man Encode::ZapCP1252 (3): Zap Windows Western Gremlins

Synopsis

  use Encode::ZapCP1252;
  # Zap or fix in-place.
  zap_cp1252 $latin1_text;
  fix_cp1252 $utf8_text;
  # Zap or fix copy.
  my $clean_latin1 = zap_cp1252 $latin1_text;
  my $fixed_utf8   = fix_cp1252 $utf8_text;

Description

Have you ever been processing a Web form submit for feed, assuming that the incoming text was encoded as specified in the Content-Type header, or in the XML declaration, only to end up with a bunch of junk because someone pasted in content from Microsoft Word? Well, this is because Microsoft uses a superset of the Latin-1 encoding called ``Windows Western'' or ``CP1252''. If the specified encoding is Latin-1, mostly things will come out right, but a few things---like curly quotes, m-dashes, ellipses, and the like---may not. The differences are well-known; you see a nice chart at documenting the differences on Wikipedia <http://en.wikipedia.org/wiki/Windows-1252>.

Of course, that won't really help you. What will help you is to quit using Latin-1 and switch to UTF-8. Then you can just convert from CP1252 to UTF-8 without losing a thing, just like this:

  use Encode;
  $text = decode 'cp1252', $text, 1;

But I know that there are those of you out there stuck with Latin-1 and who don't want any junk characters from Word users. That's where this module comes in. Its "zap_cp1252" function will zap those CP1252 gremlins for you, turning them into their appropriate ASCII approximations.

Another case that can occasionally come up is when you're reading reading in text that claims to be UTF-8, but it still ends up with some CP1252 gremlins mixed in with properly encoded characters. I've seen examples of just this sort of thing when processing GMail messages and attempting to insert them into a UTF-8 database, as well as in some feeds processed by, say Yahoo! Pipes <http://pipes.yahoo.com>. Doesn't work so well. For such cases, there's "fix_cp1252", which converts those CP1252 gremlins into their UTF-8 equivalents.

Usage

This module exports two subroutines: "zap_cp1252()" and "fix_cp1252()", each of which accept a single argument:

  zap_cp1252 $text;
  fix_cp1252 $text;

When called in a void context, as in these examples, "zap_cp1252()" and "fix_cp1252()" subroutine perform in place conversions of any CP1252 gremlins into their appropriate ASCII approximations or UTF-8 equivalents, respectively. Note that because the conversion happens in place, the data to be converted cannot be a string constant; it must be a scalar variable.

When called in a scalar or list context, on the other hand, a copy will be modifed and returned. The original string will be unchanged:

  my $clean_latin1 = zap_cp1252 $latin1_text;
  my $fixed_utf8   = fix_cp1252 $utf8_text;

In this case, even constant values can be processed. Either way, "undef"s will be ignored.

In Perl 5.8.8 and higher, the conversion will work even when the string is decoded to Perl's internal form (usually via "decode 'ISO-8859-1', $text") or the string is encoded (and thus simply processed by Perl as a series of bytes). The conversion will even work on a string that has not been decoded but has had its "utf8" flag flipped anyway (usually by an injudicious use of "Encode::_utf8_on()". This is to enable the highest possible likelihood of removing those CP1252 gremlins no matter what kind of processing has already been executed on the string.

In Perl 5.10 and higher, the functions may optionally be called with no arguments, in which case $_ will be converted, instead:

  zap_cp1252; # Modify $_ in-place.
  fix_cp1252; # Modify $_ in-place.
  my $zapped = zap_cp1252; # Copy $_ and return zapped
  my $fixed = zap_cp1252; # Copy $_ and return fixed

Conversion Table

Here's how the characters are converted to ASCII and UTF-8. The ASCII conversions are not perfect, but they should be good enough for general cleanup. If you want perfect, switch to UTF-8 and be done with it!

   Hex | Char  | ASCII | UTF-8 Name
  -----+-------+-------+-------------------------------------------
  0x80 |   X   |   e   | EURO SIGN
  0x82 |   X   |   ,   | SINGLE LOW-9 QUOTATION MARK
  0x83 |   X   |   f   | LATIN SMALL LETTER F WITH HOOK
  0x84 |   X   |   ,,  | DOUBLE LOW-9 QUOTATION MARK
  0x85 |   X   |  ...  | HORIZONTAL ELLIPSIS
  0x86 |   X   |   +   | DAGGER
  0x87 |   X   |   ++  | DOUBLE DAGGER
  0x88 |   X   |   ^   | MODIFIER LETTER CIRCUMFLEX ACCENT
  0x89 |   X   |   %   | PER MILLE SIGN
  0x8a |   X   |   S   | LATIN CAPITAL LETTER S WITH CARON
  0x8b |   X   |   <   | SINGLE LEFT-POINTING ANGLE QUOTATION MARK
  0x8c |   X   |   OE  | LATIN CAPITAL LIGATURE OE
  0x8e |   X   |   Z   | LATIN CAPITAL LETTER Z WITH CARON
  0x91 |   X   |   '   | LEFT SINGLE QUOTATION MARK
  0x92 |   X   |   '   | RIGHT SINGLE QUOTATION MARK
  0x93 |   X   |   "   | LEFT DOUBLE QUOTATION MARK
  0x94 |   X   |   "   | RIGHT DOUBLE QUOTATION MARK
  0x95 |   X   |   *   | BULLET
  0x96 |   X   |   -   | EN DASH
  0x97 |   X   |   --  | EM DASH
  0x98 |   X   |   ~   | SMALL TILDE
  0x99 |   X   |  (tm) | TRADE MARK SIGN
  0x9a |   X   |   s   | LATIN SMALL LETTER S WITH CARON
  0x9b |   X   |   >   | SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
  0x9c |   X   |   oe  | LATIN SMALL LIGATURE OE
  0x9e |   X   |   z   | LATIN SMALL LETTER Z WITH CARON
  0x9f |   X   |   Y   | LATIN CAPITAL LETTER Y WITH DIAERESIS

Changing the Tables

Don't like these conversions? You can modify them to your heart's content by accessing this module's internal conversion tables. For example, if you wanted "zap_cp1252()" to use an uppercase ``E'' for the euro sign, just do this:

  local $Encode::ZapCP1252::ascii_for{"\x80"} = 'E';

Or if, for some bizarre reason, you wanted the UTF-8 equivalent for a bullet converted by "fix_cp1252()" to really be an asterisk (why would you? Just use "zap_cp1252" for that!), you can do this:

  local $Encode::ZapCP1252::utf8_for{"\x95"} = '*';

Just remember, without "local" this would be a global change. In that case, be careful if your code zaps CP1252 elsewhere. Of course, it shouldn't really be doing that. These functions are just for cleaning up messes in one spot in your code, not for making a fundamental part of your text handling. For that, use Encode.

Support

This module is stored in an open GitHub repository <http://github.com/theory/encode-cp1252/tree/>. Feel free to fork and contribute!

Please file bug reports via GitHub Issues <http://github.com/theory/encode-cp1252/issues/> or by sending mail to [email protected] <mailto:[email protected]>.

Author

David E. Wheeler <[email protected]>

Acknowledgments

My thanks to Sean Burke for sending me his original method for converting CP1252 gremlins to more-or-less appropriate ASCII characters.

Copyright and License

This module is free software; you can redistribute it and/or modify it under the same terms as Perl itself.