SERVER-54929 Helper functions to convert underlying buckets collection index specs to time-series

collection index specs

4 files changed, 111 insertions, 194 deletions

diff --git a/src/mongo/db/commands/dbcommands.cpp b/src/mongo/db/commands/dbcommands.cpp
index 34ba441a488..b6c1790794f 100644
--- a/src/mongo/db/commands/dbcommands.cpp
+++ b/src/mongo/db/commands/dbcommands.cpp
@@ -116,87 +116,6 @@ using std::unique_ptr;
 namespace {
 
 /**
- * Returns time-series options if 'ns' refers to a time-series collection.
- */
-boost::optional<TimeseriesOptions> getTimeseriesOptions(OperationContext* opCtx,
-                                                        const NamespaceString& ns) {
-    auto viewCatalog = DatabaseHolder::get(opCtx)->getViewCatalog(opCtx, ns.db());
-    if (!viewCatalog) {
-        return {};
-    }
-
-    auto view = viewCatalog->lookupWithoutValidatingDurableViews(opCtx, ns.ns());
-    if (!view) {
-        return {};
-    }
-
-    // Return a copy of the time-series options so that we don't refer to the internal state of
-    // 'viewCatalog' once it goes out of scope.
-    return view->timeseries();
-}
-
-/**
- * Returns an index key with field names mapped to the bucket collection schema.
- */
-BSONObj makeTimeseriesIndexSpecKey(const TimeseriesOptions& timeseriesOptions,
-                                   const CollMod& origCmd,
-                                   const BSONObj& origKey) {
-    auto timeField = timeseriesOptions.getTimeField();
-    auto metaField = timeseriesOptions.getMetaField();
-
-    BSONObjBuilder builder;
-    for (const auto& elem : origKey) {
-        // Determine if the index requested on the time field is ascending or descending.
-        // The final index spec will be subjected to a more complete validation in
-        // index_key_validate::validateKeyPattern().
-        if (elem.fieldNameStringData() == timeField) {
-            uassert(
-                ErrorCodes::IndexNotFound,
-                str::stream() << "Invalid index spec for time-series collection: "
-                              << redact(origCmd.toBSON({}))  // 'origKey' is included in 'origCmd'
-                              << ". Indexes on the time field must be ascending or descending "
-                                 "(numbers only): "
-                              << elem,
-                elem.isNumber());
-            if (elem.number() >= 0) {
-                builder.appendAs(elem, str::stream() << "control.min." << timeField);
-                builder.appendAs(elem, str::stream() << "control.max." << timeField);
-            } else {
-                builder.appendAs(elem, str::stream() << "control.max." << timeField);
-                builder.appendAs(elem, str::stream() << "control.min." << timeField);
-            }
-            continue;
-        }
-
-        uassert(ErrorCodes::IndexNotFound,
-                str::stream() << "Invalid index spec for time-series collection: "
-                              << redact(origCmd.toBSON({}))  // 'origKey' is included in 'origCmd'
-                              << ". Index must be on the '" << timeField << "' field: " << elem,
-                metaField);
-
-        if (elem.fieldNameStringData() == *metaField) {
-            builder.appendAs(elem, BucketUnpacker::kBucketMetaFieldName);
-            continue;
-        }
-
-        if (elem.fieldNameStringData().startsWith(*metaField + ".")) {
-            builder.appendAs(elem,
-                             str::stream()
-                                 << BucketUnpacker::kBucketMetaFieldName << "."
-                                 << elem.fieldNameStringData().substr(metaField->size() + 1));
-            continue;
-        }
-
-        uasserted(ErrorCodes::IndexNotFound,
-                  str::stream() << "Invalid index spec for time-series collection: "
-                                << redact(origCmd.toBSON({}))  // 'origKey' is included in 'origCmd'
-                                << ". Index must be either on the '" << *metaField << "' or '"
-                                << timeField << "' fields: " << elem);
-    }
-    return builder.obj();
-}
-
-/**
  * Returns a CollMod on the underlying buckets collection of the time-series collection.
  * Returns null if 'origCmd' is not for a time-series collection.
  */
@@ -204,18 +123,24 @@ std::unique_ptr<CollMod> makeTimeseriesCollModCommand(OperationContext* opCtx,
                                                       const CollMod& origCmd) {
     const auto& origNs = origCmd.getNamespace();
 
-    auto timeseriesOptions = getTimeseriesOptions(opCtx, origNs);
+    auto timeseriesOptions = timeseries::getTimeseriesOptions(opCtx, origNs);
 
     // Return early with null if we are not working with a time-series collection.
     if (!timeseriesOptions) {
         return {};
     }
 
