code.gitea.io/gitea@v1.22.3/modules/indexer/issues/bleve/bleve.go (about)

     1  // Copyright 2018 The Gitea Authors. All rights reserved.
     2  // SPDX-License-Identifier: MIT
     3  
     4  package bleve
     5  
     6  import (
     7  	"context"
     8  
     9  	indexer_internal "code.gitea.io/gitea/modules/indexer/internal"
    10  	inner_bleve "code.gitea.io/gitea/modules/indexer/internal/bleve"
    11  	"code.gitea.io/gitea/modules/indexer/issues/internal"
    12  
    13  	"github.com/blevesearch/bleve/v2"
    14  	"github.com/blevesearch/bleve/v2/analysis/analyzer/custom"
    15  	"github.com/blevesearch/bleve/v2/analysis/token/camelcase"
    16  	"github.com/blevesearch/bleve/v2/analysis/token/lowercase"
    17  	"github.com/blevesearch/bleve/v2/analysis/token/unicodenorm"
    18  	"github.com/blevesearch/bleve/v2/analysis/tokenizer/unicode"
    19  	"github.com/blevesearch/bleve/v2/mapping"
    20  	"github.com/blevesearch/bleve/v2/search/query"
    21  )
    22  
    23  const (
    24  	issueIndexerAnalyzer      = "issueIndexer"
    25  	issueIndexerDocType       = "issueIndexerDocType"
    26  	issueIndexerLatestVersion = 4
    27  )
    28  
    29  const unicodeNormalizeName = "unicodeNormalize"
    30  
    31  func addUnicodeNormalizeTokenFilter(m *mapping.IndexMappingImpl) error {
    32  	return m.AddCustomTokenFilter(unicodeNormalizeName, map[string]any{
    33  		"type": unicodenorm.Name,
    34  		"form": unicodenorm.NFC,
    35  	})
    36  }
    37  
    38  const maxBatchSize = 16
    39  
    40  // IndexerData an update to the issue indexer
    41  type IndexerData internal.IndexerData
    42  
    43  // Type returns the document type, for bleve's mapping.Classifier interface.
    44  func (i *IndexerData) Type() string {
    45  	return issueIndexerDocType
    46  }
    47  
    48  // generateIssueIndexMapping generates the bleve index mapping for issues
    49  func generateIssueIndexMapping() (mapping.IndexMapping, error) {
    50  	mapping := bleve.NewIndexMapping()
    51  	docMapping := bleve.NewDocumentMapping()
    52  
    53  	numericFieldMapping := bleve.NewNumericFieldMapping()
    54  	numericFieldMapping.Store = false
    55  	numericFieldMapping.IncludeInAll = false
    56  	docMapping.AddFieldMappingsAt("repo_id", numericFieldMapping)
    57  
    58  	textFieldMapping := bleve.NewTextFieldMapping()
    59  	textFieldMapping.Store = false
    60  	textFieldMapping.IncludeInAll = false
    61  
    62  	boolFieldMapping := bleve.NewBooleanFieldMapping()
    63  	boolFieldMapping.Store = false
    64  	boolFieldMapping.IncludeInAll = false
    65  
    66  	numberFieldMapping := bleve.NewNumericFieldMapping()
    67  	numberFieldMapping.Store = false
    68  	numberFieldMapping.IncludeInAll = false
    69  
    70  	docMapping.AddFieldMappingsAt("is_public", boolFieldMapping)
    71  
    72  	docMapping.AddFieldMappingsAt("title", textFieldMapping)
    73  	docMapping.AddFieldMappingsAt("content", textFieldMapping)
    74  	docMapping.AddFieldMappingsAt("comments", textFieldMapping)
    75  
    76  	docMapping.AddFieldMappingsAt("is_pull", boolFieldMapping)
    77  	docMapping.AddFieldMappingsAt("is_closed", boolFieldMapping)
    78  	docMapping.AddFieldMappingsAt("label_ids", numberFieldMapping)
    79  	docMapping.AddFieldMappingsAt("no_label", boolFieldMapping)
    80  	docMapping.AddFieldMappingsAt("milestone_id", numberFieldMapping)
    81  	docMapping.AddFieldMappingsAt("project_id", numberFieldMapping)
    82  	docMapping.AddFieldMappingsAt("project_board_id", numberFieldMapping)
    83  	docMapping.AddFieldMappingsAt("poster_id", numberFieldMapping)
    84  	docMapping.AddFieldMappingsAt("assignee_id", numberFieldMapping)
    85  	docMapping.AddFieldMappingsAt("mention_ids", numberFieldMapping)
    86  	docMapping.AddFieldMappingsAt("reviewed_ids", numberFieldMapping)
    87  	docMapping.AddFieldMappingsAt("review_requested_ids", numberFieldMapping)
    88  	docMapping.AddFieldMappingsAt("subscriber_ids", numberFieldMapping)
    89  	docMapping.AddFieldMappingsAt("updated_unix", numberFieldMapping)
    90  
    91  	docMapping.AddFieldMappingsAt("created_unix", numberFieldMapping)
    92  	docMapping.AddFieldMappingsAt("deadline_unix", numberFieldMapping)
    93  	docMapping.AddFieldMappingsAt("comment_count", numberFieldMapping)
    94  
    95  	if err := addUnicodeNormalizeTokenFilter(mapping); err != nil {
