aiex-cli 0.0.4-beta.1 → 0.0.4-beta.3

package/dist/cli.mjs

@@ -1,4 +1,4 @@
-import { A as formatDoctorDiagnosticsJson, C as seedConfig, D as version, E as package_default, S as createConfig, T as name, _ as DEFAULT_MINERU_CONFIG, a as parseJsonSchema, b as PLACEHOLDER_TEXT, c as recognizeImageText, d as t, f as getDefaultAIConfig, g as DEFAULT_MARKITDOWN_CONFIG, h as DEFAULT_MARKER_CONFIG, i as JsonSchemaDefinitionSchema, k as doctorDiagnosticsTableRows, l as shouldUseImageOcrFallback, m as writeAIConfig, n as createMigrationConfig, o as toSnakeCase, p as readAIConfig, s as generateDrizzleSchema, t as collectDoctorDiagnostics, u as initI18n, v as DEFAULT_PROMPT_CONFIG, w as description, x as AIConfigSchema, y as PLACEHOLDER_SCHEMA } from "./doctor-collector-xRnW5Rj3.mjs";
+import { A as formatDoctorDiagnosticsJson, C as seedConfig, D as version, E as package_default, S as createConfig, T as name, _ as DEFAULT_MINERU_CONFIG, a as parseJsonSchema, b as PLACEHOLDER_TEXT, c as recognizeImageText, d as t, f as getDefaultAIConfig, g as DEFAULT_MARKITDOWN_CONFIG, h as DEFAULT_MARKER_CONFIG, i as JsonSchemaDefinitionSchema, k as doctorDiagnosticsTableRows, l as shouldUseImageOcrFallback, m as writeAIConfig, n as createMigrationConfig, o as toSnakeCase, p as readAIConfig, s as generateDrizzleSchema, t as collectDoctorDiagnostics, u as initI18n, v as DEFAULT_PROMPT_CONFIG, w as description, x as AIConfigSchema, y as PLACEHOLDER_SCHEMA } from "./doctor-collector-8fLyh9lK.mjs";
 import { createRequire } from "node:module";
 import fs from "node:fs/promises";
 import os from "node:os";
@@ -15,20 +15,20 @@ import fs$1 from "node:fs";
 import { intro, isCancel, outro, select, spinner, text } from "@clack/prompts";
 import Database from "better-sqlite3";
 import pc from "picocolors";
+import { Buffer } from "node:buffer";
 import * as XLSX from "xlsx";
-import { glob, globSync } from "tinyglobby";
 import { createOpenAICompatible } from "@ai-sdk/openai-compatible";
-import { LangfuseSpanProcessor } from "@langfuse/otel";
-import { NodeTracerProvider } from "@opentelemetry/sdk-trace-node";
 import { APICallError, Output, generateText, jsonSchema } from "ai";
-import mime from "mime";
 import pRetry from "p-retry";
+import mime from "mime";
 import { jsonrepair } from "jsonrepair";
+import { LangfuseSpanProcessor } from "@langfuse/otel";
+import { NodeTracerProvider } from "@opentelemetry/sdk-trace-node";
+import crypto from "node:crypto";
 import { Client, extractNotionId } from "@notionhq/client";
-import { Buffer } from "node:buffer";
 import { execa } from "execa";
+import { glob, globSync } from "tinyglobby";
 import { extractText, getDocumentProxy, getMeta } from "unpdf";
-import crypto from "node:crypto";
 import { execFile } from "node:child_process";
 import { promisify } from "node:util";
 import * as chokidar from "chokidar";
@@ -215,6 +215,50 @@ function failCommand(message) {
 	process.exitCode = 1;
 }
 
+//#endregion
+//#region src/core/export-manager.ts
+function formatRowsConformingToSchema(rows, columns, schema, format) {
+	return rows.map((row) => {
+		const newRow = {};
+		columns.forEach((col) => {
+			const colName = col.name;
+			const val = row[colName];
+			const type = (schema?.properties?.[colName])?.type || "";
+			if (val === null || val === void 0) newRow[colName] = "";
+			else if (type === "boolean") if (format === "xlsx") newRow[colName] = val === 1 || val === "1" || val === true;
+			else newRow[colName] = val === 1 || val === "1" || val === true ? "true" : "false";
+			else if (type === "number" || type === "integer") if (val === "") newRow[colName] = "";
+			else {
+				const num = Number(val);
+				newRow[colName] = Number.isNaN(num) ? val : num;
+			}
+			else if (typeof val === "object") newRow[colName] = JSON.stringify(val);
+			else {
+				const dbType = (col.type || "").toLowerCase();
+				if ((dbType.includes("int") || dbType.includes("real") || dbType.includes("num") || dbType.includes("double") || dbType.includes("float")) && typeof val === "string" && val !== "") {
+					const num = Number(val);
+					newRow[colName] = Number.isNaN(num) ? val : num;
+				} else newRow[colName] = val;
+			}
+		});
+		return newRow;
+	});
+}
+function generateExportBuffer(tableName, formattedRows, columns, format) {
+	const ws = XLSX.utils.json_to_sheet(formattedRows, { header: columns.map((col) => col.name) });
+	if (format === "xlsx") {
+		const wb = XLSX.utils.book_new();
+		XLSX.utils.book_append_sheet(wb, ws, tableName.slice(0, 31));
+		return XLSX.write(wb, {
+			bookType: "xlsx",
+			type: "buffer"
+		});
+	} else {
+		const csv = XLSX.utils.sheet_to_csv(ws);
+		return Buffer.from("" + csv, "utf8");
+	}
+}
+
 //#endregion
 //#region src/core/ai-extraction/model-capabilities.json
 var model_capabilities_default = {
@@ -12814,6 +12858,28 @@ async function withRetry(fn, onRetry, maxRetries = 5) {
 	});
 }
 
+//#endregion
+//#region src/core/ai-extraction/file-utils.ts
+function detectMimeType(filePath) {
+	return mime.getType(filePath) ?? "application/octet-stream";
+}
+async function readFilePart(filePath) {
+	const mimeStr = detectMimeType(filePath);
+	const buffer = await fs.readFile(filePath);
+	const name$1 = path.basename(filePath);
+	if (mimeStr.startsWith("image/")) return {
+		type: "image",
+		image: buffer,
+		mimeType: mimeStr
+	};
+	return {
+		type: "file",
+		data: buffer,
+		mediaType: mimeStr,
+		filename: name$1
+	};
+}
+
 //#endregion
 //#region src/core/ai-extraction/json-utils.ts
 function parseJsonLike(text$1) {
@@ -12993,7 +13059,34 @@ function generatePromptSnapshot(schema, promptConfig = DEFAULT_PROMPT_CONFIG) {
 }
 
 //#endregion
-//#region src/core/ai-extraction/extractor.ts
+//#region src/core/ai-extraction/snapshot.ts
+const SYSTEM_PROMPT_REGEX = /## System Prompt\n([\s\S]*?)(?=## User Prompt|$)/;
+const USER_PROMPT_REGEX = /## User Prompt Template\n([\s\S]*)$/;
+async function loadPromptSnapshot(aiexDir, tableName) {
+	const snapshotPath = path.join(aiexDir, "extracted", `${tableName}.prompt.md`);
+	try {
+		const content = await fs.readFile(snapshotPath, "utf-8");
+		const systemMatch = content.match(SYSTEM_PROMPT_REGEX);
+		const userMatch = content.match(USER_PROMPT_REGEX);
+		if (systemMatch && userMatch) return {
+			system: systemMatch[1].trim(),
+			user: userMatch[1].trim()
+		};
+	} catch {}
+	return null;
+}
+async function savePromptSnapshot(schema, aiexDir) {
+	const content = generatePromptSnapshot(schema, (await readAIConfig(aiexDir))?.prompt ?? DEFAULT_PROMPT_CONFIG);
+	const outputDir = path.join(aiexDir, "extracted");
+	await fs.mkdir(outputDir, { recursive: true });
+	const fileName = `${schema.table.name}.prompt.md`;
+	const outputPath = path.join(outputDir, fileName);
+	await fs.writeFile(outputPath, content);
+	return outputPath;
+}
+
+//#endregion
+//#region src/core/ai-extraction/telemetry.ts
 let langfuseInitialized = false;
 function initLangfuse(config) {
 	if (!config.langfuse?.publicKey || !config.langfuse.secretKey) return;
@@ -13010,28 +13103,9 @@ function initLangfuse(config) {
