HTML cleaner utility.

4 files changed, 749 insertions, 0 deletions

diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..0b9b82a
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,14 @@
+CFLAGS := -g -Wall -W -ansi -D_XOPEN_SOURCE=500 $(EXTRA_CFLAGS)
+
+hc: lex.yy.o hc.o
+	$(CC) $(CFLAGS) $(OUR_CFLAGS) $^ -o $@ -lfl 
+
+lex.yy.o: lex.yy.c hc.h
+
+hc.o: hc.c hc.h
+
+lex.yy.c: hc.l hc.h
+	$(LEX) -i -8 hc.l
+
+clean:
+	-rm hc lex.yy.o lex.yy.c
diff --git a/hc.c b/hc.c
new file mode 100644
index 0000000..4c14e04
--- /dev/null
+++ b/hc.c
@@ -0,0 +1,265 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "hc.h"
+
+
+static int allowed_el_spec[] = {
+  tok_el_a,
+  tok_el_abbr,
+  tok_el_acronym,
+  tok_el_address,
+  /* tok_el_applet, */
+  /* tok_el_area, */
+  tok_el_b,
+  /* tok_el_base, */
+  tok_el_basefont,
+  tok_el_bdo,
+  tok_el_big,
+  tok_el_blockquote,
+  /* tok_el_body, */
+  tok_el_br,
+  /* tok_el_button, */
+  tok_el_caption,
+  tok_el_center,
+  tok_el_cite,
+  tok_el_code,
+  tok_el_col,
+  tok_el_colgroup,
+  tok_el_dd,
+  tok_el_del,
+  tok_el_dfn,
+  tok_el_dir,
+  tok_el_div,
+  tok_el_dl,
+  tok_el_dt,
+  tok_el_em,
+  /* tok_el_fieldset, */
+  tok_el_font,
+  tok_el_form,
+  /* tok_el_frame, */
+  /* tok_el_frameset, */
+  tok_el_h1,
+  tok_el_h2,
+  tok_el_h3,
+  tok_el_h4,
+  tok_el_h5,
+  tok_el_h6,
+  /* tok_el_head, */
+  tok_el_hr,
+  /* tok_el_html, */
+  tok_el_i,
+  /* tok_el_iframe, */
+  tok_el_img,
+  /* tok_el_input, */
+  tok_el_ins,
+  tok_el_kbd,
+  /* tok_el_label, */
+  /* tok_el_legend, */
+  tok_el_li,
+  /* tok_el_link, */
+  /* tok_el_map, */
+  /* tok_el_menu, */
+  /* tok_el_meta, */
+  /* tok_el_noframes, */
+  /* tok_el_noscript, */
+  /* tok_el_object, */
+  tok_el_ol,
+  tok_el_optgroup,
+  /* tok_el_option, */
+  tok_el_p,
+  /* tok_el_param, */
+  tok_el_pre,
+  tok_el_q,
+  tok_el_samp,
+  /* tok_el_script,*/
+  /* tok_el_select, */
+  tok_el_small,
+  tok_el_span,
+  tok_el_strike,
+  tok_el_strong,
+  tok_el_style,
+  tok_el_sub,
+  tok_el_sup,
+  tok_el_table,
+  tok_el_tbody,
+  tok_el_td,
+  /* tok_el_textarea, */
+  tok_el_tfoot,
+  tok_el_th,
+  tok_el_thead,
+  /* tok_el_title, */
+  tok_el_tr,
+  tok_el_tt,
+  tok_el_u,
+  tok_el_ul,
+  /* tok_el_var, */
+  tok_eof,
+};
+
+static int allowed_el[tok_max];
+
+static const token_t blank;
+static token_t pushback;
+
+static void bail()
+{
+  fprintf(stderr, "bad html\n");
+  exit(EXIT_FAILURE);
+}
+
+static token_t mktok(toktype_t type, char *text)
+{
+  token_t tok = { 0, 0, 0, 0 };
+  tok.type = type;
+  tok.lexeme = strdup(text);
+  return tok;
+}
+
+static void deltok(token_t tok)
+{
+  free(tok.lexeme);
+}
+
+static int null(token_t tok)
