Fixed various bugs; added various improvements.

src/calibre/ebooks/mobi/mobiml.py

@@ -104,7 +104,7 @@ class MobiMLizer(object):
             return ptsize
         fbase = self.profile.fbase
         if ptsize < fbase:
-            return "%dpt" % int(round(ptsize * 2))
+            return "%dpt" % int(round(ptsize))
         return "%dem" % int(round(ptsize / fbase))
 
     def preize_text(self, text):
@@ -284,9 +284,9 @@ class MobiMLizer(object):
         else:
             istate.family = 'serif'
         valign = style['vertical-align']
-        if valign in ('super', 'sup') and asfloat(valign) > 0:
+        if valign in ('super', 'sup') or asfloat(valign) > 0:
             istate.valign = 'super'
-        elif valign == 'sub'  and asfloat(valign) < 0:
+        elif valign == 'sub'  or asfloat(valign) < 0:
             istate.valign = 'sub'
         else:
             istate.valign = 'baseline'
@@ -300,6 +300,15 @@ class MobiMLizer(object):
         if tag == 'img' and 'src' in elem.attrib:
             istate.attrib['src'] = elem.attrib['src']
             istate.attrib['align'] = 'baseline'
+            for prop in ('width', 'height'):
+                if style[prop] != 'auto':
+                    value = style[prop]
+                    if value == getattr(self.profile, prop):
+                        result = '100%'
+                    else:
+                        ems = int(round(value / self.profile.fbase))
+                        result = "%dem" % ems
+                    istate.attrib[prop] = result
         elif tag == 'hr' and asfloat(style['width']) > 0:
             prop = style['width'] / self.profile.width
             istate.attrib['width'] = "%d%%" % int(round(prop * 100))

src/calibre/ebooks/mobi/writer.py

@@ -225,11 +225,22 @@ class MobiWriter(object):
         self._oeb = oeb
         self._stream = stream
         self._records = [None]
+        self._remove_html_cover()
         self._generate_content()
         self._generate_record0()
         self._write_header()
         self._write_content()
 
+    def _remove_html_cover(self):
+        oeb = self._oeb
+        if not oeb.metadata.cover \
+           or 'cover' not in oeb.guide:
+            return
+        href = oeb.guide['cover'].href
+        del oeb.guide['cover']
+        item = oeb.manifest.hrefs[href]
+        oeb.manifest.remove(item)
+
     def _generate_content(self):
         self._map_image_names()
         self._generate_text()
@@ -391,7 +402,7 @@ class MobiWriter(object):
                 nrecs += 1
         if oeb.metadata.cover:
             id = str(oeb.metadata.cover[0])
-            item = oeb.manifest[id]
+            item = oeb.manifest.ids[id]
             href = item.href
             index = self._images[href] - 1
             exth.write(pack('>III', 0xc9, 0x0c, index))

src/calibre/ebooks/oeb/base.py

@@ -17,6 +17,7 @@ import logging
 import re
 import htmlentitydefs
 import uuid
+import copy
 from lxml import etree
 from calibre import LoggingInterface
 
@@ -32,10 +33,11 @@ XSI_NS = 'http://www.w3.org/2001/XMLSchema-instance'
 DCTERMS_NS = 'http://purl.org/dc/terms/'
 NCX_NS = 'http://www.daisy.org/z3986/2005/ncx/'
 SVG_NS = 'http://www.w3.org/2000/svg'
+XLINK_NS = 'http://www.w3.org/1999/xlink'
 XPNSMAP = {'h': XHTML_NS, 'o1': OPF1_NS, 'o2': OPF2_NS,
            'd09': DC09_NS, 'd10': DC10_NS, 'd11': DC11_NS,
            'xsi': XSI_NS, 'dt': DCTERMS_NS, 'ncx': NCX_NS,
-           'svg': SVG_NS}
+           'svg': SVG_NS, 'xl': XLINK_NS}
 
 def XML(name): return '{%s}%s' % (XML_NS, name)
 def XHTML(name): return '{%s}%s' % (XHTML_NS, name)
@@ -43,6 +45,7 @@ def OPF(name): return '{%s}%s' % (OPF2_NS, name)
 def DC(name): return '{%s}%s' % (DC11_NS, name)
 def NCX(name): return '{%s}%s' % (NCX_NS, name)
 def SVG(name): return '{%s}%s' % (SVG_NS, name)
+def XLINK(name): return '{%s}%s' % (XLINK_NS, name)
 