-    // TODO(SERVER-54639): Map index specs to bucket collection using helper function.
     auto index = origCmd.getIndex();
     if (index && index->getKeyPattern()) {
-        index->setKeyPattern(
-            makeTimeseriesIndexSpecKey(*timeseriesOptions, origCmd, *index->getKeyPattern()));
+        auto bucketsIndexSpecWithStatus = timeseries::convertTimeseriesIndexSpecToBucketsIndexSpec(
+            *timeseriesOptions, *index->getKeyPattern());
+
+        uassert(ErrorCodes::IndexNotFound,
+                str::stream() << bucketsIndexSpecWithStatus.getStatus().toString()
+                              << " Command request: " << redact(origCmd.toBSON({})),
+                bucketsIndexSpecWithStatus.isOK());
+
+        index->setKeyPattern(std::move(bucketsIndexSpecWithStatus.getValue()));
     }
 
     auto ns = origNs.makeTimeseriesBucketsNamespace();
diff --git a/src/mongo/db/commands/list_indexes.cpp b/src/mongo/db/commands/list_indexes.cpp
index 699efe6555c..5d207392b6a 100644
--- a/src/mongo/db/commands/list_indexes.cpp
+++ b/src/mongo/db/commands/list_indexes.cpp
@@ -36,7 +36,6 @@
 
 #include "mongo/bson/bsonobjbuilder.h"
 #include "mongo/db/auth/authorization_session.h"
-#include "mongo/db/catalog/database_holder.h"
 #include "mongo/db/catalog/list_indexes.h"
 #include "mongo/db/clientcursor.h"
 #include "mongo/db/commands.h"
@@ -49,7 +48,6 @@
 #include "mongo/db/exec/working_set.h"
 #include "mongo/db/index/index_descriptor.h"
 #include "mongo/db/list_indexes_gen.h"
-#include "mongo/db/pipeline/document_source_internal_unpack_bucket.h"
 #include "mongo/db/query/cursor_request.h"
 #include "mongo/db/query/cursor_response.h"
 #include "mongo/db/query/find_common.h"
@@ -57,7 +55,7 @@
 #include "mongo/db/service_context.h"
 #include "mongo/db/storage/durable_catalog.h"
 #include "mongo/db/storage/storage_engine.h"
-#include "mongo/db/views/view_catalog.h"
+#include "mongo/db/timeseries/timeseries_index_schema_conversion_functions.h"
 #include "mongo/logv2/log.h"
 #include "mongo/util/uuid.h"
 
