sensemaking 0.19.0 → 0.19.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/cjs/index.d.cts +1 -1
- package/dist/cjs/index.d.ts +1 -1
- package/dist/cjs/index.js +3 -0
- package/dist/cjs/index.js.map +1 -1
- package/dist/cjs/store/duckdb/fieldStats.js +87 -10
- package/dist/cjs/store/duckdb/fieldStats.js.map +1 -1
- package/dist/cjs/store/duckdb/open.js +3 -2
- package/dist/cjs/store/duckdb/open.js.map +1 -1
- package/dist/cjs/store/duckdb/reconcile.d.cts +2 -6
- package/dist/cjs/store/duckdb/reconcile.d.ts +2 -6
- package/dist/cjs/store/duckdb/reconcile.js +41 -581
- package/dist/cjs/store/duckdb/reconcile.js.map +1 -1
- package/dist/cjs/store/duckdb/store.js +5 -3
- package/dist/cjs/store/duckdb/store.js.map +1 -1
- package/dist/cjs/store/index.d.cts +1 -0
- package/dist/cjs/store/index.d.ts +1 -0
- package/dist/cjs/store/index.js +4 -0
- package/dist/cjs/store/index.js.map +1 -1
- package/dist/cjs/store/reconcile.d.cts +6 -0
- package/dist/cjs/store/reconcile.d.ts +6 -0
- package/dist/cjs/store/reconcile.js +750 -0
- package/dist/cjs/store/reconcile.js.map +1 -0
- package/dist/cjs/store/shared.d.cts +1 -0
- package/dist/cjs/store/shared.d.ts +1 -0
- package/dist/cjs/store/shared.js +37 -0
- package/dist/cjs/store/shared.js.map +1 -1
- package/dist/cjs/store/sqlite/connection.js +1 -1
- package/dist/cjs/store/sqlite/connection.js.map +1 -1
- package/dist/cjs/store/sqlite/open.js +33 -11
- package/dist/cjs/store/sqlite/open.js.map +1 -1
- package/dist/cjs/store/sqlite/reconcile.d.cts +5 -10
- package/dist/cjs/store/sqlite/reconcile.d.ts +5 -10
- package/dist/cjs/store/sqlite/reconcile.js +48 -840
- package/dist/cjs/store/sqlite/reconcile.js.map +1 -1
- package/dist/cjs/store/sqlite/signature.d.cts +6 -0
- package/dist/cjs/store/sqlite/signature.d.ts +6 -0
- package/dist/cjs/store/sqlite/signature.js +380 -0
- package/dist/cjs/store/sqlite/signature.js.map +1 -0
- package/dist/cjs/store/sqlite/store.js +5 -3
- package/dist/cjs/store/sqlite/store.js.map +1 -1
- package/dist/cjs/store/transaction.d.cts +2 -1
- package/dist/cjs/store/transaction.d.ts +2 -1
- package/dist/cjs/store/transaction.js +16 -6
- package/dist/cjs/store/transaction.js.map +1 -1
- package/dist/cjs/store/turso/connection.js +1 -1
- package/dist/cjs/store/turso/connection.js.map +1 -1
- package/dist/cjs/store/turso/open.d.cts +2 -0
- package/dist/cjs/store/turso/open.d.ts +2 -0
- package/dist/cjs/store/turso/open.js +115 -57
- package/dist/cjs/store/turso/open.js.map +1 -1
- package/dist/cjs/store/turso/reconcile.d.cts +3 -6
- package/dist/cjs/store/turso/reconcile.d.ts +3 -6
- package/dist/cjs/store/turso/reconcile.js +159 -596
- package/dist/cjs/store/turso/reconcile.js.map +1 -1
- package/dist/cjs/store/turso/store.js +3 -2
- package/dist/cjs/store/turso/store.js.map +1 -1
- package/dist/cjs/store/types.d.cts +12 -2
- package/dist/cjs/store/types.d.ts +12 -2
- package/dist/cjs/store/types.js +14 -0
- package/dist/cjs/store/types.js.map +1 -1
- package/dist/esm/index.d.ts +1 -1
- package/dist/esm/index.js +1 -1
- package/dist/esm/index.js.map +1 -1
- package/dist/esm/store/duckdb/fieldStats.js +32 -19
- package/dist/esm/store/duckdb/fieldStats.js.map +1 -1
- package/dist/esm/store/duckdb/open.js +3 -2
- package/dist/esm/store/duckdb/open.js.map +1 -1
- package/dist/esm/store/duckdb/reconcile.d.ts +2 -6
- package/dist/esm/store/duckdb/reconcile.js +20 -117
- package/dist/esm/store/duckdb/reconcile.js.map +1 -1
- package/dist/esm/store/duckdb/store.js +5 -3
- package/dist/esm/store/duckdb/store.js.map +1 -1
- package/dist/esm/store/index.d.ts +1 -0
- package/dist/esm/store/index.js +1 -0
- package/dist/esm/store/index.js.map +1 -1
- package/dist/esm/store/reconcile.d.ts +6 -0
- package/dist/esm/store/reconcile.js +130 -0
- package/dist/esm/store/reconcile.js.map +1 -0
- package/dist/esm/store/shared.d.ts +1 -0
- package/dist/esm/store/shared.js +9 -0
- package/dist/esm/store/shared.js.map +1 -1
- package/dist/esm/store/sqlite/connection.js +2 -2
- package/dist/esm/store/sqlite/connection.js.map +1 -1
- package/dist/esm/store/sqlite/open.js +28 -8
- package/dist/esm/store/sqlite/open.js.map +1 -1
- package/dist/esm/store/sqlite/reconcile.d.ts +5 -10
- package/dist/esm/store/sqlite/reconcile.js +31 -198
- package/dist/esm/store/sqlite/reconcile.js.map +1 -1
- package/dist/esm/store/sqlite/signature.d.ts +6 -0
- package/dist/esm/store/sqlite/signature.js +58 -0
- package/dist/esm/store/sqlite/signature.js.map +1 -0
- package/dist/esm/store/sqlite/store.js +5 -3
- package/dist/esm/store/sqlite/store.js.map +1 -1
- package/dist/esm/store/transaction.d.ts +2 -1
- package/dist/esm/store/transaction.js +11 -6
- package/dist/esm/store/transaction.js.map +1 -1
- package/dist/esm/store/turso/connection.js +2 -2
- package/dist/esm/store/turso/connection.js.map +1 -1
- package/dist/esm/store/turso/open.d.ts +2 -0
- package/dist/esm/store/turso/open.js +14 -4
- package/dist/esm/store/turso/open.js.map +1 -1
- package/dist/esm/store/turso/reconcile.d.ts +3 -6
- package/dist/esm/store/turso/reconcile.js +31 -132
- package/dist/esm/store/turso/reconcile.js.map +1 -1
- package/dist/esm/store/turso/store.js +3 -2
- package/dist/esm/store/turso/store.js.map +1 -1
- package/dist/esm/store/types.d.ts +12 -2
- package/dist/esm/store/types.js +12 -1
- package/dist/esm/store/types.js.map +1 -1
- package/package.json +1 -1
package/dist/cjs/index.d.cts
CHANGED
|
@@ -7,7 +7,7 @@ export { SenseError } from './errors.js';
|
|
|
7
7
|
export type { Format, Row, RowFormat } from './output/output.js';
|
|
8
8
|
export { printRows } from './output/output.js';
|
|
9
9
|
export type { OpenResult } from './store/index.js';
|
|
10
|
-
export { clearCache, openStore as open } from './store/index.js';
|
|
10
|
+
export { CAPABILITY_NAMES, clearCache, openStore as open } from './store/index.js';
|
|
11
11
|
export type { Capability, DocumentStore, FieldStat, LexicalHit, LexicalIndex, LexicalQueryOptions, RawStatement, RunResult, SqlSession, Statement, Store, VectorCandidate, VectorSimilar, VectorStore, VectorWriteRow } from './store/types.js';
|
|
12
12
|
export type { WatchEvent, WatchOptions } from './watch.js';
|
|
13
13
|
export { runWatch } from './watch.js';
|
package/dist/cjs/index.d.ts
CHANGED
|
@@ -7,7 +7,7 @@ export { SenseError } from './errors.js';
|
|
|
7
7
|
export type { Format, Row, RowFormat } from './output/output.js';
|
|
8
8
|
export { printRows } from './output/output.js';
|
|
9
9
|
export type { OpenResult } from './store/index.js';
|
|
10
|
-
export { clearCache, openStore as open } from './store/index.js';
|
|
10
|
+
export { CAPABILITY_NAMES, clearCache, openStore as open } from './store/index.js';
|
|
11
11
|
export type { Capability, DocumentStore, FieldStat, LexicalHit, LexicalIndex, LexicalQueryOptions, RawStatement, RunResult, SqlSession, Statement, Store, VectorCandidate, VectorSimilar, VectorStore, VectorWriteRow } from './store/types.js';
|
|
12
12
|
export type { WatchEvent, WatchOptions } from './watch.js';
|
|
13
13
|
export { runWatch } from './watch.js';
|
