@cogenta/schema 0.1.0

package/dist/search/extract.js

@@ -0,0 +1,188 @@
+import { condense } from './text.js';
+/**
+ * Turning an entry into the text that gets indexed.
+ *
+ * Three sources, and only three (L1 task 16): `text` fields, the spans of a
+ * `richText` document, and the data of the blocks. Everything else is either
+ * not text — a number, a date, a colour, a media id, a relation id — or is
+ * machine data that would pollute the ranking. A `json` field in particular is
+ * configuration, not prose: indexing it makes every entry match `true`.
+ */
+/**
+ * Keys whose values are structure, not content.
+ *
+ * Without this list a block would contribute `block`, `span`, `normal`,
+ * `strong` and a URL to its own relevance, and an entry would rank on how many
+ * bold runs it contains.
+ */
+const STRUCTURAL_KEYS = new Set([
+    '_key',
+    '_type',
+    'key',
+    'type',
+    'id',
+    'collection',
+    'style',
+    'listItem',
+    'level',
+    'marks',
+    'markDefs',
+    'href',
+    'rel',
+    'url',
+    'src',
+    'mimeType',
+    'align',
+    'variant',
+    'theme',
+    'icon',
+]);
+function isRecord(value) {
+    return typeof value === 'object' && value !== null && !Array.isArray(value);
+}
+/**
+ * Every string in a value, minus the structural keys.
+ *
+ * Recursive because contract B blocks nest — a `columns` block holds blocks,
+ * and a `faq` block holds a list of question/answer pairs. Bounded by depth so
+ * that a hand-written or imported block cannot turn indexing into a hang.
+ */
+function collectStrings(value, into, depth = 0) {
+    if (depth > 8)
+        return;
+    if (typeof value === 'string') {
+        if (value.length > 0)
+            into.push(value);
+        return;
+    }
+    if (Array.isArray(value)) {
+        for (const item of value)
+            collectStrings(item, into, depth + 1);
+        return;
+    }
+    if (!isRecord(value))
+        return;
+    for (const [key, nested] of Object.entries(value)) {
+        if (STRUCTURAL_KEYS.has(key))
+            continue;
+        collectStrings(nested, into, depth + 1);
+    }
+}
+/**
+ * The prose of a rich text document: the `text` of every span, plus the
+ * captions of the media nodes.
+ *
+ * Deliberately not `JSON.stringify`: the raw document is roughly half syntax,
+ * and indexing it would let a search for `span` return every article on the
+ * site while a search for a genuinely rare word would be diluted by it.
+ */
+export function extractRichText(value) {
+    if (!Array.isArray(value))
+        return '';
+    const parts = [];
+    for (const node of value) {
+        if (!isRecord(node))
+            continue;
+        if (node['_type'] === 'block') {
+            const children = node['children'];
+            if (!Array.isArray(children))
+                continue;
+            for (const child of children) {
+                if (isRecord(child) && typeof child['text'] === 'string')
+                    parts.push(child['text']);
+            }
+            continue;
+        }
+        if (node['_type'] === 'media') {
+            const caption = node['caption'];
+            if (typeof caption === 'string')
+                parts.push(caption);
+            continue;
+        }
+        // An unknown node type is still content: a future contract B addition must
+        // become searchable on the day it ships, not on the day this file is
+        // remembered. The structural keys are filtered out all the same.
+        collectStrings(node, parts);
+    }
+    return condense(parts.join(' '));
+}
+/** The prose of one block. Its `data` is contract B, so its shape is open. */
+export function extractBlockText(block) {
+    const parts = [];
+    collectStrings(block.data, parts);
+    return condense(parts.join(' '));
+}
+function extractZones(zones) {
+    const parts = [];
+    for (const blocks of Object.values(zones)) {
+        for (const block of blocks) {
+            const text = extractBlockText(block);
+            if (text.length > 0)
+                parts.push(text);
+        }
+    }
+    return condense(parts.join(' '));
+}
+/**
+ * The field whose value labels a result.
+ *
+ * `title` by name when the collection has one — contract A's own example uses
+ * it — and the first declared `text` field otherwise, which is what an editor
+ * sees at the top of the form.
+ */
+function titleOf(collection, entry) {
+    const named = collection.fields['title'];
+    if (named?.kind === 'text') {
+        const value = entry.values['title'];
+        if (typeof value === 'string' && value.length > 0)
+            return condense(value);
+    }
+    for (const [name, field] of Object.entries(collection.fields)) {
+        if (field.kind !== 'text')
+            continue;
+        const value = entry.values[name];
+        if (typeof value === 'string' && value.length > 0)
+            return condense(value);
+    }
+    return '';
+}
+/**
+ * The indexable document for an entry.
+ *
+ * The title is repeated inside the body on purpose: none of the three engines
+ * offers per-column weighting in a form the other two can match — Postgres has
+ * `setweight`, MySQL has nothing, FTS5 has a per-column BM25 weight vector — so
+ * the one portable way to make a title term count for more is to let it appear
+ * twice in the text every engine ranks.
+ */
+export function searchDocumentFor(collection, entry) {
+    const title = titleOf(collection, entry);
+    const parts = [];
+    if (title.length > 0)
+        parts.push(title);
+    for (const [name, field] of Object.entries(collection.fields)) {
+        const value = entry.values[name];
+        if (field.kind === 'text') {
+            if (typeof value === 'string' && value.length > 0)
+                parts.push(condense(value));
+            continue;
+        }
+        if (field.kind === 'richText') {
+            const text = extractRichText(value);
+            if (text.length > 0)
+                parts.push(text);
+        }
+    }
+    const blocks = extractZones(entry.blocks);
+    if (blocks.length > 0)
+        parts.push(blocks);
+    return {
+        id: entry.id,
+        collection: collection.name,
+        locale: entry.locale,
+        status: entry.status,
+        title,
+        body: condense(parts.join(' ')),
+    };
+}
+//# sourceMappingURL=extract.js.map

