Fix handling of reflowable covers in EPUB input. Also try to extract svg embedded raster covers. Misc. minor fixes

src/calibre/__init__.py

@@ -361,6 +361,8 @@ def entity_to_unicode(match, exceptions=[], encoding='cp1252'):
         return '&'+ent+';'
     if ent == 'apos':
         return "'"
+    if ent == 'hellips':
+        ent = 'hellip'
     if ent.startswith(u'#x'):
         num = int(ent[2:], 16)
         if encoding is None or num > 255:

src/calibre/ebooks/__init__.py

@@ -57,6 +57,35 @@ class HTMLRenderer(object):
             self.loop.exit(0)
 
 
+def extract_cover_from_embedded_svg(html, base, log):
+    from lxml import etree
+    from calibre.ebooks.oeb.base import XPath, SVG, XLINK
+    root = etree.fromstring(html)
+
+    svg = XPath('//svg:svg')(root)
+    if len(svg) == 1 and len(svg[0]) == 1 and svg[0][0].tag == SVG('image'):
+        image = svg[0][0]
+        href = image.get(XLINK('href'), None)
+        path = os.path.join(base, *href.split('/'))
+        if href and os.access(path, os.R_OK):
+            return open(path, 'rb').read()
+
+def render_html_svg_workaround(path_to_html, log, width=590, height=750):
+    from calibre.ebooks.oeb.base import SVG_NS
+    raw = open(path_to_html, 'rb').read()
+    data = None
+    if SVG_NS in raw:
+        try:
+            data = extract_cover_from_embedded_svg(raw,
+                   os.path.dirname(path_to_html), log)
+        except:
+            pass
+    if data is None:
+        renderer = render_html(path_to_html, width, height)
+        data = getattr(renderer, 'data', None)
+    return data
+
+
 def render_html(path_to_html, width=590, height=750):
     from PyQt4.QtWebKit import QWebPage
     from PyQt4.Qt import QEventLoop, QPalette, Qt, SIGNAL, QUrl, QSize

src/calibre/ebooks/epub/input.py

@@ -54,7 +54,7 @@ class EPUBInput(InputFormatPlugin):
         return False
 
     @classmethod
-    def rationalize_cover(self, opf):
+    def rationalize_cover(self, opf, log):
         guide_cover, guide_elem = None, None
         for guide_elem in opf.iterguide():
             if guide_elem.get('type', '').lower() == 'cover':
@@ -65,28 +65,37 @@ class EPUBInput(InputFormatPlugin):
         spine = list(opf.iterspine())
         if not spine:
             return
+        # Check if the cover specified in the guide is also
+        # the first element in spine
         idref = spine[0].get('idref', '')
         manifest = list(opf.itermanifest())
         if not manifest:
             return
-        if manifest[0].get('id', False) != idref:
+        elem = [x for x in manifest if x.get('id', '') == idref]
+        if not elem or elem[0].get('href', None) != guide_cover:
             return
+        log('Found HTML cover', guide_cover)
+
+        # Remove from spine as covers must be treated
+        # specially
         spine[0].getparent().remove(spine[0])
         guide_elem.set('href', 'calibre_raster_cover.jpg')
+        from calibre.ebooks.oeb.base import OPF
+        t = etree.SubElement(elem[0].getparent(), OPF('item'),
+        href=guide_elem.get('href'), id='calibre_raster_cover')
+        t.set('media-type', 'image/jpeg')
         for elem in list(opf.iterguide()):
             if elem.get('type', '').lower() == 'titlepage':
                 elem.getparent().remove(elem)
-        from calibre.ebooks.oeb.base import OPF
         t = etree.SubElement(guide_elem.getparent(), OPF('reference'))
         t.set('type', 'titlepage')
         t.set('href', guide_cover)
         t.set('title', 'Title Page')
-        from calibre.ebooks import render_html
+        from calibre.ebooks import render_html_svg_workaround
-        renderer = render_html(guide_cover)
+        renderer = render_html_svg_workaround(guide_cover, log)
         if renderer is not None:
             open('calibre_raster_cover.jpg', 'wb').write(
-                renderer.data)
+                renderer)
-
 
     def convert(self, stream, options, file_ext, log, accelerators):
         from calibre.utils.zipfile import ZipFile
@@ -121,7 +130,7 @@ class EPUBInput(InputFormatPlugin):
             for elem in opf.iterguide():
                 elem.set('href', delta+elem.get('href'))
 
-        self.rationalize_cover(opf)
+        self.rationalize_cover(opf, log)
 
         with open('content.opf', 'wb') as nopf:
             nopf.write(opf.render())

src/calibre/ebooks/metadata/epub.py

@@ -5,14 +5,10 @@ __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
 
 '''Read meta information from epub files'''
 
-import os, time
+import os
 from cStringIO import StringIO
 from contextlib import closing
 
-from PyQt4.Qt import QUrl, QEventLoop, QSize, QByteArray, QBuffer, \
-                     SIGNAL, QPainter, QImage, QObject, QApplication, Qt, QPalette
-from PyQt4.QtWebKit import QWebPage
-
 from calibre.utils.zipfile import ZipFile, BadZipfile, safe_replace
 from calibre.ebooks.BeautifulSoup import BeautifulStoneSoup
 from calibre.ebooks.metadata import MetaInformation
@@ -102,64 +98,9 @@ class OCFDirReader(OCFReader):
     def open(self, path, *args, **kwargs):
         return open(os.path.join(self.root, path), *args, **kwargs)
 