 		console.warn("[Langfuse] Failed to initialize tracing:", e instanceof Error ? e.message : e);
 	}
 }
-const SYSTEM_PROMPT_REGEX = /## System Prompt\n([\s\S]*?)(?=## User Prompt|$)/;
-const USER_PROMPT_REGEX = /## User Prompt Template\n([\s\S]*)$/;
-const OPENAI_COMPATIBLE_PROVIDER_NAME = "openai-compatible";
-function detectMimeType(filePath) {
-	return mime.getType(filePath) ?? "application/octet-stream";
-}
-async function readFilePart(filePath) {
-	const mime$1 = detectMimeType(filePath);
-	const buffer = await fs.readFile(filePath);
-	const name$1 = path.basename(filePath);
-	if (mime$1.startsWith("image/")) return {
-		type: "image",
-		image: buffer,
-		mimeType: mime$1
-	};
-	return {
-		type: "file",
-		data: buffer,
-		mediaType: mime$1,
-		filename: name$1
-	};
-}
+
+//#endregion
+//#region src/core/ai-extraction/validator.ts
 function nullableType(type) {
 	return type === "null" ? ["null"] : [type, "null"];
 }
@@ -13126,19 +13200,10 @@ function validateExtractedData(schema, data) {
 	};
 	return { success: true };
 }
-async function loadPromptSnapshot(aiexDir, tableName) {
-	const snapshotPath = path.join(aiexDir, "extracted", `${tableName}.prompt.md`);
-	try {
-		const content = await fs.readFile(snapshotPath, "utf-8");
-		const systemMatch = content.match(SYSTEM_PROMPT_REGEX);
-		const userMatch = content.match(USER_PROMPT_REGEX);
-		if (systemMatch && userMatch) return {
-			system: systemMatch[1].trim(),
-			user: userMatch[1].trim()
-		};
-	} catch {}
-	return null;
-}
+
+//#endregion
+//#region src/core/ai-extraction/extractor.ts
+const OPENAI_COMPATIBLE_PROVIDER_NAME = "openai-compatible";
 async function extractStructuredData(input) {
 	const { config, schema, text: text$1, aiexDir, file, modelOverride } = input;
 	if (!config.provider.apiKey) return {
@@ -13188,66 +13253,118 @@ async function extractStructuredData(input) {
 			user = generated.user;
 		}
 		const outputSchema = jsonSchema(schemaToExtractionOutputSchema(schema));
-		let result;
 		const timeoutMs = (config.provider.timeout ?? 300) * 1e3;
-		if (useFileContent) {
-			const filePart = await readFilePart(file);
-			const fileName = filePart.type === "file" ? filePart.filename : path.basename(file);
-			const contentParts = [{
-				type: "text",
-				text: user.includes(PLACEHOLDER_TEXT) ? user.replaceAll(PLACEHOLDER_TEXT, text$1 || `Data is contained in the attached file: ${fileName}`) : user
-			}, filePart];
-			const fileOpts = {
-				model: provider.chatModel(selected.name),
-				system,
-				messages: [{
-					role: "user",
-					content: contentParts
-				}],
-				abortSignal: AbortSignal.timeout(timeoutMs),
-				maxRetries: 0,
-				experimental_telemetry: { isEnabled: useTelemetry }
-			};
-			if (useStructuredOutput) fileOpts.output = Output.object({ schema: outputSchema });
-			result = await withRetry(() => generateText(fileOpts), input.onRetry);
-		} else {
-			const textOpts = {
-				model: provider.chatModel(selected.name),
-				system,
-				prompt: user,
-				abortSignal: AbortSignal.timeout(timeoutMs),
-				maxRetries: 0,
-				experimental_telemetry: { isEnabled: useTelemetry }
-			};
-			if (useStructuredOutput) textOpts.output = Output.object({ schema: outputSchema });
-			result = await withRetry(() => generateText(textOpts), input.onRetry);
+		let systemPrompt = system;
+		let userPrompt = user;
+		const maxAttempts = 3;
+		let lastError = "";
+		let totalPromptTokens = 0;
+		let totalCompletionTokens = 0;
+		for (let attempt = 1; attempt <= maxAttempts; attempt++) {
+			let result = null;
+			let data;
+			let parseError;
+			let validationError;
+			try {
+				if (useFileContent) {
+					const filePart = await readFilePart(file);
+					const fileName = filePart.type === "file" ? filePart.filename : path.basename(file);
+					const contentParts = [{
+						type: "text",
+						text: userPrompt.includes(PLACEHOLDER_TEXT) ? userPrompt.replaceAll(PLACEHOLDER_TEXT, text$1 || `Data is contained in the attached file: ${fileName}`) : userPrompt
+					}, filePart];
+					const fileOpts = {
+						model: provider.chatModel(selected.name),
+						system: systemPrompt,
+						messages: [{
+							role: "user",
+							content: contentParts
+						}],
+						abortSignal: AbortSignal.timeout(timeoutMs),
+						maxRetries: 0,
+						experimental_telemetry: { isEnabled: useTelemetry }
+					};
+					if (useStructuredOutput) fileOpts.output = Output.object({ schema: outputSchema });
+					result = await withRetry(() => generateText(fileOpts), input.onRetry);
+				} else {
+					const textOpts = {
+						model: provider.chatModel(selected.name),
+						system: systemPrompt,
+						prompt: userPrompt,
+						abortSignal: AbortSignal.timeout(timeoutMs),
+						maxRetries: 0,
+						experimental_telemetry: { isEnabled: useTelemetry }
+					};
+					if (useStructuredOutput) textOpts.output = Output.object({ schema: outputSchema });
+					result = await withRetry(() => generateText(textOpts), input.onRetry);
+				}
+				if (result.usage) {
+					totalPromptTokens += result.usage.inputTokens ?? 0;
+					totalCompletionTokens += result.usage.outputTokens ?? 0;
+				}
+				if (useStructuredOutput) data = result.output;
+				else try {
+					data = safeParseJSON(result.text);
+				} catch (e) {
+					parseError = e instanceof Error ? e.message : String(e);
+				}
+			} catch (error) {
+				parseError = getErrorMessage(error);
+			}
+			if (!parseError && data !== void 0) {
+				const validation = validateExtractedData(schema, data);
+				if (validation.success) {
+					const outputDir = path.resolve(aiexDir, config.extraction.outputDir.replace(".aiex/", ""));
+					await fs.mkdir(outputDir, { recursive: true });
+					const timestamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-");
+					const outputFileName = `${schema.table.name}-${timestamp}.json`;
+					const outputPath = path.join(outputDir, outputFileName);
+					await writeFile(outputPath, data, {
+						spaces: 2,
+						EOL: "\n"
+					});
+					return {
+						success: true,
+						outputPath,
+						data,
+						tokensUsed: {
+							prompt: totalPromptTokens,
+							completion: totalCompletionTokens,
+							total: totalPromptTokens + totalCompletionTokens
+						}
+					};
+				} else validationError = validation.error;
+			}
+			const errorMsg = parseError || validationError || "Unknown validation error";
+			lastError = errorMsg;
+			if (attempt < maxAttempts) {
+				const invalidJson = data !== void 0 ? JSON.stringify(data, null, 2) : result ? result.text : "";
+				systemPrompt = `You are a precise data correction assistant. Your task is to correct validation errors in a previously generated JSON object to make it comply with the provided JSON Schema.
+        
+CRITICAL RULES:
+1. Only correct the fields that failed validation.
+2. Preserve all other correctly extracted fields and their values exactly.
+3. Return ONLY the corrected JSON object. No explanations, no markdown blocks other than JSON.`;
+				userPrompt = `The JSON data you generated previously failed validation. Please correct it.