+{
+  return tok.type == tok_eof;
+}
+
+static token_t gettok(void)
+{
+  if (null(pushback)) {
+    int type = yylex();
+    return mktok(type, yytext);
+  } else {
+    token_t tok = pushback;
+    pushback = blank;
+    return tok;
+  }
+}
+
+static void ungettok(token_t tok)
+{
+  deltok(pushback);
+  pushback = tok;
+}
+
+static token_t printtok(token_t tok)
+{
+  if (!null(tok))
+    fputs(tok.lexeme, stdout);
+  return tok;
+}
+
+static token_t match(int type)
+{
+  token_t tok = gettok();
+  if (tok.type != type)
+    bail();
+  return tok;
+}
+
+static token_t optmatch(int type)
+{
+  token_t tok = gettok();
+  if (tok.type != type) {
+    ungettok(tok);
+    return blank;
+  }
+  return tok;
+}
+
+static token_t lookfor(int type)
+{
+  token_t tok;
+  for (;;) {
+    tok = gettok();
+    if (tok.type == type || null(tok))
+      break;
+    deltok(tok);
+  }
+  return tok;
+}
+
+static token_t printuntil(int type)
+{
+  token_t tok;
+  for (;;) {
+    tok = gettok();
+    if (tok.type == type || null(tok))
+      break;
+    deltok(printtok(tok));
+  }
+  printtok(tok);
+  return tok;
+}
+
+static void parse_element(token_t in)
+{
+  token_t end = optmatch('/');
+  token_t name = gettok();
+
+  switch (name.type) {
+  case '/': 
+    if (!null(end))
+      bail();
+    printtok(in);
+    printtok(name);
+    deltok(printtok(lookfor('>')));
+    goto out; 
+    return;
+  case '>':
+    bail();
+  default:
+    break;
+  }
+
+  if (allowed_el[name.type]) {
+    printtok(in);
+    printtok(end);
+    printtok(name);
+    deltok(printuntil('>'));
+  } else {
+    deltok(lookfor('>'));
+  }
+
+out:
+  deltok(end);
+  deltok(name);
+}
+
+static void parse(void)
+{
+  for (;;) {
+    token_t tok = gettok();
+
+    switch (tok.type) {
+    case '<':
+      parse_element(tok);
+      break;
+    case tok_eof:
+      deltok(tok);
+      return;
+    default:
+      printtok(tok);
+      break;
+    }
+
+    deltok(tok);
+  }
+}
+
+static void init(void)
+{
+  int i;
+
+  for (i = 0; allowed_el_spec[i] != tok_eof; i++)
+    allowed_el[allowed_el_spec[i]] = 1;
+}
+
+int main(void)
+{
+  init();
+  parse();
+  return 0;
+}
diff --git a/hc.h b/hc.h
new file mode 100644
index 0000000..05bedfb
--- /dev/null
+++ b/hc.h
@@ -0,0 +1,227 @@
+
+typedef enum {
+  tok_eof = 0,
+  tok_doctype = 256,
+  tok_text,
+  tok_el_unknown,
+  tok_el_a,
+  tok_el_abbr,
+  tok_el_acronym,
+  tok_el_address,
+  tok_el_applet,
+  tok_el_area,
+  tok_el_b,
+  tok_el_base,
+  tok_el_basefont,
+  tok_el_bdo,
+  tok_el_big,
+  tok_el_blockquote,
+  tok_el_body,
+  tok_el_br,
+  tok_el_button,
+  tok_el_caption,
+  tok_el_center,
+  tok_el_cite,
+  tok_el_code,
+  tok_el_col,
