Fix #1712 (MOBI error dump from Mac) and make detection if title in MOBI files more robust

src/calibre/ebooks/mobi/reader.py

@@ -28,7 +28,7 @@ from calibre import sanitize_file_name
 
 class EXTHHeader(object):
     
-    def __init__(self, raw, codec):
+    def __init__(self, raw, codec, title):
         self.doctype = raw[:4]
         self.length, self.num_items = struct.unpack('>LL', raw[4:12])
         raw = raw[12:]
@@ -52,16 +52,8 @@ class EXTHHeader(object):
                 self.thumbnail_offset, = struct.unpack('>L', content)
             #else:
             #    print 'unknown record', id, repr(content)
-        title = re.search(r'\0+([^\0]+)\0+', raw[pos:])
         if title:
-            title = title.group(1).decode(codec, 'replace')
-            if len(title) > 2:
             self.mi.title = title
-            else:
-                title = re.search(r'\0+([^\0]+)\0+', ''.join(reversed(raw[pos:])))
-                if title:
-                    self.mi.title = ''.join(reversed(title.group(1).decode(codec, 'replace')))
-            
     
     def process_metadata(self, id, content, codec):
         if id == 100:
@@ -121,6 +113,9 @@ class BookHeader(object):
             if self.compression_type == 'DH':
                 self.huff_offset, self.huff_number = struct.unpack('>LL', raw[0x70:0x78]) 
             
+            toff, tlen = struct.unpack('>II', raw[0x54:0x5c])
+            tend = toff + tlen
+            self.title = raw[toff:tend] if tend < len(raw) else _('Unknown')
             langcode  = struct.unpack('!L', raw[0x5C:0x60])[0]
             langid    = langcode & 0xFF
             sublangid = (langcode >> 10) & 0xFF
@@ -131,7 +126,7 @@ class BookHeader(object):
             self.exth_flag, = struct.unpack('>L', raw[0x80:0x84])
             self.exth = None
             if self.exth_flag & 0x40:
-                self.exth = EXTHHeader(raw[16+self.length:], self.codec)
+                self.exth = EXTHHeader(raw[16+self.length:], self.codec, self.title)
                 self.exth.mi.uid = self.unique_id
                 self.exth.mi.language = self.language
             

src/calibre/ebooks/mobi/writer.py

@@ -23,6 +23,7 @@ from PIL import Image
 from calibre.ebooks.oeb.base import XML_NS, XHTML, XHTML_NS, OEB_DOCS, \
     OEB_RASTER_IMAGES
 from calibre.ebooks.oeb.base import xpath, barename, namespace, prefixname
+from calibre.ebooks.oeb.base import urlnormalize
 from calibre.ebooks.oeb.base import Logger, OEBBook
 from calibre.ebooks.oeb.profile import Context
 from calibre.ebooks.oeb.transforms.flatcss import CSSFlattener
@@ -178,7 +179,7 @@ class Serializer(object):
 
     def serialize_href(self, href, base=None):
         hrefs = self.oeb.manifest.hrefs
-        path, frag = urldefrag(href)
+        path, frag = urldefrag(urlnormalize(href))
         if path and base:
             path = base.abshref(path)
         if path and path not in hrefs:

src/calibre/ebooks/oeb/base.py

@@ -447,7 +447,7 @@ class Manifest(object):
             return cmp(skey, okey)
         
         def relhref(self, href):
-            if '/' not in self.href:
+            if '/' not in self.href or ':' in href:
                 return href
             base = os.path.dirname(self.href).split('/')
             target, frag = urldefrag(href)
@@ -463,7 +463,7 @@ class Manifest(object):
             return relhref
 
         def abshref(self, href):
-            if '/' not in self.href:
+            if '/' not in self.href or ':' in href:
                 return href
             dirname = os.path.dirname(self.href)
             href = os.path.join(dirname, href)
@@ -796,6 +796,10 @@ class TOC(object):
 
     
 class OEBBook(object):
+    COVER_SVG_XP = XPath('h:body//svg:svg[position() = 1]')
+    COVER_OBJECT_XP = XPath('h:body//h:object[@data][position() = 1]')
+    COVER_IMG_XP = XPath('h:body//h:img[@src][position() = 1]')
+
     def __init__(self, opfpath=None, container=None, encoding=None,
                  logger=FauxLogger()):
         if opfpath and not container:
@@ -971,7 +975,7 @@ class OEBBook(object):
         ncx = item.data
         self.manifest.remove(item)
         title = xpath(ncx, 'ncx:docTitle/ncx:text/text()')
-        title = title[0].strip() if title else unicode(self.metadata.title)
+        title = title[0].strip() if title else unicode(self.metadata.title[0])
         self.toc = toc = TOC(title)
         navmaps = xpath(ncx, 'ncx:navMap')
         for navmap in navmaps:
@@ -1051,40 +1055,54 @@ class OEBBook(object):
         if self._toc_from_html(opf): return
         self._toc_from_spine(opf)
 