+
+[Original Text]
+${text$1 || "Data is contained in the attached file."}
+
+[JSON Schema Definition]
+${JSON.stringify(schemaToExtractionOutputSchema(schema), null, 2)}
+
+[Previously Generated Invalid JSON]
+${invalidJson}
+
+[Validation Error Details]
+${errorMsg}
+
+Please output the corrected JSON object now:`;
+			}
 		}
-		let data;
-		if (useStructuredOutput) data = result.output;
-		else data = safeParseJSON(result.text);
-		const validation = validateExtractedData(schema, data);
-		if (!validation.success) return {
-			success: false,
-			error: validation.error
-		};
-		const outputDir = path.resolve(aiexDir, config.extraction.outputDir.replace(".aiex/", ""));
-		await fs.mkdir(outputDir, { recursive: true });
-		const timestamp = (/* @__PURE__ */ new Date()).toISOString().replace(/[:.]/g, "-");
-		const outputFileName = `${schema.table.name}-${timestamp}.json`;
-		const outputPath = path.join(outputDir, outputFileName);
-		await writeFile(outputPath, data, {
-			spaces: 2,
-			EOL: "\n"
-		});
 		return {
-			success: true,
-			outputPath,
-			data,
-			tokensUsed: result.usage ? {
-				prompt: result.usage.inputTokens ?? 0,
-				completion: result.usage.outputTokens ?? 0,
-				total: (result.usage.inputTokens ?? 0) + (result.usage.outputTokens ?? 0)
-			} : void 0
+			success: false,
+			error: lastError || "Extraction failed after self-reflection retries"
 		};
 	} catch (error) {
 		return {
@@ -13394,18 +13511,6 @@ function insertExtractedData(db, schema, data) {
 	}
 }
 
-//#endregion
-//#region src/core/ai-extraction/snapshot.ts
-async function savePromptSnapshot(schema, aiexDir) {
-	const content = generatePromptSnapshot(schema, (await readAIConfig(aiexDir))?.prompt ?? DEFAULT_PROMPT_CONFIG);
-	const outputDir = path.join(aiexDir, "extracted");
-	await fs.mkdir(outputDir, { recursive: true });
-	const fileName = `${schema.table.name}.prompt.md`;
-	const outputPath = path.join(outputDir, fileName);
-	await fs.writeFile(outputPath, content);
-	return outputPath;
-}
-
 //#endregion
 //#region src/core/extraction-audit.ts
 const AUDIT_ID_RE = /^[\w.-]+$/;
@@ -13542,78 +13647,19 @@ async function findSucceededAuditByHash(aiexDir, schemaName, fileHash) {
 }
 
 //#endregion
-//#region src/core/file-constants.ts
-const MAX_UPLOAD_SIZE = 30 * 1024 * 1024;
-const MAX_UPLOAD_SIZE_TEXT = "30MB";
-const SUPPORTED_FILE_TYPES_TEXT = "images, PDF, text, markdown, CSV, JSON, HTML, XML, YAML";
-const MISSING_UPLOAD_FILE_TEXT = t("errors.file.missingUpload");
-const SUPPORTED_MIME_TYPES = new Set([
-	"image/png",
-	"image/jpeg",
-	"image/gif",
-	"image/webp",
-	"image/bmp",
-	"image/svg+xml",
-	"application/pdf",
-	"text/plain",
-	"text/markdown",
-	"text/csv",
-	"application/json",
-	"text/html",
-	"text/xml",
-	"application/x-yaml",
-	"text/yaml"
-]);
-const MIME_TO_EXT = {
-	"image/png": "png",
-	"image/jpeg": "jpg",
-	"image/gif": "gif",
-	"image/webp": "webp",
-	"image/bmp": "bmp",
-	"image/svg+xml": "svg",
-	"application/pdf": "pdf",
-	"text/plain": "txt",
-	"text/markdown": "md",
-	"text/csv": "csv",
-	"application/json": "json",
-	"text/html": "html",
-	"text/xml": "xml",
-	"application/x-yaml": "yaml",
-	"text/yaml": "yaml"
-};
-function bytesToMB(bytes) {
-	return bytes / (1024 * 1024);
-}
-function getExtensionFromMime(mimeType) {
-	return MIME_TO_EXT[mimeType];
-}
-function isAllowedMimeType(mimeType) {
-	return SUPPORTED_MIME_TYPES.has(mimeType);
-}
-function unsupportedFileTypeMessage(mimeType) {
-	return t("errors.file.unsupportedType", {
-		type: mimeType,
-		supported: SUPPORTED_FILE_TYPES_TEXT
+//#region src/utils/hash.ts
+/**
+* Helper to compute SHA-256 hash of a file asynchronously.
+*/
+function getFileHash(filePath) {
+	return new Promise((resolve, reject) => {
+		const hash = crypto.createHash("sha256");
+		const stream = fs$1.createReadStream(filePath);
+		stream.on("data", (data) => hash.update(data));
+		stream.on("end", () => resolve(hash.digest("hex")));
+		stream.on("error", (err) => reject(err));
 	});
 }
-function isMissingUploadFileError(error) {
-	return !!error && typeof error === "object" && error.code === "ENOENT";
-}
-var FileValidationError = class extends Error {
-	constructor(message) {
-		super(message);
-		this.name = "FileValidationError";
-	}
-};
-function validateFileUpload(file) {
-	if (file.size === 0) throw new FileValidationError(t("errors.file.empty"));
-	if (file.size > MAX_UPLOAD_SIZE) throw new FileValidationError(t("errors.file.sizeExceeded", {
-		size: bytesToMB(file.size).toFixed(1),
-		limit: MAX_UPLOAD_SIZE_TEXT,
-		file: file.name
-	}));
-	if (!isAllowedMimeType(file.type)) throw new FileValidationError(unsupportedFileTypeMessage(file.type));
-}
 
 //#endregion
 //#region src/core/notion-sink.ts
@@ -13847,20 +13893,148 @@ async function writeNotionPage(config, schemaName, data) {
 }
 
 //#endregion
-//#region src/core/pdf-converter/external.ts
-function applyTemplate(value, context) {
-	return value.replaceAll("{input}", context.input).replaceAll("{outputDir}", context.outputDir).replaceAll("{basename}", context.basename);
+//#region src/core/webhook-sink.ts
+async function sendWebhook(config, payload) {
+	if (!config || !config.enabled || !config.url) return;
+	const body = JSON.stringify(payload);
+	const headers = {
+		"Content-Type": "application/json",
+		"User-Agent": "aiex-webhook-dispatcher"
+	};
+	if (config.secret) headers["X-Aiex-Signature"] = `sha256=${crypto.createHmac("sha256", config.secret).update(body).digest("hex")}`;
+	const response = await fetch(config.url, {
+		method: "POST",
+		headers,
+		body
+	});
+	if (!response.ok) throw new Error(`Webhook request failed with status: ${response.status} ${response.statusText}`);
 }
-function isError(error) {
-	return error instanceof Error;
+
+//#endregion
+//#region src/core/integration/dispatcher.ts
+async function syncResultToNotion(aiConfig, schemaName, data) {
+	if (!data || typeof data !== "object" || Array.isArray(data)) throw new Error(t("errors.ai.extractionNotObject"));
+	const page = await writeNotionPage(aiConfig.notion, schemaName, data);
+	return [{
+		databaseId: page.databaseId,
+		pageId: page.pageId
+	}];
 }
-async function pathExists(filePath) {
-	try {
-		await fs.access(filePath);
-		return true;
-	} catch {
-		return false;
-	}
+function shouldSyncNotion(aiConfig, schemaName) {
+	return !!aiConfig.notion?.enabled && !!aiConfig.notion.schemas?.[schemaName]?.databaseId?.trim();
+}
+async function triggerWebhook(aiConfig, auditId, schemaName, event, source, data, error, tokensUsed, quiet = false) {
+	if (!aiConfig.webhook?.enabled) return;
+	try {
+		await sendWebhook(aiConfig.webhook, {
+			event,
+			schemaName,
+			auditId,
+			timestamp: (/* @__PURE__ */ new Date()).toISOString(),
+			source: {
+				type: source.type,
+				fileName: source.filePath ? path.basename(source.filePath) : void 0,
+				filePath: source.filePath
+			},
+			data,
+			error,
+			tokensUsed
+		});
+		if (!quiet) consola.success(t("extract.file.webhookSynced"));
+	} catch (err) {
+		if (!quiet) consola.error(t("extract.file.webhookSyncFail", { error: err instanceof Error ? err.message : String(err) }));
+	}
+}
+
+//#endregion
+//#region src/core/file-constants.ts
+const MAX_UPLOAD_SIZE = 30 * 1024 * 1024;
+const MAX_UPLOAD_SIZE_TEXT = "30MB";
+const SUPPORTED_FILE_TYPES_TEXT = "images, PDF, text, markdown, CSV, JSON, HTML, XML, YAML";
+const MISSING_UPLOAD_FILE_TEXT = t("errors.file.missingUpload");
+const SUPPORTED_MIME_TYPES = new Set([
+	"image/png",
+	"image/jpeg",
+	"image/gif",
+	"image/webp",
+	"image/bmp",
+	"image/svg+xml",
+	"application/pdf",
+	"text/plain",
+	"text/markdown",
+	"text/csv",
+	"application/json",
+	"text/html",
+	"text/xml",
+	"application/x-yaml",
+	"text/yaml"
+]);
+const MIME_TO_EXT = {
+	"image/png": "png",
+	"image/jpeg": "jpg",
+	"image/gif": "gif",
+	"image/webp": "webp",
+	"image/bmp": "bmp",
+	"image/svg+xml": "svg",
+	"application/pdf": "pdf",
+	"text/plain": "txt",
+	"text/markdown": "md",
+	"text/csv": "csv",
+	"application/json": "json",
+	"text/html": "html",
+	"text/xml": "xml",
+	"application/x-yaml": "yaml",
+	"text/yaml": "yaml"
+};
+function bytesToMB(bytes) {
+	return bytes / (1024 * 1024);
+}
+function getExtensionFromMime(mimeType) {
+	return MIME_TO_EXT[mimeType];
+}
+function isAllowedMimeType(mimeType) {
+	return SUPPORTED_MIME_TYPES.has(mimeType);
+}
+function unsupportedFileTypeMessage(mimeType) {
+	return t("errors.file.unsupportedType", {
+		type: mimeType,
+		supported: SUPPORTED_FILE_TYPES_TEXT
+	});
+}
+function isMissingUploadFileError(error) {
+	return !!error && typeof error === "object" && error.code === "ENOENT";
+}
+var FileValidationError = class extends Error {
+	constructor(message) {
+		super(message);
+		this.name = "FileValidationError";
+	}
+};
+function validateFileUpload(file) {
+	if (file.size === 0) throw new FileValidationError(t("errors.file.empty"));
+	if (file.size > MAX_UPLOAD_SIZE) throw new FileValidationError(t("errors.file.sizeExceeded", {
+		size: bytesToMB(file.size).toFixed(1),
+		limit: MAX_UPLOAD_SIZE_TEXT,
+		file: file.name
+	}));
+	if (!isAllowedMimeType(file.type)) throw new FileValidationError(unsupportedFileTypeMessage(file.type));
+}
+
+//#endregion
+//#region src/core/pdf-converter/external.ts
+function applyTemplate(value, context) {
+	return value.replaceAll("{input}", context.input).replaceAll("{outputDir}", context.outputDir).replaceAll("{basename}", context.basename);
+}
+function isError(error) {
+	return error instanceof Error;
+}
+async function pathExists(filePath) {
+	try {
+		await fs.access(filePath);
+		return true;
+	} catch {
+		return false;
+	}
 }
 async function collectMarkdownFiles(dir) {
 	return (await glob("**/*.md", {
@@ -14022,22 +14196,7 @@ function createPdfConverter(config) {
 }
 
 //#endregion
-//#region src/utils/hash.ts
-/**
-* Helper to compute SHA-256 hash of a file asynchronously.