+  tok_el_colgroup,
+  tok_el_dd,
+  tok_el_del,
+  tok_el_dfn,
+  tok_el_dir,
+  tok_el_div,
+  tok_el_dl,
+  tok_el_dt,
+  tok_el_em,
+  tok_el_fieldset,
+  tok_el_font,
+  tok_el_form,
+  tok_el_frame,
+  tok_el_frameset,
+  tok_el_h1,
+  tok_el_h2,
+  tok_el_h3,
+  tok_el_h4,
+  tok_el_h5,
+  tok_el_h6,
+  tok_el_head,
+  tok_el_hr,
+  tok_el_html,
+  tok_el_i,
+  tok_el_iframe,
+  tok_el_img,
+  tok_el_input,
+  tok_el_ins,
+  tok_el_kbd,
+  tok_el_label,
+  tok_el_legend,
+  tok_el_li,
+  tok_el_link,
+  tok_el_map,
+  tok_el_menu,
+  tok_el_meta,
+  tok_el_noframes,
+  tok_el_noscript,
+  tok_el_object,
+  tok_el_ol,
+  tok_el_optgroup,
+  tok_el_option,
+  tok_el_p,
+  tok_el_param,
+  tok_el_pre,
+  tok_el_q,
+  tok_el_samp,
+  tok_el_script,
+  tok_el_select,
+  tok_el_small,
+  tok_el_span,
+  tok_el_strike,
+  tok_el_strong,
+  tok_el_style,
+  tok_el_sub,
+  tok_el_sup,
+  tok_el_table,
+  tok_el_tbody,
+  tok_el_td,
+  tok_el_textarea,
+  tok_el_tfoot,
+  tok_el_th,
+  tok_el_thead,
+  tok_el_title,
+  tok_el_tr,
+  tok_el_tt,
+  tok_el_u,
+  tok_el_ul,
+  tok_el_var,
+  tok_at_unknown,
+  tok_at_accept,
+  tok_at_accept_charset,
+  tok_at_accesskey,
+  tok_at_action,
+  tok_at_align,
+  tok_at_alink,
+  tok_at_alt,
+  tok_at_archive,
+  tok_at_axis,
+  tok_at_background,
+  tok_at_bbr,
+  tok_at_bgcolor,
+  tok_at_border,
+  tok_at_cellpadding,
+  tok_at_cellspacing,
+  tok_at_char,
+  tok_at_charoff,
+  tok_at_charset,
+  tok_at_checked,
+  tok_at_cite,
+  tok_at_class,
+  tok_at_classid,
+  tok_at_clear,
+  tok_at_code,
+  tok_at_codebase,
+  tok_at_codetype,
+  tok_at_color,
+  tok_at_cols,
+  tok_at_colspan,
+  tok_at_compact,
+  tok_at_content,
+  tok_at_coords,
+  tok_at_data,
+  tok_at_datetime,
+  tok_at_declare,
+  tok_at_defer,
+  tok_at_dir,
+  tok_at_disabled,
+  tok_at_enctype,
+  tok_at_face,
+  tok_at_for,
+  tok_at_frame,
+  tok_at_frameborder,
+  tok_at_headers,
+  tok_at_height,
+  tok_at_href,
+  tok_at_hreflang,
+  tok_at_hspace,
+  tok_at_http_equiv,
+  tok_at_id,
+  tok_at_ismap,
+  tok_at_label,
+  tok_at_lang,
+  tok_at_language,
+  tok_at_link,
+  tok_at_longdesc,
+  tok_at_marginheight,
+  tok_at_marginwidth,
+  tok_at_maxlength,
+  tok_at_media,
+  tok_at_method,
+  tok_at_multiple,
+  tok_at_name,
+  tok_at_nohref,
+  tok_at_noresize,
+  tok_at_noshade,
+  tok_at_nowrap,
+  tok_at_object,
+  tok_at_onblur,
+  tok_at_onchange,
+  tok_at_onclick,
+  tok_at_ondblclick,
+  tok_at_onfocus,
+  tok_at_onkeydown,
+  tok_at_onkeypress,
+  tok_at_onkeyup,
+  tok_at_onload,
+  tok_at_onmousedown,
+  tok_at_onmousemove,
+  tok_at_onmouseout,
+  tok_at_onmouseover,
+  tok_at_onmouseup,
+  tok_at_onreset,
+  tok_at_onselect,
