%{
#include "parse.h"
#include <stdio.h>
#define YYMAXDEPTH 100000
char currentTitle[1024];
%}

%token START_COMMENT
%token CLOSE_TAG PLAIN_TEXT TAG_TEXT TAG_ID
%token CHARACTER_ENTITY EQUALS COMMENT_TEXT NEWLINE

%token START_HTML END_HTML START_HEAD END_HEAD START_BODY END_BODY
%token ISINDEX START_TITLE END_TITLE
%token START_UNKNOWN END_UNKNOWN

%token START_H1 END_H1 START_H2 END_H2 START_H3 END_H3
%token START_H4 END_H4 START_H5 END_H5 START_H6 END_H6
%token START_A END_A START_I END_I START_B END_B START_U END_U
%token START_S END_S START_SUP END_SUP START_SUB END_SUB 
%token START_TT END_TT START_EM END_EM START_STRONG END_STRONG
%token START_PRE END_PRE START_LIT END_LIT START_QUOTE END_QUOTE
%token START_ABSTRACT END_ABSTRACT START_BYLINE END_BYLINE START_NOTE END_NOTE
%token START_ADDRESS END_ADDRESS START_BLOCKQUOTE END_BLOCKQUOTE
%token START_OL END_OL START_UL END_UL START_LI END_LI
%token START_MENU END_MENU START_DIR END_DIR START_CITE END_CITE
%token START_DL END_DL START_DT END_DT START_DD END_DD START_FONT END_FONT
%token START_TABLE END_TABLE START_TH END_TH START_TD END_TD
%token START_TR END_TR START_TB END_TB START_CAPTION END_CAPTION

%token HR BR P TAB IMG IMAGE

%%

document : doccomments documentstart

doccomments :
    | doccomments comment
    ;

documentstart : starthtml doccontent endhtml
    | doccontent
    ;

starthtml : START_HTML optattributes CLOSE_TAG
    ;

endhtml :
    | END_HTML CLOSE_TAG {ignoreWhitespace = 1;}
    ;

doccontent : head body
    ;

head : { currentTitle[0] = '\0'; } headcontent
    | starthead { currentTitle[0] = '\0'; } headcontent endhead
    ;

starthead : START_HEAD optattributes CLOSE_TAG
    ;

endhead : END_HEAD CLOSE_TAG
    ;

headcontent :
    | headterm headcontent 
    ;

headterm : simpleheadtag
    | headtag
      {ignoreWhitespace = 0;}  headtext  {ignoreWhitespace = 1;}
      endheadtag
      { 
        if ($1.yytoken == START_TITLE) {
			strcpy(currentTitle, $3.text);
		}
	  }
    ;

headtext :
      { $$.text[0] = '\0'; }
    | PLAIN_TEXT headtext
      { strcpy($$.text, $1.text);
        strcat($$.text, $2.text); }
    | NEWLINE headtext
      { strcpy($$.text, " ");
		strcat($$.text, $2.text); }
    ;

simpleheadtag : ISINDEX optattributes CLOSE_TAG
    | comment
    ;

headtag : START_TITLE optattributes CLOSE_TAG
     { $$.yytoken = START_TITLE; }
    ;

endheadtag : END_TITLE CLOSE_TAG
    | END_UNKNOWN CLOSE_TAG
    ;

body : startbody {ignoreWhitespace = 0;} bodycontent endbody
    | {ignoreWhitespace = 0;} bodycontent
    ;

startbody : START_BODY optattributes CLOSE_TAG
    ;

endbody :
    | END_BODY CLOSE_TAG {ignoreWhitespace = 1;}
    ;

bodycontent :
    | bodycontent bodytag
    | bodycontent endbodytag
    | bodycontent PLAIN_TEXT
      { indexbodytext($2.text, currentLine); }
    | bodycontent NEWLINE
    | bodycontent CHARACTER_ENTITY
    ;