-*/
-function getFileHash(filePath) {
-	return new Promise((resolve, reject) => {
-		const hash = crypto.createHash("sha256");
-		const stream = fs$1.createReadStream(filePath);
-		stream.on("data", (data) => hash.update(data));
-		stream.on("end", () => resolve(hash.digest("hex")));
-		stream.on("error", (err) => reject(err));
-	});
-}
-
-//#endregion
-//#region src/core/extract-runner.ts
+//#region src/core/pdf-converter/orchestrator.ts
 const FILE_PART_EXTENSIONS = new Set([
 	"png",
 	"jpg",
@@ -14047,6 +14206,51 @@ const FILE_PART_EXTENSIONS = new Set([
 	"bmp",
 	"svg"
 ]);
+const PDF_EXT_RE = /\.pdf$/i;
+async function readExtractFileInput(filePath, aiConfig, modelOverride) {
+	const stat = fs$1.statSync(filePath);
+	if (stat.size > MAX_UPLOAD_SIZE) throw new Error(t("errors.file.sizeExceeded", {
+		size: bytesToMB(stat.size).toFixed(1),
+		limit: MAX_UPLOAD_SIZE_TEXT,
+		file: filePath
+	}));
+	const ext = path.extname(filePath).toLowerCase().replace(".", "");
+	if (FILE_PART_EXTENSIONS.has(ext)) {
+		if (shouldUseImageOcrFallback(aiConfig, modelOverride)) {
+			const result = await recognizeImageText(filePath, aiConfig?.image);
+			consola.info(t("extract.file.ocrText", { confidence: (result.confidence * 100).toFixed(1) }));
+			return { text: result.text };
+		}
+		return {
+			text: "",
+			filePath
+		};
+	}
+	if (ext === "pdf") {
+		const buffer = await fs.readFile(filePath);
+		const converter = createPdfConverter(aiConfig?.pdf);
+		const result = await converter.convert(buffer, filePath);
+		if (result.metadata?.fallback === "true") consola.info(t("extract.file.pdfFallback", { count: result.pageCount }));
+		else consola.info(t("extract.file.pdfConverted", {
+			name: converter.name,
+			count: result.pageCount
+		}));
+		const mdPath = filePath.replace(PDF_EXT_RE, ".md");
+		try {
+			await fs.writeFile(mdPath, result.text);
+			consola.info(t("extract.file.markdownSaved", { path: mdPath }));
+		} catch {
+			const fallbackMd = path.join(os.tmpdir(), `${path.basename(filePath, ".pdf")}.md`);
+			await fs.writeFile(fallbackMd, result.text);
+			consola.info(t("extract.file.markdownSaved", { path: fallbackMd }));
+		}
+		return { text: result.text };
+	}
+	return { text: await fs.readFile(filePath, "utf-8") };
+}
+
+//#endregion
+//#region src/core/batch/batch-processor.ts
 const SUPPORTED_EXTENSIONS$1 = new Set([
 	...FILE_PART_EXTENSIONS,
 	"pdf",
@@ -14059,20 +14263,89 @@ const SUPPORTED_EXTENSIONS$1 = new Set([
 	"yaml",
 	"yml"
 ]);
-const PDF_EXT_RE = /\.pdf$/i;
-const JSON_EXT_RE$1 = /\.json$/;
 const SUPPORTED_FILE_PATTERN = `*.{${[...SUPPORTED_EXTENSIONS$1].join(",")}}`;
-async function syncResultToNotion(aiConfig, schemaName, data) {
-	if (!data || typeof data !== "object" || Array.isArray(data)) throw new Error(t("errors.ai.extractionNotObject"));
-	const page = await writeNotionPage(aiConfig.notion, schemaName, data);
-	return [{
-		databaseId: page.databaseId,
-		pageId: page.pageId
-	}];
+function listSupportedFiles(dir, pattern) {
+	if (!fs$1.statSync(dir).isDirectory()) throw new Error(t("errors.file.notADirectory", { dir }));
+	return globSync(pattern ?? SUPPORTED_FILE_PATTERN, {
+		cwd: dir,
+		absolute: true,
+		onlyFiles: true
+	}).filter((file) => {
+		const ext = path.extname(file).toLowerCase().replace(".", "");
+		return SUPPORTED_EXTENSIONS$1.has(ext);
+	}).sort();
 }
-function shouldSyncNotion(aiConfig, schemaName) {
-	return !!aiConfig.notion?.enabled && !!aiConfig.notion.schemas?.[schemaName]?.databaseId?.trim();
+async function processOneFile(aiexDir, config, aiConfig, schemaName, filePath, modelOverride, options) {
+	const result = await runAuditedExtraction({
+		aiexDir,
+		config,
+		aiConfig,
+		schemaName,
+		source: {
+			type: "file",
+			filePath
+		},
+		modelOverride,
+		insert: options?.insert,
+		force: options?.force,
+		quiet: false
+	});
+	if (result.success) {
+		if (!result.skipped) consola.success(t("extract.file.processSuccess", { file: path.basename(filePath) }));
+		return true;
+	}
+	return false;
+}
+async function runBatchExtraction(aiexDir, config, aiConfig, schemaName, dir, globPattern, modelOverride, options) {
+	consola.info(t("extract.batch.scanning", { dir: pc.cyan(dir) }));
+	let files;
+	try {
+		files = listSupportedFiles(dir, globPattern);
+	} catch {
+		return {
+			ok: false,
+			successCount: 0,
+			failCount: 0,
+			error: t("extract.batch.errors.cannotReadDir", { dir })
+		};
+	}
+	if (files.length === 0) return {
+		ok: false,
+		successCount: 0,
+		failCount: 0,
+		error: t("extract.batch.errors.noSupportedFiles", { dir })
+	};
+	consola.info(t("extract.batch.found", { count: files.length }));
+	let successCount = 0;
+	let failCount = 0;
+	for (let i = 0; i < files.length; i++) {
+		const file = files[i];
+		consola.info(`\n${t("extract.batch.processing", {
+			current: i + 1,
+			total: files.length,
+			file: pc.cyan(path.basename(file))
+		})}`);
+		if (await processOneFile(aiexDir, config, aiConfig, schemaName, file, modelOverride, {
+			insert: options?.insert,
+			force: options?.force
+		})) successCount++;
+		else failCount++;
+	}
+	consola.info(`\n${t("extract.batch.complete", {
+		success: pc.green(successCount),
+		fail: pc.red(failCount),
+		total: files.length
+	})}`);
+	return {
+		ok: true,
+		successCount,
+		failCount
+	};
 }
+
+//#endregion
+//#region src/core/extract-runner.ts
+const JSON_EXT_RE$1 = /\.json$/;
 async function ensureDatabaseReady(dbPath, schema) {
 	try {
 		await fs.access(dbPath);
@@ -14098,17 +14371,6 @@ async function ensureDatabaseReady(dbPath, schema) {
 	}
 	return null;
 }
-function listSupportedFiles(dir, pattern) {
-	if (!fs$1.statSync(dir).isDirectory()) throw new Error(t("errors.file.notADirectory", { dir }));
-	return globSync(pattern ?? SUPPORTED_FILE_PATTERN, {
-		cwd: dir,
-		absolute: true,
-		onlyFiles: true
-	}).filter((file) => {
-		const ext = path.extname(file).toLowerCase().replace(".", "");
-		return SUPPORTED_EXTENSIONS$1.has(ext);
-	}).sort();
-}
 async function loadSchema(config, schemaName) {
 	const schemaPath = path.join(config.schemaPath, `${schemaName}.json`);
 	try {
@@ -14122,68 +14384,27 @@ async function loadSchema(config, schemaName) {
 				issues: e.issues.map((i) => `  - ${i.path.join(".")}: ${i.message}`).join("\n")
 			})
 		};
-		if (e.code === "ENOENT") return {
-			schema: null,
-			error: t("errors.schema.cannotRead", { name: `${schemaName}.json` })
-		};
-		if (e instanceof SyntaxError) return {
-			schema: null,
-			error: t("errors.schema.invalidJson", { name: `${schemaName}.json` })
-		};
-		return {
-			schema: null,
-			error: String(e)
-		};
-	}
-}
-async function listSchemas(aiexDir) {
-	try {
-		const dir = path.join(aiexDir, "schema");
-		return (await fs.readdir(dir)).filter((f) => f.endsWith(".json")).map((f) => f.replace(JSON_EXT_RE$1, "")).sort();
-	} catch {
-		return [];
-	}
-}
-async function readExtractFileInput(filePath, aiConfig, modelOverride) {
-	const stat = fs$1.statSync(filePath);
-	if (stat.size > MAX_UPLOAD_SIZE) throw new Error(t("errors.file.sizeExceeded", {
