@gobing-ai/ts-llm-jsonl-importer 0.4.23 → 0.4.25
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/importer.d.ts.map +1 -1
- package/dist/importer.js +30 -4
- package/dist/index.d.ts +2 -1
- package/dist/index.d.ts.map +1 -1
- package/dist/index.js +1 -0
- package/dist/jsonl-importer-dao.d.ts +54 -2
- package/dist/jsonl-importer-dao.d.ts.map +1 -1
- package/dist/jsonl-importer-dao.js +153 -1
- package/dist/mappers.d.ts +7 -1
- package/dist/mappers.d.ts.map +1 -1
- package/dist/mappers.js +150 -6
- package/dist/opencode-importer.d.ts +17 -0
- package/dist/opencode-importer.d.ts.map +1 -0
- package/dist/opencode-importer.js +272 -0
- package/dist/sources.d.ts.map +1 -1
- package/dist/sources.js +2 -2
- package/dist/types.d.ts +20 -1
- package/dist/types.d.ts.map +1 -1
- package/package.json +4 -4
- package/src/importer.ts +50 -4
- package/src/index.ts +2 -0
- package/src/jsonl-importer-dao.ts +229 -1
- package/src/mappers.ts +166 -6
- package/src/opencode-importer.ts +339 -0
- package/src/sources.ts +13 -2
- package/src/types.ts +21 -1
package/dist/importer.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"importer.d.ts","sourceRoot":"","sources":["../src/importer.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"importer.d.ts","sourceRoot":"","sources":["../src/importer.ts"],"names":[],"mappings":"AAuBA,OAAO,KAAK,EAER,aAAa,EACb,YAAY,EAGZ,gBAAgB,EAEnB,MAAM,SAAS,CAAC;AAOjB;;;;;;;GAOG;AACH,wBAAsB,cAAc,CAAC,MAAM,EAAE,MAAM,GAAG,gBAAgB,EAAE,OAAO,EAAE,aAAa,GAAG,OAAO,CAAC,YAAY,CAAC,CA4KrH"}
|
package/dist/importer.js
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
import { ambientRuntimePaths, createNodeFileSystem, resolvePath, walkDir, } from '@gobing-ai/ts-runtime';
|
|
2
2
|
import { sha256 } from './hash.js';
|
|
3
|
-
import { applyHistoryImportSchema, ensureTargetTables, insertLedger, insertRecord, ledgerExists, readCheckpoint, resetCheckpoints, targetTableFor, writeCheckpoint, } from './jsonl-importer-dao.js';
|
|
3
|
+
import { applyHistoryImportSchema, ensureTargetTables, insertLedger, insertRecord, ledgerExists, readCheckpoint, reconcileFullImport, resetCheckpoints, targetTableFor, writeCheckpoint, } from './jsonl-importer-dao.js';
|
|
4
4
|
import { redactRecord } from './redaction.js';
|
|
5
5
|
import { resolveSourceDefinition } from './sources.js';
|
|
6
6
|
/** Run the JSONL import pipeline for one source.
|
|
@@ -29,6 +29,9 @@ export async function runJsonlImport(source, options) {
|
|
|
29
29
|
let skippedDuplicates = 0;
|
|
30
30
|
let unknownRecords = 0;
|
|
31
31
|
let checkpointUpdates = 0;
|
|
32
|
+
// Full-mode desired set (R1, task 0504): every record hash the current source produces.
|
|
33
|
+
// ReconcileFullImport diffs this against the persisted ledger to retire stale rows.
|
|
34
|
+
const desiredHashes = new Set();
|
|
32
35
|
for (const file of files) {
|
|
33
36
|
const checkpoint = mode === 'incremental' ? await readCheckpoint(options.db, resolvedSource, file) : 0;
|
|
34
37
|
let lineNumber = 0;
|
|
@@ -41,8 +44,17 @@ export async function runJsonlImport(source, options) {
|
|
|
41
44
|
const raw = parseJsonLine(line, file, lineNumber, parseErrors);
|
|
42
45
|
if (raw === undefined)
|
|
43
46
|
continue;
|
|
44
|
-
const splitRecords = splitRawRecord(definition, raw
|
|
47
|
+
const splitRecords = splitRawRecord(definition, raw, {
|
|
48
|
+
source: resolvedSource,
|
|
49
|
+
sourceFile: file,
|
|
50
|
+
sourceLine: lineNumber,
|
|
51
|
+
splitIndex: 0,
|
|
52
|
+
});
|
|
45
53
|
let lineSucceeded = false;
|
|
54
|
+
// Atomic per-record acceptance (R2, task 0504): a schema-invalid split rejects the
|
|
55
|
+
// WHOLE line, so no partially accepted rows (or orphaned tool-call rows with a
|
|
56
|
+
// dangling message_hash) can be left behind by one bad split.
|
|
57
|
+
let lineRejected = false;
|
|
46
58
|
const prepared = [];
|
|
47
59
|
for (let splitIndex = 0; splitIndex < splitRecords.length; splitIndex += 1) {
|
|
48
60
|
const split = splitRecords[splitIndex];
|
|
@@ -61,6 +73,7 @@ export async function runJsonlImport(source, options) {
|
|
|
61
73
|
sourceLine: lineNumber,
|
|
62
74
|
reason: parsed.error.issues.map((issue) => issue.message).join('; '),
|
|
63
75
|
});
|
|
76
|
+
lineRejected = true;
|
|
64
77
|
continue;
|
|
65
78
|
}
|
|
66
79
|
const redacted = redactRecord(parsed.data, options.redactionRules);
|
|
@@ -72,12 +85,17 @@ export async function runJsonlImport(source, options) {
|
|
|
72
85
|
splitIndex,
|
|
73
86
|
record: redacted,
|
|
74
87
|
});
|
|
88
|
+
desiredHashes.add(recordHash);
|
|
75
89
|
if (await ledgerExists(options.db, recordHash)) {
|
|
76
90
|
skippedDuplicates += 1;
|
|
77
91
|
continue;
|
|
78
92
|
}
|
|
79
93
|
prepared.push({ split, splitIndex, normalized: redacted, recordHash });
|
|
80
94
|
}
|
|
95
|
+
// A rejected line is reported (validationErrors above) but never persisted — no
|
|
96
|
+
// second pass, no checkpoint advance, no partial rows from this record.
|
|
97
|
+
if (lineRejected)
|
|
98
|
+
continue;
|
|
81
99
|
// Second pass: resolve _messageSplitIndex → message_hash, then insert.
|
|
82
100
|
const recordHashBySplitIndex = new Map(prepared.map((e) => [e.splitIndex, e.recordHash]));
|
|
83
101
|
for (const entry of prepared) {
|
|
@@ -106,6 +124,13 @@ export async function runJsonlImport(source, options) {
|
|
|
106
124
|
}
|
|
107
125
|
}
|
|
108
126
|
}
|
|
127
|
+
// Full-mode reconciliation (R1, task 0504): diff the desired hash set against the
|
|
128
|
+
// persisted ledger and retire stale derived rows (target, tool, ledger, checkpoint) in
|
|
129
|
+
// one source-scoped batch. Dry-run computes the identical counts without mutation.
