@hatk/hatk 0.0.1-alpha.4 → 0.0.1-alpha.40

package/dist/database/dialect.d.ts

@@ -0,0 +1,45 @@
+import type { Dialect } from './ports.ts';
+export interface SqlDialect {
+    /** Map from lexicon type key to SQL column type */
+    typeMap: Record<string, string>;
+    /** Timestamp type name */
+    timestampType: string;
+    /** JSON type name */
+    jsonType: string;
+    /** Parameter placeholder for index (1-based). DuckDB/Postgres: $1  SQLite: ? */
+    param(index: number): string;
+    /** Whether the engine supports native bulk appenders (DuckDB) vs batched INSERT */
+    supportsAppender: boolean;
+    /** SQL for upsert — 'INSERT OR REPLACE' (DuckDB/SQLite) vs 'ON CONFLICT DO UPDATE' */
+    upsertPrefix: string;
+    /** Extract a string value from a JSON column. Returns SQL expression. */
+    jsonExtractString(column: string, path: string): string;
+    /** Aggregate strings from a JSON array. Returns SQL expression. */
+    jsonArrayStringAgg(column: string, jsonPath: string): string;
+    /** Information schema query to list user tables */
+    listTablesQuery: string;
+    /** CHECKPOINT or equivalent (for WAL compaction). null if not needed. */
+    checkpointSQL: string | null;
+    /** Current timestamp expression */
+    currentTimestamp: string;
+    /** ILIKE or equivalent for case-insensitive matching */
+    ilike: string;
+    /** Cast expression for safe timestamp parsing. DuckDB: TRY_CAST(x AS TIMESTAMP), SQLite: x */
+    tryCastTimestamp(expr: string): string;
+    /** COUNT(*)::INTEGER or equivalent */
+    countAsInteger: string;
+    /** GREATEST(...) or MAX(...) for multi-arg max */
+    greatest(exprs: string[]): string;
+    /** jaro_winkler_similarity or null if unsupported */
+    jaroWinklerSimilarity: string | null;
+    /** string_agg or group_concat */
+    stringAgg(column: string, separator: string): string;
+    /** CREATE SEQUENCE support */
+    supportsSequences: boolean;
+    /** SQL to get columns for a table. Returns rows with column_name/name and data_type/type. */
+    introspectColumnsQuery(tableName: string): string;
+}
+export declare const DUCKDB_DIALECT: SqlDialect;
+export declare const SQLITE_DIALECT: SqlDialect;
+export declare function getDialect(dialect: Dialect): SqlDialect;
+//# sourceMappingURL=dialect.d.ts.map

package/dist/database/dialect.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"dialect.d.ts","sourceRoot":"","sources":["../../src/database/dialect.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,OAAO,EAAE,MAAM,YAAY,CAAA;AAEzC,MAAM,WAAW,UAAU;IACzB,mDAAmD;IACnD,OAAO,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IAE/B,0BAA0B;IAC1B,aAAa,EAAE,MAAM,CAAA;IAErB,qBAAqB;IACrB,QAAQ,EAAE,MAAM,CAAA;IAEhB,gFAAgF;IAChF,KAAK,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAAA;IAE5B,mFAAmF;IACnF,gBAAgB,EAAE,OAAO,CAAA;IAEzB,sFAAsF;IACtF,YAAY,EAAE,MAAM,CAAA;IAEpB,yEAAyE;IACzE,iBAAiB,CAAC,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,CAAA;IAEvD,mEAAmE;IACnE,kBAAkB,CAAC,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,MAAM,GAAG,MAAM,CAAA;IAE5D,mDAAmD;IACnD,eAAe,EAAE,MAAM,CAAA;IAEvB,yEAAyE;IACzE,aAAa,EAAE,MAAM,GAAG,IAAI,CAAA;IAE5B,mCAAmC;IACnC,gBAAgB,EAAE,MAAM,CAAA;IAExB,wDAAwD;IACxD,KAAK,EAAE,MAAM,CAAA;IAEb,8FAA8F;IAC9F,gBAAgB,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,CAAA;IAEtC,sCAAsC;IACtC,cAAc,EAAE,MAAM,CAAA;IAEtB,kDAAkD;IAClD,QAAQ,CAAC,KAAK,EAAE,MAAM,EAAE,GAAG,MAAM,CAAA;IAEjC,qDAAqD;IACrD,qBAAqB,EAAE,MAAM,GAAG,IAAI,CAAA;IAEpC,iCAAiC;IACjC,SAAS,CAAC,MAAM,EAAE,MAAM,EAAE,SAAS,EAAE,MAAM,GAAG,MAAM,CAAA;IAEpD,8BAA8B;IAC9B,iBAAiB,EAAE,OAAO,CAAA;IAE1B,6FAA6F;IAC7F,sBAAsB,CAAC,SAAS,EAAE,MAAM,GAAG,MAAM,CAAA;CAClD;AAED,eAAO,MAAM,cAAc,EAAE,UA8B5B,CAAA;AAED,eAAO,MAAM,cAAc,EAAE,UA+B5B,CAAA;AAED,wBAAgB,UAAU,CAAC,OAAO,EAAE,OAAO,GAAG,UAAU,CASvD"}

