SERVER-43304 Modernize mr1.js

This reverts commit be1be55d991f7bed4d3896f5be53afbb4e579729.

12 files changed, 206 insertions, 213 deletions

diff --git a/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_100ms_refresh_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_100ms_refresh_jscore_passthrough.yml
index 66bb144b1e8..40c941c36b0 100644
--- a/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_100ms_refresh_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_100ms_refresh_jscore_passthrough.yml
@@ -91,7 +91,7 @@ selector:
   - jstests/core/sort_array.js
   # TODO: SERVER-16605
   - jstests/core/mr_index.js
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
   - jstests/core/mr5.js
diff --git a/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_10sec_refresh_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_10sec_refresh_jscore_passthrough.yml
index 4c8be72d6d8..6a5c3202daf 100644
--- a/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_10sec_refresh_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_10sec_refresh_jscore_passthrough.yml
@@ -91,7 +91,7 @@ selector:
   - jstests/core/sort_array.js
   # TODO: SERVER-16605
   - jstests/core/mr_index.js
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
   - jstests/core/mr5.js
diff --git a/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_1sec_refresh_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_1sec_refresh_jscore_passthrough.yml
index b102942e52c..a75bc546375 100644
--- a/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_1sec_refresh_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_1sec_refresh_jscore_passthrough.yml
@@ -91,7 +91,7 @@ selector:
   - jstests/core/sort_array.js
   # TODO: SERVER-16605
   - jstests/core/mr_index.js
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
   - jstests/core/mr5.js
diff --git a/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_default_refresh_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_default_refresh_jscore_passthrough.yml
index 89ea274bd8b..4ec3bf5be73 100644
--- a/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_default_refresh_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/logical_session_cache_sharding_default_refresh_jscore_passthrough.yml
@@ -91,7 +91,7 @@ selector:
   - jstests/core/sort_array.js
   # TODO: SERVER-16605
   - jstests/core/mr_index.js
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
   - jstests/core/mr5.js
diff --git a/buildscripts/resmokeconfig/suites/multi_shard_local_read_write_multi_stmt_txn_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/multi_shard_local_read_write_multi_stmt_txn_jscore_passthrough.yml
index 1072d1c9570..50ec6ff4690 100644
--- a/buildscripts/resmokeconfig/suites/multi_shard_local_read_write_multi_stmt_txn_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/multi_shard_local_read_write_multi_stmt_txn_jscore_passthrough.yml
@@ -85,7 +85,7 @@ selector:
   - jstests/core/optimized_match_explain.js
   - jstests/core/sort_array.js
   # TODO: SERVER-16605 MapReduce into sharded collection with hashed index fails
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
 
diff --git a/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_jscore_passthrough.yml
index 2593d8ffc0b..207f84cd928 100644
--- a/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_jscore_passthrough.yml
@@ -85,7 +85,7 @@ selector:
   - jstests/core/optimized_match_explain.js
   - jstests/core/sort_array.js
   # TODO: SERVER-16605 MapReduce into sharded collection with hashed index fails
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
 
diff --git a/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_kill_primary_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_kill_primary_jscore_passthrough.yml
index 5a154c63cc4..b1983e9082f 100644
--- a/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_kill_primary_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_kill_primary_jscore_passthrough.yml
@@ -85,7 +85,7 @@ selector:
   - jstests/core/optimized_match_explain.js
   - jstests/core/sort_array.js
   # TODO: SERVER-16605 MapReduce into sharded collection with hashed index fails
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
 
diff --git a/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_stepdown_primary_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_stepdown_primary_jscore_passthrough.yml
index 2cca2cfbcdd..8850efcc8d0 100644
--- a/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_stepdown_primary_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/multi_shard_multi_stmt_txn_stepdown_primary_jscore_passthrough.yml
@@ -85,7 +85,7 @@ selector:
   - jstests/core/optimized_match_explain.js
   - jstests/core/sort_array.js
   # TODO: SERVER-16605 MapReduce into sharded collection with hashed index fails
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
 
diff --git a/buildscripts/resmokeconfig/suites/multi_stmt_txn_jscore_passthrough_with_migration.yml b/buildscripts/resmokeconfig/suites/multi_stmt_txn_jscore_passthrough_with_migration.yml
index f3a6c31672b..fd2921b5ef7 100644
--- a/buildscripts/resmokeconfig/suites/multi_stmt_txn_jscore_passthrough_with_migration.yml
+++ b/buildscripts/resmokeconfig/suites/multi_stmt_txn_jscore_passthrough_with_migration.yml
@@ -85,7 +85,7 @@ selector:
   - jstests/core/optimized_match_explain.js
   - jstests/core/sort_array.js
   # TODO: SERVER-16605 MapReduce into sharded collection with hashed index fails
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
   # Create index command arbitrarily picks one of the errors if there are multiple errors from the
diff --git a/buildscripts/resmokeconfig/suites/sharded_causally_consistent_jscore_passthrough.yml b/buildscripts/resmokeconfig/suites/sharded_causally_consistent_jscore_passthrough.yml
index 8cee5e28e21..30a3f185a2a 100644
--- a/buildscripts/resmokeconfig/suites/sharded_causally_consistent_jscore_passthrough.yml
+++ b/buildscripts/resmokeconfig/suites/sharded_causally_consistent_jscore_passthrough.yml
@@ -63,7 +63,7 @@ selector:
   # collection is sharded on anything other than {_id: 1}.
   - jstests/core/mr_undef.js
   # TODO: SERVER-16605
-  - jstests/core/mr1.js
+  - jstests/core/mr_correctness.js
   - jstests/core/mr3.js
   - jstests/core/mr4.js
   # TODO: SERVER-30052
diff --git a/jstests/core/mr1.js b/jstests/core/mr1.js
deleted file mode 100644
index d43b6b8303a..00000000000
--- a/jstests/core/mr1.js
+++ /dev/null
@@ -1,203 +0,0 @@
-// @tags: [
-//   # mapReduce does not support afterClusterTime.
-//   does_not_support_causal_consistency,
-//   does_not_support_stepdowns,
-//   requires_fastcount,
-//   requires_getmore,
-//   uses_map_reduce_with_temp_collections,
-// ]
-
-t = db.mr1;
-t.drop();
-
-t.save({x: 1, tags: ["a", "b"]});
-t.save({x: 2, tags: ["b", "c"]});
-t.save({x: 3, tags: ["c", "a"]});
-t.save({x: 4, tags: ["b", "c"]});
-
-emit = printjson;
-
-function d(x) {
-    printjson(x);
-}
-
-ks = "_id";
-if (db.version() == "1.1.1")
-    ks = "key";
-
-m = function() {
-    this.tags.forEach(function(z) {
-        emit(z, {count: 1});
-    });
-};
-
-m2 = function() {
-    for (var i = 0; i < this.tags.length; i++) {
-        emit(this.tags[i], 1);
-    }
-};
-
-r = function(key, values) {
-    var total = 0;
-    for (var i = 0; i < values.length; i++) {
-        total += values[i].count;
-    }
-    return {count: total};
-};
-
-r2 = function(key, values) {
-    var total = 0;
-    for (var i = 0; i < values.length; i++) {
-        total += values[i];
-    }
-    return total;
-};
-
-res = db.runCommand({mapreduce: "mr1", map: m, reduce: r, out: "mr1_out"});
-d(res);
-if (ks == "_id")
-    assert(res.ok, "not ok");
-assert.eq(4, res.counts.input, "A");
-x = db[res.result];
-
-assert.eq(3, x.find().count(), "B");
-x.find().forEach(d);
-z = {};
-x.find().forEach(function(a) {
-    z[a[ks]] = a.value.count;
-});
-d(z);
-assert.eq(3, Object.keySet(z).length, "C");
-assert.eq(2, z.a, "D");
-assert.eq(3, z.b, "E");
-assert.eq(3, z.c, "F");
-x.drop();
-
-res = db.runCommand({mapreduce: "mr1", map: m, reduce: r, query: {x: {"$gt": 2}}, out: "mr1_out"});
-d(res);
-assert.eq(2, res.counts.input, "B");
-x = db[res.result];
-z = {};
-x.find().forEach(function(a) {
-    z[a[ks]] = a.value.count;
-});
-assert.eq(1, z.a, "C1");
-assert.eq(1, z.b, "C2");
-assert.eq(2, z.c, "C3");
-x.drop();
-
-res =
-    db.runCommand({mapreduce: "mr1", map: m2, reduce: r2, query: {x: {"$gt": 2}}, out: "mr1_out"});
-d(res);
-assert.eq(2, res.counts.input, "B");
-x = db[res.result];
-z = {};
-x.find().forEach(function(a) {
-    z[a[ks]] = a.value;
-});
-assert.eq(1, z.a, "C1z");
-assert.eq(1, z.b, "C2z");
-assert.eq(2, z.c, "C3z");
-x.drop();
-
-res = db.runCommand({mapreduce: "mr1", out: "mr1_foo", map: m, reduce: r, query: {x: {"$gt": 2}}});
-d(res);
-assert.eq(2, res.counts.input, "B2");
-assert.eq("mr1_foo", res.result, "B2-c");
-x = db[res.result];
-z = {};
-x.find().forEach(function(a) {
