS_SUN,     (CM_OBJECT|CM_IMG|CM_INLINE|CM_PARAM), ParseBlock, null,
    "PARAM",      VERS_FROM32,  (CM_INLINE|CM_EMPTY), null, null,
    "EMBED",      VERS_NETSCAPE, (CM_INLINE|CM_IMG|CM_EMPTY), null, null,
    "IFRAME",     VERS_HTML40_LOOSE, CM_INLINE, ParseBlock, null,
    "FRAME",      VERS_FRAMES,  (CM_FRAMES|CM_EMPTY), null, null,
    "NOFRAMES",   VERS_IFRAMES, (CM_BLOCK|CM_FRAMES), ParseNoFrames,  null,
    "NOSCRIPT",   (VERS_FRAMES|VERS_HTML40),  (CM_BLOCK|CM_INLINE|CM_MIXED), ParseBlock, null,
    "B",          VERS_ALL,     CM_INLINE, ParseInline, null,
    "I",          VERS_ALL,     CM_INLINE, ParseInline, null,
    "U",          VERS_ALL,     CM_INLINE, ParseInline, null,
    "TT",         VERS_ALL,     CM_INLINE, ParseInline, null,
    "S",          VERS_HTML40,  CM_INLINE, ParseInline, null,
    "STRIKE",     VERS_FROM32,  CM_INLINE, ParseInline, null,
    "BIG",        VERS_FROM32,  CM_INLINE, ParseInline, null,
    "SMALL",      VERS_FROM32,  CM_INLINE, ParseInline, null,
    "SUB",        VERS_FROM32,  CM_INLINE, ParseInline, null,
    "SUP",        VERS_FROM32,  CM_INLINE, ParseInline, null,
    "EM",         VERS_ALL,     CM_INLINE, ParseInline, null,
    "STRONG",     VERS_ALL,     CM_INLINE, ParseInline, null,
    "DFN",        VERS_ALL,     CM_INLINE, ParseInline, null,
    "CODE",       VERS_ALL,     CM_INLINE, ParseInline, null,
    "SAMP",       VERS_ALL,     CM_INLINE, ParseInline, null,
    "KBD",        VERS_ALL,     CM_INLINE, ParseInline, null,
    "VAR",        VERS_ALL,     CM_INLINE, ParseInline, null,
    "CITE",       VERS_ALL,     CM_INLINE, ParseInline, null,
    "ABBR",       VERS_HTML40,  CM_INLINE, ParseInline, null,
    "ACRONYM",    VERS_HTML40,  CM_INLINE, ParseInline, null,
    "SPAN",       VERS_FROM32,  CM_INLINE, ParseInline, null,
    "BLINK",      VERS_PROPRIETARY, CM_INLINE, ParseInline, null,
    "NOBR",       VERS_PROPRIETARY, CM_INLINE, ParseInline, null,
    "WBR",        VERS_PROPRIETARY, (CM_INLINE|CM_EMPTY), null, null,
    "MARQUEE",    VERS_MICROSOFT, (CM_INLINE|CM_EMPTY), null, null,
    "EMBED",      VERS_NETSCAPE, (CM_INLINE|CM_EMPTY), null, null,
    "SPACER",     VERS_NETSCAPE, (CM_INLINE|CM_EMPTY), null, null,
    "MAP",        VERS_FROM32,  CM_INLINE, ParseBlock, null,
    "AREA",       VERS_ALL,     (CM_BLOCK|CM_EMPTY), null, CheckAREA,
    "INPUT",      VERS_ALL,     (CM_INLINE|CM_IMG|CM_EMPTY), null, null,
    "SELECT",     VERS_ALL,     (CM_INLINE|CM_FIELD), ParseSelect, null,
    "OPTION",     VERS_ALL,     (CM_FIELD|CM_OPT), ParseText, null,
    "OPTGROUP",   VERS_HTML40,  (CM_FIELD|CM_OPT), ParseOptGroup, null,
    "TEXTAREA",   VERS_ALL,     (CM_INLINE|CM_FIELD), ParseText, null,
    "LABEL",      VERS_HTML40,  CM_INLINE, ParseInline, null,
    "LEGEND",     VERS_HTML40,  CM_INLINE, ParseInline, null,
    "BUTTON",     VERS_HTML40,  CM_INLINE, ParseInline, null,
    "BASEFONT",   VERS_LOOSE,   (CM_INLINE|CM_EMPTY), null, null,
    "FONT",       VERS_LOOSE,   CM_INLINE, ParseInline, null,
    "BDO",        VERS_HTML40,  CM_INLINE, ParseInline, null,

  /* this must be the final entry */
    null,         0,            0,          0,       0
};

