Annotation-System/hmmlib-js/src/source.l

%{
	function ERR(yy, err){
		console.log("hmmlib error: L%d: %s\n", yy.line, err);
		throw { message: err, line: yy.line };
	}

	function CHECKESCAPE(yy, str){
		if("[]:@~\\\"".indexOf(str) == -1){
			ERR(yy, "hmmlib: Unknown backslash escape code: %s", str);
		}
	}

	function UNQUOTE(yy, str){
		var i = 0;
		var j = 0;
		while((j = str.indexOf('\\', i)) != -1){
			CHECKESCAPE(yy, str.charAt(j));
			str = str.slice(i, j) + str.slice(j+1);
			i = j+1;
		}
		return str;
	}

	function M_(yy, str, s){
		yy.an.markers.push({ type: str });
		yy.mnext = s;
	}

	function M_ADD(yy, t, n){
		yy.an.markers[yy.an.markers.length - 1].marker = t.substr(0, n);
		yy.an.markers[yy.an.markers.length - 1].offset = yy.an.text.length;
		if(yy.mnext === "TEXT"){
			yy.an.text += t.substr(0, n);
		}
	}

	function MX_ADD(yy, c){
		var m = yy.an.markers[yy.an.markers.length - 1];
		if(m.parameter){
			m.parameter += c;
		} else {
			m.parameter = c;
		}
		yy.an.text += c;
	}

	function NEWANNO(yy){
		if(!yy.first) yy.annos.push(yy.an);
		yy.an = {
			line: yy.line,
			text: "",
			references: [],
			markers: [],
		};
		yy.an.line = yy.line;
		yy.first = false;
	}
%}

%option reentrant
%option noyywrap

S               [\t \r]*
SP              [\t \r]+
ATTR_SIMPLE     [^\" \]\t\r\n][^ \]\t\r\n]*
ATTR_ALNUM      [0-9a-zA-Z][0-9a-zA-Z_]*
ATTR_QUOTED     \"([^\n\"\\]|\\.)*\"
TAG_VIDEO_OPEN  "[video"
TIMECODE        \[[0-9]{1,2}(\:[0-5][0-9]){1,2}\]
BAD_TIMECODE    \[[0-9]{1,2}(\:[6-9][0-9]){1,2}\]
TEXT_BREAK      [^\\\:\@\~\[\]\r\n\t ]
LB              \[
RB              \]

%s VIDEO
%s V_ATTR
%s V2_ATTR
%s ANNOTATION
%s TEXT_START
%s TEXT
%s MARKER
%s MARKER_XTRA
%s REF
%s R_ATTR
%s AFTERTEXT
%s AUTHOR
%s CATEGORIES
%s QUOTES

%%

<<EOF>> { ERR(yy, "Unexpected EOF, video close tag not found.");  }
\r\n|\n { yy.line++; }

<INITIAL>{TAG_VIDEO_OPEN} { yy_.begin("VIDEO"); }
<INITIAL>.                { ERR(yy, "Missing video tag.");  }

<VIDEO>{SP}                 {}
<VIDEO>member{S}\=          { yy.attr = "member";          yy_.begin("V_ATTR"); }
<VIDEO>stream_platform{S}\= { yy.attr = "stream_platform"; yy_.begin("V_ATTR"); }
<VIDEO>stream_username{S}\= { yy.attr = "stream_username"; yy_.begin("V_ATTR"); }
<VIDEO>project{S}\=         { yy.attr = "project";         yy_.begin("V_ATTR"); }
<VIDEO>title{S}\=           { yy.attr = "title";           yy_.begin("V_ATTR"); }
<VIDEO>vod_platform{S}\=    { yy.attr = "vod_platform";    yy_.begin("V_ATTR"); }
<VIDEO>id{S}\=              { yy.attr = "id";              yy_.begin("V_ATTR"); }
<VIDEO>co\-host{S}\=        { yy.attr = "co_hosts"         yy_.begin("V2_ATTR"); }
<VIDEO>guest{S}\=           { yy.attr = "guests"           yy_.begin("V2_ATTR"); }
<VIDEO>annotator{S}\=       { yy.attr = "annotators";      yy_.begin("V2_ATTR"); }
<VIDEO>\]                   { yy_.begin("ANNOTATION"); };
<VIDEO>.                    { ERR(yy, "Invalid char '"+ yytext +"' in video tag.");  }