-    z[a[ks]] = a.value.count;
-});
-assert.eq(1, z.a, "C1a");
-assert.eq(1, z.b, "C2a");
-assert.eq(2, z.c, "C3a");
-x.drop();
-
-for (i = 5; i < 1000; i++) {
-    t.save({x: i, tags: ["b", "d"]});
-}
-
-res = db.runCommand({mapreduce: "mr1", map: m, reduce: r, out: "mr1_out"});
-d(res);
-assert.eq(999, res.counts.input, "Z1");
-x = db[res.result];
-x.find().forEach(d);
-assert.eq(4, x.find().count(), "Z2");
-assert.eq("a,b,c,d", x.distinct(ks), "Z3");
-
-function getk(k) {
-    var o = {};
-    o[ks] = k;
-    return x.findOne(o);
-}
-
-assert.eq(2, getk("a").value.count, "ZA");
-assert.eq(998, getk("b").value.count, "ZB");
-assert.eq(3, getk("c").value.count, "ZC");
-assert.eq(995, getk("d").value.count, "ZD");
-x.drop();
-
-if (true) {
-    printjson(db.runCommand({mapreduce: "mr1", map: m, reduce: r, verbose: true, out: "mr1_out"}));
-}
-
-print("t1: " + Date.timeFunc(function() {
-    var out = db.runCommand({mapreduce: "mr1", map: m, reduce: r, out: "mr1_out"});
-    if (ks == "_id")
-        assert(out.ok, "XXX : " + tojson(out));
-    db[out.result].drop();
-}, 10) + " (~500 on 2.8ghz) - itcount: " + Date.timeFunc(function() {
-    db.mr1.find().itcount();
-}, 10));
-
-// test doesn't exist
-res =
-    db.runCommand({mapreduce: "lasjdlasjdlasjdjasldjalsdj12e", map: m, reduce: r, out: "mr1_out"});
-assert(!res.ok, "should be not ok");
-
-if (true) {
-    correct = {};
-
-    var bulk = t.initializeUnorderedBulkOp();
-    for (i = 0; i < 20000; i++) {
-        k = "Z" + i % 10000;
-        if (correct[k])
-            correct[k]++;
-        else
-            correct[k] = 1;
-        bulk.insert({x: i, tags: [k]});
-    }
-    assert.commandWorked(bulk.execute());
-
-    res = db.runCommand({mapreduce: "mr1", out: "mr1_foo", map: m, reduce: r});
-    d(res);
-    print("t2: " + res.timeMillis + " (~3500 on 2.8ghz) - itcount: " + Date.timeFunc(function() {
-        db.mr1.find().itcount();
-    }));
-    x = db[res.result];
-    z = {};
-    x.find().forEach(function(a) {
-        z[a[ks]] = a.value.count;
-    });
-    for (zz in z) {
-        if (zz.indexOf("Z") == 0) {
-            assert.eq(correct[zz], z[zz], "ZZ : " + zz);
-        }
-    }
-    x.drop();
-
-    res = db.runCommand({mapreduce: "mr1", out: "mr1_out", map: m2, reduce: r2});
-    d(res);
-    print("t3: " + res.timeMillis + " (~3500 on 2.8ghz)");
-
-    res = db.runCommand({mapreduce: "mr1", map: m2, reduce: r2, out: {inline: true}});
-    print("t4: " + res.timeMillis);
-}
-
-res = db.runCommand({mapreduce: "mr1", map: m, reduce: r, out: "mr1_out"});
-assert(res.ok, "should be ok");
-
-t.drop();
-t1 = db.mr1_out;
-t1.drop();
diff --git a/jstests/core/mr_correctness.js b/jstests/core/mr_correctness.js
new file mode 100644
index 00000000000..466e5bc1b32
--- /dev/null
+++ b/jstests/core/mr_correctness.js
@@ -0,0 +1,196 @@
+// Basic correctness tests for the mapReduce command.
