@gobing-ai/ts-llm-jsonl-importer 0.4.19 → 0.4.21

package/dist/mappers.js

@@ -0,0 +1,1035 @@
+import { z } from 'zod';
+import { sha256 } from './hash.js';
+// ---------------------------------------------------------------------------
+// Helper: identity field map for typed columns
+// ---------------------------------------------------------------------------
+/** Columns the mapper may produce for history_message (excluding framework-managed fields). */
+const MESSAGE_MAPPER_KEYS = [
+    'session_id',
+    'seq',
+    'turn_index',
+    'role',
+    'record_type',
+    'disposition',
+    'ts',
+    'duration_ms',
+    'model',
+    'input_tokens',
+    'output_tokens',
+    'cache_read_tokens',
+    'cache_write_tokens',
+    'cost_usd',
+    'content_text',
+    'cwd',
+    'provenance',
+    'run_id',
+    'task_wbs',
+];
+/** Columns the mapper may produce for history_tool_call. */
+const TOOL_CALL_MAPPER_KEYS = [
+    '_messageSplitIndex',
+    'session_id',
+    'seq',
+    'tool_name',
+    'args_digest',
+    'status',
+    'started_at',
+    'completed_at',
+    'duration_ms',
+    'result_bytes',
+    'error_text',
+];
+/** Build an identity fieldMap for the given column list. */
+function identityFieldMap(keys) {
+    const map = {};
+    for (const key of keys) {
+        map[key] = key;
+    }
+    return map;
+}
+// ---------------------------------------------------------------------------
+// Source-specific helpers
+// ---------------------------------------------------------------------------
+/** Build a provenance string: 'ambient' unless cwd implies a spur workspace. */
+function detectProvenance(cwd) {
+    if (cwd === undefined)
+        return 'ambient';
+    return cwd.includes('/spur') || cwd.includes('/spur-new') ? 'spur-run' : 'ambient';
+}
+/** Compute args_digest: sha256 of stable-jsonified, key-sorted, redacted args. */
+function argsDigest(args) {
+    const redacted = redactArgs(args);
+    return sha256(redacted);
+}
+/** Redact tool arguments for digest: replace string values > 80 chars or containing secrets. */
+function redactArgs(args) {
+    if (typeof args === 'string') {
+        if (args.length > 80)
+            return '[REDACTED:long]';
+        if (/[A-Za-z0-9+/]{40,}=*|[A-Za-z0-9_-]{20,}/.test(args))
+            return '[REDACTED:secret]';
+        return args;
+    }
+    if (Array.isArray(args))
+        return args.map(redactArgs);
+    if (args !== null && typeof args === 'object') {
+        const result = {};
+        for (const [key, value] of Object.entries(args)) {
+            result[key] = redactArgs(value);
+        }
+        return result;
+    }
+    return args;
+}
+// ---------------------------------------------------------------------------
+// Claude mapper
+// ---------------------------------------------------------------------------
+/** Map a Claude Code JSONL record into one or more ETL split entries. */
+export function claudeSplit(raw) {
+    const entries = [];
+    const sessionId = s(raw.sessionId, raw.conversation_uuid) ?? 'unknown';
+    const seq = typeof raw.seq === 'number' ? raw.seq : typeof raw.messageIndex === 'number' ? raw.messageIndex : 0;
+    const ts = s(raw.ts, raw.timestamp, raw.createdAt) ?? new Date(0).toISOString();
+    const role = mapRole(raw.type ?? raw.role);
+    const recordType = String(raw.type ?? '');
+    const model = s(raw.model);
+    const cwd = s(raw.cwd, raw.dir);
+    // Check for system/turn-duration records
+    if (raw.subtype === 'turn_duration' && typeof raw.durationMs === 'number') {
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'system',
+                record_type: 'turn_duration',
+                disposition: 'meta',
+                ts,
+                duration_ms: raw.durationMs,
+                content_text: null,
+                provenance: detectProvenance(cwd),
+            },
+        });
+        return entries;
+    }
+    if (recordType === 'attachment' || recordType.startsWith('file-history')) {
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'meta',
+                record_type: recordType,
+                disposition: 'meta',
+                ts,
+                provenance: detectProvenance(cwd),
+            },
+        });
+        return entries;
+    }
+    // Main message
+    const hasUsage = typeof raw.usage === 'object' && raw.usage !== null;
+    const usage = hasUsage ? raw.usage : undefined;
+    const inputTokens = typeof raw.inputTokens === 'number' ? raw.inputTokens : undefined;
+    const outputTokens = typeof raw.outputTokens === 'number' ? raw.outputTokens : undefined;
+    const cacheRead = (usage?.cacheReadTokens ?? usage?.cache_read_tokens ?? undefined);
+    const cacheWrite = (usage?.cacheWriteTokens ?? usage?.cache_write_tokens ?? undefined);
+    const costUsd = computeCost(inputTokens, outputTokens, model);
+    const contentBlocks = raw.content;
+    const contentText = extractContentText(contentBlocks);
+    const messageSplitIndex = entries.length;
