kkirby/README.md

## README.md

      
    Raw
  

              README.md
            
          
    This is a simple parser used with PegJS to parse a NETSCAPE-Bookmark-file. I've only tested this with safari export.
You have to remove these tags:
<HTML>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=UTF-8">
<Title>Bookmarks</Title>
<H1>Bookmarks</H1>
</HTML>
Example valid input:
<!DOCTYPE NETSCAPE-Bookmark-file-1>
<DT><H3 FOLDED>Favorites</H3>
<DL><p>
	<DT><A HREF="....">.....</A>
	<DT><A HREF="....">.....</A>
</DL><p>
<DT><H3 FOLDED>Bookmarks Menu</H3>
<DL><p>
	<DT><H3 FOLDED>ABC</H3>
	<DL><p>
		<DT><A HREF="....">.....</A>
		<DT><A HREF="....">.....</A>
		<DT><A HREF="....">.....</A>
		<DT><H3 FOLDED>ABC</H3>
		<DL><p>
			<DT><A HREF="....">.....</A>
		</DL><p>
	</DL><p>
</DL><p>
Output:
[
   {
      "type": "folder",
      "name": "Favorites",
      "items": [
         {
            "type": "link",
            "url": "....",
            "label": "....."
         },
         {
            "type": "link",
            "url": "....",
            "label": "....."
         }
      ]
   },
   {
      "type": "folder",
      "name": "Bookmarks Menu",
      "items": [
         {
            "type": "folder",
            "name": "ABC",
            "items": [
               {
                  "type": "link",
                  "url": "....",
                  "label": "....."
               },
               {
                  "type": "link",
                  "url": "....",
                  "label": "....."
               },
               {
                  "type": "link",
                  "url": "....",
                  "label": "....."
               },
               {
                  "type": "folder",
                  "name": "ABC",
                  "items": [
                     {
                        "type": "link",
                        "url": "....",
                        "label": "....."
                     }
                  ]
               }
            ]
         }
      ]
   }
]

  
## parser.pegjs
Document = Doctype _ root:Items _ {
	return root;
}

Doctype = "<!DOCTYPE NETSCAPE-Bookmark-file-1>"

TAG_OPEN = "<"
TAG_CLOSE = ">"
BS = "/"
EQ = "="

DT = TAG_OPEN "DT" TAG_CLOSE
DL_START = TAG_OPEN "DL" TAG_CLOSE
DL_END = TAG_OPEN BS "DL" TAG_CLOSE
P = TAG_OPEN "p" TAG_CLOSE

A_OPEN = TAG_OPEN "A" _ "HREF" EQ '"' url:$(! '"' .)+ '"' TAG_CLOSE {
	return {url};
}

A_CLOSE = TAG_OPEN BS "A" TAG_CLOSE
A = open:A_OPEN _ label:$(! A_CLOSE .)+ _ A_CLOSE {
	return {
		...open,
	    label
	}
}

FOLDED_START = TAG_OPEN "H3" _ "FOLDED" _ (! TAG_CLOSE .)* TAG_CLOSE
FOLDED_END = TAG_OPEN BS "H3" TAG_CLOSE
FOLDED = FOLDED_START name:$(! FOLDED_END .)+ FOLDED_END { return name; }

Folder = DT _ name:FOLDED _ DL_START _ P _ items:Items _ DL_END _ P {
	return {type:'folder',name,items};
}

Item = _ DT _ base:A _ {
	return {
    	type: 'link',
	    ...base
	};
}

Items = items:(_ a:Folder _ { return a; } / _ a: Item _ { return a; } )* {
	return items;
}

_ "whitespace"
  = [ \t\n\r]* {
  return undefined;
}
	Document = Doctype _ root:Items _ {
	return root;
	}

	Doctype = "<!DOCTYPE NETSCAPE-Bookmark-file-1>"

	TAG_OPEN = "<"
	TAG_CLOSE = ">"
	BS = "/"
	EQ = "="

	DT = TAG_OPEN "DT" TAG_CLOSE
	DL_START = TAG_OPEN "DL" TAG_CLOSE
	DL_END = TAG_OPEN BS "DL" TAG_CLOSE
	P = TAG_OPEN "p" TAG_CLOSE

	A_OPEN = TAG_OPEN "A" _ "HREF" EQ '"' url:$(! '"' .)+ '"' TAG_CLOSE {
	return {url};
	}

	A_CLOSE = TAG_OPEN BS "A" TAG_CLOSE
	A = open:A_OPEN _ label:$(! A_CLOSE .)+ _ A_CLOSE {
	return {
	...open,
	label
	}
	}

	FOLDED_START = TAG_OPEN "H3" _ "FOLDED" _ (! TAG_CLOSE .)* TAG_CLOSE
	FOLDED_END = TAG_OPEN BS "H3" TAG_CLOSE
	FOLDED = FOLDED_START name:$(! FOLDED_END .)+ FOLDED_END { return name; }

	Folder = DT _ name:FOLDED _ DL_START _ P _ items:Items _ DL_END _ P {
	return {type:'folder',name,items};
	}

	Item = _ DT _ base:A _ {
	return {
	type: 'link',
	...base
	};
	}

	Items = items:(_ a:Folder _ { return a; } / _ a: Item _ { return a; } )* {
	return items;
	}

	_ "whitespace"
	= [ \t\n\r]* {
	return undefined;
	}