Add support for embedded fonts and entity declarations to ebook-viewer

src/calibre/ebooks/epub/iterator.py

@@ -7,6 +7,8 @@ Iterate over the HTML files in an ebook. Useful for writing viewers.
 
 import re, os, math, copy
 
+from PyQt4.Qt import QFontDatabase
+
 from calibre.ebooks.epub.from_any import MAP
 from calibre.ebooks.epub.from_html import TITLEPAGE
 from calibre.ebooks.epub import config 
@@ -77,6 +79,25 @@ class EbookIterator(object):
                 if text in open(path, 'rb').read().decode(path.encoding).lower():
                     return i
     
+    def find_embedded_fonts(self):
+        for item in self.opf.manifest:
+            if item.mime_type and 'css' in item.mime_type.lower():
+                css = open(item.path, 'rb').read().decode('utf-8')
+                for match in re.compile(r'@font-face\s*{([^}]+)}').finditer(css):
+                    block  = match.group(1)
+                    family = re.compile(r'font-family\s*:\s*([^;]+)').search(block)
+                    url    = re.compile(r'url\s*\((.+?)\)', re.DOTALL).search(block)
+                    if url:
+                        path = url.group(1).split('/')
+                        path = os.path.join(os.path.dirname(item.path), *path) 
+                        id = QFontDatabase.addApplicationFont(path)
+                        if id != -1:
+                            families = [unicode(f) for f in QFontDatabase.applicationFontFamilies(id)]
+                            if family:
+                                family = family.group(1).strip().replace('"', '')
+                                if family not in families:
+                                    print 'WARNING: Family aliasing not supported:', block
+    
     def __enter__(self):
         self._tdir = TemporaryDirectory('_ebook_iter')
         self.base  = self._tdir.__enter__()
@@ -104,7 +125,7 @@ class EbookIterator(object):
             s.max_page = s.start_page + s.pages - 1
         self.toc = self.opf.toc
         
-         
+        self.find_embedded_fonts() 
         
         return self
         

src/calibre/ebooks/metadata/epub.py

@@ -118,11 +118,10 @@ class CoverRenderer(QObject):
         self.page.mainFrame().load(url)
         
     def render_html(self, ok):
-        from calibre.utils.PythonMagickWand import ImageMagick, NewMagickWand, MagickGetImageBlob, \
-                                           MagickReadImageBlob, MagickTrimImage, MagickSetFormat
-
         self.rendered = True
         try:
+            from calibre.utils.PythonMagickWand import ImageMagick, NewMagickWand, MagickGetImageBlob, \
+                                           MagickReadImageBlob, MagickTrimImage, MagickSetFormat
             if not ok:
                 return
             self.page.setViewportSize(QSize(1280, 1024))

src/calibre/gui2/viewer/documentview.py

@@ -5,7 +5,7 @@ __docformat__ = 'restructuredtext en'
 
 '''
 '''
-import os, math
+import os, math, re
 from PyQt4.Qt import QWidget, QSize, QSizePolicy, QUrl, SIGNAL, Qt, QTimer, \
                      QPainter, QPalette, QBrush, QFontDatabase, QDialog, \
                      QByteArray, QColor, QWheelEvent, QPoint, QImage, QRegion, QFont
@@ -215,6 +215,18 @@ class Document(QWebPage):
             return self.mainFrame().contentsSize().width() # offsetWidth gives inaccurate results
         return property(fget=fget)
 
+class EntityDeclarationProcessor(object):
+    
+    def __init__(self, html):
+        self.declared_entities = {}
+        for match in re.finditer(r'<!\s*ENTITY\s+([^>]+)>', html):
+            tokens = match.group(1).split()
+            if len(tokens) > 1:
+                self.declared_entities[tokens[0].strip()] = tokens[1].strip().replace('"', '')
+        self.processed_html = html
+        for key, val in self.declared_entities.iteritems():
+            self.processed_html = self.processed_html.replace('&%s;'%key, val)
+
 class DocumentView(QWebView):
     
     DISABLED_BRUSH = QBrush(Qt.lightGray, Qt.Dense5Pattern) 
@@ -290,6 +302,7 @@ class DocumentView(QWebView):
     def load_path(self, path, pos=0.0):
         self.initial_pos = pos
         html = open(path, 'rb').read().decode(path.encoding)
+        html = EntityDeclarationProcessor(html).processed_html
         self.setHtml(html, QUrl.fromLocalFile(path))
         
     def load_started(self):

src/calibre/gui2/viewer/main.py

@@ -363,7 +363,9 @@ class EbookViewer(MainWindow, Ui_EbookViewer):
         self.view.load_path(path, pos=pos)
     
     def viewport_resized(self, frac):
-        self.set_page_number(frac)
+        pos = self.pos.value()
+        self.view.scroll_to(pos, notify=False)
+        #self.set_page_number(frac)
     
     def close_progress_indicator(self):
         self.pi.stop()