+    entries.push({
+        targetTable: 'history_message',
+        record: {
+            session_id: sessionId,
+            seq,
+            turn_index: typeof raw.turnIndex === 'number' ? raw.turnIndex : undefined,
+            role,
+            record_type: recordType,
+            disposition: 'keep',
+            ts,
+            duration_ms: undefined,
+            model: model ?? null,
+            input_tokens: inputTokens ?? null,
+            output_tokens: outputTokens ?? null,
+            cache_read_tokens: cacheRead ?? null,
+            cache_write_tokens: cacheWrite ?? null,
+            cost_usd: costUsd ?? null,
+            content_text: contentText ?? null,
+            cwd: cwd ?? null,
+            provenance: detectProvenance(cwd),
+        },
+    });
+    // Tool calls from content blocks
+    if (Array.isArray(contentBlocks) && role === 'assistant') {
+        for (const block of contentBlocks) {
+            if (typeof block !== 'object' || block === null)
+                continue;
+            const b = block;
+            if (b.type === 'tool_use') {
+                entries.push({
+                    targetTable: 'history_tool_call',
+                    record: {
+                        _messageSplitIndex: messageSplitIndex,
+                        session_id: sessionId,
+                        seq,
+                        tool_name: String(b.name ?? ''),
+                        args_digest: argsDigest(b.input),
+                        status: 'ok',
+                        started_at: undefined,
+                        completed_at: undefined,
+                        duration_ms: undefined,
+                        result_bytes: undefined,
+                        error_text: undefined,
+                    },
+                });
+            }
+        }
+    }
+    return entries;
+}
+// ---------------------------------------------------------------------------
+// Pi mapper
+// ---------------------------------------------------------------------------
+/** Map a Pi JSONL record into one or more ETL split entries. */
+export function piSplit(raw) {
+    const entries = [];
+    const r = raw;
+    const sessionId = s(r.id, o(r.session).id) ?? 'unknown';
+    const seq = typeof r.seq === 'number' ? r.seq : 0;
+    const ts = s(r.ts, r.timestamp, r.createdAt) ?? new Date(0).toISOString();
+    const role = mapRole(r.type ?? r.role);
+    const recordType = String(r.type ?? '');
+    const model = s(r.model, o(r.message).model);
+    const cwd = s(r.cwd, r.dir);
+    const msg = r.message;
+    const usage = msg?.usage;
+    const inputTokens = (usage?.input ?? usage?.input_tokens ?? undefined);
+    const outputTokens = (usage?.output ?? usage?.output_tokens ?? undefined);
+    const cacheRead = (usage?.cacheRead ?? usage?.cache_read_tokens ?? undefined);
+    const cacheWrite = (usage?.cacheWrite ?? usage?.cache_write_tokens ?? undefined);
+    const costObj = (msg?.cost ?? r.cost);
+    const costUsd = typeof costObj?.total === 'number' ? costObj.total : computeCost(inputTokens, outputTokens, model);
+    const messageSplitIndex = entries.length;
+    entries.push({
+        targetTable: 'history_message',
+        record: {
+            session_id: sessionId,
+            seq,
+            role,
+            record_type: recordType,
+            disposition: 'keep',
+            ts,
+            duration_ms: undefined,
+            model: model ?? null,
+            input_tokens: inputTokens ?? null,
+            output_tokens: outputTokens ?? null,
+            cache_read_tokens: cacheRead ?? null,
+            cache_write_tokens: cacheWrite ?? null,
+            cost_usd: costUsd ?? null,
+            content_text: s(r.content, r.text, msg?.content) ?? null,
+            cwd: cwd ?? null,
+            provenance: detectProvenance(cwd),
+        },
+    });
+    // Tool calls from toolCall blocks in assistant content
+    if (Array.isArray(r.content) && role === 'assistant') {
+        for (const block of r.content) {
+            if (block?.toolCall) {
+                const tc = block.toolCall;
+                entries.push({
+                    targetTable: 'history_tool_call',
+                    record: {
+                        _messageSplitIndex: messageSplitIndex,
+                        session_id: sessionId,
+                        seq,
+                        tool_name: String(tc.name ?? ''),
+                        args_digest: argsDigest(tc.input ?? tc.arguments),
+                        status: 'ok',
+                        started_at: undefined,
+                        completed_at: undefined,
+                        duration_ms: undefined,
+                        result_bytes: undefined,
+                        error_text: undefined,
+                    },
+                });
+            }
+        }
+    }
+    return entries;
+}
+// ---------------------------------------------------------------------------
+// OMP mapper (near-identical to pi)