-class CoverRenderer(QObject):
-    WIDTH  = 600
-    HEIGHT = 800
-
-    def __init__(self, path):
-        if QApplication.instance() is None:
-            QApplication([])
-        QObject.__init__(self)
-        self.loop = QEventLoop()
-        self.page = QWebPage()
-        pal = self.page.palette()
-        pal.setBrush(QPalette.Background, Qt.white)
-        self.page.setPalette(pal)
-        self.page.setViewportSize(QSize(self.WIDTH, self.HEIGHT))
-        self.page.mainFrame().setScrollBarPolicy(Qt.Vertical, Qt.ScrollBarAlwaysOff)
-        self.page.mainFrame().setScrollBarPolicy(Qt.Horizontal, Qt.ScrollBarAlwaysOff)
-        QObject.connect(self.page, SIGNAL('loadFinished(bool)'), self.render_html)
-        self._image_data = None
-        self.rendered = False
-        url = QUrl.fromLocalFile(os.path.normpath(path))
-        self.page.mainFrame().load(url)
-
-    def render_html(self, ok):
-        try:
-            if not ok:
-                self.rendered = True
-                return
-            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')
-            self._image_data = str(ba.data())
-        finally:
-            self.loop.exit(0)
-        self.rendered = True
-
-    def image_data():
-        def fget(self):
-            if not self.rendered:
-                self.loop.exec_()
-                count = 0
-                while count < 50 and not self.rendered:
-                    time.sleep(0.1)
-                    count += 1
-            return self._image_data
-        return property(fget=fget)
-    image_data = image_data()
-
-
 def get_cover(opf, opf_path, stream):
-    from calibre.gui2 import is_ok_to_use_qt
+    from calibre.ebooks import render_html_svg_workaround
-    if not is_ok_to_use_qt(): return None
+    from calibre.utils.logging import default_log
     spine = list(opf.spine_items())
     if not spine:
         return
@@ -172,8 +113,7 @@ def get_cover(opf, opf_path, stream):
             cpage = os.path.join(tdir, os.path.dirname(opf_path), cpage)
             if not os.path.exists(cpage):
                 return
-            cr = CoverRenderer(cpage)
+            return render_html_svg_workaround(cpage, default_log)
-            return cr.image_data
 
 def get_metadata(stream, extract_cover=True):
     """ Return metadata as a :class:`MetaInformation` object """

src/calibre/ebooks/mobi/writer.py

@@ -1556,7 +1556,8 @@ class MobiWriter(object):
         else:
             raise NotImplementedError("missing date or timestamp needed for mobi_periodical")
 
-        if oeb.metadata.cover:
+        if oeb.metadata.cover and \
+                unicode(oeb.metadata.cover[0]) in oeb.manifest.ids:
             id = unicode(oeb.metadata.cover[0])
             item = oeb.manifest.ids[id]
             href = item.href

src/calibre/ebooks/oeb/reader.py

@@ -27,7 +27,6 @@ from calibre.ebooks.oeb.base import namespace, barename, XPath, xpath, \
                                     OEBError, OEBBook, DirContainer
 from calibre.ebooks.oeb.writer import OEBWriter
 from calibre.ebooks.oeb.entitydefs import ENTITYDEFS
-from calibre.ebooks.metadata.epub import CoverRenderer
 from calibre.startup import get_lang
 from calibre.ptempfile import TemporaryDirectory
 from calibre.constants import __appname__, __version__
@@ -346,6 +345,8 @@ class OEBReader(object):
             if descriptionElement:
                 description = etree.tostring(descriptionElement[0],
                 method='text', encoding=unicode).strip()
+                if not description:
+                    description = None
             else :
                 description = None
 
@@ -525,12 +526,14 @@ class OEBReader(object):
         return
 
     def _cover_from_html(self, hcover):
+        from calibre.ebooks import render_html_svg_workaround
         with TemporaryDirectory('_html_cover') as tdir:
             writer = OEBWriter()
             writer(self.oeb, tdir)
             path = os.path.join(tdir, urlunquote(hcover.href))
-            renderer = CoverRenderer(path)
+            data = render_html_svg_workaround(path, self.logger)
-            data = renderer.image_data
+            if not data:
+                data = ''
         id, href = self.oeb.manifest.generate('cover', 'cover.jpeg')
         item = self.oeb.manifest.add(id, href, JPEG_MIME, data=data)
         return item

src/calibre/utils/logging.py

@@ -102,3 +102,5 @@ class Log(object):
 
     def __call__(self, *args, **kwargs):
         self.prints(INFO, *args, **kwargs)
+
+default_log = Log()

src/calibre/web/feeds/recipes/recipe_publico.py

@@ -17,7 +17,7 @@ class Publico(BasicNewsRecipe):
     max_articles_per_feed = 30
     encoding='utf-8'
     no_stylesheets = True
-    language = _('Portuguese')
+    language = _('Portugese')
     preprocess_regexps = [(re.compile(u"\uFFFD", re.DOTALL|re.IGNORECASE),  lambda match: ''),]
 
     feeds          = [