SERVER-113442: Add PathArrayness to ExpCtx in find_cmd for main collection (#43741)

GitOrigin-RevId: 65ff07f82713f35638925be185ce4847b9e3603b

src/mongo/db/commands/query_cmd/find_cmd.cpp

@@ -73,6 +73,7 @@
 #include "mongo/db/query/client_cursor/cursor_response.h"
 #include "mongo/db/query/collation/collator_factory_interface.h"
 #include "mongo/db/query/collation/collator_interface.h"
+#include "mongo/db/query/collection_query_info.h"
 #include "mongo/db/query/compiler/parsers/matcher/expression_parser.h"
 #include "mongo/db/query/explain.h"
 #include "mongo/db/query/explain_diagnostic_printer.h"
@@ -188,10 +189,13 @@ std::unique_ptr<CanonicalQuery> parseQueryAndBeginOperation(
 
     const auto& collection = collOrViewAcquisition.getCollectionPtr();
     const auto* collator = collection ? collection->getDefaultCollator() : nullptr;
-    auto expCtx = ExpressionContextBuilder{}
+    auto expCtx =
-                      .fromRequest(opCtx, *findCommand, collator, allowDiskUseByDefault.load())
+        ExpressionContextBuilder{}
-                      .tmpDir(boost::filesystem::path(storageGlobalParams.dbpath) / "_tmp")
+            .fromRequest(opCtx, *findCommand, collator, allowDiskUseByDefault.load())
-                      .build();
+            .tmpDir(boost::filesystem::path(storageGlobalParams.dbpath) / "_tmp")
+            .mainCollPathArrayness(
+                collection ? CollectionQueryInfo::get(collection).getPathArrayness() : nullptr)
+            .build();
     expCtx->startExpressionCounters();
     auto parsedRequest = uassertStatusOK(parsed_find_command::parse(
         expCtx,
@@ -521,6 +525,9 @@ public:
                     .fromRequest(opCtx, *_cmdRequest, collator, allowDiskUseByDefault.load())
                     .explain(verbosity)
                     .tmpDir(boost::filesystem::path(storageGlobalParams.dbpath) / "_tmp")
+                    .mainCollPathArrayness(
+                        collectionPtr ? CollectionQueryInfo::get(collectionPtr).getPathArrayness()
+                                      : nullptr)
                     .build();
             expCtx->startExpressionCounters();
 

src/mongo/db/pipeline/expression_context.h

@@ -48,6 +48,7 @@
 #include "mongo/db/pipeline/variables.h"
 #include "mongo/db/query/collation/collation_spec.h"
 #include "mongo/db/query/collation/collator_interface.h"
+#include "mongo/db/query/compiler/metadata/path_arrayness.h"
 #include "mongo/db/query/datetime/date_time_support.h"
 #include "mongo/db/query/explain_options.h"
 #include "mongo/db/query/query_feature_flags_gen.h"
@@ -1019,6 +1020,17 @@ public:
         return _featureFlagMqlJsEngineGap.get(VersionContext::getDecoration(getOperationContext()));
     }
 
+    const PathArrayness& getMainCollPathArrayness() const {
+        // mainCollPathArrayness will be unset in cases where we do not do a collection acquisition,
+        // e.g. if running on a 'mongos'. In this case, we return an empty instance of
+        // 'PathArrayness' that denotes all paths as arrays.
+        if (!_params.mainCollPathArrayness) {
+            static const auto kEmptyPathArrayness = PathArrayness();
+            return kEmptyPathArrayness;
+        }
+        return *_params.mainCollPathArrayness;
+    }
+
 protected:
     struct ExpressionContextParams {
         OperationContext* opCtx = nullptr;
@@ -1160,6 +1172,10 @@ protected:
 
         // Indicates that the query is replanned after being rate-limited.
         bool wasRateLimited = false;
+
+        // The PathArrayness information for the main collection. This may remain unset if a
+        // collection acquisition is not possible, e.g. when running on mongos.
+        std::shared_ptr<const PathArrayness> mainCollPathArrayness = nullptr;
     };
 
     ExpressionContextParams _params;

src/mongo/db/pipeline/expression_context_builder.cpp

@@ -392,6 +392,12 @@ ExpressionContextBuilder& ExpressionContextBuilder::tailableMode(TailableModeEnu
     return *this;
 }
 
+ExpressionContextBuilder& ExpressionContextBuilder::mainCollPathArrayness(
+    std::shared_ptr<const PathArrayness> mainCollPathArrayness) {
+    params.mainCollPathArrayness = mainCollPathArrayness;
+    return *this;
+}
+
 ExpressionContextBuilder& ExpressionContextBuilder::fromRequest(
     OperationContext* operationContext,
     const FindCommandRequest& request,

src/mongo/db/pipeline/expression_context_builder.h

@@ -105,6 +105,8 @@ public:
         boost::optional<std::pair<NamespaceString, std::vector<BSONObj>>>);
     ExpressionContextBuilder& originalNs(NamespaceString);
     ExpressionContextBuilder& isHybridSearch(bool);
+    ExpressionContextBuilder& mainCollPathArrayness(
+        std::shared_ptr<const PathArrayness> mainCollPathArrayness);
 
     /**
      * Add kSessionTransactionsTableNamespace, and kRsOplogNamespace

src/mongo/db/query/compiler/metadata/path_arrayness.cpp

@@ -29,6 +29,8 @@
 
 #include "mongo/db/query/compiler/metadata/path_arrayness.h"
 
+#include "mongo/logv2/log.h"
+
 #include <stack>
 
 using namespace mongo::multikey_paths;
@@ -126,10 +128,4 @@ void PathArrayness::TrieNode::insertPath(const FieldPath& path,
     // Recursively invoke the remaining path.
     _children.at(fieldNameToInsert).insertPath(path, multikeyPath, ++depth);
 }
-
-PathArrayness build(std::vector<IndexEntry> entries) {
-    PathArrayness root;
-    return root;
-}
-
 }  // namespace mongo

src/mongo/db/query/compiler/metadata/path_arrayness.h

@@ -29,8 +29,8 @@
 
 #pragma once
 
+#include "mongo/db/index/multikey_paths.h"
 #include "mongo/db/pipeline/field_path.h"
-#include "mongo/db/query/compiler/metadata/index_entry.h"
 #include "mongo/util/modules.h"
 
 namespace mongo {
@@ -140,10 +140,4 @@ private:
      */
     TrieNode _root;
 };
-
-/**
- * Build an arrayness trie from a given vector of index entries.
- */
-PathArrayness build(std::vector<IndexEntry> entries);
-
 }  // namespace mongo