@clazic/kordoc 2.4.13 → 2.4.15

package/dist/index.js

@@ -3115,7 +3115,7 @@ import JSZip2 from "jszip";
 import { DOMParser } from "@xmldom/xmldom";
 
 // src/utils.ts
-var VERSION = true ? "2.4.12" : "0.0.0-dev";
+var VERSION = true ? "2.4.14" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -11309,6 +11309,7 @@ var ApiKeyRotationPool = class _ApiKeyRotationPool {
 import { mkdir, readdir, readFile, stat, writeFile } from "fs/promises";
 import { basename as basename2, dirname as dirname3, extname, join as join4, resolve as resolve3 } from "path";
 import { spawn as spawn2 } from "child_process";
+import { performance } from "perf_hooks";
 import libre from "libreoffice-convert";
 init_logger();
 var libreConvert = libre.convert;
@@ -11357,6 +11358,9 @@ var PROOFREAD_PROMPT = [
   "- \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"
 ].join("\n");
+function elapsedMs(startAt) {
+  return Math.round(performance.now() - startAt);
+}
 async function runUnifiedOcrPipeline(inputPath, options = {}) {
   const absInput = resolve3(inputPath);
   const stem = basename2(absInput, extname(absInput));
@@ -11396,7 +11400,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
   try {
     ensureSupportedInput(absInput);
     let workingPdfPath = absInput;
-    const convertStart = Date.now();
+    const convertStart = performance.now();
     currentStage = "convert";
     markStageStart("convert", "\uBB38\uC11C\uB97C PDF\uB85C \uBCC0\uD658 \uC911");
     logStage("info", "convert", "start", "\uBB38\uC11C\uB97C PDF\uB85C \uBCC0\uD658 \uC2DC\uC791", { input: absInput });
@@ -11407,10 +11411,10 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
       const out = await convertWithLibreOffice(inputBuffer, ".pdf");
       await writeFile(workingPdfPath, out);
     }
-    timingsMs.convert = Date.now() - convertStart;
+    timingsMs.convert = elapsedMs(convertStart);
     markStageDone("convert", "PDF \uBCC0\uD658 \uC644\uB8CC");
     logStage("info", "convert", "done", "PDF \uBCC0\uD658 \uC644\uB8CC", { elapsedMs: timingsMs.convert });
-    const renderStart = Date.now();
+    const renderStart = performance.now();
     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 });
@@ -11418,57 +11422,41 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
     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`);
-    timingsMs.render = Date.now() - renderStart;
+    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 });
-    const probeStart = Date.now();
+    const probeStart = performance.now();
     currentStage = "probe";
     markStageStart("probe", "\uBAA8\uB378 \uC18D\uB3C4 \uD504\uB85C\uBE0C \uC218\uD589 \uC911");
     logStage("info", "probe", "start", "\uBAA8\uB378 \uC18D\uB3C4 \uD504\uB85C\uBE0C \uC2DC\uC791", { models, probeConcurrency });
     const probeImage = await pickRepresentativeImage(images);
     let probeDone = 0;
-    const probeResultsByIndex = await mapWithConcurrency(models, probeConcurrency, async (model, index) => {
-      const t0 = Date.now();
-      try {
-        await ocrImageViaNim({
-          imagePath: probeImage,
-          prompt: OCR_PROMPT2,
-          model,
-          maxTokens: modelMaxTokens[model] ?? 8192,
-          baseUrl,
-          keyPool,
-          timeoutMs,
-          maxRetries: 2,
-          logger,
-          stage: "probe"
-        });
-        const result = { model, durationMs: Date.now() - t0, success: true };
-        probeDone += 1;
-        markStageProgress("probe", Math.round(probeDone / models.length * 100), probeDone, models.length, `\uBAA8\uB378 \uD504\uB85C\uBE0C ${probeDone}/${models.length}`);
-        logStage("debug", "probe", "progress", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC9C4\uD589", { index: index + 1, total: models.length, model, result });