|
|
130
|
+
let reconciliation;
|
|
131
|
+
if (mode === 'full') {
|
|
132
|
+
reconciliation = await reconcileFullImport(options.db, resolvedSource, desiredHashes, files, options.dryRun ?? false);
|
|
133
|
+
}
|
|
109
134
|
return {
|
|
110
135
|
source: resolvedSource,
|
|
111
136
|
mode,
|
|
@@ -117,6 +142,7 @@ export async function runJsonlImport(source, options) {
|
|
|
117
142
|
parseErrors,
|
|
118
143
|
validationErrors,
|
|
119
144
|
checkpointUpdates,
|
|
145
|
+
reconciliation,
|
|
120
146
|
};
|
|
121
147
|
}
|
|
122
148
|
function parseJsonLine(line, sourceFile, sourceLine, parseErrors) {
|
|
@@ -133,7 +159,7 @@ function parseJsonLine(line, sourceFile, sourceLine, parseErrors) {
|
|
|
133
159
|
return undefined;
|
|
134
160
|
}
|
|
135
161
|
}
|
|
136
|
-
function splitRawRecord(definition, raw) {
|
|
162
|
+
function splitRawRecord(definition, raw, context) {
|
|
137
163
|
const targetTable = targetTableFor(definition.targetTable);
|
|
138
164
|
if (definition.splitConfig.mode === 'one-to-one') {
|
|
139
165
|
return [{ targetTable, raw }];
|
|
@@ -141,7 +167,7 @@ function splitRawRecord(definition, raw) {
|
|
|
141
167
|
if (definition.splitConfig.mode === 'custom') {
|
|
142
168
|
const config = definition.splitConfig;
|
|
143
169
|
const configTable = config.targetTable !== undefined ? targetTableFor(config.targetTable) : undefined;
|
|
144
|
-
return config.split(raw).map((entry) => {
|
|
170
|
+
return config.split(raw, context).map((entry) => {
|
|
145
171
|
// Normalize: SplitEntry has a `record` property, bare object is the record itself.
|
|
146
172
|
const hasTargetTable = 'targetTable' in entry && typeof entry.targetTable === 'string';
|
|
147
173
|
const record = hasTargetTable ? entry.record : entry;
|
package/dist/index.d.ts
CHANGED
|
@@ -2,8 +2,9 @@ export { HistoryImportError } from './errors';
|
|
|
2
2
|
export { sha256, stableJson } from './hash';
|
|
3
3
|
export { runJsonlImport } from './importer';
|
|
4
4
|
export { applyHistoryImportSchema, normalizeSourceFilePaths } from './jsonl-importer-dao';
|
|
5
|
+
export { type OpenCodeImportOptions, runOpenCodeImport } from './opencode-importer';
|
|
5
6
|
export { DEFAULT_REDACTION_RULES, redactRecord, redactValue } from './redaction';
|
|
6
7
|
export { HISTORY_IMPORT_SCHEMA_SQL } from './schema-sql';
|
|
7
8
|
export { getSourceDefinition, resolveSourceDefinition, SOURCE_DEFINITIONS, VALID_TABLE_NAME, validateSourceDefinition, } from './sources';
|
|
8
|
-
export type { FieldTransform, ImportIssue, ImportMode, ImportOptions, ImportResult, JsonObject, LlmJsonlSource, RedactionRule, SourceDefinition, SplitConfig, SplitEntry, TransformContext, } from './types';
|
|
9
|
+
export type { FieldTransform, ImportIssue, ImportMode, ImportOptions, ImportResult, JsonObject, LlmJsonlSource, ReconcileSummary, RedactionRule, SourceDefinition, SplitConfig, SplitEntry, TransformContext, } from './types';
|
|
9
10
|
//# sourceMappingURL=index.d.ts.map
|
package/dist/index.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,MAAM,UAAU,CAAC;AAC9C,OAAO,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,QAAQ,CAAC;AAC5C,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAC;AAC5C,OAAO,EAAE,wBAAwB,EAAE,wBAAwB,EAAE,MAAM,sBAAsB,CAAC;AAC1F,OAAO,EAAE,uBAAuB,EAAE,YAAY,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AACjF,OAAO,EAAE,yBAAyB,EAAE,MAAM,cAAc,CAAC;AACzD,OAAO,EACH,mBAAmB,EACnB,uBAAuB,EACvB,kBAAkB,EAClB,gBAAgB,EAChB,wBAAwB,GAC3B,MAAM,WAAW,CAAC;AACnB,YAAY,EACR,cAAc,EACd,WAAW,EACX,UAAU,EACV,aAAa,EACb,YAAY,EACZ,UAAU,EACV,cAAc,EACd,aAAa,EACb,gBAAgB,EAChB,WAAW,EACX,UAAU,EACV,gBAAgB,GACnB,MAAM,SAAS,CAAC"}
|
|
1
|
+
{"version":3,"file":"index.d.ts","sourceRoot":"","sources":["../src/index.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,kBAAkB,EAAE,MAAM,UAAU,CAAC;AAC9C,OAAO,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,QAAQ,CAAC;AAC5C,OAAO,EAAE,cAAc,EAAE,MAAM,YAAY,CAAC;AAC5C,OAAO,EAAE,wBAAwB,EAAE,wBAAwB,EAAE,MAAM,sBAAsB,CAAC;AAC1F,OAAO,EAAE,KAAK,qBAAqB,EAAE,iBAAiB,EAAE,MAAM,qBAAqB,CAAC;AACpF,OAAO,EAAE,uBAAuB,EAAE,YAAY,EAAE,WAAW,EAAE,MAAM,aAAa,CAAC;AACjF,OAAO,EAAE,yBAAyB,EAAE,MAAM,cAAc,CAAC;AACzD,OAAO,EACH,mBAAmB,EACnB,uBAAuB,EACvB,kBAAkB,EAClB,gBAAgB,EAChB,wBAAwB,GAC3B,MAAM,WAAW,CAAC;AACnB,YAAY,EACR,cAAc,EACd,WAAW,EACX,UAAU,EACV,aAAa,EACb,YAAY,EACZ,UAAU,EACV,cAAc,EACd,gBAAgB,EAChB,aAAa,EACb,gBAAgB,EAChB,WAAW,EACX,UAAU,EACV,gBAAgB,GACnB,MAAM,SAAS,CAAC"}
|
package/dist/index.js
CHANGED
|
@@ -2,6 +2,7 @@ export { HistoryImportError } from './errors.js';
|
|
|
2
2
|
export { sha256, stableJson } from './hash.js';
|
|
3
3
|
export { runJsonlImport } from './importer.js';
|
|
4
4
|
export { applyHistoryImportSchema, normalizeSourceFilePaths } from './jsonl-importer-dao.js';
|
|
5
|
+
export { runOpenCodeImport } from './opencode-importer.js';
|
|
5
6
|
export { DEFAULT_REDACTION_RULES, redactRecord, redactValue } from './redaction.js';
|
|
6
7
|
export { HISTORY_IMPORT_SCHEMA_SQL } from './schema-sql.js';
|
|
7
8
|
export { getSourceDefinition, resolveSourceDefinition, SOURCE_DEFINITIONS, VALID_TABLE_NAME, validateSourceDefinition, } from './sources.js';
|
|
@@ -1,4 +1,5 @@
|
|
|
1
|
-
import type {
|
|
1
|
+
import type { DbAdapter, DbBatchOp } from '@gobing-ai/ts-db';
|
|
2
|
+
import type { ImportOptions, JsonObject, ReconcileSummary, SourceDefinition } from './types';
|
|
2
3
|
interface CheckpointRow {
|
|
3
4
|
readonly last_imported_line: number;
|
|
4
5
|
}
|
|
@@ -46,6 +47,57 @@ declare function insertLedger(db: ImportOptions['db'], recordHash: string, sourc
|
|
|
46
47
|
* runtime `FileSystem` seam.
