When cleaning viewer cache remove all but the newest entry for a particular book path

src/calibre/gui2/viewer/convert_book.py

@@ -90,21 +90,46 @@ def expire_cache(path, instances, max_age):
             instances.remove(instance)
 
 
-def expire_cache_and_temp(temp_path, finished_path, metadata, max_age):
+def expire_old_versions(path, instances):
+    instances = filter(lambda x: x['status'] == 'finished', instances)
+    remove = sorted(instances, key=lambda x: x['atime'], reverse=True)[1:]
+    for instance in remove:
+        if robust_rmtree(os.path.join(path, instance['path'])):
+            yield instance
+
+
+def expire_cache_and_temp(temp_path, finished_path, metadata, max_age, force_expire):
     now = time.time()
-    if now - metadata['last_clear_at'] < DAY and max_age >= 0:
+    if now - metadata['last_clear_at'] < DAY and max_age >= 0 and not force_expire:
         return
     clear_temp(temp_path)
     entries = metadata['entries']
+    path_key_map = {}
     for key, instances in tuple(entries.items()):
         if instances:
             expire_cache(finished_path, instances, max_age)
             if not instances:
                 del entries[key]
+            else:
+                for x in instances:
+                    book_path = x.get('book_path')
+                    if book_path:
+                        path_key_map.setdefault(book_path, []).append(key)
+    for keys in path_key_map.values():
+        instances = []
+        for key in keys:
+            instances += entries[key]
+        if len(instances) > 1:
+            removed = tuple(expire_old_versions(finished_path, instances))
+            if removed:
+                for r in removed:
+                    entries[r['key']].remove(r)
+                    if not entries[r['key']]:
+                        del entries[r['key']]
     metadata['last_clear_at'] = now
 
 
-def prepare_convert(temp_path, key, st):
+def prepare_convert(temp_path, key, st, book_path):
     tdir = tempfile.mkdtemp(dir=temp_path)
     now = time.time()
     return {
@@ -117,6 +142,7 @@ def prepare_convert(temp_path, key, st):
         'file_mtime': st.st_mtime,
         'file_size': st.st_size,
         'cache_size': 0,
+        'book_path': book_path,
     }
 
 
@@ -170,7 +196,7 @@ def save_metadata(metadata, f):
     f.seek(0), f.truncate(), f.write(as_bytes(json.dumps(metadata, indent=2)))
 
 
-def prepare_book(path, convert_func=do_convert, max_age=30 * DAY, force=False, prepare_notify=None):
+def prepare_book(path, convert_func=do_convert, max_age=30 * DAY, force=False, prepare_notify=None, force_expire=False):
     st = os.stat(path)
     key = book_hash(path, st.st_size, st.st_mtime)
     finished_path = safe_makedirs(os.path.join(book_cache_dir(), 'f'))
@@ -194,7 +220,7 @@ def prepare_book(path, convert_func=do_convert, max_age=30 * DAY, force=False, p
                     return os.path.join(finished_path, instance['path'])
         if prepare_notify:
             prepare_notify()
-        instance = prepare_convert(temp_path, key, st)
+        instance = prepare_convert(temp_path, key, st, path)
         instances.append(instance)
         save_metadata(metadata, f)
     convert_func(path, temp_path, key, instance)
@@ -220,7 +246,7 @@ def prepare_book(path, convert_func=do_convert, max_age=30 * DAY, force=False, p
             if q['id'] == instance['id']:
                 q.update(instance)
                 break
-        expire_cache_and_temp(temp_path, finished_path, metadata, max_age)
+        expire_cache_and_temp(temp_path, finished_path, metadata, max_age, force_expire)
         save_metadata(metadata, f)
     return ans
 
@@ -304,6 +330,23 @@ def find_tests():
             prepare_book(book_src, convert_func=convert_mock, max_age=-1000)
             self.ae([], os.listdir(os.path.join(book_cache_dir(), 'f')))
 
+            # Test modifying a book and opening it repeatedly leaves only
+            # a single entry for it in the cache
+            prepare_book(book_src, convert_func=convert_mock, force_expire=True)
+            finished_entries = os.listdir(os.path.join(book_cache_dir(), 'f'))
+            self.ae(len(finished_entries), 1)
+            open(book_src, 'wb').write(b'bcde')
+            prepare_book(book_src, convert_func=convert_mock, force_expire=True)
+            new_finished_entries = os.listdir(os.path.join(book_cache_dir(), 'f'))
+            self.ae(len(new_finished_entries), 1)
+            self.assertNotEqual(finished_entries, new_finished_entries)
+            open(book_src, 'wb').write(b'bcdef')
+            prepare_book(book_src, convert_func=convert_mock, max_age=-1000, force_expire=True)
+            self.ae([], os.listdir(os.path.join(book_cache_dir(), 'f')))
+            with cache_lock() as f:
+                metadata = json.loads(f.read())
+                self.assertEqual(metadata['entries'], {})
+
             # Test updating cached book
             book_src = os.path.join(self.tdir, 'book2.epub')
             open(book_src, 'wb').write(b'bb')