kordoc 4.7.0 → 4.7.1

package/dist/chunks-C7DJL4RE.js

@@ -0,0 +1,10 @@
+#!/usr/bin/env node
+import {
+  blocksToChunks
+} from "./chunk-AFX6ZXER.js";
+import "./chunk-YWDN5PQC.js";
+import "./chunk-J3W6Q3AX.js";
+export {
+  blocksToChunks
+};
+//# sourceMappingURL=chunks-C7DJL4RE.js.map

package/dist/cli.js

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   parseFormatProfileJson
-} from "./chunk-YXXHH6WC.js";
+} from "./chunk-UZB3UGC5.js";
 import {
   BODY_FONTS,
   BULLET2_CHARS,
@@ -27,24 +27,24 @@ import {
   readBuiltinTemplate,
   resolveBuiltinTemplate,
   unknownFontWarnings
-} from "./chunk-6PIR74EA.js";
+} from "./chunk-5KPNXBNR.js";
 import "./chunk-7FCNOKOR.js";
-import "./chunk-MVTX6ODZ.js";
+import "./chunk-4VNUQYUE.js";
 import "./chunk-MOL7MDBG.js";
-import "./chunk-IDHBO6AG.js";
+import "./chunk-VECVLIWE.js";
 import "./chunk-DZIXKL7E.js";
-import "./chunk-XZDJXYUC.js";
+import "./chunk-AFX6ZXER.js";
 import {
   blocksToMarkdown
-} from "./chunk-3VTMSSDW.js";
-import "./chunk-PV2EXXGI.js";
-import "./chunk-AFPD74JJ.js";
+} from "./chunk-YWDN5PQC.js";
+import "./chunk-JEX5P66D.js";
+import "./chunk-7ZIOZ6PU.js";
 import {
   VERSION,
   classifyError,
   sanitizeError,
   toArrayBuffer
-} from "./chunk-ZYVULRYF.js";
+} from "./chunk-J3W6Q3AX.js";
 
