@clazic/kordoc 2.4.3 → 2.4.5

package/dist/auto-detect-2YGFYQCN.js

@@ -0,0 +1,15 @@
+#!/usr/bin/env node
+import {
+  detectAvailableOcr,
+  getAutoFallbackChain,
+  getTesseractFallbackMessage,
+  validateOcrMode
+} from "./chunk-7NOZFYH6.js";
+import "./chunk-ZWE3DS7E.js";
+export {
+  detectAvailableOcr,
+  getAutoFallbackChain,
+  getTesseractFallbackMessage,
+  validateOcrMode
+};
+//# sourceMappingURL=auto-detect-2YGFYQCN.js.map

package/dist/chunk-7NOZFYH6.js

@@ -0,0 +1,63 @@
+#!/usr/bin/env node
+
+// src/ocr/auto-detect.ts
+import { execSync } from "child_process";
+var CLI_PRIORITY = ["codex", "gemini", "claude", "ollama"];
+function detectAvailableOcr() {
+  for (const cli of CLI_PRIORITY) {
+    if (isCliInstalled(cli)) return cli;
+  }
+  return "tesseract";
+}
+function isCliInstalled(name) {
+  try {
+    const cmd = process.platform === "win32" ? "where" : "which";
+    execSync(`${cmd} ${name}`, { stdio: "ignore", timeout: 3e3 });
+    return true;
+  } catch {
+    return false;
+  }
+}
+function getAutoFallbackChain() {
+  const chain = [];
+  for (const cli of CLI_PRIORITY) {
+    if (isCliInstalled(cli)) chain.push(cli);
+  }
+  chain.push("tesseract");
+  return chain;
+}
+function validateOcrMode(mode) {
+  if (mode === "auto" || mode === "off" || mode === "tesseract") return;
+  if (!isCliInstalled(mode)) {
+    throw new Error(`'${mode}' CLI\uAC00 \uC124\uCE58\uB418\uC9C0 \uC54A\uC558\uC2B5\uB2C8\uB2E4.
+${getInstallGuide(mode)}`);
+  }
+}
+function getInstallGuide(mode) {
+  const guides = {
+    gemini: "\uC124\uCE58: https://ai.google.dev/gemini-api/docs/cli",
+    claude: "\uC124\uCE58: npm install -g @anthropic-ai/claude-code \uB610\uB294 https://claude.ai/code",
+    codex: "\uC124\uCE58: npm install -g @openai/codex \uB610\uB294 https://github.com/openai/codex",
+    ollama: "\uC124\uCE58: brew install ollama \uB610\uB294 https://ollama.com/download"
+  };
+  return guides[mode] || `'${mode}'\uC744(\uB97C) \uC124\uCE58\uD574\uC8FC\uC138\uC694.`;
+}
+function getTesseractFallbackMessage() {
+  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:",
+    "",
+    "  [\uAD8C\uC7A5] Codex CLI:   npm install -g @openai/codex",
+    "         Gemini CLI:  https://ai.google.dev/gemini-api/docs/cli",
+    "         Claude CLI:  npm install -g @anthropic-ai/claude-code",
+    "         Ollama:      brew install ollama (+ ollama pull gemma4:27b)"
+  ].join("\n");
+}
+
+export {
+  detectAvailableOcr,
+  getAutoFallbackChain,
+  validateOcrMode,
+  getTesseractFallbackMessage
+};
+//# sourceMappingURL=chunk-7NOZFYH6.js.map