+// ---------------------------------------------------------------------------
+/** Map an OMP JSONL record into one or more ETL split entries. */
+export function ompSplit(raw) {
+    const recordType = String(raw.type ?? '');
+    if (recordType === 'title' ||
+        recordType === 'title_change' ||
+        recordType === 'service_tier_change' ||
+        recordType === 'ttsr_injection' ||
+        recordType === 'session_init' ||
+        recordType === 'compaction') {
+        return [
+            {
+                targetTable: 'history_message',
+                record: {
+                    session_id: s(raw.id, o(raw.session).id) ?? 'unknown',
+                    seq: typeof raw.seq === 'number' ? raw.seq : 0,
+                    role: 'meta',
+                    record_type: recordType,
+                    disposition: 'meta',
+                    ts: s(raw.ts, raw.timestamp) ?? new Date(0).toISOString(),
+                    provenance: 'ambient',
+                },
+            },
+        ];
+    }
+    const entries = [];
+    const sessionId = s(raw.id, o(raw.session).id) ?? 'unknown';
+    const seq = typeof raw.seq === 'number' ? raw.seq : 0;
+    const ts = s(raw.ts, raw.timestamp, raw.createdAt) ?? new Date(0).toISOString();
+    const role = mapRole(raw.type ?? raw.role);
+    const model = s(raw.model, o(raw.message).model);
+    const cwd = s(raw.cwd, raw.dir);
+    const msg = raw.message;
+    const usage = msg?.usage;
+    const inputTokens = (usage?.input ?? usage?.input_tokens ?? undefined);
+    const outputTokens = (usage?.output ?? usage?.output_tokens ?? undefined);
+    const cacheRead = (usage?.cacheRead ?? usage?.cache_read_tokens ?? undefined);
+    const cacheWrite = (usage?.cacheWrite ?? usage?.cache_write_tokens ?? undefined);
+    const costObj = (msg?.cost ?? raw.cost);
+    const costUsd = typeof costObj?.total === 'number' ? costObj.total : computeCost(inputTokens, outputTokens, model);
+    const messageSplitIndex = entries.length;
+    entries.push({
+        targetTable: 'history_message',
+        record: {
+            session_id: sessionId,
+            seq,
+            role,
+            record_type: recordType,
+            disposition: 'keep',
+            ts,
+            duration_ms: undefined,
+            model: model ?? null,
+            input_tokens: inputTokens ?? null,
+            output_tokens: outputTokens ?? null,
+            cache_read_tokens: cacheRead ?? null,
+            cache_write_tokens: cacheWrite ?? null,
+            cost_usd: costUsd ?? null,
+            content_text: s(raw.content, raw.text, msg?.content) ?? null,
+            cwd: cwd ?? null,
+            provenance: detectProvenance(cwd),
+        },
+    });
+    if (Array.isArray(raw.content) && role === 'assistant') {
+        for (const block of raw.content) {
+            if (block?.toolCall) {
+                const tc = block.toolCall;
+                entries.push({
+                    targetTable: 'history_tool_call',
+                    record: {
+                        _messageSplitIndex: messageSplitIndex,
+                        session_id: sessionId,
+                        seq,
+                        tool_name: String(tc.name ?? ''),
+                        args_digest: argsDigest(tc.input ?? tc.arguments),
+                        status: 'ok',
+                        started_at: undefined,
+                        completed_at: undefined,
+                        duration_ms: undefined,
+                        result_bytes: undefined,
+                        error_text: undefined,
+                    },
+                });
+            }
+        }
+    }
+    return entries;
+}
+// ---------------------------------------------------------------------------
+// Codex mapper
+// ---------------------------------------------------------------------------
+/** Map a Codex JSONL record into one or more ETL split entries. */
+export function codexSplit(raw) {
+    const entries = [];
+    const payload = (raw.payload ?? raw);
+    const sessionId = s(raw.session_id, o(raw.session_meta).id, raw.id) ?? 'unknown';
+    const seq = typeof raw.seq === 'number' ? raw.seq : 0;
+    const ts = s(raw.timestamp, raw.ts, payload.ts) ?? new Date(0).toISOString();
+    const recordType = String(raw.type ?? '');
+    // Check for older short format
+    if (recordType === '' && raw.id && raw.timestamp && raw.instructions) {
+        return [
+            {
+                targetTable: 'history_message',
+                record: {
+                    session_id: sessionId,
+                    seq,
+                    role: 'unknown',
+                    record_type: 'short_format',
+                    disposition: 'unknown',
+                    ts,
+                    provenance: 'ambient',
+                },
+            },
+        ];
+    }
+    const role = mapRole(recordType);
+    const model = s(raw.model, payload.model, o(o(raw.turn_context).payload).model);
+    const cwd = s(raw.cwd, raw.dir);