package/dist/search/extract.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extract.js","sourceRoot":"","sources":["../../src/search/extract.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAA;AAGpC;;;;;;;;GAQG;AAEH;;;;;;GAMG;AACH,MAAM,eAAe,GAAG,IAAI,GAAG,CAAC;IAC9B,MAAM;IACN,OAAO;IACP,KAAK;IACL,MAAM;IACN,IAAI;IACJ,YAAY;IACZ,OAAO;IACP,UAAU;IACV,OAAO;IACP,OAAO;IACP,UAAU;IACV,MAAM;IACN,KAAK;IACL,KAAK;IACL,KAAK;IACL,UAAU;IACV,OAAO;IACP,SAAS;IACT,OAAO;IACP,MAAM;CACP,CAAC,CAAA;AAEF,SAAS,QAAQ,CAAC,KAAc;IAC9B,OAAO,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,KAAK,IAAI,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,CAAA;AAC7E,CAAC;AAED;;;;;;GAMG;AACH,SAAS,cAAc,CAAC,KAAc,EAAE,IAAc,EAAE,KAAK,GAAG,CAAC;IAC/D,IAAI,KAAK,GAAG,CAAC;QAAE,OAAM;IAErB,IAAI,OAAO,KAAK,KAAK,QAAQ,EAAE,CAAC;QAC9B,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC;YAAE,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;QACtC,OAAM;IACR,CAAC;IAED,IAAI,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QACzB,KAAK,MAAM,IAAI,IAAI,KAAK;YAAE,cAAc,CAAC,IAAI,EAAE,IAAI,EAAE,KAAK,GAAG,CAAC,CAAC,CAAA;QAC/D,OAAM;IACR,CAAC;IAED,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC;QAAE,OAAM;IAE5B,KAAK,MAAM,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,KAAK,CAAC,EAAE,CAAC;QAClD,IAAI,eAAe,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,SAAQ;QACtC,cAAc,CAAC,MAAM,EAAE,IAAI,EAAE,KAAK,GAAG,CAAC,CAAC,CAAA;IACzC,CAAC;AACH,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,eAAe,CAAC,KAAc;IAC5C,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,KAAK,CAAC;QAAE,OAAO,EAAE,CAAA;IACpC,MAAM,KAAK,GAAa,EAAE,CAAA;IAE1B,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QACzB,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC;YAAE,SAAQ;QAE7B,IAAI,IAAI,CAAC,OAAO,CAAC,KAAK,OAAO,EAAE,CAAC;YAC9B,MAAM,QAAQ,GAAG,IAAI,CAAC,UAAU,CAAC,CAAA;YACjC,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,QAAQ,CAAC;gBAAE,SAAQ;YACtC,KAAK,MAAM,KAAK,IAAI,QAAQ,EAAE,CAAC;gBAC7B,IAAI,QAAQ,CAAC,KAAK,CAAC,IAAI,OAAO,KAAK,CAAC,MAAM,CAAC,KAAK,QAAQ;oBAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC,CAAA;YACrF,CAAC;YACD,SAAQ;QACV,CAAC;QAED,IAAI,IAAI,CAAC,OAAO,CAAC,KAAK,OAAO,EAAE,CAAC;YAC9B,MAAM,OAAO,GAAG,IAAI,CAAC,SAAS,CAAC,CAAA;YAC/B,IAAI,OAAO,OAAO,KAAK,QAAQ;gBAAE,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAA;YACpD,SAAQ;QACV,CAAC;QAED,2EAA2E;QAC3E,qEAAqE;QACrE,iEAAiE;QACjE,cAAc,CAAC,IAAI,EAAE,KAAK,CAAC,CAAA;IAC7B,CAAC;IAED,OAAO,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAA;AAClC,CAAC;AAED,8EAA8E;AAC9E,MAAM,UAAU,gBAAgB,CAAC,KAAmB;IAClD,MAAM,KAAK,GAAa,EAAE,CAAA;IAC1B,cAAc,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,CAAA;IACjC,OAAO,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAA;AAClC,CAAC;AAED,SAAS,YAAY,CAAC,KAAiB;IACrC,MAAM,KAAK,GAAa,EAAE,CAAA;IAC1B,KAAK,MAAM,MAAM,IAAI,MAAM,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,CAAC;QAC1C,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,MAAM,IAAI,GAAG,gBAAgB,CAAC,KAAK,CAAC,CAAA;YACpC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC;gBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACvC,CAAC;IACH,CAAC;IACD,OAAO,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAA;AAClC,CAAC;AAED;;;;;;GAMG;AACH,SAAS,OAAO,CAAC,UAAgC,EAAE,KAAmB;IACpE,MAAM,KAAK,GAAG,UAAU,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;IACxC,IAAI,KAAK,EAAE,IAAI,KAAK,MAAM,EAAE,CAAC;QAC3B,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,OAAO,CAAC,CAAA;QACnC,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,QAAQ,CAAC,KAAK,CAAC,CAAA;IAC3E,CAAC;IAED,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC,EAAE,CAAC;QAC9D,IAAI,KAAK,CAAC,IAAI,KAAK,MAAM;YAAE,SAAQ;QACnC,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,CAAA;QAChC,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC;YAAE,OAAO,QAAQ,CAAC,KAAK,CAAC,CAAA;IAC3E,CAAC;IAED,OAAO,EAAE,CAAA;AACX,CAAC;AAED;;;;;;;;GAQG;AACH,MAAM,UAAU,iBAAiB,CAC/B,UAAgC,EAChC,KAAmB;IAEnB,MAAM,KAAK,GAAG,OAAO,CAAC,UAAU,EAAE,KAAK,CAAC,CAAA;IACxC,MAAM,KAAK,GAAa,EAAE,CAAA;IAC1B,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC;QAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;IAEvC,KAAK,MAAM,CAAC,IAAI,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC,EAAE,CAAC;QAC9D,MAAM,KAAK,GAAG,KAAK,CAAC,MAAM,CAAC,IAAI,CAAC,CAAA;QAEhC,IAAI,KAAK,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;YAC1B,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC;gBAAE,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAA;YAC9E,SAAQ;QACV,CAAC;QAED,IAAI,KAAK,CAAC,IAAI,KAAK,UAAU,EAAE,CAAC;YAC9B,MAAM,IAAI,GAAG,eAAe,CAAC,KAAK,CAAC,CAAA;YACnC,IAAI,IAAI,CAAC,MAAM,GAAG,CAAC;gBAAE,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;QACvC,CAAC;IACH,CAAC;IAED,MAAM,MAAM,GAAG,YAAY,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;IACzC,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC;QAAE,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;IAEzC,OAAO;QACL,EAAE,EAAE,KAAK,CAAC,EAAE;QACZ,UAAU,EAAE,UAAU,CAAC,IAAI;QAC3B,MAAM,EAAE,KAAK,CAAC,MAAM;QACpB,MAAM,EAAE,KAAK,CAAC,MAAM;QACpB,KAAK;QACL,IAAI,EAAE,QAAQ,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;KAChC,CAAA;AACH,CAAC"}

package/dist/search/index.d.ts

@@ -0,0 +1,36 @@
+import { type DatabaseHandle } from '@cogenta/core';
+import type { SearchDriver } from './types.js';
+/**
+ * Full-text search, one interface over three engines (L1, task 16).
+ *
+ * Nothing outside this folder knows whether the site is ranked by `ts_rank`,
+ * by an InnoDB fulltext score or by BM25 — or, on a SQLite build without FTS5,
+ * not ranked at all. What every caller does know is that the locale and the
+ * state it asks for are the only ones it can be given.
+ */
+export { extractBlockText, extractRichText, searchDocumentFor } from './extract.js';
+export type { MysqlSearchOptions } from './mysql.js';
+export { createMysqlSearch } from './mysql.js';
+export type { PostgresSearchOptions } from './postgres.js';
+export { configurationFor, createPostgresSearch } from './postgres.js';
+export type { SqliteSearchOptions } from './sqlite.js';
+export { createSqliteSearch } from './sqlite.js';
+export { SEARCH_COLUMNS, SEARCH_FTS_TABLE, SEARCH_TABLE } from './table.js';
+export { condense, foldText, MIN_TOKEN_LENGTH, queryTokens, tokenize } from './text.js';
+export type { SearchDocument, SearchDriver, SearchHit, SearchQuery, SearchReference, SearchResults, } from './types.js';
+export interface SearchIndexOptions {
+    readonly db: DatabaseHandle;
+    /** SQLite only: see `SqliteSearchOptions.fts5`. Ignored elsewhere. */
+    readonly fts5?: boolean;
+}
+/**
+ * The search index for whichever database is connected.
+ *
+ * Creating it also creates the physical index — a table plus whatever the
+ * engine needs around it — so a fresh install can index its first entry without
+ * a migration having run. That is deliberate: search is derived data, it can be
+ * rebuilt from the content at any time, and coupling it to the migration
+ * history would make a rebuild a schema change.
+ */
+export declare function createSearchIndex(options: SearchIndexOptions): Promise<SearchDriver>;
+//# sourceMappingURL=index.d.ts.map

