@struktur/sdk 1.2.1 → 2.1.0

package/src/strategies/DoublePassAutoMergeStrategy.ts

@@ -84,6 +84,20 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
 
   async run(options: ExtractionOptions<T>): Promise<ExtractionResult<T>> {
     const debug = options.debug;
+    const { telemetry } = options;
+    
+    // Create strategy-level span
+    const strategySpan = telemetry?.startSpan({
+      name: "strategy.double-pass-auto-merge",
+      kind: "CHAIN",
+      attributes: {
+        "strategy.name": this.name,
+        "strategy.artifacts.count": options.artifacts.length,
+        "strategy.chunk_size": this.config.chunkSize,
+        "strategy.concurrency": this.config.concurrency,
+      },
+    });
+    
     const batches = getBatches(
       options.artifacts,
       {
@@ -91,11 +105,24 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
         maxImages: this.config.maxImages,
       },
       debug,
+      telemetry ?? undefined,
+      strategySpan,
     );
 
     const schema = serializeSchema(options.schema);
     const totalSteps = this.getEstimatedSteps(options.artifacts);
     let step = 1;
+    
+    // Create pass 1 span
+    const pass1Span = telemetry?.startSpan({
+      name: "struktur.pass_1",
+      kind: "CHAIN",
+      parentSpan: strategySpan,
+      attributes: {
+        "pass.number": 1,
+        "pass.type": "parallel_extraction",
+      },
+    });
 
     const tasks = batches.map((batch, index) => async () => {
       const prompt = buildExtractorPrompt(
@@ -114,6 +141,8 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
         strict: options.strict ?? this.config.strict,
         debug,
         callId: `double_pass_auto_1_batch_${index + 1}`,
+        telemetry: telemetry ?? undefined,
+        parentSpan: pass1Span,
       });
       step += 1;
       await options.events?.onStep?.({
@@ -145,6 +174,17 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       inputCount: results.length,
       strategy: this.name,
     });
+    
+    // Create smart merge span
+    const mergeSpan = telemetry?.startSpan({
+      name: "struktur.smart_merge",
+      kind: "CHAIN",
+      parentSpan: pass1Span,
+      attributes: {
+        "merge.strategy": "smart",
+        "merge.input_count": results.length,
+      },
+    });
 
     for (let i = 0; i < results.length; i++) {
       const result = results[i]!;
@@ -168,12 +208,54 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
           leftCount: leftArray,
           rightCount: rightArray,
         });
+        
+        // Record merge event in telemetry
+        if (mergeSpan && telemetry) {
+          telemetry.recordEvent(mergeSpan, {
+            type: "merge",
+            strategy: "smart",
+            inputCount: rightArray ?? 1,
+            outputCount: leftArray ?? 1,
+          });
+        }
       }
     }
 
     debug?.mergeComplete({ mergeId: "double_pass_auto_merge", success: true });
+    
+    // End merge span
+    if (mergeSpan && telemetry) {
+      telemetry.endSpan(mergeSpan, {
+        status: "ok",
+        output: merged,
+      });
+    }
 
     merged = dedupeArrays(merged);
+    
+    // Create exact dedupe span
+    const exactDedupeSpan = telemetry?.startSpan({
+      name: "struktur.exact_dedupe",
+      kind: "CHAIN",
+      parentSpan: pass1Span,
+      attributes: {
+        "dedupe.method": "exact_hashing",
+      },
+    });
+    
+    // End exact dedupe span
+    if (exactDedupeSpan && telemetry) {
+      telemetry.recordEvent(exactDedupeSpan, {
+        type: "merge",
+        strategy: "exact_hash_dedupe",
+        inputCount: Object.keys(merged).length,
+        outputCount: Object.keys(merged).length,
+      });
+      telemetry.endSpan(exactDedupeSpan, {
+        status: "ok",
+        output: merged,
+      });
+    }
 
     const dedupePrompt = buildDeduplicationPrompt(schema, merged);
 
