Files
ragflow/internal/ingestion/knowledge_compile/dedup_test.go
Zhichang Yu 01d667296d refactor(knowledge_compile): global compile pool, token-budget batching, and DocEngine-only deletion (#17679)
## Summary

This PR refactors the Go knowledge-compilation ingestion pipeline
(`internal/ingestion/knowledge_compile` +
`internal/ingestion/component/knowledge_compiler`) with three related
changes:

- **Token-budget batching for LLM merge decisions.**
`LLMMergeDecider.DecideBatch` previously stuffed every `(existing,
candidate)` pair into a single LLM call, risking `max_token` overflow.
It now splits pairs into token-bounded sub-batches (budget =
`llmMaxTokens * 0.85`) via `tokenizer.NumTokensFromString`, runs them
concurrently while preserving the global pair index, and never
reindexes.
- **Process-level global compile pool.** Introduces a single vCPU-sized
goroutine pool (`pool.go`, env `KC_COMPILE_CONCURRENCY`) dedicated to
*all* knowledge-compilation stages. KNN search loop, `DecideBatch`
sub-batches, `WriteMerged`/`DeleteMerged` internals, and the
component-level (structure/mindmap) per-call pools are all unified into
it via an injected submitter. No more per-job short-lived goroutines in
`runCompilerJobs` (futures are collected then awaited on the caller).
Fan-out stays bounded by the pool worker count; these stages are
docengine-bounded / LLM-bounded, not CPU-bounded.
- **DocEngine-only deletion.** `Consumer.processBatch` deletion no
longer loads the deleted docs' products into memory. Two sequential
DocEngine calls replace the old in-memory surgery:
- `DeleteDocLevelForDocs` — one `DeleteChunks` over `doc_id IN
deletedDocIDs` (merged rows carry `doc_id == kb`, so only per-doc
products match).
- `StripMergedSources` — one `Search` of `kc_merged=1` rows filtered by
`source_doc_ids IN deletedDocIDs` (intersection pushed down to the
engine), `UpdateChunks` the source array of survivors, and
`DeleteChunks` the rows whose array became empty.

## Changes

- `internal/ingestion/knowledge_compile/pool.go` (new): global
`compilerPool` +
`runCompilerJobs`/`SubmitCompilerJob`/`SubmitCompilerJobs`.
- `internal/ingestion/knowledge_compile/consumer.go`: deletion rewritten
to the two DocEngine calls;
`mergedBase`/`toDelete`/`stripDeletedSources` removed.
- `internal/ingestion/knowledge_compile/writer.go`:
`DeleteDocLevelForDocs` + `StripMergedSources` replace
`DeleteMergedForDoc`/`DeleteMerged`.
- `internal/ingestion/knowledge_compile/reader.go`: drop
`LoadMergedBySourceDoc` + `containsString` (keep `LoadDocProducts` for
the completion branch).
- `internal/ingestion/knowledge_compile/dedup.go`: `NewLLMDeduper` takes
`llmMaxTokens`; wires `SetMaxBatchTokens`/`SetSubmitter`.
- `internal/ingestion/knowledge_compiler/{structure,merge}.go`,
`mindmap/mindmap.go`, `pool_wiring.go`: token-budget split + submitter
injection.
- Tests: `structure_test.go` (token-budget split), `dedup_test.go`,
`consumer_test.go` (tombstone + DocEngine deletion assertions) updated.

## Validation

`bash build.sh --test -race ./internal/ingestion/knowledge_compile/...
./internal/ingestion/component/knowledge_compiler/...` passes (unit
tier, no external services).

🤖 Generated with [CodeBuddy](https://www.codebuddy.ai)

---------

Co-authored-by: yuzhichang <yuzhichang@infiniflow.ai>
2026-08-02 17:06:29 +08:00

123 lines
3.6 KiB
Go

//
// Copyright 2026 The InfiniFlow Authors. All Rights Reserved.
//
// Licensed 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.
package knowledge_compile
import (
"context"
"testing"
kccommon "ragflow/internal/ingestion/component/knowledge_compiler/common"
)
// recordingDeduper is a fake Deduper that records the Decide/DecideBatch calls
// and applies a simple policy: a candidate whose content is "dup" is merged
// into the existing row, otherwise it stays distinct.
type recordingDeduper struct {
noopDeduper
decideBatchCalls int
lastGroups []MergeGroup
}
func (r *recordingDeduper) DecideBatch(ctx context.Context, groups []MergeGroup) ([]MergeGroup, error) {
r.decideBatchCalls++
r.lastGroups = groups
for gi := range groups {
existing := groups[gi].Existing
var distinct []kccommon.Product
dup := false
for _, cand := range groups[gi].Candidates {
if stringOf(cand.Content) == "dup" {
merged := existing
merged.Content = "merged-" + existing.Content
existing = merged
dup = true
continue
}
c := cand
c.Merged = true
c.DocID = existing.DocID
distinct = append(distinct, c)
}
groups[gi].Merged = existing
groups[gi].Duplicate = dup
groups[gi].Distinct = distinct
}
return groups, nil
}
func stringOf(s any) string {
if v, ok := s.(string); ok {
return v
}
return ""
}
func TestDeduperDecideBatchFoldsGroups(t *testing.T) {
d := &recordingDeduper{}
existing := kccommon.Product{ID: "row-1", DocID: "kb1", Content: "base"}
groups := []MergeGroup{
{
Existing: existing,
Candidates: []kccommon.Product{{ID: "a", Content: "dup"}, {ID: "b", Content: "fresh"}},
Score: 1.0,
},
}
out, err := d.DecideBatch(context.Background(), groups)
if err != nil {
t.Fatalf("DecideBatch: %v", err)
}
if d.decideBatchCalls != 1 {
t.Errorf("expected exactly one DecideBatch call, got %d", d.decideBatchCalls)
}
g := out[0]
if !g.Duplicate {
t.Errorf("group with a dup candidate should report Duplicate=true")
}
if g.Merged.Content != "merged-base" {
t.Errorf("merged content = %q, want merged-base", g.Merged.Content)
}
if len(g.Distinct) != 1 || stringOf(g.Distinct[0].Content) != "fresh" {
t.Errorf("distinct = %v, want one fresh candidate", g.Distinct)
}
if g.Distinct[0].DocID != "kb1" || !g.Distinct[0].Merged {
t.Errorf("distinct candidate should be a new merged row under the KB")
}
}
func TestNoopDeduperDecideBatchNoMerge(t *testing.T) {
d := NewNoopDeduper()
existing := kccommon.Product{ID: "row-1", DocID: "kb1", Content: "base"}
groups := []MergeGroup{
{Existing: existing, Candidates: []kccommon.Product{{ID: "a", Content: "x"}, {ID: "b", Content: "y"}}},
}
out, err := d.DecideBatch(context.Background(), groups)
if err != nil {
t.Fatalf("DecideBatch: %v", err)
}
g := out[0]
if g.Duplicate {
t.Errorf("noop deduper must never merge")
}
if len(g.Distinct) != 2 {
t.Fatalf("noop deduper should report both candidates distinct, got %d", len(g.Distinct))
}
for _, c := range g.Distinct {
if c.DocID != "kb1" || !c.Merged {
t.Errorf("distinct candidate should be a new merged row: %+v", c)
}
}
}