package/dist/search/index.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../../src/search/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAgB,KAAK,cAAc,EAAE,MAAM,eAAe,CAAA;AAIjE,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,YAAY,CAAA;AAE9C;;;;;;;GAOG;AAEH,OAAO,EAAE,gBAAgB,EAAE,eAAe,EAAE,iBAAiB,EAAE,MAAM,cAAc,CAAA;AACnF,YAAY,EAAE,kBAAkB,EAAE,MAAM,YAAY,CAAA;AACpD,OAAO,EAAE,iBAAiB,EAAE,MAAM,YAAY,CAAA;AAC9C,YAAY,EAAE,qBAAqB,EAAE,MAAM,eAAe,CAAA;AAC1D,OAAO,EAAE,gBAAgB,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAA;AACtE,YAAY,EAAE,mBAAmB,EAAE,MAAM,aAAa,CAAA;AACtD,OAAO,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAA;AAChD,OAAO,EAAE,cAAc,EAAE,gBAAgB,EAAE,YAAY,EAAE,MAAM,YAAY,CAAA;AAC3E,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,gBAAgB,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAA;AACvF,YAAY,EACV,cAAc,EACd,YAAY,EACZ,SAAS,EACT,WAAW,EACX,eAAe,EACf,aAAa,GACd,MAAM,YAAY,CAAA;AAEnB,MAAM,WAAW,kBAAkB;IACjC,QAAQ,CAAC,EAAE,EAAE,cAAc,CAAA;IAC3B,sEAAsE;IACtE,QAAQ,CAAC,IAAI,CAAC,EAAE,OAAO,CAAA;CACxB;AAED;;;;;;;;GAQG;AACH,wBAAsB,iBAAiB,CAAC,OAAO,EAAE,kBAAkB,GAAG,OAAO,CAAC,YAAY,CAAC,CAe1F"}

package/dist/search/index.js

@@ -0,0 +1,44 @@
+import { CogentaError } from '@cogenta/core';
+import { createMysqlSearch } from './mysql.js';
+import { createPostgresSearch } from './postgres.js';
+import { createSqliteSearch } from './sqlite.js';
+/**
+ * Full-text search, one interface over three engines (L1, task 16).
+ *
+ * Nothing outside this folder knows whether the site is ranked by `ts_rank`,
+ * by an InnoDB fulltext score or by BM25 — or, on a SQLite build without FTS5,
+ * not ranked at all. What every caller does know is that the locale and the
+ * state it asks for are the only ones it can be given.
+ */
+export { extractBlockText, extractRichText, searchDocumentFor } from './extract.js';
+export { createMysqlSearch } from './mysql.js';
+export { configurationFor, createPostgresSearch } from './postgres.js';
+export { createSqliteSearch } from './sqlite.js';
+export { SEARCH_COLUMNS, SEARCH_FTS_TABLE, SEARCH_TABLE } from './table.js';
+export { condense, foldText, MIN_TOKEN_LENGTH, queryTokens, tokenize } from './text.js';
+/**
+ * The search index for whichever database is connected.
+ *
+ * Creating it also creates the physical index — a table plus whatever the
+ * engine needs around it — so a fresh install can index its first entry without
+ * a migration having run. That is deliberate: search is derived data, it can be
+ * rebuilt from the content at any time, and coupling it to the migration
+ * history would make a rebuild a schema change.
+ */
+export async function createSearchIndex(options) {
+    const { db } = options;
+    if (db.dialect === 'postgres')
+        return createPostgresSearch({ db });
+    if (db.dialect === 'mysql')
+        return createMysqlSearch({ db });
+    if (db.dialect === 'sqlite') {
+        return createSqliteSearch(options.fts5 === undefined ? { db } : { db, fts5: options.fts5 });
+    }
+    throw new CogentaError({
+        code: 'DB_DIALECT_UNSUPPORTED',
+        message: `No full-text search implementation for the "${String(db.dialect)}" dialect.`,
+        hint: 'Cogenta indexes content on Postgres, MySQL/MariaDB and SQLite.',
+        details: { dialect: db.dialect },
+    });
+}
+//# sourceMappingURL=index.js.map

package/dist/search/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../src/search/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,YAAY,EAAuB,MAAM,eAAe,CAAA;AACjE,OAAO,EAAE,iBAAiB,EAAE,MAAM,YAAY,CAAA;AAC9C,OAAO,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAA;AACpD,OAAO,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAA;AAGhD;;;;;;;GAOG;AAEH,OAAO,EAAE,gBAAgB,EAAE,eAAe,EAAE,iBAAiB,EAAE,MAAM,cAAc,CAAA;AAEnF,OAAO,EAAE,iBAAiB,EAAE,MAAM,YAAY,CAAA;AAE9C,OAAO,EAAE,gBAAgB,EAAE,oBAAoB,EAAE,MAAM,eAAe,CAAA;AAEtE,OAAO,EAAE,kBAAkB,EAAE,MAAM,aAAa,CAAA;AAChD,OAAO,EAAE,cAAc,EAAE,gBAAgB,EAAE,YAAY,EAAE,MAAM,YAAY,CAAA;AAC3E,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,gBAAgB,EAAE,WAAW,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAA;AAgBvF;;;;;;;;GAQG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,OAA2B;IACjE,MAAM,EAAE,EAAE,EAAE,GAAG,OAAO,CAAA;IAEtB,IAAI,EAAE,CAAC,OAAO,KAAK,UAAU;QAAE,OAAO,oBAAoB,CAAC,EAAE,EAAE,EAAE,CAAC,CAAA;IAClE,IAAI,EAAE,CAAC,OAAO,KAAK,OAAO;QAAE,OAAO,iBAAiB,CAAC,EAAE,EAAE,EAAE,CAAC,CAAA;IAC5D,IAAI,EAAE,CAAC,OAAO,KAAK,QAAQ,EAAE,CAAC;QAC5B,OAAO,kBAAkB,CAAC,OAAO,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,IAAI,EAAE,OAAO,CAAC,IAAI,EAAE,CAAC,CAAA;IAC7F,CAAC;IAED,MAAM,IAAI,YAAY,CAAC;QACrB,IAAI,EAAE,wBAAwB;QAC9B,OAAO,EAAE,+CAA+C,MAAM,CAAC,EAAE,CAAC,OAAO,CAAC,YAAY;QACtF,IAAI,EAAE,gEAAgE;QACtE,OAAO,EAAE,EAAE,OAAO,EAAE,EAAE,CAAC,OAAO,EAAE;KACjC,CAAC,CAAA;AACJ,CAAC"}

package/dist/search/mysql.d.ts

@@ -0,0 +1,7 @@
+import { type DatabaseHandle } from '@cogenta/core';
+import type { SearchDriver } from './types.js';
+export interface MysqlSearchOptions {
+    readonly db: DatabaseHandle;
+}
+export declare function createMysqlSearch(options: MysqlSearchOptions): Promise<SearchDriver>;
+//# sourceMappingURL=mysql.d.ts.map