-        return result;
-      } catch (err) {
-        const result = {
-          model,
-          durationMs: Date.now() - t0,
-          success: false,
-          error: err instanceof Error ? err.message : String(err)
-        };
+    const probeRuns = startParallelProbeRuns({
+      models,
+      probeConcurrency,
+      probeImage,
+      modelMaxTokens,
+      baseUrl,
+      keyPool,
+      timeoutMs,
+      logger,
+      onProbeResult: ({ index, model, result }) => {
         probeDone += 1;
         markStageProgress("probe", Math.round(probeDone / models.length * 100), probeDone, models.length, `\uBAA8\uB378 \uD504\uB85C\uBE0C ${probeDone}/${models.length}`);
         logStage("debug", "probe", "progress", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC9C4\uD589", { index: index + 1, total: models.length, model, result });
-        return result;
       }
     });
-    const probeResults = probeResultsByIndex;
-    const selectedModel = chooseFastestModel(probeResults);
-    if (!selectedModel) throw new UnifiedOcrError("PROBE_FAILED", "probe", "\uC18D\uB3C4 \uD504\uB85C\uBE0C \uC2E4\uD328: \uC0AC\uC6A9 \uAC00\uB2A5\uD55C OCR \uBAA8\uB378\uC774 \uC5C6\uC2B5\uB2C8\uB2E4.");
-    const fallbackModelOrder = probeResults.filter((r) => r.success).sort((a, b) => a.durationMs - b.durationMs).map((r) => r.model);
-    timingsMs.probe = Date.now() - probeStart;
-    await updateModelCache(modelCachePath, probeResults);
+    const selected = await probeRuns.firstSuccess;
+    const selectedModel = selected.selectedModel;
+    const fallbackModelOrder = [selectedModel, ...models.filter((model) => model !== selectedModel)];
+    timingsMs.probe = elapsedMs(probeStart);
     markStageDone("probe", `\uD504\uB85C\uBE0C \uC644\uB8CC: ${selectedModel}`);
-    logStage("info", "probe", "done", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC644\uB8CC", { selectedModel, probeResults, elapsedMs: timingsMs.probe, modelCachePath });
-    const ocrStart = Date.now();
+    logStage("info", "probe", "done", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC644\uB8CC(\uCCAB \uC131\uACF5 \uBAA8\uB378 \uC6B0\uC120)", { selectedModel, firstDurationMs: selected.firstDurationMs, elapsedMs: timingsMs.probe });
+    const probeResultsPromise = probeRuns.allResults.then(async (results) => {
+      await updateModelCache(modelCachePath, results);
+      return results;
+    });
+    const ocrStart = performance.now();
     currentStage = "ocr";
     markStageStart("ocr", `OCR \uC9C4\uD589 \uC911 (${selectedModel})`);
     logStage("info", "ocr", "start", "\uD398\uC774\uC9C0 OCR \uC2DC\uC791", { selectedModel, pageCount: images.length });
@@ -11492,10 +11480,10 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
       markStageProgress("ocr", Math.round((i + 1) / images.length * 100), i + 1, images.length, `OCR ${i + 1}/${images.length}`);
       logStage("debug", "ocr", "progress", "\uD398\uC774\uC9C0 OCR \uC644\uB8CC", { page: i + 1, total: images.length });
     }
-    timingsMs.ocr = Date.now() - ocrStart;
+    timingsMs.ocr = elapsedMs(ocrStart);
     markStageDone("ocr", "OCR \uC644\uB8CC");
     logStage("info", "ocr", "done", "\uD398\uC774\uC9C0 OCR \uC644\uB8CC", { elapsedMs: timingsMs.ocr });
-    const proofStart = Date.now();
+    const proofStart = performance.now();
     currentStage = "proofread";
     markStageStart("proofread", "\uBE44\uD30C\uAD34 \uAD50\uC815 \uC9C4\uD589 \uC911");
     logStage("info", "proofread", "start", "\uBE44\uD30C\uAD34 \uAD50\uC815 \uC2DC\uC791", { pages: rawPagePaths.length });
@@ -11531,16 +11519,16 @@ ${rawMd}
