aiex-cli 0.0.7-beta.2 → 0.1.0-beta.1

package/dist/cli.mjs

@@ -1,4 +1,4 @@
-import { C as PLACEHOLDER_TEXT, D as name, E as description, O as package_default, S as PLACEHOLDER_SCHEMA, T as seedConfig, _ as doctorDiagnosticsTableRows, a as recognizeImageText, b as DEFAULT_MINERU_CONFIG, c as t, d as writeAIConfig, f as AIConfigSchema, h as toSnakeCase, k as version, l as getDefaultAIConfig, m as parseJsonSchema, n as collectDoctorDiagnostics, o as shouldUseImageOcrFallback, p as JsonSchemaDefinitionSchema, r as createMigrationConfig, s as initI18n, t as generateDrizzleSchema, u as readAIConfig, v as formatDoctorDiagnosticsJson, w as createConfig, x as DEFAULT_PROMPT_CONFIG, y as DEFAULT_MINERU_API_CONFIG } from "./generate-drizzle-schema-D_oDmf4J.mjs";
+import { A as package_default, C as DEFAULT_PROMPT_CONFIG, D as seedConfig, E as createConfig, O as description, S as DEFAULT_MINERU_CONFIG, T as PLACEHOLDER_TEXT, _ as doctorDiagnosticsSeverityRows, a as recognizeImageText, b as DEFAULT_LITEPARSE_CONFIG, c as t, d as writeAIConfig, f as AIConfigSchema, h as toSnakeCase, j as version, k as name, l as getDefaultAIConfig, m as parseJsonSchema, n as collectDoctorDiagnostics, o as shouldUseImageOcrFallback, p as JsonSchemaDefinitionSchema, r as createMigrationConfig, s as initI18n, t as generateDrizzleSchema, u as readAIConfig, v as doctorDiagnosticsTableRows, w as PLACEHOLDER_SCHEMA, x as DEFAULT_MINERU_API_CONFIG, y as formatDoctorDiagnosticsJson } from "./generate-drizzle-schema-BAMq_Ufp.mjs";
 import { createRequire } from "node:module";
 import fs from "node:fs/promises";
 import os from "node:os";
@@ -128,6 +128,16 @@ const doctorCommand = defineCommand({
 			});
 			table.push(...doctorDiagnosticsTableRows(diagnostics));
 			process.stdout.write(`${table.toString()}\n`);
+			const severityRows = doctorDiagnosticsSeverityRows(diagnostics);
+			if (severityRows.length) {
+				const summary = new CliTable3({
+					head: ["status", "diagnostic"],
+					colAligns: ["right", "left"],
+					style: { compact: true }
+				});
+				summary.push(...severityRows);
+				process.stdout.write(`${summary.toString()}\n`);
+			}
 		} catch (err) {
 			consola.error(t("command.doctor.diagnosticsFailed", { error: err }));
 		}
@@ -609,6 +619,66 @@ var ExternalCommandPdfConverter = class {
 	}
 };
 
