Fix regression in epub2lrf when converting EPUB files that specify a HTML page as a cover

src/calibre/ebooks/lrf/epub/convert_from.py

@@ -2,14 +2,14 @@ __license__   = 'GPL v3'
 __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
 
 import os, sys, shutil, logging
-from tempfile import mkdtemp
 from calibre.ebooks.lrf import option_parser as lrf_option_parser
 from calibre.ebooks import ConversionError, DRMError
 from calibre.ebooks.lrf.html.convert_from import process_file as html_process_file
 from calibre.ebooks.metadata.opf import OPF
 from calibre.ebooks.metadata.epub import OCFDirReader
 from calibre.utils.zipfile import ZipFile
-from calibre import __appname__, setup_cli_handlers
+from calibre import setup_cli_handlers
+from calibre.ptempfile import PersistentTemporaryDirectory
 
 
 def option_parser():
@@ -22,17 +22,16 @@ _('''Usage: %prog [options] mybook.epub
 
 def generate_html(pathtoepub, logger):
     if not os.access(pathtoepub, os.R_OK):
-        raise ConversionError, 'Cannot read from ' + pathtoepub
-    tdir = mkdtemp(prefix=__appname__+'_')
-    os.rmdir(tdir)
+        raise ConversionError('Cannot read from ' + pathtoepub)
+    tdir = PersistentTemporaryDirectory('_epub2lrf')
+    #os.rmdir(tdir)
     try:
         ZipFile(pathtoepub).extractall(tdir)
-        if os.path.exists(os.path.join(tdir, 'META-INF', 'encryption.xml')):
-            raise DRMError(os.path.basename(pathtoepub))
     except:
-        if os.path.exists(tdir) and os.path.isdir(tdir):
-            shutil.rmtree(tdir)
         raise ConversionError, '.epub extraction failed'
+    if os.path.exists(os.path.join(tdir, 'META-INF', 'encryption.xml')):
+            raise DRMError(os.path.basename(pathtoepub))
+    
     return tdir
 
 def process_file(path, options, logger=None):

src/calibre/ebooks/lrf/html/convert_from.py

@@ -1982,17 +1982,7 @@ def try_opf(path, options, logger):
                         PILImage.open(cover)
                         options.cover = cover
                     except:
-                        for prefix in opf.possible_cover_prefixes():
-                            if options.cover:
-                                break
-                            for suffix in ['.jpg', '.jpeg', '.gif', '.png', '.bmp']:
-                                cpath = os.path.join(os.path.dirname(path), prefix+suffix)
-                                try:
-                                    PILImage.open(cpath)
-                                    options.cover = cpath
-                                    break
-                                except:
-                                    continue
+                        pass
             if not getattr(options, 'cover', None) and orig_cover is not None:
                 options.cover = orig_cover
         if getattr(opf, 'spine', False):

src/calibre/ebooks/metadata/opf2.py

@@ -418,7 +418,8 @@ class OPF(object):
     tags_path       = XPath('descendant::*[re:match(name(), "subject", "i")]')
     isbn_path       = XPath('descendant::*[re:match(name(), "identifier", "i") and '+
                             '(re:match(@scheme, "isbn", "i") or re:match(@opf:scheme, "isbn", "i"))]')
-    application_id_path= XPath('descendant::*[re:match(name(), "identifier", "i") and '+
+    identifier_path = XPath('descendant::*[re:match(name(), "identifier", "i")]')
+    application_id_path = XPath('descendant::*[re:match(name(), "identifier", "i") and '+
                             '(re:match(@opf:scheme, "calibre|libprs500", "i") or re:match(@scheme, "calibre|libprs500", "i"))]')
     manifest_path   = XPath('descendant::*[re:match(name(), "manifest", "i")]/*[re:match(name(), "item", "i")]')
     manifest_ppath  = XPath('descendant::*[re:match(name(), "manifest", "i")]') 
@@ -719,6 +720,27 @@ class OPF(object):
         return property(fget=fget, fset=fset)
     
     
+    def guess_cover(self):
+        '''
+        Try to guess a cover. Needed for some old/badly formed OPF files.
+        '''
+        if self.base_dir and os.path.exists(self.base_dir):
+            for item in self.identifier_path(self.metadata):
+                scheme = None
+                for key in item.attrib.keys():
+                    if key.endswith('scheme'):
+                        scheme = item.get(key)
+                        break
+                if scheme is None:
+                    continue
+                if item.text:
+                    prefix = item.text.replace('-', '')
+                    for suffix in ['.jpg', '.jpeg', '.gif', '.png', '.bmp']:
+                        cpath = os.access(os.path.join(self.base_dir, prefix+suffix), os.R_OK)
+                        if os.access(os.path.join(self.base_dir, prefix+suffix), os.R_OK):
+                            return cpath
+            
+    
     @apply
     def cover():
         
@@ -728,6 +750,10 @@ class OPF(object):
                     for item in self.guide:
                         if item.type.lower() == t:
                             return item.path
+            try:
+                return self.guess_cover()
+            except:
+                pass
                         
         def fset(self, path):
             if self.guide is not None: