Convert::UUlib(3pm)

NAME

Convert::UUlib - decode uu/xx/b64/mime/yenc/etc-encoded data from a massive number of files

SYNOPSIS

 use Convert::UUlib ':all';
 
 # read all the files named on the commandline and decode them
 # into the CURRENT directory. See below for a longer example.
 LoadFile $_ for @ARGV;
 for my $uu (GetFileList) {
    if ($uu->state & FILE_OK) {
      $uu->decode;
      print $uu->filename, "\n";
    }
 }

DESCRIPTION

This module started as an interface to the uulib/uudeview library by Frank Pilhofer that can be used to decode all kinds of usenet (and other) binary messages.

After upstream abondoned the project, th library was continuously bugfixed and improved in this module, with major focuses on security fixes, correctness and speed (that does not mean that this library is considered safe with untrusted data, but it surely is safer than the poriginal uudeview).

Read the file doc/library.pdf from the distribution for in-depth information about the C-library used in this interface, and the rest of this document and especially the non-trivial decoder program at the end.

EXPORTED CONSTANTS

Action code constants

  ACT_IDLE      we don't do anything
  ACT_SCANNING  scanning an input file
  ACT_DECODING  decoding into a temp file
  ACT_COPYING   copying temp to target
  ACT_ENCODING  encoding a file

Message severity levels

  MSG_MESSAGE   just a message, nothing important
  MSG_NOTE      something that should be noticed
  MSG_WARNING   important msg, processing continues
  MSG_ERROR     processing has been terminated
  MSG_FATAL     decoder cannot process further requests
  MSG_PANIC     recovery impossible, app must terminate

Options

  OPT_VERSION   version number MAJOR.MINORplPATCH (ro)
  OPT_FAST      assumes only one part per file
  OPT_DUMBNESS  switch off the program's intelligence
  OPT_BRACKPOL  give numbers in [] higher precendence
  OPT_VERBOSE   generate informative messages
  OPT_DESPERATE try to decode incomplete files
  OPT_IGNREPLY  ignore RE:plies (off by default)
  OPT_OVERWRITE whether it's OK to overwrite ex. files
  OPT_SAVEPATH  prefix to save-files on disk
  OPT_IGNMODE   ignore the original file mode
  OPT_DEBUG     print messages with FILE/LINE info
  OPT_ERRNO     get last error code for RET_IOERR (ro)
  OPT_PROGRESS  retrieve progress information
  OPT_USETEXT   handle text messages
  OPT_PREAMB    handle Mime preambles/epilogues
  OPT_TINYB64   detect short B64 outside of Mime
  OPT_ENCEXT    extension for single-part encoded files
  OPT_REMOVE    remove input files after decoding (dangerous)
  OPT_MOREMIME  strict MIME adherence
  OPT_DOTDOT    ".."-unescaping has not yet been done on input files
  OPT_RBUF      set default read I/O buffer size in bytes
  OPT_WBUF      set default write I/O buffer size in bytes
  OPT_AUTOCHECK automatically check file list after every loadfile

Result/Error codes

  RET_OK        everything went fine
  RET_IOERR     I/O Error - examine errno
  RET_NOMEM     not enough memory
  RET_ILLVAL    illegal value for operation
  RET_NODATA    decoder didn't find any data
  RET_NOEND     encoded data wasn't ended properly
  RET_UNSUP     unsupported function (encoding)
  RET_EXISTS    file exists (decoding)
  RET_CONT      continue -- special from ScanPart
  RET_CANCEL    operation canceled

File States

 This code is zero, i.e. "false":
  UUFILE_READ   Read in, but not further processed
 The following state codes are or'ed together:
  FILE_MISPART  Missing Part(s) detected
  FILE_NOBEGIN  No 'begin' found
  FILE_NOEND    No 'end' found
  FILE_NODATA   File does not contain valid uudata
  FILE_OK       All Parts found, ready to decode
  FILE_ERROR    Error while decoding
  FILE_DECODED  Successfully decoded
  FILE_TMPFILE  Temporary decoded file exists

Encoding types

  UU_ENCODED    UUencoded data
  B64_ENCODED   Mime-Base64 data
  XX_ENCODED    XXencoded data
  BH_ENCODED    Binhex encoded
  PT_ENCODED    Plain-Text encoded (MIME)
  QP_ENCODED    Quoted-Printable (MIME)
  YENC_ENCODED  yEnc encoded (non-MIME)

EXPORTED FUNCTIONS

Initializing and cleanup

Initialize is automatically called when the module is loaded and allocates quite a small amount of memory for todays machines ;) CleanUp releases that again.

On my machine, a fairly complete decode with DBI backend needs about 10MB RSS to decode 20000 files.

CleanUp: Release memory, file items and clean up files. Should be called after a decoidng run, if you want to start a new one.

Setting and querying options

$option = GetOption OPT_xxx

SetOption OPT_xxx, opt-value

See the "OPT_xxx" constants above to see which options exist.