package/dist/chunk-7NOZFYH6.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/ocr/auto-detect.ts"],"sourcesContent":["/**\n * OCR CLI 자동 탐색\n *\n * 탐색 순서: codex → gemini → claude → ollama → tesseract.js\n * CLI는 which(unix) / where(win) 명령어로 PATH 존재 확인.\n * tesseract.js는 bundled 의존성이므로 항상 사용 가능 (최후 fallback).\n */\n\nimport { execSync } from \"child_process\"\nimport type { OcrMode } from \"../types.js\"\n\n/** CLI 탐색 우선순위 */\nconst CLI_PRIORITY = [\"codex\", \"gemini\", \"claude\", \"ollama\"] as const\n\n/**\n * 시스템에 설치된 OCR 도구를 우선순위대로 탐색.\n * tesseract.js는 bundled 의존성이므로 CLI를 찾지 못해도 항상 \"tesseract\" 반환.\n * @returns 사용 가능한 OcrMode (null 반환 없음)\n */\nexport function detectAvailableOcr(): OcrMode {\n  // 1. CLI 프로그램 탐색 (codex → gemini → claude → ollama)\n  for (const cli of CLI_PRIORITY) {\n    if (isCliInstalled(cli)) return cli\n  }\n\n  // 2. tesseract.js — bundled 의존성, 항상 사용 가능\n  return \"tesseract\"\n}\n\n/**\n * 특정 CLI가 시스템 PATH에 있는지 확인.\n * which(unix) 또는 where(win32) 사용.\n */\nfunction isCliInstalled(name: string): boolean {\n  try {\n    const cmd = process.platform === \"win32\" ? \"where\" : \"which\"\n    execSync(`${cmd} ${name}`, { stdio: \"ignore\", timeout: 3000 })\n    return true\n  } catch {\n    return false\n  }\n}\n\n/**\n * auto 모드에서 시도할 fallback 체인 반환.\n * 설치된 CLI만 포함하며, tesseract는 항상 마지막에 추가.\n */\nexport function getAutoFallbackChain(): OcrMode[] {\n  const chain: OcrMode[] = []\n  for (const cli of CLI_PRIORITY) {\n    if (isCliInstalled(cli)) chain.push(cli)\n  }\n  chain.push(\"tesseract\")\n  return chain\n}\n\n/**\n * 수동 지정된 OcrMode 유효성 검증.\n * --ocr gemini 등 강제 지정 시 호출.\n * @throws 해당 CLI가 설치되지 않은 경우 Error (tesseract는 항상 통과)\n */\nexport function validateOcrMode(mode: OcrMode): void {\n  if (mode === \"auto\" || mode === \"off\" || mode === \"tesseract\") return\n\n  if (!isCliInstalled(mode)) {\n    throw new Error(`'${mode}' CLI가 설치되지 않았습니다.\\n${getInstallGuide(mode)}`)\n  }\n}\n\n/** CLI별 설치 안내 메시지 */\nfunction getInstallGuide(mode: string): string {\n  const guides: Record<string, string> = {\n    gemini: \"설치: https://ai.google.dev/gemini-api/docs/cli\",\n    claude: \"설치: npm install -g @anthropic-ai/claude-code 또는 https://claude.ai/code\",\n    codex:  \"설치: npm install -g @openai/codex 또는 https://github.com/openai/codex\",\n    ollama: \"설치: brew install ollama 또는 https://ollama.com/download\",\n  }\n  return guides[mode] || `'${mode}'을(를) 설치해주세요.`\n}\n\n/**\n * AI CLI가 없어 tesseract.js로 fallback할 때 표시할 안내 메시지.\n */\nexport function getTesseractFallbackMessage(): string {\n  return [\n    \"설치된 AI CLI가 없어 내장 tesseract.js로 OCR을 수행합니다.\",\n    \"더 나은 품질(테이블/헤딩 구조 보존)을 위해 AI CLI 설치를 권장합니다:\",\n    \"\",\n    \"  [권장] Codex CLI:   npm install -g @openai/codex\",\n    \"         Gemini CLI:  https://ai.google.dev/gemini-api/docs/cli\",\n    \"         Claude CLI:  npm install -g @anthropic-ai/claude-code\",\n    \"         Ollama:      brew install ollama (+ ollama pull gemma4:27b)\",\n  ].join(\"\\n\")\n}\n"],"mappings":";;;AAQA,SAAS,gBAAgB;AAIzB,IAAM,eAAe,CAAC,SAAS,UAAU,UAAU,QAAQ;AAOpD,SAAS,qBAA8B;AAE5C,aAAW,OAAO,cAAc;AAC9B,QAAI,eAAe,GAAG,EAAG,QAAO;AAAA,EAClC;AAGA,SAAO;AACT;AAMA,SAAS,eAAe,MAAuB;AAC7C,MAAI;AACF,UAAM,MAAM,QAAQ,aAAa,UAAU,UAAU;AACrD,aAAS,GAAG,GAAG,IAAI,IAAI,IAAI,EAAE,OAAO,UAAU,SAAS,IAAK,CAAC;AAC7D,WAAO;AAAA,EACT,QAAQ;AACN,WAAO;AAAA,EACT;AACF;AAMO,SAAS,uBAAkC;AAChD,QAAM,QAAmB,CAAC;AAC1B,aAAW,OAAO,cAAc;AAC9B,QAAI,eAAe,GAAG,EAAG,OAAM,KAAK,GAAG;AAAA,EACzC;AACA,QAAM,KAAK,WAAW;AACtB,SAAO;AACT;AAOO,SAAS,gBAAgB,MAAqB;AACnD,MAAI,SAAS,UAAU,SAAS,SAAS,SAAS,YAAa;AAE/D,MAAI,CAAC,eAAe,IAAI,GAAG;AACzB,UAAM,IAAI,MAAM,IAAI,IAAI;AAAA,EAAuB,gBAAgB,IAAI,CAAC,EAAE;AAAA,EACxE;AACF;AAGA,SAAS,gBAAgB,MAAsB;AAC7C,QAAM,SAAiC;AAAA,IACrC,QAAQ;AAAA,IACR,QAAQ;AAAA,IACR,OAAQ;AAAA,IACR,QAAQ;AAAA,EACV;AACA,SAAO,OAAO,IAAI,KAAK,IAAI,IAAI;AACjC;AAKO,SAAS,8BAAsC;AACpD,SAAO;AAAA,IACL;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,EACF,EAAE,KAAK,IAAI;AACb;","names":[]}

