Implement get_metadata3()

src/calibre/ebooks/metadata/opf.py

@@ -8,7 +8,7 @@ from __future__ import (unicode_literals, division, absolute_import,
 from lxml import etree
 
 from calibre.ebooks.metadata.opf2 import OPF, pretty_print
-from calibre.ebooks.metadata.opf3 import apply_metadata
+from calibre.ebooks.metadata.opf3 import apply_metadata, read_metadata
 from calibre.ebooks.metadata.utils import parse_opf, normalize_languages, create_manifest_item, parse_opf_version
 from calibre.ebooks.metadata import MetaInformation
 
@@ -20,13 +20,19 @@ class DummyFile(object):
     def read(self):
         return self.raw
 
+def get_metadata2(root, ver):
+    opf = OPF(None, preparsed_opf=root, read_toc=False)
+    return opf.to_book_metadata(), ver, opf.raster_cover, opf.first_spine_item()
+
+def get_metadata3(root, ver):
+    return read_metadata(root, ver=ver, return_extra_data=True)
+
 def get_metadata(stream):
     if isinstance(stream, bytes):
         stream = DummyFile(stream)
     root = parse_opf(stream)
     ver = parse_opf_version(root.get('version'))
-    opf = OPF(None, preparsed_opf=root, read_toc=False)
-    return opf.to_book_metadata(), ver, opf.raster_cover, opf.first_spine_item()
+    return get_metadata2(root, ver)
 
 def set_metadata_opf2(root, cover_prefix, mi, opf_version,
                       cover_data=None, apply_null=False, update_timestamp=False, force_identifiers=False, add_missing_cover=True):

src/calibre/ebooks/metadata/opf3.py

@@ -825,7 +825,13 @@ def ensure_is_only_raster_cover(root, prefixes, refines, raster_cover_item_href)
 
 # Reading/setting Metadata objects {{{
 
-def read_metadata(root):
+def first_spine_item(root, prefixes, refines):
+    for i in XPath('./opf:spine/opf:itemref/@idref')(root):
+        for item in XPath('./opf:manifest/opf:item')(root):
+            if item.get('id') == i:
+                return item.get('href') or None
+
+def read_metadata(root, ver=None, return_extra_data=False):
     ans = Metadata(_('Unknown'), [_('Unknown')])
     prefixes, refines = read_prefixes(root), read_refines(root)
     identifiers = read_identifiers(root, prefixes, refines)
@@ -869,6 +875,8 @@ def read_metadata(root):
     ans.user_categories = read_user_categories(root, prefixes, refines) or ans.user_categories
     for name, fm in (read_user_metadata(root, prefixes, refines) or {}).iteritems():
         ans.set_user_metadata(name, fm)
+    if return_extra_data:
+        ans = ans, ver, read_raster_cover(root, prefixes, refines), first_spine_item(root, prefixes, refines)
     return ans
 
 def get_metadata(stream):

src/calibre/ebooks/metadata/opf3_test.py

@@ -486,20 +486,22 @@ class TestOPF3(unittest.TestCase):
         "label": "date", "table":
         "custom_column_2", "is_multiple": null,
         "is_category": false}"/>
-    </metadata><manifest/>
+    </metadata><manifest><item href="start.html" media-type="text/html" id="m1"/></manifest><spine><itemref idref="m1"/></spine>
 </package>'''  # }}}
 
         def compare_metadata(mi2, mi3):
             self.ae(mi2.get_all_user_metadata(False), mi3.get_all_user_metadata(False))
             for field in ALL_METADATA_FIELDS:
-                if field != 'manifest':
+                if field not in 'manifest spine':
                     v2, v3 = getattr(mi2, field, None), getattr(mi3, field, None)
                     self.ae(v2, v3, '%s: %r != %r' % (field, v2, v3))
 
         mi2 = OPF(BytesIO(raw.encode('utf-8'))).to_book_metadata()
         root = etree.fromstring(raw)
         root.set('version', '3.0')
-        mi3 = read_metadata(root)
+        mi3, _, raster_cover, first_spine_item  = read_metadata(root, return_extra_data=True)
+        self.assertIsNone(raster_cover)
+        self.ae('start.html', first_spine_item)
         compare_metadata(mi2, mi3)
         apply_metadata(root, mi3, force_identifiers=True)
         nmi = read_metadata(root)