@clazic/kordoc 2.4.12 → 2.4.14

package/dist/index.d.cts

@@ -433,6 +433,7 @@ interface UnifiedOcrOptions {
     stageWeights?: Partial<Record<UnifiedStage, number>>;
     timeoutMs?: number;
     maxRetriesPerPage?: number;
+    probeConcurrency?: number;
     logger?: Logger;
     runId?: string;
 }

package/dist/index.d.ts

@@ -433,6 +433,7 @@ interface UnifiedOcrOptions {
     stageWeights?: Partial<Record<UnifiedStage, number>>;
     timeoutMs?: number;
     maxRetriesPerPage?: number;
+    probeConcurrency?: number;
     logger?: Logger;
     runId?: string;
 }

package/dist/index.js

@@ -3115,7 +3115,7 @@ import JSZip2 from "jszip";
 import { DOMParser } from "@xmldom/xmldom";
 
 // src/utils.ts
-var VERSION = true ? "2.4.12" : "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;
@@ -11373,6 +11373,7 @@ 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 ?? {} };
@@ -11423,44 +11424,34 @@ 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})`);
@@ -11542,7 +11533,7 @@ ${rawMd}
       workspaceDir,
       selectedModel,
       probeImage,
-      probeResults,
+      probeResults: await probeResultsPromise,
       pageCount: images.length,
       keyHealth: keyPool.snapshot(),
       timingsMs,
@@ -11668,9 +11659,71 @@ 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 {