@@ -181,6 +263,16 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       dedupeId: "double_pass_auto_dedupe",
       itemCount: Object.keys(merged).length,
     });
+    
+    // Create LLM dedupe span
+    const llmDedupeSpan = telemetry?.startSpan({
+      name: "struktur.llm_dedupe",
+      kind: "CHAIN",
+      parentSpan: pass1Span,
+      attributes: {
+        "dedupe.method": "llm",
+      },
+    });
 
     const dedupeResponse = await runWithRetries<{ keys: string[] }>({
       model: this.config.dedupeModel ?? this.config.model,
@@ -192,6 +284,8 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       strict: this.config.strict,
       debug,
       callId: "double_pass_auto_dedupe",
+      telemetry: telemetry ?? undefined,
+      parentSpan: llmDedupeSpan,
     });
 
     step += 1;
@@ -217,9 +311,41 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       duplicatesFound: dedupeResponse.data.keys.length,
       itemsRemoved: dedupeResponse.data.keys.length,
     });
+    
+    // End LLM dedupe span
+    if (llmDedupeSpan && telemetry) {
+      telemetry.recordEvent(llmDedupeSpan, {
+        type: "merge",
+        strategy: "llm_dedupe",
+        inputCount: Object.keys(merged).length,
+        outputCount: Object.keys(deduped).length,
+        deduped: dedupeResponse.data.keys.length,
+      });
+      telemetry.endSpan(llmDedupeSpan, {
+        status: "ok",
+        output: deduped,
+      });
+    }
+    
+    // End pass 1 span
+    telemetry?.endSpan(pass1Span!, {
+      status: "ok",
+      output: deduped,
+    });
 
     let currentData = deduped as T;
     const usages = [...results.map((r) => r.usage), dedupeResponse.usage];
+    
+    // Create pass 2 span
+    const pass2Span = telemetry?.startSpan({
+      name: "struktur.pass_2",
+      kind: "CHAIN",
+      parentSpan: strategySpan,
+      attributes: {
+        "pass.number": 2,
+        "pass.type": "sequential_refinement",
+      },
+    });
 
     for (const [index, batch] of batches.entries()) {
       const prompt = buildSequentialPrompt(
@@ -240,6 +366,8 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
         strict: this.config.strict,
         debug,
         callId: `double_pass_auto_2_batch_${index + 1}`,
+        telemetry: telemetry ?? undefined,
+        parentSpan: pass2Span,
       });
 
       currentData = result.data;
@@ -258,6 +386,18 @@ export class DoublePassAutoMergeStrategy<T> implements ExtractionStrategy<T> {
         strategy: this.name,
       });
     }
+    
+    // End pass 2 span
+    telemetry?.endSpan(pass2Span!, {
+      status: "ok",
+      output: currentData,
+    });
+    
+    // End strategy span
+    telemetry?.endSpan(strategySpan!, {
+      status: "ok",
+      output: currentData,
+    });
 
     return { data: currentData, usage: mergeUsage(usages) };
   }

package/src/strategies/DoublePassStrategy.ts

@@ -41,6 +41,20 @@ export class DoublePassStrategy<T> implements ExtractionStrategy<T> {
 
   async run(options: ExtractionOptions<T>): Promise<ExtractionResult<T>> {
     const debug = options.debug;
+    const { telemetry } = options;
+    
+    // Create strategy-level span
+    const strategySpan = telemetry?.startSpan({
+      name: "strategy.double-pass",
+      kind: "CHAIN",
+      attributes: {
+        "strategy.name": this.name,
+        "strategy.artifacts.count": options.artifacts.length,
+        "strategy.chunk_size": this.config.chunkSize,
+        "strategy.concurrency": this.config.concurrency,
+      },
+    });
+    
     const batches = getBatches(
       options.artifacts,
       {
@@ -48,11 +62,24 @@ export class DoublePassStrategy<T> implements ExtractionStrategy<T> {
         maxImages: this.config.maxImages,
       },
       debug,
+      telemetry ?? undefined,
+      strategySpan,
     );
 
