Mobipocket support improvements:

- Initial SVG rasterization support.
  - Various minor improvements.

src/calibre/ebooks/mobi/mobiml.py

@@ -55,7 +55,7 @@ class FormatState(object):
         self.valign = 'baseline'
         self.italic = False
         self.bold = False
-        self.preserve = True
+        self.preserve = False
         self.family = 'serif'
         self.href = None
         self.list_num = 0
@@ -278,6 +278,10 @@ class MobiMLizer(object):
         istate.preserve = (style['white-space'] in ('pre', 'pre-wrap'))
         if 'monospace' in style['font-family']:
             istate.family = 'monospace'
+        elif 'sans-serif' in style['font-family']:
+            istate.family = 'sans-serif'
+        else:
+            istate.family = 'serif'
         valign = style['vertical-align']
         if valign in ('super', 'sup') and asfloat(valign) > 0:
             istate.valign = 'super'

src/calibre/ebooks/mobi/writer.py

@@ -19,11 +19,13 @@ from collections import defaultdict
 from urlparse import urldefrag
 from lxml import etree
 from PIL import Image
-from calibre.ebooks.oeb.base import XML_NS, XHTML, XHTML_NS, OEB_DOCS
+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 FauxLogger, OEBBook
 from calibre.ebooks.oeb.profile import Context
 from calibre.ebooks.oeb.transforms.flatcss import CSSFlattener
+from calibre.ebooks.oeb.transforms.rasterize import SVGRasterizer
 from calibre.ebooks.mobi.palmdoc import compress_doc
 from calibre.ebooks.mobi.langcodes import iana2mobi
 from calibre.ebooks.mobi.mobiml import MBP_NS, MBP, MobiMLizer
@@ -168,19 +170,30 @@ class Serializer(object):
                     index = self.images[val]
                     buffer.write('recindex="%05d"' % index)
                     continue
-                buffer.write('%s="%s"' % (attr, val))
+                buffer.write(attr)
+                buffer.write('="')
+                self.serialize_text(val, quot=True)
+                buffer.write('"')
         if elem.text or len(elem) > 0:
             buffer.write('>')
             if elem.text:
-                buffer.write(encode(elem.text))
+                self.serialize_text(elem.text)
             for child in elem:
                 self.serialize_elem(child, item)
                 if child.tail:
-                    buffer.write(encode(child.tail))
+                    self.serialize_text(child.tail)
             buffer.write('</%s>' % tag)
         else:
             buffer.write('/>')
 
+    def serialize_text(self, text, quot=False):
+        text = text.replace('&', '&amp;')
+        text = text.replace('<', '&lt;')
+        text = text.replace('>', '&gt;')
+        if quot:
+            text = text.replace('"', '&quot;')
+        self.buffer.write(encode(text))
+
     def fixup_links(self):
         buffer = self.buffer
         for id, hoffs in self.href_offsets.items():
@@ -226,7 +239,7 @@ class MobiWriter(object):
         index = 1
         self._images = images = {}
         for item in self._oeb.manifest.values():
-            if item.media_type.startswith('image/'):
+            if item.media_type in OEB_RASTER_IMAGES:
                 images[item.href] = index
                 index += 1
 
@@ -298,8 +311,8 @@ class MobiWriter(object):
         image = Image.open(StringIO(data))
         format = image.format
         changed = False
-        if image.format not in ('JPEG', 'GIF'):
+        if image.format not in ('JPEG', 'GIF', 'PNG'):
-            format = 'GIF'
+            format = 'PNG'
             changed = True
         if dimen is not None:
             image.thumbnail(dimen, Image.ANTIALIAS)
@@ -434,8 +447,10 @@ def main(argv=sys.argv):
     fbase = context.dest.fbase
     fkey = context.dest.fnums.values()
     flattener = CSSFlattener(unfloat=True, fbase=fbase, fkey=fkey)
+    rasterizer = SVGRasterizer()
     mobimlizer = MobiMLizer()
     flattener.transform(oeb, context)
+    rasterizer.transform(oeb, context)
     mobimlizer.transform(oeb, context)    
     writer.dump(oeb, outpath)
     return 0

src/calibre/ebooks/oeb/html.css

@@ -399,3 +399,8 @@ br {
   display: block;
 }
 
+/* Images and embedded object size defaults */
+img, object {
+    width: auto;
+    height: auto;
+}

src/calibre/ebooks/oeb/profile.py

@@ -41,7 +41,7 @@ PROFILES = {
 
     # Not really, but let's pretend
     'MobiDesktop':
-        Profile(width=280, height=300, dpi=100, fbase=12,
+        Profile(width=280, height=300, dpi=96, fbase=12,
                 fsizes=[9, 10, 11, 12, 14, 17, 20, 24]),
     
