@kecan0406/aninames 0.3.0 → 0.3.1

package/README.md

@@ -56,7 +56,7 @@ Built sizes: `core.sqlite` 610 MB, `kowiki.sqlite` 25 MB, `unlicensed.sqlite` 83
 | `<layer>/<table>/<hash>.json.gz` | about 32 KB of JSON rows sorted by key (16 KB for `entity`), named by the first 48 bits of the SHA-256 of the file |
 | `NOTICE.md`, `index.html` | each layer's sources and terms |
 
-The key tables are `flat` (folded names), `latin` (romaji keys), `sound` (sound keys across kana, Hangul and romaji), `part` (name tokens) and `initial` (Korean initials); their rows are `[key, entity number, name weight, flags, popularity, standing, bonus]`. `tag` holds the same rows for Danbooru tags, keyed by the tag folded like a name without its qualifier and weighted by the tag's posts. `home` holds them for the copyright tag each character comes from, weighted by the character's own posts; a `core` row whose character the `unlicensed` layer gives another home carries flag 64, which a client that reads `unlicensed` skips. `entity` rows, keyed by entity number (which follows the entity ids' order), carry the id, names, Danbooru tags, works, family, the ids other databases give one entity in that layer (TMDB's `tmdb:tv/…` and `tmdb:movie/…`, IGDB's `igdb:…` and Steam's `steam:…` among them, read from Wikidata) and a character's home tag. Layers stay in separate files; the only values a file carries from other layers are derived numbers and flags (popularity, family standing, adult, which layers know the entity).
+The key tables are `flat` (folded names), `latin` (romaji keys), `sound` (sound keys across kana, Hangul and romaji), `part` (name tokens) and `initial` (Korean initials); their rows are `[key, entity number, name weight, flags, popularity, standing, bonus]`. `tag` holds the same rows for Danbooru tags, keyed by the tag folded like a name without its qualifier and weighted by the tag's posts. `home` holds them for the copyright tag each character comes from, weighted by the character's own posts; a `core` row whose character the `unlicensed` layer gives another home carries flag 64, which a client that reads `unlicensed` skips. `entity` rows, keyed by entity number (which follows the entity ids' order), carry the id, names, Danbooru tags, works, family, the ids other databases give one entity in that layer (TMDB's `tmdb:tv/…` and `tmdb:movie/…`, IGDB's `igdb:…` and Steam's `steam:…` among them, read from Wikidata), a character's home tag and a work's countries of origin (Wikidata P495, as ISO 3166-1 alpha-2 codes such as `JP`). Layers stay in separate files; the only values a file carries from other layers are derived numbers and flags (popularity, family standing, adult, which layers know the entity).
 
 On the 10-04 build: 19,167 files, 149 MB gzip'd, a manifest of 236 KB gzip'd. `mise run compare` searches the probe sample (4,370 queries) both ways: the files find the entity first 87% of the time, SQLite 86% (FTS substring search is not in the files; typo search reads the rows around the query's position instead). A cold query needs a median of 2 rounds (p90 3), 31 files (p90 78) and 287 KB (p90 584 KB).
 