 // src/cli.ts
 import { readFileSync, writeFileSync, mkdirSync, statSync } from "fs";
@@ -141,7 +141,7 @@ program.name("kordoc").description("\uBAA8\uB450 \uD30C\uC2F1\uD574\uBC84\uB9AC\
 `);
         }
       } else if (opts.format === "chunks") {
-        const { blocksToChunks } = await import("./chunks-4HEG6ZPG.js");
+        const { blocksToChunks } = await import("./chunks-C7DJL4RE.js");
         output = JSON.stringify(blocksToChunks(result.blocks), null, 2);
       } else {
         output = markdown;
@@ -196,7 +196,7 @@ program.command("watch <dir>").description("\uB514\uB809\uD1A0\uB9AC \uAC10\uC2D
   if (command.getOptionValueSource("format") === "default" && program.getOptionValueSource("format") === "cli") {
     opts.format = rootOpts.format;
   }
-  const { watchDirectory } = await import("./watch-MONT6CGI.js");
+  const { watchDirectory } = await import("./watch-5ST75Q7Q.js");
   await watchDirectory({
     dir,
     outDir: opts.outDir,
@@ -418,7 +418,7 @@ program.command("fill [file]").description("\uC11C\uC2DD \uBB38\uC11C\uC758 \uBE
 });
 program.command("seal <file>").description('\uB3C4\uC7A5/\uC11C\uBA85 \uC774\uBBF8\uC9C0\uB97C \uC575\uCEE4 \uBB38\uAD6C \uC704\uC5D0 \uBD80\uC720 \uBC30\uCE58 (\uD45C/\uD398\uC774\uC9C0 \uBD88\uD655\uC7A5) \u2014 kordoc seal \uC2E0\uCCAD\uC11C.hwpx --image \uB3C4\uC7A5.png --anchor "(\uC778)" -o \uACB0\uACFC.hwpx').requiredOption("--image <path>", "\uB3C4\uC7A5/\uC11C\uBA85 \uC774\uBBF8\uC9C0 (\uD22C\uBA85 \uBC30\uACBD PNG \uAD8C\uC7A5)").option("--anchor <text>", "\uC575\uCEE4 \uBB38\uAD6C", "(\uC778)").option("-n, --occurrence <num>", "\uAC19\uC740 \uC575\uCEE4\uAC00 \uC5EC\uB7FF\uC77C \uB54C 0-based \uC120\uD0DD", "0").option("--size-mm <num>", "\uB3C4\uC7A5 \uD55C \uBCC0 \uD06C\uAE30 mm (\uAE30\uBCF8: \uC904\uB192\uC774\xD71.6, 7~18 \uD074\uB7A8\uD504)").option("--mode <mode>", "overlap(\uBB38\uAD6C \uC704 \uACB9\uCE68) | right(\uC624\uB978\uCABD \uC606) | auto", "auto").option("--dx <mm>", "x \uBBF8\uC138\uC870\uC815 mm", "0").option("--dy <mm>", "y \uBBF8\uC138\uC870\uC815 mm", "0").option("-o, --output <path>", "\uCD9C\uB825 \uACBD\uB85C (\uAE30\uBCF8: <\uC785\uB825>.sealed.hwpx)").option("--silent", "\uC9C4\uD589 \uBA54\uC2DC\uC9C0 \uC228\uAE30\uAE30").action(async (file, opts) => {
   try {
-    const { placeSealHwpx, detectFormat: detectFormat2 } = await import("./-MK2223RL.js");
+    const { placeSealHwpx, detectFormat: detectFormat2 } = await import("./-O42BMYNX.js");
     const rootOpts = program.opts();
     const output = opts.output ?? rootOpts.output;
     const silent = opts.silent ?? rootOpts.silent;
@@ -496,7 +496,7 @@ program.command("seal <file>").description('\uB3C4\uC7A5/\uC11C\uBA85 \uC774\uBB
 });
 program.command("patch <original> <edited>").description("\uC11C\uC2DD \uBCF4\uC874 \uB77C\uC6B4\uB4DC\uD2B8\uB9BD \uD328\uCE58 \u2014 \uD3B8\uC9D1\uB41C \uB9C8\uD06C\uB2E4\uC6B4\uC744 \uC6D0\uBCF8 HWPX/HWP\uC5D0 in-place \uBC18\uC601 (kordoc patch \uC6D0\uBCF8.hwpx \uD3B8\uC9D1.md -o \uCD9C\uB825.hwpx). \uBBF8\uC801\uC6A9(skip) \uD3B8\uC9D1\uC774 \uC788\uC73C\uBA74 exit 2").option("-o, --output <path>", "\uCD9C\uB825 \uACBD\uB85C (\uAE30\uBCF8: <\uC6D0\uBCF8>.patched.hwpx|.hwp)").option("--no-verify", "\uD328\uCE58 \uD6C4 \uC7AC\uD30C\uC2F1 \uC790\uB3D9 \uAC80\uC99D \uC0DD\uB7B5").option("--silent", "\uC9C4\uD589 \uBA54\uC2DC\uC9C0 \uC228\uAE30\uAE30").action(async (original, edited, opts) => {
   try {
-    const { patchHwpx, patchHwp, detectFormat: detectFormat2 } = await import("./-MK2223RL.js");
+    const { patchHwpx, patchHwp, detectFormat: detectFormat2 } = await import("./-O42BMYNX.js");
     const rootOpts = program.opts();
     const output = opts.output ?? rootOpts.output;
     const silent = opts.silent ?? rootOpts.silent;
@@ -537,7 +537,7 @@ program.command("patch <original> <edited>").description("\uC11C\uC2DD \uBCF4\uC
 });
 program.command("validate <file>").description("HWPX \uAD6C\uC870 \uAC80\uC99D \u2014 ZIP\xB7mimetype\xB7\uD544\uC218 \uD30C\uC77C\xB7XML \uC6F0\uD3FC\uB4DC\xB7secCnt\xB7manifest \uCC38\uC870 (\uD55C\uCEF4\uB3C5\uC2A4 \uAC70\uBD80 \uC694\uC778 \uC0AC\uC804 \uCC28\uB2E8)").option("--json", "\uACB0\uACFC\uB97C JSON\uC73C\uB85C stdout\uC5D0 \uCD9C\uB825").action(async (file, opts) => {
   try {
-    const { validateHwpx } = await import("./-MK2223RL.js");
+    const { validateHwpx } = await import("./-O42BMYNX.js");
     const buf = new Uint8Array(readFileSync(resolve(file)));
     const result = await validateHwpx(buf);
     if (opts.json) {
@@ -751,7 +751,7 @@ program.command("lint <file>").description("\uACF5\uBB38\uC11C \uD45C\uAE30\uBC9
   }
 });
 program.command("redact <files...>").description("\uAC1C\uC778\uC815\uBCF4 \uC11C\uC2DD \uBCF4\uC874 \uB9C8\uC2A4\uD0B9 \u2014 \uC8FC\uBBFC\uBC88\uD638\xB7\uC804\uD654\xB7\uC774\uBA54\uC77C\xB7\uCE74\uB4DC\xB7\uACC4\uC88C\uB97C \uD0D0\uC9C0\uD574 HWPX/HWP\uB294 \uC6D0\uBCF8 \uC11C\uC2DD \uADF8\uB300\uB85C patch, \uADF8 \uC678 \uD3EC\uB9F7\uC740 \uB9C8\uC2A4\uD0B9\uB41C \uB9C8\uD06C\uB2E4\uC6B4 \uCD9C\uB825. \uC790\uB3D9 \uAC80\uCD9C \uBCF4\uC870 \uB3C4\uAD6C\uC774\uBBC0\uB85C \uACB0\uACFC\uB294 \uBC18\uB4DC\uC2DC \uC0AC\uB78C\uC774 \uCD5C\uC885 \uD655\uC778\uD558\uC138\uC694 (\uC774\uBBF8\uC9C0 \uC548 \uD14D\uC2A4\uD2B8\uB294 \uD0D0\uC9C0 \uBD88\uAC00)").option("--rules <csv>", "\uC801\uC6A9 \uB8F0 (\uAE30\uBCF8: rrn,phone,email,card,account \u2014 passport,driver\uB294 opt-in)").option("--mask-char <ch>", "\uB9C8\uC2A4\uD06C \uBB38\uC790 1\uAE00\uC790 (\uAE30\uBCF8: \u25CF)").option("-o, --output <path>", "\uCD9C\uB825 \uACBD\uB85C (\uB2E8\uC77C \uD30C\uC77C \uC2DC)").option("-d, --out-dir <dir>", "\uCD9C\uB825 \uB514\uB809\uD1A0\uB9AC (\uB2E4\uC911 \uD30C\uC77C \uC2DC)").option("--dry-run", "\uD0D0\uC9C0 \uB9AC\uD3EC\uD2B8\uB9CC \uCD9C\uB825, \uD30C\uC77C \uBBF8\uC0DD\uC131").option("--json", "\uB9AC\uD3EC\uD2B8\uB97C JSON\uC73C\uB85C stdout \uCD9C\uB825").option("--silent", "\uC9C4\uD589 \uBA54\uC2DC\uC9C0 \uC228\uAE30\uAE30").action(async (files, opts) => {
-  const { redactMarkdown, DEFAULT_REDACT_RULES, patchHwpx, patchHwp } = await import("./-MK2223RL.js");
+  const { redactMarkdown, DEFAULT_REDACT_RULES, patchHwpx, patchHwp } = await import("./-O42BMYNX.js");
   const rootOpts = program.opts();
   const output = opts.output ?? rootOpts.output;
   const silent = opts.silent ?? rootOpts.silent;
@@ -840,7 +840,7 @@ program.command("render <file>").description("\uB808\uC774\uC544\uC6C3 \uBCF4\uC
     const rootOpts = program.opts();
     const output = opts.output ?? rootOpts.output;
     const silent = opts.silent ?? rootOpts.silent;
-    const { renderHwpxToSvg } = await import("./render-CAFQT7S4.js");
+    const { renderHwpxToSvg } = await import("./render-IQ4IZLY4.js");
     const absPath = resolve(file);
     const buffer = readFileSync(absPath);
     const highlights = opts.highlight ? String(opts.highlight).split(",") : void 0;
@@ -862,7 +862,7 @@ program.command("render <file>").description("\uB808\uC774\uC544\uC6C3 \uBCF4\uC
 });
 program.command("render-worker").description("persistent \uB80C\uB354 \uC6CC\uCEE4 \u2014 stdin NDJSON \uC694\uCCAD \u2192 \uC870\uD310 SVG \uD30C\uC77C \uCD9C\uB825 (\uD504\uB85C\uC138\uC2A4 \uC720\uC9C0, \uCF5C\uB4DC\uC2A4\uD0C0\uD2B8 \uC81C\uAC70)").action(async () => {
   const { createInterface } = await import("readline");
-  const { renderHwpxToSvg } = await import("./render-CAFQT7S4.js");
+  const { renderHwpxToSvg } = await import("./render-IQ4IZLY4.js");
   const rl = createInterface({ input: process.stdin });
   const write = (o) => void process.stdout.write(JSON.stringify(o) + "\n");
   write({ ready: true, version: VERSION });