@@ -65,105 +63,20 @@ namespace mongo {
 namespace {
 
 /**
- * Returns time-series options if 'ns' refers to a time-series collection.
- */
-boost::optional<TimeseriesOptions> getTimeseriesOptions(
-    OperationContext* opCtx, const boost::optional<NamespaceString>& ns) {
-    if (!ns) {
-        return {};
-    }
-
-    auto viewCatalog = DatabaseHolder::get(opCtx)->getViewCatalog(opCtx, ns->db());
-    if (!viewCatalog) {
-        return {};
-    }
-
-    auto view = viewCatalog->lookupWithoutValidatingDurableViews(opCtx, ns->ns());
-    if (!view) {
-        return {};
-    }
-
-    // Return a copy of the time-series options so that we don't refer to the internal state of
-    // 'viewCatalog' once it goes out of scope.
-    return view->timeseries();
-}
-
-/**
- * Returns an index key with field names mapped from the bucket collection schema.
- * Returns an empty BSONObj if the index key cannot be converted.
- */
-BSONObj makeTimeseriesIndexSpecKey(const TimeseriesOptions& timeseriesOptions,
-                                   const BSONObj& origKey) {
-    auto timeField = timeseriesOptions.getTimeField();
-    auto metaField = timeseriesOptions.getMetaField();
-
-    std::string controlMinTimeField = str::stream() << "control.min." << timeField;
-    std::string controlMaxTimeField = str::stream() << "control.max." << timeField;
-
-    BSONObjBuilder builder;
-    for (const auto& elem : origKey) {
-        // Determine if the index requested on the time field is ascending or descending.
-        // The final index spec will be subjected to a more complete validation in
-        // index_key_validate::validateKeyPattern().
-        if (elem.fieldNameStringData() == controlMinTimeField) {
-            if (!elem.isNumber()) {
-                return {};
-            }
-            builder.appendAs(elem, timeField);
-            continue;
-        } else if (elem.fieldNameStringData() == controlMaxTimeField) {
-            // Skip control.max.<timeField> since the control.min.<timeField> field is enough to
-            // figure out the direction of the index (ascending/descending).
-            continue;
-        }
-
-        if (!metaField) {
-            return {};
-        }
-
-        if (elem.fieldNameStringData() == BucketUnpacker::kBucketMetaFieldName) {
-            builder.appendAs(elem, *metaField);
-            continue;
-        }
-
-        if (elem.fieldNameStringData().startsWith(BucketUnpacker::kBucketMetaFieldName + ".")) {
-            builder.appendAs(elem,
-                             str::stream() << *metaField << "."
-                                           << elem.fieldNameStringData().substr(
-                                                  BucketUnpacker::kBucketMetaFieldName.size() + 1));
-            continue;
-        }
-
-        return {};
-    }
-    return builder.obj();
-}
-
-/**
+ * Converts buckets collection index specs to the time-series collection schema.
  * Returns a list of index specs mapped from the bucket collection schema.
  */
 std::list<BSONObj> makeTimeseriesIndexSpecs(const TimeseriesOptions& timeseriesOptions,
-                                            const std::list<BSONObj>& bucketIndexSpecs) {
+                                            const std::list<BSONObj>& bucketsIndexSpecs) {
     std::list<BSONObj> indexSpecs;
-    for (const auto& bucketIndexSpec : bucketIndexSpecs) {
+    for (const auto& bucketsIndexSpec : bucketsIndexSpecs) {
         // TODO(SERVER-54639): Map index specs from bucket collection using helper function.
         BSONObjBuilder builder;
         bool skip = false;
-        for (const auto& elem : bucketIndexSpec) {
+        for (const auto& elem : bucketsIndexSpec) {
             if (elem.fieldNameStringData() == ListIndexesReplyItem::kKeyFieldName) {
-                // On a time-series collection with 'tm' time field and 'mm' metadata field,
-                // we may see a compound index on the underlying bucket collection mapped from:
-                // {
-                //     'meta.tag1': 1,
-                //     'control.min.tm': 1,
-                //     'control.max.tm': 1
-                // }
-                // to an index on the time-series collection:
-                // {
-                //     'mm.tag1': 1,
-                //     tm: 1
-                // }
-                auto key = makeTimeseriesIndexSpecKey(timeseriesOptions, elem.Obj());
+                auto key = timeseries::convertBucketsIndexSpecToTimeseriesIndexSpec(
+                    timeseriesOptions, elem.Obj());
                 if (key.isEmpty()) {
                     // Skip index spec due to failed conversion.
                     skip = true;
@@ -204,21 +117,22 @@ IndexSpecsWithNamespaceString getIndexSpecsWithNamespaceString(OperationContext*
 
     // Since time-series collections don't have UUIDs, we skip the time-series lookup
     // if the target collection is specified as a UUID.
-    if (const auto& origNss = origNssOrUUID.nss();
-        auto timeseriesOptions = getTimeseriesOptions(opCtx, origNss)) {
-        auto bucketsNss = origNss->makeTimeseriesBucketsNamespace();
-        AutoGetCollectionForReadCommandMaybeLockFree autoColl(opCtx, bucketsNss);
-
-        const CollectionPtr& coll = autoColl.getCollection();
-        uassert(ErrorCodes::NamespaceNotFound,
-                str::stream() << "ns does not exist: " << bucketsNss,
-                coll);
-
-        return std::make_pair(
-            makeTimeseriesIndexSpecs(
-                *timeseriesOptions,
-                listIndexesInLock(opCtx, coll, bucketsNss, cmd.getIncludeBuildUUIDs())),
-            *origNss);
+    if (const auto& origNss = origNssOrUUID.nss()) {
+        if (auto timeseriesOptions = timeseries::getTimeseriesOptions(opCtx, *origNss)) {
+            auto bucketsNss = origNss->makeTimeseriesBucketsNamespace();
+            AutoGetCollectionForReadCommandMaybeLockFree autoColl(opCtx, bucketsNss);
+
+            const CollectionPtr& coll = autoColl.getCollection();
+            uassert(ErrorCodes::NamespaceNotFound,
+                    str::stream() << "ns does not exist: " << bucketsNss,
+                    coll);
+
+            return std::make_pair(
+                makeTimeseriesIndexSpecs(
+                    *timeseriesOptions,
+                    listIndexesInLock(opCtx, coll, bucketsNss, cmd.getIncludeBuildUUIDs())),
+                *origNss);
+        }
     }
 