+//#endregion
+//#region src/infrastructure/pdf/liteparse-converter.ts
+const TESSERACT_FAILURE_RE = /tesseract|tessdata|traineddata|language/i;
+function textFromPages(pages = []) {
+	return pages.map((page) => {
+		if (typeof page.text === "string") return page.text;
+		return page.textItems?.map((item) => item.text).filter(Boolean).join("\n") ?? "";
+	}).filter(Boolean).join("\n\n");
+}
+function hasBoundingBoxes(pages = []) {
+	return pages.some((page) => page.textItems?.some((item) => typeof item.x === "number" && typeof item.y === "number" && typeof item.width === "number" && typeof item.height === "number"));
+}
+async function loadLiteParse() {
+	try {
+		return (await import("@llamaindex/liteparse")).LiteParse;
+	} catch (error) {
+		const details = error instanceof Error ? error.message : String(error);
+		throw new Error(`LiteParse is selected but @llamaindex/liteparse is not available. Install optional dependencies or switch the PDF converter to unpdf, mineru, mineru_api, or external. ${details}`);
+	}
+}
+function formatLiteparseError(error, config) {
+	const message = error instanceof Error ? error.message : String(error);
+	if (!config.ocrEnabled) return new Error(message);
+	if (!TESSERACT_FAILURE_RE.test(message)) return new Error(message);
+	return /* @__PURE__ */ new Error(`LiteParse OCR is enabled but Tesseract language data could not be loaded. Install the traineddata file for "${config.ocrLanguage ?? DEFAULT_LITEPARSE_CONFIG.ocrLanguage}" and set pdf.liteparse.tessdataPath to the directory that contains it, or disable pdf.liteparse.ocrEnabled. Original error: ${message}`);
+}
+var LiteparsePdfConverter = class {
+	name = "liteparse";
+	constructor(config = DEFAULT_LITEPARSE_CONFIG) {
+		this.config = config;
+	}
+	async convert(input, filePath) {
+		const LiteParse = await loadLiteParse();
+		const config = {
+			...DEFAULT_LITEPARSE_CONFIG,
+			...this.config
+		};
+		const result = await new LiteParse({
+			ocrEnabled: config.ocrEnabled,
+			ocrLanguage: config.ocrLanguage,
+			ocrServerUrl: config.ocrServerUrl,
+			tessdataPath: config.tessdataPath,
+			quiet: true
+		}).parse(filePath ?? input).catch((error) => {
+			throw formatLiteparseError(error, config);
+		});
+		const pages = Array.isArray(result.pages) ? result.pages : [];
+		return {
+			text: typeof result.text === "string" ? result.text : textFromPages(pages),
+			pageCount: pages.length,
+			metadata: {
+				converter: this.name,
+				ocrEnabled: String(config.ocrEnabled ?? false),
+				...config.ocrLanguage ? { ocrLanguage: config.ocrLanguage } : {},
+				hasBoundingBoxes: String(hasBoundingBoxes(pages))
+			}
+		};
+	}
+};
+
 //#endregion
 //#region src/infrastructure/pdf/mineru-api-converter.ts
 const TRAILING_SLASH_REGEXP = /\/+$/;
@@ -770,6 +840,7 @@ function withFallback(converter, config) {
 }
 function createPdfConverter(config) {
 	if (typeof config === "object") {
+		if (config.converter === "liteparse") return new LiteparsePdfConverter(config.liteparse ?? DEFAULT_LITEPARSE_CONFIG);
 		if (config.converter === "mineru") {
 			const mineruConfig = config.mineru ?? DEFAULT_MINERU_CONFIG;
 			return withFallback(new ExternalCommandPdfConverter("mineru", mineruConfig), mineruConfig);
@@ -783,8 +854,9 @@ function createPdfConverter(config) {
 	const key = typeof config === "string" ? config : "unpdf";
 	let instance = registry$1.get(key);
 	if (!instance) {
-		if (key !== "unpdf") throw new Error(t("errors.pdf.converterRequiresConfig", { name: key }));
-		instance = new UnpdfConverter();
+		if (key === "liteparse") instance = new LiteparsePdfConverter();
+		else if (key === "unpdf") instance = new UnpdfConverter();
+		else throw new Error(t("errors.pdf.converterRequiresConfig", { name: key }));
 		registry$1.set(key, instance);
 	}
 	return instance;
@@ -3061,9 +3133,9 @@ function resolveTsxPath() {
 }
 function resolveHelperPath() {
 	try {
-		return path.join(resolvePackageRoot(), "src/core/schema-sqlite/migrate-helper.ts");
+		return path.join(resolvePackageRoot(), "src/infrastructure/schema/migrate-helper.ts");
 	} catch {
-		return path.join(__dirname, "../../core/schema-sqlite/migrate-helper.ts");
+		return path.join(__dirname, "../schema/migrate-helper.ts");
 	}
 }
 

package/dist/{generate-drizzle-schema-D_oDmf4J.mjs → generate-drizzle-schema-BAMq_Ufp.mjs}

