Revanth14 commented on code in PR #1213: URL: https://github.com/apache/iceberg-go/pull/1213#discussion_r3449183112
########## catalog/rest/scan_planning.go: ########## @@ -0,0 +1,211 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +// This file is a PROPOSED public API surface for REST server-side scan +// planning (apache/iceberg-go#1178). The bodies are intentionally +// unimplemented; the file exists so the REST surface can be reviewed as Go. +// Endpoint capability discovery (Endpoint, SupportsEndpoint) lands separately +// in the Phase 0 PR and is intentionally not redeclared here. + +package rest + +import ( + "context" + "encoding/json" + "fmt" + "time" + + "github.com/apache/iceberg-go/table" +) + +// Compile-time proof that the REST catalog satisfies the table planner seam. +var _ table.ScanPlanner = (*Catalog)(nil) + +// ErrPlanExpired is returned when polling a plan-id the server no longer knows +// about (HTTP 404 while polling), distinct from a table-not-found 404. +var ErrPlanExpired = fmt.Errorf("%w: scan plan expired", ErrRESTError) + +// --- Capability gating (Open Question 2) ------------------------------------ +// +// A single capability check is too coarse: requiring all four endpoints falls +// back to local against sync-only servers, while requiring only the plan +// endpoint false-positives, because planTableScan can return `submitted` or +// `plan-tasks` that need the poll/fetch endpoints. The split below lets `auto` +// use a sync-only server while reserving the async/fanout path for servers +// that advertise everything. + +// SupportsPlanTableScan reports whether the server advertised the synchronous +// plan endpoint. +func (r *Catalog) SupportsPlanTableScan() bool { + panic("unimplemented: proposed API for #1178") +} + +// SupportsFullRemoteScanPlanning reports whether the server advertised all four +// scan-planning endpoints (plan, fetch-result, cancel, fetch-tasks). +func (r *Catalog) SupportsFullRemoteScanPlanning() bool { + panic("unimplemented: proposed API for #1178") +} + +// --- table.ScanPlanner implementation --------------------------------------- + +// SupportsRemoteScanPlanning reports whether this catalog can complete a remote +// plan end-to-end; backed by the split capability checks above. +func (r *Catalog) SupportsRemoteScanPlanning() bool { + panic("unimplemented: proposed API for #1178") +} + +// PlanFiles plans a scan server-side and returns tasks (and, optionally, a +// plan-scoped FileIO) for the table to read. +func (r *Catalog) PlanFiles(ctx context.Context, req table.ScanPlanningRequest) (table.ScanPlanningResult, error) { + panic("unimplemented: proposed API for #1178") +} + +// --- Low-level client methods ----------------------------------------------- + +// PlanTableScan submits a scan plan. The result is either completed inline, +// submitted (returns a plan-id to poll), or failed. +func (r *Catalog) PlanTableScan(ctx context.Context, ident table.Identifier, req PlanTableScanRequest) (PlanTableScanResponse, error) { + panic("unimplemented: proposed API for #1178") +} + +// FetchPlanningResult polls a previously submitted plan. +func (r *Catalog) FetchPlanningResult(ctx context.Context, ident table.Identifier, planID string) (FetchPlanningResultResponse, error) { + panic("unimplemented: proposed API for #1178") +} + +// CancelPlanning cancels a server-side plan. Callers should cancel on context +// cancellation using a detached context with a short timeout. +func (r *Catalog) CancelPlanning(ctx context.Context, ident table.Identifier, planID string) error { + panic("unimplemented: proposed API for #1178") +} + +// FetchScanTasks fetches the scan tasks for a plan-task handle returned by a +// completed plan. +func (r *Catalog) FetchScanTasks(ctx context.Context, ident table.Identifier, req FetchScanTasksRequest) (FetchScanTasksResponse, error) { + panic("unimplemented: proposed API for #1178") +} + +// WaitForPlan polls a submitted plan to completion using jittered backoff, +// cancelling the server-side plan if the context is cancelled. It returns an +// error if the plan is still submitted after the wait, cancelled, failed, or +// expired. +func (r *Catalog) WaitForPlan(ctx context.Context, ident table.Identifier, planID string, opts WaitForPlanOptions) (CompletedPlanningResult, error) { + panic("unimplemented: proposed API for #1178") +} + +// --- Wire types (sketch) ---------------------------------------------------- +// +// Content-file, delete-file, and residual decoding lands with the scan-task +// decoder PR; these sketch the request/response envelopes so the client +// surface compiles and reads. + +// PlanStatus is the status of a server-side plan. +type PlanStatus string + +const ( + PlanStatusCompleted PlanStatus = "completed" + PlanStatusSubmitted PlanStatus = "submitted" + // PlanStatusCancelled is valid when polling a submitted plan, but invalid + // as a planTableScan response. PlanTableScan and WaitForPlan should treat a + // cancelled initial planning response as an error. + PlanStatusCancelled PlanStatus = "cancelled" + PlanStatusFailed PlanStatus = "failed" +) + +// PlanningError is the ErrorModel payload carried by a failed planning result. +type PlanningError struct { + Message string `json:"message"` + Type string `json:"type"` + Code int `json:"code"` + Stack []string `json:"stack,omitempty"` +} + +// ScanTasks carries the task payload shared by completed planning responses and +// fetchScanTasks responses. Task/delete payload decoding lands with the +// scan-task decoder PR. +type ScanTasks struct { + PlanTasks []string `json:"plan-tasks,omitempty"` + FileScanTasks json.RawMessage `json:"file-scan-tasks,omitempty"` + DeleteFiles json.RawMessage `json:"delete-files,omitempty"` +} + +// CompletedPlanningResult is the completed arm of the planning-result union. +// PlanID is required by the initial planTableScan response's +// CompletedPlanningWithIDResult arm and omitted by fetchPlanningResult. +type CompletedPlanningResult struct { + Status PlanStatus `json:"status"` + PlanID *string `json:"plan-id,omitempty"` + ScanTasks + StorageCredentials []StorageCredential `json:"storage-credentials,omitempty"` +} + +// PlanTableScanRequest is the POST .../plan request body. Filter is the +// ExpressionParser-format JSON produced by iceberg.MarshalExpressionJSON. +type PlanTableScanRequest struct { + SnapshotID *int64 `json:"snapshot-id,omitempty"` + StartSnapshotID *int64 `json:"start-snapshot-id,omitempty"` + EndSnapshotID *int64 `json:"end-snapshot-id,omitempty"` + Select []string `json:"select,omitempty"` + Filter json.RawMessage `json:"filter,omitempty"` + MinRowsRequested *int64 `json:"min-rows-requested,omitempty"` + CaseSensitive *bool `json:"case-sensitive,omitempty"` + UseSnapshotSchema *bool `json:"use-snapshot-schema,omitempty"` + StatsFields []string `json:"stats-fields,omitempty"` +} + +// PlanTableScanResponse is the POST .../plan response. The spec models this as +// a `status`-discriminated union; the flat struct carries every arm's fields +// with omitempty so none are discarded. Task/delete RawMessage payloads are +// decoded by the scan-task decoder PR. PlanID is required for submitted and +// completed responses from planTableScan; implementations must reject a missing +// PlanID for either status. A cancelled status is invalid here and must be +// treated as an error. +type PlanTableScanResponse struct { + Status PlanStatus `json:"status"` + PlanID *string `json:"plan-id,omitempty"` Review Comment: Agreed. I kept the flat-union shape, but added structural validation instead of relying on the pointer alone. `PlanTableScanResponse` now has a custom `UnmarshalJSON`: `completed` and `submitted` responses must carry `plan-id`, and `cancelled` is rejected for `planTableScan` since the spec marks it invalid for this endpoint. The doc comment spells out the same contract, including why the id matters for `CancelPlanning` / releasing server resources. I also added focused tests for missing `plan-id` and invalid `cancelled`, so malformed tracked responses won't silently decode with a nil `PlanID`. This is the one spot where I added real tested behavior rather than a stub, since the validation is exactly what you flagged. I noted that exception in the file header so the "stubs only" framing stays honest. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