package/dist/cjs/index.js
CHANGED
package/dist/cjs/index.js.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/index.ts"],"sourcesContent":["// Public library API. Deliberately small: every export is a stability promise;\n// internals (feature registry, graph, scan, meta) stay module-private.\n\nexport type { Peek, PresetCoverage, SearchOptions, TreeMap } from './commands/index.ts';\nexport { mapTree, peek, presetCoverage, search } from './commands/index.ts';\nexport type { Config, EmbedConfig, FeatureName, InitOverrides, Preset, ResolvedConfig, SavedQuery, SavedSearch, SearchOverrides, SignalName, SignalWeights, StoreName } from './config/index.ts';\nexport { CONFIG_FILENAME, initConfig, loadConfig, migrateConfig, SIGNAL_NAMES, STATE_DIR, STORE_NAMES, SUPPORTED_CONFIG_VERSION } from './config/index.ts';\nexport type { SenseErrorCode } from './errors.ts';\nexport { SenseError } from './errors.ts';\nexport type { Format, Row, RowFormat } from './output/output.ts';\nexport { printRows } from './output/output.ts';\nexport type { OpenResult } from './store/index.ts';\nexport { clearCache, openStore as open } from './store/index.ts';\nexport type { Capability, DocumentStore, FieldStat, LexicalHit, LexicalIndex, LexicalQueryOptions, RawStatement, RunResult, SqlSession, Statement, Store, VectorCandidate, VectorSimilar, VectorStore, VectorWriteRow } from './store/types.ts';\n\nexport type { WatchEvent, WatchOptions } from './watch.ts';\nexport { runWatch } from './watch.ts';\n"],"names":["CONFIG_FILENAME","SIGNAL_NAMES","STATE_DIR","STORE_NAMES","SUPPORTED_CONFIG_VERSION","SenseError","clearCache","initConfig","loadConfig","mapTree","migrateConfig","open","openStore","peek","presetCoverage","printRows","runWatch","search"],"mappings":"AAAA,+EAA+E;AAC/E,uEAAuE;;;;;;;;;;;;
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/index.ts"],"sourcesContent":["// Public library API. Deliberately small: every export is a stability promise;\n// internals (feature registry, graph, scan, meta) stay module-private.\n\nexport type { Peek, PresetCoverage, SearchOptions, TreeMap } from './commands/index.ts';\nexport { mapTree, peek, presetCoverage, search } from './commands/index.ts';\nexport type { Config, EmbedConfig, FeatureName, InitOverrides, Preset, ResolvedConfig, SavedQuery, SavedSearch, SearchOverrides, SignalName, SignalWeights, StoreName } from './config/index.ts';\nexport { CONFIG_FILENAME, initConfig, loadConfig, migrateConfig, SIGNAL_NAMES, STATE_DIR, STORE_NAMES, SUPPORTED_CONFIG_VERSION } from './config/index.ts';\nexport type { SenseErrorCode } from './errors.ts';\nexport { SenseError } from './errors.ts';\nexport type { Format, Row, RowFormat } from './output/output.ts';\nexport { printRows } from './output/output.ts';\nexport type { OpenResult } from './store/index.ts';\nexport { CAPABILITY_NAMES, clearCache, openStore as open } from './store/index.ts';\nexport type { Capability, DocumentStore, FieldStat, LexicalHit, LexicalIndex, LexicalQueryOptions, RawStatement, RunResult, SqlSession, Statement, Store, VectorCandidate, VectorSimilar, VectorStore, VectorWriteRow } from './store/types.ts';\n\nexport type { WatchEvent, WatchOptions } from './watch.ts';\nexport { runWatch } from './watch.ts';\n"],"names":["CAPABILITY_NAMES","CONFIG_FILENAME","SIGNAL_NAMES","STATE_DIR","STORE_NAMES","SUPPORTED_CONFIG_VERSION","SenseError","clearCache","initConfig","loadConfig","mapTree","migrateConfig","open","openStore","peek","presetCoverage","printRows","runWatch","search"],"mappings":"AAAA,+EAA+E;AAC/E,uEAAuE;;;;;;;;;;;;QAW9DA;eAAAA,0BAAgB;;QANhBC;eAAAA,yBAAe;;QAAyCC;eAAAA,sBAAY;;QAAEC;eAAAA,mBAAS;;QAAEC;eAAAA,qBAAW;;QAAEC;eAAAA,kCAAwB;;QAEtHC;eAAAA,oBAAU;;QAIQC;eAAAA,oBAAU;;QANXC;eAAAA,oBAAU;;QAAEC;eAAAA,oBAAU;;QAFvCC;eAAAA,gBAAO;;QAEkCC;eAAAA,uBAAa;;QAMXC;eAAbC,mBAAS;;QAR9BC;eAAAA,aAAI;;QAAEC;eAAAA,uBAAc;;QAM7BC;eAAAA,mBAAS;;QAMTC;eAAAA,iBAAQ;;QAZuBC;eAAAA,eAAM;;;uBAAQ;wBAEiF;wBAE5G;wBAED;wBAEsC;uBAIvC"}
|
|
@@ -14,6 +14,9 @@ function _array_like_to_array(arr, len) {
|
|
|
14
14
|
for(var i = 0, arr2 = new Array(len); i < len; i++)arr2[i] = arr[i];
|
|
15
15
|
return arr2;
|
|
16
16
|
}
|
|
17
|
+
function _array_with_holes(arr) {
|
|
18
|
+
if (Array.isArray(arr)) return arr;
|
|
19
|
+
}
|
|
17
20
|
function _array_without_holes(arr) {
|
|
18
21
|
if (Array.isArray(arr)) return _array_like_to_array(arr);
|
|
19
22
|
}
|
|
@@ -48,9 +51,39 @@ function _iterable_to_array(iter) {
|
|
|
48
51
|
return Array.from(iter);
|
|
49
52
|
}
|
|
50
53
|
}
|
|
54
|
+
function _iterable_to_array_limit(arr, i) {
|
|
55
|
+
var _i = arr == null ? null : typeof Symbol !== "undefined" && arr[Symbol.iterator] || arr["@@iterator"];
|
|
56
|
+
if (_i == null) return;
|
|
57
|
+
var _arr = [];
|
|
58
|
+
var _n = true;
|
|
59
|
+
var _d = false;
|
|
60
|
+
var _s, _e;
|
|
61
|
+
try {
|
|
62
|
+
for(_i = _i.call(arr); !(_n = (_s = _i.next()).done); _n = true){
|
|
63
|
+
_arr.push(_s.value);
|
|
64
|
+
if (i && _arr.length === i) break;
|
|
65
|
+
}
|
|
66
|
+
} catch (err) {
|
|
67
|
+
_d = true;
|
|
68
|
+
_e = err;
|
|
69
|
+
} finally{
|
|
70
|
+
try {
|
|
71
|
+
if (!_n && _i["return"] != null) _i["return"]();
|
|
72
|
+
} finally{
|
|
73
|
+
if (_d) throw _e;
|
|
74
|
+
}
|
|
75
|
+
}
|
|
76
|
+
return _arr;
|
|
77
|
+
}
|
|
78
|
+
function _non_iterable_rest() {
|
|
79
|
+
throw new TypeError("Invalid attempt to destructure non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method.");
|
|
80
|
+
}
|
|
51
81
|
function _non_iterable_spread() {
|
|
52
82
|
throw new TypeError("Invalid attempt to spread non-iterable instance.\nIn order to be iterable, non-array objects must have a [Symbol.iterator]() method.");
|
|
53
83
|
}
|
|
84
|
+
function _sliced_to_array(arr, i) {
|
|
85
|
+
return _array_with_holes(arr) || _iterable_to_array_limit(arr, i) || _unsupported_iterable_to_array(arr, i) || _non_iterable_rest();
|
|
86
|
+
}
|
|
54
87
|
function _to_consumable_array(arr) {
|
|
55
88
|
return _array_without_holes(arr) || _iterable_to_array(arr) || _unsupported_iterable_to_array(arr) || _non_iterable_spread();
|
|
56
89
|
}
|
|
@@ -169,13 +202,30 @@ function classifyVariantType(name, field) {
|
|
|
169
202
|
if (name === 'VARCHAR') return 'text';
|
|
170
203
|
throw new Error('duckdb: variant_typeof() returned unrecognized type "'.concat(name, '" for column "').concat(field, '"'));
|
|
171
204
|
}
|
|
205
|
+
// duckdb degrades superlinearly in the number of string_agg(DISTINCT variant_typeof()) aggregates in
|
|
206
|
+
// one projection: 301 columns cost 1148ms in one query against 10ms for COUNT alone over the same
|
|
207
|
+
// columns, so it is the aggregate count, not the width. Measured chunk sizes at 301 columns:
|
|
208
|
+
// 10 -> 71ms, 16 -> 81ms, 25 -> 88ms, 40 -> 110ms, 64 -> 158ms, flat between 10 and 25. Chunking
|
|
209
|
+
// also wins on an ordinary tree (31 columns: 17ms in one query, 11ms chunked), so there is no
|
|
210
|
+
// small-tree cost to trade against. sqlite and turso need none of this: GROUP_CONCAT over 300
|
|
211
|
+
// columns runs in 7ms.