Setting various callbacks

SetMsgCallback [callback-function]

SetBusyCallback [callback-function]

SetFileCallback [callback-function]

SetFNameFilter [callback-function]

Call the currently selected FNameFilter

$file = FNameFilter $file

Loading sourcefiles, optionally fuzzy merge and start decoding

($retval, $count) = LoadFile $fname, [$id, [$delflag, [$partno]]]

Load the given file and scan it for encoded contents. Optionally tag it with the given id, and if $delflag is true, delete the file after it is no longer necessary. If you are certain of the part number, you can specify it as the last argument.

A better (usually faster) way of doing this is using the "SetFNameFilter" functionality.

$retval = Smerge $pass

If you are desperate, try to call "Smerge" with increasing $pass values, beginning at 0, to try to merge parts that usually would not have been merged.

Most probably this will result in garbled files, so never do this by default, except:

If the "OPT_AUTOCHECK" option has been disabled (by default it is enabled) to speed up file loading, then you have to call "Smerge -1" after loading all files as an additional pre-pass (which is normally done by "LoadFile").

$item = GetFileListItem $item_number

Return the $item structure for the $item_number'th found file, or "undef" of no file with that number exists.

The first file has number 0, and the series has no holes, so you can iterate over all files by starting with zero and incrementing until you hit "undef".

This function has to walk the linear list of fils on each access, so if you want to iterate over all items, it is usually faster to use "GetFileList".

@items = GetFileList: Similar to "GetFileListItem", but returns all files in one go, which is very much faster for large number of items, and has no drawbacks when used for a small number of items.

Decoding files

$retval = $item->rename ($newname): Change the ondisk filename where the decoded file will be saved.

$retval = $item->decode_temp: Decode the file into a temporary location, use "$item->infile" to retrieve the temporary filename.

$retval = $item->remove_temp: Remove the temporarily decoded file again.

$retval = $item->decode ([$target_path]): Decode the file to its destination, or the given target path.

$retval = $item->info (callback-function)

Querying (and setting) item attributes

$state = $item->state

$mode = $item->mode ([newmode])

$uudet = $item->uudet

$size = $item->size

$filename = $item->filename ([newfilename})

$subfname = $item->subfname

$mimeid = $item->mimeid

$mimetype = $item->mimetype

$binfile = $item->binfile

Information about source parts

$parts = $item->parts

Return information about all parts (source files) used to decode the file as a list of hashrefs with the following structure:

 {
   partno   => <integer describing the part number, starting with 1>,
   # the following member sonly exist when they contain useful information
   sfname   => <local pathname of the file where this part is from>,
   filename => <the ondisk filename of the decoded file>,
   subfname => <used to cluster postings, possibly the posting filename>,
   subject  => <the subject of the posting/mail>,
   origin   => <the possible source (From) address>,
   mimetype => <the possible mimetype of the decoded file>,
   mimeid   => <the id part of the Content-Type>,
 }

Usually you are interested mostly the "sfname" and possibly the "partno" and "filename" members.

Functions below are not documented and not very well tested - feedback welcome

  QuickDecode
  EncodeMulti
  EncodePartial
  EncodeToStream
  EncodeToFile
  E_PrepSingle
  E_PrepPartial

EXTENSION FUNCTIONS

Functions found in this module but not documented in the uulib documentation:

$msg = straction ACT_xxx: Return a human readable string representing the given action code.

$msg = strerror RET_xxx: Return a human readable string representing the given error code.

$str = strencoding xxx_ENCODED: Return the name of the encoding type as a string.

$str = strmsglevel MSG_xxx: Returns the message level as a string.

SetFileNameCallback $cb

Sets (or queries) the FileNameCallback, which is called whenever the decoding library can't find a filename and wants to extract a filename from the subject line of a posting. The callback will be called with two arguments, the subject line and the current candidate for the filename. The latter argument can be "undef", which means that no filename could be found (and likely no one exists, so it is safe to also return "undef" in this case). If it doesn't return anything (not even "undef"!), then nothing happens, so this is a no-op callback:

   sub cb {
      return ();
   }

If it returns "undef", then this indicates that no filename could be found. In all other cases, the return value is taken to be the filename.

This is a slightly more useful callback:

  sub cb {
     return unless $_[1]; # skip "Re:"-plies et al.
     my ($subject, $filename) = @_;
     # if we find some *.rar, take it
     return $1 if $subject =~ /(\w+\.rar)/;
     # otherwise just pass what we have
     return ();
  }

LARGE EXAMPLE DECODER

The general workflow for decoding is like this:

1. Configure options with "SetOption" or "SetXXXCallback".

2. Load all source files with "LoadFile".

3. Optionally "Smerge".

4. Iterate over all "GetFileList" items (i.e. result files).

5. "CleanUp" to delete files and free items.

What follows is the file "example-decoder" from the distribution that illustrates the above worklfow in a non-trivial example.