|
|
47
48
|
*/
|
|
48
49
|
export declare function normalizeSourceFilePaths(db: ImportOptions['db'], resolveRealPath: (sourceFile: string) => string | null | undefined): Promise<void>;
|
|
50
|
+
/**
|
|
51
|
+
* Reconcile a source's persisted derived rows against the desired set produced by a full
|
|
52
|
+
* import (task 0504 R1). Rows whose `record_hash` is no longer reproduced by current source
|
|
53
|
+
* data or mapper output are stale: their target rows (typed contract tables — including tool
|
|
54
|
+
* calls — and ETL tables), ledger rows, and checkpoints for vanished source files are removed
|
|
55
|
+
* in ONE source-scoped batch (atomic). Dry-run returns the exact same counts without mutating
|
|
56
|
+
* the database, so `--dry-run` is a safe preview and a second full write reports zero changes.
|
|
57
|
+
*
|
|
58
|
+
* Deletes are keyed by `record_hash` (PK) and checkpoint `(source, source_file)` (PK) — never
|
|
59
|
+
* by an unindexed ledger predicate — so reconciliation cost scales with the stale count, not
|
|
60
|
+
* with ledger size.
|
|
61
|
+
*/
|
|
62
|
+
declare function reconcileFullImport(db: ImportOptions['db'], source: string, desiredHashes: ReadonlySet<string>, discoveredFiles: readonly string[], dryRun: boolean): Promise<ReconcileSummary>;
|
|
63
|
+
/** OpenCode message row projected from its SQLite store. */
|
|
64
|
+
export interface OpenCodeMessageRow {
|
|
65
|
+
id: string;
|
|
66
|
+
session_id: string;
|
|
67
|
+
time_created: number;
|
|
68
|
+
data: string;
|
|
69
|
+
directory: string;
|
|
70
|
+
}
|
|
71
|
+
/** OpenCode message-part row projected from its SQLite store. */
|
|
72
|
+
export interface OpenCodePartRow {
|
|
73
|
+
id: string;
|
|
74
|
+
message_id: string;
|
|
75
|
+
time_created: number;
|
|
76
|
+
data: string;
|
|
77
|
+
}
|
|
78
|
+
/** Previously imported OpenCode ledger entry. */
|
|
79
|
+
export interface OpenCodeExistingEntry {
|
|
80
|
+
record_hash: string;
|
|
81
|
+
target_table: string;
|
|
82
|
+
}
|
|
83
|
+
/** Normalized OpenCode entry awaiting a history-table write. */
|
|
84
|
+
export interface OpenCodeQueuedEntry {
|
|
85
|
+
targetTable: 'history_message' | 'history_tool_call';
|
|
86
|
+
splitIndex: number;
|
|
87
|
+
record: JsonObject;
|
|
88
|
+
recordHash: string;
|
|
89
|
+
sourceFile: string;
|
|
90
|
+
}
|
|
91
|
+
/** Read one ordered page of OpenCode messages after the supplied cursor. */
|
|
92
|
+
export declare function readOpenCodeMessages(db: DbAdapter, lastTime: number, lastId: string, limit: number): Promise<OpenCodeMessageRow[]>;
|
|
93
|
+
/** Read all parts belonging to the supplied OpenCode message IDs. */
|
|
94
|
+
export declare function readOpenCodeParts(db: DbAdapter, messageIds: readonly string[]): Promise<OpenCodePartRow[]>;
|
|
95
|
+
/** Load existing OpenCode ledger entries grouped by source message ID. */
|
|
96
|
+
export declare function readOpenCodeExistingEntries(db: DbAdapter): Promise<Map<string, OpenCodeExistingEntry[]>>;
|
|
97
|
+
/** Build deletion operations for records superseded by a forced import. */
|
|
98
|
+
export declare function openCodeDeleteOperations(entries: readonly OpenCodeExistingEntry[]): DbBatchOp[];
|
|
99
|
+
/** Build batched history, ledger, and checkpoint writes for OpenCode entries. */
|
|
100
|
+
export declare function openCodeBulkWriteOperations(entries: readonly OpenCodeQueuedEntry[], checkpointFiles: readonly string[], now: ImportOptions['now']): DbBatchOp[];
|
|
49
101
|
export type { CheckpointRow };
|
|
50
|
-
export { ETL_TABLE_DDL, ensureTargetTables, insertLedger, insertRecord, ledgerExists, readCheckpoint, resetCheckpoints, targetTableFor, timestamp, writeCheckpoint, };
|
|
102
|
+
export { ETL_TABLE_DDL, ensureTargetTables, insertLedger, insertRecord, ledgerExists, readCheckpoint, reconcileFullImport, resetCheckpoints, targetTableFor, timestamp, writeCheckpoint, };
|
|
51
103
|
//# sourceMappingURL=jsonl-importer-dao.d.ts.map
|
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"jsonl-importer-dao.d.ts","sourceRoot":"","sources":["../src/jsonl-importer-dao.ts"],"names":[],"mappings":"
|
|
1
|
+