package/dist/{image-ocr-4MHEX4MN.js → image-ocr-5QRZBNOF.js}

@@ -4,12 +4,12 @@ import {
   getOcrEngine,
   ocrItemsToBlocks,
   rulingToPdfLines
-} from "./chunk-P2NWNJMR.js";
-import "./chunk-PEVPKD6Z.js";
-import "./chunk-MVTX6ODZ.js";
+} from "./chunk-PLEGO474.js";
+import "./chunk-BQENBAUC.js";
+import "./chunk-4VNUQYUE.js";
 import {
   KordocError
-} from "./chunk-ZYVULRYF.js";
+} from "./chunk-J3W6Q3AX.js";
 import {
   ensureOcrModels
 } from "./chunk-GZDOWYJM.js";
@@ -76,4 +76,4 @@ function detectImageMime(buffer) {
 export {
   parseImageDocument
 };
-//# sourceMappingURL=image-ocr-4MHEX4MN.js.map
+//# sourceMappingURL=image-ocr-5QRZBNOF.js.map

package/dist/{image-ocr-OB5PCQUU.cjs → image-ocr-MKD6MKWO.cjs}

@@ -4,12 +4,12 @@
 
 
 
-var _chunk5QTHHQQEcjs = require('./chunk-5QTHHQQE.cjs');
+var _chunkTZ3WCAJZcjs = require('./chunk-TZ3WCAJZ.cjs');
 require('./chunk-ZV5OKQ3Z.cjs');
-require('./chunk-C27PPXRZ.cjs');
+require('./chunk-ZDWP3LLY.cjs');
 
 
-var _chunkTUU535RBcjs = require('./chunk-TUU535RB.cjs');
+var _chunkDJKLOLBTcjs = require('./chunk-DJKLOLBT.cjs');
 require('./chunk-GS7T56RP.cjs');
 