 EPUB_MIME = 'application/epub+zip'
 XHTML_MIME = 'application/xhtml+xml'
@@ -246,10 +249,10 @@ class Metadata(object):
         self.oeb = oeb
         self.items = defaultdict(list)
 
-    def add(self, term, value, attrib={}, **kwargs):
+    def add(self, term, value, attrib={}, index=-1, **kwargs):
         item = self.Item(term, value, attrib, **kwargs)
         items = self.items[barename(item.term)]
-        items.append(item)
+        items.insert(index, item)
         return item
 
     def iterkeys(self):
@@ -323,8 +326,7 @@ class Manifest(object):
                 data = self._loader(self.href)
                 if self.media_type in OEB_DOCS:
                     data = self._force_xhtml(data)
-                elif self.media_type[-4:] in ('+xml', '/xml') \
+                elif self.media_type[-4:] in ('+xml', '/xml'):
-                     and self.media_type != SVG_MIME:
                     data = etree.fromstring(data, parser=XML_PARSER)
                 self._data = data
                 return data
@@ -341,6 +343,9 @@ class Manifest(object):
                 return xml2str(data)
             return str(data)
 
+        def __eq__(self, other):
+            return id(self) == id(other)
+
         def __cmp__(self, other):
             result = cmp(self.spine_position, other.spine_position)
             if result != 0:
@@ -558,9 +563,12 @@ class Guide(object):
         for type, ref in self.refs.items():
             yield type, ref
     
-    def __getitem__(self, index):
+    def __getitem__(self, key):
-        return self.refs[index]
+        return self.refs[key]
 
+    def __delitem__(self, key):
+        del self.refs[key]
+    
     def __contains__(self, key):
         return key in self.refs
 
@@ -891,20 +899,27 @@ class OEBBook(object):
 
     def _ensure_cover_image(self):
         cover = None
+        spine0 = self.spine[0]
+        html = spine0.data
         if self.metadata.cover:
             id = str(self.metadata.cover[0])
-            cover = self.manifest[id]
+            cover = self.manifest.ids[id]
         elif MS_COVER_TYPE in self.guide:
             href = self.guide[MS_COVER_TYPE].href
             cover = self.manifest.hrefs[href]
-        elif 'cover' in self.guide:
+        elif xpath(html, '//h:img[position()=1]'):
-            href = self.guide['cover'].href
+            img = xpath(html, '//h:img[position()=1]')[0]
+            href = img.get('src')
             cover = self.manifest.hrefs[href]
-        else:
+        elif xpath(html, '//h:object[position()=1]'):
-            html = self.spine[0].data
+            object = xpath(html, '//h:object[position()=1]')[0]
-            imgs = xpath(html, '//h:img[position()=1]')
+            href = object.get('data')
-            href = imgs[0].get('src') if imgs else None
+            cover = self.manifest.hrefs[href]
-            cover = self.manifest.hrefs[href] if href else None
+        elif xpath(html, '//svg:svg[position()=1]'):
+            svg = copy.deepcopy(xpath(html, '//svg:svg[position()=1]')[0])
+            href = os.path.splitext(spine0.href)[0] + '.svg'
+            id, href = self.manifest.generate(spine0.id, href)
+            cover = self.manifest.add(id, href, SVG_MIME, data=svg)
         if cover and not self.metadata.cover:
             self.metadata.add('cover', cover.id)
             

src/calibre/ebooks/oeb/html.css

@@ -35,7 +35,8 @@
  *
  * ***** END LICENSE BLOCK ***** */
 
-@namespace url(http://www.w3.org/1999/xhtml); /* set default namespace to HTML */
+@namespace url(http://www.w3.org/1999/xhtml);
+@namespace svg url(http://www.w3.org/2000/svg);
 