+// @tags: [
+//   # mapReduce does not support afterClusterTime.
+//   does_not_support_causal_consistency,
+//   does_not_support_stepdowns,
+//   requires_fastcount,
+//   requires_getmore,
+//   uses_map_reduce_with_temp_collections,
+// ]
+(function() {
+"use strict";
+load("jstests/libs/fixture_helpers.js");  // For "FixtureHelpers".
+
+const coll = db.mr_correctness;
+coll.drop();
+
+assert.commandWorked(coll.insert({x: 1, tags: ["a", "b"]}));
+assert.commandWorked(coll.insert({x: 2, tags: ["b", "c"]}));
+assert.commandWorked(coll.insert({x: 3, tags: ["c", "a"]}));
+assert.commandWorked(coll.insert({x: 4, tags: ["b", "c"]}));
+
+function mapToObj() {
+    this.tags.forEach(function(z) {
+        emit(z, {count: 1});
+    });
+}
+
+function reduceObjs(key, values) {
+    let total = 0;
+    for (let i = 0; i < values.length; i++) {
+        total += values[i].count;
+    }
+    return {count: total};
+}
+
+const outColl = db[coll.getName() + "_out"];
+outColl.drop();
+(function testBasicMapReduce() {
+    const res = db.runCommand(
+        {mapReduce: coll.getName(), map: mapToObj, reduce: reduceObjs, out: outColl.getName()});
+    assert.commandWorked(res);
+    assert.eq(4, res.counts.input);
+    assert.eq(res.result, outColl.getName());
+
+    assert.eq(
+        3,
+        outColl.find().count(),
+        () =>
+            `expected 3 distinct tags: ['a', 'b', 'c'], found ${tojson(outColl.find().toArray())}`);
+    const keys = {};
+    for (let result of outColl.find().toArray()) {
+        keys[result._id] = result.value.count;
+    }
+    assert.eq(3, Object.keySet(keys).length, Object.keySet(keys));
+    assert.eq(2, keys.a, () => `Expected 2 occurences of the tag 'a': ${tojson(keys)}`);
+    assert.eq(3, keys.b, () => `Expected 3 occurences of the tag 'b': ${tojson(keys)}`);
+    assert.eq(3, keys.c, () => `Expected 3 occurences of the tag 'c': ${tojson(keys)}`);
+    outColl.drop();
+})();
+
+(function testMapReduceWithPredicate() {
+    const res = db.runCommand({
+        mapReduce: coll.getName(),
+        map: mapToObj,
+        reduce: reduceObjs,
+        query: {x: {$gt: 2}},
+        out: outColl.getName()
+    });
+    assert.commandWorked(res);
+    assert.eq(2, res.counts.input, () => tojson(res));
+    assert.eq(res.result, outColl.getName());
+    const keys = {};
+    for (let result of outColl.find().toArray()) {
+        keys[result._id] = result.value.count;
+    }
+    assert.eq(3, Object.keySet(keys).length, Object.keySet(keys));
+    assert.eq(1, keys.a, () => `Expected 1 occurence of the tag 'a': ${tojson(keys)}`);
+    assert.eq(1, keys.b, () => `Expected 1 occurence of the tag 'b': ${tojson(keys)}`);
+    assert.eq(2, keys.c, () => `Expected 2 occurences of the tag 'c': ${tojson(keys)}`);
+    outColl.drop();
+}());
+
+function mapToNumber() {
+    for (let tag of this.tags) {
+        emit(tag, 1);
+    }
+}
+
+function reduceNumbers(key, values) {
+    let total = 0;
+    for (let val of values) {
+        total += val;
+    }
+    return total;
+}
+
+// Now do a similar test but using the above map and reduce functions which use numbers as the value
+// instead of objects.