 // src/ocr/image-ocr.ts
@@ -25,13 +25,13 @@ async function parseImageDocument(buffer, options) {
     return { blocks: [{ type: "paragraph", text: text.trim(), pageNumber: 1 }], warnings };
   }
   const { data, width, height } = await decodeToRgba(buffer);
-  await _chunk5QTHHQQEcjs.ensureOcrModels.call(void 0, (p) => {
+  await _chunkTZ3WCAJZcjs.ensureOcrModels.call(void 0, (p) => {
     if (p.phase === "download" && p.downloaded === 0) {
       process.stderr.write(`[kordoc-ocr] ${p.spec.name} \uB2E4\uC6B4\uB85C\uB4DC \uC911 (~${p.spec.sizeMb}MB)...
 `);
     }
   });
-  const engine = await _chunk5QTHHQQEcjs.getOcrEngine.call(void 0, );
+  const engine = await _chunkTZ3WCAJZcjs.getOcrEngine.call(void 0, );
   const stats = { droppedLowConf: 0 };
   const items = await engine.recognizePage(data, width, height, stats);
   if (stats.droppedLowConf > 0) {
@@ -47,9 +47,9 @@ async function parseImageDocument(buffer, options) {
   }
   const pdfW = width / IMAGE_SCALE;
   const pdfH = height / IMAGE_SCALE;
-  const ruling = _chunk5QTHHQQEcjs.detectRulingLines.call(void 0, data, width, height, IMAGE_SCALE);
-  const extraLines = _chunk5QTHHQQEcjs.rulingToPdfLines.call(void 0, ruling, IMAGE_SCALE, pdfH);
-  return { blocks: _chunk5QTHHQQEcjs.ocrItemsToBlocks.call(void 0, items, 1, pdfW, pdfH, IMAGE_SCALE, extraLines, _optionalChain([options, 'optionalAccess', _2 => _2.tables]) !== false), warnings };
+  const ruling = _chunkTZ3WCAJZcjs.detectRulingLines.call(void 0, data, width, height, IMAGE_SCALE);
+  const extraLines = _chunkTZ3WCAJZcjs.rulingToPdfLines.call(void 0, ruling, IMAGE_SCALE, pdfH);
+  return { blocks: _chunkTZ3WCAJZcjs.ocrItemsToBlocks.call(void 0, items, 1, pdfW, pdfH, IMAGE_SCALE, extraLines, _optionalChain([options, 'optionalAccess', _2 => _2.tables]) !== false), warnings };
 }
 async function decodeToRgba(buffer) {
   let sharp;
@@ -57,7 +57,7 @@ async function decodeToRgba(buffer) {
     const mod = await Promise.resolve().then(() => _interopRequireWildcard(require("sharp")));
     sharp = typeof mod === "function" ? mod : _nullishCoalesce(mod.default, () => ( mod));
   } catch (e) {
-    throw new (0, _chunkTUU535RBcjs.KordocError)(
+    throw new (0, _chunkDJKLOLBTcjs.KordocError)(
       `\uC774\uBBF8\uC9C0 \uD30C\uC2F1\uC5D0\uB294 optional dependency 'sharp' \uAC00 \uD544\uC694\uD569\uB2C8\uB2E4. \`npm install sharp\` \uD6C4 \uB2E4\uC2DC \uC2E4\uD589\uD558\uC138\uC694. \uC6D0\uC778: ${e.message}`
     );
   }
@@ -73,4 +73,4 @@ function detectImageMime(buffer) {
 
 
 exports.parseImageDocument = parseImageDocument;
-//# sourceMappingURL=image-ocr-OB5PCQUU.cjs.map
+//# sourceMappingURL=image-ocr-MKD6MKWO.cjs.map