-		size: bytesToMB(stat.size).toFixed(1),
-		limit: MAX_UPLOAD_SIZE_TEXT,
-		file: filePath
-	}));
-	const ext = path.extname(filePath).toLowerCase().replace(".", "");
-	if (FILE_PART_EXTENSIONS.has(ext)) {
-		if (shouldUseImageOcrFallback(aiConfig, modelOverride)) {
-			const result = await recognizeImageText(filePath, aiConfig?.image);
-			consola.info(t("extract.file.ocrText", { confidence: (result.confidence * 100).toFixed(1) }));
-			return { text: result.text };
-		}
-		return {
-			text: "",
-			filePath
-		};
-	}
-	if (ext === "pdf") {
-		const buffer = await fs.readFile(filePath);
-		const converter = createPdfConverter(aiConfig?.pdf);
-		const result = await converter.convert(buffer, filePath);
-		if (result.metadata?.fallback === "true") consola.info(t("extract.file.pdfFallback", { count: result.pageCount }));
-		else consola.info(t("extract.file.pdfConverted", {
-			name: converter.name,
-			count: result.pageCount
-		}));
-		const mdPath = filePath.replace(PDF_EXT_RE, ".md");
-		try {
-			await fs.writeFile(mdPath, result.text);
-			consola.info(t("extract.file.markdownSaved", { path: mdPath }));
-		} catch {
-			const fallbackMd = path.join(os.tmpdir(), `${path.basename(filePath, ".pdf")}.md`);
-			await fs.writeFile(fallbackMd, result.text);
-			consola.info(t("extract.file.markdownSaved", { path: fallbackMd }));
-		}
-		return { text: result.text };
+		if (e.code === "ENOENT") return {
+			schema: null,
+			error: t("errors.schema.cannotRead", { name: `${schemaName}.json` })
+		};
+		if (e instanceof SyntaxError) return {
+			schema: null,
+			error: t("errors.schema.invalidJson", { name: `${schemaName}.json` })
+		};
+		return {
+			schema: null,
+			error: String(e)
+		};
+	}
+}
+async function listSchemas(aiexDir) {
+	try {
+		const dir = path.join(aiexDir, "schema");
+		return (await fs.readdir(dir)).filter((f) => f.endsWith(".json")).map((f) => f.replace(JSON_EXT_RE$1, "")).sort();
+	} catch {
+		return [];
 	}
-	return { text: await fs.readFile(filePath, "utf-8") };
 }
 async function extractSingle(aiexDir, config, aiConfig, schemaName, text$1, filePath, modelOverride, options) {
 	const schemaLoad = await loadSchema(config, schemaName);
@@ -14367,6 +14588,7 @@ async function runAuditedExtraction(options) {
 					error: error instanceof Error ? error.message : String(error)
 				});
 				if (!quiet) consola.error(t("extract.file.notionSyncFail", { error: error instanceof Error ? error.message : String(error) }));
+				await triggerWebhook(aiConfig, audit.id, schemaName, "extraction.failed", source, r.data, error instanceof Error ? error.message : String(error), r.tokensUsed, quiet);
 				return {
 					success: false,
 					error: error instanceof Error ? error.message : String(error),
@@ -14382,6 +14604,7 @@ async function runAuditedExtraction(options) {
 				notionPages,
 				tokensUsed: r.tokensUsed
 			});
+			await triggerWebhook(aiConfig, audit.id, schemaName, "extraction.success", source, r.data, void 0, r.tokensUsed, quiet);
 			return {
 				success: true,
 				outputPath: updated.outputPath,
@@ -14398,6 +14621,7 @@ async function runAuditedExtraction(options) {
 				error: r.error || "Extraction failed"
 			});
 			if (!quiet) consola.error(t("extract.file.extractionFailed", { error: r.error }));
+			await triggerWebhook(aiConfig, audit.id, schemaName, "extraction.failed", source, void 0, r.error || "Extraction failed", void 0, quiet);
 			return {
 				success: false,
 				error: r.error,
@@ -14417,6 +14641,7 @@ async function runAuditedExtraction(options) {
 				error: e instanceof Error ? e.message : String(e)
 			}));
 		}
+		await triggerWebhook(aiConfig, audit.id, schemaName, "extraction.failed", source, void 0, e instanceof Error ? e.message : String(e), void 0, quiet);
 		return {
 			success: false,
 			error: e instanceof Error ? e.message : String(e),
@@ -14425,73 +14650,6 @@ async function runAuditedExtraction(options) {
 		};
 	}
 }
-async function processOneFile(aiexDir, config, aiConfig, schemaName, filePath, modelOverride, options) {
-	const result = await runAuditedExtraction({
-		aiexDir,
-		config,
-		aiConfig,
-		schemaName,
-		source: {
-			type: "file",
-			filePath
-		},
-		modelOverride,
-		insert: options?.insert,
-		force: options?.force,
-		quiet: false
-	});
-	if (result.success) {
-		if (!result.skipped) consola.success(t("extract.file.processSuccess", { file: path.basename(filePath) }));
-		return true;
-	}
-	return false;
-}
-async function runBatchExtraction(aiexDir, config, aiConfig, schemaName, dir, globPattern, modelOverride, options) {
-	consola.info(t("extract.batch.scanning", { dir: pc.cyan(dir) }));
-	let files;
-	try {
-		files = listSupportedFiles(dir, globPattern);
-	} catch {
-		return {
-			ok: false,
-			successCount: 0,
-			failCount: 0,
-			error: t("extract.batch.errors.cannotReadDir", { dir })
-		};
-	}
-	if (files.length === 0) return {
-		ok: false,
-		successCount: 0,
-		failCount: 0,
-		error: t("extract.batch.errors.noSupportedFiles", { dir })
-	};
-	consola.info(t("extract.batch.found", { count: files.length }));
-	let successCount = 0;
-	let failCount = 0;
-	for (let i = 0; i < files.length; i++) {
-		const file = files[i];
-		consola.info(`\n${t("extract.batch.processing", {
-			current: i + 1,
-			total: files.length,
-			file: pc.cyan(path.basename(file))
-		})}`);
-		if (await processOneFile(aiexDir, config, aiConfig, schemaName, file, modelOverride, {
-			insert: options?.insert,
-			force: options?.force
-		})) successCount++;
-		else failCount++;
-	}
-	consola.info(`\n${t("extract.batch.complete", {
-		success: pc.green(successCount),
-		fail: pc.red(failCount),
-		total: files.length
-	})}`);
-	return {
-		ok: true,
-		successCount,
-		failCount
-	};
-}
 
 //#endregion
 //#region src/commands/dump.ts
@@ -14620,49 +14778,25 @@ const dumpCommand = defineCommand({
 		} else s.stop(t("command.dump.loaded", { count: rows.length }));
 		const s2 = spinner();
 		s2.start(t("command.dump.formatting"));
-		const formattedRows = rows.map((row) => {
-			const newRow = {};
-			columns.forEach((col) => {
-				const colName = col.name;
-				const val = row[colName];
-				const type = (schema?.properties?.[colName])?.type || "";
-				if (val === null || val === void 0) newRow[colName] = "";
-				else if (type === "boolean") if (format === "xlsx") newRow[colName] = val === 1 || val === "1" || val === true;
-				else newRow[colName] = val === 1 || val === "1" || val === true ? "true" : "false";
-				else if (type === "number" || type === "integer") if (val === "") newRow[colName] = "";
-				else {
-					const num = Number(val);
-					newRow[colName] = Number.isNaN(num) ? val : num;
-				}
-				else if (typeof val === "object") newRow[colName] = JSON.stringify(val);
-				else {
-					const dbType = (col.type || "").toLowerCase();
-					if ((dbType.includes("int") || dbType.includes("real") || dbType.includes("num") || dbType.includes("double") || dbType.includes("float")) && typeof val === "string" && val !== "") {