+  tok_at_onsubmit,
+  tok_at_onunload,
+  tok_at_profile,
+  tok_at_prompt,
+  tok_at_readonly,
+  tok_at_rel,
+  tok_at_rev,
+  tok_at_rows,
+  tok_at_rowspan,
+  tok_at_rules,
+  tok_at_scheme,
+  tok_at_scope,
+  tok_at_scrolling,
+  tok_at_selected,
+  tok_at_shape,
+  tok_at_size,
+  tok_at_span,
+  tok_at_src,
+  tok_at_standby,
+  tok_at_start,
+  tok_at_style,
+  tok_at_summary,
+  tok_at_tabindex,
+  tok_at_target,
+  tok_at_text,
+  tok_at_title,
+  tok_at_type,
+  tok_at_usemap,
+  tok_at_valign,
+  tok_at_value,
+  tok_at_valuetype,
+  tok_at_version,
+  tok_at_vlink,
+  tok_at_vspace,
+  tok_at_width,
+  tok_max
+} toktype_t;
+
+typedef struct {
+  int type;
+  int is_tag;
+  int is_close;
+  char *lexeme;
+} token_t;
+
+extern int yylex(void);
+extern char *yytext;
diff --git a/hc.l b/hc.l
new file mode 100644
index 0000000..8d16781
--- /dev/null
+++ b/hc.l
@@ -0,0 +1,243 @@
+/* This flex scanner is intended to be compliled case insensitive. */
+
+%{
+
+#include <stdio.h>
+#include <string.h>
+#include "hc.h"
+
+%}
+
+wsp                     [ \t\n\r\v\t] 
+notwsp                  [^ \t\n\r\v\t] 
+ctrl                    [\x0-\x1f]
+notctrl                 [^\x0-\x1f]
+special                 ["'<>/=&]
+notspecial              [^"'<>/=&]
+elname                  [A-Za-z0-9]+
+attrname                [^"'<>/=&\x0-\x1f\t\n\r\v\t ]
+endnm                   [^A-Za-z_\-0-9]
+%x                      ELM ATT
+
+%%
+
+[<]                     { BEGIN(ELM);
+                          return '<'; }
+{notspecial}+           { return tok_text; }
+<ELM>a/{endnm}          { BEGIN(ATT); return tok_el_a; }
+<ELM>abbr/{endnm}       { BEGIN(ATT); return tok_el_abbr; }
+<ELM>acronym/{endnm}    { BEGIN(ATT); return tok_el_acronym; }
+<ELM>address/{endnm}    { BEGIN(ATT); return tok_el_address; }
+<ELM>applet/{endnm}     { BEGIN(ATT); return tok_el_applet; }
+<ELM>area/{endnm}       { BEGIN(ATT); return tok_el_area; }
+<ELM>b/{endnm}          { BEGIN(ATT); return tok_el_b; }
+<ELM>base/{endnm}       { BEGIN(ATT); return tok_el_base; }
+<ELM>basefont/{endnm}   { BEGIN(ATT); return tok_el_basefont; }
+<ELM>bdo/{endnm}        { BEGIN(ATT); return tok_el_bdo; }
+<ELM>big/{endnm}        { BEGIN(ATT); return tok_el_big; }
+<ELM>blockquote/{endnm} { BEGIN(ATT); return tok_el_blockquote; }
+<ELM>body/{endnm}       { BEGIN(ATT); return tok_el_body; }
+<ELM>br/{endnm}         { BEGIN(ATT); return tok_el_br; }
+<ELM>button/{endnm}     { BEGIN(ATT); return tok_el_button; }
+<ELM>caption/{endnm}    { BEGIN(ATT); return tok_el_caption; }
+<ELM>center/{endnm}     { BEGIN(ATT); return tok_el_center; }
+<ELM>cite/{endnm}       { BEGIN(ATT); return tok_el_cite; }
