EPUB Metadata: Implement updating identifiers

EPUB Metadata: Implementing updating identifiers in the epub file from
calibre when polishing or exporting the epub

ebook-meta: Add an --identifier option to set identifiers.

src/calibre/customize/builtins.py

@@ -435,7 +435,7 @@ class EPUBMetadataWriter(MetadataWriterPlugin):
 
     def set_metadata(self, stream, mi, type):
         from calibre.ebooks.metadata.epub import set_metadata
-        set_metadata(stream, mi, apply_null=self.apply_null)
+        set_metadata(stream, mi, apply_null=self.apply_null, force_identifiers=self.force_identifiers)
 
 class FB2MetadataWriter(MetadataWriterPlugin):
 
@@ -1695,7 +1695,7 @@ class StoreWHSmithUKStore(StoreBase):
 class StoreWolneLekturyStore(StoreBase):
     name = 'Wolne Lektury'
     author = u'Tomasz Długosz'
-    description = u'Wolne Lektury to biblioteka internetowa czynna 24 godziny na dobę, 365 dni w roku, której zasoby dostępne są całkowicie za darmo. Wszystkie dzieła są odpowiednio opracowane - opatrzone przypisami, motywami i udostępnione w kilku formatach - HTML, TXT, PDF, EPUB, MOBI, FB2.'
+    description = u'Wolne Lektury to biblioteka internetowa czynna 24 godziny na dobę, 365 dni w roku, której zasoby dostępne są całkowicie za darmo. Wszystkie dzieła są odpowiednio opracowane - opatrzone przypisami, motywami i udostępnione w kilku formatach - HTML, TXT, PDF, EPUB, MOBI, FB2.'  # noqa
     actual_plugin = 'calibre.gui2.store.stores.wolnelektury_plugin:WolneLekturyStore'
 
     headquarters = 'PL'

src/calibre/customize/ui.py

@@ -319,6 +319,19 @@ class ApplyNullMetadata(object):
 
 apply_null_metadata = ApplyNullMetadata()
 
+class ForceIdentifiers(object):
+
+    def __init__(self):
+        self.force_identifiers = False
+
+    def __enter__(self):
+        self.force_identifiers = True
+
+    def __exit__(self, *args):
+        self.force_identifiers = False
+
+force_identifiers = ForceIdentifiers()
+
 def get_file_type_metadata(stream, ftype):
     mi = MetaInformation(None, None)
 
@@ -346,6 +359,7 @@ def set_file_type_metadata(stream, mi, ftype):
                 with plugin:
                     try:
                         plugin.apply_null = apply_null_metadata.apply_null
+                        plugin.force_identifiers = force_identifiers.force_identifiers
                         plugin.set_metadata(stream, mi, ftype.lower().strip())
                         break
                     except:

src/calibre/ebooks/metadata/cli.py

@@ -9,7 +9,7 @@ ebook-meta
 import sys, os
 
 from calibre.utils.config import StringConfig
-from calibre.customize.ui import metadata_readers, metadata_writers
+from calibre.customize.ui import metadata_readers, metadata_writers, force_identifiers
 from calibre.ebooks.metadata.meta import get_metadata, set_metadata
 from calibre.ebooks.metadata import string_to_authors, authors_to_sort_string, \
                     title_sort, MetaInformation
@@ -63,6 +63,11 @@ def config():
               help=_('Set the rating. Should be a number between 1 and 5.'))
     c.add_opt('isbn', ['--isbn'],
               help=_('Set the ISBN of the book.'))
+    c.add_opt('identifiers', ['--identifier'], action='append',
+              help=_('Set the identifiers for the book, can be specified multiple times.'
+                     ' For example: --identifier uri:http://acme.com --identifier isbn:12345'
+                     ' To remove an identifier, specify no value, --identifier isbn:'
+                     ' Note that for EPUB files, an identifier marked as the package identifier cannot be removed.'))
     c.add_opt('tags', ['--tags'],
               help=_('Set the tags for the book. Should be a comma separated list.'))
     c.add_opt('book_producer', ['-k', '--book-producer'],
@@ -114,7 +119,7 @@ def do_set_metadata(opts, mi, stream, stream_type):
     for pref in config().option_set.preferences:
         if pref.name in ('to_opf', 'from_opf', 'authors', 'title_sort',
                          'author_sort', 'get_cover', 'cover', 'tags',
-                         'lrf_bookid'):
+                         'lrf_bookid', 'identifiers'):
             continue
         val = getattr(opts, pref.name, None)
         if val is not None:
@@ -136,11 +141,19 @@ def do_set_metadata(opts, mi, stream, stream_type):
         mi.series_index = float(opts.series_index.strip())
     if getattr(opts, 'pubdate', None) is not None:
         mi.pubdate = parse_date(opts.pubdate, assume_utc=False, as_utc=False)
+    if getattr(opts, 'identifiers', None):
+        val = {k.strip():v.strip() for k, v in (x.partition(':')[0::2] for x in opts.identifiers)}
+        if val:
+            orig = mi.get_identifiers()
+            orig.update(val)
+            val = {k:v for k, v in orig.iteritems() if k and v}
+            mi.set_identifiers(val)
 
