view release on metacpan or  search on metacpan

t/02-mime.t  view on Meta::CPAN

#!perl

use Test::More;
use FindBin;
BEGIN { unshift @INC, "$1/../blib/lib" if $FindBin::Bin =~ m{(.*)} };
use File::Unpack2;
use Data::Dumper;

diag("File::MimeInfo::Magic missing\n") unless $INC{'File/MimeInfo/Magic.pm'};
diag("File::LibMagic missing\n") unless $INC{'File/LibMagic.pm'};
my $shared_mime_info_db = '/usr/share/mime/magic';

my $u = File::Unpack2->new();

my $d = "data"; $d = "t/data" unless -d $d;
opendir DIR, $d or diag("where is my test data?");
my @f = sort grep { !/^\./ } readdir DIR;
closedir DIR;

my $sample = 'monotone.info';	# one of the below files, without regexps, for further tests.
%exp = 
(
  # Filename => [ mimetype, encoding, ... comments ]

  ## these two are from SUSE:Factory:Head/qpdf%5.1.0%r23/qpdf-5.1.0/qpdf/qtest/qpdf/
  'bad34.pdf' => 
  	[ 'application/pdf', 'us-ascii', 'PDF document, version 1.3' ],
  'good10.pdf' => 
  	[ 'application/pdf', 'us-ascii', 'PDF document, version 1.3' ],

  ## 0.22 used to say application/x-lzma, but true binary data. Not even compressed.
  'lxknf09SCc0.bin' => 
  	[ 'application/octet-stream', qr{^(binary|unknown|)$} ], 

  ## actually 'application/x-desktop' or 'text/x-desktop'
  'Desktop.directory' => 
  	[ 'text/plain', 'utf-8', 'UTF-8 Unicode text' ],

  ## text/plain seen on 12.1, was text/x-desktop before
  'xterm-snippet.desktop' => 
  	[ qr{^text/(plain|x\-desktop)$}, 'utf-8', 
	 'UTF-8 Unicode Pascal program text', ['text/x-pascal','application/x-desktop']],

  'IPA-snippet.pfa' => 
  	[ 'text/x-font-type1', qr{^(us-ascii|)$}, 
	  'PostScript Type 1 font text (OmegaSerifIPA 001.000)', 
	  [ 'text/plain', 'application/x-font-type1' ] ],

  'Times-Roman-snippet.afm' => 
  	[ qr{^(application|text)/x-font-sunos-news$}, 
	  'us-ascii','ASCII font metrics',['text/x-fortran','application/x-font-sunos-news']], 

  ## actually 'text/x-xslfo'
  'columns-snippet.fo' => 
    [ qr{^(text/plain|application/xml|text/x-application-xml|text/xml)$}, 'us-ascii',
	  'XML  document text'],

  ## actually 'application/x-pax
  'Archive.pax' => 
  	[ 'application/x-cpio', qr{^(binary|unknown)$},
	  'ASCII cpio archive (pre-SVR4 or odc)' ],

  'empty.odt' => 
  	[ 'application/vnd.oasis.opendocument.text+zip', qr{^(binary|unknown|)$},
	  'Zip archive data, at least v2.0 to extract, mime type application/vnd OpenDocument Text'],

  'ruhyphal.tex' => 
  	[ 'text/plain','iso-8859-1', 
	  'ISO-8859 English text'],

  # File-LibMagic-0.96 at SLE11-SP1 reports text/html, 
  # File-LibMagic-0.96 at openSUSE-12.2 reports text/plain, 
  'test.mht' => 
  	[ qr(^text/(html|plain)$), 'iso-8859-1', 
	  'multipart/related; start=<op.mhtml.1250319979062.7d507541390148, '],

  'test2.tga' => 
  	[ 'image/x-tga', qr{^(binary|unknown|)$},
	  'Targa image data - RGB - RLE 32 x 32',
	  ['application/octet-stream','image/x-tga']],

  ## actually a 'audio/x-mpegurl'; newer shared-mime-info may return text/plain
  'wzbc-2009-06-28-17-00.m3u' => 
  	[ qr{^(audio/x-mpegurl|text/plain)$}, 'us-ascii',
	  'M3U playlist text'],

  ## File::LibMagic says application/octet-stream here:
  'monotone.info' => 
  	[ 'application/x-text-mixed', qr{^(binary|unknown)$}, 
	  'data', ['application/octet-stream','application/x-text-mixed']],

  ## this is actually plain text, but we are fooled by its apparent magic.
  'pdftex-a.txt' =>  
  	[ 'application/pdf', 'utf-8', 
	  'PDF document, version 1.4' ],

  'intermediate-removal-test.tar.gz' =>
  	[ qr{^application/(x-tar\+gzip|x-gzip|gzip)$}, qr{^(binary|unknown|)$},
	  'gzip compressed data' ],

  'recursive-3level-removal-test.tar.gz' =>
  	[ qr{^application/(x-tar\+gzip|x-gzip|gzip)$}, qr{^(binary|unknown|)$},