comix 4.0.4-3 / usr / bin / comicthumb

#!/usr/bin/env python

"""comicthumb - Thumbnailer for comic book archives, bundled with Comix.

comicthumb is dependent on the Python Imaging Library (PIL).

comicthumb was originally written by Christoph Wolk, this version was
re-written from scratch for Comix 4 by Pontus Ekberg. 

Supported formats: ZIP, RAR and tar (.cbz, .cbr, .cbt)

Usage: comicthumb INFILE OUTFILE [SIZE]
"""

import os
import sys
import gc
import re
import cStringIO
import zipfile
import tarfile
import subprocess

try:
    import Image
except ImportError:
    print '! Could not import the Image module (PIL).'
    print __doc__
    sys.exit(1)

ZIP, RAR, TAR, GZIP, BZIP2 = range(5)


class Process:
    
    """The subprocess and popen2 modules in Python are broken (see issue
    #1336). The problem (i.e. complete crash) they can cause happen fairly
    often (once is too often) in Comix when calling "rar" or "unrar" to
    extract specific files from archives. We roll our own very simple
    process spawning module here instead.
    """
    # TODO: I can no longer reproduce the issue. Check if this version of
    # process.py still solves it.

    def __init__(self, args):
        """Setup a Process where <args> is a sequence of arguments that defines
        the process, e.g. ['ls', '-a'].
        """
        self._args = args
        self._proc = None
    
    def _exec(self):
        """Spawns the process, and returns its stdout.
        (NOTE: separate function to make python2.4 exception syntax happy)
        """
        try:
            self._proc = subprocess.Popen(self._args, stdout=subprocess.PIPE)
            return self._proc.stdout
        except Exception:
            return None

    def spawn(self):
        """Spawn the process defined by the args in __init__(). Return a
        file-like object linked to the spawned process' stdout.
        """
        try:
            gc.disable() # Avoid Python issue #1336!
            return self._exec()
        finally:
            gc.enable()

    def wait(self):
        """Wait for the process to terminate."""
        if self._proc is None:
            raise Exception('Process not spawned.')
        return self._proc.wait()


class Extractor:

    """Extractor is a class for extracting different archive formats.
    This is a much simplified version of the Extractor class from Comix.
    """

    def __init__(self, src):
        """Setup the extractor with archive <src>."""
        self._src = src
        self._type = archive_mime_type(src)
        self._files = []
        
        if self._type == ZIP:
            self._zfile = zipfile.ZipFile(src, 'r')
            self._files = self._zfile.namelist()
        elif self._type in [TAR, GZIP, BZIP2]:
            self._tfile = tarfile.open(src, 'r')
            self._files = self._tfile.getnames()
        elif self._type == RAR:
            self._rar = None
            for command in ('unrar', 'rar'):
                if Process([command]).spawn() is not None:
                    self._rar = command
            if self._rar == None:
                print '! Could not find the "rar" or "unrar" executable.'
                sys.exit(1)
            proc = Process([self._rar, 'vb', src])
            fobj = proc.spawn()
            self._files = fobj.readlines()
            proc.wait()
            self._files = [name.rstrip('\n') for name in self._files]
    
    def get_files(self):
        """Return a list of the files in the archive."""
        return self._files
        
    def extract(self, chosen): 
        """Extract the file <chosen> and return it as a cStringIO.StringIO
        object. The <chosen> file must be one of the files in the list
        returned by the get_files() method.
        """
        if self._type == ZIP:
            return cStringIO.StringIO(self._zfile.read(chosen))
        elif self._type in [TAR, GZIP, BZIP2]:
            return cStringIO.StringIO(self._tfile.extractfile(chosen).read())
        elif self._type == RAR:
            proc = Process([self._rar, 'p', '-inul', '-p-', '--',
                self._src, chosen])
            fobj = proc.spawn()
            return cStringIO.StringIO(fobj.read())


def archive_mime_type(path):
    """Return the archive type of <path> or None for non-archives."""
    try:
        if os.path.isfile(path):
            if not os.access(path, os.R_OK):
                return None
            if zipfile.is_zipfile(path):
                return ZIP
            fd = open(path, 'rb')
            magic = fd.read(4)
            fd.close()
            if tarfile.is_tarfile(path) and os.path.getsize(path) > 0:
                if magic.startswith('BZh'):
                    return BZIP2
                if magic.startswith('\037\213'):
                    return GZIP
                return TAR
            if magic == 'Rar!':
                return RAR
    except Exception:
        print '! Error while reading', path
    return None


def guess_cover(files):
    """Return the filename within <files> that is the most likely to be
    the cover of an archive.
    """
    alphanumeric_sort(files)
    ext_re = re.compile(r'\.(jpg|jpeg|png|gif|tif|tiff)\s*$', re.I)
    front_re = re.compile('(cover|front)', re.I)
    images = filter(ext_re.search, files)
    candidates = filter(front_re.search, images)
    candidates = [c for c in candidates if not 'back' in c.lower()]
    if candidates:
        return candidates[0]
    if images:
        return images[0]
    return None

def alphanumeric_sort(filenames):
    """Do an in-place alphanumeric sort of the strings in <filenames>,
    such that for an example "1.jpg", "2.jpg", "10.jpg" is a sorted
    ordering.
    """
    def _format_substring(s):
        if s.isdigit():
            return int(s)
        return s.lower()

    rec = re.compile("\d+|\D+")
    filenames.sort(key=lambda s: map(_format_substring, rec.findall(s)))


if __name__ == '__main__':
    try:
        in_path = sys.argv[1]
        out_path = sys.argv[2]
        if len(sys.argv) == 4:
            size = int(sys.argv[3])
        else:
            size = 128
    except:
        print __doc__
        sys.exit(1)
    extractor = Extractor(in_path)
    files = extractor.get_files()
    chosen = guess_cover(files)
    fd = extractor.extract(chosen)
    im = Image.open(fd)
    if im.size[0] > im.size[1]:
        x = size
        y = size * im.size[1] / im.size[0]
    else:
        x = size * im.size[0] / im.size[1]
        y = size
    x = max(1, x)
    y = max(1, y)
    im.thumbnail((x, y), Image.ANTIALIAS)
    im = im.convert('RGB')
    im.save(out_path, 'PNG')
    sys.exit(0)