blockdiag shorturl issue inspection

gistfile1.txt

Flow analysis
==============

in compress::

  diagram = Base64.encodeURI('\x78\x9c' + RawDeflate.deflate(diagram) + adler32(diagram));

in decompress::

  source = Base64.decode(source)
  source = Base64.utob(source);
  source = source.substring(2, source.length - 4);
  source = RawDeflate.inflate(source);
  source = Base64.btou(source);

Data analysis
==============

1st inspection
--------------

For some broken shorturl, it turns out ...

::

  >>> src = $("textarea").val()
  >>> RawDeflate.inflate(RawDeflate.deflate(src)) == src
  true

::

  (Javascript)
  >>> b4 = '\x8cc\xad\x81\xc2\xa6V\n.\x99'
  >>> $.map(b4, function(c, i){return b4.charCodeAt(i)}).join(" ")
  "140 99 173 129 194 166 86 10 46 153"
  >>> Base64.encodeURI(b4)
  "jGOtgcKmVgoumQ"

  >>> b5 = Base64.decode("jGOtgcKmVgoumQ")
  >>> $.map(b5, function(c, i){return b5.charCodeAt(i)}).join(" ")
  "140 99 173 129 166 86 10 46 153"

  >>> b6 = Base64.utob(b5)
  >>> $.map(b6, function(c, i){return b6.charCodeAt(i)}).join(" ")
  "140 99 173 129 166 86 10 46 153"

::

  (Python)
  >>> import base64
  >>> b4 = '\x8cc\xad\x81\xc2\xa6V\n.\x99'
  >>> base64.b64encode(b4)
  'jGOtgcKmVgoumQ=='

It seems Base64.decode() is wrong.

2st inspection
--------------

Base64.decode() is actually a function like this::

    decode:function(a){ 
        return btou(atob(a.replace(/[-_]/g, function(m0){
            return m0 == '-' ? '+' : '/';
        })));
    }

::

  >>> b7 = atob("jGOtgcKmVgoumQ")
  >>> $.map(b7, function(c, i){return b7.charCodeAt(i)}).join(" ")
  "140 99 173 129 194 166 86 10 46 153"

  >>> b8 = Base64.btou(b7)
  >>> $.map(b8, function(c, i){return b8.charCodeAt(i)}).join(" ")
  "140 99 173 129 166 86 10 46 153"

Base64.btou() is a function defined as::

  var btou = function(bin){
      return bin.replace(re_bytes_nonascii, sub_bytes_nonascii);
  };

and here are `re_bytes_nonascii` regex and `sub_bytes_nonascii` function::


  var re_bytes_nonascii
      = /[\xC0-\xDF][\x80-\xBF]|[\xE0-\xEF][\x80-\xBF]{2}|[\xF0-\xF7][\x80-\xBF]{3}/g;

  var sub_bytes_nonascii = function(m){
      var c0 = m.charCodeAt(0);
      var c1 = m.charCodeAt(1);
      if(c0 < 0xe0){
          return String.fromCharCode(((c0 & 0x1f) << 6) | (c1 & 0x3f));
      }else{
          var c2 = m.charCodeAt(2);
          return String.fromCharCode(
              ((c0 & 0x0f) << 12) | ((c1 & 0x3f) <<  6) | (c2 & 0x3f)
          );
      }
  };

simplified case::

  >>> s="\xc2\xa6"
  "¦"
  >>> s.replace(re_bytes_nonascii, sub_bytes_nonascii)
  "¦"

  >>> c0 = 0xc2
  >>> c1 = 0xa6
  >>> String.fromCharCode(((c0 & 0x1f) << 6) | (c1 & 0x3f))
  "¦"

Bingo!