|
|
212
|
+
var AGGREGATES_PER_QUERY = 16;
|
|
172
213
|
function fieldStats(conn, columns, scopeWhere) {
|
|
173
214
|
return _async_to_generator(function() {
|
|
174
|
-
var exprs, stmt, row;
|
|
215
|
+
var stats, start, chunk, exprs, stmt, row, _iteratorNormalCompletion, _didIteratorError, _iteratorError, _loop, _iterator, _step;
|
|
175
216
|
return _ts_generator(this, function(_state) {
|
|
176
217
|
switch(_state.label){
|
|
177
218
|
case 0:
|
|
178
|
-
|
|
219
|
+
stats = [];
|
|
220
|
+
start = 0;
|
|
221
|
+
_state.label = 1;
|
|
222
|
+
case 1:
|
|
223
|
+
if (!(start < columns.length)) return [
|
|
224
|
+
3,
|
|
225
|
+
5
|
|
226
|
+
];
|
|
227
|
+
chunk = columns.slice(start, start + AGGREGATES_PER_QUERY);
|
|
228
|
+
exprs = chunk.map(function(name, i) {
|
|
179
229
|
var quoted = (0, _sharedts.quoteIdent)(name);
|
|
180
230
|
return "COUNT(".concat(quoted, ") AS n").concat(i, ", string_agg(DISTINCT variant_typeof(").concat(quoted, "), '|') FILTER (WHERE ").concat(quoted, " IS NOT NULL) AS t").concat(i);
|
|
181
231
|
});
|
|
@@ -183,28 +233,55 @@ function fieldStats(conn, columns, scopeWhere) {
|
|
|
183
233
|
4,
|
|
184
234
|
conn.prepare("SELECT ".concat(exprs.join(', '), " FROM frontmatter ").concat(scopeWhere))
|
|
185
235
|
];
|
|
186
|
-
case
|
|
236
|
+
case 2:
|
|
187
237
|
stmt = _state.sent();
|
|
188
238
|
return [
|
|
189
239
|
4,
|
|
190
240
|
stmt.get()
|
|
191
241
|
];
|
|
192
|
-
case
|
|
242
|
+
case 3:
|
|
193
243
|
row = _state.sent();
|
|
194
|
-
|
|
195
|
-
|
|
196
|
-
|
|
244
|
+
_iteratorNormalCompletion = true, _didIteratorError = false, _iteratorError = undefined;
|
|
245
|
+
try {
|
|
246
|
+
_loop = function() {
|
|
247
|
+
var _step_value = _sliced_to_array(_step.value, 2), i = _step_value[0], name = _step_value[1];
|
|
197
248
|
var _row_, _row_1;
|
|
198
249
|
var raw = ((_row_ = row["t".concat(i)]) !== null && _row_ !== void 0 ? _row_ : '').split('|').filter(Boolean);
|
|
199
250
|
var types = _to_consumable_array(new Set(raw.map(function(t) {
|
|
200
251
|
return classifyVariantType(t, name);
|
|
201
252
|
}))).sort();
|
|
202
|
-
|
|
253
|
+
stats.push({
|
|
203
254
|
field: name,
|
|
204
255
|
coverage: Number((_row_1 = row["n".concat(i)]) !== null && _row_1 !== void 0 ? _row_1 : 0),
|
|
205
256
|
type: types.join(',')
|
|
206
|
-
};
|
|
207
|
-
}
|
|
257
|
+
});
|
|
258
|
+
};
|
|
259
|
+
for(_iterator = chunk.entries()[Symbol.iterator](); !(_iteratorNormalCompletion = (_step = _iterator.next()).done); _iteratorNormalCompletion = true)_loop();
|
|
260
|
+
} catch (err) {
|
|
261
|
+
_didIteratorError = true;
|
|
262
|
+
_iteratorError = err;
|
|
263
|
+
} finally{
|
|
264
|
+
try {
|
|
265
|
+
if (!_iteratorNormalCompletion && _iterator.return != null) {
|
|
266
|
+
_iterator.return();
|
|
267
|
+
}
|
|
268
|
+
} finally{
|
|
269
|
+
if (_didIteratorError) {
|
|
270
|
+
throw _iteratorError;
|
|
271
|
+
}
|
|
272
|
+
}
|
|
273
|
+
}
|
|
274
|
+
_state.label = 4;
|
|
275
|
+
case 4:
|
|
276
|
+
start += AGGREGATES_PER_QUERY;
|
|
277
|
+
return [
|
|
278
|
+
3,
|
|
279
|
+
1
|
|
280
|
+
];
|
|
281
|
+
case 5:
|
|
282
|
+
return [
|
|
283
|
+
2,
|
|
284
|
+
stats
|
|
208
285
|
];
|
|
209
286
|
}
|
|
210
287
|
});
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/fieldStats.ts"],"sourcesContent":["import { quoteIdent } from '../shared.ts';\nimport type { Connection, FieldStat } from '../types.ts';\n\n// DuckDB's frontmatter columns are VARIANT (reconcile.ts), so typeof() always reports VARIANT; variant_typeof() reports the boxed native\n// type instead -- INT128, DOUBLE, VARCHAR, or a literal's narrower type like DECIMAL(p, s) -- mapped onto sqlite/turso's integer/real/text vocabulary; an unrecognized name throws (no-silent-modes).\nfunction classifyVariantType(name: string, field: string): 'integer' | 'real' | 'text' {\n if (/^U?INT\\d+$/.test(name)) return 'integer';\n if (name === 'FLOAT' || name === 'DOUBLE' || name.startsWith('DECIMAL(')) return 'real';\n if (name === 'VARCHAR') return 'text';\n throw new Error(`duckdb: variant_typeof() returned unrecognized type \"${name}\" for column \"${field}\"`);\n}\n\n//
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/fieldStats.ts"],"sourcesContent":["import { quoteIdent } from '../shared.ts';\nimport type { Connection, FieldStat } from '../types.ts';\n\n// DuckDB's frontmatter columns are VARIANT (reconcile.ts), so typeof() always reports VARIANT; variant_typeof() reports the boxed native\n// type instead -- INT128, DOUBLE, VARCHAR, or a literal's narrower type like DECIMAL(p, s) -- mapped onto sqlite/turso's integer/real/text vocabulary; an unrecognized name throws (no-silent-modes).\nfunction classifyVariantType(name: string, field: string): 'integer' | 'real' | 'text' {\n if (/^U?INT\\d+$/.test(name)) return 'integer';\n if (name === 'FLOAT' || name === 'DOUBLE' || name.startsWith('DECIMAL(')) return 'real';\n if (name === 'VARCHAR') return 'text';\n throw new Error(`duckdb: variant_typeof() returned unrecognized type \"${name}\" for column \"${field}\"`);\n}\n\n// duckdb degrades superlinearly in the number of string_agg(DISTINCT variant_typeof()) aggregates in\n// one projection: 301 columns cost 1148ms in one query against 10ms for COUNT alone over the same\n// columns, so it is the aggregate count, not the width. Measured chunk sizes at 301 columns:\n// 10 -> 71ms, 16 -> 81ms, 25 -> 88ms, 40 -> 110ms, 64 -> 158ms, flat between 10 and 25. Chunking\n// also wins on an ordinary tree (31 columns: 17ms in one query, 11ms chunked), so there is no\n// small-tree cost to trade against. sqlite and turso need none of this: GROUP_CONCAT over 300\n// columns runs in 7ms.