@clazic/kordoc 2.4.15 → 2.4.17

package/dist/index.d.cts

@@ -421,6 +421,7 @@ interface UnifiedOcrProgressEvent {
     total?: number;
     code?: UnifiedOcrErrorCode;
     message?: string;
+    model?: string;
 }
 interface UnifiedOcrOptions {
     workspaceDir?: string;

package/dist/index.d.ts

@@ -421,6 +421,7 @@ interface UnifiedOcrProgressEvent {
     total?: number;
     code?: UnifiedOcrErrorCode;
     message?: string;
+    model?: string;
 }
 interface UnifiedOcrOptions {
     workspaceDir?: string;

package/dist/index.js

@@ -3115,7 +3115,7 @@ import JSZip2 from "jszip";
 import { DOMParser } from "@xmldom/xmldom";
 
 // src/utils.ts
-var VERSION = true ? "2.4.14" : "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;
@@ -11349,14 +11349,14 @@ 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(performance.now() - startAt);
@@ -11391,7 +11391,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
   await 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) => {
@@ -11418,13 +11418,32 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
     currentStage = "render";
     markStageStart("render", "PDF \uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC911");
     logStage("info", "render", "start", "PDF \uD398\uC774\uC9C0 \uB80C\uB354\uB9C1 \uC2DC\uC791", { pdf: workingPdfPath, dpi });
-    await renderPdfToPng(workingPdfPath, join4(imagesDir, "page"), dpi);
+    const renderWithProgress = await renderPdfToPngWithProgress(
+      workingPdfPath,
+      join4(imagesDir, "page"),
+      dpi,
+      (current, total) => {
+        markStageProgress(
+          "render",
+          Math.round(current / total * 100),
+          current,
+          total,
+          `\uD398\uC774\uC9C0 ${current}/${total} \uB80C\uB354\uB9C1`
+        );
+      }
+    );
     const images = await listPageImages(imagesDir);
     if (images.length === 0) throw new UnifiedOcrError("RENDER_FAILED", "render", "\uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC2E4\uD328: \uACB0\uACFC \uC774\uBBF8\uC9C0\uAC00 \uC5C6\uC2B5\uB2C8\uB2E4.");
-    markStageProgress("render", 100, images.length, images.length, `\uD398\uC774\uC9C0 ${images.length}\uC7A5 \uC0DD\uC131`);
+    if (!renderWithProgress.emittedPerPageProgress) {
+      markStageProgress("render", 100, images.length, images.length, `\uD398\uC774\uC9C0 ${images.length}\uC7A5 \uC0DD\uC131`);
+    }
     timingsMs.render = elapsedMs(renderStart);
     markStageDone("render", "\uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC644\uB8CC");
-    logStage("info", "render", "done", "\uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC644\uB8CC", { pages: images.length, elapsedMs: timingsMs.render });
+    logStage("info", "render", "done", "\uD398\uC774\uC9C0 \uC774\uBBF8\uC9C0 \uC0DD\uC131 \uC644\uB8CC", {
+      pages: images.length,
+      elapsedMs: timingsMs.render,
+      pageCountSource: renderWithProgress.pageCountSource
+    });
     const probeStart = performance.now();
     currentStage = "probe";
     markStageStart("probe", "\uBAA8\uB378 \uC18D\uB3C4 \uD504\uB85C\uBE0C \uC218\uD589 \uC911");
@@ -11477,7 +11496,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
       const pagePath = join4(rawDir, `page_${String(i + 1).padStart(4, "0")}.md`);
       await 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);
@@ -11601,7 +11620,8 @@ 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) {
@@ -11622,6 +11642,49 @@ async function renderPdfToPng(pdfPath, prefixPath, dpi) {
     throw new UnifiedOcrError("RENDER_FAILED", "render", err instanceof Error ? err.message : String(err));
   }
 }
+async function getPdfPageCount(pdfPath) {
+  const stdout = await runCommandWithStdout("pdfinfo", [pdfPath]);
+  const m = stdout.match(/^\s*Pages:\s*(\d+)\s*$/mi);
+  if (!m) {
+    throw new Error("pdfinfo \uCD9C\uB825\uC5D0\uC11C \uD398\uC774\uC9C0 \uC218\uB97C \uCC3E\uC9C0 \uBABB\uD588\uC2B5\uB2C8\uB2E4.");
+  }
+  const n = Number(m[1]);
+  if (!Number.isFinite(n) || n <= 0) {
+    throw new Error(`\uC798\uBABB\uB41C \uD398\uC774\uC9C0 \uC218: ${m[1]}`);
+  }
+  return n;
+}
+async function renderPdfToPngWithProgress(pdfPath, prefixPath, dpi, onPageDone) {
+  let totalPages = 0;
+  try {
+    totalPages = await getPdfPageCount(pdfPath);
+  } catch {
+    totalPages = 0;
+  }
+  if (totalPages > 0) {
+    try {
+      for (let page = 1; page <= totalPages; page++) {
+        await runCommand("pdftoppm", [
+          "-png",
+          "-r",
+          String(dpi),
+          "-f",
+          String(page),
+          "-l",
+          String(page),
+          pdfPath,
+          prefixPath
+        ]);
+        onPageDone(page, totalPages);
+      }
+      return { emittedPerPageProgress: true, pageCountSource: "pdfinfo" };
+    } catch (err) {
+      throw new UnifiedOcrError("RENDER_FAILED", "render", err instanceof Error ? err.message : String(err));
+    }
+  }
+  await renderPdfToPng(pdfPath, prefixPath, dpi);
+  return { emittedPerPageProgress: false, pageCountSource: "fallback" };
+}
 async function runCommand(cmd, args) {
   await new Promise((resolvePromise, reject) => {
     const child = spawn2(cmd, args, { stdio: "pipe" });
@@ -11636,6 +11699,24 @@ async function runCommand(cmd, args) {
     });
   });
 }
+async function runCommandWithStdout(cmd, args) {
+  return await new Promise((resolvePromise, reject) => {
+    const child = spawn2(cmd, args, { stdio: "pipe" });
+    let stdout = "";
+    let stderr = "";
+    child.stdout.on("data", (d) => {
+      stdout += String(d);
+    });
+    child.stderr.on("data", (d) => {
+      stderr += String(d);
+    });
+    child.on("error", reject);
+    child.on("close", (code) => {
+      if (code === 0) resolvePromise(stdout);
+      else reject(new Error(`${cmd} \uC2E4\uD328 (code=${code}): ${stderr.trim()}`));
+    });
+  });
+}
 async function assertSofficeAvailable() {
   try {
     await runCommand("soffice", ["--version"]);