@heripo/document-processor 0.1.14 → 0.1.15

package/dist/index.js

@@ -3039,6 +3039,9 @@ var PageRangeParser = class extends VisionLLMComponent {
       };
     }
     const sampledPages = /* @__PURE__ */ new Set();
+    let consecutiveAllNullAttempts = 0;
+    const MAX_ALL_NULL_ATTEMPTS = 3;
+    const allSamples = [];
     for (let attempt = 0; attempt <= this.MAX_PATTERN_RETRIES; attempt++) {
       const samplePageNos = this.selectRandomSamples(
         pageNos,
@@ -3059,6 +3062,20 @@ var PageRangeParser = class extends VisionLLMComponent {
       );
       usageList.push(result.usage);
       const samples = result.samples;
+      allSamples.push(...samples);
+      const allNull = samples.every((s) => s.startPageNo === null);
+      if (allNull) {
+        consecutiveAllNullAttempts++;
+        if (consecutiveAllNullAttempts >= MAX_ALL_NULL_ATTEMPTS) {
+          this.log(
+            "warn",
+            `All samples returned null for ${MAX_ALL_NULL_ATTEMPTS} consecutive attempts, pages likely have no page numbers`
+          );
+          break;
+        }
+      } else {
+        consecutiveAllNullAttempts = 0;
+      }
       const pattern = this.detectPattern(samples);
       if (pattern.pattern !== "unknown" /* UNKNOWN */) {
         this.log(
@@ -3075,9 +3092,29 @@ var PageRangeParser = class extends VisionLLMComponent {
         `Pattern detection failed, attempt ${attempt + 1}/${this.MAX_PATTERN_RETRIES + 1}`
       );
     }
-    throw new PageRangeParseError(
-      `Failed to detect page pattern after ${this.MAX_PATTERN_RETRIES + 1} attempts for size group with ${pageNos.length} pages`
+    const allValidSamples = allSamples.filter((s) => s.startPageNo !== null);
+    if (allValidSamples.length >= 2) {
+      const finalPattern = this.detectPattern(allSamples);
+      if (finalPattern.pattern !== "unknown" /* UNKNOWN */) {
+        this.log(
+          "info",
+          `Pattern detected from accumulated samples: ${finalPattern.pattern}`
+        );
+        return {
+          pageRangeMap: this.applyPattern(pageNos, finalPattern),
+          usage: usageList
+        };
+      }
+    }
+    this.log(
+      "warn",
+      `Could not detect page pattern after ${consecutiveAllNullAttempts >= MAX_ALL_NULL_ATTEMPTS ? consecutiveAllNullAttempts : this.MAX_PATTERN_RETRIES + 1} attempts for group with ${pageNos.length} pages, marking as failed`
     );
+    const fallbackMap = {};
+    for (const pageNo of pageNos) {
+      fallbackMap[pageNo] = { startPageNo: 0, endPageNo: 0 };
+    }
+    return { pageRangeMap: fallbackMap, usage: usageList };
   }
   /**
    * Select random samples from page numbers