+<ELM>code/{endnm}       { BEGIN(ATT); return tok_el_code; }
+<ELM>col/{endnm}        { BEGIN(ATT); return tok_el_col; }
+<ELM>colgroup/{endnm}   { BEGIN(ATT); return tok_el_colgroup; }
+<ELM>dd/{endnm}         { BEGIN(ATT); return tok_el_dd; }
+<ELM>del/{endnm}        { BEGIN(ATT); return tok_el_del; }
+<ELM>dfn/{endnm}        { BEGIN(ATT); return tok_el_dfn; }
+<ELM>dir/{endnm}        { BEGIN(ATT); return tok_el_dir; }
+<ELM>div/{endnm}        { BEGIN(ATT); return tok_el_div; }
+<ELM>dl/{endnm}         { BEGIN(ATT); return tok_el_dl; }
+<ELM>dt/{endnm}         { BEGIN(ATT); return tok_el_dt; }
+<ELM>em/{endnm}         { BEGIN(ATT); return tok_el_em; }
+<ELM>fieldset/{endnm}   { BEGIN(ATT); return tok_el_fieldset; }
+<ELM>font/{endnm}       { BEGIN(ATT); return tok_el_font; }
+<ELM>form/{endnm}       { BEGIN(ATT); return tok_el_form; }
+<ELM>frame/{endnm}      { BEGIN(ATT); return tok_el_frame; }
+<ELM>frameset/{endnm}   { BEGIN(ATT); return tok_el_frameset; }
+<ELM>h1/{endnm}         { BEGIN(ATT); return tok_el_h1; }
+<ELM>h2/{endnm}         { BEGIN(ATT); return tok_el_h2; }
+<ELM>h3/{endnm}         { BEGIN(ATT); return tok_el_h3; }
+<ELM>h4/{endnm}         { BEGIN(ATT); return tok_el_h4; }
+<ELM>h5/{endnm}         { BEGIN(ATT); return tok_el_h5; }
+<ELM>h6/{endnm}         { BEGIN(ATT); return tok_el_h5; }
+<ELM>head/{endnm}       { BEGIN(ATT); return tok_el_head; }
+<ELM>hr/{endnm}         { BEGIN(ATT); return tok_el_hr; }
+<ELM>html/{endnm}       { BEGIN(ATT); return tok_el_html; }
+<ELM>i/{endnm}          { BEGIN(ATT); return tok_el_i; }
+<ELM>iframe/{endnm}     { BEGIN(ATT); return tok_el_iframe; }
+<ELM>img/{endnm}        { BEGIN(ATT); return tok_el_img; }
+<ELM>input/{endnm}      { BEGIN(ATT); return tok_el_input; }
+<ELM>ins/{endnm}        { BEGIN(ATT); return tok_el_ins; }
+<ELM>kbd/{endnm}        { BEGIN(ATT); return tok_el_kbd; }
+<ELM>label/{endnm}      { BEGIN(ATT); return tok_el_label; }
+<ELM>legend/{endnm}     { BEGIN(ATT); return tok_el_legend; }
+<ELM>li/{endnm}         { BEGIN(ATT); return tok_el_li; }
+<ELM>link/{endnm}       { BEGIN(ATT); return tok_el_link; }
+<ELM>map/{endnm}        { BEGIN(ATT); return tok_el_map; }
+<ELM>menu/{endnm}       { BEGIN(ATT); return tok_el_menu; }
+<ELM>meta/{endnm}       { BEGIN(ATT); return tok_el_meta; }
+<ELM>noframes/{endnm}   { BEGIN(ATT); return tok_el_noframes; }
+<ELM>noscript/{endnm}   { BEGIN(ATT); return tok_el_noscript; }
+<ELM>object/{endnm}     { BEGIN(ATT); return tok_el_object; }
+<ELM>ol/{endnm}         { BEGIN(ATT); return tok_el_ol; }
+<ELM>optgroup/{endnm}   { BEGIN(ATT); return tok_el_optgroup; }
+<ELM>option/{endnm}     { BEGIN(ATT); return tok_el_option; }