     const schema = serializeSchema(options.schema);
     const totalSteps = this.getEstimatedSteps(options.artifacts);
     let step = 1;
+    
+    // Create pass 1 span
+    const pass1Span = telemetry?.startSpan({
+      name: "struktur.pass_1",
+      kind: "CHAIN",
+      parentSpan: strategySpan,
+      attributes: {
+        "pass.number": 1,
+        "pass.type": "parallel_extraction",
+      },
+    });
 
     const tasks = batches.map((batch, index) => async () => {
       const prompt = buildExtractorPrompt(
@@ -71,6 +98,8 @@ export class DoublePassStrategy<T> implements ExtractionStrategy<T> {
         strict: options.strict ?? this.config.strict,
         debug,
         callId: `double_pass_1_batch_${index + 1}`,
+        telemetry: telemetry ?? undefined,
+        parentSpan: pass1Span,
       });
       step += 1;
       await options.events?.onStep?.({
@@ -97,6 +126,17 @@ export class DoublePassStrategy<T> implements ExtractionStrategy<T> {
       inputCount: results.length,
       strategy: this.name,
     });
+    
+    // Create pass 1 merge span
+    const pass1MergeSpan = telemetry?.startSpan({
+      name: "struktur.pass_1_merge",
+      kind: "CHAIN",
+      parentSpan: pass1Span,
+      attributes: {
+        "merge.strategy": "parallel",
+        "merge.input_count": results.length,
+      },
+    });
 
     const mergePrompt = buildParallelMergerPrompt(
       schema,
@@ -113,6 +153,8 @@ export class DoublePassStrategy<T> implements ExtractionStrategy<T> {
       strict: this.config.strict,
       debug,
       callId: "double_pass_1_merge",
+      telemetry: telemetry ?? undefined,
+      parentSpan: pass1MergeSpan,
     });
 
     step += 1;
@@ -128,6 +170,37 @@ export class DoublePassStrategy<T> implements ExtractionStrategy<T> {
       strategy: this.name,
     });
     debug?.mergeComplete({ mergeId: "double_pass_1_merge", success: true });
+    
+    // End pass 1 merge span
+    if (pass1MergeSpan && telemetry) {
+      telemetry.recordEvent(pass1MergeSpan, {
+        type: "merge",
+        strategy: "parallel",
+        inputCount: results.length,
+        outputCount: 1,
+      });
+      telemetry.endSpan(pass1MergeSpan, {
+        status: "ok",
+        output: merged.data,
+      });
+    }
+    
+    // End pass 1 span
+    telemetry?.endSpan(pass1Span!, {
+      status: "ok",
+      output: merged.data,
+    });
+    
+    // Create pass 2 span
+    const pass2Span = telemetry?.startSpan({
+      name: "struktur.pass_2",
+      kind: "CHAIN",
+      parentSpan: strategySpan,
+      attributes: {
+        "pass.number": 2,
+        "pass.type": "sequential_refinement",
+      },
+    });
 
     let currentData = merged.data;
     const usages = [...results.map((r) => r.usage), merged.usage];
@@ -151,6 +224,8 @@ export class DoublePassStrategy<T> implements ExtractionStrategy<T> {
         strict: this.config.strict,
         debug,
         callId: `double_pass_2_batch_${index + 1}`,
+        telemetry: telemetry ?? undefined,
+        parentSpan: pass2Span,
       });
 
       currentData = result.data;
@@ -169,6 +244,18 @@ export class DoublePassStrategy<T> implements ExtractionStrategy<T> {
         strategy: this.name,
       });
     }
+    
+    // End pass 2 span
+    telemetry?.endSpan(pass2Span!, {
+      status: "ok",
+      output: currentData,
+    });
+    
+    // End strategy span
+    telemetry?.endSpan(strategySpan!, {
+      status: "ok",
+      output: currentData,
+    });
 
