this repo has no description
palomar: update hashtags ('tag') schema and remove text parse
+18
-29
+1
-1
search/post_schema.json
+1
-1
search/post_schema.json
···
59
59
"embed_img_alt_text": { "type": "text", "analyzer": "textIcu", "search_analyzer": "textIcuSearch", "copy_to": "everything" },
60
60
"self_label": { "type": "keyword", "normalizer": "default" },
61
61
62
-
"hashtag": { "type": "keyword", "normalizer": "default" },
62
+
"tag": { "type": "keyword", "normalizer": "default" },
63
63
"emoji": { "type": "keyword", "normalizer": "caseSensitive" },
64
64
65
65
"everything": { "type": "text", "analyzer": "textIcu", "search_analyzer": "textIcuSearch" },
+1
-1
search/profile_schema.json
+1
-1
search/profile_schema.json
···
50
50
"img_alt_text": { "type": "text", "analyzer": "textIcu", "search_analyzer": "textIcuSearch", "copy_to": "everything" },
51
51
"self_label": { "type": "keyword", "normalizer": "default" },
52
52
53
-
"hashtag": { "type": "keyword", "normalizer": "default" },
53
+
"tag": { "type": "keyword", "normalizer": "default" },
54
54
"emoji": { "type": "keyword", "normalizer": "caseSensitive" },
55
55
56
56
"has_avatar": { "type": "boolean" },
-1
search/testdata/transform-post-fixtures.json
-1
search/testdata/transform-post-fixtures.json
-1
search/testdata/transform-profile-fixtures.json
-1
search/testdata/transform-profile-fixtures.json
+16
-15
search/transform.go
+16
-15
search/transform.go
···
1
1
package search
2
2
3
3
import (
4
-
"regexp"
5
4
"strings"
6
5
"time"
7
6
···
20
19
Description *string `json:"description,omitempty"`
21
20
ImgAltText []string `json:"img_alt_text,omitempty"`
22
21
SelfLabel []string `json:"self_label,omitempty"`
23
-
Hashtag []string `json:"hashtag,omitempty"`
22
+
Tag []string `json:"tag,omitempty"`
24
23
Emoji []string `json:"emoji,omitempty"`
25
24
HasAvatar bool `json:"has_avatar"`
26
25
HasBanner bool `json:"has_banner"`
···
43
42
EmbedImgCount int `json:"embed_img_count"`
44
43
EmbedImgAltText []string `json:"embed_img_alt_text,omitempty"`
45
44
SelfLabel []string `json:"self_label,omitempty"`
46
-
Hashtag []string `json:"hashtag,omitempty"`
45
+
Tag []string `json:"tag,omitempty"`
47
46
Emoji []string `json:"emoji,omitempty"`
48
47
}
49
48
···
64
63
func TransformProfile(profile *appbsky.ActorProfile, ident *identity.Identity, cid string) ProfileDoc {
65
64
// TODO: placeholder for future alt text on profile blobs
66
65
var altText []string
67
-
var hashtags []string
66
+
var tags []string
68
67
var emojis []string
69
68
if profile.Description != nil {
70
-
hashtags = parseHashtags(*profile.Description)
69
+
tags = parseProfileTags(profile)
71
70
emojis = parseEmojis(*profile.Description)
72
71
}
73
72
var selfLabels []string
···
89
88
Description: profile.Description,
90
89
ImgAltText: altText,
91
90
SelfLabel: selfLabels,
92
-
Hashtag: hashtags,
91
+
Tag: tags,
93
92
Emoji: emojis,
94
93
HasAvatar: profile.Avatar != nil,
95
94
HasBanner: profile.Banner != nil,
···
173
172
EmbedImgCount: embedImgCount,
174
173
EmbedImgAltText: embedImgAltText,
175
174
SelfLabel: selfLabels,
176
-
Hashtag: parseHashtags(post.Text),
175
+
Tag: parsePostTags(post),
177
176
Emoji: parseEmojis(post.Text),
178
177
}
179
178
···
184
183
return doc
185
184
}
186
185
187
-
func parseHashtags(s string) []string {
188
-
var hashtagRegex = regexp.MustCompile(`\B#([A-Za-z]+)\b`)
186
+
func parseProfileTags(p *appbsky.ActorProfile) []string {
187
+
// TODO: waiting for profile tag lexicon support
189
188
var ret []string = []string{}
190
-
seen := make(map[string]bool)
191
-
for _, m := range hashtagRegex.FindAllStringSubmatch(s, -1) {
192
-
if seen[m[1]] == false {
193
-
ret = append(ret, m[1])
194
-
seen[m[1]] = true
195
-
}
189
+
if len(ret) == 0 {
190
+
return nil
196
191
}
192
+
return ret
193
+
}
194
+
195
+
func parsePostTags(p *appbsky.FeedPost) []string {
196
+
// TODO: waiting for post tag lexicon support (indigo codegen)
197
+
var ret []string = []string{}
197
198
if len(ret) == 0 {
198
199
return nil
199
200
}
-10
search/transform_test.go
-10
search/transform_test.go
···
13
13
"github.com/stretchr/testify/assert"
14
14
)
15
15
16
-
func TestParseHashtags(t *testing.T) {
17
-
assert := assert.New(t)
18
-
19
-
assert.Equal(parseHashtags("#basic post with #HashTag #examples"), []string{"basic", "HashTag", "examples"})
20
-
assert.Equal(parseHashtags("#dedupe #dedupe"), []string{"dedupe"})
21
-
assert.Equal(parseHashtags("##double"), []string{"double"})
22
-
assert.Equal(parseHashtags("#with-punc"), []string{"with"})
23
-
assert.True(parseHashtags("not https://example.com/thing#fragment") == nil)
24
-
}
25
-
26
16
func TestParseEmojis(t *testing.T) {
27
17
assert := assert.New(t)
28
18