Fix issues indexer document mapping (#25619)

Fix regression of #5363 (so long ago).

The old code definded a document mapping for `issueIndexerDocType`, and
assigned it to `BleveIndexerData` as its type. (`BleveIndexerData` has
been renamed to `IndexerData` in #25174, but nothing more.) But the old
code never used `BleveIndexerData`, it wrote the index with an anonymous
struct type. Nonetheless, bleve would use the default auto-mapping for
struct it didn't know, so the indexer still worked. This means the
custom document mapping was always dead code.

The custom document mapping is not useless, it can reduce index storage,
this PR brings it back and disable default mapping to prevent it from
happening again. Since `IndexerData`(`BleveIndexerData`) has JSON tags,
and bleve uses them first, so we should use `repo_id` as the field name
instead of `RepoID`.

I did a test to compare the storage size before and after this, with
about 3k real comments that were migrated from some public repos.

Before:

```text
[ 160]  .
├── [  42]  index_meta.json
├── [  13]  rupture_meta.json
└── [ 128]  store
    ├── [6.9M]  00000000005d.zap
    └── [256K]  root.bolt
```

After:

```text
[ 160]  .
├── [  42]  index_meta.json
├── [  13]  rupture_meta.json
└── [ 128]  store
    ├── [3.5M]  000000000065.zap
    └── [256K]  root.bolt
```

It saves about half the storage space.

---------

Co-authored-by: Giteabot <teabot@gitea.io>
This commit is contained in:
Jason Song 2023-07-04 17:05:28 +08:00 committed by GitHub
parent dae022ab2a
commit 9958642502
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23

View file

@ -23,7 +23,7 @@ import (
const (
issueIndexerAnalyzer = "issueIndexer"
issueIndexerDocType = "issueIndexerDocType"
issueIndexerLatestVersion = 2
issueIndexerLatestVersion = 3
)
// numericEqualityQuery a numeric equality query for the given value and field
@ -67,15 +67,16 @@ func generateIssueIndexMapping() (mapping.IndexMapping, error) {
docMapping := bleve.NewDocumentMapping()
numericFieldMapping := bleve.NewNumericFieldMapping()
numericFieldMapping.Store = false
numericFieldMapping.IncludeInAll = false
docMapping.AddFieldMappingsAt("RepoID", numericFieldMapping)
docMapping.AddFieldMappingsAt("repo_id", numericFieldMapping)
textFieldMapping := bleve.NewTextFieldMapping()
textFieldMapping.Store = false
textFieldMapping.IncludeInAll = false
docMapping.AddFieldMappingsAt("Title", textFieldMapping)
docMapping.AddFieldMappingsAt("Content", textFieldMapping)
docMapping.AddFieldMappingsAt("Comments", textFieldMapping)
docMapping.AddFieldMappingsAt("title", textFieldMapping)
docMapping.AddFieldMappingsAt("content", textFieldMapping)
docMapping.AddFieldMappingsAt("comments", textFieldMapping)
if err := addUnicodeNormalizeTokenFilter(mapping); err != nil {
return nil, err
@ -91,6 +92,7 @@ func generateIssueIndexMapping() (mapping.IndexMapping, error) {
mapping.DefaultAnalyzer = issueIndexerAnalyzer
mapping.AddDocumentMapping(issueIndexerDocType, docMapping)
mapping.AddDocumentMapping("_all", bleve.NewDocumentDisabledMapping())
mapping.DefaultMapping = bleve.NewDocumentDisabledMapping() // disable default mapping, avoid indexing unexpected structs
return mapping, nil
}
@ -116,17 +118,7 @@ func NewIndexer(indexDir string) *Indexer {
func (b *Indexer) Index(_ context.Context, issues []*internal.IndexerData) error {
batch := inner_bleve.NewFlushingBatch(b.inner.Indexer, maxBatchSize)
for _, issue := range issues {
if err := batch.Index(indexer_internal.Base36(issue.ID), struct {
RepoID int64
Title string
Content string
Comments []string
}{
RepoID: issue.RepoID,
Title: issue.Title,
Content: issue.Content,
Comments: issue.Comments,
}); err != nil {
if err := batch.Index(indexer_internal.Base36(issue.ID), (*IndexerData)(issue)); err != nil {
return err
}
}
@ -149,7 +141,7 @@ func (b *Indexer) Delete(_ context.Context, ids ...int64) error {
func (b *Indexer) Search(ctx context.Context, keyword string, repoIDs []int64, limit, start int) (*internal.SearchResult, error) {
var repoQueriesP []*query.NumericRangeQuery
for _, repoID := range repoIDs {
repoQueriesP = append(repoQueriesP, numericEqualityQuery(repoID, "RepoID"))
repoQueriesP = append(repoQueriesP, numericEqualityQuery(repoID, "repo_id"))
}
repoQueries := make([]query.Query, len(repoQueriesP))
for i, v := range repoQueriesP {
@ -159,9 +151,9 @@ func (b *Indexer) Search(ctx context.Context, keyword string, repoIDs []int64, l
indexerQuery := bleve.NewConjunctionQuery(
bleve.NewDisjunctionQuery(repoQueries...),
bleve.NewDisjunctionQuery(
newMatchPhraseQuery(keyword, "Title", issueIndexerAnalyzer),
newMatchPhraseQuery(keyword, "Content", issueIndexerAnalyzer),
newMatchPhraseQuery(keyword, "Comments", issueIndexerAnalyzer),
newMatchPhraseQuery(keyword, "title", issueIndexerAnalyzer),
newMatchPhraseQuery(keyword, "content", issueIndexerAnalyzer),
newMatchPhraseQuery(keyword, "comments", issueIndexerAnalyzer),
))
search := bleve.NewSearchRequestOptions(indexerQuery, limit, start, false)
search.SortBy([]string{"-_score"})