RTF Input: Support extraction of images (JPEG/PNG only)

src/calibre/ebooks/rtf/input.py

@@ -2,7 +2,7 @@ from __future__ import with_statement
 __license__   = 'GPL v3'
 __copyright__ = '2008, Kovid Goyal <kovid at kovidgoyal.net>'
 
-import os
+import os, glob, re
 
 from lxml import etree
 
@@ -61,6 +61,30 @@ class RTFInput(InputFormatPlugin):
         os.remove('out.xml')
         return ans
 
+    def extract_images(self, picts):
+        self.log('Extracting images...')
+        count = 0
+        raw = open(picts, 'rb').read()
+        starts = []
+        for match in re.finditer(r'\{\\pict([^}]+)\}', raw):
+            starts.append(match.start(1))
+
+        for start in starts:
+            pos, bc = start, 1
+            while bc > 0:
+                if raw[pos] == '}': bc -= 1
+                elif raw[pos] == '{': bc += 1
+                pos += 1
+            pict = raw[start:pos+1]
+            enc = re.sub(r'[^a-zA-Z0-9]', '', pict)
+            if len(enc) % 2 == 1:
+                enc = enc[:-1]
+            data = enc.decode('hex')
+            count += 1
+            name = (('%4d'%count).replace(' ', '0'))+'.jpg'
+            open(name, 'wb').write(data)
+            #open(name+'.hex', 'wb').write(enc)
+
     def convert(self, stream, options, file_ext, log,
                 accelerators):
         from calibre.ebooks.rtf.xsl import xhtml
@@ -74,6 +98,12 @@ class RTFInput(InputFormatPlugin):
         except RtfInvalidCodeException:
             raise ValueError(_('This RTF file has a feature calibre does not '
             'support. Convert it to HTML first and then try it.'))
+        d = glob.glob(os.path.join('*_rtf_pict_dir', 'picts.rtf'))
+        if d:
+            try:
+                self.extract_images(d[0])
+            except:
+                self.log.exception('Failed to extract images...')
         self.log('Parsing XML...')
         parser = etree.XMLParser(recover=True, no_network=True)
         doc = etree.fromstring(xml, parser=parser)

src/calibre/ebooks/rtf/xsl.py

@@ -523,7 +523,11 @@ xhtml = '''\
         <xsl:apply-templates/>
     </xsl:template>
 
-    <xsl:template match="rtf:pict" />
+    <xsl:template match="rtf:pict">
+        <xsl:element name="img">
+            <xsl:attribute name="src"><xsl:value-of select="@num" />.jpg</xsl:attribute>
+        </xsl:element>
+    </xsl:template>
 
     <xsl:template match="*">
         <xsl:message>

src/calibre/ebooks/rtf2xml/pict.py

@@ -55,7 +55,7 @@ class Pict:
         return "}\n"
     def __text_func(self, line):
         #tx<nu<__________<true text
-        return line[18:]
+        return line[17:]
     def __make_dir(self):
         """ Make a dirctory to put the image data in"""
         base_name = os.path.basename(getattr(self.__orig_file, 'name',