@@ -11,7 +11,7 @@ import { z } from "zod";
 
 //#region package.json
 var name = "aiex-cli";
-var version = "0.0.7-beta.2";
+var version = "0.1.0-beta.1";
 var description = "JSON Schema → SQLite with AI-powered data extraction";
 var package_default = {
 	name,
@@ -42,7 +42,6 @@ var package_default = {
 	exports: {
 		".": "./dist/index.mjs",
 		"./cli": "./dist/cli.mjs",
-		"./core/schema-sqlite/migrate-helper": "./dist/core/schema-sqlite/migrate-helper.mjs",
 		"./package.json": "./package.json"
 	},
 	main: "./dist/index.mjs",
@@ -55,8 +54,8 @@ var package_default = {
 	files: [
 		"bin",
 		"dist",
-		"src/core/schema-sqlite/migrate-helper.ts",
-		"src/core/schema-sqlite/migration-name.ts"
+		"src/infrastructure/schema/migrate-helper.ts",
+		"src/infrastructure/schema/migration-name.ts"
 	],
 	scripts: {
 		"build": "tsdown && pnpm --filter aiex-web build",
@@ -64,6 +63,7 @@ var package_default = {
 		"start": "tsx src/index.ts",
 		"test": "vitest",
 		"coverage": "vitest --coverage",
+		"smoke:package": "node scripts/package-smoke.mjs",
 		"typecheck": "tsc",
 		"lint": "eslint .",
 		"prepack": "cp ../../README.md . && node scripts/generate-completions.mjs",
@@ -111,7 +111,10 @@ var package_default = {
 		"xlsx": "catalog:",
 		"zod": "catalog:"
 	},
-	optionalDependencies: { "@napi-rs/system-ocr": "catalog:" },
+	optionalDependencies: {
+		"@llamaindex/liteparse": "catalog:",
+		"@napi-rs/system-ocr": "catalog:"
+	},
 	devDependencies: {
 		"@antfu/eslint-config": "catalog:cli",
 		"@antfu/ni": "catalog:cli",
@@ -146,6 +149,13 @@ function seedConfig(config = createConfig()) {
 
 //#endregion
 //#region src/domain/ai/types.ts
+const PDF_CONVERTER_KINDS = [
+	"unpdf",
+	"liteparse",
+	"mineru",
+	"mineru_api",
+	"external"
+];
 const PLACEHOLDER_SCHEMA = "{schema}";
 const PLACEHOLDER_TEXT = "{text}";
 const DEFAULT_MODELS = [{
@@ -192,6 +202,10 @@ const DEFAULT_MINERU_CONFIG = {
 	timeout: 600,
 	fallbackToUnpdf: true
 };
+const DEFAULT_LITEPARSE_CONFIG = {
+	ocrEnabled: false,
+	ocrLanguage: "eng"
+};
 const DEFAULT_MINERU_API_CONFIG = {
 	token: "",
 	baseURL: "https://mineru.net/api/v4",
@@ -202,6 +216,7 @@ const DEFAULT_MINERU_API_CONFIG = {
 };
 const DEFAULT_PDF_CONFIG = {
 	converter: "unpdf",
+	liteparse: DEFAULT_LITEPARSE_CONFIG,
 	mineru: DEFAULT_MINERU_CONFIG,
 	mineruApi: DEFAULT_MINERU_API_CONFIG
 };
@@ -290,6 +305,21 @@ function doctorDiagnosticsTableRows(d) {
 	for (const err of p.errors) rows.push(["error", err]);
 	return rows;
 }
+function doctorDiagnosticsSeverityRows(d) {
+	const rows = [];
+	const p = d.project;
+	rows.push([p.dirExists ? "ok" : "warn", p.dirExists ? "Project directory exists" : "Project directory is not initialized"]);
+	rows.push([p.aiConfig ? "ok" : "warn", p.aiConfig ? "AI config exists" : "AI config is missing"]);
+	if (p.aiConfig) rows.push([p.aiApiKeySet ? "ok" : "warn", p.aiApiKeySet ? "AI API key is set" : "AI API key is empty"]);
+	if (p.aiConnectionOk !== null) rows.push([p.aiConnectionOk ? "ok" : "warn", p.aiConnectionOk ? "AI provider connection succeeded" : "AI provider connection failed"]);
+	if (p.pdfConverterOk !== null) rows.push([p.pdfConverterOk ? "ok" : "error", p.pdfConverterOk ? `PDF converter is available: ${p.pdfConverter ?? "none"}` : `PDF converter is unavailable: ${p.pdfConverter ?? "none"}${p.pdfConverterError ? ` (${p.pdfConverterError})` : ""}`]);
+	else if (p.pdfConverterError) rows.push(["warn", p.pdfConverterError]);
+	if (d.imageOcr.ocrOk !== null) rows.push([d.imageOcr.ocrOk ? "ok" : "warn", d.imageOcr.ocrOk ? "Image OCR self-check passed" : `Image OCR self-check failed${d.imageOcr.error ? `: ${d.imageOcr.error}` : ""}`]);
+	if (p.databaseTablesOk !== null) rows.push([p.databaseTablesOk ? "ok" : "error", p.databaseTablesOk ? "Database tables match schemas" : `Database tables are missing: ${p.missingDatabaseTables.join(", ") || "unknown"}`]);
+	for (const invalid of p.invalidSchemas) rows.push(["error", `Invalid schema ${invalid.file}: ${invalid.error}`]);
+	for (const err of p.errors) rows.push(["error", err]);
+	return rows;
+}
 