package/dist/search/mysql.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"mysql.d.ts","sourceRoot":"","sources":["../../src/search/mysql.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,KAAK,cAAc,EAMpB,MAAM,eAAe,CAAA;AAqBtB,OAAO,KAAK,EAAkB,YAAY,EAAgC,MAAM,YAAY,CAAA;AAiC5F,MAAM,WAAW,kBAAkB;IACjC,QAAQ,CAAC,EAAE,EAAE,cAAc,CAAA;CAC5B;AAED,wBAAsB,iBAAiB,CAAC,OAAO,EAAE,kBAAkB,GAAG,OAAO,CAAC,YAAY,CAAC,CA2G1F"}

package/dist/search/mysql.js

@@ -0,0 +1,121 @@
+import { identifier, sql, limit as sqlLimit, } from '@cogenta/core';
+import { joinFragments } from '../store/fragments.js';
+import { assertDocument, assertReference, EMPTY_RESULTS, normaliseQuery, numeric, storedTitle, toResults, } from './query.js';
+import { allOf, baseColumns, primaryKey, SEARCH_COLUMNS, SEARCH_TABLE, scopeFilters, selectedColumns, } from './table.js';
+import { foldText, MIN_TOKEN_LENGTH } from './text.js';
+/**
+ * MySQL and MariaDB full-text search: a `FULLTEXT` index, read in boolean mode.
+ *
+ * Boolean mode rather than natural language mode, for two reasons that both
+ * bite in production. Natural language mode on MyISAM drops any word present in
+ * more than half the rows — on a site where every page says the brand name,
+ * searching for the brand returns nothing, with no error. And natural language
+ * mode has no prefix operator, so a half-typed word matches nothing. Boolean
+ * mode gives `+term*`: every term required, each one allowed to be a prefix,
+ * which is the same semantics this package gives on the other two dialects.
+ *
+ * What it does not give is a comparable score. `MATCH … AGAINST` in boolean
+ * mode returns a term-count-based number, not the IDF-weighted rank Postgres
+ * computes. The L1 spec accepts that explicitly, so the contract suite asserts
+ * which documents come back and never in which order they are ranked.
+ */
+const FULLTEXT_INDEX = 'cogenta_search_ft';
+const SCOPE_INDEX = 'cogenta_search_scope';
+/**
+ * `+term*` for every term.
+ *
+ * Tokens are letters and digits only (see `text.ts`), so a reader typing
+ * `-foo +bar*` gets two ordinary words rather than boolean operators, and the
+ * index cannot be steered by the query string.
+ */
+function booleanExpression(tokens) {
+    return tokens.map((token) => `+${token}*`).join(' ');
+}
+export async function createMysqlSearch(options) {
+    const { db } = options;
+    const dialect = 'mysql';
+    const table = identifier(SEARCH_TABLE, dialect);
+    const content = identifier(SEARCH_COLUMNS.content, dialect);
+    // The FULLTEXT key is declared inside `create table`: MySQL has no
+    // `create index if not exists`, so adding it afterwards means either a
+    // swallowed error or a catalogue lookup, and neither is worth a second
+    // statement when the table is created here anyway.
+    const columns = [
+        ...baseColumns(dialect),
+        primaryKey(dialect),
+        sql `fulltext key ${identifier(FULLTEXT_INDEX, dialect)} (${content})`,
+        sql `key ${identifier(SCOPE_INDEX, dialect)} (${identifier(SEARCH_COLUMNS.locale, dialect)},
+                                                 ${identifier(SEARCH_COLUMNS.status, dialect)})`,
+    ];
+    await db.query(sql `create table if not exists ${table} (${joinFragments(columns, ', ')})`);
+    const assignments = [
+        SEARCH_COLUMNS.locale,
+        SEARCH_COLUMNS.status,
+        SEARCH_COLUMNS.title,
+        SEARCH_COLUMNS.content,
+    ].map((column) => {
+        const name = identifier(column, dialect);
+        // `values(col)` rather than the newer row alias: MariaDB does not accept
+        // `as new` at all, and this package is tested against both.
+        return sql `${name} = values(${name})`;
+    });
+    return {
+        dialect,
+        index: async (entry) => {
+            assertDocument(entry);
+            await db.query(sql `insert into ${table} (
+              ${identifier(SEARCH_COLUMNS.entryId, dialect)},
+              ${identifier(SEARCH_COLUMNS.collection, dialect)},
+              ${identifier(SEARCH_COLUMNS.locale, dialect)},
+              ${identifier(SEARCH_COLUMNS.status, dialect)},
+              ${identifier(SEARCH_COLUMNS.title, dialect)},
+              ${content})
+            values (${entry.id}, ${entry.collection}, ${entry.locale}, ${entry.status},
+                    ${storedTitle(entry)}, ${foldText(`${entry.title} ${entry.body}`)})
+            on duplicate key update ${joinFragments(assignments, ', ')}`);
+        },
+        remove: async (reference) => {
+            assertReference(reference);
+            await db.query(sql `delete from ${table}
+            where ${identifier(SEARCH_COLUMNS.collection, dialect)} = ${reference.collection}
+              and ${identifier(SEARCH_COLUMNS.entryId, dialect)} = ${reference.id}`);
+        },
+        search: async (query) => {
+            const normalised = normaliseQuery(query);
+            if (normalised.tokens.length === 0)
+                return EMPTY_RESULTS;
+            const expression = booleanExpression(normalised.tokens);
+            const match = sql `match(${content}) against (${expression} in boolean mode)`;
+            const filters = scopeFilters(dialect, normalised);
+            const found = await db.query(sql `select ${selectedColumns(dialect)}, ${match} as ${identifier('score', dialect)}
+            from ${table}
+            where ${allOf([match, ...filters])}
+            order by ${identifier('score', dialect)} desc,
+                     ${identifier(SEARCH_COLUMNS.entryId, dialect)} asc
+            limit ${sqlLimit(normalised.size + 1)} offset ${sqlLimit(normalised.offset)}`);
+            return toResults(found.rows, (row) => numeric(row['score']), normalised);
+        },
+        clear: async (scope) => {
+            const where = scope?.collection === undefined
+                ? sql ``
+                : sql ` where ${identifier(SEARCH_COLUMNS.collection, dialect)} = ${scope.collection}`;
+            await db.query(sql `delete from ${table}${where}`);
+        },
+        health: async () => {
+            const started = Date.now();
+            const counted = await db.query(sql `select count(*) as ${identifier('total', dialect)} from ${table}`);
+            return {
+                status: 'ok',
+                driver: 'mysql',
+                tier: 'optimal',
+                latencyMs: Date.now() - started,
+                // Stated rather than hidden: the server, not Cogenta, decides the floor,
+                // and an administrator who wonders why a two-letter word finds nothing
+                // needs to be pointed at `innodb_ft_min_token_size`.
+                message: `FULLTEXT search in boolean mode. Words shorter than ${MIN_TOKEN_LENGTH} characters, and the server's stopwords, are not indexed.`,
+                details: { documents: numeric(counted.rows[0]?.total) },
+            };
+        },
+    };
+}
+//# sourceMappingURL=mysql.js.map

package/dist/search/mysql.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"mysql.js","sourceRoot":"","sources":["../../src/search/mysql.ts"],"names":[],"mappings":"AAAA,OAAO,EAGL,UAAU,EAEV,GAAG,EACH,KAAK,IAAI,QAAQ,GAClB,MAAM,eAAe,CAAA;AACtB,OAAO,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAA;AACrD,OAAO,EACL,cAAc,EACd,eAAe,EACf,aAAa,EACb,cAAc,EACd,OAAO,EACP,WAAW,EACX,SAAS,GACV,MAAM,YAAY,CAAA;AACnB,OAAO,EACL,KAAK,EACL,WAAW,EACX,UAAU,EACV,cAAc,EACd,YAAY,EACZ,YAAY,EACZ,eAAe,GAChB,MAAM,YAAY,CAAA;AACnB,OAAO,EAAE,QAAQ,EAAE,gBAAgB,EAAE,MAAM,WAAW,CAAA;AAGtD;;;;;;;;;;;;;;;GAeG;AAEH,MAAM,cAAc,GAAG,mBAAmB,CAAA;AAC1C,MAAM,WAAW,GAAG,sBAAsB,CAAA;AAE1C;;;;;;GAMG;AACH,SAAS,iBAAiB,CAAC,MAAyB;IAClD,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,IAAI,KAAK,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAA;AACtD,CAAC;AAMD,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,OAA2B;IACjE,MAAM,EAAE,EAAE,EAAE,GAAG,OAAO,CAAA;IACtB,MAAM,OAAO,GAAG,OAAgB,CAAA;IAChC,MAAM,KAAK,GAAG,UAAU,CAAC,YAAY,EAAE,OAAO,CAAC,CAAA;IAC/C,MAAM,OAAO,GAAG,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,CAAA;IAE3D,mEAAmE;IACnE,uEAAuE;IACvE,uEAAuE;IACvE,mDAAmD;IACnD,MAAM,OAAO,GAAkB;QAC7B,GAAG,WAAW,CAAC,OAAO,CAAC;QACvB,UAAU,CAAC,OAAO,CAAC;QACnB,GAAG,CAAA,gBAAgB,UAAU,CAAC,cAAc,EAAE,OAAO,CAAC,KAAK,OAAO,GAAG;QACrE,GAAG,CAAA,OAAO,UAAU,CAAC,WAAW,EAAE,OAAO,CAAC,KAAK,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC;mDAC1C,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG;KAC7F,CAAA;IAED,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,CAAA,8BAA8B,KAAK,KAAK,aAAa,CAAC,OAAO,EAAE,IAAI,CAAC,GAAG,CAAC,CAAA;IAE1F,MAAM,WAAW,GAAG;QAClB,cAAc,CAAC,MAAM;QACrB,cAAc,CAAC,MAAM;QACrB,cAAc,CAAC,KAAK;QACpB,cAAc,CAAC,OAAO;KACvB,CAAC,GAAG,CAAC,CAAC,MAAM,EAAE,EAAE;QACf,MAAM,IAAI,GAAG,UAAU,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;QACxC,yEAAyE;QACzE,4DAA4D;QAC5D,OAAO,GAAG,CAAA,GAAG,IAAI,aAAa,IAAI,GAAG,CAAA;IACvC,CAAC,CAAC,CAAA;IAEF,OAAO;QACL,OAAO;QAEP,KAAK,EAAE,KAAK,EAAE,KAAqB,EAAiB,EAAE;YACpD,cAAc,CAAC,KAAK,CAAC,CAAA;YAErB,MAAM,EAAE,CAAC,KAAK,CACZ,GAAG,CAAA,eAAe,KAAK;gBACf,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC;gBAC3C,UAAU,CAAC,cAAc,CAAC,UAAU,EAAE,OAAO,CAAC;gBAC9C,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC;gBAC1C,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC;gBAC1C,UAAU,CAAC,cAAc,CAAC,KAAK,EAAE,OAAO,CAAC;gBACzC,OAAO;sBACD,KAAK,CAAC,EAAE,KAAK,KAAK,CAAC,UAAU,KAAK,KAAK,CAAC,MAAM,KAAK,KAAK,CAAC,MAAM;sBAC/D,WAAW,CAAC,KAAK,CAAC,KAAK,QAAQ,CAAC,GAAG,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC;sCAC/C,aAAa,CAAC,WAAW,EAAE,IAAI,CAAC,EAAE,CACjE,CAAA;QACH,CAAC;QAED,MAAM,EAAE,KAAK,EAAE,SAA0B,EAAiB,EAAE;YAC1D,eAAe,CAAC,SAAS,CAAC,CAAA;YAC1B,MAAM,EAAE,CAAC,KAAK,CACZ,GAAG,CAAA,eAAe,KAAK;oBACX,UAAU,CAAC,cAAc,CAAC,UAAU,EAAE,OAAO,CAAC,MAAM,SAAS,CAAC,UAAU;oBACxE,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,MAAM,SAAS,CAAC,EAAE,EAAE,CAC5E,CAAA;QACH,CAAC;QAED,MAAM,EAAE,KAAK,EAAE,KAAkB,EAAE,EAAE;YACnC,MAAM,UAAU,GAAG,cAAc,CAAC,KAAK,CAAC,CAAA;YACxC,IAAI,UAAU,CAAC,MAAM,CAAC,MAAM,KAAK,CAAC;gBAAE,OAAO,aAAa,CAAA;YAExD,MAAM,UAAU,GAAG,iBAAiB,CAAC,UAAU,CAAC,MAAM,CAAC,CAAA;YACvD,MAAM,KAAK,GAAG,GAAG,CAAA,SAAS,OAAO,cAAc,UAAU,mBAAmB,CAAA;YAC5E,MAAM,OAAO,GAAG,YAAY,CAAC,OAAO,EAAE,UAAU,CAAC,CAAA;YAEjD,MAAM,KAAK,GAAG,MAAM,EAAE,CAAC,KAAK,CAC1B,GAAG,CAAA,UAAU,eAAe,CAAC,OAAO,CAAC,KAAK,KAAK,OAAO,UAAU,CAAC,OAAO,EAAE,OAAO,CAAC;mBACvE,KAAK;oBACJ,KAAK,CAAC,CAAC,KAAK,EAAE,GAAG,OAAO,CAAC,CAAC;uBACvB,UAAU,CAAC,OAAO,EAAE,OAAO,CAAC;uBAC5B,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC;oBAC9C,QAAQ,CAAC,UAAU,CAAC,IAAI,GAAG,CAAC,CAAC,WAAW,QAAQ,CAAC,UAAU,CAAC,MAAM,CAAC,EAAE,CAClF,CAAA;YAED,OAAO,SAAS,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,EAAE,UAAU,CAAC,CAAA;QAC1E,CAAC;QAED,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,EAAE;YACrB,MAAM,KAAK,GACT,KAAK,EAAE,UAAU,KAAK,SAAS;gBAC7B,CAAC,CAAC,GAAG,CAAA,EAAE;gBACP,CAAC,CAAC,GAAG,CAAA,UAAU,UAAU,CAAC,cAAc,CAAC,UAAU,EAAE,OAAO,CAAC,MAAM,KAAK,CAAC,UAAU,EAAE,CAAA;YACzF,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,CAAA,eAAe,KAAK,GAAG,KAAK,EAAE,CAAC,CAAA;QACnD,CAAC;QAED,MAAM,EAAE,KAAK,IAA2B,EAAE;YACxC,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;YAC1B,MAAM,OAAO,GAAG,MAAM,EAAE,CAAC,KAAK,CAC5B,GAAG,CAAA,sBAAsB,UAAU,CAAC,OAAO,EAAE,OAAO,CAAC,SAAS,KAAK,EAAE,CACtE,CAAA;YACD,OAAO;gBACL,MAAM,EAAE,IAAI;gBACZ,MAAM,EAAE,OAAO;gBACf,IAAI,EAAE,SAAS;gBACf,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,OAAO;gBAC/B,yEAAyE;gBACzE,uEAAuE;gBACvE,qDAAqD;gBACrD,OAAO,EAAE,uDAAuD,gBAAgB,2DAA2D;gBAC3I,OAAO,EAAE,EAAE,SAAS,EAAE,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,EAAE;aACxD,CAAA;QACH,CAAC;KACF,CAAA;AACH,CAAC"}

