From 188f24f0d61b62b9b0601f5b890b4820d7570719 Mon Sep 17 00:00:00 2001
From: Irina Yatsenko <irina.yatsenko@mongodb.com>
Date: Wed, 2 Nov 2022 15:16:52 +0000
Subject: SERVER-70944 precompute field hashes for ExpressionFieldPath

---
 src/mongo/db/pipeline/expression.cpp |  2 +-
 src/mongo/db/pipeline/field_path.cpp | 14 +++++++-------
 src/mongo/db/pipeline/field_path.h   | 23 ++++++++++-------------
 3 files changed, 18 insertions(+), 21 deletions(-)

diff --git a/src/mongo/db/pipeline/expression.cpp b/src/mongo/db/pipeline/expression.cpp
index e87a39f85ac..080faffe5c3 100644
--- a/src/mongo/db/pipeline/expression.cpp
+++ b/src/mongo/db/pipeline/expression.cpp
@@ -2414,7 +2414,7 @@ intrusive_ptr<ExpressionFieldPath> ExpressionFieldPath::createVarFromString(
 ExpressionFieldPath::ExpressionFieldPath(ExpressionContext* const expCtx,
                                          const string& theFieldPath,
                                          Variables::Id variable)
-    : Expression(expCtx), _fieldPath(theFieldPath), _variable(variable) {
+    : Expression(expCtx), _fieldPath(theFieldPath, true /*precomputeHashes*/), _variable(variable) {
     const auto varName = theFieldPath.substr(0, theFieldPath.find('.'));
     tassert(5943201,
             std::string{"Variable with $$ROOT's id is not $$CURRENT or $$ROOT as expected, "
diff --git a/src/mongo/db/pipeline/field_path.cpp b/src/mongo/db/pipeline/field_path.cpp
index c860e863512..bdbbbabfee0 100644
--- a/src/mongo/db/pipeline/field_path.cpp
+++ b/src/mongo/db/pipeline/field_path.cpp
@@ -69,10 +69,8 @@ string FieldPath::getFullyQualifiedPath(StringData prefix, StringData suffix) {
     return str::stream() << prefix << "." << suffix;
 }
 
-FieldPath::FieldPath(std::string inputPath)
-    : _fieldPath(std::move(inputPath)),
-      _fieldPathDotPosition{string::npos},
-      _fieldHash{kHashUninitialized} {
+FieldPath::FieldPath(std::string inputPath, bool precomputeHashes)
+    : _fieldPath(std::move(inputPath)), _fieldPathDotPosition{string::npos} {
     uassert(40352, "FieldPath cannot be constructed with empty string", !_fieldPath.empty());
     uassert(40353, "FieldPath must not end with a '.'.", _fieldPath[_fieldPath.size() - 1] != '.');
 
@@ -81,19 +79,21 @@ FieldPath::FieldPath(std::string inputPath)
     size_t startPos = 0;
     while (string::npos != (dotPos = _fieldPath.find('.', startPos))) {
         _fieldPathDotPosition.push_back(dotPos);
-        _fieldHash.push_back(kHashUninitialized);
         startPos = dotPos + 1;
     }
 
     _fieldPathDotPosition.push_back(_fieldPath.size());
 
-    // Validate the path length and the fields.
+    // Validate the path length and the fields, and precompute their hashes if requested.
     const auto pathLength = getPathLength();
     uassert(ErrorCodes::Overflow,
             "FieldPath is too long",
             pathLength <= BSONDepth::getMaxAllowableDepth());
+    _fieldHash.reserve(pathLength);
     for (size_t i = 0; i < pathLength; ++i) {
-        uassertValidFieldName(getFieldName(i));
+        const auto& fieldName = getFieldName(i);
+        uassertValidFieldName(fieldName);
+        _fieldHash.push_back(precomputeHashes ? FieldNameHasher()(fieldName) : kHashUninitialized);
     }
 }
 
diff --git a/src/mongo/db/pipeline/field_path.h b/src/mongo/db/pipeline/field_path.h
index d2ee93734e7..c216bdbc0a4 100644
--- a/src/mongo/db/pipeline/field_path.h
+++ b/src/mongo/db/pipeline/field_path.h
@@ -69,9 +69,11 @@ public:
      *
      * Field names are validated using uassertValidFieldName().
      */
-    /* implicit */ FieldPath(std::string inputPath);
-    /* implicit */ FieldPath(StringData inputPath) : FieldPath(inputPath.toString()) {}
-    /* implicit */ FieldPath(const char* inputPath) : FieldPath(std::string(inputPath)) {}
+    /* implicit */ FieldPath(std::string inputPath, bool precomputeHashes = false);
+    /* implicit */ FieldPath(StringData inputPath, bool precomputeHashes = false)
+        : FieldPath(inputPath.toString(), precomputeHashes) {}
+    /* implicit */ FieldPath(const char* inputPath, bool precomputeHashes = false)
+        : FieldPath(std::string(inputPath), precomputeHashes) {}
 
     /**
      * Returns the number of path elements in the field path.
@@ -117,13 +119,8 @@ public:
      */
     HashedFieldName getFieldNameHashed(size_t i) const {
         dassert(i < getPathLength());
-        const auto begin = _fieldPathDotPosition[i] + 1;
-        const auto end = _fieldPathDotPosition[i + 1];
-        StringData fieldName{&_fieldPath[begin], end - begin};
-        if (_fieldHash[i] == kHashUninitialized) {
-            _fieldHash[i] = FieldNameHasher()(fieldName);
-        }
-        return HashedFieldName{fieldName, _fieldHash[i]};
+        invariant(_fieldHash[i] != kHashUninitialized);
+        return HashedFieldName{getFieldName(i), _fieldHash[i]};
     }
 
     /**
@@ -177,9 +174,9 @@ private:
     // lookup.
     std::vector<size_t> _fieldPathDotPosition;
 
-    // Contains the cached hash value for the field. Will initially be set to 'kHashUninitialized',
-    // and only generated when it is first retrieved via 'getFieldNameHashed'.
-    mutable std::vector<size_t> _fieldHash;
+    // Contains the hash value for the field names if it was requested when creating this path.
+    // Otherwise all elements are set to 'kHashUninitialized'.
+    std::vector<size_t> _fieldHash;
     static constexpr std::size_t kHashUninitialized = std::numeric_limits<std::size_t>::max();
 };
 
-- 
cgit v1.2.1