 //#endregion
 //#region src/domain/schema/parser.ts
@@ -653,6 +683,15 @@ const MineruApiPdfConverterConfigSchema = z.object({
 	enableFormula: z.boolean().optional(),
 	enableTable: z.boolean().optional()
 });
+const LiteparsePdfConverterConfigSchema = z.object({
+	ocrEnabled: z.boolean().default(false),
+	ocrLanguage: z.string().min(1).default("eng"),
+	tessdataPath: z.string().min(1).optional(),
+	ocrServerUrl: z.string().url().refine((value) => {
+		const url = new URL(value);
+		return url.protocol === "http:" || url.protocol === "https:";
+	}, { message: "ocrServerUrl must use http or https" }).optional()
+});
 const PdfConfigSchema = z.preprocess((value) => {
 	if (!value || typeof value !== "object") return value;
 	const config = { ...value };
@@ -667,12 +706,8 @@ const PdfConfigSchema = z.preprocess((value) => {
 	delete config.marker;
 	return config;
 }, z.object({
-	converter: z.enum([
-		"unpdf",
-		"mineru",
-		"mineru_api",
-		"external"
-	]),
+	converter: z.enum(PDF_CONVERTER_KINDS),
+	liteparse: LiteparsePdfConverterConfigSchema.optional(),
 	mineru: ExternalPdfConverterConfigSchema.optional(),
 	mineruApi: MineruApiPdfConverterConfigSchema.optional(),
 	external: ExternalPdfConverterConfigSchema.optional()
@@ -1184,7 +1219,9 @@ const en = {
 	pdf: {
 		converterOptions: {
 			unpdf: "Built-in text extraction (unpdf)",
+			liteparse: "Built-in layout parsing (liteparse)",
 			mineru: "MinerU (mineru)",
+			mineru_api: "MinerU API (mineru_api)",
 			external: "Custom External Command"
 		},
 		ocrFallbackOptions: { localAuto: "Vision model or local OCR" }
@@ -1258,7 +1295,7 @@ async function initI18n(lng) {
 			fallbackLng: "en",
 			resources: {
 				"en": { translation: en },
-				"zh-CN": { translation: await import("./zh-CN-wEUNhuHM.mjs").then((m) => m.zhCN) }
+				"zh-CN": { translation: await import("./zh-CN-B2yrInX9.mjs").then((m) => m.zhCN) }
 			},
 			interpolation: { escapeValue: false },
 			returnNull: false
@@ -1419,6 +1456,14 @@ async function commandAvailable(command) {
 		return false;
 	}
 }
+async function liteparseAvailable() {
+	try {
+		await import("@llamaindex/liteparse");
+		return true;
+	} catch {
+		return false;
+	}
+}
 async function findImageOcrSelfCheckLogo() {
 	const candidates = [
 		path.resolve(MODULE_DIR, "logo.png"),
@@ -1473,7 +1518,11 @@ async function collectDoctorDiagnostics(options = {}) {
 			aiConnectionOk = await checkConnection(cfg.provider.baseURL);
 			pdfConverter = cfg.pdf?.converter ?? "unpdf";
 			if (pdfConverter === "unpdf") pdfConverterOk = true;
-			else if (pdfConverter === "mineru") {
+			else if (pdfConverter === "liteparse") {
+				pdfConverterOk = await liteparseAvailable();
+				if (!pdfConverterOk) pdfConverterError = "@llamaindex/liteparse optional dependency is not installed or cannot be loaded";
+				else if (cfg.pdf?.liteparse?.ocrEnabled && !cfg.pdf.liteparse.tessdataPath) pdfConverterError = "LiteParse OCR is enabled. If OCR fails, install Tesseract traineddata and configure pdf.liteparse.tessdataPath.";
+			} else if (pdfConverter === "mineru") {
 				const command = cfg.pdf?.mineru?.command ?? DEFAULT_MINERU_CONFIG.command;
 				pdfConverterOk = await commandAvailable(command);
 				if (!pdfConverterOk) pdfConverterError = `Command not found: ${command}`;
@@ -1671,4 +1720,4 @@ function generateDrizzleSchema(result) {
 }
 
