补充调试信息，记录包括各个阶段的比如query分析结果检索表达式各阶段耗时 ES搜索的检索表达式

tangwang
1 parent edd38328
Showing 5 changed files with 133 additions and 54 deletions Show diff stats
api/models.py
api/routes/search.py
frontend/index.html
frontend/static/js/app.js
search/searcher.py
@@ -17,6 +17,7 @@ class SearchRequest(BaseModel):
     aggregations: Optional[Dict[str, Any]] = Field(None, description="Aggregation specifications")
     sort_by: Optional[str] = Field(None, description="Sort field name")
     sort_order: Optional[str] = Field("desc", description="Sort order: 'asc' or 'desc'")
+    debug: bool = Field(False, description="Enable debug information output")
  
  
 class ImageSearchRequest(BaseModel):
@@ -35,6 +36,7 @@ class SearchResponse(BaseModel):
     aggregations: Dict[str, Any] = Field(default_factory=dict, description="Aggregation results")
     query_info: Dict[str, Any] = Field(default_factory=dict, description="Query processing information")
     performance_info: Optional[Dict[str, Any]] = Field(None, description="Detailed performance timing information")
+    debug_info: Optional[Dict[str, Any]] = Field(None, description="Debug information (only when debug=True)")
  
  
 class DocumentResponse(BaseModel):
@@ -71,7 +71,8 @@ async def search(request: SearchRequest, http_request: Request):
             context=context,
             aggregations=request.aggregations,
             sort_by=request.sort_by,
-            sort_order=request.sort_order
+            sort_order=request.sort_order,
+            debug=request.debug
         )
  
         # Include performance summary in response
@@ -85,7 +86,8 @@ async def search(request: SearchRequest, http_request: Request):
             took_ms=result.took_ms,
             aggregations=result.aggregations,
             query_info=result.query_info,
-            performance_info=performance_summary
+            performance_info=performance_summary,
+            debug_info=result.debug_info
         )
  
     except Exception as e:
@@ -109,10 +109,10 @@
         <!-- Pagination -->
         <div class="pagination" id="pagination" style="display: none;"></div>
  
-        <!-- Debug Info (Collapsible) -->
+        <!-- Debug Info -->
         <details class="debug-info">
-            <summary>Debug Info (Query Details)</summary>
-            <div id="queryInfo" class="query-info-content"></div>
+            <summary>调试信息</summary>
+            <div id="debugInfo"></div>
         </details>
     </div>
  
@@ -13,12 +13,15 @@ let state = {
     sortBy: '',
     sortOrder: 'desc',
     aggregations: null,
-    lastSearchData: null
+    lastSearchData: null,
+    debug: true  // Always enable debug mode for test frontend
 };
  
 // Initialize
 document.addEventListener('DOMContentLoaded', function() {
     console.log('SearchEngine loaded');
+    console.log('Debug mode: always enabled (test frontend)');
+    
     document.getElementById('searchInput').focus();
 });
  
@@ -100,7 +103,8 @@ async function performSearch(page = 1) {
                 filters: Object.keys(state.filters).length > 0 ? state.filters : null,
                 aggregations: aggregations,
                 sort_by: state.sortBy || null,
-                sort_order: state.sortOrder
+                sort_order: state.sortOrder,
+                debug: state.debug
             })
         });
  
@@ -116,7 +120,7 @@ async function performSearch(page = 1) {
         displayResults(data);
         displayAggregations(data.aggregations);
         displayPagination();
-        displayQueryInfo(data.query_info);
+        displayDebugInfo(data);
         updateProductCount(data.total);
         updateClearFiltersButton();
  
@@ -483,58 +487,92 @@ function goToPage(page) {
     window.scrollTo({ top: 0, behavior: 'smooth' });
 }
  