 /* blocks */
 
@@ -399,8 +400,8 @@ br {
   display: block;
 }
 
-/* Images and embedded object size defaults */
+/* Images, embedded object, and SVG size defaults */
-img, object {
+img, object, svg|svg {
     width: auto;
     height: auto;
 }

src/calibre/ebooks/oeb/profile.py

@@ -41,8 +41,8 @@ PROFILES = {
 
     # Not really, but let's pretend
     'MobiDesktop':
-        Profile(width=280, height=300, dpi=96, fbase=12,
+        Profile(width=280, height=300, dpi=96, fbase=18,
-                fsizes=[9, 10, 11, 12, 14, 17, 20, 24]),
+                fsizes=[14, 14, 16, 18, 20, 22, 22, 24]),
     
     # No clue on usable screen size and DPI
     'CybookG3':

src/calibre/ebooks/oeb/stylizer.py

@@ -268,6 +268,7 @@ class Style(object):
         self._style = {}
         self._fontSize = None
         self._width = None
+        self._height = None
         stylizer._styles[element] = self
 
     def _update_cssdict(self, cssdict):
@@ -390,17 +391,38 @@ class Style(object):
                 base = styles[self._element.getparent()].width
             else:
                 base = self._profile.width
-            if 'width' in self._style:
+            if 'width' is self._element.attrib:
+                width = self._element.attrib['width']
+            elif 'width' in self._style:
                 width = self._style['width']
-                if width == 'auto':
-                    result = base
-                else:
-                    result = self._unit_convert(width, base=base)
             else:
                 result = base
+            if not result:
+                result = self._unit_convert(width, base=base)
             self._width = result
         return self._width
     
+    @property
+    def height(self):
+        if self._height is None:
+            result = None
+            base = None
+            if self._has_parent():
+                styles = self._stylizer._styles
+                base = styles[self._element.getparent()].height
+            else:
+                base = self._profile.height
+            if 'height' is self._element.attrib:
+                height = self._element.attrib['height']
+            elif 'height' in self._style:
+                height = self._style['height']
+            else:
+                result = base
+            if not result:
+                result = self._unit_convert(height, base=base)
+            self._height = result
+        return self._height
+    
     def __str__(self):
         items = self._style.items()
         items.sort()

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

@@ -8,17 +8,21 @@ __copyright__ = '2008, Marshall T. Vandegrift <llasram@gmail.com>'
 
 import sys
 import os
+from urlparse import urldefrag
+import base64
 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 QColor
 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 XHTML_NS, XHTML, SVG_NS, SVG, XLINK
-from calibre.ebooks.oeb.base import namespace, barename
+from calibre.ebooks.oeb.base import SVG_MIME, PNG_MIME
+from calibre.ebooks.oeb.base import xml2str, xpath, namespace, barename
 from calibre.ebooks.oeb.stylizer import Stylizer
 
 IMAGE_TAGS = set([XHTML('img'), XHTML('object')])
@@ -32,8 +36,55 @@ class SVGRasterizer(object):
         self.oeb = oeb
         self.profile = context.dest
         self.images = {}
+        self.dataize_manifest()
         self.rasterize_spine()
+        self.rasterize_cover()
 