{"version":3,"file":"jsonl-importer-dao.d.ts","sourceRoot":"","sources":["../src/jsonl-importer-dao.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,kBAAkB,CAAC;AAI7D,OAAO,KAAK,EAAE,aAAa,EAAE,UAAU,EAAE,gBAAgB,EAAE,gBAAgB,EAAE,MAAM,SAAS,CAAC;AAE7F,UAAU,aAAa;IACnB,QAAQ,CAAC,kBAAkB,EAAE,MAAM,CAAC;CACvC;AAqDD,iBAAS,cAAc,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAK7C;AAED,iBAAS,SAAS,CAAC,GAAG,EAAE,aAAa,CAAC,KAAK,CAAC,GAAG,MAAM,CAEpD;AAED,iBAAS,aAAa,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAS5C;AAED,0DAA0D;AAC1D,wBAAsB,wBAAwB,CAAC,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,CAOrF;AAED;;;;;;;;;;GAUG;AACH,iBAAe,kBAAkB,CAAC,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EAAE,UAAU,EAAE,gBAAgB,GAAG,OAAO,CAAC,IAAI,CAAC,CAUtG;AAED,iBAAe,cAAc,CAAC,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,CAAC,CAO1G;AAED,iBAAe,gBAAgB,CAAC,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,MAAM,EAAE,KAAK,EAAE,SAAS,MAAM,EAAE,GAAG,OAAO,CAAC,IAAI,CAAC,CAIhH;AAED,iBAAe,eAAe,CAC1B,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EACvB,MAAM,EAAE,MAAM,EACd,UAAU,EAAE,MAAM,EAClB,IAAI,EAAE,MAAM,EACZ,GAAG,EAAE,aAAa,CAAC,KAAK,CAAC,GAC1B,OAAO,CAAC,IAAI,CAAC,CAYf;AAED,iBAAe,YAAY,CAAC,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EAAE,UAAU,EAAE,MAAM,GAAG,OAAO,CAAC,OAAO,CAAC,CAMzF;AAID,iBAAe,YAAY,CACvB,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EACvB,WAAW,EAAE,MAAM,EACnB,UAAU,EAAE,MAAM,EAClB,UAAU,EAAE,MAAM,EAClB,UAAU,EAAE,MAAM,EAClB,UAAU,EAAE,MAAM,EAClB,OAAO,EAAE,UAAU,EACnB,GAAG,EAAE,aAAa,CAAC,KAAK,CAAC,GAC1B,OAAO,CAAC,IAAI,CAAC,CAiDf;AAED,iBAAe,YAAY,CACvB,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EACvB,UAAU,EAAE,MAAM,EAClB,MAAM,EAAE,MAAM,EACd,UAAU,EAAE,MAAM,EAClB,UAAU,EAAE,MAAM,EAClB,UAAU,EAAE,MAAM,EAClB,WAAW,EAAE,MAAM,EACnB,GAAG,EAAE,aAAa,CAAC,KAAK,CAAC,GAC1B,OAAO,CAAC,IAAI,CAAC,CAYf;AAED;;;;;;;;;;;;;;;;;;;GAmBG;AACH,wBAAsB,wBAAwB,CAC1C,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EACvB,eAAe,EAAE,CAAC,UAAU,EAAE,MAAM,KAAK,MAAM,GAAG,IAAI,GAAG,SAAS,GACnE,OAAO,CAAC,IAAI,CAAC,CAWf;AA6FD;;;;;;;;;;;GAWG;AACH,iBAAe,mBAAmB,CAC9B,EAAE,EAAE,aAAa,CAAC,IAAI,CAAC,EACvB,MAAM,EAAE,MAAM,EACd,aAAa,EAAE,WAAW,CAAC,MAAM,CAAC,EAClC,eAAe,EAAE,SAAS,MAAM,EAAE,EAClC,MAAM,EAAE,OAAO,GAChB,OAAO,CAAC,gBAAgB,CAAC,CA6C3B;AAID,4DAA4D;AAC5D,MAAM,WAAW,kBAAkB;IAC/B,EAAE,EAAE,MAAM,CAAC;IACX,UAAU,EAAE,MAAM,CAAC;IACnB,YAAY,EAAE,MAAM,CAAC;IACrB,IAAI,EAAE,MAAM,CAAC;IACb,SAAS,EAAE,MAAM,CAAC;CACrB;AAED,iEAAiE;AACjE,MAAM,WAAW,eAAe;IAC5B,EAAE,EAAE,MAAM,CAAC;IACX,UAAU,EAAE,MAAM,CAAC;IACnB,YAAY,EAAE,MAAM,CAAC;IACrB,IAAI,EAAE,MAAM,CAAC;CAChB;AAED,iDAAiD;AACjD,MAAM,WAAW,qBAAqB;IAClC,WAAW,EAAE,MAAM,CAAC;IACpB,YAAY,EAAE,MAAM,CAAC;CACxB;AAED,gEAAgE;AAChE,MAAM,WAAW,mBAAmB;IAChC,WAAW,EAAE,iBAAiB,GAAG,mBAAmB,CAAC;IACrD,UAAU,EAAE,MAAM,CAAC;IACnB,MAAM,EAAE,UAAU,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;IACnB,UAAU,EAAE,MAAM,CAAC;CACtB;AAED,4EAA4E;AAC5E,wBAAsB,oBAAoB,CACtC,EAAE,EAAE,SAAS,EACb,QAAQ,EAAE,MAAM,EAChB,MAAM,EAAE,MAAM,EACd,KAAK,EAAE,MAAM,GACd,OAAO,CAAC,kBAAkB,EAAE,CAAC,CAa/B;AAED,qEAAqE;AACrE,wBAAsB,iBAAiB,CAAC,EAAE,EAAE,SAAS,EAAE,UAAU,EAAE,SAAS,MAAM,EAAE,GAAG,OAAO,CAAC,eAAe,EAAE,CAAC,CAShH;AAED,0EAA0E;AAC1E,wBAAsB,2BAA2B,CAAC,EAAE,EAAE,SAAS,GAAG,OAAO,CAAC,GAAG,CAAC,MAAM,EAAE,qBAAqB,EAAE,CAAC,CAAC,CAY9G;AAED,2EAA2E;AAC3E,wBAAgB,wBAAwB,CAAC,OAAO,EAAE,SAAS,qBAAqB,EAAE,GAAG,SAAS,EAAE,CAW/F;AAED,iFAAiF;AACjF,wBAAgB,2BAA2B,CACvC,OAAO,EAAE,SAAS,mBAAmB,EAAE,EACvC,eAAe,EAAE,SAAS,MAAM,EAAE,EAClC,GAAG,EAAE,aAAa,CAAC,KAAK,CAAC,GAC1B,SAAS,EAAE,CA0Db;AAED,YAAY,EAAE,aAAa,EAAE,CAAC;AAC9B,OAAO,EACH,aAAa,EACb,kBAAkB,EAClB,YAAY,EACZ,YAAY,EACZ,YAAY,EACZ,cAAc,EACd,mBAAmB,EACnB,gBAAgB,EAChB,cAAc,EACd,SAAS,EACT,eAAe,GAClB,CAAC"}
|
|
@@ -252,4 +252,156 @@ function resolveWithFallback(resolveRealPath, sourceFile) {
|
|
|
252
252
|
return sourceFile;
|
|
253
253
|
}
|
|
254
254
|
}
|
|
255
|
-
|
|
255
|
+
/**
|
|
256
|
+
* Reconcile a source's persisted derived rows against the desired set produced by a full
|
|
257
|
+
* import (task 0504 R1). Rows whose `record_hash` is no longer reproduced by current source
|
|
258
|
+
* data or mapper output are stale: their target rows (typed contract tables — including tool
|
|
259
|
+
* calls — and ETL tables), ledger rows, and checkpoints for vanished source files are removed
|
|
260
|
+
* in ONE source-scoped batch (atomic). Dry-run returns the exact same counts without mutating
|
|
261
|
+
* the database, so `--dry-run` is a safe preview and a second full write reports zero changes.
|
|
262
|
+
*
|
|
263
|
+
* Deletes are keyed by `record_hash` (PK) and checkpoint `(source, source_file)` (PK) — never
|
|
264
|
+
* by an unindexed ledger predicate — so reconciliation cost scales with the stale count, not
|
|
265
|
+
* with ledger size.