\nconst AGGREGATES_PER_QUERY = 16;\n\n// COUNT + string_agg(DISTINCT variant_typeof()) per column keeps this O(columns), not O(rows x columns).\n// '|' is the separator, not ',': DECIMAL's display name (\"DECIMAL(4, 2)\") contains a comma.\nexport async function fieldStats(conn: Connection, columns: string[], scopeWhere: string): Promise<FieldStat[]> {\n const stats: FieldStat[] = [];\n for (let start = 0; start < columns.length; start += AGGREGATES_PER_QUERY) {\n const chunk = columns.slice(start, start + AGGREGATES_PER_QUERY);\n const exprs = chunk.map((name, i) => {\n const quoted = quoteIdent(name);\n return `COUNT(${quoted}) AS n${i}, string_agg(DISTINCT variant_typeof(${quoted}), '|') FILTER (WHERE ${quoted} IS NOT NULL) AS t${i}`;\n });\n const stmt = await conn.prepare(`SELECT ${exprs.join(', ')} FROM frontmatter ${scopeWhere}`);\n const row = (await stmt.get()) as Record<string, number | string | null>;\n for (const [i, name] of chunk.entries()) {\n const raw = ((row[`t${i}`] as string) ?? '').split('|').filter(Boolean);\n const types = [...new Set(raw.map((t) => classifyVariantType(t, name)))].sort();\n stats.push({ field: name, coverage: Number(row[`n${i}`] ?? 0), type: types.join(',') });\n }\n }\n return stats;\n}\n"],"names":["fieldStats","classifyVariantType","name","field","test","startsWith","Error","AGGREGATES_PER_QUERY","conn","columns","scopeWhere","stats","start","chunk","exprs","stmt","row","length","slice","map","i","quoted","quoteIdent","prepare","join","get","raw","split","filter","Boolean","types","Set","t","sort","push","coverage","Number","type","entries"],"mappings":";;;;+BAuBsBA;;;eAAAA;;;wBAvBK;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAG3B,yIAAyI;AACzI,sMAAsM;AACtM,SAASC,oBAAoBC,IAAY,EAAEC,KAAa;IACtD,IAAI,aAAaC,IAAI,CAACF,OAAO,OAAO;IACpC,IAAIA,SAAS,WAAWA,SAAS,YAAYA,KAAKG,UAAU,CAAC,aAAa,OAAO;IACjF,IAAIH,SAAS,WAAW,OAAO;IAC/B,MAAM,IAAII,MAAM,AAAC,wDAA4EH,OAArBD,MAAK,kBAAsB,OAANC,OAAM;AACrG;AAEA,qGAAqG;AACrG,kGAAkG;AAClG,6FAA6F;AAC7F,iGAAiG;AACjG,8FAA8F;AAC9F,8FAA8F;AAC9F,uBAAuB;AACvB,IAAMI,uBAAuB;AAItB,SAAeP,WAAWQ,IAAgB,EAAEC,OAAiB,EAAEC,UAAkB;;YAChFC,OACGC,OACDC,OACAC,OAIAC,MACAC,KACD,2BAAA,mBAAA,uBAAA,WAAA;;;;oBATDL;oBACGC,QAAQ;;;yBAAGA,CAAAA,QAAQH,QAAQQ,MAAM,AAAD;;;;oBACjCJ,QAAQJ,QAAQS,KAAK,CAACN,OAAOA,QAAQL;oBACrCO,QAAQD,MAAMM,GAAG,CAAC,SAACjB,MAAMkB;wBAC7B,IAAMC,SAASC,IAAAA,oBAAU,EAACpB;wBAC1B,OAAO,AAAC,SAAuBkB,OAAfC,QAAO,UAAiDA,OAAzCD,GAAE,yCAAsEC,OAA/BA,QAAO,0BAAmDD,OAA3BC,QAAO,sBAAsB,OAAFD;oBACpI;oBACa;;wBAAMZ,KAAKe,OAAO,CAAC,AAAC,UAA8Cb,OAArCI,MAAMU,IAAI,CAAC,OAAM,sBAA+B,OAAXd;;;oBAAzEK,OAAO;oBACA;;wBAAMA,KAAKU,GAAG;;;oBAArBT,MAAO;oBACR,kCAAA,2BAAA;;;4BAAA,mCAAA,iBAAOI,oBAAGlB;gCACCc,OAE6BA;4BAF3C,IAAMU,MAAM,EAAEV,QAAAA,GAAG,CAAC,AAAC,IAAK,OAAFI,GAAI,cAAZJ,mBAAAA,QAA2B,IAAIW,KAAK,CAAC,KAAKC,MAAM,CAACC;4BAC/D,IAAMC,QAAQ,AAAC,qBAAG,IAAIC,IAAIL,IAAIP,GAAG,CAAC,SAACa;uCAAM/B,oBAAoB+B,GAAG9B;iCAAS+B,IAAI;4BAC7EtB,MAAMuB,IAAI,CAAC;gCAAE/B,OAAOD;gCAAMiC,UAAUC,QAAOpB,SAAAA,GAAG,CAAC,AAAC,IAAK,OAAFI,GAAI,cAAZJ,oBAAAA,SAAgB;gCAAIqB,MAAMP,MAAMN,IAAI,CAAC;4BAAK;wBACvF;wBAJA,IAAK,YAAmBX,MAAMyB,OAAO,yBAAhC,6BAAA,QAAA,yBAAA;;wBAAA;wBAAA;;;iCAAA,6BAAA;gCAAA;;;gCAAA;sCAAA;;;;;;oBARqC1B,SAASL;;;;;;oBAcrD;;wBAAOI;;;;IACT"}
|
|
@@ -27,10 +27,11 @@ var _typests = require("../../embed/types.js");
|
|
|
27
27
|
var _errorsts = require("../../errors.js");
|
|
28
28
|
var _indexts1 = require("../../features/index.js");
|
|
29
29
|
var _cachets = require("../cache.js");
|
|
30
|
+
var _reconcilets = require("../reconcile.js");
|
|
30
31
|
var _sharedts = require("../shared.js");
|
|
31
32
|
var _connectionts = require("./connection.js");
|
|
32
33
|
var _nativets = require("./native.js");
|
|
33
|
-
var
|
|
34
|
+
var _reconcilets1 = require("./reconcile.js");
|
|
34
35
|
var _sqlfunctionsts = require("./sql-functions.js");
|
|
35
36
|
var _storets = require("./store.js");
|
|
36
37
|
function asyncGeneratorStep(gen, resolve, reject, _next, _throw, key, arg) {
|
|
@@ -437,7 +438,7 @@ function connect(cfg) {
|
|
|
437
438
|
_state.sent();
|
|
438
439
|
return [
|
|
439
440
|
4,
|
|
440
|
-
(0, _reconcilets.reconcile)(conn, cfg, cfg.baseDir)
|
|
441
|
+
(0, _reconcilets.reconcile)(conn, cfg, cfg.baseDir, _reconcilets1.duckdbDialect)
|
|
441
442
|
];
|
|
442
443
|
case 15:
|
|
443
444
|
_ref = _state.sent(), parsed = _ref.parsed, warnings = _ref.warnings;
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/open.ts"],"sourcesContent":["import { mkdirSync } from 'node:fs';\nimport { join } from 'node:path';\nimport type { DuckDBConnection, DuckDBInstance } from '@duckdb/node-api';\nimport type { Config, ResolvedConfig } from '../../config/index.ts';\nimport { featureSignature, STATE_DIR } from '../../config/index.ts';\nimport { rekeyChunkText } from '../../embed/handoff.ts';\nimport { STORE_DIMS } from '../../embed/types.ts';\nimport { SenseError } from '../../errors.ts';\nimport { activeFeatures, FEATURES } from '../../features/index.ts';\nimport { clearCache } from '../cache.ts';\nimport { getMeta, setMeta } from '../shared.ts';\nimport type { Connection, Store } from '../types.ts';\nimport { createConnection } from './connection.ts';\nimport { DUCKDB_PACKAGE, duckdbApi } from './native.ts';\nimport { reconcile } from './reconcile.ts';\nimport { registerFunctions } from './sql-functions.ts';\nimport { createStore } from './store.ts';\n\nexport const DB_FILENAME = 'cache.duckdb';\n// Independent of sqlite's SCHEMA_VERSION -- the two stores' cache shapes evolve separately (VARIANT frontmatter columns vs untyped).