Preserve notes resource mtimes when importing notes db

src/calibre/db/backend.py

@@ -989,8 +989,8 @@ class DB:
         id_val = self.tables[field].id_map[item_id]
         return self.notes.unretire(self.conn, field, item_id, id_val)
 
-    def add_notes_resource(self, path_or_stream, name) -> int:
-        return self.notes.add_resource(self.conn, path_or_stream, name)
+    def add_notes_resource(self, path_or_stream, name, mtime=None) -> int:
+        return self.notes.add_resource(self.conn, path_or_stream, name, mtime=mtime)
 
     def get_notes_resource(self, resource_hash) -> Optional[dict]:
         return self.notes.get_resource_data(self.conn, resource_hash)

src/calibre/db/cache.py

@@ -20,7 +20,7 @@ from functools import partial, wraps
 from io import DEFAULT_BUFFER_SIZE, BytesIO
 from queue import Queue
 from threading import Lock
-from time import monotonic, sleep, time
+from time import mktime, monotonic, sleep, time
 from typing import NamedTuple, Optional, Tuple
 
 from calibre import as_unicode, detect_ncpus, isbytestring
@@ -724,9 +724,9 @@ class Cache:
         return self.backend.set_notes_for(field, item_id, doc, searchable_text, resource_hashes, remove_unused_resources)
 
     @write_api
-    def add_notes_resource(self, path_or_stream_or_data, name: str) -> int:
+    def add_notes_resource(self, path_or_stream_or_data, name: str, mtime: float = None) -> int:
         ' Add the specified resource so it can be referenced by notes and return its content hash '
-        return self.backend.add_notes_resource(path_or_stream_or_data, name)
+        return self.backend.add_notes_resource(path_or_stream_or_data, name, mtime)
 
     @read_api
     def get_notes_resource(self, resource_hash) -> Optional[dict]:
@@ -3460,7 +3460,10 @@ def import_library(library_key, importer, library_path, progress=None, abort=Non
         with closing(importer.start_file(metadata['notes.db'], 'notes.db for ' + library_path)) as stream:
             stream.check_hash = False
             with zipfile.ZipFile(stream) as zf:
-                zf.extractall(notes_dir)
+                for zi in zf.infolist():
+                    tpath = zf._extract_member(zi, notes_dir, None)
+                    date_time = mktime(zi.date_time + (0, 0, -1))
+                    os.utime(tpath, (date_time, date_time))
     if abort is not None and abort.is_set():
         return
     cache = Cache(DB(library_path, load_user_formatter_functions=False))

src/calibre/db/notes/connect.py

@@ -306,7 +306,7 @@ class Notes:
             for path in items[:extra]:
                 remove_with_retry(path, is_dir=True)
 
-    def add_resource(self, conn, path_or_stream_or_data, name, update_name=True):
+    def add_resource(self, conn, path_or_stream_or_data, name, update_name=True, mtime=None):
         if isinstance(path_or_stream_or_data, bytes):
             data = path_or_stream_or_data
         elif isinstance(path_or_stream_or_data, str):
@@ -332,6 +332,9 @@ class Notes:
                 f = open(path, 'wb')
             with f:
                 f.write(data)
+            if mtime is not None:
+                os.utime(f.name, (mtime, mtime))
+
         name = sanitize_file_name(name)
         base_name, ext = os.path.splitext(name)
         c = 0

src/calibre/db/tests/filesystem.py

@@ -264,8 +264,8 @@ class FilesystemTest(BaseTest):
                 bookdir = os.path.dirname(ic.format_abspath(1, '__COVER_INTERNAL__'))
                 self.assertEqual('exf', open(os.path.join(bookdir, 'exf')).read())
                 self.assertEqual('recurse', open(os.path.join(bookdir, 'sub', 'recurse')).read())
-        r1 = cache.add_notes_resource(b'res1', 'res.jpg')
-        r2 = cache.add_notes_resource(b'res2', 'res.jpg')
+        r1 = cache.add_notes_resource(b'res1', 'res.jpg', mtime=time.time()-113)
+        r2 = cache.add_notes_resource(b'res2', 'res.jpg', mtime=time.time()-1115)
         cache.set_notes_for('authors', 2, 'some notes', resource_hashes=(r1, r2))
         cache.add_format(1, 'TXT', BytesIO(b'testing exim'))
         cache.fts_indexing_sleep_time = 0.001
@@ -285,7 +285,10 @@ class FilesystemTest(BaseTest):
             ic = import_library('l', importer, idir)
             self.assertEqual(ic.fts_search('exim')[0]['id'], 1)
             self.assertEqual(cache.notes_for('authors', 2), ic.notes_for('authors', 2))
-            self.assertEqual(cache.get_notes_resource(r1), ic.get_notes_resource(r1))
+            a, b = cache.get_notes_resource(r1), ic.get_notes_resource(r1)
+            at, bt, = a.pop('mtime'), b.pop('mtime')
+            self.assertEqual(a, b)
+            self.assertLess(abs(at-bt), 2)
 
     def test_find_books_in_directory(self):
         from calibre.db.adding import find_books_in_directory, compile_rule

src/calibre/db/tests/notes.py

@@ -151,6 +151,10 @@ def test_cache_api(self: 'NotesTest'):
         self.assertGreater(note_id, 0)
         self.assertIn('<p>test simple exim <img', cache.notes_for('authors', author_id))
         res2 = tuple(cache.get_notes_resource(x) for x in cache.notes_resources_used_by('authors', author_id))
+        for x in res:
+            del x['mtime']
+        for x in res2:
+            del x['mtime']
         self.ae(sorted(res, key=itemgetter('name')), sorted(res2, key=itemgetter('name')))