<V_ATTR>{SP}          { yy_.begin("VIDEO"); }
<V_ATTR>{ATTR_SIMPLE} { yy.meta[yy.attr] = yytext; yy_.begin("VIDEO"); }
<V_ATTR>{ATTR_QUOTED} { yy.meta[yy.attr] = UNQUOTE(yy, yytext.substr(1, yyleng-2)); yy_.begin("VIDEO"); }
<V_ATTR>\]            { yy_.less(0); yy_.begin("VIDEO"); }

<V2_ATTR>{SP}          { yy_.begin("VIDEO"); }
<V2_ATTR>{ATTR_SIMPLE} { yy.meta[yy.attr].push(yytext); yy_.begin("VIDEO"); }
<V2_ATTR>{ATTR_QUOTED} { yy.meta[yy.attr].push(UNQUOTE(yy, yytext.substr(1, yyleng-2))); yy_.begin("VIDEO"); }
<V2_ATTR>\]            { yy_.less(0); yy_.begin("VIDEO"); }

<ANNOTATION>\[\/video\]      { NEWANNO(yy); return 1; }
<ANNOTATION>{TIMECODE}{LB}\@ { NEWANNO(yy); yy.an.time = yytext.substr(1, yyleng-4); yy_.begin("AUTHOR"); }
<ANNOTATION>{TIMECODE}       { NEWANNO(yy); yy.an.time = yytext.substr(1, yyleng-2); yy_.begin("TEXT_START"); }
<ANNOTATION>{BAD_TIMECODE}   { ERR(yy, "Timecode '"+ yytext +"' out of range.");  }
<ANNOTATION>{SP}             {}
<ANNOTATION>.                { ERR(yy, "Cannot parse annotation. Expected timecode.");  }

<TEXT_START>{LB}\: { M_(yy, "CATEGORY", "TEXT"); yy_.begin("MARKER"); }
<TEXT_START>{LB}\@ { M_(yy, "MEMBER"  , "TEXT"); yy_.begin("MARKER"); }
<TEXT_START>{LB}\~ { M_(yy, "PROJECT" , "TEXT"); yy_.begin("MARKER"); }
<TEXT_START>{LB}   { yy_.less(0); yy_.begin("TEXT"); }
<TEXT_START>.      { ERR(yy, "Unknown character '"+ yytext +"' after timecode.");  }

<TEXT>{TEXT_BREAK}+ { yy.an.text += yytext; }
<TEXT>\\.           { CHECKESCAPE(yy, yytext.charAt(1)); yy.an.text += yytext.substr(1, yyleng-1); }
<TEXT>[ \r\t]+\:    { yy.an.text += ' '; M_(yy, "CATEGORY", "TEXT"); yy_.begin("MARKER"); }
<TEXT>[ \r\t]+\@    { yy.an.text += ' '; M_(yy, "MEMBER"  , "TEXT"); yy_.begin("MARKER"); }
<TEXT>[ \r\t]+\~    { yy.an.text += ' '; M_(yy, "PROJECT" , "TEXT"); yy_.begin("MARKER"); }
<TEXT>{LB}\:        { M_(yy, "CATEGORY", "MARKER_XTRA"); yy_.begin("MARKER"); }
<TEXT>{LB}\@        { M_(yy, "MEMBER"  , "MARKER_XTRA"); yy_.begin("MARKER"); }
<TEXT>{LB}\~        { M_(yy, "PROJECT" , "MARKER_XTRA"); yy_.begin("MARKER"); }
<TEXT>\]            { yy_.begin("AFTERTEXT"); }
<TEXT>{LB}ref       { yy.ref.offset = yy.an.text.length; yy_.begin("REF"); }
<TEXT>{LB}          {}
<TEXT>{SP}          { yy.an.text += ' '; }
<TEXT>.             { yy.an.text += yytext; }

<MARKER>{ATTR_ALNUM}  { M_ADD(yy, yytext); yy_.begin(yy.mnext); };
<MARKER>{ATTR_QUOTED} { M_ADD(yy, yytext.substr(1, yyleng-2)); yy_.begin(yy.mnext); };
<MARKER>.             { ERR(yy, "Cannot parse Marker. Expected quoted or alphanumeric attribute.");  }