\n// The store name already joins the feature signature, so switching a config's `store` key rebuilds rather than reusing the other engine's cache.\nexport const SCHEMA_VERSION = '2';\n\nexport interface OpenResult {\n store: Store;\n cfg: ResolvedConfig;\n dbPath: string;\n parsed: number;\n warnings: string[];\n}\n\n// `content` is a plain table (not FTS-virtual): the fts index is built over it lazily, only when a lexical query runs (lexical.ts), and read directly for contains() verification either way.\n// No tokenizer resolution: this store always uses the fts extension's default (porter) stemmer; sqlite's configurable `content.tokenize` is not read here.\nasync function ensureSchema(conn: Connection, cfg: Config): Promise<void> {\n await conn.exec(`CREATE TABLE IF NOT EXISTS frontmatter (\"path\" TEXT PRIMARY KEY, \"_mtime\" DOUBLE, \"_ctime\" DOUBLE, \"_size\" INTEGER, \"_parse_error\" TEXT)`);\n await conn.exec(`CREATE TABLE IF NOT EXISTS content (\"path\" TEXT PRIMARY KEY, title TEXT, summary TEXT, text TEXT)`);\n await conn.exec(`CREATE TABLE IF NOT EXISTS preset_files (\"path\" TEXT, preset TEXT, PRIMARY KEY (\"path\", preset))`);\n await conn.exec('CREATE INDEX IF NOT EXISTS preset_files_preset ON preset_files(preset)');\n for (const feature of activeFeatures(cfg)) {\n // Native FLOAT[STORE_DIMS] instead of the embed feature's engine-neutral BLOB+scale DDL (vectors.ts). `scale` is kept, unused,\n // so the feature's shared reconcile-time INSERT/DELETE (features/embed.ts) names a column that exists on both stores.\n if (feature.name === 'embed') {\n await conn.exec(`CREATE TABLE IF NOT EXISTS embeddings (\"path\" TEXT, chunk INTEGER, start_line INTEGER, end_line INTEGER, scale REAL, vector FLOAT[${STORE_DIMS}], PRIMARY KEY (\"path\", chunk))`);\n continue;\n }\n await feature.schema(conn);\n }\n if ((await getMeta(conn, 'schema_version')) === null) await setMeta(conn, 'schema_version', SCHEMA_VERSION);\n if ((await getMeta(conn, 'features')) === null) await setMeta(conn, 'features', featureSignature(cfg, FEATURES));\n}\n\nasync function connect(cfg: ResolvedConfig): Promise<{ instance: DuckDBInstance; duckdb: DuckDBConnection; conn: Connection; cfg: ResolvedConfig; dbPath: string; parsed: number; warnings: string[] }> {\n const stateDir = join(cfg.baseDir, STATE_DIR);\n mkdirSync(stateDir, { recursive: true });\n const dbPath = join(stateDir, DB_FILENAME);\n\n // Dynamic, not a top-level import: sqlite trees must never attempt to resolve this optional peer dependency, so nothing imports\n // it as a value until a duckdb tree opens (types-only imports are erased). Installed on first use if missing, shared with sql-functions.ts and vectors.ts via native.ts's duckdbApi.\n let DuckDBInstance: typeof import('@duckdb/node-api').DuckDBInstance;\n try {\n ({ DuckDBInstance } = await duckdbApi());\n } catch (err) {\n if (err instanceof SenseError) throw err;\n throw new SenseError('STORE_DEPENDENCY_MISSING', `store \"duckdb\" needs the ${DUCKDB_PACKAGE} package (${(err as Error).message})`);\n }\n\n let duckdb: DuckDBConnection;\n let instance: DuckDBInstance | undefined;\n try {\n // On-disk files default to an older storage format for cross-version compatibility, which rejects VARIANT columns (\"VARIANT\n // columns are not supported in storage versions prior to v1.5.0\"); this store's dynamic frontmatter columns need VARIANT (reconcile.ts), so the floor is pinned explicitly.\n instance = await DuckDBInstance.create(dbPath, { storage_compatibility_version: 'v1.5.0' });\n duckdb = await instance.connect();\n } catch (err) {\n // create() may have succeeded before connect() failed: close it, or its WAL stays open.\n instance?.closeSync();\n throw new SenseError('STORE_DEPENDENCY_MISSING', `store \"duckdb\" failed to open ${dbPath}: ${(err as Error).message}`);\n }\n // A throw below (schema, reconcile's COLUMN_LIMIT) would leak the open instance, whose WAL\n // then locks the .duckdb file undeletable on Windows. closed marks the rebuild branch.\n let closed = false;\n try {\n await registerFunctions(duckdb);\n const conn = createConnection(duckdb);\n\n await conn.exec('CREATE TABLE IF NOT EXISTS meta (key TEXT PRIMARY KEY, value TEXT)');\n\n const version = await getMeta(conn, 'schema_version');\n const features = await getMeta(conn, 'features');\n const wantFeatures = featureSignature(cfg, FEATURES);\n if ((version !== null && version !== SCHEMA_VERSION) || (features !== null && features !== wantFeatures)) {\n // Reconcile only reparses changed files, so a stale cache can't be patched incrementally --\n // rebuild instead, same as sqlite's open() (see its comment for why this is announced).\n const reason = version !== null && version !== SCHEMA_VERSION ? 'cache format changed (new sensemaking version)' : 'config change (features, tokenizer, or presets)';\n console.error(`sense: ${reason}; rebuilding the index`);\n // Close before clearCache deletes the files underneath the still-open instance.