package/dist/{image-ocr-OB5PCQUU.cjs.map → image-ocr-MKD6MKWO.cjs.map}

@@ -1 +1 @@
-{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/image-ocr-OB5PCQUU.cjs","../src/ocr/image-ocr.ts"],"names":[],"mappings":"AAAA;AACE;AACA;AACA;AACA;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACA;ACKA,IAAM,YAAA,EAAc,CAAA;AAQpB,MAAA,SAAsB,kBAAA,CACpB,MAAA,EACA,OAAA,EACyB;AACzB,EAAA,MAAM,SAAA,EAA2B,CAAC,CAAA;AAGlC,EAAA,GAAA,CAAI,uBAAO,OAAA,2BAAS,MAAA,IAAQ,UAAA,EAAY;AACtC,IAAA,MAAM,KAAA,EAAO,MAAM,OAAA,CAAQ,GAAA,CAAI,IAAI,UAAA,CAAW,MAAM,CAAA,EAAG,CAAA,EAAG,eAAA,CAAgB,MAAM,CAAC,CAAA;AACjF,IAAA,GAAA,CAAI,CAAC,IAAA,CAAK,IAAA,CAAK,CAAA,EAAG;AAChB,MAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,+BAAA,EAAa,IAAA,EAAM,aAAa,CAAC,CAAA;AACnE,MAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,IAChC;AACA,IAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,EAAE,IAAA,EAAM,WAAA,EAAa,IAAA,EAAM,IAAA,CAAK,IAAA,CAAK,CAAA,EAAG,UAAA,EAAY,EAAE,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,EACvF;AAEA,EAAA,MAAM,EAAE,IAAA,EAAM,KAAA,EAAO,OAAO,EAAA,EAAI,MAAM,YAAA,CAAa,MAAM,CAAA;AAEzD,EAAA,MAAM,+CAAA,CAAgB,CAAA,EAAA,GAAK;AACzB,IAAA,GAAA,CAAI,CAAA,CAAE,MAAA,IAAU,WAAA,GAAc,CAAA,CAAE,WAAA,IAAe,CAAA,EAAG;AAChD,MAAA,OAAA,CAAQ,MAAA,CAAO,KAAA,CAAM,CAAA,aAAA,EAAgB,CAAA,CAAE,IAAA,CAAK,IAAI,CAAA,mCAAA,EAAa,CAAA,CAAE,IAAA,CAAK,MAAM,CAAA;AAAA,CAAU,CAAA;AAAA,IACtF;AAAA,EACF,CAAC,CAAA;AACD,EAAA,MAAM,OAAA,EAAS,MAAM,4CAAA,CAAa;AAClC,EAAA,MAAM,MAAA,EAAQ,EAAE,cAAA,EAAgB,EAAE,CAAA;AAClC,EAAA,MAAM,MAAA,EAAQ,MAAM,MAAA,CAAO,aAAA,CAAc,IAAA,EAAM,KAAA,EAAO,MAAA,EAAQ,KAAK,CAAA;AACnE,EAAA,GAAA,CAAI,KAAA,CAAM,eAAA,EAAiB,CAAA,EAAG;AAC5B,IAAA,QAAA,CAAS,IAAA,CAAK;AAAA,MACZ,IAAA,EAAM,CAAA;AAAA,MACN,OAAA,EAAS,CAAA,oCAAA,EAAc,KAAA,CAAM,cAAc,CAAA,4DAAA,CAAA;AAAA,MAC3C,IAAA,EAAM;AAAA,IACR,CAAC,CAAA;AAAA,EACH;AACA,EAAA,GAAA,CAAI,KAAA,CAAM,OAAA,IAAW,CAAA,EAAG;AACtB,IAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,iHAAA,EAAyB,IAAA,EAAM,aAAa,CAAC,CAAA;AAC/E,IAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,EAChC;AAEA,EAAA,MAAM,KAAA,EAAO,MAAA,EAAQ,WAAA;AACrB,EAAA,MAAM,KAAA,EAAO,OAAA,EAAS,WAAA;AACtB,EAAA,MAAM,OAAA,EAAS,iDAAA,IAAkB,EAAM,KAAA,EAAO,MAAA,EAAQ,WAAW,CAAA;AACjE,EAAA,MAAM,WAAA,EAAa,gDAAA,MAAiB,EAAQ,WAAA,EAAa,IAAI,CAAA;AAC7D,EAAA,OAAO,EAAE,MAAA,EAAQ,gDAAA,KAAiB,EAAO,CAAA,EAAG,IAAA,EAAM,IAAA,EAAM,WAAA,EAAa,UAAA,kBAAY,OAAA,6BAAS,SAAA,IAAW,KAAK,CAAA,EAAG,SAAS,CAAA;AACxH;AAGA,MAAA,SAAe,YAAA,CACb,MAAA,EAC8D;AAI9D,EAAA,IAAI,KAAA;AACJ,EAAA,IAAI;AACF,IAAA,MAAM,IAAA,EAAO,MAAM,4DAAA,CAAO,OAAO,GAAA;AACjC,IAAA,MAAA,EAAQ,OAAO,IAAA,IAAQ,WAAA,EAAa,IAAA,mBAAO,GAAA,CAAI,OAAA,UAAY,KAAA;AAAA,EAC7D,EAAA,MAAA,CAAS,CAAA,EAAG;AAEV,IAAA,MAAM,IAAI,kCAAA;AAAA,MACR,CAAA,uMAAA,EAC4C,CAAA,CAAY,OAAO,CAAA;AAAA,IAAA;AACjE,EAAA;AAEF,EAAA;AACA,EAAA;AACF;AAGA;AACE,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACF;AD5BA;AACA;AACA","file":"/Users/mong-e/workspace/kordoc/dist/image-ocr-OB5PCQUU.cjs","sourcesContent":[null,"/**\n * 이미지 직접 입력 OCR — PNG/JPEG/WebP 를 PDF 래핑 없이 바로 파싱한다.\n *\n * 이미지는 텍스트층이 없으므로 OCR 이 항상 필요하다 — `--ocr` 플래그 없이도\n * 내장 엔진(PP-OCRv5 korean)을 기본 적용하고, OcrProvider 가 주어지면 위임한다.\n * 좌표는 216dpi(스케일 3) 가정으로 PDF pt 공간에 환산해 기존 블록 파이프라인\n * (괘선 감지 → 선 기반 표 → 클러스터 → xy-cut)을 그대로 태운다.\n *\n * 디코딩은 optional dependency `sharp` — 미설치면 MISSING_DEPENDENCY 안내.