Fixes COUCHDB-1265

Backport of 1164350 from trunk. Slightly modified for an export 
declaration conflict and removing a clause that only applies to
trunk.



git-svn-id: https://svn.apache.org/repos/asf/couchdb/branches/1.1.x@1164351 13f79535-47bb-0310-9956-ffa450edef68

3 files changed, 68 insertions, 6 deletions

diff --git a/share/www/script/test/recreate_doc.js b/share/www/script/test/recreate_doc.js
index 05843558..a1cfb8f8 100644
--- a/share/www/script/test/recreate_doc.js
+++ b/share/www/script/test/recreate_doc.js
@@ -77,4 +77,45 @@ couchTests.recreate_doc = function(debug) {
   } catch (e) {
     T(e.error == "conflict");
   }
+
+  db.deleteDb();
+  db.createDb();
+
+  // COUCHDB-1265
+  // Resuscitate an unavailable old revision and make sure that it
+  // doesn't introduce duplicates into the _changes feed.
+  
+  var doc = {_id: "bar", count: 0};
+  T(db.save(doc).ok);
+  var ghost = {_id: "bar", _rev: doc._rev, count: doc.count};
+  for(var i = 0; i < 2; i++) {
+    doc.count += 1;
+    T(db.save(doc).ok);
+  }
+
+  // Compact so that the old revision to be resuscitated will be
+  // in the rev_tree as ?REV_MISSING
+  db.compact();
+  while(db.info().compact_running) {}
+
+  // Saving the ghost here puts it back in the rev_tree in such
+  // a way as to create a new update_seq but without changing a
+  // leaf revision. This would cause the #full_doc_info{} and
+  // #doc_info{} records to diverge in their idea of what the
+  // doc's update_seq is and end up introducing a duplicate in
+  // the _changes feed the next time this doc is updated.
+  T(db.save(ghost, {new_edits: false}).ok);
+
+  // The duplicate would have been introduce here becuase the #doc_info{}
+  // would not have been removed correctly.
+  T(db.save(doc).ok);
+
+  // And finally assert that there are no duplicates in _changes.
+  var req = CouchDB.request("GET", "/test_suite_db/_changes");
+  var resp = JSON.parse(req.responseText);
+  var docids = {};
+  for(var i = 0; i < resp.results.length; i++) {
+    T(docids[resp.results[i].id] === undefined, "Duplicates in _changes feed.");
+    docids[resp.results[i].id] = true;
+  }
 };
diff --git a/src/couchdb/couch_doc.erl b/src/couchdb/couch_doc.erl
index 531eb6bb..a6700d59 100644
--- a/src/couchdb/couch_doc.erl
+++ b/src/couchdb/couch_doc.erl
@@ -302,10 +302,16 @@ to_doc_info(FullDocInfo) ->
     {DocInfo, _Path} = to_doc_info_path(FullDocInfo),
     DocInfo.
 
-max_seq([], Max) ->
-    Max;
-max_seq([#rev_info{seq=Seq}|Rest], Max) ->
-    max_seq(Rest, if Max > Seq -> Max; true -> Seq end).
+max_seq(Tree) ->
+    FoldFun = fun({_Pos, _Key}, Value, _Type, MaxOldSeq) ->
+        case Value of
+            {_Deleted, _DiskPos, OldTreeSeq} ->
+                erlang:max(MaxOldSeq, OldTreeSeq);
+            _ ->
+                MaxOldSeq
+        end
+    end,
+    couch_key_tree:fold(FoldFun, 0, Tree).
 
 to_doc_info_path(#full_doc_info{id=Id,rev_tree=Tree}) ->
     RevInfosAndPath =
@@ -320,7 +326,7 @@ to_doc_info_path(#full_doc_info{id=Id,rev_tree=Tree}) ->
         end, RevInfosAndPath),
     [{_RevInfo, WinPath}|_] = SortedRevInfosAndPath,
     RevInfos = [RevInfo || {RevInfo, _Path} <- SortedRevInfosAndPath],
-    {#doc_info{id=Id, high_seq=max_seq(RevInfos, 0), revs=RevInfos}, WinPath}.
+    {#doc_info{id=Id, high_seq=max_seq(Tree), revs=RevInfos}, WinPath}.
 
 
 
diff --git a/src/couchdb/couch_key_tree.erl b/src/couchdb/couch_key_tree.erl
index 48a76b1d..367c9e33 100644
--- a/src/couchdb/couch_key_tree.erl
+++ b/src/couchdb/couch_key_tree.erl
@@ -49,7 +49,7 @@
 
 -export([merge/3, find_missing/2, get_key_leafs/2, get_full_key_paths/2, get/2]).
 -export([map/2, get_all_leafs/1, count_leafs/1, remove_leafs/2,
-    get_all_leafs_full/1,stem/2,map_leafs/2]).
+    get_all_leafs_full/1,stem/2,map_leafs/2, fold/3]).
 
 -include("couch_db.hrl").
 
@@ -325,6 +325,21 @@ count_leafs_simple([{_Key, _Value, SubTree} | RestTree]) ->
     count_leafs_simple(SubTree) + count_leafs_simple(RestTree).
 
 
+fold(_Fun, Acc, []) ->
+    Acc;
+fold(Fun, Acc0, [{Pos, Tree}|Rest]) ->
+    Acc1 = fold_simple(Fun, Acc0, Pos, [Tree]),
+    fold(Fun, Acc1, Rest).
+
+fold_simple(_Fun, Acc, _Pos, []) ->
+    Acc;
+fold_simple(Fun, Acc0, Pos, [{Key, Value, SubTree} | RestTree]) ->
+    Type = if SubTree == [] -> leaf; true -> branch end,
+    Acc1 = Fun({Pos, Key}, Value, Type, Acc0),
+    Acc2 = fold_simple(Fun, Acc1, Pos+1, SubTree),
+    fold_simple(Fun, Acc2, Pos, RestTree).
+
+
 map(_Fun, []) ->
     [];
 map(Fun, [{Pos, Tree}|Rest]) ->