[couchdb] 01/10: introduce mem3_util:docid_hash/1 and docid_hash/2

[rn...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

rnewson pushed a commit to branch user-partitioned-dbs-wip
in repository https://gitbox.apache.org/repos/asf/couchdb.git

commit 2b63513f50a364a196e3e5ba9ceb8b325cfd4517
Author: Robert Newson <rn...@apache.org>
AuthorDate: Wed Jul 4 17:15:16 2018 +0100

    introduce mem3_util:docid_hash/1 and docid_hash/2
    
    mem3_util:docid_hash/1 is identical to mem3_util:hash/1
    
    mem3_util:docid_hash/2 allows the user to control the result with
    specially-formatted doc ids.
---
 src/mem3/src/mem3.erl        |  2 +-
 src/mem3/src/mem3_shards.erl |  4 ++--
 src/mem3/src/mem3_util.erl   | 15 +++++++++++++++
 3 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/src/mem3/src/mem3.erl b/src/mem3/src/mem3.erl
index 0e5eabf..ca083c4 100644
--- a/src/mem3/src/mem3.erl
+++ b/src/mem3/src/mem3.erl
@@ -239,7 +239,7 @@ belongs(DbName, DocId) when is_binary(DbName), is_binary(DocId) ->
     true.
 
 belongs(Begin, End, DocId) ->
-    HashKey = mem3_util:hash(DocId),
+    HashKey = mem3_util:docid_hash(DocId),
     Begin =< HashKey andalso HashKey =< End.
 
 range(#shard{range = Range}) ->
diff --git a/src/mem3/src/mem3_shards.erl b/src/mem3/src/mem3_shards.erl
index da3b69a..0559149 100644
--- a/src/mem3/src/mem3_shards.erl
+++ b/src/mem3/src/mem3_shards.erl
@@ -67,7 +67,7 @@ for_docid(DbName, DocId) ->
     for_docid(DbName, DocId, []).
 
 for_docid(DbName, DocId, Options) ->
-    HashKey = mem3_util:hash(DocId),
+    HashKey = mem3_util:docid_hash(DocId),
     ShardHead = #shard{
         dbname = DbName,
         range = ['$1', '$2'],
@@ -397,7 +397,7 @@ load_shards_from_db(ShardDb, DbName) ->
 
 load_shards_from_disk(DbName, DocId)->
     Shards = load_shards_from_disk(DbName),
-    HashKey = mem3_util:hash(DocId),
+    HashKey = mem3_util:docid_hash(DocId),
     [S || S <- Shards, in_range(S, HashKey)].
 
 in_range(Shard, HashKey) ->
diff --git a/src/mem3/src/mem3_util.erl b/src/mem3/src/mem3_util.erl
index 0b69d79..fc6123d 100644
--- a/src/mem3/src/mem3_util.erl
+++ b/src/mem3/src/mem3_util.erl
@@ -16,6 +16,7 @@
     n_val/2, to_atom/1, to_integer/1, write_db_doc/1, delete_db_doc/1,
     shard_info/1, ensure_exists/1, open_db_doc/1]).
 -export([is_deleted/1, rotate_list/2]).
+-export([docid_hash/1, docid_hash/2]).
 
 %% do not use outside mem3.
 -export([build_ordered_shards/2, downcast/1]).
@@ -34,6 +35,20 @@ hash(Item) when is_binary(Item) ->
 hash(Item) ->
     erlang:crc32(term_to_binary(Item)).
 
+
+docid_hash(DocId) when is_binary(DocId) ->
+    docid_hash(DocId, []).
+
+docid_hash(DocId, Options) when is_binary(DocId), is_list(Options) ->
+    Data = case lists:member(partitioned, Options) of
+        true ->
+            hd(binary:split(DocId, <<":">>));
+        false ->
+            DocId
+    end,
+    erlang:crc32(Data).
+
+
 name_shard(Shard) ->
     name_shard(Shard, "").