package/dist/database/dialect.js

@@ -0,0 +1,72 @@
+export const DUCKDB_DIALECT = {
+    typeMap: {
+        text: 'TEXT',
+        integer: 'INTEGER',
+        bigint: 'BIGINT',
+        boolean: 'BOOLEAN',
+        blob: 'BLOB',
+        timestamp: 'TIMESTAMP',
+        timestamptz: 'TIMESTAMPTZ',
+        json: 'JSON',
+    },
+    timestampType: 'TIMESTAMP',
+    jsonType: 'JSON',
+    param: (i) => `$${i}`,
+    supportsAppender: true,
+    upsertPrefix: 'INSERT OR REPLACE INTO',
+    jsonExtractString: (col, path) => `json_extract_string(${col}, '${path}')`,
+    jsonArrayStringAgg: (col, path) => `list_string_agg(json_extract_string(${col}, '${path}'))`,
+    listTablesQuery: `SELECT table_name FROM information_schema.tables WHERE table_schema = 'main' AND table_name NOT LIKE '\\_%' ESCAPE '\\\\'`,
+    checkpointSQL: 'CHECKPOINT',
+    currentTimestamp: 'CURRENT_TIMESTAMP',
+    ilike: 'ILIKE',
+    tryCastTimestamp: (expr) => `TRY_CAST(${expr} AS TIMESTAMP)`,
+    countAsInteger: 'COUNT(*)::INTEGER',
+    greatest: (exprs) => `GREATEST(${exprs.join(', ')})`,
+    jaroWinklerSimilarity: 'jaro_winkler_similarity',
+    stringAgg: (col, sep) => `string_agg(${col}, ${sep})`,
+    supportsSequences: true,
+    introspectColumnsQuery: (tableName) => `SELECT column_name, data_type FROM information_schema.columns WHERE table_name = '${tableName}'`,
+};
+export const SQLITE_DIALECT = {
+    typeMap: {
+        text: 'TEXT',
+        integer: 'INTEGER',
+        bigint: 'INTEGER',
+        boolean: 'INTEGER',
+        blob: 'BLOB',
+        timestamp: 'TEXT',
+        timestamptz: 'TEXT',
+        json: 'TEXT',
+    },
+    timestampType: 'TEXT',
+    jsonType: 'TEXT',
+    param: (_i) => '?',
+    supportsAppender: false,
+    upsertPrefix: 'INSERT OR REPLACE INTO',
+    jsonExtractString: (col, path) => `json_extract(${col}, '${path}')`,
+    jsonArrayStringAgg: (col, path) => {
+        return `(SELECT group_concat(je.value, ' ') FROM json_each(${col}, '${path}') je)`;
+    },
+    listTablesQuery: `SELECT name AS table_name FROM sqlite_master WHERE type='table' AND name NOT LIKE '\\_%' ESCAPE '\\\\'`,
+    checkpointSQL: null,
+    currentTimestamp: 'CURRENT_TIMESTAMP',
+    ilike: 'LIKE',
+    tryCastTimestamp: (expr) => expr,
+    countAsInteger: 'CAST(COUNT(*) AS INTEGER)',
+    greatest: (exprs) => `MAX(${exprs.join(', ')})`,
+    jaroWinklerSimilarity: null,
+    stringAgg: (col, sep) => `group_concat(${col}, ${sep})`,
+    supportsSequences: false,
+    introspectColumnsQuery: (tableName) => `PRAGMA table_info("${tableName}")`,
+};
+export function getDialect(dialect) {
+    switch (dialect) {
+        case 'duckdb':
+            return DUCKDB_DIALECT;
+        case 'sqlite':
+            return SQLITE_DIALECT;
+        case 'postgres':
+            throw new Error('PostgreSQL adapter not yet implemented');
+    }
+}