|
|
266
|
+
*/
|
|
267
|
+
async function reconcileFullImport(db, source, desiredHashes, discoveredFiles, dryRun) {
|
|
268
|
+
const ledgerRows = await db.queryAll('SELECT record_hash, target_table FROM history_import_ledger WHERE source = ?', source);
|
|
269
|
+
const stale = [];
|
|
270
|
+
for (const row of ledgerRows) {
|
|
271
|
+
if (!desiredHashes.has(row.record_hash))
|
|
272
|
+
stale.push(row);
|
|
273
|
+
}
|
|
274
|
+
const checkpointRows = await db.queryAll('SELECT source_file FROM history_import_checkpoint WHERE source = ?', source);
|
|
275
|
+
const discovered = new Set(discoveredFiles);
|
|
276
|
+
const staleCheckpoints = checkpointRows
|
|
277
|
+
.map((row) => row.source_file)
|
|
278
|
+
.filter((sourceFile) => !discovered.has(sourceFile));
|
|
279
|
+
const summary = {
|
|
280
|
+
staleTargetRows: stale.length,
|
|
281
|
+
staleLedgerRows: stale.length,
|
|
282
|
+
staleCheckpointRows: staleCheckpoints.length,
|
|
283
|
+
};
|
|
284
|
+
if (dryRun || (stale.length === 0 && staleCheckpoints.length === 0))
|
|
285
|
+
return summary;
|
|
286
|
+
const operations = [];
|
|
287
|
+
for (const row of stale) {
|
|
288
|
+
operations.push({
|
|
289
|
+
sql: `DELETE FROM ${targetTableFor(row.target_table)} WHERE record_hash = ?`,
|
|
290
|
+
params: [row.record_hash],
|
|
291
|
+
});
|
|
292
|
+
operations.push({
|
|
293
|
+
sql: 'DELETE FROM history_import_ledger WHERE record_hash = ?',
|
|
294
|
+
params: [row.record_hash],
|
|
295
|
+
});
|
|
296
|
+
}
|
|
297
|
+
for (const sourceFile of staleCheckpoints) {
|
|
298
|
+
operations.push({
|
|
299
|
+
sql: 'DELETE FROM history_import_checkpoint WHERE source = ? AND source_file = ?',
|
|
300
|
+
params: [source, sourceFile],
|
|
301
|
+
});
|
|
302
|
+
}
|
|
303
|
+
await db.batch(operations);
|
|
304
|
+
return summary;
|
|
305
|
+
}
|
|
306
|
+
const OPENCODE_SOURCE = 'opencode';
|
|
307
|
+
/** Read one ordered page of OpenCode messages after the supplied cursor. */
|
|
308
|
+
export async function readOpenCodeMessages(db, lastTime, lastId, limit) {
|
|
309
|
+
return db.queryAll(`SELECT m.id, m.session_id, m.time_created, m.data, s.directory
|
|
310
|
+
FROM message m
|
|
311
|
+
JOIN session s ON s.id = m.session_id
|
|
312
|
+
WHERE m.time_created > ? OR (m.time_created = ? AND m.id > ?)
|
|
313
|
+
ORDER BY m.time_created, m.id
|
|
314
|
+
LIMIT ?`, lastTime, lastTime, lastId, limit);
|
|
315
|
+
}
|
|
316
|
+
/** Read all parts belonging to the supplied OpenCode message IDs. */
|
|
317
|
+
export async function readOpenCodeParts(db, messageIds) {
|
|
318
|
+
if (messageIds.length === 0)
|
|
319
|
+
return [];
|
|
320
|
+
return db.queryAll(`SELECT id, message_id, time_created, data
|
|
321
|
+
FROM part
|
|
322
|
+
WHERE message_id IN (${messageIds.map(() => '?').join(', ')})
|
|
323
|
+
ORDER BY time_created, id`, ...messageIds);
|
|
324
|
+
}
|
|
325
|
+
/** Load existing OpenCode ledger entries grouped by source message ID. */
|
|
326
|
+
export async function readOpenCodeExistingEntries(db) {
|
|
327
|
+
const rows = await db.queryAll('SELECT source_file, record_hash, target_table FROM history_import_ledger WHERE source = ?', OPENCODE_SOURCE);
|
|
328
|
+
const bySourceFile = new Map();
|
|
329
|
+
for (const row of rows) {
|
|
330
|
+
const entries = bySourceFile.get(row.source_file) ?? [];
|
|
331
|
+
entries.push({ record_hash: row.record_hash, target_table: row.target_table });
|
|
332
|
+
bySourceFile.set(row.source_file, entries);
|
|
333
|
+
}
|
|
334
|
+
return bySourceFile;
|
|
335
|
+
}
|
|
336
|
+
/** Build deletion operations for records superseded by a forced import. */
|
|
337
|
+
export function openCodeDeleteOperations(entries) {
|
|
338
|
+
return entries.flatMap((entry) => [
|
|
339
|
+
{
|
|
340
|
+
sql: `DELETE FROM ${targetTableFor(entry.target_table)} WHERE record_hash = ?`,
|
|
341
|
+
params: [entry.record_hash],
|
|
342
|
+
},
|
|
343
|
+
{
|
|
344
|
+
sql: 'DELETE FROM history_import_ledger WHERE record_hash = ?',
|
|
345
|
+
params: [entry.record_hash],
|
|
346
|
+
},
|
|
347
|
+
]);
|
|
348
|
+
}
|
|
349
|
+
/** Build batched history, ledger, and checkpoint writes for OpenCode entries. */
|
|
350
|
+
export function openCodeBulkWriteOperations(entries, checkpointFiles, now) {
|
|
351
|
+
const importedAt = timestamp(now);
|
|
352
|
+
const operations = [];
|
|
353
|
+
for (const table of ['history_message', 'history_tool_call']) {
|
|
354
|
+
const tableEntries = entries.filter((entry) => entry.targetTable === table);
|
|
355
|
+
for (let offset = 0; offset < tableEntries.length; offset += 100) {
|
|
356
|
+
const chunk = tableEntries.slice(offset, offset + 100);
|
|
357
|
+
const columns = TYPED_TABLE_COLUMNS[table] ?? [];
|
|
358
|
+
operations.push({
|
|
359
|
+
sql: `WITH input(record_hash, payload_json, imported_at) AS (
|
|
360
|
+
VALUES ${chunk.map(() => '(?, ?, ?)').join(', ')}
|
|
361
|
+
)
|
|
362
|
+
INSERT INTO ${table} (${columns.join(', ')})
|
|
363
|
+
SELECT ${columns
|
|
364
|
+
.map((column) => column === 'record_hash'
|
|
365
|
+
? 'record_hash'
|
|
366
|
+
: column === 'imported_at'
|
|
367
|
+
? 'imported_at'
|
|
368
|
+
: `json_extract(payload_json, '$.${column}')`)
|
|
369
|
+
.join(', ')}
|
|
370
|
+
FROM input
|
|
371
|
+
WHERE true
|
|
372
|
+
ON CONFLICT(record_hash) DO NOTHING`,
|
|
373
|
+
params: chunk.flatMap((entry) => [entry.recordHash, JSON.stringify(entry.record), importedAt]),
|
|
374
|
+
});
|
|
375
|
+
}
|
|
376
|
+
}
|
|
377
|
+
for (let offset = 0; offset < entries.length; offset += 200) {
|
|
378
|
+
const chunk = entries.slice(offset, offset + 200);
|
|
379
|
+
operations.push({
|
|
380
|
+
sql: `INSERT INTO history_import_ledger
|
|
381
|
+
