Implement setting of identifiers

src/calibre/ebooks/metadata/opf3.py

@@ -11,8 +11,8 @@ from lxml import etree
 
 from calibre.ebooks.metadata import check_isbn
 from calibre.ebooks.metadata.book.base import Metadata
-from calibre.ebooks.metadata.utils import parse_opf
-from calibre.ebooks.oeb.base import OPF2_NSMAP, OPF
+from calibre.ebooks.metadata.utils import parse_opf, pretty_print_opf
+from calibre.ebooks.oeb.base import OPF2_NSMAP, OPF, DC
 
 # Utils {{{
 # http://www.idpf.org/epub/vocab/package/pfx/
@@ -119,6 +119,32 @@ def read_identifiers(root, prefixes, refines):
             if scheme and val:
                 ans[scheme].append(val)
     return ans
+
+def set_identifiers(root, prefixes, refines, new_identifiers, force_identifiers=False):
+    uid = root.get('unique-identifier')
+    package_identifier = None
+    for ident in XPath('./opf:metadata/dc:identifier')(root):
+        if uid is not None and uid == ident.get('id'):
+            package_identifier = ident
+            continue
+        val = (ident.text or '').strip()
+        if not val:
+            ident.getparent().remove(ident)
+            continue
+        scheme, val = parse_identifier(ident, val, refines)
+        if not scheme or not val or force_identifiers or scheme in new_identifiers:
+            ident.getparent().remove(ident)
+            continue
+    metadata = XPath('./opf:metadata')(root)[0]
+    for scheme, val in new_identifiers.iteritems():
+        ident = metadata.makeelement(DC('identifier'))
+        ident.text = '%s:%s' % (scheme, val)
+        if package_identifier is None:
+            metadata.append(ident)
+        else:
+            p = package_identifier.getparent()
+            p.insert(p.index(package_identifier), ident)
+
 # }}}
 
 def read_metadata(root):
@@ -139,6 +165,18 @@ def get_metadata(stream):
     root = parse_opf(stream)
     return read_metadata(root)
 
+def apply_metadata(root, mi, cover_prefix='', cover_data=None, apply_null=False, update_timestamp=False, force_identifiers=False):
+    prefixes, refines = read_prefixes(root), read_refines(root)
+    set_identifiers(root, prefixes, refines, mi.identifiers, force_identifiers=force_identifiers)
+    pretty_print_opf(root)
+
+def set_metadata(stream, mi, cover_prefix='', cover_data=None, apply_null=False, update_timestamp=False, force_identifiers=False):
+    root = parse_opf(stream)
+    return apply_metadata(
+        root, mi, cover_prefix=cover_prefix, cover_data=cover_data,
+        apply_null=apply_null, update_timestamp=update_timestamp,
+        force_identifiers=force_identifiers)
+
 if __name__ == '__main__':
     import sys
     print(get_metadata(open(sys.argv[-1], 'rb')))

src/calibre/ebooks/metadata/opf3_test.py

@@ -10,10 +10,12 @@ import unittest
 from lxml import etree
 
 from calibre.ebooks.metadata.opf3 import (
-    parse_prefixes, reserved_prefixes, expand_prefix, read_identifiers, read_metadata
+    parse_prefixes, reserved_prefixes, expand_prefix, read_identifiers,
+    read_metadata, set_identifiers, XPath
 )
 
 TEMPLATE = '''<package xmlns="http://www.idpf.org/2007/opf" version="3.0" unique-identifier="uid"><metadata xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:opf="http://www.idpf.org/2007/opf">{metadata}</metadata></package>'''  # noqa
+default_refines = defaultdict(list)
 
 class TestOPF3(unittest.TestCase):
 
@@ -33,8 +35,11 @@ class TestOPF3(unittest.TestCase):
             self.ae(expand_prefix(raw, reserved_prefixes), expanded)
 
     def test_identifiers(self):
-        def idt(val, scheme=None):
-            return '<dc:identifier {scheme}>{val}</dc:identifier>'.format(scheme=('opf:scheme="%s"'%scheme if scheme else ''), val=val)
+        def idt(val, scheme=None, iid=''):
+            return '<dc:identifier id="{id}" {scheme}>{val}</dc:identifier>'.format(scheme=('opf:scheme="%s"'%scheme if scheme else ''), val=val, id=iid)
+        def ri(root):
+            return dict(read_identifiers(root, reserved_prefixes, default_refines))
+
         for m, result in (
                 (idt('abc', 'ISBN'), {}),
                 (idt('isbn:9780230739581'), {'isbn':['9780230739581']}),
@@ -45,11 +50,19 @@ class TestOPF3(unittest.TestCase):
                 (idt('url:http://x'), {'url':['http://x']}),
                 (idt('a:1')+idt('a:2'), {'a':['1', '2']}),
         ):
-            self.ae(result, dict(read_identifiers(self.get_opf(m), reserved_prefixes, defaultdict(list))))
+            self.ae(result, ri(self.get_opf(m)))
         mi = read_metadata(self.get_opf(
             metadata=idt('a:1')+idt('a:2')+idt('calibre:x')+idt('uuid:y')))
         self.ae(mi.application_id, 'x')
 
+        root = self.get_opf(metadata=idt('i:1', iid='uid') + idt('r:1') + idt('o:1'))
+        set_identifiers(root, reserved_prefixes, default_refines, {'i':'2', 'o':'2'})
+        self.ae({'i':['2', '1'], 'r':['1'], 'o':['2']}, ri(root))
+        self.ae(1, len(XPath('//dc:identifier[@id="uid"]')(root)))
+        root = self.get_opf(metadata=idt('i:1', iid='uid') + idt('r:1') + idt('o:1'))
+        set_identifiers(root, reserved_prefixes, default_refines, {'i':'2', 'o':'2'}, force_identifiers=True)
+        self.ae({'i':['2', '1'], 'o':['2']}, ri(root))
+
 class TestRunner(unittest.main):
 
     def createTests(self):