-						const num = Number(val);
-						newRow[colName] = Number.isNaN(num) ? val : num;
-					} else newRow[colName] = val;
-				}
-			});
-			return newRow;
-		});
-		s2.stop(t("command.dump.formatted"));
+		let formattedRows;
+		try {
+			formattedRows = formatRowsConformingToSchema(rows, columns, schema, format);
+			s2.stop(t("command.dump.formatted"));
+		} catch (error) {
+			s2.stop(t("command.dump.dbQueryFailed"));
+			failCommand(error instanceof Error ? error.message : String(error));
+			return;
+		}
 		const s3 = spinner();
 		s3.start(t("command.dump.writing", {
 			format: format.toUpperCase(),
 			path: resolvedOutput
 		}));
 		try {
-			const ws = XLSX.utils.json_to_sheet(formattedRows, { header: columns.map((col) => col.name) });
+			const buffer = generateExportBuffer(tableName, formattedRows, columns, format);
 			const outputDir = path.dirname(resolvedOutput);
 			if (!fs$1.existsSync(outputDir)) fs$1.mkdirSync(outputDir, { recursive: true });
-			if (format === "xlsx") {
-				const wb = XLSX.utils.book_new();
-				XLSX.utils.book_append_sheet(wb, ws, tableName.slice(0, 31));
-				XLSX.writeFile(wb, resolvedOutput);
-			} else {
-				const csv = XLSX.utils.sheet_to_csv(ws);
-				fs$1.writeFileSync(resolvedOutput, "" + csv, "utf8");
-			}
+			fs$1.writeFileSync(resolvedOutput, buffer);
 			s3.stop(t("command.dump.dumpCompleted"));
 			consola.success(t("command.dump.successMsg", {
 				count: rows.length,
@@ -15584,26 +15718,17 @@ function aiRoutes(config) {
 }
 
 //#endregion
-//#region src/server/routes/data.ts
+//#region src/core/data-service.ts
 const FILE_REGEX = /\.json$/;
 const EXTRACTION_TIMESTAMP_RE = /-\d{4}-\d{2}-\d{2}T/;
 const INTERNAL_ROWID_COLUMN = "__aiex_rowid";
 const TIMESTAMP_CLEANUP = /(\d{2})-(\d{2})-(\d{2})/;
 const TIMESTAMP_TZ = /(\d{3})Z/;
-const tableParamSchema = z.object({ name: z.string().regex(/^[a-z][a-z0-9_]*$/) });
-const extractionFileParamSchema = z.object({ name: z.string().regex(/^[\w.-]+\.json$/).refine((name$1) => name$1 === path.basename(name$1) && !name$1.includes("..")) });
-const tableQuerySchema = z.object({
-	page: z.coerce.number().int().min(1).catch(1),
-	pageSize: z.coerce.number().int().min(1).max(500).catch(50),
-	search: z.string().catch(""),
-	sortField: z.string().optional(),
-	sortOrder: z.preprocess((value) => typeof value === "string" ? value.toLowerCase() : value, z.enum(["asc", "desc"]).catch("asc")),
-	all: z.preprocess((value) => value === "true" || value === true, z.boolean().catch(false))
-});
-function invalidParamResponse$1(message) {
-	return (result, c) => {
-		if (!result.success) return c.json({ error: message }, 400);
-	};
+function schemaNameFromExtractionFile(name$1) {
+	const stem = name$1.replace(FILE_REGEX, "");
+	const match = stem.match(EXTRACTION_TIMESTAMP_RE);
+	if (!match || typeof match.index !== "number" || match.index <= 0) return null;
+	return stem.slice(0, match.index);
 }
 function getAuditNotionStatus(record) {
 	if (record.notionPages?.length) return "synced";
@@ -15630,50 +15755,233 @@ async function getRowExtractionActions(aiexDir, tableName) {
 	}
 	return actions;
 }
-function schemaNameFromExtractionFile(name$1) {
-	const stem = name$1.replace(FILE_REGEX, "");
-	const match = stem.match(EXTRACTION_TIMESTAMP_RE);
-	if (!match || typeof match.index !== "number" || match.index <= 0) return null;
-	return stem.slice(0, match.index);
-}
 function createReadonlyQueryDb(databasePath) {
 	return new Kysely({ dialect: new SqliteDialect({ database: new Database(databasePath, { readonly: true }) }) });
 }
+async function listExtractions(config) {
+	const aiexDir = path.dirname(config.schemaPath);
+	const extractedDir = path.join(aiexDir, "extracted");
+	await fs.mkdir(extractedDir, { recursive: true });
+	const jsonFiles = (await fs.readdir(extractedDir)).filter((f) => f.endsWith(".json") && !f.endsWith(".prompt.md"));
+	const auditRecords = await listExtractionAuditRecords(aiexDir);
+	const auditByOutputName = new Map(auditRecords.map((record) => [record.outputName, record]));
+	const records = [];
+	for (const file of jsonFiles) {
+		const schemaName = schemaNameFromExtractionFile(file);
+		if (!schemaName) continue;
+		const timestamp = file.replace(FILE_REGEX, "").slice(schemaName.length + 1).replace(/-/g, (d, i) => i === 4 || i === 7 ? "-" : d).replace(TIMESTAMP_CLEANUP, (_, h, m, s) => `${h}:${m}:${s}`).replace(TIMESTAMP_TZ, ".$1Z");
+		const filePath = path.join(extractedDir, file);
+		try {
+			const stat = await fs.stat(filePath);
+			const audit = auditByOutputName.get(file);
+			const notionPages = audit?.notionPages?.length ? audit.notionPages : void 0;
+			records.push({
+				name: file,
+				schemaName,
+				timestamp,
+				fileSize: stat.size,
+				modifiedAt: stat.mtime.toISOString(),
+				notionStatus: notionPages ? "synced" : audit?.status === "failed" ? "failed" : "not_synced",
+				notionPages,
+				notionError: !notionPages && audit?.status === "failed" ? audit.error : void 0
+			});
+		} catch {
+			continue;
+		}
+	}
+	records.sort((a, b) => b.timestamp.localeCompare(a.timestamp));
+	return records;
+}
+async function listTables(config) {
+	const schemaDir = config.schemaPath;
+	let schemaFiles = [];
+	try {
+		schemaFiles = (await fs.readdir(schemaDir)).filter((f) => f.endsWith(".json"));
+	} catch {
+		schemaFiles = [];
+	}
+	let db = null;
+	let dbTables = [];
+	try {
+		db = createReadonlyQueryDb(config.databasePath);
+		dbTables = (await sql`
+      select name
+      from sqlite_master
+      where type = 'table' and name not like 'sqlite_%' and name not like '_%'
+      order by name
+    `.execute(db)).rows.map((row) => row.name);
+	} catch {} finally {
+		await db?.destroy();
+	}
+	const tables = [];
+	for (const file of schemaFiles) try {
+		const schema = await readFile(path.join(schemaDir, file));
+		const tableName = schema.table?.name;
+		if (!tableName) continue;
+		tables.push({
+			name: tableName,
+			title: schema.title || tableName,
+			hasData: dbTables.includes(tableName)
+		});
+	} catch {
+		continue;
+	}
+	return tables;
+}
+async function getTableData(config, tableName, query) {
+	const { page, pageSize, search, sortField, sortOrder, all } = query;
+	const aiexDir = path.dirname(config.schemaPath);
+	let db;
+	try {
+		db = createReadonlyQueryDb(config.databasePath);
+	} catch {
+		throw new Error(t("server.dbNotFound"));
+	}
+	try {
+		if ((await sql`
+      select name
+      from sqlite_master
+      where type = 'table' and name = ${tableName}
+    `.execute(db)).rows.length === 0) throw new Error(t("server.tableNotFound", { name: tableName }));
+		const columns = (await sql`
+      pragma table_info(${sql.table(tableName)})
+    `.execute(db)).rows.map((col) => ({
+			name: col.name,
+			type: col.type,
+			notNull: !!col.notnull,
+			pk: !!col.pk
+		}));
+		const searchConditions = columns.map((col) => sql`${sql.ref(col.name)} like ${`%${search}%`}`);
+		const searchCondition = search ? sql`where ${sql.join(searchConditions, sql` or `)}` : sql``;