+<ELM>p/{endnm}          { BEGIN(ATT); return tok_el_p; }
+<ELM>param/{endnm}      { BEGIN(ATT); return tok_el_param; }
+<ELM>pre/{endnm}        { BEGIN(ATT); return tok_el_pre; }
+<ELM>q/{endnm}          { BEGIN(ATT); return tok_el_q; }
+<ELM>samp/{endnm}       { BEGIN(ATT); return tok_el_samp; }
+<ELM>script/{endnm}     { BEGIN(ATT); return tok_el_script; }
+<ELM>select/{endnm}     { BEGIN(ATT); return tok_el_select; }
+<ELM>small/{endnm}      { BEGIN(ATT); return tok_el_small; }
+<ELM>span/{endnm}       { BEGIN(ATT); return tok_el_span; }
+<ELM>strike/{endnm}     { BEGIN(ATT); return tok_el_strike; }
+<ELM>strong/{endnm}     { BEGIN(ATT); return tok_el_strong; }
+<ELM>style/{endnm}      { BEGIN(ATT); return tok_el_style; }
+<ELM>sub/{endnm}        { BEGIN(ATT); return tok_el_sub; }
+<ELM>sup/{endnm}        { BEGIN(ATT); return tok_el_sup; }
+<ELM>table/{endnm}      { BEGIN(ATT); return tok_el_table; }
+<ELM>tbody/{endnm}      { BEGIN(ATT); return tok_el_tbody; }
+<ELM>td/{endnm}         { BEGIN(ATT); return tok_el_td; }
+<ELM>textarea/{endnm}   { BEGIN(ATT); return tok_el_textarea; }
+<ELM>tfoot/{endnm}      { BEGIN(ATT); return tok_el_tfoot; }
+<ELM>th/{endnm}         { BEGIN(ATT); return tok_el_th; }
+<ELM>thead/{endnm}      { BEGIN(ATT); return tok_el_thead; }
+<ELM>title/{endnm}      { BEGIN(ATT); return tok_el_title; }
+<ELM>tr/{endnm}         { BEGIN(ATT); return tok_el_tr; }
+<ELM>tt/{endnm}         { BEGIN(ATT); return tok_el_tt; }
+<ELM>u/{endnm}          { BEGIN(ATT); return tok_el_u; }
+<ELM>ul/{endnm}         { BEGIN(ATT); return tok_el_ul; }
+<ELM>var/{endnm}        { BEGIN(ATT); return tok_el_var; }
+<ELM>{elname}           { BEGIN(ATT); return tok_el_unknown; }
+<ELM>.                  { return yytext[0]; }
+
+<ATT>accept/{endnm}     { return tok_at_accept; }
+<ATT>accept-charset/{endnm}     { return tok_at_accept_charset; }
+<ATT>accesskey/{endnm}  { return tok_at_accesskey; }
+<ATT>action/{endnm}     { return tok_at_action; }
+<ATT>align/{endnm}      { return tok_at_align; }
+<ATT>alink/{endnm}      { return tok_at_alink; }
+<ATT>alt/{endnm}        { return tok_at_alt; }
+<ATT>archive/{endnm}    { return tok_at_archive; }
+<ATT>axis/{endnm}       { return tok_at_axis; }
+<ATT>background/{endnm} { return tok_at_background; }
+<ATT>bbr/{endnm}        { return tok_at_bbr; }
+<ATT>bgcolor/{endnm}    { return tok_at_bgcolor; }
+<ATT>border/{endnm}     { return tok_at_border; }
+<ATT>cellpadding/{endnm}        { return tok_at_cellpadding; }
+<ATT>cellspacing/{endnm}        { return tok_at_cellspacing; }
+<ATT>char/{endnm}       { return tok_at_char; }
+<ATT>charoff/{endnm}    { return tok_at_charoff; }
+<ATT>charset/{endnm}    { return tok_at_charset; }