+(function testBasicMapReduceWithNumberValues() {
+    const res = db.runCommand({
+        mapReduce: coll.getName(),
+        map: mapToNumber,
+        reduce: reduceNumbers,
+        query: {x: {$gt: 2}},
+        out: outColl.getName()
+    });
+    assert.commandWorked(res);
+    assert.eq(2, res.counts.input, () => tojson(res));
+    assert.eq(res.result, outColl.getName());
+    const keys = {};
+    for (let result of outColl.find().toArray()) {
+        keys[result._id] = result.value;
+    }
+    assert.eq(3, Object.keySet(keys).length, Object.keySet(keys));
+    assert.eq(1, keys.a, () => `Expected 1 occurence of the tag 'a': ${tojson(keys)}`);
+    assert.eq(1, keys.b, () => `Expected 1 occurence of the tag 'b': ${tojson(keys)}`);
+    assert.eq(2, keys.c, () => `Expected 2 occurences of the tag 'c': ${tojson(keys)}`);
+    outColl.drop();
+}());
+
+(function testMapReduceWithManyValuesGrouped() {
+    const bulk = coll.initializeUnorderedBulkOp();
+    for (let i = 5; i < 1000; i++) {
+        bulk.insert({x: i, tags: ["b", "d"]});
+    }
+    assert.commandWorked(bulk.execute());
+
+    const res = db.runCommand(
+        {mapReduce: coll.getName(), map: mapToObj, reduce: reduceObjs, out: outColl.getName()});
+    assert.commandWorked(res);
+    assert.eq(999, res.counts.input, () => tojson(res));
+    assert.eq(res.result, outColl.getName());
+    assert.eq(4,
+              outColl.find().count(),
+              () => `expected 4 distinct tags: ['a', 'b', 'c', 'd'], found ${
+                  tojson(outColl.find().toArray())}`);
+    assert.eq("a,b,c,d", outColl.distinct("_id"));
+
+    assert.eq(2, outColl.findOne({_id: "a"}).value.count, () => outColl.findOne({_id: "a"}));
+    assert.eq(998, outColl.findOne({_id: "b"}).value.count, () => outColl.findOne({_id: "b"}));
+    assert.eq(3, outColl.findOne({_id: "c"}).value.count, () => outColl.findOne({_id: "c"}));
+    assert.eq(995, outColl.findOne({_id: "d"}).value.count, () => outColl.findOne({_id: "d"}));
+    outColl.drop();
+}());
+
+(function testThatVerboseOptionIncludesTimingInformation() {
+    const cmd =
+        {mapReduce: coll.getName(), map: mapToObj, reduce: reduceObjs, out: outColl.getName()};
+    const withoutVerbose = assert.commandWorked(db.runCommand(cmd));
+    // TODO SERVER-43290 The verbose option should have the same effect on mongos.
+    assert(FixtureHelpers.isMongos(db) || !withoutVerbose.hasOwnProperty("timing"));
+    const withVerbose = assert.commandWorked(db.runCommand(Object.merge(cmd, {verbose: true})));
+    assert(withVerbose.hasOwnProperty("timing"));
+}());
+
+(function testMapReduceAgainstNonExistentCollection() {
+    assert.commandFailedWithCode(db.runCommand({
+        mapReduce: "lasjdlasjdlasjdjasldjalsdj12e",
+        map: mapToObj,
+        reduce: reduceObjs,
+        out: outColl.getName()
+    }),
+                                 ErrorCodes.NamespaceNotFound);
+}());
+
+(function testHighCardinalityKeySet() {
+    let correctValues = {};
+
+    coll.drop();
+    const bulk = coll.initializeUnorderedBulkOp();
+    for (let i = 0; i < 20000; i++) {
+        const tag = "Z" + i % 10000;
+        if (!correctValues[tag])
+            correctValues[tag] = 1;
+        else
+            correctValues[tag]++;
+        bulk.insert({x: i, tags: [tag]});
+    }
+    assert.commandWorked(bulk.execute());
+
+    const res = db.runCommand(
+        {mapReduce: coll.getName(), out: outColl.getName(), map: mapToObj, reduce: reduceObjs});
+    assert.commandWorked(res);
+    assert.eq(res.result, outColl.getName());
+    let actualValues = {};
+    outColl.find().forEach(function(resultDoc) {
+        actualValues[resultDoc._id] = resultDoc.value.count;
+    });
+    for (let key in actualValues) {
+        assert.eq(correctValues[key], actualValues[key], key);
+    }
+}());
+
+coll.drop();
+outColl.drop();
+}());