       markStageProgress("proofread", Math.round((i + 1) / rawPagePaths.length * 100), i + 1, rawPagePaths.length, `\uAD50\uC815 ${i + 1}/${rawPagePaths.length}`);
       logStage("debug", "proofread", "progress", "\uD398\uC774\uC9C0 \uAD50\uC815 \uC644\uB8CC", { page: i + 1, total: rawPagePaths.length });
     }
-    timingsMs.proofread = Date.now() - proofStart;
+    timingsMs.proofread = elapsedMs(proofStart);
     markStageDone("proofread", "\uAD50\uC815 \uC644\uB8CC");
     logStage("info", "proofread", "done", "\uBE44\uD30C\uAD34 \uAD50\uC815 \uC644\uB8CC", { elapsedMs: timingsMs.proofread });
-    const mergeStart = Date.now();
+    const mergeStart = performance.now();
     currentStage = "merge";
     markStageStart("merge", "\uCD5C\uC885 Markdown \uBCD1\uD569 \uC911");
     logStage("info", "merge", "start", "\uCD5C\uC885 \uBCD1\uD569 \uC2DC\uC791", { pages: proofedPaths.length });
     const merged = await mergeMarkdownPages(proofedPaths);
     await writeFile(outputPath, merged, "utf-8");
-    timingsMs.merge = Date.now() - mergeStart;
+    timingsMs.merge = elapsedMs(mergeStart);
     markStageDone("merge", "\uBCD1\uD569 \uC644\uB8CC");
     logStage("info", "merge", "done", "\uCD5C\uC885 \uBCD1\uD569 \uC644\uB8CC", { outputPath, elapsedMs: timingsMs.merge });
     const report = {
@@ -11549,7 +11537,7 @@ ${rawMd}
       workspaceDir,
       selectedModel,
       probeImage,
-      probeResults,
+      probeResults: await probeResultsPromise,
       pageCount: images.length,
       keyHealth: keyPool.snapshot(),
       timingsMs,
@@ -11690,9 +11678,56 @@ async function mapWithConcurrency(items, concurrency, mapper) {
   await Promise.all(Array.from({ length: workerCount }, () => worker()));
   return results;
 }
-function chooseFastestModel(results) {
-  const ok = results.filter((r) => r.success).sort((a, b) => a.durationMs - b.durationMs);
-  return ok[0]?.model ?? null;
+function startParallelProbeRuns(input) {
+  let firstResolved = false;
+  let doneCount = 0;
+  let resolveFirst;
+  let rejectFirst;
+  const firstSuccess = new Promise((resolve4, reject) => {
+    resolveFirst = resolve4;
+    rejectFirst = reject;
+  });
+  let lastErr = "\uC18D\uB3C4 \uD504\uB85C\uBE0C \uC2E4\uD328: \uC0AC\uC6A9 \uAC00\uB2A5\uD55C OCR \uBAA8\uB378\uC774 \uC5C6\uC2B5\uB2C8\uB2E4.";
+  const allResults = mapWithConcurrency(input.models, input.probeConcurrency, async (model, index) => {
+    const t0 = performance.now();
+    try {
+      await ocrImageViaNim({
+        imagePath: input.probeImage,
+        prompt: OCR_PROMPT2,
+        model,
+        maxTokens: input.modelMaxTokens[model] ?? 8192,
+        baseUrl: input.baseUrl,
+        keyPool: input.keyPool,
+        timeoutMs: input.timeoutMs,
+        maxRetries: 2,
+        logger: input.logger,
+        stage: "probe"
+      });
+      const result = { model, durationMs: elapsedMs(t0), success: true };
+      input.onProbeResult?.({ index, model, result });
+      if (!firstResolved) {
+        firstResolved = true;
+        resolveFirst?.({ selectedModel: model, firstDurationMs: result.durationMs });
+      }
+      return result;
+    } catch (err) {
+      const result = {
+        model,
+        durationMs: elapsedMs(t0),
+        success: false,
+        error: err instanceof Error ? err.message : String(err)
+      };
+      lastErr = result.error ?? lastErr;
+      input.onProbeResult?.({ index, model, result });
+      return result;
+    } finally {
+      doneCount += 1;
+      if (doneCount === input.models.length && !firstResolved) {
+        rejectFirst?.(new UnifiedOcrError("PROBE_FAILED", "probe", `\uC18D\uB3C4 \uD504\uB85C\uBE0C \uC2E4\uD328: ${lastErr}`));
+      }
+    }
+  });
+  return { firstSuccess, allResults };
 }
 async function loadModelCache(path) {
   try {