package/dist/search/postgres.d.ts

@@ -0,0 +1,9 @@
+import { type DatabaseHandle } from '@cogenta/core';
+import type { SearchDriver } from './types.js';
+/** `fr-CA`, `fr_CA` and `FR` all mean the French dictionary. */
+export declare function configurationFor(locale: string): string;
+export interface PostgresSearchOptions {
+    readonly db: DatabaseHandle;
+}
+export declare function createPostgresSearch(options: PostgresSearchOptions): Promise<SearchDriver>;
+//# sourceMappingURL=postgres.d.ts.map

package/dist/search/postgres.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"postgres.d.ts","sourceRoot":"","sources":["../../src/search/postgres.ts"],"names":[],"mappings":"AAAA,OAAO,EACL,KAAK,cAAc,EAMpB,MAAM,eAAe,CAAA;AAqBtB,OAAO,KAAK,EAAkB,YAAY,EAAgC,MAAM,YAAY,CAAA;AAwD5F,gEAAgE;AAChE,wBAAgB,gBAAgB,CAAC,MAAM,EAAE,MAAM,GAAG,MAAM,CAGvD;AAeD,MAAM,WAAW,qBAAqB;IACpC,QAAQ,CAAC,EAAE,EAAE,cAAc,CAAA;CAC5B;AAED,wBAAsB,oBAAoB,CAAC,OAAO,EAAE,qBAAqB,GAAG,OAAO,CAAC,YAAY,CAAC,CAqHhG"}

package/dist/search/postgres.js

@@ -0,0 +1,167 @@
+import { identifier, sql, limit as sqlLimit, } from '@cogenta/core';
+import { joinFragments } from '../store/fragments.js';
+import { assertDocument, assertReference, EMPTY_RESULTS, normaliseQuery, numeric, storedTitle, toResults, } from './query.js';
+import { allOf, baseColumns, primaryKey, SEARCH_COLUMNS, SEARCH_TABLE, scopeFilters, selectedColumns, } from './table.js';
+import { foldText } from './text.js';
+/**
+ * Postgres full-text search: a stored `tsvector` and a GIN index.
+ *
+ * The vector is a real column rather than an expression index, because the
+ * text search configuration is **per row**: an entry in French must be stemmed
+ * by the French dictionary and an entry in English by the English one, and an
+ * expression index can only name one configuration. Storing the vector costs
+ * space and buys the right stemmer per language, which is the whole reason to
+ * be on Postgres for this feature.
+ */
+const DOCUMENT_COLUMN = 'document';
+const GIN_INDEX = 'cogenta_search_document_gin';
+const SCOPE_INDEX = 'cogenta_search_scope';
+/**
+ * Locale to text search configuration.
+ *
+ * Only configurations that ship with a default Postgres build are listed: a
+ * missing one is not a warning but a `text search configuration "…" does not
+ * exist` at query time. Anything unlisted falls back to `simple`, which does no
+ * stemming and no stopword removal — worse recall, never an error, which is the
+ * right trade for a language the server was not built to handle.
+ */
+const CONFIGURATIONS = {
+    ar: 'arabic',
+    da: 'danish',
+    de: 'german',
+    el: 'greek',
+    en: 'english',
+    es: 'spanish',
+    eu: 'basque',
+    fi: 'finnish',
+    fr: 'french',
+    ga: 'irish',
+    hi: 'hindi',
+    hu: 'hungarian',
+    hy: 'armenian',
+    id: 'indonesian',
+    it: 'italian',
+    lt: 'lithuanian',
+    ne: 'nepali',
+    nl: 'dutch',
+    no: 'norwegian',
+    pt: 'portuguese',
+    ro: 'romanian',
+    ru: 'russian',
+    sr: 'serbian',
+    sv: 'swedish',
+    ta: 'tamil',
+    tr: 'turkish',
+    yi: 'yiddish',
+};
+/** `fr-CA`, `fr_CA` and `FR` all mean the French dictionary. */
+export function configurationFor(locale) {
+    const primary = locale.toLowerCase().split(/[-_]/u)[0] ?? '';
+    return CONFIGURATIONS[primary] ?? 'simple';
+}
+/**
+ * The `tsquery` text: every term required, every term a prefix.
+ *
+ * Requiring all terms (`&`) rather than any of them matches what a reader
+ * expects of a site search — more terms should narrow, not widen. The `:*`
+ * makes a half-typed last word still match, which is what a search-as-you-type
+ * box needs. Tokens are letters and digits only (see `text.ts`), so nothing in
+ * here can be read as `tsquery` syntax.
+ */
+function tsquery(tokens) {
+    return tokens.map((token) => `${token}:*`).join(' & ');
+}
+export async function createPostgresSearch(options) {
+    const { db } = options;
+    const dialect = 'postgres';
+    const table = identifier(SEARCH_TABLE, dialect);
+    const document = identifier(DOCUMENT_COLUMN, dialect);
+    const columns = [
+        ...baseColumns(dialect),
+        sql `${document} tsvector not null`,
+        primaryKey(dialect),
+    ];
+    await db.query(sql `create table if not exists ${table} (${joinFragments(columns, ', ')})`);
+    // GIN rather than GiST: slower to update, far faster to search, and a content
+    // index is read orders of magnitude more often than it is written.
+    await db.query(sql `create index if not exists ${identifier(GIN_INDEX, dialect)}
+        on ${table} using gin (${document})`);
+    await db.query(sql `create index if not exists ${identifier(SCOPE_INDEX, dialect)}
+        on ${table} (${identifier(SEARCH_COLUMNS.locale, dialect)},
+                     ${identifier(SEARCH_COLUMNS.status, dialect)})`);
+    return {
+        dialect,
+        index: async (entry) => {
+            assertDocument(entry);
+            const configuration = configurationFor(entry.locale);
+            const content = foldText(`${entry.title} ${entry.body}`);
+            // `on conflict` rather than delete-then-insert: re-indexing an entry after
+            // an edit is the common path, and two statements would leave the entry
+            // missing from the index for the instant between them.
+            await db.query(sql `insert into ${table} (
+              ${identifier(SEARCH_COLUMNS.entryId, dialect)},
+              ${identifier(SEARCH_COLUMNS.collection, dialect)},
+              ${identifier(SEARCH_COLUMNS.locale, dialect)},
+              ${identifier(SEARCH_COLUMNS.status, dialect)},
+              ${identifier(SEARCH_COLUMNS.title, dialect)},
+              ${identifier(SEARCH_COLUMNS.content, dialect)},
+              ${document})
+            values (${entry.id}, ${entry.collection}, ${entry.locale}, ${entry.status},
+                    ${storedTitle(entry)}, ${content},
+                    to_tsvector(${configuration}::regconfig, ${content}))
+            on conflict (${identifier(SEARCH_COLUMNS.collection, dialect)},
+                         ${identifier(SEARCH_COLUMNS.entryId, dialect)})
+            do update set
+              ${identifier(SEARCH_COLUMNS.locale, dialect)} = excluded.${identifier(SEARCH_COLUMNS.locale, dialect)},
+              ${identifier(SEARCH_COLUMNS.status, dialect)} = excluded.${identifier(SEARCH_COLUMNS.status, dialect)},
+              ${identifier(SEARCH_COLUMNS.title, dialect)} = excluded.${identifier(SEARCH_COLUMNS.title, dialect)},
+              ${identifier(SEARCH_COLUMNS.content, dialect)} = excluded.${identifier(SEARCH_COLUMNS.content, dialect)},
+              ${document} = excluded.${document}`);
+        },
+        remove: async (reference) => {
+            assertReference(reference);
+            await db.query(sql `delete from ${table}
+            where ${identifier(SEARCH_COLUMNS.collection, dialect)} = ${reference.collection}
+              and ${identifier(SEARCH_COLUMNS.entryId, dialect)} = ${reference.id}`);
+        },
+        search: async (query) => {
+            const normalised = normaliseQuery(query);
+            if (normalised.tokens.length === 0)
+                return EMPTY_RESULTS;
+            const configuration = configurationFor(normalised.locale);
+            const expression = tsquery(normalised.tokens);
+            const filters = scopeFilters(dialect, normalised);
+            // The query is built twice — once to match, once to rank. Postgres
+            // evaluates `to_tsquery` as an immutable function, so this is one parse
+            // and no extra work, and it keeps the statement readable.
+            const match = sql `${document} @@ to_tsquery(${configuration}::regconfig, ${expression})`;
+            const rank = sql `ts_rank(${document}, to_tsquery(${configuration}::regconfig, ${expression}))`;
+            const found = await db.query(sql `select ${selectedColumns(dialect)}, ${rank} as ${identifier('score', dialect)}
+            from ${table}
+            where ${allOf([match, ...filters])}
+            order by ${identifier('score', dialect)} desc,
+                     ${identifier(SEARCH_COLUMNS.entryId, dialect)} asc
+            limit ${sqlLimit(normalised.size + 1)} offset ${sqlLimit(normalised.offset)}`);
+            return toResults(found.rows, (row) => numeric(row['score']), normalised);
+        },
+        clear: async (scope) => {
+            const where = scope?.collection === undefined
+                ? sql ``
+                : sql ` where ${identifier(SEARCH_COLUMNS.collection, dialect)} = ${scope.collection}`;
+            await db.query(sql `delete from ${table}${where}`);
+        },
+        health: async () => {
+            const started = Date.now();
+            const counted = await db.query(sql `select count(*) as ${identifier('total', dialect)} from ${table}`);
+            return {
+                status: 'ok',
+                driver: 'postgres',
+                tier: 'optimal',
+                latencyMs: Date.now() - started,
+                message: 'Full-text search on tsvector with a GIN index, stemmed per locale.',
+                details: { documents: numeric(counted.rows[0]?.total) },
+            };
+        },
+    };
+}
+//# sourceMappingURL=postgres.js.map