package/dist/{chunk-IAU7NTTA.js → chunk-ATB6T3SG.js}

@@ -1,4 +1,7 @@
 #!/usr/bin/env node
+import {
+  markdownToBlocks
+} from "./chunk-4PP34NVQ.js";
 import {
   KordocError,
   classifyError,
@@ -6,7 +9,7 @@ import {
   precheckZipSize,
   sanitizeHref,
   toArrayBuffer
-} from "./chunk-HOUVJPR7.js";
+} from "./chunk-CG3DV7QG.js";
 import {
   parsePageRange
 } from "./chunk-MOL7MDBG.js";
@@ -16,9 +19,6 @@ import {
 import {
   createCliOcrProvider
 } from "./chunk-JOGAFNIL.js";
-import {
-  markdownToBlocks
-} from "./chunk-4PP34NVQ.js";
 import {
   __commonJS,
   __require,
@@ -5456,53 +5456,86 @@ async function parsePdfDocument(buffer, options) {
     }
     const parsedPageCount = parsedPages || (pageFilter ? pageFilter.size : effectivePageCount);
     if (isImageBased) {
-      let ocrProvider = options?.ocr ?? null;
       const ocrMode = options?.ocrMode ?? "auto";
-      if (!ocrProvider && ocrMode !== "off") {
-        try {
-          const { resolveOcrProvider } = await import("./resolve-UOAOPQ4H.js");
-          const concurrency = options?.ocrConcurrency ?? 1;
-          const batchSize = options?.ocrBatchSize;
-          ocrProvider = await resolveOcrProvider(ocrMode, warnings, concurrency, batchSize);
-        } catch (resolveErr) {
-          if (ocrMode !== "auto") {
-            throw Object.assign(
-              new KordocError(resolveErr instanceof Error ? resolveErr.message : "OCR \uD504\uB85C\uBC14\uC774\uB354 \uCD08\uAE30\uD654 \uC2E4\uD328"),
-              { isImageBased: true }
-            );
-          }
-        }
+      const concurrency = options?.ocrConcurrency ?? 1;
+      const batchSize = options?.ocrBatchSize;
+      if (ocrMode === "off") {
+        throw Object.assign(new KordocError(`\uC774\uBBF8\uC9C0 \uAE30\uBC18 PDF (${pageCount}\uD398\uC774\uC9C0, ${totalChars}\uC790)`), { isImageBased: true });
       }
-      if (ocrProvider) {
-        let ocrBlocks = [];
+      const { resolveOcrProvider } = await import("./resolve-TZVGVOVD.js");
+      const { ocrPages } = await import("./provider-7F7NEDTN.js");
+      const tryProvider = async (provider, filter) => {
         try {
-          const { ocrPages } = await import("./provider-HE727F7Z.js");
-          const concurrency = options?.ocrConcurrency ?? 1;
-          ocrBlocks = await ocrPages(doc, ocrProvider, pageFilter, effectivePageCount, warnings, concurrency, options?.onProgress);
+          return await ocrPages(doc, provider, filter, effectivePageCount, warnings, concurrency, options?.onProgress);
         } catch {
+          return [];
         } finally {
-          const terminable = ocrProvider;
+          const terminable = provider;
           if (typeof terminable.terminate === "function") {
             await terminable.terminate().catch(() => {
             });
           }
         }
-        if (ocrBlocks.length > 0) {
-          const ocrMarkdown = blocksToMarkdown(ocrBlocks);
-          return {
-            markdown: ocrMarkdown,
-            blocks: ocrBlocks,
-            metadata,
-            warnings: warnings.length > 0 ? warnings : void 0,
-            isImageBased: true
-          };
+      };
+      let ocrBlocks = [];
+      if (options?.ocr) {
+        ocrBlocks = await tryProvider(options.ocr, pageFilter);
+      } else if (ocrMode === "auto") {
+        const { getAutoFallbackChain } = await import("./auto-detect-2YGFYQCN.js");