     return { data: currentData, usage: mergeUsage(usages) };
   }

package/src/strategies/ParallelAutoMergeStrategy.ts

@@ -83,6 +83,20 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
 
   async run(options: ExtractionOptions<T>): Promise<ExtractionResult<T>> {
     const debug = options.debug;
+    const { telemetry } = options;
+    
+    // Create strategy-level span
+    const strategySpan = telemetry?.startSpan({
+      name: "strategy.parallel-auto-merge",
+      kind: "CHAIN",
+      attributes: {
+        "strategy.name": this.name,
+        "strategy.artifacts.count": options.artifacts.length,
+        "strategy.chunk_size": this.config.chunkSize,
+        "strategy.concurrency": this.config.concurrency,
+      },
+    });
+    
     const batches = getBatches(
       options.artifacts,
       {
@@ -90,6 +104,8 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
         maxImages: this.config.maxImages,
       },
       debug,
+      telemetry ?? undefined,
+      strategySpan,
     );
 
     const schema = serializeSchema(options.schema);
@@ -113,6 +129,8 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
         strict: options.strict ?? this.config.strict,
         debug,
         callId: `parallel_auto_batch_${index + 1}`,
+        telemetry: telemetry ?? undefined,
+        parentSpan: strategySpan,
       });
       step += 1;
       await options.events?.onStep?.({
@@ -144,6 +162,17 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       inputCount: results.length,
       strategy: this.name,
     });
+    
+    // Create smart merge span
+    const mergeSpan = telemetry?.startSpan({
+      name: "struktur.smart_merge",
+      kind: "CHAIN",
+      parentSpan: strategySpan,
+      attributes: {
+        "merge.strategy": "smart",
+        "merge.input_count": results.length,
+      },
+    });
 
     for (let i = 0; i < results.length; i++) {
       const result = results[i]!;
@@ -169,6 +198,16 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
           leftCount: leftArray,
           rightCount: rightArray,
         });
+        
+        // Record merge event in telemetry
+        if (mergeSpan && telemetry) {
+          telemetry.recordEvent(mergeSpan, {
+            type: "merge",
+            strategy: "smart",
+            inputCount: rightArray ?? 1,
+            outputCount: leftArray ?? 1,
+          });
+        }
       }
     }
 
@@ -176,8 +215,40 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       mergeId: "parallel_auto_smart_merge",
       success: true,
     });
+    
+    // End merge span
+    if (mergeSpan && telemetry) {
+      telemetry.endSpan(mergeSpan, {
+        status: "ok",
+        output: merged,
+      });
+    }
 
     merged = dedupeArrays(merged);
+    
+    // Create exact dedupe span
+    const exactDedupeSpan = telemetry?.startSpan({
+      name: "struktur.exact_dedupe",
+      kind: "CHAIN",
+      parentSpan: strategySpan,
+      attributes: {
+        "dedupe.method": "exact_hashing",
+      },
+    });
+    
+    // End exact dedupe span
+    if (exactDedupeSpan && telemetry) {
+      telemetry.recordEvent(exactDedupeSpan, {
+        type: "merge",
+        strategy: "exact_hash_dedupe",
+        inputCount: Object.keys(merged).length,
+        outputCount: Object.keys(merged).length,
+      });
+      telemetry.endSpan(exactDedupeSpan, {
+        status: "ok",
+        output: merged,
+      });
+    }
 
     const dedupePrompt = buildDeduplicationPrompt(schema, merged);
 
@@ -185,6 +256,16 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       dedupeId: "parallel_auto_dedupe",
       itemCount: Object.keys(merged).length,
     });
+    
+    // Create LLM dedupe span
+    const llmDedupeSpan = telemetry?.startSpan({
+      name: "struktur.llm_dedupe",
+      kind: "CHAIN",
+      parentSpan: strategySpan,
+      attributes: {
+        "dedupe.method": "llm",
+      },
+    });
 
     const dedupeResponse = await runWithRetries<{ keys: string[] }>({
       model: this.config.dedupeModel ?? this.config.model,
@@ -196,6 +277,8 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       strict: this.config.strict,
       debug,
       callId: "parallel_auto_dedupe",
+      telemetry: telemetry ?? undefined,
+      parentSpan: llmDedupeSpan,
     });
 
