@clazic/kordoc 2.4.12 → 2.4.14

package/dist/{chunk-5R37N6KE.js → chunk-X6NIA6BK.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 
 // src/utils.ts
-var VERSION = true ? "2.4.12" : "0.0.0-dev";
+var VERSION = true ? "2.4.13" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -105,4 +105,4 @@ export {
   classifyError,
   normalizeKordocError
 };
-//# sourceMappingURL=chunk-5R37N6KE.js.map
+//# sourceMappingURL=chunk-X6NIA6BK.js.map

package/dist/cli.js

@@ -4,12 +4,12 @@ import {
   markdownToHwpx,
   markdownToXlsx,
   parse
-} from "./chunk-KJEZPVEK.js";
+} from "./chunk-6CADPLGJ.js";
 import "./chunk-YW5G6BCJ.js";
 import {
   VERSION,
   toArrayBuffer
-} from "./chunk-5R37N6KE.js";
+} from "./chunk-X6NIA6BK.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-XLLXVB7V.js");
+      const { sanitizeError } = await import("./utils-EK3CPEZG.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-XLLXVB7V.js");
+    const { sanitizeError } = await import("./utils-EK3CPEZG.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-SOMS2KR7.js");
+  const { watchDirectory } = await import("./watch-NSWBVKQZ.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 ? "2.4.12" : "0.0.0-dev";
+var VERSION = true ? "2.4.13" : "0.0.0-dev";
 function toArrayBuffer(buf) {
   if (buf.byteOffset === 0 && buf.byteLength === buf.buffer.byteLength) {
     return buf.buffer;
@@ -11396,6 +11396,7 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
   const maxRetriesPerPage = options.maxRetriesPerPage ?? 5;
   const dpi = options.dpi ?? 300;
   const modelsInput = options.modelCandidates?.length ? options.modelCandidates : DEFAULT_MODELS;
+  const probeConcurrency = Math.max(1, Math.floor(options.probeConcurrency ?? Math.min(3, modelsInput.length)));
   const modelCache = await loadModelCache(modelCachePath);
   const models = sortModelsByCache(modelsInput, modelCache);
   const modelMaxTokens = { ...DEFAULT_MODEL_MAX_TOKENS, ...options.modelMaxTokens ?? {} };
@@ -11446,44 +11447,34 @@ async function runUnifiedOcrPipeline(inputPath, options = {}) {
     const probeStart = Date.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 });
+    logStage("info", "probe", "start", "\uBAA8\uB378 \uC18D\uB3C4 \uD504\uB85C\uBE0C \uC2DC\uC791", { models, probeConcurrency });
     const probeImage = await pickRepresentativeImage(images);
-    const probeResults = [];
-    for (let i = 0; i < models.length; i++) {
-      const model = models[i];
-      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"
-        });
-        probeResults.push({ model, durationMs: Date.now() - t0, success: true });
-      } catch (err) {
-        probeResults.push({
-          model,
-          durationMs: Date.now() - t0,
-          success: false,
-          error: err instanceof Error ? err.message : String(err)
-        });
+    let probeDone = 0;
+    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 });
       }
-      markStageProgress("probe", Math.round((i + 1) / models.length * 100), i + 1, models.length, `\uBAA8\uB378 \uD504\uB85C\uBE0C ${i + 1}/${models.length}`);
-      logStage("debug", "probe", "progress", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC9C4\uD589", { index: i + 1, total: models.length, model, result: probeResults.at(-1) });
-    }
-    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);
+    });
+    const selected = await probeRuns.firstSuccess;
+    const selectedModel = selected.selectedModel;
+    const fallbackModelOrder = [selectedModel, ...models.filter((model) => model !== selectedModel)];
     timingsMs.probe = Date.now() - probeStart;
-    await updateModelCache(modelCachePath, probeResults);
     markStageDone("probe", `\uD504\uB85C\uBE0C \uC644\uB8CC: ${selectedModel}`);
-    logStage("info", "probe", "done", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC644\uB8CC", { selectedModel, probeResults, elapsedMs: timingsMs.probe, modelCachePath });
+    logStage("info", "probe", "done", "\uBAA8\uB378 \uD504\uB85C\uBE0C \uC644\uB8CC(\uCCAB \uC131\uACF5 \uBAA8\uB378 \uC6B0\uC120)", { selectedModel, firstDurationMs: selected.firstDurationMs, elapsedMs: timingsMs.probe });
+    const probeResultsPromise = probeRuns.allResults.then(async (results) => {
+      await updateModelCache(modelCachePath, results);
+      return results;
+    });
     const ocrStart = Date.now();
     currentStage = "ocr";
     markStageStart("ocr", `OCR \uC9C4\uD589 \uC911 (${selectedModel})`);
@@ -11565,7 +11556,7 @@ ${rawMd}
       workspaceDir,
       selectedModel,
       probeImage,
-      probeResults,
+      probeResults: await probeResultsPromise,
       pageCount: images.length,
       keyHealth: keyPool.snapshot(),
       timingsMs,
@@ -11691,9 +11682,71 @@ async function pickRepresentativeImage(images) {
   use.sort((a, b) => a.size - b.size);
   return use[Math.floor(use.length / 2)].path;
 }
-function chooseFastestModel(results) {
-  const ok = results.filter((r) => r.success).sort((a, b) => a.durationMs - b.durationMs);
-  return ok[0]?.model ?? null;
+async function mapWithConcurrency(items, concurrency, mapper) {
+  const results = new Array(items.length);
+  let nextIndex = 0;
+  async function worker() {
+    while (true) {
+      const idx = nextIndex;
+      if (idx >= items.length) return;
+      nextIndex += 1;
+      results[idx] = await mapper(items[idx], idx);
+    }
+  }
+  const workerCount = Math.max(1, Math.min(concurrency, items.length));
+  await Promise.all(Array.from({ length: workerCount }, () => worker()));
+  return results;
+}
+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 = Date.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: Date.now() - 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: Date.now() - 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 {