+    const tokenCount = payload.token_count;
+    const inputTokens = (tokenCount?.input ?? tokenCount?.input_tokens ?? undefined);
+    const outputTokens = (tokenCount?.output ?? tokenCount?.output_tokens ?? undefined);
+    const turnCtx = raw.turn_context;
+    const turnPayload = turnCtx?.payload;
+    const turnInputTokens = turnPayload?.input_tokens;
+    const turnOutputTokens = turnPayload?.output_tokens;
+    const finalInput = inputTokens !== undefined ? inputTokens : turnInputTokens;
+    const finalOutput = outputTokens !== undefined ? outputTokens : turnOutputTokens;
+    const costUsd = computeCost(finalInput, finalOutput, model);
+    const messageSplitIndex = entries.length;
+    entries.push({
+        targetTable: 'history_message',
+        record: {
+            session_id: sessionId,
+            seq,
+            role,
+            record_type: recordType,
+            disposition: 'keep',
+            ts,
+            duration_ms: undefined,
+            model: model ?? null,
+            input_tokens: finalInput ?? null,
+            output_tokens: finalOutput ?? null,
+            cache_read_tokens: null,
+            cache_write_tokens: null,
+            cost_usd: costUsd ?? null,
+            content_text: s(payload.content, payload.text, raw.content, raw.text) ?? null,
+            cwd: cwd ?? null,
+            provenance: detectProvenance(cwd),
+        },
+    });
+    // Tool calls: function_call in response_item
+    const responseItems = payload.response_item
+        ? [payload.response_item]
+        : Array.isArray(payload.response_items)
+            ? payload.response_items
+            : [];
+    for (const item of responseItems) {
+        if (item.function_call) {
+            const fc = item.function_call;
+            entries.push({
+                targetTable: 'history_tool_call',
+                record: {
+                    _messageSplitIndex: messageSplitIndex,
+                    session_id: sessionId,
+                    seq,
+                    tool_name: String(fc.name ?? ''),
+                    args_digest: argsDigest(fc.arguments),
+                    status: 'ok',
+                    started_at: undefined,
+                    completed_at: undefined,
+                    duration_ms: undefined,
+                    result_bytes: undefined,
+                    error_text: undefined,
+                },
+            });
+        }
+    }
+    return entries;
+}
+// ---------------------------------------------------------------------------
+// AGY mapper (Antigravity CLI)
+// ---------------------------------------------------------------------------
+/** Map an Antigravity (AGY) JSONL record into one or more ETL split entries. */
+export function agySplit(raw) {
+    const entries = [];
+    const recordType = String(raw.type ?? '');
+    const sessionId = s(raw.session_id, raw.conversation_id) ?? 'unknown';
+    const seq = typeof raw.seq === 'number' ? raw.seq : typeof raw.step_index === 'number' ? raw.step_index : 0;
+    const ts = s(raw.created_at, raw.timestamp, raw.ts) ?? new Date(0).toISOString();
+    let role;
+    let disposition;
+    let contentText = null;
+    // Classification from task 0463 field map (agy transcript types).
+    switch (recordType) {
+        case 'USER_INPUT':
+        case 'ASK_QUESTION':
+            role = 'user';
+            disposition = 'keep';
+            contentText = s(raw.content, raw.text, raw.message) ?? null;
+            break;
+        case 'PLANNER_RESPONSE':
+            role = 'assistant';
+            disposition = 'keep';
+            contentText = s(raw.content, raw.text, raw.message, raw.plan) ?? null;
+            break;
+        case 'ERROR_MESSAGE':
+            role = 'system';
+            disposition = 'keep';
+            contentText = s(raw.content, raw.text, raw.message) ?? null;
+            break;
+        case 'RUN_COMMAND':
+        case 'VIEW_FILE':
+        case 'GREP_SEARCH':
+        case 'LIST_DIRECTORY':
+        case 'READ_URL_CONTENT':
+            role = 'tool';
+            disposition = 'keep';
+            contentText = s(raw.content, raw.command, raw.file_path, raw.pattern) ?? null;
+            break;
+        case 'CODE_ACTION':
+        case 'INVOKE_SUBAGENT':
+            role = 'tool';
+            disposition = 'keep';
+            contentText = s(raw.content, raw.text) ?? null;
+            break;
+        case 'CONVERSATION_HISTORY':
+        case 'CHECKPOINT':
+        case 'GENERIC':
+            role = 'meta';
+            disposition = 'meta';
+            contentText = s(raw.content, raw.text) ?? null;
+            break;
+        default:
+            if (recordType.length === 0) {
+                role = 'unknown';
+                disposition = 'unknown';
+            }
+            else {
+                // Known discriminator present but not in the frozen 0463 keep/meta table —
+                // still a determined type; treat as meta bookkeeping rather than unknown.
