Read cover image from EPUB files

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

@@ -44,7 +44,7 @@ class HTMLTableRenderer(QObject):
             cutoff_height = int(self.height/factor)-3
             image = QImage(self.page.viewportSize(), QImage.Format_ARGB32)
             image.setDotsPerMeterX(self.dpi*(100/2.54))
-            image.setDotsPerMeterX(self.dpi*(100/2.54))
+            image.setDotsPerMeterY(self.dpi*(100/2.54))
             painter = QPainter(image)
             self.page.mainFrame().render(painter)
             painter.end()

src/calibre/ebooks/metadata/epub.py

@@ -5,15 +5,23 @@ __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
 
 '''Read meta information from epub files'''
 
-import sys, os
+import sys, os, time
 from cStringIO import StringIO
 from contextlib import closing
+from ctypes import c_long, byref
 
+from PyQt4.Qt import QWebPage, QUrl, QEventLoop, QSize, QByteArray, QBuffer, \
+                     SIGNAL, QPainter, QImage, QObject, QApplication
+
+from calibre.utils.PythonMagickWand import ImageMagick, NewMagickWand, MagickGetImageBlob, \
+                                           MagickReadImageBlob, MagickTrimImage, MagickSetFormat
 
 from calibre.utils.zipfile import ZipFile, BadZipfile, safe_replace
 from calibre.ebooks.BeautifulSoup import BeautifulStoneSoup
 from calibre.ebooks.metadata import get_parser, MetaInformation
 from calibre.ebooks.metadata.opf2 import OPF
+from calibre.ptempfile import TemporaryDirectory
+from calibre import CurrentDir
 
 class EPubException(Exception):
     pass
@@ -65,9 +73,9 @@ class OCFReader(OCF):
                 self.container = Container(f)
         except KeyError:
             raise EPubException("missing OCF container.xml file")
-
+        self.opf_path = self.container[OPF.MIMETYPE] 
         try:
-            with closing(self.open(self.container[OPF.MIMETYPE])) as f:
+            with closing(self.open(self.opf_path)) as f:
                 self.opf = OPF(f, self.root)
         except KeyError:
             raise EPubException("missing OPF package file")
@@ -97,12 +105,95 @@ class OCFDirReader(OCFReader):
     def open(self, path, *args, **kwargs):
         return open(os.path.join(self.root, path), *args, **kwargs)
 
+class CoverRenderer(QObject):
     
-def get_metadata(stream):
-    """ Return metadata as a L{MetaInfo} object """
-    return OCFZipReader(stream).opf
+    def __init__(self, url, size, loop):
+        QObject.__init__(self)
+        self.loop = loop
+        self.page = QWebPage()
+        QObject.connect(self.page, SIGNAL('loadFinished(bool)'), self.render_html)
+        self.image_data = None
+        self.rendered = False
+        self.page.mainFrame().load(url)
+        
+    def render_html(self, ok):
+        self.rendered = True
+        try:
+            if not ok:
+                return
+            self.page.setViewportSize(QSize(1280, 1024))
+            image = QImage(self.page.viewportSize(), QImage.Format_ARGB32)
+            image.setDotsPerMeterX(96*(100/2.54))
+            image.setDotsPerMeterY(96*(100/2.54))
+            painter = QPainter(image)
+            self.page.mainFrame().render(painter)
+            painter.end()
+            
+            ba = QByteArray()
+            buf = QBuffer(ba)
+            buf.open(QBuffer.WriteOnly)
+            image.save(buf, 'JPEG')
+            raw = str(ba.data())
+            with ImageMagick():
+                wand = NewMagickWand()
+                if not MagickReadImageBlob(wand, raw, len(raw)):
+                    raise ValueError('Failed to load cover image')
+                if not MagickTrimImage(wand, 10.):
+                    raise ValueError('Failed to process cover image')
+                x = c_long(0)
+                if not MagickSetFormat(wand, 'JPEG'):
+                    raise Exception()
+                dat = MagickGetImageBlob(wand, byref(x))
+                obuf = StringIO()
+                for i in xrange(x.value):
+                    obuf.write(chr(dat[i]))
+                self.image_data = obuf.getvalue()
+        finally:
+            self.loop.exit(0)
+        
+
+def get_cover(opf, opf_path, stream):
+    spine = list(opf.spine_items())
+    if not spine:
+        return
+    cpage = spine[0]
+    with TemporaryDirectory('_epub_meta') as tdir:
+        with CurrentDir(tdir):
+            stream.seek(0)
+            ZipFile(stream).extractall()
+            opf_path = opf_path.replace('/', os.sep)
+            cpage = os.path.join(tdir, os.path.dirname(opf_path), *cpage.split('/'))
+            if not os.path.exists(cpage):
+                return
+            if QApplication.instance() is None:
+                QApplication([])
+            url = QUrl.fromLocalFile(cpage)
+            loop = QEventLoop()
+            cr = CoverRenderer(url, os.stat(cpage).st_size, loop)
+            loop.exec_()
+            count = 0
+            while count < 50 and not cr.rendered:
+                time.sleep(0.1)
+                count += 1
+    return cr.image_data 
+    
+def get_metadata(stream, extract_cover=True):
+    """ Return metadata as a :class:`MetaInformation` object """
+    stream.seek(0)
+    reader = OCFZipReader(stream)
+    mi = MetaInformation(reader.opf)
+    if extract_cover:
+        try:
+            cdata = get_cover(reader.opf, reader.opf_path, stream)
+            if cdata is not None:
+                mi.cover_data = ('jpg', cdata)
+        except:
+            import traceback
+            traceback.print_exc()
+    return mi
 