   #!/usr/bin/perl
   # decode all the files in the directory uusrc/ and copy
   # the resulting files to uudst/
   use Convert::UUlib ':all';
   sub namefilter {
      my ($path) = @_;
      $path=~s/^.*[\/\\]//;
      $path
   }
   sub busycb {
      my ($action, $curfile, $partno, $numparts, $percent, $fsize) = @_;
      $_[0]=straction($action);
      print "busy_callback(", (join ",",@_), ")\n";
      0
   }
   SetOption OPT_RBUF, 128*1024;
   SetOption OPT_WBUF, 1024*1024;
   SetOption OPT_IGNMODE, 1;
   SetOption OPT_IGNMODE, 1;
   SetOption OPT_VERBOSE, 1;
   SetOption OPT_AUTOCHK, 0;
   # show the three ways you can set callback functions. I normally
   # prefer the one with the sub inplace.
   SetFNameFilter \&namefilter;
   SetBusyCallback "busycb", 333;
   SetMsgCallback sub {
      my ($msg, $level) = @_;
      print uc strmsglevel $_[1], ": $msg\n";
   };
   # the following non-trivial FileNameCallback takes care
   # of some subject lines not detected properly by uulib:
   SetFileNameCallback sub {
      return unless $_[1]; # skip "Re:"-plies et al.
      local $_ = $_[0];
      # the following rules are rather effective on some newsgroups,
      # like alt.binaries.games.anime, where non-mime, uuencoded data
      # is very common
      # if we find some *.rar, take it as the filename
      return $1 if /(\S{3,}\.(?:[rstuvwxyz]\d\d|rar))\s/i;
      # one common subject format
      return $1 if /- "(.{2,}?\..+?)" (?:yenc )?\(\d+\/\d+\)/i;
      # - filename.par (04/55)
      return $1 if /- "?(\S{3,}\.\S+?)"? (?:yenc )?\(\d+\/\d+\)/i;
      # - (xxx) No. 1 sayuri81.jpg 756565 bytes
      # - (20 files) No.17 Roseanne.jpg [2/2]
      return $1 if /No\.[ 0-9]+ (\S+\....) (?:\d+ bytes )?\[/;
      # try to detect some common forms of filenames
      return $1 if /([a-z0-9_\-+.]{3,}\.[a-z]{3,4}(?:.\d+))/i;
      # otherwise just pass what we have
      ()
   };
   # now read all files in the directory uusrc/*
   for (<uusrc/*>) {
      my ($retval, $count) = LoadFile ($_, $_, 1);
      print "file($_), status(", strerror $retval, ") parts($count)\n";
   }
   Smerge -1;
   SetOption OPT_SAVEPATH, "uudst/";
   # now wade through all files and their source parts
   for my $uu (GetFileList) {
      print "file ", $uu->filename, "\n";
      print " state ", $uu->state, "\n";
      print " mode ", $uu->mode, "\n";
      print " uudet ", strencoding $uu->uudet, "\n";
      print " size ", $uu->size, "\n";
      print " subfname ", $uu->subfname, "\n";
      print " mimeid ", $uu->mimeid, "\n";
      print " mimetype ", $uu->mimetype, "\n";
      # print additional info about all parts
      print " parts";
      for ($uu->parts) {
         for my $k (sort keys %$_) {
            print " $k=$_->{$k}";
         }
         print "\n";
      }
      $uu->remove_temp;
      if (my $err = $uu->decode) {
         print " ERROR ", strerror $err, "\n";
      } else {
         print " successfully saved as uudst/", $uu->filename, "\n";
      }
   }
   print "cleanup...\n";
   CleanUp;

PERLMULTICORE SUPPORT

This module supports the perlmulticore standard (see <http://perlmulticore.schmorp.de/> for more info) for the following functions - generally these are functions accessing the disk and/or using considerable CPU time:

   LoadFile
   $item->decode
   $item->decode_temp
   $item->remove_temp
   $item->info

The perl interpreter will be reacquired/released on every callback invocation, so for performance reasons, callbacks should be avoided if that is costly.

Future versions might enable multicore support for more functions.

BUGS AND LIMITATIONS

The original uulib library this module uses was written at a time where main memory of measured in megabytes and buffer overflows as a security thign didn't exist. While a lot of security fixes have been applied over the years (includign some defense in depth mechanism that can shield against a lot of as-of-yet undetected bugs), using this library for security purposes requires care.

Likewise, file sizes when the uulib library was written were tiny compared to today, so do not expect this library to handle files larger than 2GB.

Lastly, this module uses a very "C-like" interface, which means it doesn't protect you from invalid points as you might expect from "more perlish" modules - for example, accessing a file item object after callinbg "CleanUp" will likely result in crashes, memory corruption, or worse.

AUTHOR

Marc Lehmann <[email protected]>, the original uulib library was written by Frank Pilhofer <[email protected]>, and later heavily bugfixed by Marc Lehmann.