(record_hash, source, source_file, source_line, split_index, target_table, imported_at)
|
|
382
|
+
VALUES ${chunk.map(() => '(?, ?, ?, ?, ?, ?, ?)').join(', ')}`,
|
|
383
|
+
params: chunk.flatMap((entry) => [
|
|
384
|
+
entry.recordHash,
|
|
385
|
+
OPENCODE_SOURCE,
|
|
386
|
+
entry.sourceFile,
|
|
387
|
+
1,
|
|
388
|
+
entry.splitIndex,
|
|
389
|
+
entry.targetTable,
|
|
390
|
+
importedAt,
|
|
391
|
+
]),
|
|
392
|
+
});
|
|
393
|
+
}
|
|
394
|
+
for (let offset = 0; offset < checkpointFiles.length; offset += 200) {
|
|
395
|
+
const chunk = checkpointFiles.slice(offset, offset + 200);
|
|
396
|
+
operations.push({
|
|
397
|
+
sql: `INSERT INTO history_import_checkpoint (source, source_file, last_imported_line, updated_at)
|
|
398
|
+
VALUES ${chunk.map(() => '(?, ?, ?, ?)').join(', ')}
|
|
399
|
+
ON CONFLICT(source, source_file) DO UPDATE SET
|
|
400
|
+
last_imported_line = excluded.last_imported_line,
|
|
401
|
+
updated_at = excluded.updated_at`,
|
|
402
|
+
params: chunk.flatMap((sourceFile) => [OPENCODE_SOURCE, sourceFile, 1, importedAt]),
|
|
403
|
+
});
|
|
404
|
+
}
|
|
405
|
+
return operations;
|
|
406
|
+
}
|
|
407
|
+
export { ETL_TABLE_DDL, ensureTargetTables, insertLedger, insertRecord, ledgerExists, readCheckpoint, reconcileFullImport, resetCheckpoints, targetTableFor, timestamp, writeCheckpoint, };
|
package/dist/mappers.d.ts
CHANGED
|
@@ -1,5 +1,5 @@
|
|
|
1
1
|
import { z } from 'zod';
|
|
2
|
-
import type { JsonObject, SplitEntry } from './types';
|
|
2
|
+
import type { JsonObject, SplitEntry, TransformContext } from './types';
|
|
3
3
|
/** Map a Claude Code JSONL record into one or more ETL split entries. */
|
|
4
4
|
export declare function claudeSplit(raw: Record<string, unknown>): readonly SplitEntry[];
|
|
5
5
|
/** Map a Pi JSONL record into one or more ETL split entries. */
|
|
@@ -10,6 +10,8 @@ export declare function ompSplit(raw: Record<string, unknown>): readonly SplitEn
|
|
|
10
10
|
export declare function codexSplit(raw: Record<string, unknown>): readonly SplitEntry[];
|
|
11
11
|
/** Map an Antigravity (AGY) JSONL record into one or more ETL split entries. */
|
|
12
12
|
export declare function agySplit(raw: Record<string, unknown>): readonly SplitEntry[];
|
|
13
|
+
/** Map Gemini CLI message, tool, and session-state events into the forensic contract tables. */
|
|
14
|
+
export declare function geminiSplit(raw: Record<string, unknown>, context?: TransformContext): readonly SplitEntry[];
|
|
13
15
|
/** Map a Grok JSONL record into one or more ETL split entries. */
|
|
14
16
|
export declare function grokSplit(raw: Record<string, unknown>): readonly SplitEntry[];
|
|
15
17
|
/** Compute a stable field-shape key from a JSON object: sorted, lowercased, +-joined top-level keys. */
|
|
@@ -26,6 +28,8 @@ export declare const CODEX_FIELD_MAP: Record<string, string>;
|
|
|
26
28
|
export declare const AGY_FIELD_MAP: Record<string, string>;
|
|
27
29
|
/** Identity field map for the Grok mapper. */
|
|
28
30
|
export declare const GROK_FIELD_MAP: Record<string, string>;
|
|
31
|
+
/** Identity field map for the Gemini mapper. */
|
|
32
|
+
export declare const GEMINI_FIELD_MAP: Record<string, string>;
|
|
29
33
|
/** Passthrough Zod schema for the Claude mapper. */
|
|
30
34
|
export declare const CLAUDE_SCHEMA: z.ZodObject<{}, z.core.$loose>;
|
|
31
35
|
/** Passthrough Zod schema for the Pi mapper. */
|
|
@@ -38,4 +42,6 @@ export declare const CODEX_SCHEMA: z.ZodObject<{}, z.core.$loose>;
|
|
|
38
42
|
export declare const AGY_SCHEMA: z.ZodObject<{}, z.core.$loose>;
|
|
39
43
|
/** Passthrough Zod schema for the Grok mapper. */
|
|
40
44
|
export declare const GROK_SCHEMA: z.ZodObject<{}, z.core.$loose>;
|
|
45
|
+
/** Passthrough Zod schema for the Gemini mapper. */
|
|
46
|
+
export declare const GEMINI_SCHEMA: z.ZodObject<{}, z.core.$loose>;
|
|
41
47
|
//# sourceMappingURL=mappers.d.ts.map
|
package/dist/mappers.d.ts.map
CHANGED
|
@@ -1 +1 @@
|
|
|
1
|
-
{"version":3,"file":"mappers.d.ts","sourceRoot":"","sources":["../src/mappers.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,OAAO,KAAK,EAAE,UAAU,EAAE,UAAU,EAAE,MAAM,SAAS,CAAC;
|
|
1
|
+
{"version":3,"file":"mappers.d.ts","sourceRoot":"","sources":["../src/mappers.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAExB,OAAO,KAAK,EAAE,UAAU,EAAE,UAAU,EAAE,gBAAgB,EAAE,MAAM,SAAS,CAAC;AA2FxE,yEAAyE;AACzE,wBAAgB,WAAW,CAAC,GAAG,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,SAAS,UAAU,EAAE,CA4G/E;AAMD,gEAAgE;AAChE,wBAAgB,OAAO,CAAC,GAAG,EAAE,OAAO,GAAG,SAAS,UAAU,EAAE,CAsE3D;AAMD,kEAAkE;AAClE,wBAAgB,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,SAAS,UAAU,EAAE,CA4F5E;AAMD,mEAAmE;AACnE,wBAAgB,UAAU,CAAC,GAAG,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,SAAS,UAAU,EAAE,CAiG9E;AAMD,gFAAgF;AAChF,wBAAgB,QAAQ,CAAC,GAAG,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,SAAS,UAAU,EAAE,CAgH5E;AAMD,gGAAgG;AAChG,wBAAgB,WAAW,CAAC,GAAG,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,EAAE,OAAO,CAAC,EAAE,gBAAgB,GAAG,SAAS,UAAU,EAAE,CAgF3G;AAyMD,kEAAkE;AAClE,wBAAgB,SAAS,CAAC,GAAG,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,SAAS,UAAU,EAAE,CAyS7E;AAwDD,wGAAwG;AACxG,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,UAAU,GAAG,MAAM,CAKxD;AAMD,gDAAgD;AAChD,eAAO,MAAM,gBAAgB,wBAAsE,CAAC;AACpG,4CAA4C;AAC5C,eAAO,MAAM,YAAY,wBAAsE,CAAC;AAChG,6CAA6C;AAC7C,eAAO,MAAM,aAAa,wBAAsE,CAAC;AACjG,+CAA+C;AAC/C,eAAO,MAAM,eAAe,wBAAsE,CAAC;AACnG,6CAA6C;AAC7C,eAAO,MAAM,aAAa,wBAAsE,CAAC;AACjG,8CAA8C;AAC9C,eAAO,MAAM,cAAc,wBAAsE,CAAC;AAClG,gDAAgD;AAChD,eAAO,MAAM,gBAAgB,wBAAsE,CAAC;AAQpG,oDAAoD;AACpD,eAAO,MAAM,aAAa,gCAAoB,CAAC;AAC/C,gDAAgD;AAChD,eAAO,MAAM,SAAS,gCAAoB,CAAC;AAC3C,iDAAiD;AACjD,eAAO,MAAM,UAAU,gCAAoB,CAAC;AAC5C,mDAAmD;AACnD,eAAO,MAAM,YAAY,gCAAoB,CAAC;AAC9C,iDAAiD;AACjD,eAAO,MAAM,UAAU,gCAAoB,CAAC;AAC5C,kDAAkD;AAClD,eAAO,MAAM,WAAW,gCAAoB,CAAC;AAC7C,oDAAoD;AACpD,eAAO,MAAM,aAAa,gCAAoB,CAAC"}