\n closed = true;\n duckdb.disconnectSync();\n instance.closeSync();\n clearCache(cfg);\n return connect(cfg);\n }\n\n await ensureSchema(conn, cfg);\n\n const { parsed, warnings } = await reconcile(conn, cfg, cfg.baseDir);\n return { instance, duckdb, conn, cfg, dbPath, parsed, warnings };\n } catch (err) {\n if (!closed) {\n duckdb.disconnectSync();\n instance?.closeSync();\n }\n throw err;\n }\n}\n\nexport async function openDuckdb(cfg: ResolvedConfig): Promise<OpenResult> {\n const { instance, duckdb, conn, cfg: resolvedCfg, dbPath, parsed, warnings } = await connect(cfg);\n const store = createStore(instance, duckdb, conn, resolvedCfg, resolvedCfg.baseDir);\n // reconcile ran before this object existed, so its chunk text is keyed by the connection.\n rekeyChunkText(conn, store);\n return { store: store, cfg: resolvedCfg, dbPath, parsed, warnings };\n}\n"],"names":["DB_FILENAME","SCHEMA_VERSION","openDuckdb","ensureSchema","conn","cfg","feature","exec","activeFeatures","name","STORE_DIMS","schema","getMeta","setMeta","featureSignature","FEATURES","connect","stateDir","dbPath","DuckDBInstance","err","duckdb","instance","closed","version","features","wantFeatures","reason","parsed","warnings","join","baseDir","STATE_DIR","mkdirSync","recursive","duckdbApi","SenseError","DUCKDB_PACKAGE","message","create","storage_compatibility_version","closeSync","registerFunctions","createConnection","console","error","disconnectSync","clearCache","reconcile","resolvedCfg","store","createStore","rekeyChunkText"],"mappings":";;;;;;;;;;;QAkBaA;eAAAA;;QAGAC;eAAAA;;QA+FSC;eAAAA;;;sBApHI;wBACL;uBAGuB;yBACb;uBACJ;wBACA;wBACc;uBACd;wBACM;4BAEA;wBACS;2BAChB;8BACQ;uBACN;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAErB,IAAMF,cAAc;AAGpB,IAAMC,iBAAiB;AAU9B,8LAA8L;AAC9L,2JAA2J;AAC3J,SAAeE,aAAaC,IAAgB,EAAEC,GAAW;;YAKlD,2BAAA,mBAAA,gBAAA,WAAA,OAAMC;;;;oBAJX;;wBAAMF,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACK,kCAAA,2BAAA;;;;;;;;;oBAAA,YAAiBC,IAAAA,wBAAc,EAACH;;;2BAAhC,6BAAA,QAAA;;;;oBAAMC,UAAN;yBAGCA,CAAAA,QAAQG,IAAI,KAAK,OAAM,GAAvBH;;;;oBACF;;wBAAMF,KAAKG,IAAI,CAAC,AAAC,qIAA+I,OAAXG,mBAAU,EAAC;;;oBAAhK;oBACA;;;;;oBAEF;;wBAAMJ,QAAQK,MAAM,CAACP;;;oBAArB;;;oBAPG;;;;;;;;;;;;oBAAA;oBAAA;;;;;;;6BAAA,6BAAA;4BAAA;;;4BAAA;kCAAA;;;;;;;oBASA;;wBAAMQ,IAAAA,iBAAO,EAACR,MAAM;;;yBAArB,CAAA,AAAC,kBAA2C,IAAG,GAA/C;;;;oBAAkD;;wBAAMS,IAAAA,iBAAO,EAACT,MAAM,kBAAkBH;;;oBAAtC;;;oBACjD;;wBAAMW,IAAAA,iBAAO,EAACR,MAAM;;;yBAArB,CAAA,AAAC,kBAAqC,IAAG,GAAzC;;;;oBAA4C;;wBAAMS,IAAAA,iBAAO,EAACT,MAAM,YAAYU,IAAAA,yBAAgB,EAACT,KAAKU,kBAAQ;;;oBAA9D;;;;;;;;IAClD;;AAEA,SAAeC,QAAQX,GAAmB;;YAClCY,UAEAC,QAIFC,gBAGKC,KAKLC,QACAC,UAMKF,MAOLG,QAGInB,MAIAoB,SACAC,UACAC,cAIEC,QAYqB,MAArBC,QAAQC,UAETT;;;;oBAvDHH,WAAWa,IAAAA,cAAI,EAACzB,IAAI0B,OAAO,EAAEC,kBAAS;oBAC5CC,IAAAA,iBAAS,EAAChB,UAAU;wBAAEiB,WAAW;oBAAK;oBAChChB,SAASY,IAAAA,cAAI,EAACb,UAAUjB;;;;;;;;;oBAMN;;wBAAMmC,IAAAA,mBAAS;;;oBAAlChB,iBAAmB,cAAnBA;;;;;;oBACIC;oBACP,IAAIA,AAAG,YAAHA,KAAegB,oBAAU,GAAE,MAAMhB;oBACrC,MAAM,IAAIgB,oBAAU,CAAC,4BAA4B,AAAC,4BAAsD,OAA3BC,wBAAc,EAAC,cAAmC,OAAvB,AAACjB,IAAckB,OAAO,EAAC;;;;;;;;oBAQpH;;wBAAMnB,eAAeoB,MAAM,CAACrB,QAAQ;4BAAEsB,+BAA+B;wBAAS;;;oBAFzF,4HAA4H;oBAC5H,4KAA4K;oBAC5KlB,WAAW;oBACF;;wBAAMA,SAASN,OAAO;;;oBAA/BK,SAAS;;;;;;oBACFD;oBACP,wFAAwF;oBACxFE,qBAAAA,+BAAAA,SAAUmB,SAAS;oBACnB,MAAM,IAAIL,oBAAU,CAAC,4BAA4B,AAAC,iCAA2C,OAAXlB,QAAO,MAA2B,OAAvB,AAACE,KAAckB,OAAO;;oBAErH,2FAA2F;oBAC3F,uFAAuF;oBACnFf,SAAS;;;;;;;;;oBAEX;;wBAAMmB,IAAAA,iCAAiB,EAACrB;;;oBAAxB;oBACMjB,OAAOuC,IAAAA,8BAAgB,EAACtB;oBAE9B;;wBAAMjB,KAAKG,IAAI,CAAC;;;oBAAhB;oBAEgB;;wBAAMK,IAAAA,iBAAO,EAACR,MAAM;;;oBAA9BoB,UAAU;oBACC;;wBAAMZ,IAAAA,iBAAO,EAACR,MAAM;;;oBAA/BqB,WAAW;oBACXC,eAAeZ,IAAAA,yBAAgB,EAACT,KAAKU,kBAAQ;oBACnD,IAAI,AAACS,YAAY,QAAQA,YAAYvB,kBAAoBwB,aAAa,QAAQA,aAAaC,cAAe;wBACxG,4FAA4F;wBAC5F,wFAAwF;wBAClFC,SAASH,YAAY,QAAQA,YAAYvB,iBAAiB,mDAAmD;wBACnH2C,QAAQC,KAAK,CAAC,AAAC,UAAgB,OAAPlB,QAAO;wBAC/B,gFAAgF;wBAChFJ,SAAS;wBACTF,OAAOyB,cAAc;wBACrBxB,SAASmB,SAAS;wBAClBM,IAAAA,mBAAU,EAAC1C;wBACX;;4BAAOW,QAAQX;;oBACjB;oBAEA;;wBAAMF,aAAaC,MAAMC;;;oBAAzB;oBAE6B;;wBAAM2C,IAAAA,sBAAS,EAAC5C,MAAMC,KAAKA,IAAI0B,OAAO;;;oBAAtC,OAAA,eAArBH,SAAqB,KAArBA,QAAQC,WAAa,KAAbA;oBAChB;;wBAAO;4BAAEP,UAAAA;4BAAUD,QAAAA;4BAAQjB,MAAAA;4BAAMC,KAAAA;4BAAKa,QAAAA;4BAAQU,QAAAA;4BAAQC,UAAAA;wBAAS;;;oBACxDT;oBACP,IAAI,CAACG,QAAQ;wBACXF,OAAOyB,cAAc;wBACrBxB,qBAAAA,+BAAAA,SAAUmB,SAAS;oBACrB;oBACA,MAAMrB;;;;;;;IAEV;;AAEO,SAAelB,WAAWG,GAAmB;;YAC6B,MAAvEiB,UAAUD,QAAQjB,MAAW6C,aAAa/B,QAAQU,QAAQC,UAC5DqB;;;;oBADyE;;wBAAMlC,QAAQX;;;oBAAd,OAAA,eAAvEiB,WAAuE,KAAvEA,UAAUD,SAA6D,KAA7DA,QAAQjB,OAAqD,KAArDA,MAAW6C,cAA0C,KAA/C5C,KAAkBa,SAA6B,KAA7BA,QAAQU,SAAqB,KAArBA,QAAQC,WAAa,KAAbA;oBAC5DqB,QAAQC,IAAAA,oBAAW,EAAC7B,UAAUD,QAAQjB,MAAM6C,aAAaA,YAAYlB,OAAO;oBAClF,0FAA0F;oBAC1FqB,IAAAA,yBAAc,EAAChD,MAAM8C;oBACrB;;wBAAO;4BAAEA,OAAOA;4BAAO7C,KAAK4C;4BAAa/B,QAAAA;4BAAQU,QAAAA;4BAAQC,UAAAA;wBAAS;;;;IACpE"}
|
|
1
|
+
{"version":3,"sources":["/Users/kevin/Dev/OpenSource/ai/sensemaking/src/store/duckdb/open.ts"],"sourcesContent":["import { mkdirSync } from 'node:fs';\nimport { join } from 'node:path';\nimport type { DuckDBConnection, DuckDBInstance } from '@duckdb/node-api';\nimport type { Config, ResolvedConfig } from '../../config/index.ts';\nimport { featureSignature, STATE_DIR } from '../../config/index.ts';\nimport { rekeyChunkText } from '../../embed/handoff.ts';\nimport { STORE_DIMS } from '../../embed/types.ts';\nimport { SenseError } from '../../errors.ts';\nimport { activeFeatures, FEATURES } from '../../features/index.ts';\nimport { clearCache } from '../cache.ts';\nimport { reconcile } from '../reconcile.ts';\nimport { getMeta, setMeta } from '../shared.ts';\nimport type { Connection, Store } from '../types.ts';\nimport { createConnection } from './connection.ts';\nimport { DUCKDB_PACKAGE, duckdbApi } from './native.ts';\nimport { duckdbDialect } from './reconcile.ts';\nimport { registerFunctions } from './sql-functions.ts';\nimport { createStore } from './store.ts';\n\nexport const DB_FILENAME = 'cache.duckdb';\n// Independent of sqlite's SCHEMA_VERSION -- the two stores' cache shapes evolve separately (VARIANT frontmatter columns vs untyped).