Skip to content

Commit 5d10f1e

Browse files
committed
fix mandarin search
1 parent 55ac596 commit 5d10f1e

File tree

7 files changed

+8261
-5205
lines changed

7 files changed

+8261
-5205
lines changed

.gitignore

+3-1
Original file line numberDiff line numberDiff line change
@@ -30,4 +30,6 @@ yarn-error.log*
3030
# others
3131
.env*.local
3232
.vercel
33-
next-env.d.ts
33+
next-env.d.ts
34+
35+
.yarn/install-state.gz
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,85 @@
1+
diff --git a/dist/commonjs/index.js b/dist/commonjs/index.js
2+
index 6a3f12542f84448bc9c3bd90e7a494ac9ca863a2..ddd907c66d09c2cb2118e14b5ffd488d11965896 100644
3+
--- a/dist/commonjs/index.js
4+
+++ b/dist/commonjs/index.js
5+
@@ -1,7 +1,9 @@
6+
"use strict";
7+
Object.defineProperty(exports, "__esModule", { value: true });
8+
const japanese_js_1 = require("./japanese.js");
9+
+const mandarin_js_1 = require("./mandarin.js");
10+
exports.default = {
11+
japanese: japanese_js_1.createTokenizer,
12+
+ mandarin: mandarin_js_1.createTokenizer,
13+
};
14+
//# sourceMappingURL=index.js.map
15+
diff --git a/dist/commonjs/mandarin.d.ts b/dist/commonjs/mandarin.d.ts
16+
index a5c44e17802c6b4d17203e59cf8563120751a193..b3910a611bdff11a800b56eac173e3d06c1110ff 100644
17+
--- a/dist/commonjs/mandarin.d.ts
18+
+++ b/dist/commonjs/mandarin.d.ts
19+
@@ -1,9 +1,9 @@
20+
import type { DefaultTokenizer, DefaultTokenizerConfig } from "@orama/orama";
21+
-declare const tokenizerLanguage = "japanese";
22+
+declare const tokenizerLanguage = "mandarin";
23+
type TLanguage = typeof tokenizerLanguage;
24+
-type JapaneseTokenizerConfig = DefaultTokenizerConfig & {
25+
+type MandarinTokenizerConfig = DefaultTokenizerConfig & {
26+
language: TLanguage;
27+
};
28+
-export declare function createTokenizer(config?: JapaneseTokenizerConfig): DefaultTokenizer;
29+
+export declare function createTokenizer(config?: MandarinTokenizerConfig): DefaultTokenizer;
30+
export {};
31+
//# sourceMappingURL=mandarin.d.ts.map
32+
diff --git a/dist/commonjs/mandarin.js b/dist/commonjs/mandarin.js
33+
index 8015b8f30f2a4198761e1f2ee469f0c6fadb3f4a..5b745cb384c1269fb547302420c7ad6016796599 100644
34+
--- a/dist/commonjs/mandarin.js
35+
+++ b/dist/commonjs/mandarin.js
36+
@@ -2,7 +2,7 @@
37+
Object.defineProperty(exports, "__esModule", { value: true });
38+
exports.createTokenizer = createTokenizer;
39+
const internals_1 = require("@orama/orama/internals");
40+
-const tokenizerLanguage = "japanese";
41+
+const tokenizerLanguage = "mandarin";
42+
const defaultConfig = {
43+
language: tokenizerLanguage,
44+
};
45+
diff --git a/dist/esm/index.js b/dist/esm/index.js
46+
index 8ecf1802ba11bcc60573095b3794eaa690a3dcf7..944d4f4b40680adab70cc8f22a40351ba9115064 100644
47+
--- a/dist/esm/index.js
48+
+++ b/dist/esm/index.js
49+
@@ -1,5 +1,7 @@
50+
import { createTokenizer as createJapaneseTokenizer } from "./japanese.js";
51+
+import { createTokenizer as createMandarinTokenizer } from "./mandarin.js";
52+
export default {
53+
japanese: createJapaneseTokenizer,
54+
+ mandarin: createMandarinTokenizer,
55+
};
56+
//# sourceMappingURL=index.js.map
57+
\ No newline at end of file
58+
diff --git a/dist/esm/mandarin.d.ts b/dist/esm/mandarin.d.ts
59+
index a5c44e17802c6b4d17203e59cf8563120751a193..b3910a611bdff11a800b56eac173e3d06c1110ff 100644
60+
--- a/dist/esm/mandarin.d.ts
61+
+++ b/dist/esm/mandarin.d.ts
62+
@@ -1,9 +1,9 @@
63+
import type { DefaultTokenizer, DefaultTokenizerConfig } from "@orama/orama";
64+
-declare const tokenizerLanguage = "japanese";
65+
+declare const tokenizerLanguage = "mandarin";
66+
type TLanguage = typeof tokenizerLanguage;
67+
-type JapaneseTokenizerConfig = DefaultTokenizerConfig & {
68+
+type MandarinTokenizerConfig = DefaultTokenizerConfig & {
69+
language: TLanguage;
70+
};
71+
-export declare function createTokenizer(config?: JapaneseTokenizerConfig): DefaultTokenizer;
72+
+export declare function createTokenizer(config?: MandarinTokenizerConfig): DefaultTokenizer;
73+
export {};
74+
//# sourceMappingURL=mandarin.d.ts.map
75+
diff --git a/dist/esm/mandarin.js b/dist/esm/mandarin.js
76+
index 487a9ad00354b0c02cf8cd5c74fc28ac1d10b9d8..0c0d103e99bf8d81e58e71a7a6e98f5b336eac56 100644
77+
--- a/dist/esm/mandarin.js
78+
+++ b/dist/esm/mandarin.js
79+
@@ -1,5 +1,5 @@
80+
import { normalizeToken } from "@orama/orama/internals";
81+
-const tokenizerLanguage = "japanese";
82+
+const tokenizerLanguage = "mandarin";
83+
const defaultConfig = {
84+
language: tokenizerLanguage,
85+
};

