@clazic/kordoc 2.4.18 → 2.4.19

package/dist/index.d.cts

@@ -106,17 +106,16 @@ interface ParseOptions {
     ocr?: OcrProvider;
     /**
      * OCR 모드 (CLI 자동 탐색용).
-     * - "auto": 설치된 CLI 자동 탐색 (gemini→claude→codex→ollama→tesseract)
-     * - "gemini"|"claude"|"codex"|"ollama"|"tesseract": 특정 도구 강제 지정
+     * - "auto": 설치된 CLI 자동 탐색 (codex→gemini→claude→ollama)
+     * - "gemini"|"claude"|"codex"|"ollama": 특정 도구 강제 지정
      * - "off": OCR 비활성화 (이미지 기반 PDF면 에러)
      * - undefined: 라이브러리 API 기존 동작 유지 (자동 탐색 안 함)
      */
     ocrMode?: OcrMode;
     /**
      * OCR 병렬 처리 수.
-     * - tesseract: 기본값은 CPU 코어 수 (병렬 처리로 속도 향상)
-     * - CLI 제공 프로바이더(gemini/claude/codex): 기본 1 (rate limit 보호)
-     * - 1: 순차 처리 (기존 동작)
+     * - CLI 제공 프로바이더(gemini/claude/codex): 기본 4 (배치 병렬 실행)
+     * - 1: 순차 처리
      */
     ocrConcurrency?: number;
     /** 진행률 콜백 — current: 현재 페이지/섹션, total: 전체 수 */
@@ -241,7 +240,7 @@ interface StructuredOcrResult {
     markdown: string;
 }
 /** OCR 모드 — CLI --ocr 옵션 허용값 */
-type OcrMode = "auto" | "gemini" | "claude" | "codex" | "ollama" | "tesseract" | "off";
+type OcrMode = "auto" | "gemini" | "claude" | "codex" | "ollama" | "off";
 /** 사용자 제공 OCR 함수 — 페이지 이미지를 받아 텍스트 또는 구조화된 결과 반환 */
 type OcrProvider = (pageImage: Uint8Array, pageNumber: number, mimeType: "image/png") => Promise<string | StructuredOcrResult>;
 interface WatchOptions {

package/dist/index.d.ts

@@ -106,17 +106,16 @@ interface ParseOptions {
     ocr?: OcrProvider;
     /**
      * OCR 모드 (CLI 자동 탐색용).
-     * - "auto": 설치된 CLI 자동 탐색 (gemini→claude→codex→ollama→tesseract)
-     * - "gemini"|"claude"|"codex"|"ollama"|"tesseract": 특정 도구 강제 지정
+     * - "auto": 설치된 CLI 자동 탐색 (codex→gemini→claude→ollama)
+     * - "gemini"|"claude"|"codex"|"ollama": 특정 도구 강제 지정
      * - "off": OCR 비활성화 (이미지 기반 PDF면 에러)
      * - undefined: 라이브러리 API 기존 동작 유지 (자동 탐색 안 함)
      */
     ocrMode?: OcrMode;
     /**
      * OCR 병렬 처리 수.
-     * - tesseract: 기본값은 CPU 코어 수 (병렬 처리로 속도 향상)
-     * - CLI 제공 프로바이더(gemini/claude/codex): 기본 1 (rate limit 보호)
-     * - 1: 순차 처리 (기존 동작)
+     * - CLI 제공 프로바이더(gemini/claude/codex): 기본 4 (배치 병렬 실행)
+     * - 1: 순차 처리
      */
     ocrConcurrency?: number;
     /** 진행률 콜백 — current: 현재 페이지/섹션, total: 전체 수 */
@@ -241,7 +240,7 @@ interface StructuredOcrResult {
     markdown: string;
 }
 /** OCR 모드 — CLI --ocr 옵션 허용값 */
-type OcrMode = "auto" | "gemini" | "claude" | "codex" | "ollama" | "tesseract" | "off";
+type OcrMode = "auto" | "gemini" | "claude" | "codex" | "ollama" | "off";
 /** 사용자 제공 OCR 함수 — 페이지 이미지를 받아 텍스트 또는 구조화된 결과 반환 */
 type OcrProvider = (pageImage: Uint8Array, pageNumber: number, mimeType: "image/png") => Promise<string | StructuredOcrResult>;
 interface WatchOptions {

package/dist/index.js

@@ -2183,7 +2183,7 @@ var auto_detect_exports = {};
 __export(auto_detect_exports, {
   detectAvailableOcr: () => detectAvailableOcr,
   getAutoFallbackChain: () => getAutoFallbackChain,
-  getTesseractFallbackMessage: () => getTesseractFallbackMessage,
+  getNoCliMessage: () => getNoCliMessage,
   validateOcrMode: () => validateOcrMode
 });
 import { execSync } from "child_process";
@@ -2191,7 +2191,7 @@ function detectAvailableOcr() {
   for (const cli of CLI_PRIORITY) {
     if (isCliInstalled(cli)) return cli;
   }
-  return "tesseract";
+  return null;
 }
 function isCliInstalled(name) {
   try {
@@ -2207,11 +2207,10 @@ function getAutoFallbackChain() {
   for (const cli of CLI_PRIORITY) {
     if (isCliInstalled(cli)) chain.push(cli);
   }
-  chain.push("tesseract");
   return chain;
 }
 function validateOcrMode(mode) {
-  if (mode === "auto" || mode === "off" || mode === "tesseract") return;
+  if (mode === "auto" || mode === "off") return;
   if (!isCliInstalled(mode)) {
     throw new Error(`'${mode}' CLI\uAC00 \uC124\uCE58\uB418\uC9C0 \uC54A\uC558\uC2B5\uB2C8\uB2E4.
 ${getInstallGuide(mode)}`);
@@ -2226,10 +2225,10 @@ function getInstallGuide(mode) {
   };
   return guides[mode] || `'${mode}'\uC744(\uB97C) \uC124\uCE58\uD574\uC8FC\uC138\uC694.`;
 }
-function getTesseractFallbackMessage() {
+function getNoCliMessage() {
   return [
-    "\uC124\uCE58\uB41C AI CLI\uAC00 \uC5C6\uC5B4 \uB0B4\uC7A5 tesseract.js\uB85C OCR\uC744 \uC218\uD589\uD569\uB2C8\uB2E4.",
-    "\uB354 \uB098\uC740 \uD488\uC9C8(\uD14C\uC774\uBE14/\uD5E4\uB529 \uAD6C\uC870 \uBCF4\uC874)\uC744 \uC704\uD574 AI CLI \uC124\uCE58\uB97C \uAD8C\uC7A5\uD569\uB2C8\uB2E4:",
+    "\uC124\uCE58\uB41C AI CLI\uAC00 \uC5C6\uC5B4 OCR\uC744 \uC218\uD589\uD560 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4.",
+    "\uC774\uBBF8\uC9C0 \uAE30\uBC18 PDF \uCC98\uB9AC\uB97C \uC704\uD574 AI CLI \uC124\uCE58\uB97C \uAD8C\uC7A5\uD569\uB2C8\uB2E4:",
     "",
     "  [\uAD8C\uC7A5] Codex CLI:   npm install -g @openai/codex",
     "         Gemini CLI:  https://ai.google.dev/gemini-api/docs/cli",
@@ -2412,65 +2411,6 @@ var init_cli_provider = __esm({
   }
 });
 
