back round 7 view raw
wip: add indexer mappings #495

merged
opened by
boltless.me 4 months ago targeting master from boltless.me/core: feat/search
Signed-off-by: Seongmin Lee boltlessengineer@proton.me
options
unified split
Changed files
+78 -4
appview
indexer
issues
indexer.go
+78 -4
appview/indexer/issues/indexer.go
···

       7
       7
        
       	"os"

     

       8
       8
        
       

     

       9
       9
        
       	"github.com/blevesearch/bleve/v2"

     

       10
       10
       +
       	"github.com/blevesearch/bleve/v2/analysis/analyzer/custom"

     

       11
       11
       +
       	"github.com/blevesearch/bleve/v2/analysis/token/camelcase"

     

       12
       12
       +
       	"github.com/blevesearch/bleve/v2/analysis/token/lowercase"

     

       13
       13
       +
       	"github.com/blevesearch/bleve/v2/analysis/token/unicodenorm"

     

       14
       14
       +
       	"github.com/blevesearch/bleve/v2/analysis/tokenizer/unicode"

     

       10
       15
        
       	"github.com/blevesearch/bleve/v2/index/upsidedown"

     

       16
       16
       +
       	"github.com/blevesearch/bleve/v2/mapping"

     

       11
       17
        
       	"github.com/blevesearch/bleve/v2/search/query"

     

       12
       18
        
       	"tangled.org/core/appview/db"

     

       13
       19
        
       	"tangled.org/core/appview/indexer/base36"

     
···

       17
       23
        
       	tlog "tangled.org/core/log"

     

       18
       24
        
       )

     

       19
       25
        
       

     

       26
       26
       +
       const (

     

       27
       27
       +
       	issueIndexerAnalyzer = "issueIndexer"

     

       28
       28
       +
       	issueIndexerDocType  = "issueIndexerDocType"

     

       29
       29
       +
       

     

       30
       30
       +
       	unicodeNormalizeName = "uicodeNormalize"

     

       31
       31
       +
       )

     

       32
       32
       +
       

     

       20
       33
        
       type Indexer struct {

     

       21
       34
        
       	indexer bleve.Index

     

       22
       35
        
       	path    string

     
···

       45
       58
        
       	l.Info("Initialized the issue indexer")

     

       46
       59
        
       }

     

       47
       60
        
       

     

       61
       61
       +
       func generateIssueIndexMapping() (mapping.IndexMapping, error) {

     

       62
       62
       +
       	mapping := bleve.NewIndexMapping()

     

       63
       63
       +
       	docMapping := bleve.NewDocumentMapping()

     

       64
       64
       +
       

     

       65
       65
       +
       	textFieldMapping := bleve.NewTextFieldMapping()

     

       66
       66
       +
       	textFieldMapping.Store = false

     

       67
       67
       +
       	textFieldMapping.IncludeInAll = false

     

       68
       68
       +
       

     

       69
       69
       +
       	boolFieldMapping := bleve.NewBooleanFieldMapping()

     

       70
       70
       +
       	boolFieldMapping.Store = false

     

       71
       71
       +
       	boolFieldMapping.IncludeInAll = false

     

       72
       72
       +
       

     

       73
       73
       +
       	keywordFieldMapping := bleve.NewKeywordFieldMapping()

     

       74
       74
       +
       	keywordFieldMapping.Store = false

     

       75
       75
       +
       	keywordFieldMapping.IncludeInAll = false

     

       76
       76
       +
       

     

       77
       77
       +
       	// numericFieldMapping := bleve.NewNumericFieldMapping()

     

       78
       78
       +
       

     

       79
       79
       +
       	docMapping.AddFieldMappingsAt("title", textFieldMapping)

     

       80
       80
       +
       	docMapping.AddFieldMappingsAt("body", textFieldMapping)

     

       81
       81
       +
       

     

       82
       82
       +
       	docMapping.AddFieldMappingsAt("repo_at", keywordFieldMapping)

     

       83
       83
       +
       	docMapping.AddFieldMappingsAt("is_open", boolFieldMapping)

     

       84
       84
       +
       

     

       85
       85
       +
       	err := mapping.AddCustomTokenFilter(unicodeNormalizeName, map[string]any{

     

       86
       86
       +
       		"type": unicodenorm.Name,

     

       87
       87
       +
       		"form": unicodenorm.NFC,

     