|
package/dist/mappers.js
CHANGED
|
@@ -366,9 +366,9 @@ export function codexSplit(raw) {
|
|
|
366
366
|
record: {
|
|
367
367
|
session_id: sessionId,
|
|
368
368
|
seq,
|
|
369
|
-
role: '
|
|
369
|
+
role: 'meta',
|
|
370
370
|
record_type: 'short_format',
|
|
371
|
-
disposition: '
|
|
371
|
+
disposition: 'meta',
|
|
372
372
|
ts,
|
|
373
373
|
provenance: 'ambient',
|
|
374
374
|
},
|
|
@@ -552,6 +552,109 @@ export function agySplit(raw) {
|
|
|
552
552
|
return entries;
|
|
553
553
|
}
|
|
554
554
|
// ---------------------------------------------------------------------------
|
|
555
|
+
// Gemini mapper
|
|
556
|
+
// ---------------------------------------------------------------------------
|
|
557
|
+
/** Map Gemini CLI message, tool, and session-state events into the forensic contract tables. */
|
|
558
|
+
export function geminiSplit(raw, context) {
|
|
559
|
+
const recordType = String(raw.type ?? (raw.kind !== undefined ? 'session' : raw.$set !== undefined ? 'state' : ''));
|
|
560
|
+
const state = o(raw.$set);
|
|
561
|
+
const sessionId = sessionIdFromContext(context, raw);
|
|
562
|
+
const seq = context?.sourceLine ?? 0;
|
|
563
|
+
const ts = s(raw.timestamp, raw.startTime, raw.lastUpdated, state.lastUpdated) ?? new Date(0).toISOString();
|
|
564
|
+
if (recordType === 'session' || recordType === 'state') {
|
|
565
|
+
return [
|
|
566
|
+
{
|
|
567
|
+
targetTable: 'history_message',
|
|
568
|
+
record: {
|
|
569
|
+
session_id: sessionId,
|
|
570
|
+
seq,
|
|
571
|
+
role: 'meta',
|
|
572
|
+
record_type: recordType,
|
|
573
|
+
disposition: 'meta',
|
|
574
|
+
ts,
|
|
575
|
+
content_text: s(state.summary) ?? null,
|
|
576
|
+
provenance: detectProvenance(context?.sourceFile),
|
|
577
|
+
},
|
|
578
|
+
},
|
|
579
|
+
];
|
|
580
|
+
}
|
|
581
|
+
const role = recordType === 'gemini' ? 'assistant' : recordType === 'user' ? 'user' : 'system';
|
|
582
|
+
const tokens = o(raw.tokens);
|
|
583
|
+
const content = geminiContent(raw);
|
|
584
|
+
const entries = [
|
|
585
|
+
{
|
|
586
|
+
targetTable: 'history_message',
|
|
587
|
+
record: {
|
|
588
|
+
session_id: sessionId,
|
|
589
|
+
seq,
|
|
590
|
+
role,
|
|
591
|
+
record_type: recordType,
|
|
592
|
+
disposition: 'keep',
|
|
593
|
+
ts,
|
|
594
|
+
duration_ms: null,
|
|
595
|
+
model: s(raw.model) ?? null,
|
|
596
|
+
input_tokens: numberOrNull(tokens.input),
|
|
597
|
+
output_tokens: numberOrNull(tokens.output),
|
|
598
|
+
cache_read_tokens: numberOrNull(tokens.cached),
|
|
599
|
+
cache_write_tokens: null,
|
|
600
|
+
cost_usd: null,
|
|
601
|
+
content_text: content,
|
|
602
|
+
cwd: null,
|
|
603
|
+
provenance: detectProvenance(context?.sourceFile),
|
|
604
|
+
},
|
|
605
|
+
},
|
|
606
|
+
];
|
|
607
|
+
if (Array.isArray(raw.toolCalls)) {
|
|
608
|
+
for (const value of raw.toolCalls) {
|
|
609
|
+
if (value === null || typeof value !== 'object' || Array.isArray(value))
|
|
610
|
+
continue;
|
|
611
|
+
const tool = value;
|
|
612
|
+
const status = s(tool.status) ?? 'unknown';
|
|
613
|
+
const timestamp = s(tool.timestamp);
|
|
614
|
+
const result = tool.result ?? tool.resultDisplay;
|
|
615
|
+
entries.push({
|
|
616
|
+
targetTable: 'history_tool_call',
|
|
617
|
+
record: {
|
|
618
|
+
_messageSplitIndex: 0,
|
|
619
|
+
session_id: sessionId,
|
|
620
|
+
seq,
|
|
621
|
+
tool_name: s(tool.name, tool.displayName) ?? 'unknown',
|
|
622
|
+
args_digest: argsDigest(tool.args),
|
|
623
|
+
status,
|
|
624
|
+
started_at: timestamp,
|
|
625
|
+
completed_at: timestamp,
|
|
626
|
+
duration_ms: null,
|
|
627
|
+
result_bytes: result === undefined ? null : new TextEncoder().encode(JSON.stringify(result)).byteLength,
|
|
628
|
+
error_text: status === 'error' ? (s(tool.resultDisplay, tool.result) ?? null) : null,
|
|
629
|
+
},
|
|
630
|
+
});
|
|
631
|
+
}
|
|
632
|
+
}
|
|
633
|
+
return entries;
|
|
634
|
+
}
|
|
635
|
+
function sessionIdFromContext(context, raw) {
|
|
636
|
+
if (context !== undefined) {
|
|
637
|
+
const file = context.sourceFile.split('/').at(-1);
|
|
638
|
+
if (file !== undefined)
|
|
639
|
+
return file.replace(/\.jsonl$/, '');
|
|
640
|
+
}
|
|
641
|
+
return s(raw.sessionId, raw.id) ?? 'unknown';
|
|
642
|
+
}
|
|
643
|
+
function geminiContent(raw) {
|
|
644
|
+
const content = extractContentText(raw.content);
|
|
645
|
+
const thoughts = Array.isArray(raw.thoughts)
|
|
646
|
+
? raw.thoughts
|
|
647
|
+
.map((value) => (value !== null && typeof value === 'object' ? s(o(value).description) : undefined))
|
|
648
|
+
.filter((value) => value !== undefined)
|
|
649
|
+
.join('\n')
|
|
650
|
+
: '';
|
|
651
|
+
return ([content, thoughts].filter((value) => value !== undefined && value.length > 0).join('\n') ||
|
|
652
|
+
null);
|
|
653
|
+
}
|
|
654
|
+
function numberOrNull(value) {
|
|
655
|
+
return typeof value === 'number' && Number.isFinite(value) ? value : null;
|
|
656
|
+
}
|
|
657
|
+
// ---------------------------------------------------------------------------
|
|
555
658
|
// Grok mapper
|
|
556
659
|
// ---------------------------------------------------------------------------
|
|
557
660
|
/**
|
|
@@ -566,13 +669,29 @@ function normalizeGrokRecord(raw) {
|
|
|
566
669
|
const update = params?.update;
|
|
567
670
|
const isUpdatesShape = typeof raw.method === 'string' && params !== undefined && update !== undefined;
|
|
568
671
|
const body = isUpdatesShape ? update : raw;
|
|
569
|
-
const recordType = String((isUpdatesShape ? update?.sessionUpdate : raw.type) ??