     if getattr(opts, 'cover', None) is not None:
         ext = os.path.splitext(opts.cover)[1].replace('.', '').upper()
         mi.cover_data = (ext, open(opts.cover, 'rb').read())
 
+    with force_identifiers:
         set_metadata(stream, mi, stream_type)
 
 

src/calibre/ebooks/metadata/epub.py

@@ -250,7 +250,7 @@ def _write_new_cover(new_cdata, cpath):
     save_cover_data_to(new_cdata, new_cover.name)
     return new_cover
 
-def update_metadata(opf, mi, apply_null=False, update_timestamp=False):
+def update_metadata(opf, mi, apply_null=False, update_timestamp=False, force_identifiers=False):
     for x in ('guide', 'toc', 'manifest', 'spine'):
         setattr(mi, x, None)
     if mi.languages:
@@ -274,10 +274,16 @@ def update_metadata(opf, mi, apply_null=False, update_timestamp=False):
             opf.isbn = None
         if not getattr(mi, 'comments', None):
             opf.comments = None
+    if apply_null or force_identifiers:
+        opf.set_identifiers(mi.get_identifiers())
+    else:
+        orig = opf.get_identifiers()
+        orig.update(mi.get_identifiers())
+        opf.set_identifiers({k:v for k, v in orig.iteritems() if k and v})
     if update_timestamp and mi.timestamp is not None:
         opf.timestamp = mi.timestamp
 
-def set_metadata(stream, mi, apply_null=False, update_timestamp=False):
+def set_metadata(stream, mi, apply_null=False, update_timestamp=False, force_identifiers=False):
     stream.seek(0)
     reader = get_zip_reader(stream, root=os.getcwdu())
     raster_cover = reader.opf.raster_cover
@@ -308,7 +314,7 @@ def set_metadata(stream, mi, apply_null=False, update_timestamp=False):
             traceback.print_exc()
 
     update_metadata(reader.opf, mi, apply_null=apply_null,
-                    update_timestamp=update_timestamp)
+                    update_timestamp=update_timestamp, force_identifiers=force_identifiers)
 
     newopf = StringIO(reader.opf.render())
     if isinstance(reader.archive, LocalZipFile):

src/calibre/ebooks/metadata/opf2.py

@@ -959,6 +959,33 @@ class OPF(object):  # {{{
                         identifiers['isbn'] = val
         return identifiers
 
+    def set_identifiers(self, identifiers):
+        identifiers = identifiers.copy()
+        uuid_id = None
+        for attr in self.root.attrib:
+            if attr.endswith('unique-identifier'):
+                uuid_id = self.root.attrib[attr]
+                break
+
+        for x in self.XPath(
+            'descendant::*[local-name() = "identifier"]')(
+                    self.metadata):
+            xid = x.get('id', None)
+            is_package_identifier = uuid_id is not None and uuid_id == xid
+            typ = {val for attr, val in x.attrib.iteritems() if attr.endswith('scheme')}
+            if is_package_identifier:
+                typ = tuple(typ)
+                if typ and typ[0].lower() in identifiers:
+                    self.set_text(x, identifiers.pop(typ[0].lower()))
+                continue
+            if typ and not (typ & {'calibre', 'uuid'}):
+                x.getparent().remove(x)
+
+        for typ, val in identifiers.iteritems():
+            attrib = {'{%s}scheme'%self.NAMESPACES['opf']: typ.upper()}
+            self.set_text(self.create_metadata_element(
+                'identifier', attrib=attrib), unicode(val))
+
     @dynamic_property
     def application_id(self):