+        const pendingPages = /* @__PURE__ */ new Set();
+        for (let i = 1; i <= effectivePageCount; i++) {
+          if (!pageFilter || pageFilter.has(i)) pendingPages.add(i);
+        }
+        const allOcrBlocks = [];
+        for (const mode of getAutoFallbackChain()) {
+          if (pendingPages.size === 0) break;
+          try {
+            const modeFilter = pendingPages.size < effectivePageCount ? new Set(pendingPages) : pageFilter;
+            const provider = await resolveOcrProvider(mode, warnings, concurrency, batchSize);
+            const blocks2 = await tryProvider(provider, modeFilter);
+            if (blocks2.length > 0) {
+              for (const b of blocks2) {
+                if (b.pageNumber !== void 0) pendingPages.delete(b.pageNumber);
+              }
+              for (const b of blocks2) allOcrBlocks.push(b);
+              if (pendingPages.size > 0) {
+                warnings.push({
+                  message: `OCR: '${mode}' \uC644\uB8CC (${pendingPages.size}\uD398\uC774\uC9C0 \uBBF8\uCC98\uB9AC \u2192 \uB2E4\uC74C \uC5D4\uC9C4\uC73C\uB85C \uC7AC\uC2DC\uB3C4)`,
+                  code: "OCR_CLI_FALLBACK"
+                });
+              }
+            } else {
+              warnings.push({ message: `OCR: '${mode}' \uACB0\uACFC \uC5C6\uC74C, \uB2E4\uC74C \uC5D4\uC9C4\uC73C\uB85C \uC2DC\uB3C4`, code: "OCR_CLI_FALLBACK" });
+            }
+          } catch {
+          }
+        }
+        ocrBlocks = allOcrBlocks;
+      } else {
+        try {
+          const provider = await resolveOcrProvider(ocrMode, warnings, concurrency, batchSize);
+          ocrBlocks = await tryProvider(provider, pageFilter);
+        } catch (resolveErr) {
+          throw Object.assign(
+            new KordocError(resolveErr instanceof Error ? resolveErr.message : "OCR \uD504\uB85C\uBC14\uC774\uB354 \uCD08\uAE30\uD654 \uC2E4\uD328"),
+            { isImageBased: true }
+          );
         }
       }
-      if (ocrMode === "off") {
-        throw Object.assign(new KordocError(`\uC774\uBBF8\uC9C0 \uAE30\uBC18 PDF (${pageCount}\uD398\uC774\uC9C0, ${totalChars}\uC790)`), { isImageBased: true });
+      if (ocrBlocks.length > 0) {
+        const ocrMarkdown = blocksToMarkdown(ocrBlocks);
+        return {
+          markdown: ocrMarkdown,
+          blocks: ocrBlocks,
+          metadata,
+          warnings: warnings.length > 0 ? warnings : void 0,
+          isImageBased: true
+        };
       }
-      const errMsg = ocrMode ? `\uC774\uBBF8\uC9C0 \uAE30\uBC18 PDF \u2014 OCR \uC2E4\uD328 (${pageCount}\uD398\uC774\uC9C0, ${totalChars}\uC790)` : `\uC774\uBBF8\uC9C0 \uAE30\uBC18 PDF (${pageCount}\uD398\uC774\uC9C0, ${totalChars}\uC790)`;
-      throw Object.assign(new KordocError(errMsg), { isImageBased: true });
+      throw Object.assign(
+        new KordocError(`\uC774\uBBF8\uC9C0 \uAE30\uBC18 PDF \u2014 OCR \uC2E4\uD328 (${pageCount}\uD398\uC774\uC9C0, ${totalChars}\uC790)`),
+        { isImageBased: true }
+      );
     }
     if (options?.removeHeaderFooter !== false && parsedPageCount >= 3) {
       const removed = removeHeaderFooterBlocks(blocks, pageHeights, warnings);
@@ -9779,4 +9812,4 @@ export {
 cfb/cfb.js:
   (*! crc32.js (C) 2014-present SheetJS -- http://sheetjs.com *)
 */
-//# sourceMappingURL=chunk-IAU7NTTA.js.map
+//# sourceMappingURL=chunk-ATB6T3SG.js.map