\n */\n\nimport type { IRBlock, ParseOptions, ParseWarning } from \"../types.js\"\nimport { KordocError } from \"../utils.js\"\nimport { getOcrEngine } from \"./engine.js\"\nimport { ensureOcrModels } from \"./models.js\"\nimport { detectRulingLines, rulingToPdfLines } from \"./ruling-lines.js\"\nimport { ocrItemsToBlocks } from \"./pdf-ocr.js\"\n\n/** 좌표 환산 스케일 (px/pt) — PDF OCR 렌더(216dpi)와 동일 기준 */\nconst IMAGE_SCALE = 3\n\nexport interface ImageOcrResult {\n  blocks: IRBlock[]\n  warnings: ParseWarning[]\n}\n\n/** 이미지 버퍼 → OCR → IRBlock[] (표 괘선 감지 포함) */\nexport async function parseImageDocument(\n  buffer: ArrayBuffer,\n  options?: ParseOptions,\n): Promise<ImageOcrResult> {\n  const warnings: ParseWarning[] = []\n\n  // 사용자 OcrProvider — 원본 바이트 그대로 위임 (종전 PDF 경로와 같은 계약)\n  if (typeof options?.ocr === \"function\") {\n    const text = await options.ocr(new Uint8Array(buffer), 1, detectImageMime(buffer))\n    if (!text.trim()) {\n      warnings.push({ page: 1, message: \"OCR 결과 없음\", code: \"OCR_FAILED\" })\n      return { blocks: [], warnings }\n    }\n    return { blocks: [{ type: \"paragraph\", text: text.trim(), pageNumber: 1 }], warnings }\n  }\n\n  const { data, width, height } = await decodeToRgba(buffer)\n\n  await ensureOcrModels(p => {\n    if (p.phase === \"download\" && p.downloaded === 0) {\n      process.stderr.write(`[kordoc-ocr] ${p.spec.name} 다운로드 중 (~${p.spec.sizeMb}MB)...\\n`)\n    }\n  })\n  const engine = await getOcrEngine()\n  const stats = { droppedLowConf: 0 }\n  const items = await engine.recognizePage(data, width, height, stats)\n  if (stats.droppedLowConf > 0) {\n    warnings.push({\n      page: 1,\n      message: `저신뢰 OCR 라인 ${stats.droppedLowConf}개 폐기 (인식 결손 가능)`,\n      code: \"OCR_LOW_CONF\",\n    })\n  }\n  if (items.length === 0) {\n    warnings.push({ page: 1, message: \"이미지에서 텍스트를 인식하지 못했습니다\", code: \"OCR_FAILED\" })\n    return { blocks: [], warnings }\n  }\n\n  const pdfW = width / IMAGE_SCALE\n  const pdfH = height / IMAGE_SCALE\n  const ruling = detectRulingLines(data, width, height, IMAGE_SCALE)\n  const extraLines = rulingToPdfLines(ruling, IMAGE_SCALE, pdfH)\n  return { blocks: ocrItemsToBlocks(items, 1, pdfW, pdfH, IMAGE_SCALE, extraLines, options?.tables !== false), warnings }\n}\n\n/** sharp 로 RGBA 디코딩 — 미설치는 MISSING_DEPENDENCY 로 분류되도록 안내 메시지 throw */\nasync function decodeToRgba(\n  buffer: ArrayBuffer,\n): Promise<{ data: Uint8Array; width: number; height: number }> {\n  type SharpFactory = (input: Buffer) => {\n    ensureAlpha(): { raw(): { toBuffer(opts: { resolveWithObject: true }): Promise<{ data: Buffer; info: { width: number; height: number } }> } }\n  }\n  let sharp: SharpFactory\n  try {\n    const mod = (await import(\"sharp\")) as unknown as SharpFactory | { default?: SharpFactory }\n    sharp = typeof mod === \"function\" ? mod : (mod.default ?? (mod as unknown as SharpFactory))\n  } catch (e) {\n    // KordocError + \"optional dependency\" 문구 → sanitizeError 메시지 보존 + MISSING_DEPENDENCY 분류\n    throw new KordocError(\n      \"이미지 파싱에는 optional dependency 'sharp' 가 필요합니다. \" +\n        `\\`npm install sharp\\` 후 다시 실행하세요. 원인: ${(e as Error).message}`,\n    )\n  }\n  const { data, info } = await sharp(Buffer.from(buffer)).ensureAlpha().raw().toBuffer({ resolveWithObject: true })\n  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height }\n}\n\n/** OcrProvider 계약용 mime 판별 */\nfunction detectImageMime(buffer: ArrayBuffer): \"image/png\" | \"image/jpeg\" | \"image/webp\" {\n  const b = new Uint8Array(buffer, 0, Math.min(12, buffer.byteLength))\n  if (b[0] === 0xff && b[1] === 0xd8) return \"image/jpeg\"\n  if (b[0] === 0x52 && b[1] === 0x49) return \"image/webp\"\n  return \"image/png\"\n}\n"]}