<MARKER_XTRA>\\]  { MX_ADD(yy, ']'); }
<MARKER_XTRA>\\\# { MX_ADD(yy, '#'); }
<MARKER_XTRA>\]   { yy_.begin("TEXT"); }
<MARKER_XTRA>[ ]  { if(yy.an.markers[yy.an.markers.length - 1].parameter){ MX_ADD(yy, ' '); } }
<MARKER_XTRA>[ ]\#[0-9]+ {
	var m = yy.an.markers[yy.an.markers.length - 1];
	if(m.type == "PROJECT"){
		m.episode = yytext.substr(2);
	} else {
		MX_ADD(yytext.substr(1));
	}
}
<MARKER_XTRA>.    { MX_ADD(yy, yytext); }

<REF>{SP}           {}
<REF>site{S}\=      { yy.attr = "site";      yy_.begin("R_ATTR"); }
<REF>page{S}\=      { yy.attr = "page";      yy_.begin("R_ATTR"); }
<REF>url{S}\=       { yy.attr = "url";       yy_.begin("R_ATTR"); }
<REF>title{S}\=     { yy.attr = "title";     yy_.begin("R_ATTR"); }
<REF>article{S}\=   { yy.attr = "article";   yy_.begin("R_ATTR"); }
<REF>author{S}\=    { yy.attr = "author";    yy_.begin("R_ATTR"); }
<REF>editor{S}\=    { yy.attr = "editor";    yy_.begin("R_ATTR"); }
<REF>publisher{S}\= { yy.attr = "publisher"; yy_.begin("R_ATTR"); }
<REF>isbn{S}\=      { yy.attr = "isbn";      yy_.begin("R_ATTR"); }
<REF>\]             { yy.an.references.push(yy.ref); yy.ref = {}; yy_.begin("TEXT"); }
<REF>.              { ERR(yy, "Unexpected item in ref: " + yytext); }

<R_ATTR>{SP}          {}
<R_ATTR>{ATTR_SIMPLE} { yy.ref[yy.attr] = yytext; yy_.begin("REF"); }
<R_ATTR>{ATTR_QUOTED} { yy.ref[yy.attr] = UNQUOTE(yy, yytext.substr(1, yyleng-2)); yy_.begin("REF"); }

<AFTERTEXT>\[\/video\] { NEWANNO(yy); return 1; }

<AFTERTEXT>{SP}        {}
<AFTERTEXT>{LB}quote   { yy_.begin("QUOTES"); }
<AFTERTEXT>{LB}\:      { yy_.begin("CATEGORIES"); yy_.less(1); }
<AFTERTEXT>{LB}[0-9]   { yy_.begin("ANNOTATION"); yy_.less(0); }
<AFTERTEXT>..          { ERR(yy, "Unexpected thing after text node: " + yytext); }
<AFTERTEXT>.           { ERR(yy, "Unexpected thing after text node: " + yytext); }

<AUTHOR>[^\]\n]+\] { yy.an.author = yytext.substr(0, yyleng-1); yy_.begin("TEXT_START"); }
<AUTHOR>{SP}       {}

<CATEGORIES>{SP}            {}
<CATEGORIES>\:{ATTR_SIMPLE} { yy.an.markers.push({ type: "CATEGORY", marker: yytext.substr(1, yyleng-1), offset: -1 }); }
<CATEGORIES>\:{ATTR_QUOTED} { yy.an.markers.push({ type: "CATEGORY", marker: UNQUOTE(yy, yytext.substr(2, yyleng-3)), offset: -1 }); }
<CATEGORIES>\]{LB}          { yy_.begin("QUOTES"); }
<CATEGORIES>\]              { yy_.begin("ANNOTATION"); }
<CATEGORIES>.               { ERR(yy, "Unexpected character in category tag: " + yytext); }

<QUOTES>{SP}         {}
<QUOTES>[0-9]+{S}\]  { yy.an.quote = { author: "", id: parseInt(yytext) }; yy_.begin("ANNOTATION"); }
<QUOTES>{ATTR_ALNUM} { yy.an.quote.author += yytext; }
<QUOTES>.            { ERR(yy, "Unexpected character in quotes tag: " + yytext); }

%%