 //#endregion
-export { PLACEHOLDER_TEXT as C, name as D, description as E, package_default as O, PLACEHOLDER_SCHEMA as S, seedConfig as T, doctorDiagnosticsTableRows as _, recognizeImageText as a, DEFAULT_MINERU_CONFIG as b, t as c, writeAIConfig as d, AIConfigSchema as f, buildDoctorDiagnostics as g, toSnakeCase as h, generateDrizzleConfig as i, version as k, getDefaultAIConfig as l, parseJsonSchema as m, collectDoctorDiagnostics as n, shouldUseImageOcrFallback as o, JsonSchemaDefinitionSchema as p, createMigrationConfig as r, initI18n as s, generateDrizzleSchema as t, readAIConfig as u, formatDoctorDiagnosticsJson as v, createConfig as w, DEFAULT_PROMPT_CONFIG as x, DEFAULT_MINERU_API_CONFIG as y };
+export { package_default as A, DEFAULT_PROMPT_CONFIG as C, seedConfig as D, createConfig as E, description as O, DEFAULT_MINERU_CONFIG as S, PLACEHOLDER_TEXT as T, doctorDiagnosticsSeverityRows as _, recognizeImageText as a, DEFAULT_LITEPARSE_CONFIG as b, t as c, writeAIConfig as d, AIConfigSchema as f, buildDoctorDiagnostics as g, toSnakeCase as h, generateDrizzleConfig as i, version as j, name as k, getDefaultAIConfig as l, parseJsonSchema as m, collectDoctorDiagnostics as n, shouldUseImageOcrFallback as o, JsonSchemaDefinitionSchema as p, createMigrationConfig as r, initI18n as s, generateDrizzleSchema as t, readAIConfig as u, doctorDiagnosticsTableRows as v, PLACEHOLDER_SCHEMA as w, DEFAULT_MINERU_API_CONFIG as x, formatDoctorDiagnosticsJson as y };

