Implement raster_cover()

src/calibre/ebooks/metadata/opf3.py

@@ -115,6 +115,9 @@ def simple_text(f):
     def wrapper(*args, **kw):
         return normalize_whitespace(f(*args, **kw))
     return wrapper
+
+def items_with_property(root, q):
+    return XPath("./opf:manifest/opf:item[@properties and contains(concat(' ', normalize-space(@properties), ' '), ' %s ')]" % q)(root)
 # }}}
 
 # Prefixes {{{
@@ -782,6 +785,8 @@ def set_user_metadata(root, prefixes, refines, val):
 
 # }}}
 
+# Reading/setting Metadata objects {{{
+
 def read_metadata(root):
     ans = Metadata(_('Unknown'), [_('Unknown')])
     prefixes, refines = read_prefixes(root), read_refines(root)
@@ -911,6 +916,46 @@ def set_metadata(stream, mi, cover_prefix='', cover_data=None, apply_null=False,
         root, mi, cover_prefix=cover_prefix, cover_data=cover_data,
         apply_null=apply_null, update_timestamp=update_timestamp,
         force_identifiers=force_identifiers)
+# }}}
+
+# Covers {{{
+
+def raster_cover(root):
+
+    def get_href(item):
+        mt = item.get('media-type')
+        if mt and 'xml' not in mt and 'html' not in mt:
+            href = item.get('href')
+            if href:
+                return href
+
+    for item in items_with_property(root, 'cover-image'):
+        href = get_href(item)
+        if href:
+            return href
+
+    for item_id in XPath('./opf:metadata/opf:meta[@name="cover"]/@content')(root):
+        for item in XPath('./opf:manifest/opf:item[@id and @href and @media-type]')(root):
+            if item.get('id') == item_id:
+                href = get_href(item)
+                if href:
+                    return href
+
+def ensure_is_only_raster_cover(root, raster_cover_item_href):
+    refines = read_refines(root)
+    for item in XPath('./opf:metadata/opf:meta[@name="cover"]')(root):
+        remove_element(item, refines)
+    for item in items_with_property(root, 'cover-image'):
+        prop = normalize_whitespace(item.get('properties').replace('cover-image', ''))
+        if prop:
+            item.set('properties', prop)
+        else:
+            del item.attrib['properties']
+    for item in XPath('./opf:manifest/opf:item')(root):
+        if item.get('href') == raster_cover_item_href:
+            item.set('properties', normalize_whitespace((item.get('properties') or '') + ' cover-image'))
+
+# }}}
 
 if __name__ == '__main__':
     import sys

src/calibre/ebooks/metadata/opf3_test.py

@@ -22,22 +22,22 @@ from calibre.ebooks.metadata.opf3 import (
     set_comments, read_publisher, set_publisher, read_tags, set_tags, read_rating,
     set_rating, read_series, set_series, read_user_metadata, set_user_metadata,
     read_author_link_map, read_user_categories, set_author_link_map, set_user_categories,
-    apply_metadata
+    apply_metadata, raster_cover, ensure_is_only_raster_cover
 )
 # This import is needed to prevent a test from running slowly
 from calibre.ebooks.oeb.polish.pretty import pretty_opf, pretty_xml_tree  # noqa
 
 read_author_link_map, read_user_categories, set_author_link_map, set_user_categories
 
-TEMPLATE = '''<package xmlns="http://www.idpf.org/2007/opf" version="3.0" prefix="calibre: %s" unique-identifier="uid"><metadata xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:opf="http://www.idpf.org/2007/opf">{metadata}</metadata></package>''' % CALIBRE_PREFIX  # noqa
+TEMPLATE = '''<package xmlns="http://www.idpf.org/2007/opf" version="3.0" prefix="calibre: %s" unique-identifier="uid"><metadata xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:opf="http://www.idpf.org/2007/opf">{metadata}</metadata><manifest>{manifest}</manifest></package>''' % CALIBRE_PREFIX  # noqa
 default_refines = defaultdict(list)
 
 class TestOPF3(unittest.TestCase):
 
     ae = unittest.TestCase.assertEqual
 
-    def get_opf(self, metadata=''):
+    def get_opf(self, metadata='', manifest=''):
-        return etree.fromstring(TEMPLATE.format(metadata=metadata))
+        return etree.fromstring(TEMPLATE.format(metadata=metadata, manifest=manifest))
 
     def test_prefix_parsing(self):  # {{{
         self.ae(parse_prefixes('foaf: http://xmlns.com/foaf/spec/\n dbp: http://dbpedia.org/ontology/'),
@@ -205,6 +205,20 @@ class TestOPF3(unittest.TestCase):
         self.ae('<a>p</a>', st(root, '<a>p</a> '))
     # }}}
 
+    def test_raster_cover(self):  # {{{
+        def rt(root):
+            return raster_cover(root)
+        root = self.get_opf('<meta name="cover" content="cover"/>', '<item id="cover" media-type="image/jpeg" href="x.jpg"/>')
+        self.ae('x.jpg', rt(root))
+        root = self.get_opf('<meta name="cover" content="cover"/>',
+                            '<item id="cover" media-type="image/jpeg" href="x.jpg"/><item media-type="image/jpeg" href="y.jpg" properties="cover-image"/>')
+        self.ae('y.jpg', rt(root))
+        ensure_is_only_raster_cover(root, 'x.jpg')
+        self.ae('x.jpg', rt(root))
+        self.ae(['x.jpg'], root.xpath('//*[@properties="cover-image"]/@href'))
+        self.assertFalse(root.xpath('//*[@name]'))
+    # }}}
+
     def test_tags(self):  # {{{
         def rt(root):
             return read_tags(root, read_prefixes(root), read_refines(root))