+<ATT>checked/{endnm}    { return tok_at_checked; }
+<ATT>cite/{endnm}       { return tok_at_cite; }
+<ATT>class/{endnm}      { return tok_at_class; }
+<ATT>classid/{endnm}    { return tok_at_classid; }
+<ATT>clear/{endnm}      { return tok_at_clear; }
+<ATT>code/{endnm}       { return tok_at_code; }
+<ATT>codebase/{endnm}   { return tok_at_codebase; }
+<ATT>codetype/{endnm}   { return tok_at_codetype; }
+<ATT>color/{endnm}      { return tok_at_color; }
+<ATT>cols/{endnm}       { return tok_at_cols; }
+<ATT>colspan/{endnm}    { return tok_at_colspan; }
+<ATT>compact/{endnm}    { return tok_at_compact; }
+<ATT>content/{endnm}    { return tok_at_content; }
+<ATT>coords/{endnm}     { return tok_at_coords; }
+<ATT>data/{endnm}       { return tok_at_data; }
+<ATT>datetime/{endnm}   { return tok_at_datetime; }
+<ATT>declare/{endnm}    { return tok_at_declare; }
+<ATT>defer/{endnm}      { return tok_at_defer; }
+<ATT>dir/{endnm}        { return tok_at_dir; }
+<ATT>disabled/{endnm}   { return tok_at_disabled; }
+<ATT>enctype/{endnm}    { return tok_at_enctype; }
+<ATT>face/{endnm}       { return tok_at_face; }
+<ATT>for/{endnm}        { return tok_at_for; }
+<ATT>frame/{endnm}      { return tok_at_frame; }
+<ATT>frameborder/{endnm}        { return tok_at_frameborder; }
+<ATT>headers/{endnm}    { return tok_at_headers; }
+<ATT>height/{endnm}     { return tok_at_height; }
+<ATT>href/{endnm}       { return tok_at_href; }
+<ATT>hreflang/{endnm}   { return tok_at_hreflang; }
+<ATT>hspace/{endnm}     { return tok_at_hspace; }
+<ATT>http-equiv/{endnm} { return tok_at_http_equiv; }
+<ATT>id/{endnm} { return tok_at_id; }
+<ATT>ismap/{endnm}      { return tok_at_ismap; }
+<ATT>label/{endnm}      { return tok_at_label; }
+<ATT>lang/{endnm}       { return tok_at_lang; }
+<ATT>language/{endnm}   { return tok_at_language; }
+<ATT>link/{endnm}       { return tok_at_link; }
+<ATT>longdesc/{endnm}   { return tok_at_longdesc; }
+<ATT>marginheight/{endnm}       { return tok_at_marginheight; }
+<ATT>marginwidth/{endnm}        { return tok_at_marginwidth; }
+<ATT>maxlength/{endnm}  { return tok_at_maxlength; }
+<ATT>media/{endnm}      { return tok_at_media; }
+<ATT>method/{endnm}     { return tok_at_method; }
+<ATT>multiple/{endnm}   { return tok_at_multiple; }
+<ATT>name/{endnm}       { return tok_at_name; }
+<ATT>nohref/{endnm}     { return tok_at_nohref; }
+<ATT>noresize/{endnm}   { return tok_at_noresize; }
+<ATT>noshade/{endnm}    { return tok_at_noshade; }
+<ATT>nowrap/{endnm}     { return tok_at_nowrap; }
+<ATT>object/{endnm}     { return tok_at_object; }
+<ATT>onblur/{endnm}     { return tok_at_onblur; }
+<ATT>onchange/{endnm}   { return tok_at_onchange; }