+{"version":3,"sources":["/Users/mong-e/workspace/kordoc/dist/image-ocr-MKD6MKWO.cjs","../src/ocr/image-ocr.ts"],"names":[],"mappings":"AAAA;AACE;AACA;AACA;AACA;AACA;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B,gCAA6B;AAC7B;AACE;AACF,wDAA6B;AAC7B,gCAA6B;AAC7B;AACA;ACKA,IAAM,YAAA,EAAc,CAAA;AAQpB,MAAA,SAAsB,kBAAA,CACpB,MAAA,EACA,OAAA,EACyB;AACzB,EAAA,MAAM,SAAA,EAA2B,CAAC,CAAA;AAGlC,EAAA,GAAA,CAAI,uBAAO,OAAA,2BAAS,MAAA,IAAQ,UAAA,EAAY;AACtC,IAAA,MAAM,KAAA,EAAO,MAAM,OAAA,CAAQ,GAAA,CAAI,IAAI,UAAA,CAAW,MAAM,CAAA,EAAG,CAAA,EAAG,eAAA,CAAgB,MAAM,CAAC,CAAA;AACjF,IAAA,GAAA,CAAI,CAAC,IAAA,CAAK,IAAA,CAAK,CAAA,EAAG;AAChB,MAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,+BAAA,EAAa,IAAA,EAAM,aAAa,CAAC,CAAA;AACnE,MAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,IAChC;AACA,IAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,EAAE,IAAA,EAAM,WAAA,EAAa,IAAA,EAAM,IAAA,CAAK,IAAA,CAAK,CAAA,EAAG,UAAA,EAAY,EAAE,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,EACvF;AAEA,EAAA,MAAM,EAAE,IAAA,EAAM,KAAA,EAAO,OAAO,EAAA,EAAI,MAAM,YAAA,CAAa,MAAM,CAAA;AAEzD,EAAA,MAAM,+CAAA,CAAgB,CAAA,EAAA,GAAK;AACzB,IAAA,GAAA,CAAI,CAAA,CAAE,MAAA,IAAU,WAAA,GAAc,CAAA,CAAE,WAAA,IAAe,CAAA,EAAG;AAChD,MAAA,OAAA,CAAQ,MAAA,CAAO,KAAA,CAAM,CAAA,aAAA,EAAgB,CAAA,CAAE,IAAA,CAAK,IAAI,CAAA,mCAAA,EAAa,CAAA,CAAE,IAAA,CAAK,MAAM,CAAA;AAAA,CAAU,CAAA;AAAA,IACtF;AAAA,EACF,CAAC,CAAA;AACD,EAAA,MAAM,OAAA,EAAS,MAAM,4CAAA,CAAa;AAClC,EAAA,MAAM,MAAA,EAAQ,EAAE,cAAA,EAAgB,EAAE,CAAA;AAClC,EAAA,MAAM,MAAA,EAAQ,MAAM,MAAA,CAAO,aAAA,CAAc,IAAA,EAAM,KAAA,EAAO,MAAA,EAAQ,KAAK,CAAA;AACnE,EAAA,GAAA,CAAI,KAAA,CAAM,eAAA,EAAiB,CAAA,EAAG;AAC5B,IAAA,QAAA,CAAS,IAAA,CAAK;AAAA,MACZ,IAAA,EAAM,CAAA;AAAA,MACN,OAAA,EAAS,CAAA,oCAAA,EAAc,KAAA,CAAM,cAAc,CAAA,4DAAA,CAAA;AAAA,MAC3C,IAAA,EAAM;AAAA,IACR,CAAC,CAAA;AAAA,EACH;AACA,EAAA,GAAA,CAAI,KAAA,CAAM,OAAA,IAAW,CAAA,EAAG;AACtB,IAAA,QAAA,CAAS,IAAA,CAAK,EAAE,IAAA,EAAM,CAAA,EAAG,OAAA,EAAS,iHAAA,EAAyB,IAAA,EAAM,aAAa,CAAC,CAAA;AAC/E,IAAA,OAAO,EAAE,MAAA,EAAQ,CAAC,CAAA,EAAG,SAAS,CAAA;AAAA,EAChC;AAEA,EAAA,MAAM,KAAA,EAAO,MAAA,EAAQ,WAAA;AACrB,EAAA,MAAM,KAAA,EAAO,OAAA,EAAS,WAAA;AACtB,EAAA,MAAM,OAAA,EAAS,iDAAA,IAAkB,EAAM,KAAA,EAAO,MAAA,EAAQ,WAAW,CAAA;AACjE,EAAA,MAAM,WAAA,EAAa,gDAAA,MAAiB,EAAQ,WAAA,EAAa,IAAI,CAAA;AAC7D,EAAA,OAAO,EAAE,MAAA,EAAQ,gDAAA,KAAiB,EAAO,CAAA,EAAG,IAAA,EAAM,IAAA,EAAM,WAAA,EAAa,UAAA,kBAAY,OAAA,6BAAS,SAAA,IAAW,KAAK,CAAA,EAAG,SAAS,CAAA;AACxH;AAGA,MAAA,SAAe,YAAA,CACb,MAAA,EAC8D;AAI9D,EAAA,IAAI,KAAA;AACJ,EAAA,IAAI;AACF,IAAA,MAAM,IAAA,EAAO,MAAM,4DAAA,CAAO,OAAO,GAAA;AACjC,IAAA,MAAA,EAAQ,OAAO,IAAA,IAAQ,WAAA,EAAa,IAAA,mBAAO,GAAA,CAAI,OAAA,UAAY,KAAA;AAAA,EAC7D,EAAA,MAAA,CAAS,CAAA,EAAG;AAEV,IAAA,MAAM,IAAI,kCAAA;AAAA,MACR,CAAA,uMAAA,EAC4C,CAAA,CAAY,OAAO,CAAA;AAAA,IAAA;AACjE,EAAA;AAEF,EAAA;AACA,EAAA;AACF;AAGA;AACE,EAAA;AACA,EAAA;AACA,EAAA;AACA,EAAA;AACF;AD5BA;AACA;AACA","file":"/Users/mong-e/workspace/kordoc/dist/image-ocr-MKD6MKWO.cjs","sourcesContent":[null,"/**\n * 이미지 직접 입력 OCR — PNG/JPEG/WebP 를 PDF 래핑 없이 바로 파싱한다.\n *\n * 이미지는 텍스트층이 없으므로 OCR 이 항상 필요하다 — `--ocr` 플래그 없이도\n * 내장 엔진(PP-OCRv5 korean)을 기본 적용하고, OcrProvider 가 주어지면 위임한다.\n * 좌표는 216dpi(스케일 3) 가정으로 PDF pt 공간에 환산해 기존 블록 파이프라인\n * (괘선 감지 → 선 기반 표 → 클러스터 → xy-cut)을 그대로 태운다.\n *\n * 디코딩은 optional dependency `sharp` — 미설치면 MISSING_DEPENDENCY 안내.\n */\n\nimport type { IRBlock, ParseOptions, ParseWarning } from \"../types.js\"\nimport { KordocError } from \"../utils.js\"\nimport { getOcrEngine } from \"./engine.js\"\nimport { ensureOcrModels } from \"./models.js\"\nimport { detectRulingLines, rulingToPdfLines } from \"./ruling-lines.js\"\nimport { ocrItemsToBlocks } from \"./pdf-ocr.js\"\n\n/** 좌표 환산 스케일 (px/pt) — PDF OCR 렌더(216dpi)와 동일 기준 */\nconst IMAGE_SCALE = 3\n\nexport interface ImageOcrResult {\n  blocks: IRBlock[]\n  warnings: ParseWarning[]\n}\n\n/** 이미지 버퍼 → OCR → IRBlock[] (표 괘선 감지 포함) */\nexport async function parseImageDocument(\n  buffer: ArrayBuffer,\n  options?: ParseOptions,\n): Promise<ImageOcrResult> {\n  const warnings: ParseWarning[] = []\n\n  // 사용자 OcrProvider — 원본 바이트 그대로 위임 (종전 PDF 경로와 같은 계약)\n  if (typeof options?.ocr === \"function\") {\n    const text = await options.ocr(new Uint8Array(buffer), 1, detectImageMime(buffer))\n    if (!text.trim()) {\n      warnings.push({ page: 1, message: \"OCR 결과 없음\", code: \"OCR_FAILED\" })\n      return { blocks: [], warnings }\n    }\n    return { blocks: [{ type: \"paragraph\", text: text.trim(), pageNumber: 1 }], warnings }\n  }\n\n  const { data, width, height } = await decodeToRgba(buffer)\n\n  await ensureOcrModels(p => {\n    if (p.phase === \"download\" && p.downloaded === 0) {\n      process.stderr.write(`[kordoc-ocr] ${p.spec.name} 다운로드 중 (~${p.spec.sizeMb}MB)...\\n`)\n    }\n  })\n  const engine = await getOcrEngine()\n  const stats = { droppedLowConf: 0 }\n  const items = await engine.recognizePage(data, width, height, stats)\n  if (stats.droppedLowConf > 0) {\n    warnings.push({\n      page: 1,\n      message: `저신뢰 OCR 라인 ${stats.droppedLowConf}개 폐기 (인식 결손 가능)`,\n      code: \"OCR_LOW_CONF\",\n    })\n  }\n  if (items.length === 0) {\n    warnings.push({ page: 1, message: \"이미지에서 텍스트를 인식하지 못했습니다\", code: \"OCR_FAILED\" })\n    return { blocks: [], warnings }\n  }\n\n  const pdfW = width / IMAGE_SCALE\n  const pdfH = height / IMAGE_SCALE\n  const ruling = detectRulingLines(data, width, height, IMAGE_SCALE)\n  const extraLines = rulingToPdfLines(ruling, IMAGE_SCALE, pdfH)\n  return { blocks: ocrItemsToBlocks(items, 1, pdfW, pdfH, IMAGE_SCALE, extraLines, options?.tables !== false), warnings }\n}\n\n/** sharp 로 RGBA 디코딩 — 미설치는 MISSING_DEPENDENCY 로 분류되도록 안내 메시지 throw */\nasync function decodeToRgba(\n  buffer: ArrayBuffer,\n): Promise<{ data: Uint8Array; width: number; height: number }> {\n  type SharpFactory = (input: Buffer) => {\n    ensureAlpha(): { raw(): { toBuffer(opts: { resolveWithObject: true }): Promise<{ data: Buffer; info: { width: number; height: number } }> } }\n  }\n  let sharp: SharpFactory\n  try {\n    const mod = (await import(\"sharp\")) as unknown as SharpFactory | { default?: SharpFactory }\n    sharp = typeof mod === \"function\" ? mod : (mod.default ?? (mod as unknown as SharpFactory))\n  } catch (e) {\n    // KordocError + \"optional dependency\" 문구 → sanitizeError 메시지 보존 + MISSING_DEPENDENCY 분류\n    throw new KordocError(\n      \"이미지 파싱에는 optional dependency 'sharp' 가 필요합니다. \" +\n        `\\`npm install sharp\\` 후 다시 실행하세요. 원인: ${(e as Error).message}`,\n    )\n  }\n  const { data, info } = await sharp(Buffer.from(buffer)).ensureAlpha().raw().toBuffer({ resolveWithObject: true })\n  return { data: new Uint8Array(data.buffer, data.byteOffset, data.byteLength), width: info.width, height: info.height }\n}\n\n/** OcrProvider 계약용 mime 판별 */\nfunction detectImageMime(buffer: ArrayBuffer): \"image/png\" | \"image/jpeg\" | \"image/webp\" {\n  const b = new Uint8Array(buffer, 0, Math.min(12, buffer.byteLength))\n  if (b[0] === 0xff && b[1] === 0xd8) return \"image/jpeg\"\n  if (b[0] === 0x52 && b[1] === 0x49) return \"image/webp\"\n  return \"image/png\"\n}\n"]}

package/dist/{image-ocr-3A3HZX6O.js → image-ocr-TAJHRZ5T.js}

@@ -4,12 +4,12 @@ import {
   getOcrEngine,
   ocrItemsToBlocks,
   rulingToPdfLines
-} from "./chunk-AEYNY6FG.js";
+} from "./chunk-6663UXPE.js";
 import "./chunk-3H5X63BQ.js";
-import "./chunk-GIWKBPII.js";
+import "./chunk-3UIT4PPQ.js";
 import {
   KordocError
-} from "./chunk-YUVKU5CW.js";
+} from "./chunk-IZEUWEEY.js";
 
 // src/ocr/image-ocr.ts
 var IMAGE_SCALE = 3;
@@ -72,4 +72,4 @@ function detectImageMime(buffer) {
 export {
   parseImageDocument
 };
-//# sourceMappingURL=image-ocr-3A3HZX6O.js.map
+//# sourceMappingURL=image-ocr-TAJHRZ5T.js.map