+                role = 'meta';
+                disposition = 'meta';
+                contentText = s(raw.content, raw.text) ?? null;
+            }
+            break;
+    }
+    const messageSplitIndex = entries.length;
+    entries.push({
+        targetTable: 'history_message',
+        record: {
+            session_id: sessionId,
+            seq,
+            role,
+            record_type: recordType,
+            disposition,
+            ts,
+            duration_ms: undefined,
+            model: null,
+            input_tokens: null,
+            output_tokens: null,
+            cache_read_tokens: null,
+            cache_write_tokens: null,
+            cost_usd: null,
+            content_text: contentText,
+            cwd: null,
+            provenance: 'ambient',
+        },
+    });
+    // Tool calls from PLANNER_RESPONSE.tool_calls[]
+    if (recordType === 'PLANNER_RESPONSE' && Array.isArray(raw.tool_calls)) {
+        for (const tc of raw.tool_calls) {
+            entries.push({
+                targetTable: 'history_tool_call',
+                record: {
+                    _messageSplitIndex: messageSplitIndex,
+                    session_id: sessionId,
+                    seq: typeof tc.step_index === 'number' ? tc.step_index : seq,
+                    tool_name: String(tc.name ?? tc.tool_name ?? ''),
+                    args_digest: argsDigest(tc.input ?? tc.arguments),
+                    status: 'ok',
+                    started_at: undefined,
+                    completed_at: undefined,
+                    duration_ms: undefined,
+                    result_bytes: undefined,
+                    error_text: undefined,
+                },
+            });
+        }
+    }
+    return entries;
+}
+// ---------------------------------------------------------------------------
+// Grok mapper
+// ---------------------------------------------------------------------------
+/**
+ * Grok emits two JSONL shapes per session (task 0463):
+ * - `events.jsonl`: flat `{ts, type, ...}`
+ * - `updates.jsonl`: JSON-RPC-like `{timestamp, method, params:{sessionId, update:{sessionUpdate,...}}}`
+ *
+ * Normalize both into a single view before disposition mapping.
+ */
+function normalizeGrokRecord(raw) {
+    const params = raw.params;
+    const update = params?.update;
+    const isUpdatesShape = typeof raw.method === 'string' && params !== undefined && update !== undefined;
+    const body = isUpdatesShape ? update : raw;
+    const recordType = String((isUpdatesShape ? update?.sessionUpdate : raw.type) ?? raw.sessionUpdate ?? '');
+    const sessionId = s(params?.sessionId, raw.session_id, raw.sessionId, raw.params?.sessionId) ?? 'unknown';
+    const ts = normalizeTs(raw.ts ?? raw.timestamp ?? body.ts ?? body.timestamp);
+    const seq = typeof raw.seq === 'number' ? raw.seq : typeof body.seq === 'number' ? body.seq : 0;
+    const meta = (body._meta ?? params?._meta ?? raw._meta);
+    const model = s(meta?.modelId, body.model, raw.model);
+    const contentText = extractGrokContent(body.content ?? body.text ?? raw.content ?? raw.text);
+    const usage = (body.usage ?? raw.usage);
+    const toolMeta = meta?.['x.ai/tool'];
+    const toolName = s(body.title, body.tool_name, body.name, toolMeta?.name, toolMeta?.label, raw.tool_name, raw.name) ?? '';
+    const toolArgs = body.rawInput ?? body.input ?? body.arguments ?? toolMeta?.input ?? raw.input ?? raw.arguments;
+    const toolStatus = s(body.status, raw.outcome, raw.status) ?? 'ok';
+    const durationMs = typeof body.duration_ms === 'number'
+        ? body.duration_ms
+        : typeof raw.duration_ms === 'number'
+            ? raw.duration_ms
+            : typeof usage?.apiDurationMs === 'number'
+                ? usage.apiDurationMs
+                : undefined;
+    const errorText = body.error !== undefined ? String(body.error) : raw.error !== undefined ? String(raw.error) : undefined;
+    const hasTypeDiscriminator = recordType.length > 0 || isUpdatesShape;
+    return {
+        recordType,
+        sessionId,
+        ts,
+        seq,
+        body,
+        model,
+        contentText,
+        usage,
+        toolName,
+        toolArgs,
+        toolStatus,
+        durationMs,
+        errorText,
+        hasTypeDiscriminator,
+    };
+}
+function normalizeTs(value) {
+    if (typeof value === 'number' && Number.isFinite(value)) {
+        // Grok timestamps are unix seconds (sometimes with fractional).