|
|
570
|
-
|
|
571
|
-
|
|
672
|
+
const recordType = String((isUpdatesShape ? update?.sessionUpdate : raw.type) ??
|
|
673
|
+
raw.sessionUpdate ??
|
|
674
|
+
raw.eventType ??
|
|
675
|
+
(raw.file_snapshots !== undefined || raw.after_snapshots !== undefined
|
|
676
|
+
? 'file_snapshot'
|
|
677
|
+
: raw.prompt !== undefined
|
|
678
|
+
? 'user_message'
|
|
679
|
+
: raw.answer !== undefined || raw.question !== undefined
|
|
680
|
+
? 'assistant'
|
|
681
|
+
: ''));
|
|
682
|
+
const sessionId = s(params?.sessionId, raw.session_id, raw.sessionId, raw.parentSessionId, raw.btwSessionId, raw.params?.sessionId) ?? 'unknown';
|
|
683
|
+
const ts = normalizeTs(raw.ts ?? raw.timestamp ?? raw.created_at ?? raw.askedAt ?? body.ts ?? body.timestamp);
|
|
572
684
|
const seq = typeof raw.seq === 'number' ? raw.seq : typeof body.seq === 'number' ? body.seq : 0;
|
|
573
685
|
const meta = (body._meta ?? params?._meta ?? raw._meta);
|
|
574
686
|
const model = s(meta?.modelId, body.model, raw.model);
|
|
575
|
-
const contentText = extractGrokContent(body.content ??
|
|
687
|
+
const contentText = extractGrokContent(body.content ??
|
|
688
|
+
body.text ??
|
|
689
|
+
raw.content ??
|
|
690
|
+
raw.text ??
|
|
691
|
+
raw.prompt ??
|
|
692
|
+
(raw.answer !== undefined || raw.question !== undefined
|
|
693
|
+
? [raw.question, raw.answer].filter((value) => typeof value === 'string').join('\n')
|
|
694
|
+
: undefined));
|
|
576
695
|
const usage = (body.usage ?? raw.usage);
|
|
577
696
|
const toolMeta = meta?.['x.ai/tool'];
|
|
578
697
|
const toolName = s(body.title, body.tool_name, body.name, toolMeta?.name, toolMeta?.label, raw.tool_name, raw.name) ?? '';
|
|
@@ -756,6 +875,27 @@ export function grokSplit(raw) {
|
|
|
756
875
|
return entries;
|
|
757
876
|
}
|
|
758
877
|
if (recordType === 'tool_completed') {
|
|
878
|
+
entries.push({
|
|
879
|
+
targetTable: 'history_message',
|
|
880
|
+
record: {
|
|
881
|
+
session_id: sessionId,
|
|
882
|
+
seq,
|
|
883
|
+
role: 'assistant',
|
|
884
|
+
record_type: recordType,
|
|
885
|
+
disposition: 'keep',
|
|
886
|
+
ts,
|
|
887
|
+
duration_ms: n.durationMs ?? null,
|
|
888
|
+
model: n.model ?? null,
|
|
889
|
+
input_tokens: null,
|
|
890
|
+
output_tokens: null,
|
|
891
|
+
cache_read_tokens: null,
|
|
892
|
+
cache_write_tokens: null,
|
|
893
|
+
cost_usd: null,
|
|
894
|
+
content_text: n.contentText,
|
|
895
|
+
cwd: null,
|
|
896
|
+
provenance: 'ambient',
|
|
897
|
+
},
|
|
898
|
+
});
|
|
759
899
|
entries.push({
|
|
760
900
|
targetTable: 'history_tool_call',
|
|
761
901
|
record: {
|
|
@@ -1017,6 +1157,8 @@ export const CODEX_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_
|
|
|
1017
1157
|
export const AGY_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
|
|
1018
1158
|
/** Identity field map for the Grok mapper. */
|
|
1019
1159
|
export const GROK_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
|
|
1160
|
+
/** Identity field map for the Gemini mapper. */
|
|
1161
|
+
export const GEMINI_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
|
|
1020
1162
|
// ---------------------------------------------------------------------------
|
|
1021
1163
|
// Zod schemas for the six source mappers
|
|
1022
1164
|
// ---------------------------------------------------------------------------
|
|
@@ -1033,3 +1175,5 @@ export const CODEX_SCHEMA = passthroughSchema;
|
|
|
1033
1175
|
export const AGY_SCHEMA = passthroughSchema;
|
|
1034
1176
|
/** Passthrough Zod schema for the Grok mapper. */
|
|
1035
1177
|
export const GROK_SCHEMA = passthroughSchema;
|
|
1178
|
+
/** Passthrough Zod schema for the Gemini mapper. */
|
|
1179
|
+
export const GEMINI_SCHEMA = passthroughSchema;
|
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
import { type DbAdapter } from '@gobing-ai/ts-db';
|
|
2
|
+
import { type FileSystem, type RuntimePaths } from '@gobing-ai/ts-runtime';
|
|
3
|
+
import type { ImportMode, ImportResult, RedactionRule } from './types';
|
|
4
|
+
/** Dependencies and mode controls for importing OpenCode's SQLite history. */
|
|
5
|
+
export interface OpenCodeImportOptions {
|
|
6
|
+
db: DbAdapter;
|
|
7
|
+
sourceDatabase?: string;
|
|
8
|
+
fileSystem?: FileSystem;
|
|
9
|
+
paths?: RuntimePaths;
|
|
10
|
+
mode?: ImportMode;
|
|
11
|
+
dryRun?: boolean;
|
|
12
|
+
redactionRules?: readonly RedactionRule[];
|
|
13
|
+
now?: () => Date;
|
|
14
|
+
}
|
|
15
|
+
/** Import OpenCode's current SQLite history store into the forensic contract tables. */
|
|
16
|
+
export declare function runOpenCodeImport(options: OpenCodeImportOptions): Promise<ImportResult>;
|
|
17
|
+
//# sourceMappingURL=opencode-importer.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"opencode-importer.d.ts","sourceRoot":"","sources":["../src/opencode-importer.ts"],"names":[],"mappings":"AAAA,OAAO,EAAmB,KAAK,SAAS,EAAkB,MAAM,kBAAkB,CAAC;AACnF,OAAO,EAGH,KAAK,UAAU,EAEf,KAAK,YAAY,EACpB,MAAM,uBAAuB,CAAC;AAc/B,OAAO,KAAK,EAAe,UAAU,EAAE,YAAY,EAAgC,aAAa,EAAE,MAAM,SAAS,CAAC;AAYlH,8EAA8E;AAC9E,MAAM,WAAW,qBAAqB;IAClC,EAAE,EAAE,SAAS,CAAC;IACd,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,UAAU,CAAC,EAAE,UAAU,CAAC;IACxB,KAAK,CAAC,EAAE,YAAY,CAAC;IACrB,IAAI,CAAC,EAAE,UAAU,CAAC;IAClB,MAAM,CAAC,EAAE,OAAO,CAAC;IACjB,cAAc,CAAC,EAAE,SAAS,aAAa,EAAE,CAAC;IAC1C,GAAG,CAAC,EAAE,MAAM,IAAI,CAAC;CACpB;AAED,wFAAwF;AACxF,wBAAsB,iBAAiB,CAAC,OAAO,EAAE,qBAAqB,GAAG,OAAO,CAAC,YAAY,CAAC,CAuI7F"}
|