SERVER-66786 Add V2Compact explain format for Bonsai

8 files changed, 123 insertions, 10 deletions

diff --git a/src/mongo/db/query/cqf_command_utils.h b/src/mongo/db/query/cqf_command_utils.h
index 4ceb333d364..5529afdc126 100644
--- a/src/mongo/db/query/cqf_command_utils.h
+++ b/src/mongo/db/query/cqf_command_utils.h
@@ -30,9 +30,9 @@
 #pragma once
 
 #include "mongo/db/catalog/collection.h"
+#include "mongo/db/query/query_knobs_gen.h"
 
 #define MONGO_LOGV2_DEFAULT_COMPONENT ::mongo::logv2::LogComponent::kQuery
-constexpr bool kMongoOptimizerStdCoutDebugOutput = false;
 
 namespace mongo {
 
@@ -57,7 +57,7 @@ void coutPrint(const std::string& msg, const logv2::detail::NamedArg<Args>&... a
 
 #define OPTIMIZER_DEBUG_LOG(ID, DLEVEL, FMTSTR_MESSAGE, ...) \
     LOGV2_DEBUG(ID, DLEVEL, FMTSTR_MESSAGE, ##__VA_ARGS__);  \
-    if (kMongoOptimizerStdCoutDebugOutput)                   \
+    if (internalCascadesOptimizerStdCoutDebugOutput.load())  \
         ::mongo::coutPrint(FMTSTR_MESSAGE, __VA_ARGS__);
 
 /**
diff --git a/src/mongo/db/query/cqf_get_executor.cpp b/src/mongo/db/query/cqf_get_executor.cpp
index 8e5daaec376..799d214caae 100644
--- a/src/mongo/db/query/cqf_get_executor.cpp
+++ b/src/mongo/db/query/cqf_get_executor.cpp
@@ -274,7 +274,7 @@ static std::unique_ptr<PlanExecutor, PlanExecutor::Deleter> optimizeAndCreateExe
     }
 
     {
-        const std::string explain = ExplainGenerator::explainV2(
+        const std::string explain = ExplainGenerator::explainV2Compact(
             make<MemoPhysicalDelegatorNode>(phaseManager.getPhysicalNodeId()),
             true /*displayPhysicalProperties*/,
             &phaseManager.getMemo());
@@ -520,7 +520,7 @@ std::unique_ptr<PlanExecutor, PlanExecutor::Deleter> getSBEExecutorViaCascadesOp
     }
 
     OPTIMIZER_DEBUG_LOG(
-        6264803, 5, "Translated ABT", "explain"_attr = ExplainGenerator::explainV2(abt));
+        6264803, 5, "Translated ABT", "explain"_attr = ExplainGenerator::explainV2Compact(abt));
 
     const int64_t numRecords = collectionExists ? collection->numRecords(opCtx) : -1;
 
diff --git a/src/mongo/db/query/optimizer/explain.cpp b/src/mongo/db/query/optimizer/explain.cpp
index 206b7bf0d8b..cc4e7cfab69 100644
--- a/src/mongo/db/query/optimizer/explain.cpp
+++ b/src/mongo/db/query/optimizer/explain.cpp
@@ -43,7 +43,7 @@ BSONObj ABTPrinter::explainBSON() const {
         _abtTree, true /*displayProperties*/, nullptr /*Memo*/, _nodeToPropsMap);
 }
 
-enum class ExplainVersion { V1, V2, V3, Vmax };
+enum class ExplainVersion { V1, V2, V2Compact, V3, Vmax };
 
 bool constexpr operator<(const ExplainVersion v1, const ExplainVersion v2) {
     return static_cast<int>(v1) < static_cast<int>(v2);
@@ -72,6 +72,18 @@ struct CommandStruct {
 
 using CommandVector = std::vector<CommandStruct>;
 
+/**
+ * Helper class for building indented, multiline strings.
+ *
+ * The main operations it supports are:
+ *   - Print a single value, of any type that supports '<<' to std::ostream.
+ *   - Indent/unindent, and add newlines.
+ *   - Print another ExplainPrinterImpl, preserving its 2D layout.
+ *
+ * Being able to print another whole printer makes it easy to build these 2D strings
+ * bottom-up, without passing around a std::ostream. It also allows displaying
+ * child elements in a different order than they were visited.
+ */
 template <const ExplainVersion version = kDefaultExplainVersion>
 class ExplainPrinterImpl {
 public:
@@ -81,6 +93,7 @@ public:
           _osDirty(false),
           _indentCount(0),
           _childrenRemaining(0),
+          _inlineNextChild(false),
           _cmdInsertPos(-1) {}
 