+        const ms = value > 1e12 ? value : value * 1000;
+        return new Date(ms).toISOString();
+    }
+    if (typeof value === 'string' && value.length > 0) {
+        const asNum = Number(value);
+        if (Number.isFinite(asNum) && /^\d+(\.\d+)?$/.test(value.trim())) {
+            const ms = asNum > 1e12 ? asNum : asNum * 1000;
+            return new Date(ms).toISOString();
+        }
+        return value;
+    }
+    return new Date(0).toISOString();
+}
+function extractGrokContent(content) {
+    if (typeof content === 'string')
+        return content;
+    if (content !== null && typeof content === 'object' && !Array.isArray(content)) {
+        const c = content;
+        if (typeof c.text === 'string')
+            return c.text;
+        if (typeof c.content === 'string')
+            return c.content;
+    }
+    if (Array.isArray(content)) {
+        const parts = [];
+        for (const block of content) {
+            if (typeof block === 'string')
+                parts.push(block);
+            else if (block !== null && typeof block === 'object') {
+                const b = block;
+                if (typeof b.text === 'string')
+                    parts.push(b.text);
+            }
+        }
+        return parts.length > 0 ? parts.join('\n') : null;
+    }
+    return null;
+}
+/** Bookkeeping event types from 0463 grok field map → disposition meta. */
+const GROK_META_TYPES = new Set([
+    'phase_changed',
+    'permission_requested',
+    'permission_resolved',
+    'mcp_server_starting',
+    'mcp_server_connected',
+    'mcp_config_resolved',
+    'mcp_managed_config_result',
+    'mcp_transport_decode_error',
+    'mcp_init_completed',
+    'mcp_health_check',
+    'loop_started',
+    'first_token',
+    'agent_thought_chunk',
+    'hook_execution',
+    'plan',
+    'session_recap',
+    'task_backgrounded',
+    'task_completed',
+    'retry_state',
+    'auto_compact_started',
+    'auto_compact_completed',
+    'compaction_checkpoint',
+    'image_compressed',
+    'reasoning',
+    'system',
+    'turn_started',
+    'turn_ended',
+    'yolo_toggled',
+]);
+/** Map a Grok JSONL record into one or more ETL split entries. */
+export function grokSplit(raw) {
+    const entries = [];
+    const n = normalizeGrokRecord(raw);
+    const { recordType, sessionId, ts, seq } = n;
+    if (recordType === 'turn_completed') {
+        const usage = n.usage;
+        const modelUsage = usage?.modelUsage;
+        const modelFromUsage = modelUsage !== null && typeof modelUsage === 'object'
+            ? Object.keys(modelUsage)[0]
+            : undefined;
+        const model = s(n.model, modelFromUsage);
+        const inputTokens = (usage?.inputTokens ?? usage?.input ?? usage?.input_tokens);
+        const outputTokens = (usage?.outputTokens ?? usage?.output ?? usage?.output_tokens);
+        const cacheRead = (usage?.cachedReadTokens ?? usage?.cache_read_tokens);
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'assistant',
+                record_type: recordType,
+                disposition: 'keep',
+                ts,
+                duration_ms: n.durationMs ?? null,
+                model: model ?? null,
+                input_tokens: inputTokens ?? null,
+                output_tokens: outputTokens ?? null,
+                cache_read_tokens: cacheRead ?? null,
+                cache_write_tokens: null,
+                cost_usd: null,
+                content_text: n.contentText,
+                cwd: null,
+                provenance: 'ambient',
+            },
+        });
+        return entries;
+    }
+    if (recordType === 'tool_started') {
+        const messageSplitIndex = 0;
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'assistant',
+                record_type: recordType,
+                disposition: 'keep',
+                ts,
+                duration_ms: undefined,
+                model: n.model ?? null,
+                input_tokens: null,
+                output_tokens: null,
+                cache_read_tokens: null,
+                cache_write_tokens: null,
+                cost_usd: null,
+                content_text: n.contentText,
+                cwd: null,
+                provenance: 'ambient',
+            },
+        });
+        entries.push({
+            targetTable: 'history_tool_call',
+            record: {
+                _messageSplitIndex: messageSplitIndex,
+                session_id: sessionId,
+                seq,
+                tool_name: n.toolName,
+                args_digest: argsDigest(n.toolArgs),
+                status: n.toolStatus || 'ok',
+                started_at: ts,
+                completed_at: undefined,
+                duration_ms: undefined,
+                result_bytes: undefined,
+                error_text: n.errorText,
+            },
+        });
+        return entries;
+    }
+    if (recordType === 'tool_completed') {
+        entries.push({
+            targetTable: 'history_tool_call',
+            record: {
+                _messageSplitIndex: 0,
+                session_id: sessionId,
+                seq,
+                tool_name: n.toolName,
+                args_digest: null,
+                status: n.toolStatus || 'ok',
+                started_at: undefined,
+                completed_at: ts,
+                duration_ms: n.durationMs,
+                result_bytes: undefined,
+                error_text: n.errorText,
+            },
+        });
+        return entries;
+    }
+    if (recordType === 'tool_call' || recordType === 'tool_call_update') {
+        // Pair tool_call with a parent message row so message_hash resolution works.
