SERVER-25038 add LiteParsedPipeline

This provides a way to do pre-parse validity checks. Full
parsing of the Pipeline must be done under the collection
lock, when the collation is known.

6 files changed, 92 insertions, 14 deletions

diff --git a/jstests/aggregation/bugs/server19095.js b/jstests/aggregation/bugs/server19095.js
index 875a7d16a9d..e55539cedc7 100644
--- a/jstests/aggregation/bugs/server19095.js
+++ b/jstests/aggregation/bugs/server19095.js
@@ -375,7 +375,7 @@ load("jstests/aggregation/extras/utils.js");
         // All four fields must be specified.
         assertErrorCode(coll, [{$lookup: {foreignField: "b", from: "from", as: "same"}}], 4572);
         assertErrorCode(coll, [{$lookup: {localField: "a", from: "from", as: "same"}}], 4572);
-        assertErrorCode(coll, [{$lookup: {localField: "a", foreignField: "b", as: "same"}}], 4572);
+        assertErrorCode(coll, [{$lookup: {localField: "a", foreignField: "b", as: "same"}}], 40320);
         assertErrorCode(
             coll, [{$lookup: {localField: "a", foreignField: "b", from: "from"}}], 4572);
 
@@ -385,12 +385,15 @@ load("jstests/aggregation/extras/utils.js");
         assertErrorCode(
             coll, [{$lookup: {localField: "a", foreignField: 1, from: "from", as: "as"}}], 4570);
         assertErrorCode(
-            coll, [{$lookup: {localField: "a", foreignField: "b", from: 1, as: "as"}}], 4570);
+            coll, [{$lookup: {localField: "a", foreignField: "b", from: 1, as: "as"}}], 40321);
         assertErrorCode(
             coll, [{$lookup: {localField: "a", foreignField: "b", from: "from", as: 1}}], 4570);
 
+        // The foreign collection must be a valid namespace.
+        assertErrorCode(
+            coll, [{$lookup: {localField: "a", foreignField: "b", from: "", as: "as"}}], 40322);
         // $lookup's field must be an object.
-        assertErrorCode(coll, [{$lookup: "string"}], 4569);
+        assertErrorCode(coll, [{$lookup: "string"}], 40319);
     }
 
     // Run tests on single node.
diff --git a/jstests/aggregation/bugs/server6045.js b/jstests/aggregation/bugs/server6045.js
index 852a8bdc093..dc250ca9516 100644
--- a/jstests/aggregation/bugs/server6045.js
+++ b/jstests/aggregation/bugs/server6045.js
@@ -27,10 +27,10 @@ db.agg.insert({key: "string", value: 17});
 db.agg.insert({key: "yarn", value: 42});
 
 // As pipeline
-assertErrorCode(db.agg, [{}], 16435);
+assertErrorCode(db.agg, [{}], 40323);
 // Start of pipeline
-assertErrorCode(db.agg, [{$project: {value: 1}}, {}], 16435);
+assertErrorCode(db.agg, [{$project: {value: 1}}, {}], 40323);
 // End of pipeline
-assertErrorCode(db.agg, [{}, {$project: {value: 1}}], 16435);
+assertErrorCode(db.agg, [{}, {$project: {value: 1}}], 40323);
 // Middle of pipeline
-assertErrorCode(db.agg, [{$project: {value: 1}}, {}, {$project: {value: 1}}], 16435);
+assertErrorCode(db.agg, [{$project: {value: 1}}, {}, {$project: {value: 1}}], 40323);
diff --git a/jstests/aggregation/bugs/server6861.js b/jstests/aggregation/bugs/server6861.js
index 28d19445241..e6748bd07da 100644
--- a/jstests/aggregation/bugs/server6861.js
+++ b/jstests/aggregation/bugs/server6861.js
@@ -19,9 +19,9 @@ function assertResult(result, expression) {
 assertResult([{a: 1}], {$project: {_id: 0, a: 1}});
 
 // Incorrect number of fields.
-assertCode(16435, {});
-assertCode(16435, {$project: {_id: 0, a: 1}, $group: {_id: 0}});
-assertCode(16435, {$project: {_id: 0, a: 1}, $group: {_id: 0}, $sort: {a: 1}});
+assertCode(40323, {});
+assertCode(40323, {$project: {_id: 0, a: 1}, $group: {_id: 0}});
+assertCode(40323, {$project: {_id: 0, a: 1}, $group: {_id: 0}, $sort: {a: 1}});
 
