boltless.me
Signed-off-by: Seongmin Lee boltlessengineer@proton.me
+88
-4
appview/indexer/issues/indexer.go
+88
-4
appview/indexer/issues/indexer.go
···
1
+
// heavily inspired by gitea's model (basically copy-pasted)
1
2
package issues_indexer
2
3
3
4
import (
···
6
7
"os"
7
8
8
9
"github.com/blevesearch/bleve/v2"
10
+
"github.com/blevesearch/bleve/v2/analysis/analyzer/custom"
11
+
"github.com/blevesearch/bleve/v2/analysis/token/camelcase"
12
+
"github.com/blevesearch/bleve/v2/analysis/token/lowercase"
13
+
"github.com/blevesearch/bleve/v2/analysis/token/unicodenorm"
14
+
"github.com/blevesearch/bleve/v2/analysis/tokenizer/unicode"
9
15
"github.com/blevesearch/bleve/v2/index/upsidedown"
16
+
"github.com/blevesearch/bleve/v2/mapping"
10
17
"github.com/blevesearch/bleve/v2/search/query"
11
18
"tangled.org/core/appview/db"
12
19
"tangled.org/core/appview/indexer/base36"
···
16
23
tlog "tangled.org/core/log"
17
24
)
18
25
26
+
const (
27
+
issueIndexerAnalyzer = "issueIndexer"
28
+
issueIndexerDocType = "issueIndexerDocType"
29
+
30
+
unicodeNormalizeName = "uicodeNormalize"
31
+
)
32
+
19
33
type Indexer struct {
20
34
indexer bleve.Index
21
35
path string
···
44
58
l.Info("Initialized the issue indexer")
45
59
}
46
60
61
+
func generateIssueIndexMapping() (mapping.IndexMapping, error) {
62
+
mapping := bleve.NewIndexMapping()
63
+
docMapping := bleve.NewDocumentMapping()
64
+
65
+
textFieldMapping := bleve.NewTextFieldMapping()
66
+
textFieldMapping.Store = false
67
+
textFieldMapping.IncludeInAll = false
68
+
69
+
boolFieldMapping := bleve.NewBooleanFieldMapping()
70
+
boolFieldMapping.Store = false
71
+
boolFieldMapping.IncludeInAll = false
72
+
73
+
keywordFieldMapping := bleve.NewKeywordFieldMapping()
74
+
keywordFieldMapping.Store = false
75
+
keywordFieldMapping.IncludeInAll = false
76
+
77
+
// numericFieldMapping := bleve.NewNumericFieldMapping()
78
+
79
+
docMapping.AddFieldMappingsAt("title", textFieldMapping)
80
+
docMapping.AddFieldMappingsAt("body", textFieldMapping)
81
+
82
+
docMapping.AddFieldMappingsAt("repo_at", keywordFieldMapping)
83
+
docMapping.AddFieldMappingsAt("is_open", boolFieldMapping)
84
+
85
+
err := mapping.AddCustomTokenFilter(unicodeNormalizeName, map[string]any{
86
+
"type": unicodenorm.Name,
87
+
"form": unicodenorm.NFC,
88
+
})
89
+
if err != nil {
90
+
return nil, err
91
+
}
92
+
93
+
err = mapping.AddCustomAnalyzer(issueIndexerAnalyzer, map[string]any{
94
+
"type": custom.Name,
95
+
"char_filters": []string{},
96
+
"tokenizer": unicode.Name,
97
+
"token_filters": []string{unicodeNormalizeName, camelcase.Name, lowercase.Name},
98
+
})
99
+
if err != nil {
100
+
return nil, err
101
+
}
102
+
103
+
mapping.DefaultAnalyzer = issueIndexerAnalyzer
104
+
mapping.AddDocumentMapping(issueIndexerDocType, docMapping)
105
+
mapping.AddDocumentMapping("_all", bleve.NewDocumentDisabledMapping())
106
+
mapping.DefaultMapping = bleve.NewDocumentDisabledMapping()
107
+
108
+
return mapping, nil
109
+
}
110
+
47
111
func (ix *Indexer) intialize(ctx context.Context) (bool, error) {
48
112
if ix.indexer != nil {
49
113
return false, errors.New("indexer is already initialized")
···
58
122
return true, nil
59
123
}
60
124
61
-
mapping := bleve.NewIndexMapping()
125
+
mapping, err := generateIssueIndexMapping()
126
+
if err != nil {
127
+
return false, err
128
+
}
62
129
indexer, err = bleve.New(ix.path, mapping)
63
130
if err != nil {
64
131
return false, err
···
94
161
for _, issue := range issues {
95
162
dataList = append(dataList, &IssueData{
96
163
ID: issue.Id,
164
+
RepoAt: issue.RepoAt.String(),
97
165
IssueID: issue.IssueId,
98
166
Title: issue.Title,
99
167
Body: issue.Body,
···
114
182
// IssueData data stored and will be indexed
115
183
type IssueData struct {
116
184
ID int64 `json:"id"`
185
+
RepoAt string `json:"repo_at"`
117
186
IssueID int `json:"issue_id"`
118
187
Title string `json:"title"`
119
188
Body string `json:"body"`
···
122
191
Comments []IssueCommentData `json:"comments"`
123
192
}
124
193
194
+
// Type returns the document type, for bleve's mapping.Classifier interface.
195
+
func (i *IssueData) Type() string {
196
+
return issueIndexerDocType
197
+
}
198
+
199
+
125
200
type IssueCommentData struct {
126
201
Body string `json:"body"`
127
202
}
···
149
224
150
225
if opts.Keyword != "" {
151
226
queries = append(queries, bleve.NewDisjunctionQuery(
152
-
matchAndQuery(opts.Keyword, "title"),
153
-
matchAndQuery(opts.Keyword, "body"),
227
+
matchAndQuery(opts.Keyword, "title", issueIndexerAnalyzer, 0),
228
+
matchAndQuery(opts.Keyword, "body", issueIndexerAnalyzer, 0),
154
229
))
155
230
}
231
+
queries = append(queries, keywordFieldQuery(opts.RepoAt, "repo_at"))
156
232
queries = append(queries, boolFieldQuery(opts.IsOpen, "is_open"))
157
233
// TODO: append more queries
158
234
···
176
252
return ret, nil
177
253
}
178
254
179
-
func matchAndQuery(keyword, field string) query.Query {
255
+
func matchAndQuery(keyword, field, analyzer string, fuzziness int) query.Query {
180
256
q := bleve.NewMatchQuery(keyword)
181
257
q.FieldVal = field
258
+
q.Analyzer = analyzer
259
+
q.Fuzziness = fuzziness
182
260
return q
183
261
}
184
262
···
187
265
q.FieldVal = field
188
266
return q
189
267
}
268
+
269
+
func keywordFieldQuery(keyword, field string) query.Query {
270
+
q := bleve.NewTermQuery(keyword)
271
+
q.FieldVal = field
272
+
return q
273
+
}