package/dist/search/postgres.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"postgres.js","sourceRoot":"","sources":["../../src/search/postgres.ts"],"names":[],"mappings":"AAAA,OAAO,EAGL,UAAU,EAEV,GAAG,EACH,KAAK,IAAI,QAAQ,GAClB,MAAM,eAAe,CAAA;AACtB,OAAO,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAA;AACrD,OAAO,EACL,cAAc,EACd,eAAe,EACf,aAAa,EACb,cAAc,EACd,OAAO,EACP,WAAW,EACX,SAAS,GACV,MAAM,YAAY,CAAA;AACnB,OAAO,EACL,KAAK,EACL,WAAW,EACX,UAAU,EACV,cAAc,EACd,YAAY,EACZ,YAAY,EACZ,eAAe,GAChB,MAAM,YAAY,CAAA;AACnB,OAAO,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAA;AAGpC;;;;;;;;;GASG;AAEH,MAAM,eAAe,GAAG,UAAU,CAAA;AAClC,MAAM,SAAS,GAAG,6BAA6B,CAAA;AAC/C,MAAM,WAAW,GAAG,sBAAsB,CAAA;AAE1C;;;;;;;;GAQG;AACH,MAAM,cAAc,GAAqC;IACvD,EAAE,EAAE,QAAQ;IACZ,EAAE,EAAE,QAAQ;IACZ,EAAE,EAAE,QAAQ;IACZ,EAAE,EAAE,OAAO;IACX,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,QAAQ;IACZ,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,QAAQ;IACZ,EAAE,EAAE,OAAO;IACX,EAAE,EAAE,OAAO;IACX,EAAE,EAAE,WAAW;IACf,EAAE,EAAE,UAAU;IACd,EAAE,EAAE,YAAY;IAChB,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,YAAY;IAChB,EAAE,EAAE,QAAQ;IACZ,EAAE,EAAE,OAAO;IACX,EAAE,EAAE,WAAW;IACf,EAAE,EAAE,YAAY;IAChB,EAAE,EAAE,UAAU;IACd,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,OAAO;IACX,EAAE,EAAE,SAAS;IACb,EAAE,EAAE,SAAS;CACd,CAAA;AAED,gEAAgE;AAChE,MAAM,UAAU,gBAAgB,CAAC,MAAc;IAC7C,MAAM,OAAO,GAAG,MAAM,CAAC,WAAW,EAAE,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAA;IAC5D,OAAO,cAAc,CAAC,OAAO,CAAC,IAAI,QAAQ,CAAA;AAC5C,CAAC;AAED;;;;;;;;GAQG;AACH,SAAS,OAAO,CAAC,MAAyB;IACxC,OAAO,MAAM,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,EAAE,CAAC,GAAG,KAAK,IAAI,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,CAAA;AACxD,CAAC;AAMD,MAAM,CAAC,KAAK,UAAU,oBAAoB,CAAC,OAA8B;IACvE,MAAM,EAAE,EAAE,EAAE,GAAG,OAAO,CAAA;IACtB,MAAM,OAAO,GAAG,UAAmB,CAAA;IACnC,MAAM,KAAK,GAAG,UAAU,CAAC,YAAY,EAAE,OAAO,CAAC,CAAA;IAC/C,MAAM,QAAQ,GAAG,UAAU,CAAC,eAAe,EAAE,OAAO,CAAC,CAAA;IAErD,MAAM,OAAO,GAAkB;QAC7B,GAAG,WAAW,CAAC,OAAO,CAAC;QACvB,GAAG,CAAA,GAAG,QAAQ,oBAAoB;QAClC,UAAU,CAAC,OAAO,CAAC;KACpB,CAAA;IAED,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,CAAA,8BAA8B,KAAK,KAAK,aAAa,CAAC,OAAO,EAAE,IAAI,CAAC,GAAG,CAAC,CAAA;IAC1F,8EAA8E;IAC9E,mEAAmE;IACnE,MAAM,EAAE,CAAC,KAAK,CACZ,GAAG,CAAA,8BAA8B,UAAU,CAAC,SAAS,EAAE,OAAO,CAAC;aACtD,KAAK,eAAe,QAAQ,GAAG,CACzC,CAAA;IACD,MAAM,EAAE,CAAC,KAAK,CACZ,GAAG,CAAA,8BAA8B,UAAU,CAAC,WAAW,EAAE,OAAO,CAAC;aACxD,KAAK,KAAK,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC;uBAC1C,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,CACjE,CAAA;IAED,OAAO;QACL,OAAO;QAEP,KAAK,EAAE,KAAK,EAAE,KAAqB,EAAiB,EAAE;YACpD,cAAc,CAAC,KAAK,CAAC,CAAA;YACrB,MAAM,aAAa,GAAG,gBAAgB,CAAC,KAAK,CAAC,MAAM,CAAC,CAAA;YACpD,MAAM,OAAO,GAAG,QAAQ,CAAC,GAAG,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,IAAI,EAAE,CAAC,CAAA;YAExD,2EAA2E;YAC3E,uEAAuE;YACvE,uDAAuD;YACvD,MAAM,EAAE,CAAC,KAAK,CACZ,GAAG,CAAA,eAAe,KAAK;gBACf,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC;gBAC3C,UAAU,CAAC,cAAc,CAAC,UAAU,EAAE,OAAO,CAAC;gBAC9C,