Have RepositoryInterface return all potential matches

5 files changed, 36 insertions, 18 deletions

diff --git a/obnamlib/plugins/backup_plugin.py b/obnamlib/plugins/backup_plugin.py
index ae1c3709..1e39dc5a 100644
--- a/obnamlib/plugins/backup_plugin.py
+++ b/obnamlib/plugins/backup_plugin.py
@@ -761,7 +761,7 @@ class BackupPlugin(obnamlib.ObnamPlugin):
             # exceptions, and other errors. We don't care: we'll just
             # pretend no chunk with the checksum exists yet.
             try:
-                in_tree = [self.repo.find_chunk_id_by_content(data)]
+                in_tree = self.repo.find_chunk_ids_by_content(data)
             except larch.Error:
                 in_tree = []
             except obnamlib.RepositoryChunkContentNotInIndexes:
diff --git a/obnamlib/repo_dummy.py b/obnamlib/repo_dummy.py
index 4f9b462a..a6df73d8 100644
--- a/obnamlib/repo_dummy.py
+++ b/obnamlib/repo_dummy.py
@@ -449,10 +449,13 @@ class ChunkIndexes(object):
         self.data.set_value(chunk_id, token_is_chunk_content)
 
     def find_chunk(self, chunk_content):
+        chunk_ids = []
         for chunk_id, stored_content in self.data.items():
             if stored_content == chunk_content:
-                return chunk_id
-        raise obnamlib.RepositoryChunkContentNotInIndexes()
+                chunk_ids.append(chunk_id)
+        if not chunk_ids:
+            raise obnamlib.RepositoryChunkContentNotInIndexes()
+        return chunk_ids
 
     def remove_chunk(self, chunk_id, client_id):
         self._require_lock()
@@ -659,7 +662,7 @@ class RepositoryFormatDummy(obnamlib.RepositoryInterface):
     def put_chunk_into_indexes(self, chunk_id, token, client_id):
         self._chunk_indexes.put_chunk(chunk_id, token, client_id)
 
-    def find_chunk_id_by_content(self, chunk_content):
+    def find_chunk_ids_by_content(self, chunk_content):
         return self._chunk_indexes.find_chunk(chunk_content)
 
     def remove_chunk_from_indexes(self, chunk_id, client_id):
diff --git a/obnamlib/repo_fmt_6.py b/obnamlib/repo_fmt_6.py
index e40f2c75..e6c75173 100644
--- a/obnamlib/repo_fmt_6.py
+++ b/obnamlib/repo_fmt_6.py
@@ -674,13 +674,11 @@ class RepositoryFormat6(obnamlib.RepositoryInterface):
         self._chunksums.remove(checksum, chunk_id, client_id)
         self._chunklist.remove(chunk_id)
 
-    def find_chunk_id_by_content(self, data):
+    def find_chunk_ids_by_content(self, data):
         checksum = self._checksum(data)
         candidates = self._chunksums.find(checksum)
-        for chunk_id in candidates:
-            chunk_data = self.get_chunk_content(chunk_id)
-            if chunk_data == data:
-                return chunk_id
+        if candidates:
+            return candidates
         raise obnamlib.RepositoryChunkContentNotInIndexes()
 
     def validate_chunk_content(self, chunk_id):
diff --git a/obnamlib/repo_interface.py b/obnamlib/repo_interface.py
index b20a6f59..f4ab7531 100644
--- a/obnamlib/repo_interface.py
+++ b/obnamlib/repo_interface.py
@@ -722,13 +722,15 @@ class RepositoryInterface(object):
         '''Removes a chunk from indexes, given its id, for a given client.'''
         raise NotImplementedError()
 
-    def find_chunk_id_by_content(self, data):
-        '''Finds a chunk id given its content.
+    def find_chunk_ids_by_content(self, data):
+        '''Finds chunk ids that probably match a given content.
 
         This will raise RepositoryChunkContentNotInIndexes if the
-        chunk is not in the indexes. Otherwise it will return one
-        chunk id that has exactly the same content. If the indexes
-        contain duplicate chunks, any one of the might be returned.
+        chunk is not in the indexes. Otherwise it will return all
+        chunk ids that would have the same token (see
+        prepare_chunk_for_indexes). Note that the chunks whose ids are
+        returned do not necessarily match the given data; if the
+        caller cares, they need to verify.
 
         '''
         raise NotImplementedError()
@@ -1738,7 +1740,22 @@ class RepositoryInterfaceTests(unittest.TestCase): # pragma: no cover
         token = self.repo.prepare_chunk_for_indexes('foochunk')
         self.repo.put_chunk_into_indexes(chunk_id, token, 'fooclient')
         self.assertEqual(
-            self.repo.find_chunk_id_by_content('foochunk'), chunk_id)
+            self.repo.find_chunk_ids_by_content('foochunk'), [chunk_id])
+
+    def test_finds_all_matching_chunk_ids(self):
+        self.setup_client()
+        token = self.repo.prepare_chunk_for_indexes('foochunk')
+        self.repo.lock_chunk_indexes()
+
+        chunk_id_1 = self.repo.put_chunk_content('foochunk')
+        self.repo.put_chunk_into_indexes(chunk_id_1, token, 'fooclient')
+
+        chunk_id_2 = self.repo.put_chunk_content('foochunk')
+        self.repo.put_chunk_into_indexes(chunk_id_2, token, 'fooclient')
+
+        self.assertEqual(
+            set(self.repo.find_chunk_ids_by_content('foochunk')),
+            set([chunk_id_1, chunk_id_2]))
 
     def test_removes_chunk_from_indexes(self):
         self.setup_client()
@@ -1749,7 +1766,7 @@ class RepositoryInterfaceTests(unittest.TestCase): # pragma: no cover
         self.repo.remove_chunk_from_indexes(chunk_id, 'fooclient')
         self.assertRaises(
             obnamlib.RepositoryChunkContentNotInIndexes,
-            self.repo.find_chunk_id_by_content, 'foochunk')
+            self.repo.find_chunk_ids_by_content, 'foochunk')
 
     def test_putting_chunk_to_indexes_without_locking_them_fails(self):
         chunk_id = self.repo.put_chunk_content('foochunk')
@@ -1779,7 +1796,7 @@ class RepositoryInterfaceTests(unittest.TestCase): # pragma: no cover
         self.repo.unlock_chunk_indexes()
         self.assertRaises(
             obnamlib.RepositoryChunkContentNotInIndexes,
-            self.repo.find_chunk_id_by_content, 'foochunk')
+            self.repo.find_chunk_ids_by_content, 'foochunk')
 
     def test_committing_chunk_indexes_remembers_changes(self):
         self.setup_client()
@@ -1789,7 +1806,7 @@ class RepositoryInterfaceTests(unittest.TestCase): # pragma: no cover
         self.repo.put_chunk_into_indexes(chunk_id, token, 'fooclient')
         self.repo.commit_chunk_indexes()
         self.assertEqual(
-            self.repo.find_chunk_id_by_content('foochunk'), chunk_id)
+            self.repo.find_chunk_ids_by_content('foochunk'), [chunk_id])
 
     def test_locking_chunk_indexes_twice_fails(self):
         self.repo.lock_chunk_indexes()
diff --git a/test-gpghome/random_seed b/test-gpghome/random_seed
index 273edb2a..e03c7d57 100644
--- a/test-gpghome/random_seed
+++ b/test-gpghome/random_seed