    96  		return nil, err
    97  	} else if err = mapping.AddCustomAnalyzer(issueIndexerAnalyzer, map[string]any{
    98  		"type":          custom.Name,
    99  		"char_filters":  []string{},
   100  		"tokenizer":     unicode.Name,
   101  		"token_filters": []string{unicodeNormalizeName, camelcase.Name, lowercase.Name},
   102  	}); err != nil {
   103  		return nil, err
   104  	}
   105  
   106  	mapping.DefaultAnalyzer = issueIndexerAnalyzer
   107  	mapping.AddDocumentMapping(issueIndexerDocType, docMapping)
   108  	mapping.AddDocumentMapping("_all", bleve.NewDocumentDisabledMapping())
   109  	mapping.DefaultMapping = bleve.NewDocumentDisabledMapping() // disable default mapping, avoid indexing unexpected structs
   110  
   111  	return mapping, nil
   112  }
   113  
   114  var _ internal.Indexer = &Indexer{}
   115  
   116  // Indexer implements Indexer interface
   117  type Indexer struct {
   118  	inner                    *inner_bleve.Indexer
   119  	indexer_internal.Indexer // do not composite inner_bleve.Indexer directly to avoid exposing too much
   120  }
   121  
   122  // NewIndexer creates a new bleve local indexer
   123  func NewIndexer(indexDir string) *Indexer {
   124  	inner := inner_bleve.NewIndexer(indexDir, issueIndexerLatestVersion, generateIssueIndexMapping)
   125  	return &Indexer{
   126  		Indexer: inner,
   127  		inner:   inner,
   128  	}
   129  }
   130  
   131  // Index will save the index data
   132  func (b *Indexer) Index(_ context.Context, issues ...*internal.IndexerData) error {
   133  	batch := inner_bleve.NewFlushingBatch(b.inner.Indexer, maxBatchSize)
   134  	for _, issue := range issues {
   135  		if err := batch.Index(indexer_internal.Base36(issue.ID), (*IndexerData)(issue)); err != nil {
   136  			return err
   137  		}
   138  	}
   139  	return batch.Flush()
   140  }
   141  
   142  // Delete deletes indexes by ids
   143  func (b *Indexer) Delete(_ context.Context, ids ...int64) error {
   144  	batch := inner_bleve.NewFlushingBatch(b.inner.Indexer, maxBatchSize)
   145  	for _, id := range ids {
   146  		if err := batch.Delete(indexer_internal.Base36(id)); err != nil {
   147  			return err
   148  		}
   149  	}
   150  	return batch.Flush()
   151  }
   152  
   153  // Search searches for issues by given conditions.
   154  // Returns the matching issue IDs
   155  func (b *Indexer) Search(ctx context.Context, options *internal.SearchOptions) (*internal.SearchResult, error) {
   156  	var queries []query.Query
   157  
   158  	if options.Keyword != "" {
   159  		fuzziness := 0
   160  		if options.IsFuzzyKeyword {
   161  			fuzziness = inner_bleve.GuessFuzzinessByKeyword(options.Keyword)
   162  		}
   163  
   164  		queries = append(queries, bleve.NewDisjunctionQuery([]query.Query{
   165  			inner_bleve.MatchPhraseQuery(options.Keyword, "title", issueIndexerAnalyzer, fuzziness),
   166  			inner_bleve.MatchPhraseQuery(options.Keyword, "content", issueIndexerAnalyzer, fuzziness),
   167  			inner_bleve.MatchPhraseQuery(options.Keyword, "comments", issueIndexerAnalyzer, fuzziness),
   168  		}...))
   169  	}
   170  
   171  	if len(options.RepoIDs) > 0 || options.AllPublic {
   172  		var repoQueries []query.Query
   173  		for _, repoID := range options.RepoIDs {
   174  			repoQueries = append(repoQueries, inner_bleve.NumericEqualityQuery(repoID, "repo_id"))
   175  		}
   176  		if options.AllPublic {
   177  			repoQueries = append(repoQueries, inner_bleve.BoolFieldQuery(true, "is_public"))
   178  		}
   179  		queries = append(queries, bleve.NewDisjunctionQuery(repoQueries...))
   180  	}
   181  
   182  	if options.IsPull.Has() {
   183  		queries = append(queries, inner_bleve.BoolFieldQuery(options.IsPull.Value(), "is_pull"))
   184  	}
   185  	if options.IsClosed.Has() {
   186  		queries = append(queries, inner_bleve.BoolFieldQuery(options.IsClosed.Value(), "is_closed"))
   187  	}
   188  
   189  	if options.NoLabelOnly {
   190  		queries = append(queries, inner_bleve.BoolFieldQuery(true, "no_label"))
   191  	} else {
   192  		if len(options.IncludedLabelIDs) > 0 {
   193  			var includeQueries []query.Query
   194  			for _, labelID := range options.IncludedLabelIDs {
   195  				includeQueries = append(includeQueries, inner_bleve.NumericEqualityQuery(labelID, "label_ids"))
   196  			}
   197  			queries = append(queries, bleve.NewConjunctionQuery(includeQueries...))