-// src/ocr/tesseract-provider.ts
-var tesseract_provider_exports = {};
-__export(tesseract_provider_exports, {
-  createTesseractPoolProvider: () => createTesseractPoolProvider,
-  createTesseractProvider: () => createTesseractProvider
-});
-import { createWorker } from "tesseract.js";
-async function createTesseractProvider() {
-  const worker = await createWorker("kor+eng");
-  let terminated = false;
-  const provider = async (pageImage, _pageNumber, _mimeType) => {
-    const { data } = await worker.recognize(pageImage);
-    return data.text;
-  };
-  provider.terminate = async () => {
-    if (!terminated) {
-      await worker.terminate();
-      terminated = true;
-    }
-  };
-  return provider;
-}
-async function createTesseractPoolProvider(concurrency) {
-  const workers = await Promise.all(
-    Array.from({ length: concurrency }, () => createWorker("kor+eng"))
-  );
-  const idle = [...workers];
-  const waitQueue = [];
-  function acquire() {
-    if (idle.length > 0) return Promise.resolve(idle.pop());
-    return new Promise((resolve4) => waitQueue.push(resolve4));
-  }
-  function release(w) {
-    if (waitQueue.length > 0) {
-      waitQueue.shift()(w);
-    } else {
-      idle.push(w);
-    }
-  }
-  const provider = async (pageImage, _pageNumber, _mimeType) => {
-    const w = await acquire();
-    try {
-      const { data } = await w.recognize(pageImage);
-      return data.text;
-    } finally {
-      release(w);
-    }
-  };
-  provider.terminate = async () => {
-    await Promise.all(workers.map((w) => w.terminate()));
-  };
-  return provider;
-}
-var init_tesseract_provider = __esm({
-  "src/ocr/tesseract-provider.ts"() {
-    "use strict";
-  }
-});
-
 // src/ocr/batch-provider.ts
 var batch_provider_exports = {};
 __export(batch_provider_exports, {
@@ -2679,15 +2619,6 @@ async function resolveOcrProvider(mode, warnings, concurrency, batchSize) {
   }
   if (mode !== "auto") {
     validateOcrMode(mode);
-    if (mode === "tesseract") {
-      const { createTesseractProvider: createTesseractProvider2, createTesseractPoolProvider: createTesseractPoolProvider2 } = await Promise.resolve().then(() => (init_tesseract_provider(), tesseract_provider_exports));
-      if (concurrency && concurrency > 1) {
-        logger.log({ level: "info", event: "done", message: "Tesseract pool provider \uC120\uD0DD", meta: { concurrency } });
-        return createTesseractPoolProvider2(concurrency);
-      }
-      logger.log({ level: "info", event: "done", message: "Tesseract single provider \uC120\uD0DD" });
-      return createTesseractProvider2();
-    }
     if (mode === "gemini" || mode === "claude" || mode === "codex") {
       const { createBatchCliProvider: createBatchCliProvider2, DEFAULT_BATCH_SIZES: DEFAULT_BATCH_SIZES2 } = await Promise.resolve().then(() => (init_batch_provider(), batch_provider_exports));
       const effectiveBatch = batchSize ?? DEFAULT_BATCH_SIZES2[mode];
@@ -2703,27 +2634,16 @@ async function resolveOcrProvider(mode, warnings, concurrency, batchSize) {
   }
   const detected = detectAvailableOcr();
   logger.log({ level: "info", event: "progress", message: "OCR auto \uAC10\uC9C0 \uACB0\uACFC", meta: { detected } });
-  if (detected !== "codex") {
-    if (detected === "tesseract") {
-      warnings?.push({
-        message: getTesseractFallbackMessage(),
-        code: "OCR_CLI_FALLBACK"
-      });
-    } else {
-      warnings?.push({
-        message: `OCR: '${detected}' \uC0AC\uC6A9 \uC911 (codex CLI\uAC00 \uC5C6\uC5B4 fallback). \uB354 \uB098\uC740 \uD488\uC9C8\uC744 \uC704\uD574 codex CLI \uC124\uCE58\uB97C \uAD8C\uC7A5\uD569\uB2C8\uB2E4.`,
-        code: "OCR_CLI_FALLBACK"
-      });
-    }
+  if (!detected) {
+    throw new Error(
+      "\uC0AC\uC6A9 \uAC00\uB2A5\uD55C OCR CLI\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4. \uB2E4\uC74C \uC911 \uD558\uB098\uB97C \uC124\uCE58\uD558\uC138\uC694:\n  Codex CLI:  npm install -g @openai/codex\n  Claude CLI: npm install -g @anthropic-ai/claude-code\n  Gemini CLI: https://ai.google.dev/gemini-api/docs/cli"
+    );
   }
-  if (detected === "tesseract") {
-    const { createTesseractProvider: createTesseractProvider2, createTesseractPoolProvider: createTesseractPoolProvider2 } = await Promise.resolve().then(() => (init_tesseract_provider(), tesseract_provider_exports));
-    if (concurrency && concurrency > 1) {
-      logger.log({ level: "info", event: "done", message: "AUTO: Tesseract pool provider \uC120\uD0DD", meta: { concurrency } });
-      return createTesseractPoolProvider2(concurrency);
-    }
-    logger.log({ level: "info", event: "done", message: "AUTO: Tesseract single provider \uC120\uD0DD" });
-    return createTesseractProvider2();
+  if (detected !== "codex") {
+    warnings?.push({
+      message: `OCR: '${detected}' \uC0AC\uC6A9 \uC911 (codex CLI\uAC00 \uC5C6\uC5B4 fallback). \uB354 \uB098\uC740 \uD488\uC9C8\uC744 \uC704\uD574 codex CLI \uC124\uCE58\uB97C \uAD8C\uC7A5\uD569\uB2C8\uB2E4.`,
+      code: "OCR_CLI_FALLBACK"
+    });
   }
   if (detected === "gemini" || detected === "codex" || detected === "claude") {
     const { createBatchCliProvider: createBatchCliProvider2, DEFAULT_BATCH_SIZES: DEFAULT_BATCH_SIZES2 } = await Promise.resolve().then(() => (init_batch_provider(), batch_provider_exports));
@@ -3115,7 +3035,7 @@ import JSZip2 from "jszip";
 import { DOMParser } from "@xmldom/xmldom";
 