     ~ExplainPrinterImpl() {
@@ -101,6 +114,7 @@ public:
           _osDirty(other._osDirty),
           _indentCount(other._indentCount),
           _childrenRemaining(other._childrenRemaining),
+          _inlineNextChild(other._inlineNextChild),
           _cmdInsertPos(other._cmdInsertPos) {}
 
     template <class T>
@@ -140,12 +154,20 @@ public:
     }
 
     ExplainPrinterImpl& setChildCount(const size_t childCount) {
-        if (version > ExplainVersion::V1) {
+        if (version == ExplainVersion::V1) {
+            return *this;
+        }
+
+        if (version == ExplainVersion::V2Compact && childCount == 1) {
+            _inlineNextChild = true;
             _childrenRemaining = childCount;
-            indent("");
-            for (int i = 0; i < _childrenRemaining - 1; i++) {
-                indent("|");
-            }
+            return *this;
+        }
+
+        _childrenRemaining = childCount;
+        indent("");
+        for (int i = 0; i < _childrenRemaining - 1; i++) {
+            indent("|");
         }
         return *this;
     }
@@ -205,6 +227,10 @@ public:
         return os.str();
     }
 
+    /**
+     * Ends the current line, if there is one. Repeated calls do not create
+     * blank lines.
+     */
     void newLine() {
         if (!_osDirty) {
             return;
@@ -251,8 +277,25 @@ private:
                     _os << element._str;
                 }
             }
+        } else if (_inlineNextChild) {
+            _inlineNextChild = false;
+            // Print 'other' without starting a new line.
+            // Embed its first line into our current one, and keep the rest of its commands.
+            bool first = true;
+            for (const CommandStruct& element : other.getCommands()) {
+                if (first && element._type == CommandType::AddLine) {
+                    _os << singleLevelSpacer << element._str;
+                } else {
+                    newLine();
+                    _cmd.push_back(element);
+                }
+                first = false;
+            }
         } else {
             newLine();
+            // If 'hadChildrenRemaining' then 'other' represents a child of 'this', which means
+            // there was a prior call to setChildCount() that added indentation for it.
+            // If '! hadChildrenRemaining' then create indentation for it now.
             if (!hadChildrenRemaining) {
                 indent();
             }
@@ -281,11 +324,22 @@ private:
         _cmd.emplace_back(CommandType::Unindent, "");
     }
 
+    // Holds completed lines, and indent/unIndent commands.
+    // When '_cmdInsertPos' is nonnegative, some of these lines and commands belong
+    // after the currently-being-built line.
     CommandVector _cmd;
+    // Holds the incomplete line currently being built. Once complete this will become the last
+    // line, unless '_cmdInsertPos' is nonnegative.
     std::ostringstream _os;
+    // True means we have an incomplete line in '_os'.
+    // Once the line is completed with newLine(), this flag is false until
+    // we begin building a new one with print().
     bool _osDirty;
     int _indentCount;
     int _childrenRemaining;
+    bool _inlineNextChild;
+    // When nonnegative, indicates the insertion point where completed lines
+    // should be added to '_cmd'. -1 means completed lines will be added at the end.
     int _cmdInsertPos;
 };
 
@@ -2305,6 +2359,14 @@ std::string ExplainGenerator::explainV2(const ABT& node,
     return gen.generate(node).str();
 }
 
+std::string ExplainGenerator::explainV2Compact(const ABT& node,
+                                               const bool displayProperties,
+                                               const cascades::Memo* memo,
+                                               const NodeToGroupPropsMap& nodeMap) {
+    ExplainGeneratorTransporter<ExplainVersion::V2Compact> gen(displayProperties, memo, nodeMap);
+    return gen.generate(node).str();
+}
+
 std::string ExplainGenerator::explainNode(const ABT& node) {
     if (node.empty()) {
         return "Empty\n";
diff --git a/src/mongo/db/query/optimizer/explain.h b/src/mongo/db/query/optimizer/explain.h
index 5cc0f29fafe..dfc8cdda77d 100644
--- a/src/mongo/db/query/optimizer/explain.h
+++ b/src/mongo/db/query/optimizer/explain.h
@@ -75,6 +75,13 @@ public:
                                  const cascades::Memo* memo = nullptr,
                                  const NodeToGroupPropsMap& nodeMap = {});
 
+    // Optionally display logical and physical properties using the memo.
+    // whenever memo delegators are printed.
+    static std::string explainV2Compact(const ABT& node,
+                                        bool displayProperties = false,
+                                        const cascades::Memo* memo = nullptr,
+                                        const NodeToGroupPropsMap& nodeMap = {});
+
     static std::string explainNode(const ABT& node);
 
     static std::pair<sbe::value::TypeTags, sbe::value::Value> explainBSON(
diff --git a/src/mongo/db/query/optimizer/optimizer_test.cpp b/src/mongo/db/query/optimizer/optimizer_test.cpp
index 5e1534fe936..d3ec1ca24a5 100644
--- a/src/mongo/db/query/optimizer/optimizer_test.cpp
+++ b/src/mongo/db/query/optimizer/optimizer_test.cpp
@@ -698,5 +698,35 @@ TEST(Properties, Basic) {
     }
 }
 