   198  		} else if len(options.IncludedAnyLabelIDs) > 0 {
   199  			var includeQueries []query.Query
   200  			for _, labelID := range options.IncludedAnyLabelIDs {
   201  				includeQueries = append(includeQueries, inner_bleve.NumericEqualityQuery(labelID, "label_ids"))
   202  			}
   203  			queries = append(queries, bleve.NewDisjunctionQuery(includeQueries...))
   204  		}
   205  		if len(options.ExcludedLabelIDs) > 0 {
   206  			var excludeQueries []query.Query
   207  			for _, labelID := range options.ExcludedLabelIDs {
   208  				q := bleve.NewBooleanQuery()
   209  				q.AddMustNot(inner_bleve.NumericEqualityQuery(labelID, "label_ids"))
   210  				excludeQueries = append(excludeQueries, q)
   211  			}
   212  			queries = append(queries, bleve.NewConjunctionQuery(excludeQueries...))
   213  		}
   214  	}
   215  
   216  	if len(options.MilestoneIDs) > 0 {
   217  		var milestoneQueries []query.Query
   218  		for _, milestoneID := range options.MilestoneIDs {
   219  			milestoneQueries = append(milestoneQueries, inner_bleve.NumericEqualityQuery(milestoneID, "milestone_id"))
   220  		}
   221  		queries = append(queries, bleve.NewDisjunctionQuery(milestoneQueries...))
   222  	}
   223  
   224  	if options.ProjectID.Has() {
   225  		queries = append(queries, inner_bleve.NumericEqualityQuery(options.ProjectID.Value(), "project_id"))
   226  	}
   227  	if options.ProjectBoardID.Has() {
   228  		queries = append(queries, inner_bleve.NumericEqualityQuery(options.ProjectBoardID.Value(), "project_board_id"))
   229  	}
   230  
   231  	if options.PosterID.Has() {
   232  		queries = append(queries, inner_bleve.NumericEqualityQuery(options.PosterID.Value(), "poster_id"))
   233  	}
   234  
   235  	if options.AssigneeID.Has() {
   236  		queries = append(queries, inner_bleve.NumericEqualityQuery(options.AssigneeID.Value(), "assignee_id"))
   237  	}
   238  
   239  	if options.MentionID.Has() {
   240  		queries = append(queries, inner_bleve.NumericEqualityQuery(options.MentionID.Value(), "mention_ids"))
   241  	}
   242  
   243  	if options.ReviewedID.Has() {
   244  		queries = append(queries, inner_bleve.NumericEqualityQuery(options.ReviewedID.Value(), "reviewed_ids"))
   245  	}
   246  	if options.ReviewRequestedID.Has() {
   247  		queries = append(queries, inner_bleve.NumericEqualityQuery(options.ReviewRequestedID.Value(), "review_requested_ids"))
   248  	}
   249  
   250  	if options.SubscriberID.Has() {
   251  		queries = append(queries, inner_bleve.NumericEqualityQuery(options.SubscriberID.Value(), "subscriber_ids"))
   252  	}
   253  
   254  	if options.UpdatedAfterUnix.Has() || options.UpdatedBeforeUnix.Has() {
   255  		queries = append(queries, inner_bleve.NumericRangeInclusiveQuery(
   256  			options.UpdatedAfterUnix,
   257  			options.UpdatedBeforeUnix,
   258  			"updated_unix"))
   259  	}
   260  
   261  	var indexerQuery query.Query = bleve.NewConjunctionQuery(queries...)
   262  	if len(queries) == 0 {
   263  		indexerQuery = bleve.NewMatchAllQuery()
   264  	}
   265  
   266  	skip, limit := indexer_internal.ParsePaginator(options.Paginator)
   267  	search := bleve.NewSearchRequestOptions(indexerQuery, limit, skip, false)
   268  
   269  	if options.SortBy == "" {
   270  		options.SortBy = internal.SortByCreatedAsc
   271  	}
   272  
   273  	search.SortBy([]string{string(options.SortBy), "-_id"})
   274  
   275  	result, err := b.inner.Indexer.SearchInContext(ctx, search)
   276  	if err != nil {
   277  		return nil, err
   278  	}
   279  
   280  	ret := &internal.SearchResult{
   281  		Total: int64(result.Total),
   282  		Hits:  make([]internal.Match, 0, len(result.Hits)),
   283  	}
   284  	for _, hit := range result.Hits {
   285  		id, err := indexer_internal.ParseBase36(hit.ID)
   286  		if err != nil {
   287  			return nil, err
   288  		}
   289  		ret.Hits = append(ret.Hits, internal.Match{
   290  			ID: id,
   291  		})
   292  	}
   293  	return ret, nil
   294  }