SERVER-71718 Refactor extractTimeAndMeta to branch earlier and iterate document once

(cherry picked from commit 88d18f93e89f1124fc121d4c8c10b936c851eb1a)

4 files changed, 74 insertions, 25 deletions

diff --git a/src/mongo/db/timeseries/bucket_catalog.cpp b/src/mongo/db/timeseries/bucket_catalog.cpp
index cef4bb3a964..d7d07cb5c8c 100644
--- a/src/mongo/db/timeseries/bucket_catalog.cpp
+++ b/src/mongo/db/timeseries/bucket_catalog.cpp
@@ -1285,14 +1285,23 @@ StatusWith<std::pair<BucketCatalog::BucketKey, Date_t>> BucketCatalog::_extractB
     const StringData::ComparatorInterface* comparator,
     const TimeseriesOptions& options,
     const BSONObj& doc) const {
-    auto swDocTimeAndMeta = timeseries::extractTimeAndMeta(doc, options);
-    if (!swDocTimeAndMeta.isOK()) {
-        return swDocTimeAndMeta.getStatus();
-    }
-    auto time = swDocTimeAndMeta.getValue().first;
+    Date_t time;
     BSONElement metadata;
-    if (auto metadataValue = swDocTimeAndMeta.getValue().second) {
-        metadata = *metadataValue;
+
+    if (!options.getMetaField().has_value()) {
+        auto swTime = timeseries::extractTime(doc, options.getTimeField());
+        if (!swTime.isOK()) {
+            return swTime.getStatus();
+        }
+        time = swTime.getValue();
+    } else {
+        auto swDocTimeAndMeta = timeseries::extractTimeAndMeta(
+            doc, options.getTimeField(), options.getMetaField().value());
+        if (!swDocTimeAndMeta.isOK()) {
+            return swDocTimeAndMeta.getStatus();
+        }
+        time = swDocTimeAndMeta.getValue().first;
+        metadata = swDocTimeAndMeta.getValue().second;
     }
 
     // Buckets are spread across independently-lockable stripes to improve parallelism. We map a
diff --git a/src/mongo/db/timeseries/bucket_catalog_helpers.cpp b/src/mongo/db/timeseries/bucket_catalog_helpers.cpp
index b54518eafb7..4500212e420 100644
--- a/src/mongo/db/timeseries/bucket_catalog_helpers.cpp
+++ b/src/mongo/db/timeseries/bucket_catalog_helpers.cpp
@@ -197,22 +197,45 @@ StatusWith<Schema> generateSchemaFromBucketDoc(const BSONObj& bucketDoc,
     }
 }
 
-StatusWith<std::pair<Date_t, boost::optional<BSONElement>>> extractTimeAndMeta(
-    const BSONObj& doc, const TimeseriesOptions& options) {
-    auto timeElem = doc[options.getTimeField()];
+StatusWith<Date_t> extractTime(const BSONObj& doc, StringData timeFieldName) {
+    auto timeElem = doc[timeFieldName];
     if (!timeElem || BSONType::Date != timeElem.type()) {
         return {ErrorCodes::BadValue,
-                str::stream() << "'" << options.getTimeField() << "' must be present and contain a "
+                str::stream() << "'" << timeFieldName << "' must be present and contain a "
                               << "valid BSON UTC datetime value"};
     }
+    return timeElem.Date();
+}
 
-    auto time = timeElem.Date();
-    auto metaFieldName = options.getMetaField();
 
-    if (metaFieldName) {
-        return std::make_pair(time, doc[*metaFieldName]);
+StatusWith<std::pair<Date_t, BSONElement>> extractTimeAndMeta(const BSONObj& doc,
+                                                              StringData timeFieldName,
+                                                              StringData metaFieldName) {
+    // Iterate the document once, checking for both fields.
+    BSONElement timeElem;
+    BSONElement metaElem;
+    for (auto&& el : doc) {
+        if (!timeElem && el.fieldNameStringData() == timeFieldName) {
+            timeElem = el;
+            if (metaElem) {
+                break;
+            }
+        } else if (!metaElem && el.fieldNameStringData() == metaFieldName) {
+            metaElem = el;
+            if (timeElem) {
+                break;
+            }
+        }
+    }
+
+    if (!timeElem || BSONType::Date != timeElem.type()) {
+        return {ErrorCodes::BadValue,
+                str::stream() << "'" << timeFieldName << "' must be present and contain a "
+                              << "valid BSON UTC datetime value"};
     }
-    return std::make_pair(time, boost::none);
+    auto time = timeElem.Date();
+
+    return std::make_pair(time, metaElem);
 }
 