+		const sortColumn = columns.find((col) => col.name === sortField);
+		const orderBy = sortColumn ? sql`order by ${sql.ref(sortColumn.name)} ${sql.raw(sortOrder === "desc" ? "desc" : "asc")}` : sql``;
+		const total = (await sql`
+      select count(*) as count
+      from ${sql.table(tableName)}
+      ${searchCondition}
+    `.execute(db)).rows[0]?.count ?? 0;
+		const offset = (page - 1) * pageSize;
+		const totalPages = all ? 1 : Math.max(1, Math.ceil(total / pageSize));
+		const result = all ? await sql`
+          select rowid as ${sql.raw(INTERNAL_ROWID_COLUMN)}, *
+          from ${sql.table(tableName)}
+          ${searchCondition}
+          ${orderBy}
+        `.execute(db) : await sql`
+          select rowid as ${sql.raw(INTERNAL_ROWID_COLUMN)}, *
+          from ${sql.table(tableName)}
+          ${searchCondition}
+          ${orderBy}
+          limit ${pageSize}
+          offset ${offset}
+        `.execute(db);
+		const actionsByRowId = await getRowExtractionActions(aiexDir, tableName);
+		const rowActions = Object.fromEntries(result.rows.map((row, index) => {
+			const rowId = row[INTERNAL_ROWID_COLUMN];
+			const action = rowId === null || rowId === void 0 ? void 0 : actionsByRowId.get(String(rowId));
+			return action ? [String(index), action] : null;
+		}).filter((entry) => !!entry));
+		const rows = result.rows.map(({ [INTERNAL_ROWID_COLUMN]: _rowid, ...row }) => row);
+		const schemaDir = config.schemaPath;
+		let schema = null;
+		try {
+			const schemaFiles = (await fs.readdir(schemaDir)).filter((f) => f.endsWith(".json"));
+			for (const file of schemaFiles) {
+				const s = await readFile(path.join(schemaDir, file));
+				if (s.table?.name === tableName) {
+					schema = s;
+					break;
+				}
+			}
+		} catch {}
+		return {
+			columns,
+			rows,
+			rowActions,
+			total,
+			page: all ? 1 : page,
+			pageSize: all ? total : pageSize,
+			totalPages,
+			schema
+		};
+	} finally {
+		await db.destroy();
+	}
+}
+async function retryNotionSync(config, fileName) {
+	const aiexDir = path.dirname(config.schemaPath);
+	const extractedDir = path.join(aiexDir, "extracted");
+	const filePath = path.join(extractedDir, fileName);
+	const schemaName = schemaNameFromExtractionFile(fileName);
+	if (!schemaName) throw new Error(t("server.cannotInferSchema"));
+	const aiConfig = await readAIConfig(aiexDir);
+	if (!aiConfig?.notion?.enabled) throw new Error(t("errors.notion.notEnabled"));
+	if (!aiConfig.notion.schemas?.[schemaName]?.databaseId?.trim()) throw new Error(t("errors.notion.noSchemaConfig", { name: schemaName }));
+	try {
+		const data = await readFile(filePath);
+		if (!data || typeof data !== "object" || Array.isArray(data)) throw new Error(t("errors.ai.extractionNotObject"));
+		const page = await writeNotionPage(aiConfig.notion, schemaName, data);
+		const notionPages = [{
+			databaseId: page.databaseId,
+			pageId: page.pageId
+		}];
+		let record = (await listExtractionAuditRecords(aiexDir)).find((record$1) => record$1.outputName === fileName);
+		if (!record) record = await createExtractionAuditRecord(aiexDir, {
+			schemaName,
+			source: {
+				type: "file",
+				filePath,
+				fileName
+			}
+		});
+		if (record) await updateExtractionAuditRecord(aiexDir, record.id, {
+			status: "succeeded",
+			outputPath: filePath,
+			outputName: fileName,
+			notionPages,
+			error: void 0
+		});
+		return {
+			success: true,
+			notionPages
+		};
+	} catch (error) {
+		const message = error instanceof Error ? error.message : String(error);
+		const record = (await listExtractionAuditRecords(aiexDir)).find((record$1) => record$1.outputName === fileName);
+		if (record) await updateExtractionAuditRecord(aiexDir, record.id, {
+			status: "failed",
+			outputPath: filePath,
+			outputName: fileName,
+			error: message
+		});
+		throw error;
+	}
+}
+
+//#endregion
+//#region src/server/routes/data.ts
+const tableParamSchema = z.object({ name: z.string().regex(/^[a-z][a-z0-9_]*$/) });
+const extractionFileParamSchema = z.object({ name: z.string().regex(/^[\w.-]+\.json$/).refine((name$1) => name$1 === path.basename(name$1) && !name$1.includes("..")) });
+const tableQuerySchema = z.object({
+	page: z.coerce.number().int().min(1).catch(1),
+	pageSize: z.coerce.number().int().min(1).max(500).catch(50),
+	search: z.string().catch(""),
+	sortField: z.string().optional(),
+	sortOrder: z.preprocess((value) => typeof value === "string" ? value.toLowerCase() : value, z.enum(["asc", "desc"]).catch("asc")),
+	all: z.preprocess((value) => value === "true" || value === true, z.boolean().catch(false))
+});
+function invalidParamResponse$1(message) {
+	return (result, c) => {
+		if (!result.success) return c.json({ error: message }, 400);
+	};
+}
 function dataRoutes(config) {
 	const app = new Hono();
 	const aiexDir = path.dirname(config.schemaPath);
 	const extractedDir = path.join(aiexDir, "extracted");
 	app.get("/data", async (c) => {
 		try {
-			await fs.mkdir(extractedDir, { recursive: true });
-			const jsonFiles = (await fs.readdir(extractedDir)).filter((f) => f.endsWith(".json") && !f.endsWith(".prompt.md"));
-			const auditRecords = await listExtractionAuditRecords(aiexDir);
-			const auditByOutputName = new Map(auditRecords.map((record) => [record.outputName, record]));
-			const records = [];
-			for (const file of jsonFiles) {
-				const schemaName = schemaNameFromExtractionFile(file);
-				if (!schemaName) continue;
-				const timestamp = file.replace(FILE_REGEX, "").slice(schemaName.length + 1).replace(/-/g, (d, i) => i === 4 || i === 7 ? "-" : d).replace(TIMESTAMP_CLEANUP, (_, h, m, s) => `${h}:${m}:${s}`).replace(TIMESTAMP_TZ, ".$1Z");
-				const filePath = path.join(extractedDir, file);
-				try {
-					const stat = await fs.stat(filePath);
-					const audit = auditByOutputName.get(file);
-					const notionPages = audit?.notionPages?.length ? audit.notionPages : void 0;
-					records.push({
-						name: file,
-						schemaName,
-						timestamp,
-						fileSize: stat.size,
-						modifiedAt: stat.mtime.toISOString(),
-						notionStatus: notionPages ? "synced" : audit?.status === "failed" ? "failed" : "not_synced",
-						notionPages,
-						notionError: !notionPages && audit?.status === "failed" ? audit.error : void 0
-					});
-				} catch {
-					continue;
-				}
-			}
-			records.sort((a, b) => b.timestamp.localeCompare(a.timestamp));
+			const records = await listExtractions(config);
 			return c.json(records);
 		} catch (error) {
 			return c.json({ error: error instanceof Error ? error.message : String(error) }, 500);
@@ -15681,39 +15989,7 @@ function dataRoutes(config) {
 	});
 	app.get("/data/tables", async (c) => {
 		try {
-			const schemaDir = config.schemaPath;
-			let schemaFiles = [];
-			try {
-				schemaFiles = (await fs.readdir(schemaDir)).filter((f) => f.endsWith(".json"));
-			} catch {
-				schemaFiles = [];
-			}
-			let db = null;
-			let dbTables = [];
-			try {
-				db = createReadonlyQueryDb(config.databasePath);
-				dbTables = (await sql`
-          select name