     AutoGetCollectionForReadCommandMaybeLockFree autoColl(opCtx, origNssOrUUID);
diff --git a/src/mongo/db/timeseries/timeseries_index_schema_conversion_functions.cpp b/src/mongo/db/timeseries/timeseries_index_schema_conversion_functions.cpp
index e9212d3b212..57841a7b83f 100644
--- a/src/mongo/db/timeseries/timeseries_index_schema_conversion_functions.cpp
+++ b/src/mongo/db/timeseries/timeseries_index_schema_conversion_functions.cpp
@@ -123,5 +123,59 @@ StatusWith<BSONObj> convertTimeseriesIndexSpecToBucketsIndexSpec(
     return builder.obj();
 }
 
+BSONObj convertBucketsIndexSpecToTimeseriesIndexSpec(const TimeseriesOptions& timeseriesOptions,
+                                                     const BSONObj& bucketsIndexSpecBSON) {
+    auto timeField = timeseriesOptions.getTimeField();
+    auto metaField = timeseriesOptions.getMetaField();
+
+    const std::string controlMinTimeField = str::stream() << "control.min." << timeField;
+    const std::string controlMaxTimeField = str::stream() << "control.max." << timeField;
+
+    BSONObjBuilder builder;
+    for (const auto& elem : bucketsIndexSpecBSON) {
+        // The index specification on the time field is ascending or descending.
+        if (elem.fieldNameStringData() == controlMinTimeField) {
+            if (!elem.isNumber()) {
+                // This index spec on the underlying buckets collection is not valid for
+                // time-series. Therefore, we will not convert the index spec..
+                return {};
+            }
+
+            builder.appendAs(elem, timeField);
+            continue;
+        } else if (elem.fieldNameStringData() == controlMaxTimeField) {
+            // Skip 'control.max.<timeField>' since the 'control.min.<timeField>' field is
+            // sufficient to determine whether the index is ascending or descending.
+            continue;
+        }
+
+        if (!metaField) {
+            // 'elem' is an invalid index spec field for this time-series collection. It does not
+            // match the time field and there is no metaField set. Therefore, we will not convert
+            // the index spec.
+            return {};
+        }
+
+        if (elem.fieldNameStringData() == BucketUnpacker::kBucketMetaFieldName) {
+            builder.appendAs(elem, *metaField);
+            continue;
+        }
+
+        if (elem.fieldNameStringData().startsWith(BucketUnpacker::kBucketMetaFieldName + ".")) {
+            builder.appendAs(elem,
+                             str::stream() << *metaField << "."
+                                           << elem.fieldNameStringData().substr(
+                                                  BucketUnpacker::kBucketMetaFieldName.size() + 1));
+            continue;
+        }
+
+        // 'elem' is an invalid index spec field for this time-series collection. It matches neither
+        // the time field  nor the metaField field. Therefore, we will not convert the index spec.
+        return {};
+    }
+
+    return builder.obj();
+}
+
 }  // namespace timeseries
 }  // namespace mongo
diff --git a/src/mongo/db/timeseries/timeseries_index_schema_conversion_functions.h b/src/mongo/db/timeseries/timeseries_index_schema_conversion_functions.h
index 151837bdce9..aec7cda44c7 100644
--- a/src/mongo/db/timeseries/timeseries_index_schema_conversion_functions.h
+++ b/src/mongo/db/timeseries/timeseries_index_schema_conversion_functions.h
@@ -58,5 +58,29 @@ boost::optional<TimeseriesOptions> getTimeseriesOptions(OperationContext* opCtx,
 StatusWith<BSONObj> convertTimeseriesIndexSpecToBucketsIndexSpec(
     const TimeseriesOptions& timeseriesOptions, const BSONObj& timeseriesIndexSpecBSON);
 
+/**
+ * Maps the buckets collection index spec 'bucketsIndexSpecBSON' to the index schema of the
+ * time-series collection using the information provided in 'timeseriesOptions'.
+ *
+ * If 'bucketsIndexSpecBSON' does not match a valid time-series index format, then an empty BSONObj
+ * is returned.
+ *
+ * Conversion Example:
+ * On a time-series collection with 'tm' time field and 'mm' metadata field,
+ * we may see a compound index on the underlying bucket collection mapped from:
+ * {
+ *     'meta.tag1': 1,
+ *     'control.min.tm': 1,
+ *     'control.max.tm': 1
+ * }
+ * to an index on the time-series collection:
+ * {
+ *     'mm.tag1': 1,
+ *     'tm': 1
+ * }
+ */
+BSONObj convertBucketsIndexSpecToTimeseriesIndexSpec(const TimeseriesOptions& timeseriesOptions,
+                                                     const BSONObj& bucketsIndexSpecBSON);
+
 }  // namespace timeseries
 }  // namespace mongo