 void normalizeMetadata(BSONObjBuilder* builder,
diff --git a/src/mongo/db/timeseries/bucket_catalog_helpers.h b/src/mongo/db/timeseries/bucket_catalog_helpers.h
index f81004ca147..aa00c9541da 100644
--- a/src/mongo/db/timeseries/bucket_catalog_helpers.h
+++ b/src/mongo/db/timeseries/bucket_catalog_helpers.h
@@ -57,12 +57,20 @@ StatusWith<Schema> generateSchemaFromBucketDoc(const BSONObj& bucketDoc,
                                                const StringData::ComparatorInterface* comparator);
 
 /**
- * Extracts the time field of a measurement document and its meta field if it is present.
+ * Extracts the time field of a measurement document.
  *
  * Returns a bad status if the document is malformed.
  */
-StatusWith<std::pair<Date_t, boost::optional<BSONElement>>> extractTimeAndMeta(
-    const BSONObj& doc, const TimeseriesOptions& options);
+StatusWith<Date_t> extractTime(const BSONObj& doc, StringData timeFieldName);
+
+/**
+ * Extracts the time field of a measurement document and its meta field.
+ *
+ * Returns a bad status if the document is malformed.
+ */
+StatusWith<std::pair<Date_t, BSONElement>> extractTimeAndMeta(const BSONObj& doc,
+                                                              StringData timeFieldName,
+                                                              StringData metaFieldName);
 
 
 /**
diff --git a/src/mongo/db/timeseries/bucket_catalog_helpers_test.cpp b/src/mongo/db/timeseries/bucket_catalog_helpers_test.cpp
index 150f6e0377f..0a5442d4080 100644
--- a/src/mongo/db/timeseries/bucket_catalog_helpers_test.cpp
+++ b/src/mongo/db/timeseries/bucket_catalog_helpers_test.cpp
@@ -76,18 +76,27 @@ BSONObj BucketCatalogHelpersTest::_findSuitableBucket(OperationContext* opCtx,
             "Missing bucketMaxSpanSeconds option.",
             options.getBucketMaxSpanSeconds());
 
-    auto swDocTimeAndMeta = timeseries::extractTimeAndMeta(measurementDoc, options);
-    if (!swDocTimeAndMeta.isOK()) {
-        return BSONObj();
+    Date_t time;
+    BSONElement metadata;
+    if (!options.getMetaField().has_value()) {
+        auto swTime = timeseries::extractTime(measurementDoc, options.getTimeField());
+        ASSERT_OK(swTime);
+        time = swTime.getValue();
+    } else {
+        auto swDocTimeAndMeta = timeseries::extractTimeAndMeta(
+            measurementDoc, options.getTimeField(), options.getMetaField().value());
+        ASSERT_OK(swDocTimeAndMeta);
+        time = swDocTimeAndMeta.getValue().first;
+        metadata = swDocTimeAndMeta.getValue().second;
     }
-    auto [time, metadata] = swDocTimeAndMeta.getValue();
+
     auto controlMinTimePath =
         timeseries::kControlMinFieldNamePrefix.toString() + options.getTimeField();
 
     boost::optional<BSONObj> normalizedMetadata;
-    if (metadata && (*metadata).ok()) {
+    if (metadata.ok()) {
         BSONObjBuilder builder;
-        timeseries::normalizeMetadata(&builder, *metadata, timeseries::kBucketMetaFieldName);
+        timeseries::normalizeMetadata(&builder, metadata, timeseries::kBucketMetaFieldName);
         normalizedMetadata = builder.obj();
     }