     # No clue on usable screen size and DPI

src/calibre/ebooks/oeb/stylizer.py

@@ -93,10 +93,12 @@ def xpath(elem, expr):
 
 class CSSSelector(etree.XPath):
     MIN_SPACE_RE = re.compile(r' *([>~+]) *')
+    LOCAL_NAME_RE = re.compile(r"(?<!local-)name[(][)] *= *'[^:]+:")
     
     def __init__(self, css, namespaces=XPNSMAP):
         css = self.MIN_SPACE_RE.sub(r'\1', css)
         path = css_to_xpath(css)
+        path = self.LOCAL_NAME_RE.sub(r"local-name() = '", path)
         etree.XPath.__init__(self, path, namespaces=namespaces)
         self.css = css
 
@@ -164,7 +166,6 @@ class Stylizer(object):
         for elem in xpath(tree, '//h:*[@style]'):
             self.style(elem)._apply_style_attr()
     
-
     def flatten_rule(self, rule, href, index):
         results = []
         if isinstance(rule, CSSStyleRule):

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

@@ -179,7 +179,8 @@ class CSSFlattener(object):
                 percent = (margin - style['text-indent']) / style['width']
                 cssdict['margin-left'] = "%d%%" % (percent * 100)
                 left -= style['text-indent']
-            if self.unfloat and 'float' in cssdict and tag != 'img':
+            if self.unfloat and 'float' in cssdict \
+               and tag not in ('img', 'object'):
                 del cssdict['float']
                 if cssdict.get('display', 'none') != 'none':
                     del cssdict['display']

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

@@ -0,0 +1,96 @@
+'''
+SVG rasterization transform.
+'''
+from __future__ import with_statement
+
+__license__   = 'GPL v3'
+__copyright__ = '2008, Marshall T. Vandegrift <llasram@gmail.com>'
+
+import sys
+import os
+from lxml import etree
+from PyQt4.QtCore import Qt
+from PyQt4.QtCore import QByteArray
+from PyQt4.QtCore import QBuffer
+from PyQt4.QtCore import QIODevice
+from PyQt4.QtGui import QImage
+from PyQt4.QtGui import QPainter
+from PyQt4.QtSvg import QSvgRenderer
+from PyQt4.QtGui import QApplication
+from calibre.ebooks.oeb.base import XHTML, SVG, SVG_NS, SVG_MIME
+from calibre.ebooks.oeb.base import namespace, barename
+from calibre.ebooks.oeb.stylizer import Stylizer
+
+IMAGE_TAGS = set([XHTML('img'), XHTML('object')])
+
+class SVGRasterizer(object):
+    def __init__(self):
+        if QApplication.instance() is None:
+            QApplication([])
+
+    def transform(self, oeb, context):
+        self.oeb = oeb
+        self.profile = context.dest
+        self.images = {}
+        self.rasterize_spine()
+
+    def rasterize_spine(self):
+        for item in self.oeb.spine:
+            html = item.data
+            stylizer = Stylizer(html, item.href, self.oeb, self.profile)
+            self.rasterize_elem(html.find(XHTML('body')), item, stylizer)
+
+    def rasterize_elem(self, elem, item, stylizer):
+        if not isinstance(elem.tag, basestring): return
+        style = stylizer.style(elem)
+        if namespace(elem.tag) == SVG_NS:
+            return self.rasterize_inline(elem, style)
+        if elem.tag in IMAGE_TAGS:
+            manifest = self.oeb.manifest
+            src = elem.get('src', None) or elem.get('data', None)
+            image = manifest.hrefs[item.abshref(src)] if src else None
+            if image and image.media_type == SVG_MIME:
+                return self.rasterize_external(elem, style, item, image)
+        for child in elem:
+            self.rasterize_elem(child, item, stylizer)
+
+    def rasterize_inline(self, elem, style):
+        pass
+
+    def rasterize_external(self, elem, style, item, svgitem):
+        data = QByteArray(svgitem.data)
+        svg = QSvgRenderer(data)
+        size = svg.defaultSize()
+        height = style['height']
+        if height == 'auto':
+            height = self.profile.height
+        width = style['width']
+        if width == 'auto':
+            width = self.profile.width
+        width = (width / 72) * self.profile.dpi
+        height = (height / 72) * self.profile.dpi
+        size.scale(width, height, Qt.KeepAspectRatio)
+        key = (svgitem.href, size.width(), size.height())
+        if key in self.images:
+            href = self.images[key]
+        else:
+            image = QImage(size, QImage.Format_ARGB32_Premultiplied)
+            painter = QPainter(image)
+            svg.render(painter)
+            painter.end()
+            array = QByteArray()
+            buffer = QBuffer(array)
+            buffer.open(QIODevice.WriteOnly)
+            image.save(buffer, 'PNG')
+            data = str(array)
+            manifest = self.oeb.manifest
+            href = os.path.splitext(svgitem.href)[0] + '.png'
+            id, href = manifest.generate(svgitem.id, href)
+            manifest.add(id, href, 'image/png', data=data)
+            self.images[key] = href
+        elem.tag = XHTML('img')
+        elem.attrib['src'] = item.relhref(href)
+        elem.text = None
+        for child in elem:
+            elem.remove(child)
+