Add cross-partition DISTINCT queries

sdk/cosmos/.cspell.json

@@ -88,6 +88,9 @@
     "Daad",
     "dcount",
     "dedicatedgateway",
+    "dedup",
+    "dedupe",
+    "dedups",
     "deprioritized",
     "dkunda",
     "deprioritizes",
@@ -155,6 +158,7 @@
     "Gson",
     "Gtle",
     "hedgeable",
+    "hexdigit",
     "HRESULT",
     "hresult",
     "hostnames",
@@ -166,6 +170,8 @@
     "injective",
     "inlines",
     "inmemory",
+    "inspectable",
+    "intprop",
     "ints",
     "intptr",
     "INVALIDARG",
@@ -186,6 +192,7 @@
     "LIBCLANG",
     "libfuzzer",
     "libqueryplaninterop",
+    "queryplan",
     "QUERYPLANINTEROP",
     "linearizability",
     "livesite",
@@ -276,6 +283,8 @@
     "readfeed",
     "recompiles",
     "redecoded",
+    "reemit",
+    "reemitted",
     "reencode",
     "reencoded",
     "refetch",
@@ -287,6 +296,7 @@
     "Replicaset",
     "reqs",
     "restype",
+    "resumability",
     "Retriable",
     "retryable",
     "replayable",

sdk/cosmos/azure_data_cosmos/CHANGELOG.md

@@ -9,6 +9,7 @@
 - Added `DatabaseClient::name()` and `DatabaseClient::rid()` to inspect how a database client was addressed. ([#4687](https://github.com/Azure/azure-sdk-for-rust/pull/4687))
 - RID-addressed databases skip the extra database read when resolving throughput offers, reusing the addressed RID directly. ([#4687](https://github.com/Azure/azure-sdk-for-rust/pull/4687))
 - Reading and querying items by RID now works end-to-end, including a parent-database cross-check that rejects a container RID belonging to a different database. ([#4687](https://github.com/Azure/azure-sdk-for-rust/pull/4687))
+- Added cross-partition `DISTINCT` query support. `SELECT DISTINCT` now deduplicates structurally equal values across every physical partition and page, rather than failing as an unsupported query feature. A `DISTINCT` query whose `ORDER BY` prefixes its projection (for example `SELECT DISTINCT VALUE c.city FROM c ORDER BY c.city`) is resumable from a continuation token; one without such an `ORDER BY` is not, and requesting a token for it returns an error explaining how to rewrite the query. ([#5026](https://github.com/Azure/azure-sdk-for-rust/pull/5026))
 - Added an SDK-generated `x-ms-client-id` header that remains stable for each `CosmosClient`. ([#4844](https://github.com/Azure/azure-sdk-for-rust/pull/4844))
 - Added opt-in Cosmos binary JSON encoding for item operations (`create`/`read`/`replace`/`upsert`). Enable it via `CosmosClientBuilder::with_binary_encoding_options` (or the `AZURE_COSMOS_BINARY_ENCODING_ENABLED` environment-variable fallback). Off by default; when disabled, requests and responses are byte-for-byte unchanged. ([#4671](https://github.com/Azure/azure-sdk-for-rust/pull/4671))
 - Added `FeedOptions::max_fan_out` (and `FeedOptions::with_max_fan_out`) to cap how many physical partitions a cross-partition query or change feed may fan out to. Applies to `ContainerClient::query_items` and `ContainerClient::query_change_feed`. The cap is enforced only at initial query setup; a partition that splits mid-execution and pushes the fan-out higher does not abort the operation. ([#4855](https://github.com/Azure/azure-sdk-for-rust/pull/4855))

sdk/cosmos/azure_data_cosmos/tests/emulator_tests/cosmos_hpk.rs

@@ -802,15 +802,17 @@ pub async fn hpk_query_single_partition_order_by_servable() -> Result<(), Box<dy
     .await
 }
 
-/// B9: cross-partition advanced query operators (`DISTINCT`, aggregates,
-/// `GROUP BY`, `OFFSET/LIMIT`) over a hierarchical container are not servable.
+/// B9: cross-partition advanced query operators over a hierarchical container.
 ///
-/// The Rust SDK advertises no cross-partition query features
-/// (`SUPPORTED_QUERY_FEATURES=""`), so any full-container query that needs a
-/// client-side pipeline is rejected. HPK fan-out itself works since #4729 (see
-/// B6); these cases fail solely because the SDK lacks the cross-partition
-/// operator pipeline. Each case is expected to error with 400 BadRequest /
-/// 1004 CrossPartitionQueryNotServable.
+/// The SDK advertises only the features it has a client-side pipeline for (see
+/// `query::SUPPORTED_QUERY_FEATURES`), and the service rejects any query needing
+/// one it did not advertise with 400 BadRequest / 1004
+/// CrossPartitionQueryNotServable. HPK fan-out itself already works (see B6),
+/// so these cases turn on operator support alone.
+///
+/// `DISTINCT` moved from the rejected set to the servable set when the DISTINCT
+/// stage landed; aggregates, `GROUP BY`, and `OFFSET`/`LIMIT` are still
+/// pipeline-less and remain rejected.
 #[tokio::test]
 #[cfg_attr(
     not(any(test_category = "emulator", test_category = "emulator_vnext")),
@@ -825,8 +827,26 @@ pub async fn hpk_query_cross_partition_advanced_not_servable() -> Result<(), Box
         async |run_context, db_client| {
             let container = seed_three_level(run_context, db_client).await?;
 
+            // Servable: DISTINCT has a client-side stage, and it must
+            // deduplicate correctly across the container's physical partitions.
+            let mut countries = collect_query::<serde_json::Value>(
+                &container,
+                "SELECT DISTINCT VALUE c.country FROM c",
+                FeedScope::full_container(),
+            )
+            .await?
+            .into_iter()
+            .map(|v| v.as_str().unwrap_or_default().to_owned())
+            .collect::<Vec<_>>();
+            countries.sort();
+            assert_eq!(
+                countries,
+                vec!["CANADA".to_string(), "USA".to_string()],
+                "cross-partition DISTINCT over an HPK container must return each country once"
+            );
+
+            // Still rejected: no client-side pipeline for these yet.
             let advanced = [
-                "SELECT DISTINCT c.country FROM c",
                 "SELECT VALUE COUNT(1) FROM c",
                 "SELECT c.state, COUNT(1) AS n FROM c GROUP BY c.state",
                 "SELECT * FROM c OFFSET 1 LIMIT 2",