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC;gBAC1C,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC;gBAC1C,UAAU,CAAC,cAAc,CAAC,KAAK,EAAE,OAAO,CAAC;gBACzC,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC;gBAC3C,QAAQ;sBACF,KAAK,CAAC,EAAE,KAAK,KAAK,CAAC,UAAU,KAAK,KAAK,CAAC,MAAM,KAAK,KAAK,CAAC,MAAM;sBAC/D,WAAW,CAAC,KAAK,CAAC,KAAK,OAAO;kCAClB,aAAa,gBAAgB,OAAO;2BAC3C,UAAU,CAAC,cAAc,CAAC,UAAU,EAAE,OAAO,CAAC;2BAC9C,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC;;gBAEtD,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC,eAAe,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC;gBACnG,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC,eAAe,UAAU,CAAC,cAAc,CAAC,MAAM,EAAE,OAAO,CAAC;gBACnG,UAAU,CAAC,cAAc,CAAC,KAAK,EAAE,OAAO,CAAC,eAAe,UAAU,CAAC,cAAc,CAAC,KAAK,EAAE,OAAO,CAAC;gBACjG,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,eAAe,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC;gBACrG,QAAQ,eAAe,QAAQ,EAAE,CAC1C,CAAA;QACH,CAAC;QAED,MAAM,EAAE,KAAK,EAAE,SAA0B,EAAiB,EAAE;YAC1D,eAAe,CAAC,SAAS,CAAC,CAAA;YAC1B,MAAM,EAAE,CAAC,KAAK,CACZ,GAAG,CAAA,eAAe,KAAK;oBACX,UAAU,CAAC,cAAc,CAAC,UAAU,EAAE,OAAO,CAAC,MAAM,SAAS,CAAC,UAAU;oBACxE,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC,MAAM,SAAS,CAAC,EAAE,EAAE,CAC5E,CAAA;QACH,CAAC;QAED,MAAM,EAAE,KAAK,EAAE,KAAkB,EAAE,EAAE;YACnC,MAAM,UAAU,GAAG,cAAc,CAAC,KAAK,CAAC,CAAA;YACxC,IAAI,UAAU,CAAC,MAAM,CAAC,MAAM,KAAK,CAAC;gBAAE,OAAO,aAAa,CAAA;YAExD,MAAM,aAAa,GAAG,gBAAgB,CAAC,UAAU,CAAC,MAAM,CAAC,CAAA;YACzD,MAAM,UAAU,GAAG,OAAO,CAAC,UAAU,CAAC,MAAM,CAAC,CAAA;YAC7C,MAAM,OAAO,GAAG,YAAY,CAAC,OAAO,EAAE,UAAU,CAAC,CAAA;YAEjD,mEAAmE;YACnE,wEAAwE;YACxE,0DAA0D;YAC1D,MAAM,KAAK,GAAG,GAAG,CAAA,GAAG,QAAQ,kBAAkB,aAAa,gBAAgB,UAAU,GAAG,CAAA;YACxF,MAAM,IAAI,GAAG,GAAG,CAAA,WAAW,QAAQ,gBAAgB,aAAa,gBAAgB,UAAU,IAAI,CAAA;YAE9F,MAAM,KAAK,GAAG,MAAM,EAAE,CAAC,KAAK,CAC1B,GAAG,CAAA,UAAU,eAAe,CAAC,OAAO,CAAC,KAAK,IAAI,OAAO,UAAU,CAAC,OAAO,EAAE,OAAO,CAAC;mBACtE,KAAK;oBACJ,KAAK,CAAC,CAAC,KAAK,EAAE,GAAG,OAAO,CAAC,CAAC;uBACvB,UAAU,CAAC,OAAO,EAAE,OAAO,CAAC;uBAC5B,UAAU,CAAC,cAAc,CAAC,OAAO,EAAE,OAAO,CAAC;oBAC9C,QAAQ,CAAC,UAAU,CAAC,IAAI,GAAG,CAAC,CAAC,WAAW,QAAQ,CAAC,UAAU,CAAC,MAAM,CAAC,EAAE,CAClF,CAAA;YAED,OAAO,SAAS,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,EAAE,UAAU,CAAC,CAAA;QAC1E,CAAC;QAED,KAAK,EAAE,KAAK,EAAE,KAAK,EAAE,EAAE;YACrB,MAAM,KAAK,GACT,KAAK,EAAE,UAAU,KAAK,SAAS;gBAC7B,CAAC,CAAC,GAAG,CAAA,EAAE;gBACP,CAAC,CAAC,GAAG,CAAA,UAAU,UAAU,CAAC,cAAc,CAAC,UAAU,EAAE,OAAO,CAAC,MAAM,KAAK,CAAC,UAAU,EAAE,CAAA;YACzF,MAAM,EAAE,CAAC,KAAK,CAAC,GAAG,CAAA,eAAe,KAAK,GAAG,KAAK,EAAE,CAAC,CAAA;QACnD,CAAC;QAED,MAAM,EAAE,KAAK,IAA2B,EAAE;YACxC,MAAM,OAAO,GAAG,IAAI,CAAC,GAAG,EAAE,CAAA;YAC1B,MAAM,OAAO,GAAG,MAAM,EAAE,CAAC,KAAK,CAC5B,GAAG,CAAA,sBAAsB,UAAU,CAAC,OAAO,EAAE,OAAO,CAAC,SAAS,KAAK,EAAE,CACtE,CAAA;YACD,OAAO;gBACL,MAAM,EAAE,IAAI;gBACZ,MAAM,EAAE,UAAU;gBAClB,IAAI,EAAE,SAAS;gBACf,SAAS,EAAE,IAAI,CAAC,GAAG,EAAE,GAAG,OAAO;gBAC/B,OAAO,EAAE,oEAAoE;gBAC7E,OAAO,EAAE,EAAE,SAAS,EAAE,OAAO,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,KAAK,CAAC,EAAE;aACxD,CAAA;QACH,CAAC;KACF,CAAA;AACH,CAAC"}