\n// The store name already joins the feature signature, so switching a config's `store` key rebuilds rather than reusing the other engine's cache.\nexport const SCHEMA_VERSION = '2';\n\nexport interface OpenResult {\n store: Store;\n cfg: ResolvedConfig;\n dbPath: string;\n parsed: number;\n warnings: string[];\n}\n\n// `content` is a plain table (not FTS-virtual): the fts index is built over it lazily, only when a lexical query runs (lexical.ts), and read directly for contains() verification either way.\n// No tokenizer resolution: this store always uses the fts extension's default (porter) stemmer; sqlite's configurable `content.tokenize` is not read here.\nasync function ensureSchema(conn: Connection, cfg: Config): Promise<void> {\n await conn.exec(`CREATE TABLE IF NOT EXISTS frontmatter (\"path\" TEXT PRIMARY KEY, \"_mtime\" DOUBLE, \"_ctime\" DOUBLE, \"_size\" INTEGER, \"_parse_error\" TEXT)`);\n await conn.exec(`CREATE TABLE IF NOT EXISTS content (\"path\" TEXT PRIMARY KEY, title TEXT, summary TEXT, text TEXT)`);\n await conn.exec(`CREATE TABLE IF NOT EXISTS preset_files (\"path\" TEXT, preset TEXT, PRIMARY KEY (\"path\", preset))`);\n await conn.exec('CREATE INDEX IF NOT EXISTS preset_files_preset ON preset_files(preset)');\n for (const feature of activeFeatures(cfg)) {\n // Native FLOAT[STORE_DIMS] instead of the embed feature's engine-neutral BLOB+scale DDL (vectors.ts). `scale` is kept, unused,\n // so the feature's shared reconcile-time INSERT/DELETE (features/embed.ts) names a column that exists on both stores.\n if (feature.name === 'embed') {\n await conn.exec(`CREATE TABLE IF NOT EXISTS embeddings (\"path\" TEXT, chunk INTEGER, start_line INTEGER, end_line INTEGER, scale REAL, vector FLOAT[${STORE_DIMS}], PRIMARY KEY (\"path\", chunk))`);\n continue;\n }\n await feature.schema(conn);\n }\n if ((await getMeta(conn, 'schema_version')) === null) await setMeta(conn, 'schema_version', SCHEMA_VERSION);\n if ((await getMeta(conn, 'features')) === null) await setMeta(conn, 'features', featureSignature(cfg, FEATURES));\n}\n\nasync function connect(cfg: ResolvedConfig): Promise<{ instance: DuckDBInstance; duckdb: DuckDBConnection; conn: Connection; cfg: ResolvedConfig; dbPath: string; parsed: number; warnings: string[] }> {\n const stateDir = join(cfg.baseDir, STATE_DIR);\n mkdirSync(stateDir, { recursive: true });\n const dbPath = join(stateDir, DB_FILENAME);\n\n // Dynamic, not a top-level import: sqlite trees must never attempt to resolve this optional peer dependency, so nothing imports\n // it as a value until a duckdb tree opens (types-only imports are erased). Installed on first use if missing, shared with sql-functions.ts and vectors.ts via native.ts's duckdbApi.\n let DuckDBInstance: typeof import('@duckdb/node-api').DuckDBInstance;\n try {\n ({ DuckDBInstance } = await duckdbApi());\n } catch (err) {\n if (err instanceof SenseError) throw err;\n throw new SenseError('STORE_DEPENDENCY_MISSING', `store \"duckdb\" needs the ${DUCKDB_PACKAGE} package (${(err as Error).message})`);\n }\n\n let duckdb: DuckDBConnection;\n let instance: DuckDBInstance | undefined;\n try {\n // On-disk files default to an older storage format for cross-version compatibility, which rejects VARIANT columns (\"VARIANT\n // columns are not supported in storage versions prior to v1.5.0\"); this store's dynamic frontmatter columns need VARIANT (reconcile.ts), so the floor is pinned explicitly.\n instance = await DuckDBInstance.create(dbPath, { storage_compatibility_version: 'v1.5.0' });\n duckdb = await instance.connect();\n } catch (err) {\n // create() may have succeeded before connect() failed: close it, or its WAL stays open.\n instance?.closeSync();\n throw new SenseError('STORE_DEPENDENCY_MISSING', `store \"duckdb\" failed to open ${dbPath}: ${(err as Error).message}`);\n }\n // A throw below (schema, reconcile's COLUMN_LIMIT) would leak the open instance, whose WAL\n // then locks the .duckdb file undeletable on Windows. closed marks the rebuild branch.\n let closed = false;\n try {\n await registerFunctions(duckdb);\n const conn = createConnection(duckdb);\n\n await conn.exec('CREATE TABLE IF NOT EXISTS meta (key TEXT PRIMARY KEY, value TEXT)');\n\n const version = await getMeta(conn, 'schema_version');\n const features = await getMeta(conn, 'features');\n const wantFeatures = featureSignature(cfg, FEATURES);\n if ((version !== null && version !== SCHEMA_VERSION) || (features !== null && features !== wantFeatures)) {\n // Reconcile only reparses changed files, so a stale cache can't be patched incrementally --\n // rebuild instead, same as sqlite's open() (see its comment for why this is announced).\n const reason = version !== null && version !== SCHEMA_VERSION ? 'cache format changed (new sensemaking version)' : 'config change (features, tokenizer, or presets)';\n console.error(`sense: ${reason}; rebuilding the index`);\n // Close before clearCache deletes the files underneath the still-open instance.