package/dist/database/fts.d.ts

@@ -0,0 +1,27 @@
+import type { SearchPort } from './ports.ts';
+export declare function setSearchPort(port: SearchPort | null): void;
+export declare function hasSearchPort(): boolean;
+export declare function getSearchPort(): SearchPort | null;
+export declare function getSearchColumns(collection: string): string[];
+export declare function getLastRebuiltAt(collection: string): string | null;
+/**
+ * DuckDB FTS can't handle dots in table names (interprets them as catalog.schema.table).
+ * We create a shadow table with underscored names for FTS indexing.
+ */
+export declare function ftsTableName(collection: string): string;
+/**
+ * Build FTS index for a collection.
+ * Creates a shadow table copy and indexes all TEXT NOT NULL columns
+ * using Porter stemmer with English stopwords.
+ */
+export declare function buildFtsIndex(collection: string): Promise<void>;
+export declare function buildFtsRow(collection: string, uri: string): Promise<Record<string, string | null> | null>;
+export declare function updateFtsRecord(collection: string, uri: string): Promise<void>;
+export declare function deleteFtsRecord(collection: string, uri: string): Promise<void>;
+/**
+ * Strip English stop words from a search query, preserving non-stop-word terms.
+ * Returns the cleaned query string. If all words are stop words, returns the original query.
+ */
+export declare function stripStopWords(query: string): string;
+export declare function rebuildAllIndexes(collections: string[]): Promise<void>;
+//# sourceMappingURL=fts.d.ts.map

package/dist/database/fts.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"fts.d.ts","sourceRoot":"","sources":["../../src/database/fts.ts"],"names":[],"mappings":"AAGA,OAAO,KAAK,EAAE,UAAU,EAAE,MAAM,YAAY,CAAA;AAuE5C,wBAAgB,aAAa,CAAC,IAAI,EAAE,UAAU,GAAG,IAAI,GAAG,IAAI,CAE3D;AAED,wBAAgB,aAAa,IAAI,OAAO,CAEvC;AAED,wBAAgB,aAAa,IAAI,UAAU,GAAG,IAAI,CAEjD;AAWD,wBAAgB,gBAAgB,CAAC,UAAU,EAAE,MAAM,GAAG,MAAM,EAAE,CAE7D;AAED,wBAAgB,gBAAgB,CAAC,UAAU,EAAE,MAAM,GAAG,MAAM,GAAG,IAAI,CAElE;AAED;;;GAGG;AACH,wBAAgB,YAAY,CAAC,UAAU,EAAE,MAAM,GAAG,MAAM,CAEvD;AAgFD;;;;GAIG;AACH,wBAAsB,aAAa,CAAC,UAAU,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CAmBrE;AAED,wBAAsB,WAAW,CAAC,UAAU,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,MAAM,EAAE,MAAM,GAAG,IAAI,CAAC,GAAG,IAAI,CAAC,CAehH;AAED,wBAAsB,eAAe,CAAC,UAAU,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CAepF;AAED,wBAAsB,eAAe,CAAC,UAAU,EAAE,MAAM,EAAE,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC,CAYpF;AAokBD;;;GAGG;AACH,wBAAgB,cAAc,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAIpD;AAED,wBAAsB,iBAAiB,CAAC,WAAW,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC,IAAI,CAAC,CA4B5E"}