/* choose what version to use for new doctype */
int HTMLVersion(Lexer *lexer)
{
    uint versions;

    versions = lexer->versions;

    if (versions & VERS_HTML20)
        return VERS_HTML20;

    if (versions & VERS_HTML32)
        return VERS_HTML32;

    if (versions & VERS_HTML40_STRICT)
        return VERS_HTML40_STRICT;

    if (versions & VERS_HTML40_LOOSE)
        return VERS_HTML40_LOOSE;

    if (versions & VERS_FRAMES)
        return VERS_FRAMES;

    return VERS_UNKNOWN;
}

static unsigned hash(char *s)
{
    unsigned hashval;

    for (hashval = 0; *s != '\0'; s++)
        hashval = *s + 31*hashval;

    return hashval % HASHSIZE;
}

static Dict *lookup(char *s)
{
    Dict *np;

    for (np = hashtab[hash(s)]; np != null; np = np->next)
        if (wstrcmp(s, np->name) == 0)
            return np;
    return null;
}

static Dict *install(char *name, uint versions, uint model, 
                     Parser *parser, CheckAttribs *chkattrs)
{
    Dict *np;
    unsigned hashval;

    if ((np = lookup(name)) == null)
    {
        np = (Dict *)MemAlloc(sizeof(*np));

        if (np == null || (np->name = wstrdup(name)) == null)
            return null;

        hashval = hash(name);
        np->next = hashtab[hashval];
        hashtab[hashval] = np;
    }

    np->versions = versions;
    np->model = model;
    np->parser = parser;
    np->chkattrs = chkattrs;
    return np;
}

/* public interface for finding tag by name */
Bool FindTag(Node *node)
{
    Dict *np;

    if (node->element && (np = lookup(node->element)))
    {
        node->tag = np;
        return yes;
    }

    if (XmlTags)
    {
        node->tag = xml_tags;
        return yes;
    }

    return no;
}

void DefineInlineTag(char *name)
{
    install(name, VERS_PROPRIETARY, (CM_INLINE|CM_NO_INDENT|CM_NEW), ParseBlock, null);
}

void DefineBlockTag(char *name)
{
    install(name, VERS_PROPRIETARY, (CM_BLOCK|CM_NO_INDENT|CM_NEW), ParseBlock, null);
}

void InitTags(void)
{
    struct tag *tp;
    
    for(tp = tags; tp->name != null; ++tp)
        install(tp->name, tp->versions, tp->model, tp->parser, tp->chkattrs);

    tag_html = lookup("HTML");
    tag_head = lookup("HEAD");
    tag_body = lookup("BODY");
    tag_frameset = lookup("FRAMESET");
    tag_noframes = lookup("NOFRAMES");
    tag_title = lookup("TITLE");
    tag_hr = lookup("HR");
    tag_pre = lookup("PRE");
    tag_listing = lookup("LISTING");
    tag_h1 = lookup("H1");
    tag_h2 = lookup("H2");
    tag_p  = lookup("P");
    tag_ul = lookup("UL");
    tag_ol = lookup("OL");
    tag_dir = lookup("DIR");
    tag_li = lookup("LI");
    tag_dt = lookup("DT");
    tag_dd = lookup("DD");
    tag_td = lookup("TD");
    tag_th = lookup("TH");
    tag_tr = lookup("TR");
    tag_col = lookup("COL");
    tag_br = lookup("BR");
    tag_a = lookup("A");
    tag_link = lookup("LINK");
    tag_b = lookup("B");
    tag_i = lookup("I");
    tag_param = lookup("PARAM");
    tag_option = lookup("OPTION");
    tag_optgroup = lookup("OPTGROUP");
    tag_map = lookup("MAP");
    tag_area = lookup("AREA");
    tag_nobr = lookup("NOBR");
    tag_wbr = lookup("WBR");
    tag_font = lookup("FONT");
    tag_spacer = lookup("SPACER");
    tag_layer = lookup("LAYER");
    tag_center = lookup("CENTER");
    tag_style = lookup("STYLE");
    tag_script = lookup("SCRIPT");
    tag_noscript = lookup("NOSCRIPT");
    tag_table = lookup("TABLE");
    tag_caption = lookup("CAPTION");
    tag_form = lookup("FORM");
    tag_blockquote = lookup("BLOCKQUOTE");
    tag_div = lookup("DIV");
    tag_span = lookup("SPAN");

    /* create dummy entry for all xml tags */
    xml_tags = (Dict *)MemAlloc(sizeof(*xml_tags));
    xml_tags->name = null;
    xml_tags->versions = VERS_ALL;
    xml_tags->model = CM_BLOCK;
    xml_tags->parser = null;
    xml_tags->chkattrs = null;
}

void FreeTags(void)
{
    Dict *prev, *next;
    int i;

    MemFree(xml_tags);

    for (i = 0; i < HASHSIZE; ++i)
    {
        prev = null;
        next = hashtab[i];

        while(next)
        {
            prev = next->next;
            MemFree(next);
            next = prev;
        }
    }
}