+TEST(Explain, ExplainV2Compact) {
+    ABT pathNode =
+        make<PathGet>("a",
+                      make<PathTraverse>(
+                          make<PathComposeM>(
+                              make<PathCompare>(Operations::Gte,
+                                                make<UnaryOp>(Operations::Neg, Constant::int64(2))),
+                              make<PathCompare>(Operations::Lt, Constant::int64(7))),
+                          1));
+    ABT scanNode = make<ScanNode>("x1", "test");
+    ABT evalNode = make<EvaluationNode>(
+        "x2", make<EvalPath>(pathNode, make<Variable>("a")), std::move(scanNode));
+
+    ASSERT_EXPLAIN_V2Compact(
+        "Evaluation []\n"
+        "|   BindBlock:\n"
+        "|       [x2]\n"
+        "|           EvalPath []\n"
+        "|           |   Variable [a]\n"
+        "|           PathGet [a] PathTraverse [1] PathComposeM []\n"
+        "|           |   PathCompare [Lt] Const [7]\n"
+        "|           PathCompare [Gte] UnaryOp [Neg] Const [2]\n"
+        "Scan [test]\n"
+        "    BindBlock:\n"
+        "        [x1]\n"
+        "            Source []\n",
+        evalNode);
+}
+
+
 }  // namespace
 }  // namespace mongo::optimizer
diff --git a/src/mongo/db/query/optimizer/utils/unit_test_utils.cpp b/src/mongo/db/query/optimizer/utils/unit_test_utils.cpp
index 01882a4ff22..1bd82ce80f5 100644
--- a/src/mongo/db/query/optimizer/utils/unit_test_utils.cpp
+++ b/src/mongo/db/query/optimizer/utils/unit_test_utils.cpp
@@ -45,12 +45,14 @@ void maybePrintABT(const ABT& abt) {
     // Always print using the supported versions to make sure we don't crash.
     const std::string strV1 = ExplainGenerator::explain(abt);
     const std::string strV2 = ExplainGenerator::explainV2(abt);
+    const std::string strV2Compact = ExplainGenerator::explainV2Compact(abt);
     auto [tag, val] = ExplainGenerator::explainBSON(abt);
     sbe::value::ValueGuard vg(tag, val);
 
     if constexpr (kDebugAsserts) {
         std::cout << "V1: " << strV1 << "\n";
         std::cout << "V2: " << strV2 << "\n";
+        std::cout << "V2Compact: " << strV2Compact << "\n";
         std::cout << "BSON: " << ExplainGenerator::printBSON(tag, val) << "\n";
     }
 }
diff --git a/src/mongo/db/query/optimizer/utils/unit_test_utils.h b/src/mongo/db/query/optimizer/utils/unit_test_utils.h
index 24a32017416..f25df1a11b0 100644
--- a/src/mongo/db/query/optimizer/utils/unit_test_utils.h
+++ b/src/mongo/db/query/optimizer/utils/unit_test_utils.h
@@ -49,6 +49,10 @@ void maybePrintABT(const ABT& abt);
     maybePrintABT(abt);                  \
     ASSERT_EQ(expected, ExplainGenerator::explainV2(abt))
 
+#define ASSERT_EXPLAIN_V2Compact(expected, abt) \
+    maybePrintABT(abt);                         \
+    ASSERT_EQ(expected, ExplainGenerator::explainV2Compact(abt))
+
 #define ASSERT_EXPLAIN_BSON(expected, abt) \
     maybePrintABT(abt);                    \
     ASSERT_EQ(expected, ExplainGenerator::explainBSON(abt))
diff --git a/src/mongo/db/query/query_knobs.idl b/src/mongo/db/query/query_knobs.idl
index 3edf0e05cc9..efbe81f8501 100644
--- a/src/mongo/db/query/query_knobs.idl
+++ b/src/mongo/db/query/query_knobs.idl
@@ -790,6 +790,14 @@ server_parameters:
     cpp_vartype: AtomicWord<bool>
     default: false
 
+  internalCascadesOptimizerStdCoutDebugOutput:
+    description: "Enables verbose, non-JSON, debug output for Cascades optimizer."
+    set_at: [ startup, runtime ]
+    cpp_varname: "internalCascadesOptimizerStdCoutDebugOutput"
+    cpp_vartype: AtomicWord<bool>
+    default: false
+    test_only: true
+
   internalQueryFrameworkControl:
     description: "Knob to control the optimizer/execution engine to use."
     set_at: [ startup, runtime ]