     step += 1;
@@ -221,6 +304,27 @@ export class ParallelAutoMergeStrategy<T> implements ExtractionStrategy<T> {
       duplicatesFound: dedupeResponse.data.keys.length,
       itemsRemoved: dedupeResponse.data.keys.length,
     });
+    
+    // End LLM dedupe span
+    if (llmDedupeSpan && telemetry) {
+      telemetry.recordEvent(llmDedupeSpan, {
+        type: "merge",
+        strategy: "llm_dedupe",
+        inputCount: Object.keys(merged).length,
+        outputCount: Object.keys(deduped).length,
+        deduped: dedupeResponse.data.keys.length,
+      });
+      telemetry.endSpan(llmDedupeSpan, {
+        status: "ok",
+        output: deduped,
+      });
+    }
+
+    // End strategy span
+    telemetry?.endSpan(strategySpan!, {
+      status: "ok",
+      output: deduped,
+    });
 
     return {
       data: deduped as T,

package/src/strategies/ParallelStrategy.ts

@@ -40,6 +40,20 @@ export class ParallelStrategy<T> implements ExtractionStrategy<T> {
 
   async run(options: ExtractionOptions<T>): Promise<ExtractionResult<T>> {
     const debug = options.debug;
+    const { telemetry } = options;
+    
+    // Create strategy-level span
+    const strategySpan = telemetry?.startSpan({
+      name: "strategy.parallel",
+      kind: "CHAIN",
+      attributes: {
+        "strategy.name": this.name,
+        "strategy.artifacts.count": options.artifacts.length,
+        "strategy.chunk_size": this.config.chunkSize,
+        "strategy.concurrency": this.config.concurrency,
+      },
+    });
+    
     const batches = getBatches(
       options.artifacts,
       {
@@ -47,6 +61,8 @@ export class ParallelStrategy<T> implements ExtractionStrategy<T> {
         maxImages: this.config.maxImages,
       },
       debug,
+      telemetry ?? undefined,
+      strategySpan,
     );
 
     const schema = serializeSchema(options.schema);
@@ -83,6 +99,8 @@ export class ParallelStrategy<T> implements ExtractionStrategy<T> {
         strict: options.strict ?? this.config.strict,
         debug,
         callId: `parallel_batch_${index + 1}`,
+        telemetry: telemetry ?? undefined,
+        parentSpan: strategySpan,
       });
       // Emit progress after batch completes (if there are more batches)
       const completedIndex = index + 1;
@@ -113,6 +131,17 @@ export class ParallelStrategy<T> implements ExtractionStrategy<T> {
       inputCount: results.length,
       strategy: this.name,
     });
+    
+    // Create merge span
+    const mergeSpan = telemetry?.startSpan({
+      name: "struktur.merge",
+      kind: "CHAIN",
+      parentSpan: strategySpan,
+      attributes: {
+        "merge.strategy": "parallel",
+        "merge.input_count": results.length,
+      },
+    });
 
     const mergePrompt = buildParallelMergerPrompt(
       schema,
@@ -129,6 +158,8 @@ export class ParallelStrategy<T> implements ExtractionStrategy<T> {
       strict: this.config.strict,
       debug,
       callId: "parallel_merge",
+      telemetry: telemetry ?? undefined,
+      parentSpan: mergeSpan,
     });
 
     step += 1;
@@ -144,6 +175,26 @@ export class ParallelStrategy<T> implements ExtractionStrategy<T> {
       strategy: this.name,
     });
     debug?.mergeComplete({ mergeId: "parallel_merge", success: true });
+    
+    // End merge span
+    if (mergeSpan && telemetry) {
+      telemetry.recordEvent(mergeSpan, {
+        type: "merge",
+        strategy: "parallel",
+        inputCount: results.length,
+        outputCount: 1,
+      });
+      telemetry.endSpan(mergeSpan, {
+        status: "ok",
+        output: merged.data,
+      });
+    }
+    
+    // End strategy span
+    telemetry?.endSpan(strategySpan!, {
+      status: "ok",
+      output: merged.data,
+    });
 
     return {
       data: merged.data,