KF8 Output: Handle links ot binary resources

src/calibre/ebooks/conversion/plugins/mobi_output.py

@@ -170,13 +170,13 @@ class MOBIOutput(OutputFormatPlugin):
         resources = Resources(oeb, opts, self.is_periodical,
                 add_fonts=create_kf8)
 
-        kf8 = self.create_kf8() if create_kf8 else None
+        kf8 = self.create_kf8(resources) if create_kf8 else None
 
         self.write_mobi(input_plugin, output_path, kf8, resources)
 
-    def create_kf8(self):
+    def create_kf8(self, resources):
         from calibre.ebooks.mobi.writer8.main import KF8Writer
-        return KF8Writer(self.oeb, self.opts)
+        return KF8Writer(self.oeb, self.opts, resources)
 
     def write_mobi(self, input_plugin, output_path, kf8, resources):
         from calibre.ebooks.mobi.mobiml import MobiMLizer

src/calibre/ebooks/mobi/writer2/resources.py

@@ -7,6 +7,8 @@ __license__   = 'GPL v3'
 __copyright__ = '2012, Kovid Goyal <kovid@kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
+import imghdr
+
 from calibre.ebooks.mobi import MAX_THUMB_DIMEN, MAX_THUMB_SIZE
 from calibre.ebooks.mobi.utils import (rescale_image, mobify_image,
         write_font_record)
@@ -23,6 +25,7 @@ class Resources(object):
 
         self.item_map = {}
         self.records = []
+        self.mime_map = {}
         self.masthead_offset = 0
         self.used_image_indices = set()
         self.image_indices = set()
@@ -76,6 +79,7 @@ class Resources(object):
                 self.image_indices.add(len(self.records))
                 self.records.append(data)
                 self.item_map[item.href] = index
+                self.mime_map[item.href] = 'image/%s'%imghdr.what(None, data)
                 index += 1
 
                 if cover_href and item.href == cover_href:

src/calibre/ebooks/mobi/writer8/main.py

@@ -8,20 +8,25 @@ __copyright__ = '2012, Kovid Goyal <kovid@kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
 import copy
+from functools import partial
 
 import cssutils
 
-from calibre.ebooks.oeb.base import (OEB_DOCS, OEB_STYLES, SVG_MIME)
+from calibre import isbytestring
+from calibre.ebooks.oeb.base import (OEB_DOCS, OEB_STYLES, SVG_MIME, XPath)
 
 XML_DOCS = OEB_DOCS | {SVG_MIME}
 
 class KF8Writer(object):
 
-    def __init__(self, oeb, opts):
+    def __init__(self, oeb, opts, resources):
         self.oeb, self.opts, self.log = oeb, opts, oeb.log
         self.used_images = set()
+        self.resources = resources
         self.dup_data()
 
+        self.replace_resource_links()
+
         self.create_pieces()
 
     def dup_data(self):
@@ -41,6 +46,51 @@ class KF8Writer(object):
     def data(self, item):
         return self._data_cache.get(item.href, item.data)
 
+    def replace_resource_links(self):
+        ''' Replace links to resources (raster images/fonts) with pointers to
+        the MOBI record containing the resource. The pointers are of the form:
+        kindle:embed:XXXX?mime=image/* The ?mime= is apparently optional and
+        not used for fonts. '''
+
+        def pointer(item, oref):
+            ref = item.abshref(oref)
+            idx = self.resources.item_map.get(ref, None)
+            if idx is not None:
+                is_image = self.resources.records[idx-1][:4] not in {b'FONT'}
+                if is_image:
+                    self.used_images.add(ref)
+                    return 'kindle:embed:%04d?mime=%s'%(idx,
+                            self.resources.mime_map[ref])
+                else:
+                    return 'kindle:embed:%04d'%idx
+            return oref
+
+        for item in self.oeb.manifest:
+
+            if item.media_type in XML_DOCS:
+                root = self.data(item)
+                for tag in XPath('//h:img|//svg:image')(root):
+                    for attr, ref in tag.attrib.iteritems():
+                        if attr.split('}')[-1].lower() in {'src', 'href'}:
+                            tag.attrib[attr] = pointer(item, ref)
+
+                for tag in XPath('//h:style')(root):
+                    if tag.text:
+                        sheet = cssutils.parseString(tag.text)
+                        replacer = partial(pointer, item)
+                        cssutils.replaceUrls(sheet, replacer,
+                                ignoreImportRules=True)
+                        repl = sheet.cssText
+                        if isbytestring(repl):
+                            repl = repl.decode('utf-8')
+                        tag.text = '\n'+ repl + '\n'
+
+            elif item.media_type in OEB_STYLES:
+                sheet = self.data(item)
+                replacer = partial(pointer, item)
+                cssutils.replaceUrls(sheet, replacer, ignoreImportRules=True)
+
+
     def create_pieces(self):
         self.flows = [None] # First flow item is reserved for the text