-          from sqlite_master
-          where type = 'table' and name not like 'sqlite_%' and name not like '_%'
-          order by name
-        `.execute(db)).rows.map((row) => row.name);
-			} catch {} finally {
-				await db?.destroy();
-			}
-			const tables = [];
-			for (const file of schemaFiles) try {
-				const schema = await readFile(path.join(schemaDir, file));
-				const tableName = schema.table?.name;
-				if (!tableName) continue;
-				tables.push({
-					name: tableName,
-					title: schema.title || tableName,
-					hasData: dbTables.includes(tableName)
-				});
-			} catch {
-				continue;
-			}
+			const tables = await listTables(config);
 			return c.json(tables);
 		} catch (error) {
 			return c.json({ error: error instanceof Error ? error.message : String(error) }, 500);
@@ -15721,84 +15997,14 @@ function dataRoutes(config) {
 	});
 	app.get("/data/tables/:name", zValidator("param", tableParamSchema, invalidParamResponse$1(t("server.invalidTableName"))), zValidator("query", tableQuerySchema), async (c) => {
 		const { name: tableName } = c.req.valid("param");
-		const { page, pageSize, search, sortField, sortOrder, all } = c.req.valid("query");
-		let db;
+		const query = c.req.valid("query");
 		try {
-			db = createReadonlyQueryDb(config.databasePath);
-		} catch {
-			return c.json({ error: t("server.dbNotFound") }, 400);
-		}
-		try {
-			if ((await sql`
-          select name
-          from sqlite_master
-          where type = 'table' and name = ${tableName}
-        `.execute(db)).rows.length === 0) return c.json({ error: t("server.tableNotFound", { name: tableName }) }, 404);
-			const columns = (await sql`
-          pragma table_info(${sql.table(tableName)})
-        `.execute(db)).rows.map((col) => ({
-				name: col.name,
-				type: col.type,
-				notNull: !!col.notnull,
-				pk: !!col.pk
-			}));
-			const searchConditions = columns.map((col) => sql`${sql.ref(col.name)} like ${`%${search}%`}`);
-			const searchCondition = search ? sql`where ${sql.join(searchConditions, sql` or `)}` : sql``;
-			const sortColumn = columns.find((col) => col.name === sortField);
-			const orderBy = sortColumn ? sql`order by ${sql.ref(sortColumn.name)} ${sql.raw(sortOrder === "desc" ? "desc" : "asc")}` : sql``;
-			const total = (await sql`
-          select count(*) as count
-          from ${sql.table(tableName)}
-          ${searchCondition}
-        `.execute(db)).rows[0]?.count ?? 0;
-			const offset = (page - 1) * pageSize;
-			const totalPages = all ? 1 : Math.max(1, Math.ceil(total / pageSize));
-			const result = all ? await sql`
-              select rowid as ${sql.raw(INTERNAL_ROWID_COLUMN)}, *
-              from ${sql.table(tableName)}
-              ${searchCondition}
-              ${orderBy}
-            `.execute(db) : await sql`
-              select rowid as ${sql.raw(INTERNAL_ROWID_COLUMN)}, *
-              from ${sql.table(tableName)}
-              ${searchCondition}
-              ${orderBy}
-              limit ${pageSize}
-              offset ${offset}
-            `.execute(db);
-			const actionsByRowId = await getRowExtractionActions(aiexDir, tableName);
-			const rowActions = Object.fromEntries(result.rows.map((row, index) => {
-				const rowId = row[INTERNAL_ROWID_COLUMN];
-				const action = rowId === null || rowId === void 0 ? void 0 : actionsByRowId.get(String(rowId));
-				return action ? [String(index), action] : null;
-			}).filter((entry) => !!entry));
-			const rows = result.rows.map(({ [INTERNAL_ROWID_COLUMN]: _rowid, ...row }) => row);
-			const schemaDir = config.schemaPath;
-			let schema = null;
-			try {
-				const schemaFiles = (await fs.readdir(schemaDir)).filter((f) => f.endsWith(".json"));
-				for (const file of schemaFiles) {
-					const s = await readFile(path.join(schemaDir, file));
-					if (s.table?.name === tableName) {
-						schema = s;
-						break;
-					}
-				}
-			} catch {}
-			return c.json({
-				columns,
-				rows,
-				rowActions,
-				total,
-				page: all ? 1 : page,
-				pageSize: all ? total : pageSize,
-				totalPages,
-				schema
-			});
+			const result = await getTableData(config, tableName, query);
+			return c.json(result);
 		} catch (error) {
-			return c.json({ error: error instanceof Error ? error.message : String(error) }, 500);
-		} finally {
-			await db.destroy();
+			const errMessage = error instanceof Error ? error.message : String(error);
+			const status = errMessage.includes("not found") ? 404 : 500;
+			return c.json({ error: errMessage }, status);
 		}
 	});
 	app.get("/data/:name", zValidator("param", extractionFileParamSchema, invalidParamResponse$1(t("server.invalidFileName"))), async (c) => {
@@ -15817,61 +16023,15 @@ function dataRoutes(config) {
 	});
 	app.post("/data/:name/notion/retry", zValidator("param", extractionFileParamSchema, invalidParamResponse$1(t("server.invalidFileName"))), async (c) => {
 		const { name: name$1 } = c.req.valid("param");
-		const filePath = path.join(extractedDir, name$1);
-		const schemaName = schemaNameFromExtractionFile(name$1);
-		if (!schemaName) return c.json({
+		if (!schemaNameFromExtractionFile(name$1)) return c.json({
 			success: false,
 			error: t("server.cannotInferSchema")
 		}, 400);
-		const aiConfig = await readAIConfig(aiexDir);
-		if (!aiConfig?.notion?.enabled) return c.json({
-			success: false,
-			error: t("errors.notion.notEnabled")
-		}, 400);
-		if (!aiConfig.notion.schemas?.[schemaName]?.databaseId?.trim()) return c.json({
-			success: false,
-			error: t("errors.notion.noSchemaConfig", { name: schemaName })
-		}, 400);
 		try {
-			const data = await readFile(filePath);
-			if (!data || typeof data !== "object" || Array.isArray(data)) return c.json({
-				success: false,
-				error: t("errors.ai.extractionNotObject")
-			}, 400);
-			const page = await writeNotionPage(aiConfig.notion, schemaName, data);
-			const notionPages = [{
-				databaseId: page.databaseId,
-				pageId: page.pageId
-			}];
-			let record = (await listExtractionAuditRecords(aiexDir)).find((record$1) => record$1.outputName === name$1);
-			if (!record) record = await createExtractionAuditRecord(aiexDir, {
-				schemaName,
-				source: {
-					type: "file",
-					filePath,
-					fileName: name$1
-				}
-			});
-			if (record) await updateExtractionAuditRecord(aiexDir, record.id, {
-				status: "succeeded",
-				outputPath: filePath,
-				outputName: name$1,
-				notionPages,
-				error: void 0
-			});
-			return c.json({
-				success: true,
-				notionPages
-			});
+			const result = await retryNotionSync(config, name$1);
+			return c.json(result);
 		} catch (error) {
 			const message = error instanceof Error ? error.message : String(error);
-			const record = (await listExtractionAuditRecords(aiexDir)).find((record$1) => record$1.outputName === name$1);
-			if (record) await updateExtractionAuditRecord(aiexDir, record.id, {
-				status: "failed",
-				outputPath: filePath,
-				outputName: name$1,
-				error: message
-			});
 			return c.json({
 				success: false,
 				error: message