+    def rasterize_svg(self, elem, width=0, height=0):
+        data = QByteArray(xml2str(elem))
+        svg = QSvgRenderer(data)
+        size = svg.defaultSize()
+        if size.width() == 100 and size.height() == 100 \
+           and 'viewBox' in elem.attrib:
+            box = [float(x) for x in elem.attrib['viewBox'].split()]
+            size.setWidth(box[2] - box[0])
+            size.setHeight(box[3] - box[1])
+        if width or height:
+            size.scale(width, height, Qt.KeepAspectRatio)
+        image = QImage(size, QImage.Format_ARGB32_Premultiplied)
+        image.fill(QColor("white").rgb())
+        painter = QPainter(image)
+        svg.render(painter)
+        painter.end()
+        array = QByteArray()
+        buffer = QBuffer(array)
+        buffer.open(QIODevice.WriteOnly)
+        image.save(buffer, 'PNG')
+        return str(array)
+
+    def dataize_manifest(self):
+        for item in self.oeb.manifest.values():
+            if item.media_type == SVG_MIME:
+                self.dataize_svg(item)
+
+    def dataize_svg(self, item, svg=None):
+        if svg is None:
+            svg = item.data
+        hrefs = self.oeb.manifest.hrefs
+        for elem in xpath(svg, '//svg:*[@xl:href]'):
+            href = elem.attrib[XLINK('href')]
+            path, frag = urldefrag(href)
+            if not path:
+                continue
+            abshref = item.abshref(path)
+            if abshref not in hrefs:
+                continue
+            linkee = hrefs[abshref]
+            data = base64.encodestring(str(linkee))
+            data = "data:%s;base64,%s" % (linkee.media_type, data)
+            elem.attrib[XLINK('href')] = data
+        return svg
+            
     def rasterize_spine(self):
         for item in self.oeb.spine:
             html = item.data
@@ -44,7 +95,7 @@ class SVGRasterizer(object):
         if not isinstance(elem.tag, basestring): return
         style = stylizer.style(elem)
         if namespace(elem.tag) == SVG_NS:
-            return self.rasterize_inline(elem, style)
+            return self.rasterize_inline(elem, style, item)
         if elem.tag in IMAGE_TAGS:
             manifest = self.oeb.manifest
             src = elem.get('src', None) or elem.get('data', None)
@@ -54,27 +105,46 @@ class SVGRasterizer(object):
         for child in elem:
             self.rasterize_elem(child, item, stylizer)
 
-    def rasterize_inline(self, elem, style):
+    def rasterize_inline(self, elem, style, item):
-        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
+        height = style['height']
+        if height == 'auto':
+            height = self.profile.height
         width = (width / 72) * self.profile.dpi
         height = (height / 72) * self.profile.dpi
+        elem = self.dataize_svg(item, elem)
+        data = self.rasterize_svg(elem, width, height)
+        manifest = self.oeb.manifest
+        href = os.path.splitext(item.href)[0] + '.png'
+        id, href = manifest.generate(item.id, href)
+        manifest.add(id, href, PNG_MIME, data=data)
+        img = etree.Element(XHTML('img'), src=item.relhref(href))
+        elem.getparent().replace(elem, img)
+        for prop in ('width', 'height'):
+            if prop in elem.attrib:
+                img.attrib[prop] = elem.attrib[prop]
+
+    def rasterize_external(self, elem, style, item, svgitem):
+        width = style['width']
+        if width == 'auto':
+            width = self.profile.width
+        height = style['height']
+        if height == 'auto':
+            height = self.profile.height
+        width = (width / 72) * self.profile.dpi
+        height = (height / 72) * self.profile.dpi
+        data = QByteArray(str(svgitem))
+        svg = QSvgRenderer(data)
+        size = svg.defaultSize()
         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)
+            image.fill(QColor("white").rgb())
             painter = QPainter(image)
             svg.render(painter)
             painter.end()
@@ -86,7 +156,7 @@ class SVGRasterizer(object):
             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)
+            manifest.add(id, href, PNG_MIME, data=data)
             self.images[key] = href
         elem.tag = XHTML('img')
         elem.attrib['src'] = item.relhref(href)
@@ -94,3 +164,15 @@ class SVGRasterizer(object):
         for child in elem:
             elem.remove(child)
             
+    def rasterize_cover(self):
+        covers = self.oeb.metadata.cover
+        if not covers:
+            return
+        cover = self.oeb.manifest.ids[str(covers[0])]
+        if not cover.media_type == SVG_MIME:
+            return
+        data = self.rasterize_svg(cover.data, 600, 800)
+        href = os.path.splitext(cover.href)[0] + '.png'
+        id, href = self.oeb.manifest.generate(cover.id, href)
+        self.oeb.manifest.add(id, href, PNG_MIME, data=data)
+        covers[0].value = id