@@ -71,7 +71,7 @@ const [first] = await names.search('봇치', { limit: 10, signal })
 const works = await names.lookup(first?.works ?? [])
 ```
 
-`search` returns `Found` cards (`id`, `kind`, `label`, `names` with their `lang` and `kind`, `danbooru` tags with the most posts first, `works`, `refs`, `home`, the copyright tag a character comes from, `score`, `adult`); `search(query, { kind, limit, adult: false })` narrows it, `lookup(ids)` returns the cards of given ids in one round, and `tagged(tag, { kind, adult })` the cards of the entities that carry a Danbooru tag, most posts first, whatever its case, punctuation or qualifier (`soul-eater`, `soul_eater` and `soul_eater_(series)` find the same works); `prefix: true` also finds the tags that start with it. `cast(tag, { adult, prefix, limit })` returns the characters whose home is that tag, most posts first. `cachedSource` keeps the files it fetched forever (their names are hashes of their contents), revalidates the manifest with its ETag or Last-Modified once a day (`maxAge`), answers from the cache when offline, refetches the manifest when the site has moved on to a new build, and deletes files a new manifest no longer names. `fileSource(dir)` reads a local `dist/site`.
+`search` returns `Found` cards (`id`, `kind`, `label`, `names` with their `lang` and `kind`, `danbooru` tags with the most posts first, `works`, `refs`, `home`, the copyright tag a character comes from, `origin`, the countries a work comes from, `score`, `adult`); `search(query, { kind, limit, adult: false })` narrows it, `lookup(ids)` returns the cards of given ids in one round, and `tagged(tag, { kind, adult })` the cards of the entities that carry a Danbooru tag, most posts first, whatever its case, punctuation or qualifier (`soul-eater`, `soul_eater` and `soul_eater_(series)` find the same works); `prefix: true` also finds the tags that start with it. `cast(tag, { adult, prefix, limit })` returns the characters whose home is that tag, most posts first. `cachedSource` keeps the files it fetched forever (their names are hashes of their contents), revalidates the manifest with its ETag or Last-Modified once a day (`maxAge`), answers from the cache when offline, refetches the manifest when the site has moved on to a new build, and deletes files a new manifest no longer names. `fileSource(dir)` reads a local `dist/site`.
 
 ## Development
 

package/lib/client/format.d.ts

@@ -76,5 +76,6 @@ export interface EntityRow {
     adult: boolean;
     refs: string[];
     home: string | null;
+    origin: string[];
 }
 export declare function readRows(table: Table, value: unknown): KeyRow[] | EntityRow[] | undefined;

package/lib/client/format.js

@@ -192,10 +192,11 @@ const tagOf = (value) => Array.isArray(value) && typeof value[0] === 'string' &&
 const workOf = (value) => Array.isArray(value) && isCount(value[0]) && typeof value[1] === 'string' ? [value[0], value[1]] : undefined;
 const eidOf = (value) => (isCount(value) ? value : undefined);
 const refOf = (value) => typeof value === 'string' && value.indexOf(':') > 0 ? value : undefined;
+const countryOf = (value) => typeof value === 'string' && /^[A-Z]{2}$/.test(value) ? value : undefined;
 function entityRow(value) {
     if (!Array.isArray(value) || value.length < 10)
         return undefined;
-    const [eid, id, rawNames, rawTags, rawWorks, rawReach, rank, standing, extra, adult, rawRefs = [], home = null] = value;
+    const [eid, id, rawNames, rawTags, rawWorks, rawReach, rank, standing, extra, adult, rawRefs = [], home = null, rawOrigin = [],] = value;
     if (!isCount(eid) || typeof id !== 'string')
         return undefined;
     const kind = id.startsWith('w/') ? 'work' : id.startsWith('c/') ? 'character' : undefined;
@@ -206,9 +207,12 @@ function entityRow(value) {
     const works = list(rawWorks, workOf);
     const reach = list(rawReach, eidOf);
     const refs = list(rawRefs, refOf);
-    if (!names || !danbooru || !works || !reach || !refs || (home !== null && typeof home !== 'string'))
+    const origin = list(rawOrigin, countryOf);
+    if (!names || !danbooru || !works || !reach || !refs || !origin)
         return undefined;
-    return { eid, id, kind, names, danbooru, works, reach, rank, standing, extra, adult: adult === 1, refs, home };
+    if (home !== null && typeof home !== 'string')
+        return undefined;
+    return { eid, id, kind, names, danbooru, works, reach, rank, standing, extra, adult: adult === 1, refs, home, origin };
 }
 export function readRows(table, value) {
     if (!Array.isArray(value))

package/lib/client/names.d.ts

@@ -14,6 +14,7 @@ export interface Found {
     works: string[];
     refs: string[];
     home: string | null;
+    origin: string[];
     score: number;
     adult: boolean;
 }

package/lib/client/names.js

@@ -79,7 +79,20 @@ function card(rows, score, adult) {
     const works = head.kind === 'character' ? [...new Set(rows.flatMap((row) => row.works.map(([, id]) => id)))] : [];
     const refs = [...new Set(rows.flatMap((row) => row.refs))].sort();
     const home = rows.findLast((row) => row.home !== null)?.home ?? null;
-    return { id: head.id, kind: head.kind, label: label(head.id, rows), names, danbooru, works, refs, home, score, adult };
+    const origin = [...new Set(rows.flatMap((row) => row.origin))];
+    return {
+        id: head.id,
+        kind: head.kind,
+        label: label(head.id, rows),
+        names,
+        danbooru,
+        works,
+        refs,
+        home,
+        origin,
+        score,
+        adult,
+    };
 }
 export class Names {
     #source;

package/lib/model.d.ts

@@ -26,6 +26,7 @@ export interface Entry {
     adult?: boolean;
     type?: string;
     year?: number;
+    origin?: string[];
 }
 export interface Source {
     name: string;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kecan0406/aninames",
-  "version": "0.3.0",
+  "version": "0.3.1",
   "type": "module",
   "description": "Multilingual alias index for anime titles and characters",
   "license": "MIT",