maximvl · July 22, 2017 10:16
diff --git a/html-parser.red b/html-parser.red
 Red []
 {
 grammar HTML
    document          <-   (doctype / text / tag)*
    tag               <-   open_tag (text / tag)* close_tag
    open_tag          <-   "<" [0-9a-zA-Z \"'=-]+ ">"
    close_tag         <-   "</" [0-9a-zA-Z]+ ">"
    doctype           <-   "<!DOCTYPE " [0-9a-zA-Z]+ ">"
    text              <-   [^<]+
 }

 ws: charset reduce [newline space tab]
 digits: charset {0123456789}
 chars: union charset [#"a" - #"z"] charset [#"A" - #"Z"]
 alphanum: union digits chars
 alphanum-with-specials: union ws union alphanum charset {"'=-}

 tags-stack: copy []

 handle-open-tag:  func [name] [
 	append tags-stack name
 	;print ["open" name]
 	print tags-stack
 ]
 handle-close-tag: func [name] [
 	take/last tags-stack
 	;print ["close" name]
 	print tags-stack
 ]

 document: [any [ahead "<" [ tag | doctype ] | text]]
 tag: [whitespace open-tag any [ahead not "<" text | tag] close-tag]
 open-tag: ["<" copy name tag-name (handle-open-tag name) any tag-parameter ">"]
 tag-name: [some alphanum]
 tag-parameter: [whitespace some alphanum opt ["=" "^"" some [not "^"" skip] "^""] ]
 close-tag: ["</" copy name tag-name (handle-close-tag name) ">"]
 doctype: ["<!DOCTYPE " some alphanum ">"]
 text: [any [not "<" skip]]
 whitespace: [any ws]

 html: {

 <html>
 <body>
 <img src="picture1.jpg" alt="<title>"></img>тут точно не тайтл<img src="picture2.jpg" alt="</title>"></img>

 <img src="picture1.jpg" alt="<u>"></img>тут точно не подчеркнуто<img src="picture2.jpg" alt="</u>"></img>
 <u>а тут подчеркнуто</u>
 </body>
 </html>

 }

 probe parse html document
	Red []
	{
	grammar HTML
	document <- (doctype / text / tag)*
	tag <- open_tag (text / tag)* close_tag
	open_tag <- "<" [0-9a-zA-Z \"'=-]+ ">"
	close_tag <- "</" [0-9a-zA-Z]+ ">"
	doctype <- "<!DOCTYPE " [0-9a-zA-Z]+ ">"
	text <- [^<]+
	}

	ws: charset reduce [newline space tab]
	digits: charset {0123456789}
	chars: union charset [#"a" - #"z"] charset [#"A" - #"Z"]
	alphanum: union digits chars
	alphanum-with-specials: union ws union alphanum charset {"'=-}

	tags-stack: copy []

	handle-open-tag: func [name] [
	append tags-stack name
	;print ["open" name]
	print tags-stack
	]
	handle-close-tag: func [name] [
	take/last tags-stack
	;print ["close" name]
	print tags-stack
	]

	document: [any [ahead "<" [ tag \| doctype ] \| text]]
	tag: [whitespace open-tag any [ahead not "<" text \| tag] close-tag]
	open-tag: ["<" copy name tag-name (handle-open-tag name) any tag-parameter ">"]
	tag-name: [some alphanum]
	tag-parameter: [whitespace some alphanum opt ["=" "^"" some [not "^"" skip] "^""] ]
	close-tag: ["</" copy name tag-name (handle-close-tag name) ">"]
	doctype: ["<!DOCTYPE " some alphanum ">"]
	text: [any [not "<" skip]]
	whitespace: [any ws]

	html: {

	<html>
	<body>
	<img src="picture1.jpg" alt="<title>"></img>тут точно не тайтл<img src="picture2.jpg" alt="</title>"></img>

	<img src="picture1.jpg" alt="<u>"></img>тут точно не подчеркнуто<img src="picture2.jpg" alt="</u>"></img>
	<u>а тут подчеркнуто</u>
	</body>
	</html>

	}

	probe parse html document