package/dist/index.d.mts

@@ -81,6 +81,7 @@ declare function buildDoctorDiagnostics(input: {
 }): DoctorDiagnostics;
 declare function formatDoctorDiagnosticsJson(d: DoctorDiagnostics): string;
 declare function doctorDiagnosticsTableRows(d: DoctorDiagnostics): [string, string][];
+declare function doctorDiagnosticsSeverityRows(d: DoctorDiagnostics): [string, string][];
 //#endregion
 //#region src/config.d.ts
 interface AppConfig {
@@ -363,4 +364,4 @@ declare function generateDrizzleSchema(result: ParseResult): string;
 declare function createMigrationConfig(cwd: string): MigrationConfig;
 declare function generateDrizzleConfig(): string;
 //#endregion
-export { type DoctorDiagnostics, type JsonSchemaDefinition, JsonSchemaDefinitionSchema, type JsonSchemaProperty, type MigrationConfig, type ParseResult, type ParsedColumn, type ParsedRelation, type ParsedTable, buildDoctorDiagnostics, collectDoctorDiagnostics, createMigrationConfig, doctorDiagnosticsTableRows, formatDoctorDiagnosticsJson, generateDrizzleConfig, generateDrizzleSchema, parseJsonSchema };
+export { type DoctorDiagnostics, type JsonSchemaDefinition, JsonSchemaDefinitionSchema, type JsonSchemaProperty, type MigrationConfig, type ParseResult, type ParsedColumn, type ParsedRelation, type ParsedTable, buildDoctorDiagnostics, collectDoctorDiagnostics, createMigrationConfig, doctorDiagnosticsSeverityRows, doctorDiagnosticsTableRows, formatDoctorDiagnosticsJson, generateDrizzleConfig, generateDrizzleSchema, parseJsonSchema };

package/dist/index.mjs

@@ -1,3 +1,3 @@
-import { _ as doctorDiagnosticsTableRows, g as buildDoctorDiagnostics, i as generateDrizzleConfig, m as parseJsonSchema, n as collectDoctorDiagnostics, p as JsonSchemaDefinitionSchema, r as createMigrationConfig, t as generateDrizzleSchema, v as formatDoctorDiagnosticsJson } from "./generate-drizzle-schema-D_oDmf4J.mjs";
+import { _ as doctorDiagnosticsSeverityRows, g as buildDoctorDiagnostics, i as generateDrizzleConfig, m as parseJsonSchema, n as collectDoctorDiagnostics, p as JsonSchemaDefinitionSchema, r as createMigrationConfig, t as generateDrizzleSchema, v as doctorDiagnosticsTableRows, y as formatDoctorDiagnosticsJson } from "./generate-drizzle-schema-BAMq_Ufp.mjs";
 
-export { JsonSchemaDefinitionSchema, buildDoctorDiagnostics, collectDoctorDiagnostics, createMigrationConfig, doctorDiagnosticsTableRows, formatDoctorDiagnosticsJson, generateDrizzleConfig, generateDrizzleSchema, parseJsonSchema };
+export { JsonSchemaDefinitionSchema, buildDoctorDiagnostics, collectDoctorDiagnostics, createMigrationConfig, doctorDiagnosticsSeverityRows, doctorDiagnosticsTableRows, formatDoctorDiagnosticsJson, generateDrizzleConfig, generateDrizzleSchema, parseJsonSchema };

package/dist/{core/schema-sqlite → infrastructure/schema}/migrate-helper.mjs

@@ -15,7 +15,7 @@ function sanitizeMigrationName(name) {
 }
 
 //#endregion
-//#region src/core/schema-sqlite/migrate-helper.ts
+//#region src/infrastructure/schema/migrate-helper.ts
 const require = createRequire(import.meta.url);
 const __filename = fileURLToPath(import.meta.url);
 const __dirname = path.dirname(__filename);
@@ -40,7 +40,7 @@ async function loadSchemaExports(schemaPath) {
 		const pkgPath = require.resolve("aiex-cli/package.json");
 		cliDir = path.dirname(pkgPath);
 	} catch {
-		cliDir = path.dirname(path.dirname(path.dirname(path.dirname(__dirname))));
+		cliDir = path.dirname(path.dirname(path.dirname(__dirname)));
 	}
 	const result = await esbuild.build({
 		entryPoints: [schemaPath],