@clazic/kordoc 2.4.16 → 2.4.17

package/dist/cli.js

@@ -4,12 +4,12 @@ import {
   markdownToHwpx,
   markdownToXlsx,
   parse
-} from "./chunk-RH6IBTHH.js";
+} from "./chunk-WM3XI23V.js";
 import "./chunk-YW5G6BCJ.js";
 import {
   VERSION,
   toArrayBuffer
-} from "./chunk-QR27D67R.js";
+} from "./chunk-W2KDIKDF.js";
 import "./chunk-MOL7MDBG.js";
 import "./chunk-7FMKAV4P.js";
 import "./chunk-34WIGIQC.js";
@@ -177,7 +177,7 @@ async function runParse(files, opts) {
         saveImages(absPath);
       }
     } catch (err) {
-      const { sanitizeError } = await import("./utils-HHJDSSR6.js");
+      const { sanitizeError } = await import("./utils-DHOODYKU.js");
       process.stderr.write(`
 [kordoc] ERROR: ${fileName} \u2014 ${sanitizeError(err)}
 `);
@@ -259,7 +259,7 @@ program.command("convert <input>").description("\uB9C8\uD06C\uB2E4\uC6B4 \uD30C\
 `));
     }
   } catch (err) {
-    const { sanitizeError } = await import("./utils-HHJDSSR6.js");
+    const { sanitizeError } = await import("./utils-DHOODYKU.js");
     process.stderr.write(` FAIL
 `);
     process.stderr.write(`  \u2192 ${sanitizeError(err)}
@@ -291,7 +291,7 @@ program.command("init-env").description("kordoc\uC6A9 .env \uD15C\uD50C\uB9BF \u
   }
 });
 program.command("watch <dir>").description("\uB514\uB809\uD1A0\uB9AC \uAC10\uC2DC \u2014 \uC0C8 \uBB38\uC11C \uC790\uB3D9 \uBCC0\uD658").option("--webhook <url>", "\uACB0\uACFC \uC804\uC1A1 \uC6F9\uD6C5 URL").option("-d, --out-dir <dir>", "\uBCC0\uD658 \uACB0\uACFC \uCD9C\uB825 \uB514\uB809\uD1A0\uB9AC").option("-p, --pages <range>", "\uD398\uC774\uC9C0/\uC139\uC158 \uBC94\uC704").option("--format <type>", "\uCD9C\uB825 \uD615\uC2DD: markdown \uB610\uB294 json", "markdown").option("--silent", "\uC9C4\uD589 \uBA54\uC2DC\uC9C0 \uC228\uAE30\uAE30").action(async (dir, opts) => {
-  const { watchDirectory } = await import("./watch-YAILKKKP.js");
+  const { watchDirectory } = await import("./watch-RM4VNOL4.js");
   await watchDirectory({
     dir,
     outDir: opts.outDir,

package/dist/index.cjs

@@ -3138,7 +3138,7 @@ var import_jszip2 = __toESM(require("jszip"), 1);
 var import_xmldom = require("@xmldom/xmldom");
 
 // src/utils.ts
-var VERSION = true ? "2.4.15" : "0.0.0-dev";
+var VERSION = true ? "2.4.17" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -11372,14 +11372,14 @@ var DEFAULT_STAGE_WEIGHTS = {
   proofread: 10,
   merge: 5
 };
-var OCR_PROMPT2 = "\uC774 \uC774\uBBF8\uC9C0 1\uC7A5\uC758 \uD14D\uC2A4\uD2B8\uC640 \uD45C\uB97C \uC694\uC57D \uC5C6\uC774 \uADF8\uB300\uB85C Markdown\uC73C\uB85C \uCD94\uCD9C\uD558\uC138\uC694. \uC808\uB300\uB85C \uB0B4\uC6A9\uC744 \uCD94\uCE21\uD558\uAC70\uB098 \uBC14\uAFB8\uC9C0 \uB9C8\uC138\uC694.";
+var OCR_PROMPT2 = "Extract all text and tables from this image exactly as-is into Markdown. Do not summarize, infer, or alter the content in any way.";
 var PROOFREAD_PROMPT = [
-  "\uC544\uB798 Markdown\uC744 \uBE44\uD30C\uAD34 \uAD50\uC815\uB9CC \uC218\uD589\uD558\uC138\uC694.",
-  "\uADDC\uCE59:",
-  "- \uC0AC\uC2E4 \uCD94\uAC00/\uC0AD\uC81C/\uCD94\uCE21 \uAE08\uC9C0",
-  "- \uC22B\uC790, \uB2E8\uC704, \uACE0\uC720\uBA85\uC0AC \uBCC0\uACBD \uAE08\uC9C0",
-  "- \uC624\uD0C8\uC790, \uB744\uC5B4\uC4F0\uAE30, \uC904\uBC14\uAFC8, Markdown \uAD6C\uC870\uB9CC \uAD50\uC815",
-  "- \uACB0\uACFC\uB294 Markdown \uBCF8\uBB38\uB9CC \uCD9C\uB825"
+  "Perform non-destructive proofreading only on the Markdown below.",
+  "Rules:",
+  "- Do not add, remove, or infer any facts",
+  "- Do not change numbers, units, or proper nouns",
+  "- Correct only typos, spacing, line breaks, and Markdown structure",
+  "- Output the corrected Markdown body only"
 ].join("\n");
 function elapsedMs(startAt) {
   return Math.round(import_node_perf_hooks.performance.now() - startAt);
@@ -11414,7 +11414,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
   await (0, import_promises2.mkdir)(diffDir, { recursive: true });
   const timingsMs = {};
   const markStageStart = (stage, message) => emitProgress(options.onEvent, stage, 0, stageWeights, { message, type: "stage_start" });
-  const markStageProgress = (stage, stagePercent, current, total, message) => emitProgress(options.onEvent, stage, stagePercent, stageWeights, { type: "stage_progress", current, total, message });
+  const markStageProgress = (stage, stagePercent, current, total, message, model) => emitProgress(options.onEvent, stage, stagePercent, stageWeights, { type: "stage_progress", current, total, message, model });
   const markStageDone = (stage, message) => emitProgress(options.onEvent, stage, 100, stageWeights, { message, type: "stage_done" });
   let currentStage = "convert";
   const logStage = (level, stage, event, message, meta) => {
@@ -11519,7 +11519,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
       const pagePath = (0, import_path5.join)(rawDir, `page_${String(i + 1).padStart(4, "0")}.md`);
       await (0, import_promises2.writeFile)(pagePath, markdown, "utf-8");
       rawPagePaths.push(pagePath);
-      markStageProgress("ocr", Math.round((i + 1) / images.length * 100), i + 1, images.length, `OCR ${i + 1}/${images.length}`);
+      markStageProgress("ocr", Math.round((i + 1) / images.length * 100), i + 1, images.length, `OCR ${i + 1}/${images.length}`, selectedModel);
       logStage("debug", "ocr", "progress", "\uD398\uC774\uC9C0 OCR \uC644\uB8CC", { page: i + 1, total: images.length });
     }
     timingsMs.ocr = elapsedMs(ocrStart);
@@ -11643,7 +11643,8 @@ function emitProgress(cb, stage, stagePercent, weights, extra) {
     current: extra.current,
     total: extra.total,
     code: extra.code,
-    message: extra.message
+    message: extra.message,
+    model: extra.model
   });
 }
 async function convertWithLibreOffice(buffer, ext) {