.yarn/releases/yarn-4.7.0.cjs

+935
Large diffs are not rendered by default.

.yarnrc.yml

+3
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,3 @@
1+
nodeLinker: node-modules
2+
3+
yarnPath: .yarn/releases/yarn-4.7.0.cjs

app/api/search/route.ts

+15-12
Original file line numberDiff line numberDiff line change
@@ -6,24 +6,27 @@ import { stopwords as mandarinStopwords } from "@orama/stopwords/mandarin";
66
export const revalidate = false;
77

88
const tokenizer = createTokenizer({
9+
language: 'mandarin',
910
stopWords: mandarinStopwords,
1011
});
12+
1113
const search = {
1214
tokenizer,
13-
language: 'mandarin',
1415
components: {
1516
tokenizer,
1617
},
1718
search: {
18-
threshold: 0,
19-
tolerance: 0,
19+
threshold: 1.5,
20+
tolerance: 2,
21+
boost: {
22+
title: 2,
23+
content: 1,
24+
},
2025
},
21-
}
22-
export const { staticGET: GET } = createFromSource(source, undefined, {
23-
localeMap: {
24-
cn: search,
25-
en: search,
26-
mandarin: search,
27-
english: search,
28-
}
29-
});
26+
insertOptions: {
27+
batchSize: 100,
28+
async: true,
29+
},
30+
};
31+
32+
export const { staticGET: GET } = createFromSource(source, undefined, search);

package.json

+3-2
Original file line numberDiff line numberDiff line change
@@ -12,7 +12,7 @@
1212
},
1313
"dependencies": {
1414
"@orama/stopwords": "^3.1.1",
15-
"@orama/tokenizers": "^3.1.1",
15+
"@orama/tokenizers": "patch:@orama/tokenizers@npm%3A3.1.1#~/.yarn/patches/@orama-tokenizers-npm-3.1.1-d353c0f792.patch",
1616
"fumadocs-core": "15.0.15",
1717
"fumadocs-mdx": "11.5.6",
1818
"fumadocs-twoslash": "^3.1.0",
@@ -36,5 +36,6 @@
3636
"postcss": "^8.5.3",
3737
"tailwindcss": "^4.0.9",
3838
"typescript": "^5.8.2"
39-
}
39+
},
40+
"packageManager": "yarn@4.7.0"
4041
}

0 commit comments

Comments
 (0)