 // Invalid stage specification.
-assertCode(16436, {$noSuchStage: {a: 1}});
+assertCode(40324, {$noSuchStage: {a: 1}});
diff --git a/jstests/aggregation/sources/bucket/collation_bucket.js b/jstests/aggregation/sources/bucket/collation_bucket.js
index 8fb791e0901..b21ae778569 100644
--- a/jstests/aggregation/sources/bucket/collation_bucket.js
+++ b/jstests/aggregation/sources/bucket/collation_bucket.js
@@ -49,4 +49,53 @@
     assert.throws(
         () => coll.aggregate([{$bucket: {groupBy: "$num", boundaries: ["1", "10", "100", "1000"]}}],
                              {collation: {locale: "simple"}}));
+
+    // Test that $bucket rejects boundaries that are not sorted according to the collation.
+    assert.throws(
+        () => coll.aggregate([{$bucket: {groupBy: "$num", boundaries: ["100", "20", "4"]}}]));
+
+    assert.throws(() =>
+                      coll.aggregate([{$bucket: {groupBy: "$num", boundaries: ["4", "20", "100"]}}],
+                                     {collation: {locale: "simple"}}));
+
+    // Test that $bucket rejects a default value that falls within the boundaries.
+    assert.throws(
+        () => coll.aggregate(
+            [{$bucket: {groupBy: "$num", boundaries: ["1", "10", "100"], default: "40"}}]));
+
+    assert.throws(() => coll.aggregate(
+                      [{$bucket: {groupBy: "$num", boundaries: ["100", "999"], default: "2"}}],
+                      {collation: {locale: "simple"}}));
+
+    // Test that $bucket accepts a default value that falls outside the boundaries according to the
+    // collation.
+    results =
+        coll.aggregate([{
+                $bucket: {
+                    groupBy: "$num",
+                    boundaries: ["100", "999"],
+                    default: "2"  // Would fall between 100 and 999 if using the simple collation.
+                }
+            }])
+            .toArray();
+    assert.eq(2, results.length);
+    assert.eq({_id: "2", count: 6}, results[0]);
+    assert.eq({_id: "100", count: 3}, results[1]);  // "100", "200", and "500".
+
+    results =
+        coll.aggregate(
+                [{
+                   $bucket: {
+                       groupBy: "$num",
+                       boundaries: ["1", "19999"],  // Will include all numbers that start with "1"
+                       default: "2"                 // Would fall between boundaries if using the
+                                                    // collection-default collation with numeric
+                                                    // ordering.
+                   }
+                }],
+                {collation: {locale: "simple"}})
+            .toArray();
+    assert.eq(2, results.length);
+    assert.eq({_id: "1", count: 3}, results[0]);  // "1", "10", and "100".
+    assert.eq({_id: "2", count: 6}, results[1]);
 })();
diff --git a/jstests/aggregation/sources/graphLookup/error.js b/jstests/aggregation/sources/graphLookup/error.js
index 2728e1abd75..a686fb30945 100644
--- a/jstests/aggregation/sources/graphLookup/error.js
+++ b/jstests/aggregation/sources/graphLookup/error.js
@@ -9,7 +9,10 @@ load("jstests/aggregation/extras/utils.js");  // For "assertErrorCode".
     local.drop();
     local.insert({});
 
-    var pipeline = {
+    var pipeline = {$graphLookup: 4};
+    assertErrorCode(local, pipeline, 40327, "$graphLookup spec must be an object");
+
+    pipeline = {
         $graphLookup: {
             from: "foreign",
             startWith: {$literal: 0},
@@ -54,7 +57,18 @@ load("jstests/aggregation/extras/utils.js");  // For "assertErrorCode".
             as: "output"
         }
     };
-    assertErrorCode(local, pipeline, 40103, "from must be a string");
+    assertErrorCode(local, pipeline, 40329, "from must be a string");
+
+    pipeline = {
+        $graphLookup: {
+            from: "",
+            startWith: {$literal: 0},
+            connectToField: "a",
+            connectFromField: "b",
+            as: "output"
+        }
+    };
+    assertErrorCode(local, pipeline, 40330, "from must be a valid namespace");
 
     pipeline = {
         $graphLookup: {
@@ -201,7 +215,7 @@ load("jstests/aggregation/extras/utils.js");  // For "assertErrorCode".
         $graphLookup:
             {startWith: {$literal: 0}, connectToField: "a", connectFromField: "b", as: "output"}
     };
-    assertErrorCode(local, pipeline, 40105, "from was not specified");
+    assertErrorCode(local, pipeline, 40328, "from was not specified");
 
     // restrictSearchWithMatch must be a valid match expression.
     pipeline = {
diff --git a/jstests/views/views_collation.js b/jstests/views/views_collation.js
index 4dfe3e59d55..04abf106292 100644
--- a/jstests/views/views_collation.js
+++ b/jstests/views/views_collation.js
@@ -77,14 +77,26 @@
             viewsDB.runCommand({aggregate: "filView", pipeline: [], collation: {locale: "en"}}),
             ErrorCodes.OptionNotSupportedOnView);
         assert.commandFailedWithCode(
+            viewsDB.runCommand({aggregate: "filView", pipeline: [], collation: {locale: "simple"}}),
+            ErrorCodes.OptionNotSupportedOnView);
+        assert.commandFailedWithCode(
             viewsDB.runCommand({find: "filView", collation: {locale: "fr"}}),
             ErrorCodes.OptionNotSupportedOnView);
         assert.commandFailedWithCode(
+            viewsDB.runCommand({find: "filView", collation: {locale: "simple"}}),
+            ErrorCodes.OptionNotSupportedOnView);
+        assert.commandFailedWithCode(
             viewsDB.runCommand({count: "filView", collation: {locale: "zh"}}),
             ErrorCodes.OptionNotSupportedOnView);
         assert.commandFailedWithCode(
+            viewsDB.runCommand({count: "filView", collation: {locale: "simple"}}),
+            ErrorCodes.OptionNotSupportedOnView);
+        assert.commandFailedWithCode(
             viewsDB.runCommand({distinct: "filView", key: "x", collation: {locale: "es"}}),
             ErrorCodes.OptionNotSupportedOnView);
+        assert.commandFailedWithCode(
+            viewsDB.runCommand({distinct: "filView", key: "x", collation: {locale: "simple"}}),
+            ErrorCodes.OptionNotSupportedOnView);
 
         const lookupSimpleView = {
             $lookup: {from: "simpleView", localField: "x", foreignField: "x", as: "result"}