 // src/utils.ts
-var VERSION = true ? "2.4.17" : "0.0.0-dev";
+var VERSION = true ? "2.4.19" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -8711,7 +8631,6 @@ async function parseDocxDocument(buffer, options, existingZip) {
 
 // src/index.ts
 init_cli_provider();
-init_tesseract_provider();
 init_markdown_to_blocks();
 init_logger();
 
@@ -12070,9 +11989,6 @@ async function parseImage(buffer, options) {
     if (ocrMode === "gemini" || ocrMode === "claude" || ocrMode === "codex" || ocrMode === "ollama") {
       ocrProvider = createCliOcrProvider(ocrMode);
       actualOcrMode = ocrMode;
-    } else if (ocrMode === "tesseract") {
-      ocrProvider = await createTesseractProvider();
-      actualOcrMode = ocrMode;
     } else if (ocrMode === "auto") {
       const modesToTry = ["gemini", "claude", "codex", "ollama"];
       for (const mode of modesToTry) {
@@ -12084,10 +12000,6 @@ async function parseImage(buffer, options) {
           console.warn(`[kordoc] OCR auto-detection: ${mode} CLI not available or failed. Trying next.`, e);
         }
       }
-      if (!ocrProvider) {
-        ocrProvider = await createTesseractProvider();
-        actualOcrMode = "tesseract";
-      }
     }
     if (!ocrProvider) {
       return { success: false, fileType: "image", error: "\uC0AC\uC6A9 \uAC00\uB2A5\uD55C OCR \uD504\uB85C\uBC14\uC774\uB354\uB97C \uCC3E\uC744 \uC218 \uC5C6\uC2B5\uB2C8\uB2E4.", code: "PARSE_ERROR" };