\n closed = true;\n duckdb.disconnectSync();\n instance.closeSync();\n clearCache(cfg);\n return connect(cfg);\n }\n\n await ensureSchema(conn, cfg);\n\n const { parsed, warnings } = await reconcile(conn, cfg, cfg.baseDir, duckdbDialect);\n return { instance, duckdb, conn, cfg, dbPath, parsed, warnings };\n } catch (err) {\n if (!closed) {\n duckdb.disconnectSync();\n instance?.closeSync();\n }\n throw err;\n }\n}\n\nexport async function openDuckdb(cfg: ResolvedConfig): Promise<OpenResult> {\n const { instance, duckdb, conn, cfg: resolvedCfg, dbPath, parsed, warnings } = await connect(cfg);\n const store = createStore(instance, duckdb, conn, resolvedCfg, resolvedCfg.baseDir);\n // reconcile ran before this object existed, so its chunk text is keyed by the connection.\n rekeyChunkText(conn, store);\n return { store: store, cfg: resolvedCfg, dbPath, parsed, warnings };\n}\n"],"names":["DB_FILENAME","SCHEMA_VERSION","openDuckdb","ensureSchema","conn","cfg","feature","exec","activeFeatures","name","STORE_DIMS","schema","getMeta","setMeta","featureSignature","FEATURES","connect","stateDir","dbPath","DuckDBInstance","err","duckdb","instance","closed","version","features","wantFeatures","reason","parsed","warnings","join","baseDir","STATE_DIR","mkdirSync","recursive","duckdbApi","SenseError","DUCKDB_PACKAGE","message","create","storage_compatibility_version","closeSync","registerFunctions","createConnection","console","error","disconnectSync","clearCache","reconcile","duckdbDialect","resolvedCfg","store","createStore","rekeyChunkText"],"mappings":";;;;;;;;;;;QAmBaA;eAAAA;;QAGAC;eAAAA;;QA+FSC;eAAAA;;;sBArHI;wBACL;uBAGuB;yBACb;uBACJ;wBACA;wBACc;uBACd;2BACD;wBACO;4BAEA;wBACS;4BACZ;8BACI;uBACN;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAErB,IAAMF,cAAc;AAGpB,IAAMC,iBAAiB;AAU9B,8LAA8L;AAC9L,2JAA2J;AAC3J,SAAeE,aAAaC,IAAgB,EAAEC,GAAW;;YAKlD,2BAAA,mBAAA,gBAAA,WAAA,OAAMC;;;;oBAJX;;wBAAMF,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACA;;wBAAMH,KAAKG,IAAI,CAAC;;;oBAAhB;oBACK,kCAAA,2BAAA;;;;;;;;;oBAAA,YAAiBC,IAAAA,wBAAc,EAACH;;;2BAAhC,6BAAA,QAAA;;;;oBAAMC,UAAN;yBAGCA,CAAAA,QAAQG,IAAI,KAAK,OAAM,GAAvBH;;;;oBACF;;wBAAMF,KAAKG,IAAI,CAAC,AAAC,qIAA+I,OAAXG,mBAAU,EAAC;;;oBAAhK;oBACA;;;;;oBAEF;;wBAAMJ,QAAQK,MAAM,CAACP;;;oBAArB;;;oBAPG;;;;;;;;;;;;oBAAA;oBAAA;;;;;;;6BAAA,6BAAA;4BAAA;;;4BAAA;kCAAA;;;;;;;oBASA;;wBAAMQ,IAAAA,iBAAO,EAACR,MAAM;;;yBAArB,CAAA,AAAC,kBAA2C,IAAG,GAA/C;;;;oBAAkD;;wBAAMS,IAAAA,iBAAO,EAACT,MAAM,kBAAkBH;;;oBAAtC;;;oBACjD;;wBAAMW,IAAAA,iBAAO,EAACR,MAAM;;;yBAArB,CAAA,AAAC,kBAAqC,IAAG,GAAzC;;;;oBAA4C;;wBAAMS,IAAAA,iBAAO,EAACT,MAAM,YAAYU,IAAAA,yBAAgB,EAACT,KAAKU,kBAAQ;;;oBAA9D;;;;;;;;IAClD;;AAEA,SAAeC,QAAQX,GAAmB;;YAClCY,UAEAC,QAIFC,gBAGKC,KAKLC,QACAC,UAMKF,MAOLG,QAGInB,MAIAoB,SACAC,UACAC,cAIEC,QAYqB,MAArBC,QAAQC,UAETT;;;;oBAvDHH,WAAWa,IAAAA,cAAI,EAACzB,IAAI0B,OAAO,EAAEC,kBAAS;oBAC5CC,IAAAA,iBAAS,EAAChB,UAAU;wBAAEiB,WAAW;oBAAK;oBAChChB,SAASY,IAAAA,cAAI,EAACb,UAAUjB;;;;;;;;;oBAMN;;wBAAMmC,IAAAA,mBAAS;;;oBAAlChB,iBAAmB,cAAnBA;;;;;;oBACIC;oBACP,IAAIA,AAAG,YAAHA,KAAegB,oBAAU,GAAE,MAAMhB;oBACrC,MAAM,IAAIgB,oBAAU,CAAC,4BAA4B,AAAC,4BAAsD,OAA3BC,wBAAc,EAAC,cAAmC,OAAvB,AAACjB,IAAckB,OAAO,EAAC;;;;;;;;oBAQpH;;wBAAMnB,eAAeoB,MAAM,CAACrB,QAAQ;4BAAEsB,+BAA+B;wBAAS;;;oBAFzF,4HAA4H;oBAC5H,4KAA4K;oBAC5KlB,WAAW;oBACF;;wBAAMA,SAASN,OAAO;;;oBAA/BK,SAAS;;;;;;oBACFD;oBACP,wFAAwF;oBACxFE,qBAAAA,+BAAAA,SAAUmB,SAAS;oBACnB,MAAM,IAAIL,oBAAU,CAAC,4BAA4B,AAAC,iCAA2C,OAAXlB,QAAO,MAA2B,OAAvB,AAACE,KAAckB,OAAO;;oBAErH,2FAA2F;oBAC3F,uFAAuF;oBACnFf,SAAS;;;;;;;;;oBAEX;;wBAAMmB,IAAAA,iCAAiB,EAACrB;;;oBAAxB;oBACMjB,OAAOuC,IAAAA,8BAAgB,EAACtB;oBAE9B;;wBAAMjB,KAAKG,IAAI,CAAC;;;oBAAhB;oBAEgB;;wBAAMK,IAAAA,iBAAO,EAACR,MAAM;;;oBAA9BoB,UAAU;oBACC;;wBAAMZ,IAAAA,iBAAO,EAACR,MAAM;;;oBAA/BqB,WAAW;oBACXC,eAAeZ,IAAAA,yBAAgB,EAACT,KAAKU,kBAAQ;oBACnD,IAAI,AAACS,YAAY,QAAQA,YAAYvB,kBAAoBwB,aAAa,QAAQA,aAAaC,cAAe;wBACxG,4FAA4F;wBAC5F,wFAAwF;wBAClFC,SAASH,YAAY,QAAQA,YAAYvB,iBAAiB,mDAAmD;wBACnH2C,QAAQC,KAAK,CAAC,AAAC,UAAgB,OAAPlB,QAAO;wBAC/B,gFAAgF;wBAChFJ,SAAS;wBACTF,OAAOyB,cAAc;wBACrBxB,SAASmB,SAAS;wBAClBM,IAAAA,mBAAU,EAAC1C;wBACX;;4BAAOW,QAAQX;;oBACjB;oBAEA;;wBAAMF,aAAaC,MAAMC;;;oBAAzB;oBAE6B;;wBAAM2C,IAAAA,sBAAS,EAAC5C,MAAMC,KAAKA,IAAI0B,OAAO,EAAEkB,2BAAa;;;oBAArD,OAAA,eAArBrB,SAAqB,KAArBA,QAAQC,WAAa,KAAbA;oBAChB;;wBAAO;4BAAEP,UAAAA;4BAAUD,QAAAA;4BAAQjB,MAAAA;4BAAMC,KAAAA;4BAAKa,QAAAA;4BAAQU,QAAAA;4BAAQC,UAAAA;wBAAS;;;oBACxDT;oBACP,IAAI,CAACG,QAAQ;wBACXF,OAAOyB,cAAc;wBACrBxB,qBAAAA,+BAAAA,SAAUmB,SAAS;oBACrB;oBACA,MAAMrB;;;;;;;IAEV;;AAEO,SAAelB,WAAWG,GAAmB;;YAC6B,MAAvEiB,UAAUD,QAAQjB,MAAW8C,aAAahC,QAAQU,QAAQC,UAC5DsB;;;;oBADyE;;wBAAMnC,QAAQX;;;oBAAd,OAAA,eAAvEiB,WAAuE,KAAvEA,UAAUD,SAA6D,KAA7DA,QAAQjB,OAAqD,KAArDA,MAAW8C,cAA0C,KAA/C7C,KAAkBa,SAA6B,KAA7BA,QAAQU,SAAqB,KAArBA,QAAQC,WAAa,KAAbA;oBAC5DsB,QAAQC,IAAAA,oBAAW,EAAC9B,UAAUD,QAAQjB,MAAM8C,aAAaA,YAAYnB,OAAO;oBAClF,0FAA0F;oBAC1FsB,IAAAA,yBAAc,EAACjD,MAAM+C;oBACrB;;wBAAO;4BAAEA,OAAOA;4BAAO9C,KAAK6C;4BAAahC,QAAAA;4BAAQU,QAAAA;4BAAQC,UAAAA;wBAAS;;;;IACpE"}
|
|
@@ -1,6 +1,2 @@
|
|
|
1
|
-
import type {
|
|
2
|
-
|
|
3
|
-
export declare function reconcile(conn: Connection, cfg: Config, baseDir: string): Promise<{
|
|
4
|
-
parsed: number;
|
|
5
|
-
warnings: string[];
|
|
6
|
-
}>;
|
|
1
|
+
import type { ReconcileDialect } from '../types.js';
|
|
2
|
+
export declare const duckdbDialect: ReconcileDialect;
|
|
@@ -1,6 +1,2 @@
|
|
|
1
|
-
import type {
|
|
2
|
-
|
|
3
|
-
export declare function reconcile(conn: Connection, cfg: Config, baseDir: string): Promise<{
|
|
4
|
-
parsed: number;
|
|
5
|
-
warnings: string[];
|
|
6
|
-
}>;
|
|
1
|
+
import type { ReconcileDialect } from '../types.js';
|
|
2
|
+
export declare const duckdbDialect: ReconcileDialect;
|