+<ATT>onclick/{endnm}    { return tok_at_onclick; }
+<ATT>ondblclick/{endnm} { return tok_at_ondblclick; }
+<ATT>onfocus/{endnm}    { return tok_at_onfocus; }
+<ATT>onkeydown/{endnm}  { return tok_at_onkeydown; }
+<ATT>onkeypress/{endnm} { return tok_at_onkeypress; }
+<ATT>onkeyup/{endnm}    { return tok_at_onkeyup; }
+<ATT>onload/{endnm}     { return tok_at_onload; }
+<ATT>onmousedown/{endnm}        { return tok_at_onmousedown; }
+<ATT>onmousemove/{endnm}        { return tok_at_onmousemove; }
+<ATT>onmouseout/{endnm} { return tok_at_onmouseout; }
+<ATT>onmouseover/{endnm}        { return tok_at_onmouseover; }
+<ATT>onmouseup/{endnm}  { return tok_at_onmouseup; }
+<ATT>onreset/{endnm}    { return tok_at_onreset; }
+<ATT>onselect/{endnm}   { return tok_at_onselect; }
+<ATT>onsubmit/{endnm}   { return tok_at_onsubmit; }
+<ATT>onunload/{endnm}   { return tok_at_onunload; }
+<ATT>profile/{endnm}    { return tok_at_profile; }
+<ATT>prompt/{endnm}     { return tok_at_prompt; }
+<ATT>readonly/{endnm}   { return tok_at_readonly; }
+<ATT>rel/{endnm}        { return tok_at_rel; }
+<ATT>rev/{endnm}        { return tok_at_rev; }
+<ATT>rows/{endnm}       { return tok_at_rows; }
+<ATT>rowspan/{endnm}    { return tok_at_rowspan; }
+<ATT>rules/{endnm}      { return tok_at_rules; }
+<ATT>scheme/{endnm}     { return tok_at_scheme; }
+<ATT>scope/{endnm}      { return tok_at_scope; }
+<ATT>scrolling/{endnm}  { return tok_at_scrolling; }
+<ATT>selected/{endnm}   { return tok_at_selected; }
+<ATT>shape/{endnm}      { return tok_at_shape; }
+<ATT>size/{endnm}       { return tok_at_size; }
+<ATT>span/{endnm}       { return tok_at_span; }
+<ATT>src/{endnm}        { return tok_at_src; }
+<ATT>standby/{endnm}    { return tok_at_standby; }
+<ATT>start/{endnm}      { return tok_at_start; }
+<ATT>style/{endnm}      { return tok_at_style; }
+<ATT>summary/{endnm}    { return tok_at_summary; }
+<ATT>tabindex/{endnm}   { return tok_at_tabindex; }
+<ATT>target/{endnm}     { return tok_at_target; }
+<ATT>text/{endnm}       { return tok_at_text; }
+<ATT>title/{endnm}      { return tok_at_title; }
+<ATT>type/{endnm}       { return tok_at_type; }
+<ATT>usemap/{endnm}     { return tok_at_usemap; }
+<ATT>valign/{endnm}     { return tok_at_valign; }
+<ATT>value/{endnm}      { return tok_at_value; }
+<ATT>valuetype/{endnm}  { return tok_at_valuetype; }
+<ATT>version/{endnm}    { return tok_at_version; }
+<ATT>vlink/{endnm}      { return tok_at_vlink; }
+<ATT>vspace/{endnm}     { return tok_at_vspace; }
+<ATT>width/{endnm}      { return tok_at_width; }
+<ATT>{attrname}         { return tok_at_unknown; }
+
+<ATT>[>]                { BEGIN(INITIAL); return yytext[0]; }
+<ATT>.                  { return yytext[0]; }
+
+%%