       88
       88
       +
       	})

     

       89
       89
       +
       	if err != nil {

     

       90
       90
       +
       		return nil, err

     

       91
       91
       +
       	}

     

       92
       92
       +
       

     

       93
       93
       +
       	err = mapping.AddCustomAnalyzer(issueIndexerAnalyzer, map[string]any{

     

       94
       94
       +
       		"type":          custom.Name,

     

       95
       95
       +
       		"char_filters":  []string{},

     

       96
       96
       +
       		"tokenizer":     unicode.Name,

     

       97
       97
       +
       		"token_filters": []string{unicodeNormalizeName, camelcase.Name, lowercase.Name},

     

       98
       98
       +
       	})

     

       99
       99
       +
       	if err != nil {

     

       100
       100
       +
       		return nil, err

     

       101
       101
       +
       	}

     

       102
       102
       +
       

     

       103
       103
       +
       	mapping.DefaultAnalyzer = issueIndexerAnalyzer

     

       104
       104
       +
       	mapping.AddDocumentMapping(issueIndexerDocType, docMapping)

     

       105
       105
       +
       	mapping.AddDocumentMapping("_all", bleve.NewDocumentDisabledMapping())

     

       106
       106
       +
       	mapping.DefaultMapping = bleve.NewDocumentDisabledMapping()

     

       107
       107
       +
       

     

       108
       108
       +
       	return mapping, nil

     

       109
       109
       +
       }

     

       110
       110
       +
       

     

       48
       111
        
       func (ix *Indexer) intialize(ctx context.Context) (bool, error) {

     

       49
       112
        
       	if ix.indexer != nil {

     

       50
       113
        
       		return false, errors.New("indexer is already initialized")

     
···

       59
       122
        
       		return true, nil

     

       60
       123
        
       	}

     

       61
       124
        
       

     

       62
       62
       -
       	mapping := bleve.NewIndexMapping()

     

       125
       125
       +
       	mapping, err := generateIssueIndexMapping()

     

       126
       126
       +
       	if err != nil {

     

       127
       127
       +
       		return false, err

     

       128
       128
       +
       	}

     

       63
       129
        
       	indexer, err = bleve.New(ix.path, mapping)

     

       64
       130
        
       	if err != nil {

     

       65
       131
        
       		return false, err

     
···

       122
       188
        
       	}

     

       123
       189
        
       }

     

       124
       190
        
       

     

       191
       191
       +
       // Type returns the document type, for bleve's mapping.Classifier interface.

     

       192
       192
       +
       func (i *issueData) Type() string {

     

       193
       193
       +
       	return issueIndexerDocType

     

       194
       194
       +
       }

     

       195
       195
       +
       

     

       196
       196
       +
       

     

       125
       197
        
       type IssueCommentData struct {

     

       126
       198
        
       	Body string `json:"body"`

     

       127
       199
        
       }

     
···

       150
       222
        
       

     

       151
       223
        
       	if opts.Keyword != "" {

     

       152
       224
        
       		queries = append(queries, bleve.NewDisjunctionQuery(

     

       153
       153
       -
       			matchAndQuery(opts.Keyword, "title"),

     

       154
       154
       -
       			matchAndQuery(opts.Keyword, "body"),

     

       225
       225
       +
       			matchAndQuery(opts.Keyword, "title", issueIndexerAnalyzer, 0),

     

       226
       226
       +
       			matchAndQuery(opts.Keyword, "body", issueIndexerAnalyzer, 0),

     

       155
       227
        
       		))

     

       156
       228
        
       	}

     

       157
       229
        
       	queries = append(queries, keywordFieldQuery(opts.RepoAt, "repo_at"))

     
···

       178
       250
        
       	return ret, nil

     

       179
       251
        
       }

     

       180
       252
        
       

     

       181
       181
       -
       func matchAndQuery(keyword, field string) query.Query {

     

       253
       253
       +
       func matchAndQuery(keyword, field, analyzer string, fuzziness int) query.Query {

     

       182
       254
        
       	q := bleve.NewMatchQuery(keyword)

     

       183
       255
        
       	q.FieldVal = field

     

       256
       256
       +
       	q.Analyzer = analyzer

     

       257
       257
       +
       	q.Fuzziness = fuzziness

     

       184
       258
        
       	return q

     

       185
       259
        
       }

     

       186
       260