+        const messageSplitIndex = 0;
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'assistant',
+                record_type: recordType,
+                disposition: 'keep',
+                ts,
+                duration_ms: undefined,
+                model: n.model ?? null,
+                input_tokens: null,
+                output_tokens: null,
+                cache_read_tokens: null,
+                cache_write_tokens: null,
+                cost_usd: null,
+                content_text: n.contentText,
+                cwd: null,
+                provenance: 'ambient',
+            },
+        });
+        entries.push({
+            targetTable: 'history_tool_call',
+            record: {
+                _messageSplitIndex: messageSplitIndex,
+                session_id: sessionId,
+                seq,
+                tool_name: n.toolName,
+                args_digest: argsDigest(n.toolArgs),
+                status: n.toolStatus || 'ok',
+                started_at: recordType === 'tool_call' ? ts : undefined,
+                completed_at: recordType === 'tool_call_update' ? ts : undefined,
+                duration_ms: n.durationMs,
+                result_bytes: undefined,
+                error_text: n.errorText,
+            },
+        });
+        return entries;
+    }
+    if (recordType === 'user' ||
+        recordType === 'user_message' ||
+        recordType === 'user_message_chunk' ||
+        recordType === 'message') {
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'user',
+                record_type: recordType || 'user_message',
+                disposition: 'keep',
+                ts,
+                duration_ms: undefined,
+                model: null,
+                input_tokens: null,
+                output_tokens: null,
+                cache_read_tokens: null,
+                cache_write_tokens: null,
+                cost_usd: null,
+                content_text: n.contentText,
+                cwd: null,
+                provenance: 'ambient',
+            },
+        });
+        return entries;
+    }
+    if (recordType === 'assistant' || recordType === 'agent_message_chunk') {
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'assistant',
+                record_type: recordType,
+                disposition: 'keep',
+                ts,
+                duration_ms: undefined,
+                model: n.model ?? null,
+                input_tokens: null,
+                output_tokens: null,
+                cache_read_tokens: null,
+                cache_write_tokens: null,
+                cost_usd: null,
+                content_text: n.contentText,
+                cwd: null,
+                provenance: 'ambient',
+            },
+        });
+        return entries;
+    }
+    if (GROK_META_TYPES.has(recordType)) {
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'meta',
+                record_type: recordType,
+                disposition: 'meta',
+                ts,
+                duration_ms: n.durationMs ?? null,
+                model: n.model ?? null,
+                input_tokens: null,
+                output_tokens: null,
+                cache_read_tokens: null,
+                cache_write_tokens: null,
+                cost_usd: null,
+                content_text: n.contentText,
+                cwd: null,
+                provenance: 'ambient',
+            },
+        });
+        return entries;
+    }
+    // True unknown: no type discriminator (or an unlisted type with empty discriminator).
+    // Capture loudly with a stable field-shape key (R6).
+    if (!n.hasTypeDiscriminator || recordType.length === 0) {
+        entries.push({
+            targetTable: 'history_message',
+            record: {
+                session_id: sessionId,
+                seq,
+                role: 'unknown',
+                record_type: stableFieldShape(raw),
+                disposition: 'unknown',
+                ts,
+                duration_ms: undefined,
+                model: null,
+                input_tokens: null,
+                output_tokens: null,
+                cache_read_tokens: null,
+                cache_write_tokens: null,
+                cost_usd: null,
+                content_text: n.contentText,
+                cwd: null,
+                provenance: 'ambient',
+            },
+        });
+        return entries;
+    }
+    // Determined but unlisted type — keep as meta bookkeeping, not unknown.
