Implement #1021 (CBR and CBZ cover detection)

src/calibre/ebooks/metadata/meta.py

@@ -20,8 +20,8 @@ from calibre.ebooks.metadata.opf  import OPFReader
 from calibre.ebooks.metadata.rtf  import set_metadata as set_rtf_metadata
 from calibre.ebooks.lrf.meta      import set_metadata as set_lrf_metadata
 from calibre.ebooks.metadata.epub import set_metadata as set_epub_metadata
-from calibre.libunrar import extract_first as rar_extract_first
+from calibre.libunrar import extract_member as rar_extract_first
-from calibre.libunzip import extract_first as zip_extract_first
+from calibre.libunzip import extract_member as zip_extract_first
 
 from calibre.ebooks.metadata import MetaInformation
 from calibre.ptempfile import TemporaryDirectory
@@ -111,18 +111,13 @@ def get_metadata(stream, stream_type='lrf', use_libprs_metadata=False):
     return base
 
 def get_comic_cover(stream, type):
-    with TemporaryDirectory('_comic_cover') as tdir:
+    extract_first = zip_extract_first if type == 'zip' else rar_extract_first
-        extract_first = zip_extract_first if type == 'zip' else rar_extract_first
+    ret = extract_first(stream)
-        extract_first(stream, tdir)
+    if ret is not None:
-        files = os.listdir(tdir)
+        path, data = ret
-        print tdir, files
+        ext = os.path.splitext(path)[1][1:]
-        if files:
+        return (ext.lower(), data)
-            path = os.path.join(tdir, files[0])
+        
-            ext = os.path.splitext(path)[1].lower()
-            if ext:
-                ext = ext[1:]
-                return (ext, open(path, 'rb').read())
-            
         
 
 def set_metadata(stream, mi, stream_type='lrf'):

src/calibre/libunrar.py

@@ -6,13 +6,14 @@ This module provides a thin ctypes based wrapper around libunrar.
 
 See  ftp://ftp.rarlabs.com/rar/unrarsrc-3.7.5.tar.gz
 """
-import os, ctypes, sys
+import os, ctypes, sys, re
 from ctypes import Structure, c_char_p, c_uint, c_void_p, POINTER, \
                     byref, c_wchar_p, c_int, c_char, c_wchar
 from tempfile import NamedTemporaryFile
 from StringIO import StringIO
 
 from calibre import iswindows, load_library, CurrentDir
+from calibre.ptempfile import TemporaryDirectory
 
 _librar_name = 'libunrar'
 cdll = ctypes.cdll
@@ -185,28 +186,31 @@ def extract(path, dir):
         os.chdir(cwd)
         _libunrar.RARCloseArchive(arc_data)
         
-def extract_first(path, dir):
+def extract_member(path, match=re.compile(r'\.(jpg|jpeg|gif|png)\s*$', re.I)):
     if hasattr(path, 'read'):
         data = path.read()
         f = NamedTemporaryFile(suffix='.rar')
         f.write(data)
         f.flush()
         path = f.name
-    if not os.path.isdir( dir ):
+    with TemporaryDirectory('_libunrar') as dir:
-        os.makedirs(dir)
+        with CurrentDir(dir):
-    with CurrentDir(dir):
+            open_archive_data = RAROpenArchiveDataEx(ArcName=path, OpenMode=RAR_OM_EXTRACT, CmtBuf=None)
-        open_archive_data = RAROpenArchiveDataEx(ArcName=path, OpenMode=RAR_OM_EXTRACT, CmtBuf=None)
+            arc_data = _libunrar.RAROpenArchiveEx(byref(open_archive_data))
-        arc_data = _libunrar.RAROpenArchiveEx(byref(open_archive_data))
+            try:
-        try:
+                if open_archive_data.OpenResult != 0:
-            if open_archive_data.OpenResult != 0:
+                    raise UnRARException(_interpret_open_error(open_archive_data.OpenResult, path))
-                raise UnRARException(_interpret_open_error(open_archive_data.OpenResult, path))
+                header_data = RARHeaderDataEx(CmtBuf=None)
-            header_data = RARHeaderDataEx(CmtBuf=None)
+                while True:
-            if _libunrar.RARReadHeaderEx(arc_data, byref(header_data)) != 0:
+                    if _libunrar.RARReadHeaderEx(arc_data, byref(header_data)) != 0:
-                raise UnRARException('%s has no files'%path)
+                        raise UnRARException('%s has no files'%path)
-            PFCode = _libunrar.RARProcessFileW(arc_data, RAR_EXTRACT, None, None)
+                    PFCode = _libunrar.RARProcessFileW(arc_data, RAR_EXTRACT, None, None)
-            if PFCode != 0:
+                    if PFCode != 0:
-                raise UnRARException(_interpret_process_file_error(PFCode))
+                        raise UnRARException(_interpret_process_file_error(PFCode))
-        finally:
+                    if match.search(header_data.FileNameW):
-            _libunrar.RARCloseArchive(arc_data)
+                        return header_data.FileNameW.replace('/', os.sep), \
-        
+                                open(os.path.join(dir, *header_data.FileNameW.split('/')), 'rb').read()
+            finally:
+                _libunrar.RARCloseArchive(arc_data)
+            
     

src/calibre/libunzip.py

@@ -3,7 +3,7 @@ __license__   = 'GPL v3'
 __copyright__ = '2008, Kovid Goyal kovid@kovidgoyal.net'
 __docformat__ = 'restructuredtext en'
 
-import os
+import os, re
 from calibre.utils import zipfile
 
 def update(pathtozip, patterns, filepaths, names, compression=zipfile.ZIP_DEFLATED, verbose=True):
@@ -43,10 +43,9 @@ def extract(filename, dir):
     zf = zipfile.ZipFile( filename )
     zf.extractall(dir)
     
-def extract_first(filename, dir):
+def extract_member(filename, match=re.compile(r'\.(jpg|jpeg|gif|png)\s*$', re.I)):
     zf = zipfile.ZipFile(filename)
     names = zf.namelist()
-    if not names:
+    for name in names:
-        raise ValueError('%s has no files'%filename)
+        if match.search(name):
-    bytes = zf.read(names[0])
+            return name, zf.read(name)
-    open(os.path.join(dir, names[0]), 'wb').write(bytes)