bodytag : START_UNKNOWN optattributes CLOSE_TAG
    | comment
    | START_H1 optattributes CLOSE_TAG
    | START_H2 optattributes CLOSE_TAG
    | START_H3 optattributes CLOSE_TAG
    | START_H4 optattributes CLOSE_TAG
    | START_H5 optattributes CLOSE_TAG
    | START_H6 optattributes CLOSE_TAG
    | START_A optattributes CLOSE_TAG
    | START_I optattributes CLOSE_TAG
    | START_B optattributes CLOSE_TAG
    | START_U optattributes CLOSE_TAG
    | START_S optattributes CLOSE_TAG
    | START_SUP optattributes CLOSE_TAG
    | START_SUB optattributes CLOSE_TAG
    | START_TT optattributes CLOSE_TAG
    | START_EM optattributes CLOSE_TAG
    | START_STRONG optattributes CLOSE_TAG
    | START_PRE optattributes CLOSE_TAG
    | START_LIT optattributes CLOSE_TAG
    | START_QUOTE optattributes CLOSE_TAG
    | START_ABSTRACT optattributes CLOSE_TAG
    | START_BYLINE optattributes CLOSE_TAG
    | START_NOTE optattributes CLOSE_TAG
    | START_ADDRESS optattributes CLOSE_TAG
    | START_BLOCKQUOTE optattributes CLOSE_TAG
    | START_CITE optattributes CLOSE_TAG
    | START_OL optattributes CLOSE_TAG
    | START_UL optattributes CLOSE_TAG
    | START_LI optattributes CLOSE_TAG
    | START_MENU optattributes CLOSE_TAG
    | START_DIR optattributes CLOSE_TAG
    | START_DL optattributes CLOSE_TAG
    | START_DT optattributes CLOSE_TAG
    | START_DD optattributes CLOSE_TAG
    | START_FONT optattributes CLOSE_TAG
    | START_TABLE optattributes CLOSE_TAG
    | START_TH optattributes CLOSE_TAG
    | START_TD optattributes CLOSE_TAG
    | START_TR optattributes CLOSE_TAG
    | START_TB optattributes CLOSE_TAG
    | START_CAPTION optattributes CLOSE_TAG
    | HR optattributes CLOSE_TAG
    | BR optattributes CLOSE_TAG
    | P optattributes CLOSE_TAG
    | TAB optattributes CLOSE_TAG
    | IMG optattributes CLOSE_TAG
    | IMAGE optattributes CLOSE_TAG
    ;

optcomment :
    | COMMENT_TEXT
    ;

comment : START_COMMENT COMMENT_TEXT CLOSE_TAG
      { indexcomment($2.text); }

optattributes : attributes optcomment
    ;

attributes :
    | TAG_ID attributes
    | TAG_ID EQUALS TAG_TEXT attributes
    | TAG_ID EQUALS TAG_ID attributes
    ;

endbodytag : END_UNKNOWN CLOSE_TAG
    | END_H1 CLOSE_TAG
    | END_H2 CLOSE_TAG
    | END_H3 CLOSE_TAG
    | END_H4 CLOSE_TAG
    | END_H5 CLOSE_TAG
    | END_H6 CLOSE_TAG
    | END_A CLOSE_TAG
    | END_I CLOSE_TAG
    | END_B CLOSE_TAG
    | END_U CLOSE_TAG
    | END_S CLOSE_TAG
    | END_SUP CLOSE_TAG
    | END_SUB CLOSE_TAG
    | END_TT CLOSE_TAG
    | END_EM CLOSE_TAG
    | END_STRONG CLOSE_TAG
    | END_PRE CLOSE_TAG
    | END_LIT CLOSE_TAG
    | END_QUOTE CLOSE_TAG
    | END_ABSTRACT CLOSE_TAG
    | END_BYLINE CLOSE_TAG
    | END_NOTE CLOSE_TAG
    | END_ADDRESS CLOSE_TAG
    | END_BLOCKQUOTE CLOSE_TAG
    | END_CITE CLOSE_TAG
    | END_OL CLOSE_TAG
    | END_UL CLOSE_TAG
    | END_LI CLOSE_TAG
    | END_MENU CLOSE_TAG
    | END_DIR CLOSE_TAG
    | END_DL CLOSE_TAG
    | END_DT CLOSE_TAG
    | END_DD CLOSE_TAG
    | END_FONT CLOSE_TAG
    | END_TABLE CLOSE_TAG
    | END_TH CLOSE_TAG
    | END_TD CLOSE_TAG
    | END_TR CLOSE_TAG
    | END_TB CLOSE_TAG
    | END_CAPTION CLOSE_TAG
    ;

%%