 def set_metadata(stream, mi):
+    stream.seek(0)
     reader = OCFZipReader(stream, root=os.getcwdu())
     reader.opf.smart_update(mi)
     newopf = StringIO(reader.opf.render())
@@ -119,6 +210,8 @@ def option_parser():
                       help=_('The series index'))
     parser.add_option('--language', default=None,
                       help=_('The book language'))
+    parser.add_option('--get-cover', default=False, action='store_true',
+                      help=_('Extract the cover'))
     return parser
 
 def main(args=sys.argv):
@@ -127,8 +220,8 @@ def main(args=sys.argv):
     if len(args) != 2:
         parser.print_help()
         return 1
-    stream = open(args[1], 'r+b')
-    mi = MetaInformation(OCFZipReader(stream, root=os.getcwdu()).opf)
+    with open(args[1], 'r+b') as stream:
+        mi = get_metadata(stream, extract_cover=opts.get_cover)
         changed = False
         if opts.title:
             mi.title = opts.title
@@ -153,11 +246,15 @@ def main(args=sys.argv):
             changed = True
         
         if changed:
-        stream.seek(0)
             set_metadata(stream, mi)
-    stream.seek(0)
-    print unicode(MetaInformation(OCFZipReader(stream, root=os.getcwdu()).opf))
-    stream.close()
+        print unicode(get_metadata(stream, extract_cover=False))
+        
+    if mi.cover_data[1] is not None:
+        cpath = os.path.splitext(os.path.basename(args[1]))[0] + '_cover.jpg'
+        with open(cpath, 'wb') as f:
+            f.write(mi.cover_data[1])
+            print 'Cover saved to', f.name
+    
     return 0
 
 if __name__ == '__main__':

src/calibre/ebooks/metadata/opf2.py

@@ -34,6 +34,7 @@ class Resource(object):
     '''
     
     def __init__(self, href_or_path, basedir=os.getcwd(), is_path=True):
+        self.orig = href_or_path
         self._href = None
         self._basedir = basedir
         self.path = None
@@ -522,6 +523,13 @@ class OPF(object):
     def iterspine(self):
         return self.spine_path(self.root)
     
+    def spine_items(self):
+        for item in self.iterspine():
+            idref = item.get('idref', '')
+            for x in self.itermanifest():
+                if x.get('id', None) == idref:
+                    yield x.get('href', '')
+    
     def create_spine_item(self, idref):
         ans = etree.Element('{%s}itemref'%self.NAMESPACES['opf'], idref=idref)
         ans.tail = '\n\t\t'