+    entries.push({
+        targetTable: 'history_message',
+        record: {
+            session_id: sessionId,
+            seq,
+            role: 'meta',
+            record_type: recordType,
+            disposition: 'meta',
+            ts,
+            duration_ms: n.durationMs ?? null,
+            model: n.model ?? null,
+            input_tokens: null,
+            output_tokens: null,
+            cache_read_tokens: null,
+            cache_write_tokens: null,
+            cost_usd: null,
+            content_text: n.contentText,
+            cwd: null,
+            provenance: 'ambient',
+        },
+    });
+    return entries;
+}
+// ---------------------------------------------------------------------------
+// Shared utilities
+// ---------------------------------------------------------------------------
+/** Narrow an unknown JSON value to an object bag so nested lookups type-check. */
+function o(value) {
+    return typeof value === 'object' && value !== null ? value : {};
+}
+/** Safe string accessor: returns the first non-empty string value. */
+function s(...values) {
+    for (const value of values) {
+        if (typeof value === 'string' && value.length > 0)
+            return value;
+        if (typeof value === 'number' && Number.isFinite(value))
+            return String(value);
+    }
+    return undefined;
+}
+function mapRole(type) {
+    const t = String(type ?? '').toLowerCase();
+    if (t === 'user' || t === 'human')
+        return 'user';
+    if (t === 'assistant' || t === 'ai' || t === 'model')
+        return 'assistant';
+    if (t === 'system' || t === 'error')
+        return 'system';
+    return t || 'unknown';
+}
+/** Compute cost_usd from token counts and model. Conservative estimate using known rates. */
+function computeCost(inputTokens, outputTokens, model) {
+    if (inputTokens === undefined || inputTokens === null)
+        return undefined;
+    if (outputTokens === undefined || outputTokens === null)
+        return undefined;
+    const inputRate = model?.toLowerCase().includes('claude') ? 3 : 1;
+    const outputRate = model?.toLowerCase().includes('claude') ? 15 : 5;
+    return (inputTokens * inputRate + outputTokens * outputRate) / 1_000_000;
+}
+/** Extract text content from Claude-style content blocks. */
+function extractContentText(content) {
+    if (typeof content === 'string')
+        return content;
+    if (!Array.isArray(content))
+        return undefined;
+    const parts = [];
+    for (const block of content) {
+        if (typeof block !== 'object' || block === null)
+            continue;
+        const b = block;
+        if (b.type === 'text' && typeof b.text === 'string') {
+            parts.push(b.text);
+        }
+    }
+    return parts.length > 0 ? parts.join('\n') : undefined;
+}
+/** Compute a stable field-shape key from a JSON object: sorted, lowercased, +-joined top-level keys. */
+export function stableFieldShape(raw) {
+    return Object.keys(raw)
+        .map((k) => k.toLowerCase())
+        .sort()
+        .join('+');
+}
+// ---------------------------------------------------------------------------
+// Identity field maps for the six source mappers
+// ---------------------------------------------------------------------------
+/** Identity field map for the Claude mapper. */
+export const CLAUDE_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
+/** Identity field map for the Pi mapper. */
+export const PI_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
+/** Identity field map for the OMP mapper. */
+export const OMP_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
+/** Identity field map for the Codex mapper. */
+export const CODEX_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
+/** Identity field map for the AGY mapper. */
+export const AGY_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
+/** Identity field map for the Grok mapper. */
+export const GROK_FIELD_MAP = identityFieldMap(MESSAGE_MAPPER_KEYS.concat(TOOL_CALL_MAPPER_KEYS));
+// ---------------------------------------------------------------------------
+// Zod schemas for the six source mappers
+// ---------------------------------------------------------------------------
+const passthroughSchema = z.object({}).passthrough();
+/** Passthrough Zod schema for the Claude mapper. */
+export const CLAUDE_SCHEMA = passthroughSchema;
+/** Passthrough Zod schema for the Pi mapper. */
+export const PI_SCHEMA = passthroughSchema;
+/** Passthrough Zod schema for the OMP mapper. */
+export const OMP_SCHEMA = passthroughSchema;
+/** Passthrough Zod schema for the Codex mapper. */
+export const CODEX_SCHEMA = passthroughSchema;
+/** Passthrough Zod schema for the AGY mapper. */
+export const AGY_SCHEMA = passthroughSchema;
+/** Passthrough Zod schema for the Grok mapper. */
+export const GROK_SCHEMA = passthroughSchema;