-    def _ensure_cover_image(self):
+    def _locate_cover_image(self):
-        cover = None
+        if self.metadata.cover:
+            id = str(self.metadata.cover[0])
+            item = self.manifest.ids.get(id, None)
+            if item is not None:
+                return item
         hcover = self.spine[0]
         if 'cover' in self.guide:
             href = self.guide['cover'].href
             item = self.manifest.hrefs[href]
             media_type = item.media_type
-            if media_type in OEB_RASTER_IMAGES:
+            if media_type in OEB_IMAGES:
-                cover = item
+                return item
             elif media_type in OEB_DOCS:
                 hcover = item
         html = hcover.data
-        if cover is not None:
+        if MS_COVER_TYPE in self.guide:
-            pass
-        elif self.metadata.cover:
-            id = str(self.metadata.cover[0])
-            cover = self.manifest.ids[id]
-        elif MS_COVER_TYPE in self.guide:
             href = self.guide[MS_COVER_TYPE].href
-            cover = self.manifest.hrefs[href]
+            item = self.manifest.hrefs.get(href, None)
-        elif xpath(html, '//h:img[position()=1]'):
+            if item is not None and item.media_type in OEB_IMAGES:
-            img = xpath(html, '//h:img[position()=1]')[0]
+                return item
-            href = hcover.abshref(img.get('src'))
+        if self.COVER_SVG_XP(html):
-            cover = self.manifest.hrefs[href]
+            svg = copy.deepcopy(self.COVER_SVG_XP(html)[0])
-        elif xpath(html, '//h:object[position()=1]'):
-            object = xpath(html, '//h:object[position()=1]')[0]
-            href = hcover.abshref(object.get('data'))
-            cover = self.manifest.hrefs[href]
-        elif xpath(html, '//svg:svg[position()=1]'):
-            svg = copy.deepcopy(xpath(html, '//svg:svg[position()=1]')[0])
             href = os.path.splitext(hcover.href)[0] + '.svg'
             id, href = self.manifest.generate(hcover.id, href)
-            cover = self.manifest.add(id, href, SVG_MIME, data=svg)
+            item = self.manifest.add(id, href, SVG_MIME, data=svg)
-        if cover and not self.metadata.cover:
+            return item
+        if self.COVER_OBJECT_XP(html):
+            object = self.COVER_OBJECT_XP(html)[0]
+            href = hcover.abshref(object.get('data'))
+            item = self.manifest.hrefs.get(href, None)
+            if item is not None and item.media_type in OEB_IMAGES:
+                return item
+        if self.COVER_IMG_XP(html):
+            img = self.COVER_IMG_XP(html)[0]
+            href = hcover.abshref(img.get('src'))
+            item = self.manifest.hrefs.get(href, None)
+            if item is not None and item.media_type in OEB_IMAGES:
+                return item            
+        return None
+        
+    def _ensure_cover_image(self):
+        cover = self._locate_cover_image()
+        if not cover:
+            return
+        if self.metadata.cover:
+            self.metadata.cover[0].value = cover.id
+            return
         self.metadata.add('cover', cover.id)
     
     def _all_from_opf(self, opf):

src/calibre/ebooks/oeb/transforms/rasterize.py

@@ -23,6 +23,7 @@ from PyQt4.QtGui import QApplication
 from calibre.ebooks.oeb.base import XHTML_NS, XHTML, SVG_NS, SVG, XLINK
 from calibre.ebooks.oeb.base import SVG_MIME, PNG_MIME, JPEG_MIME
 from calibre.ebooks.oeb.base import xml2str, xpath, namespace, barename
+from calibre.ebooks.oeb.base import urlnormalize
 from calibre.ebooks.oeb.stylizer import Stylizer
 
 IMAGE_TAGS = set([XHTML('img'), XHTML('object')])
@@ -78,7 +79,7 @@ class SVGRasterizer(object):
             svg = item.data
         hrefs = self.oeb.manifest.hrefs
         for elem in xpath(svg, '//svg:*[@xl:href]'):
-            href = elem.attrib[XLINK('href')]
+            href = urlnormalize(elem.attrib[XLINK('href')])
             path, frag = urldefrag(href)
             if not path:
                 continue
@@ -100,15 +101,15 @@ class SVGRasterizer(object):
     def rasterize_item(self, item, stylizer):
         html = item.data
         hrefs = self.oeb.manifest.hrefs
-        for elem in xpath(html, '//h:img'):
+        for elem in xpath(html, '//h:img[@src]'):
-            src = elem.get('src', None)
+            src = urlnormalize(elem.attrib['src'])
-            image = hrefs.get(item.abshref(src), None) if src else None
+            image = hrefs.get(item.abshref(src), None)
             if image and image.media_type == SVG_MIME:
                 style = stylizer.style(elem)
                 self.rasterize_external(elem, style, item, image)
-        for elem in xpath(html, '//h:object[@type="%s"]' % SVG_MIME):
+        for elem in xpath(html, '//h:object[@type="%s" and @data]' % SVG_MIME):
-            data = elem.get('data', None)
+            data = urlnormalize(elem.attrib['data'])
-            image = hrefs.get(item.abshref(data), None) if data else None
+            image = hrefs.get(item.abshref(data), None)
             if image and image.media_type == SVG_MIME:
                 style = stylizer.style(elem)
                 self.rasterize_external(elem, style, item, image)

src/calibre/ebooks/oeb/transforms/trimmanifest.py

@@ -54,7 +54,7 @@ class ManifestTrimmer(object):
                                 new.add(found)
                 elif item.media_type == CSS_MIME:
                     def replacer(uri):
-                        absuri = item.abshref(uri)
+                        absuri = item.abshref(urlnormalize(uri))
                         if absuri in oeb.manifest.hrefs:
                             found = oeb.manifest.hrefs[href]
                             if found not in used: