Various stability improvements.

src/calibre/ebooks/oeb/base.py

@@ -15,10 +15,12 @@ from urlparse import urldefrag, urlparse, urlunparse
 from urllib import unquote as urlunquote
 import logging
 import re
+from htmlentitydefs import entitydefs
+import uuid
 from lxml import etree
 from calibre import LoggingInterface
 
-XML_PARSER = etree.XMLParser(recover=True, resolve_entities=False)
+XML_PARSER = etree.XMLParser(recover=True)
 XML_NS = 'http://www.w3.org/XML/1998/namespace'
 XHTML_NS = 'http://www.w3.org/1999/xhtml'
 OPF1_NS = 'http://openebook.org/namespaces/oeb-package/1.0/'
@@ -29,15 +31,18 @@ DC11_NS = 'http://purl.org/dc/elements/1.1/'
 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'
 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}
+           'xsi': XSI_NS, 'dt': DCTERMS_NS, 'ncx': NCX_NS,
+           'svg': SVG_NS}
 
 def XML(name): return '{%s}%s' % (XML_NS, name)
 def XHTML(name): return '{%s}%s' % (XHTML_NS, name)
 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)
 
 EPUB_MIME = 'application/epub+zip'
 XHTML_MIME = 'application/xhtml+xml'
@@ -47,9 +52,15 @@ OPF_MIME = 'application/oebps-package+xml'
 OEB_DOC_MIME = 'text/x-oeb1-document'
 OEB_CSS_MIME = 'text/x-oeb1-css'
 OPENTYPE_MIME = 'font/opentype'
+GIF_MIME = 'image/gif'
+JPEG_MIME = 'image/jpeg'
+PNG_MIME = 'image/png'
+SVG_MIME = 'image/svg+xml'
 
 OEB_STYLES = set([CSS_MIME, OEB_CSS_MIME, 'text/x-oeb-css'])
 OEB_DOCS = set([XHTML_MIME, 'text/html', OEB_DOC_MIME, 'text/x-oeb-document'])
+OEB_RASTER_IMAGES = set([GIF_MIME, JPEG_MIME, PNG_MIME])
+OEB_IMAGES = set([GIF_MIME, JPEG_MIME, PNG_MIME, SVG_MIME])
 
 MS_COVER_TYPE = 'other.ms-coverimage-standard'
 
@@ -102,6 +113,9 @@ def urlnormalize(href):
     return urlunparse(parts)
 
 
+class OEBError(Exception):
+    pass
+
 class FauxLogger(object):
     def __getattr__(self, name):
         return self
@@ -162,8 +176,9 @@ class Metadata(object):
                   'xsi': XSI_NS}
     
     class Item(object):
-        def __init__(self, term, value, fq_attrib={}):
+        def __init__(self, term, value, fq_attrib={}, **kwargs):
-            self.fq_attrib = dict(fq_attrib)
+            self.fq_attrib = fq_attrib = dict(fq_attrib)
+            fq_attrib.update(kwargs)
             if term == OPF('meta') and not value:
                 term = self.fq_attrib.pop('name')
                 value = self.fq_attrib.pop('content')
@@ -225,8 +240,8 @@ class Metadata(object):
         self.oeb = oeb
         self.items = defaultdict(list)
 
-    def add(self, term, value, attrib={}):
+    def add(self, term, value, attrib={}, **kwargs):
-        item = self.Item(term, value, attrib)
+        item = self.Item(term, value, attrib, **kwargs)
         items = self.items[barename(item.term)]
         items.append(item)
         return item
@@ -267,6 +282,7 @@ class Metadata(object):
 
 class Manifest(object):
     class Item(object):
+        ENTITY_RE = re.compile(r'&([a-zA-Z_:][a-zA-Z0-9.-_:]+);')
         NUM_RE = re.compile('^(.*)([0-9][0-9.]*)(?=[.]|$)')
     
         def __init__(self, id, href, media_type,
@@ -284,19 +300,25 @@ class Manifest(object):
             return 'Item(id=%r, href=%r, media_type=%r)' \
                 % (self.id, self.href, self.media_type)
 
+        def _force_xhtml(self, data):
+            repl = lambda m: entitydefs.get(m.group(1), m.group(0))
+            data = self.ENTITY_RE.sub(repl, data)
+            data = etree.fromstring(data, parser=XML_PARSER)
+            if namespace(data.tag) != XHTML_NS:
+                data.attrib['xmlns'] = XHTML_NS
+                data = etree.tostring(data)
+                data = etree.fromstring(data, parser=XML_PARSER)
+            return data
+        
         def data():
             def fget(self):
                 if self._data is not None:
                     return self._data
                 data = self._loader(self.href)
                 if self.media_type in OEB_DOCS:
-                    data = etree.fromstring(data, parser=XML_PARSER)
+                    data = self._force_xhtml(data)
-                    if namespace(data.tag) != XHTML_NS:
+                elif self.media_type[-4:] in ('+xml', '/xml') \
-                        data.attrib['xmlns'] = XHTML_NS
+                     and self.media_type != SVG_MIME:
-                        data = etree.tostring(data)
-                        data = etree.fromstring(data, parser=XML_PARSER)
-                elif self.media_type.startswith('application/') \
-                     and self.media_type.endswith('+xml'):
                     data = etree.fromstring(data, parser=XML_PARSER)
                 self._data = data
                 return data
@@ -636,13 +658,22 @@ class OEBBook(object):
             self._all_from_opf(opf)
     
     def _convert_opf1(self, opf):
+        # Seriously, seriously wrong
+        if namespace(opf.tag) == OPF1_NS:
+            opf.tag = barename(opf.tag)
+            for elem in opf.iterdescendants():
+                if isinstance(elem.tag, basestring) \
+                   and namespace(elem.tag) == OPF1_NS:
+                    elem.tag = barename(elem.tag)
+        attrib = dict(opf.attrib)
+        attrib['version'] = '2.0'
         nroot = etree.Element(OPF('package'),
-            nsmap={None: OPF2_NS}, version="2.0", **dict(opf.attrib))
+            nsmap={None: OPF2_NS}, attrib=attrib)
         metadata = etree.SubElement(nroot, OPF('metadata'),
             nsmap={'opf': OPF2_NS, 'dc': DC11_NS,
                    'xsi': XSI_NS, 'dcterms': DCTERMS_NS})
         for prefix in ('d11', 'd10', 'd09'):
-            elements = xpath(opf, 'metadata/dc-metadata/%s:*' % prefix)
+            elements = xpath(opf, 'metadata//%s:*' % prefix)
             if elements: break
         for element in elements:
             if not element.text: continue
@@ -654,7 +685,7 @@ class OEBBook(object):
                     element.attrib[nsname] = element.attrib[name]
                     del element.attrib[name]
             metadata.append(element)
-        for element in opf.xpath('metadata/x-metadata/meta'):
+        for element in opf.xpath('metadata//meta'):
             metadata.append(element)
         for item in opf.xpath('manifest/item'):
             media_type = item.attrib['media-type'].lower()
@@ -671,23 +702,40 @@ class OEBBook(object):
     def _read_opf(self, opfpath):
         opf = self.container.read_xml(opfpath)
         version = float(opf.get('version', 1.0))
-        if version < 2.0:
+        ns = namespace(opf.tag)
+        if ns not in ('', OPF1_NS, OPF2_NS):
+            raise OEBError('Invalid namespace %r for OPF document' % ns)
+        if ns != OPF2_NS or version < 2.0:
             opf = self._convert_opf1(opf)
         return opf
     
     def _metadata_from_opf(self, opf):
-        uid = opf.attrib['unique-identifier']
+        uid = opf.get('unique-identifier', 'calibre-uuid')
+        self.uid = None
         self.metadata = metadata = Metadata(self)
-        for elem in xpath(opf, '/o2:package/o2:metadata/*'):
+        ignored = (OPF('dc-metadata'), OPF('x-metadata'))
-            if elem.text or elem.attrib:
+        for elem in xpath(opf, '/o2:package/o2:metadata//*'):
+            if elem.tag not in ignored and (elem.text or elem.attrib):
                 metadata.add(elem.tag, elem.text, elem.attrib)
+        haveuuid = haveid = False
+        for ident in metadata.identifier:
+            if unicode(ident).startswith('urn:uuid:'):
+                haveuuid = True
+            if 'id' in ident.attrib:
+                haveid = True
+        if not haveuuid and haveid:
+            bookid = "urn:uuid:%s" % str(uuid.uuid4())
+            metadata.add('identifier', bookid, id='calibre-uuid')
         for item in metadata.identifier:
             if item.id == uid:
                 self.uid = item
                 break
         else:
             self.logger.log_warn(u'Unique-identifier %r not found.' % uid)
-            self.uid = metadata.identifier[0]
+            for ident in metadata.identifier:
+                if 'id' in ident.attrib:
+                    self.uid = metadata.identifier[0]
+                    break
         if not metadata.language:
             self.logger.log_warn(u'Language not specified.')
             metadata.add('language', 'en')