-// Display query info
-function displayQueryInfo(queryInfo) {
-    if (!queryInfo) return;
+// Display debug info
+function displayDebugInfo(data) {
+    const debugInfoDiv = document.getElementById('debugInfo');
+    
+    if (!state.debug || !data.debug_info) {
+        // If debug mode is off or no debug info, show basic query info
+        if (data.query_info) {
+            let html = '<div style="padding: 10px;">';
+            html += `<div><strong>查询:</strong> ${escapeHtml(data.query_info.original_query || 'N/A')}</div>`;
+            html += `<div><strong>语言:</strong> ${getLanguageName(data.query_info.detected_language)}</div>`;
+            html += '</div>';
+            debugInfoDiv.innerHTML = html;
+        } else {
+            debugInfoDiv.innerHTML = '';
+        }
+        return;
+    }
  
-    const queryInfoDiv = document.getElementById('queryInfo');
+    // Display comprehensive debug info when debug mode is on
+    const debugInfo = data.debug_info;
+    let html = '<div style="padding: 10px; font-family: monospace; font-size: 12px;">';
+    
+    // Query Analysis
+    if (debugInfo.query_analysis) {
+        html += '<div style="margin-bottom: 15px;"><strong style="font-size: 14px;">查询分析:</strong>';
+        html += `<div>原始查询: ${escapeHtml(debugInfo.query_analysis.original_query || 'N/A')}</div>`;
+        html += `<div>标准化: ${escapeHtml(debugInfo.query_analysis.normalized_query || 'N/A')}</div>`;
+        html += `<div>重写后: ${escapeHtml(debugInfo.query_analysis.rewritten_query || 'N/A')}</div>`;
+        html += `<div>检测语言: ${getLanguageName(debugInfo.query_analysis.detected_language)}</div>`;
+        html += `<div>域: ${escapeHtml(debugInfo.query_analysis.domain || 'default')}</div>`;
+        html += `<div>简单查询: ${debugInfo.query_analysis.is_simple_query ? '是' : '否'}</div>`;
+        
+        if (debugInfo.query_analysis.translations && Object.keys(debugInfo.query_analysis.translations).length > 0) {
+            html += '<div>翻译: ';
+            for (const [lang, translation] of Object.entries(debugInfo.query_analysis.translations)) {
+                if (translation) {
+                    html += `${getLanguageName(lang)}: ${escapeHtml(translation)}; `;
+                }
+            }
+            html += '</div>';
+        }
+        
+        if (debugInfo.query_analysis.boolean_ast) {
+            html += `<div>布尔AST: ${escapeHtml(debugInfo.query_analysis.boolean_ast)}</div>`;
+        }
+        
+        html += `<div>向量嵌入: ${debugInfo.query_analysis.has_vector ? '已启用' : '未启用'}</div>`;
+        html += '</div>';
+    }
  
-    let html = '<div class="info-grid">';
+    // Feature Flags
+    if (debugInfo.feature_flags) {
+        html += '<div style="margin-bottom: 15px;"><strong style="font-size: 14px;">功能开关:</strong>';
+        html += `<div>翻译: ${debugInfo.feature_flags.translation_enabled ? '启用' : '禁用'}</div>`;
+        html += `<div>嵌入: ${debugInfo.feature_flags.embedding_enabled ? '启用' : '禁用'}</div>`;
+        html += `<div>重排序: ${debugInfo.feature_flags.rerank_enabled ? '启用' : '禁用'}</div>`;
+        html += '</div>';
+    }
  
-    html += `
-        <div class="info-item">
-            <strong>Original Query</strong>
-            <span>${escapeHtml(queryInfo.original_query || 'N/A')}</span>
-        </div>
-        <div class="info-item">
-            <strong>Rewritten Query</strong>
-            <span>${escapeHtml(queryInfo.rewritten_query || 'N/A')}</span>
-        </div>
-        <div class="info-item">
-            <strong>Detected Language</strong>
-            <span>${getLanguageName(queryInfo.detected_language)}</span>
-        </div>
-        <div class="info-item">
-            <strong>Domain</strong>
-            <span>${escapeHtml(queryInfo.domain || 'default')}</span>
-        </div>
-    `;
+    // ES Response
+    if (debugInfo.es_response) {
+        html += '<div style="margin-bottom: 15px;"><strong style="font-size: 14px;">ES响应:</strong>';
+        html += `<div>耗时: ${debugInfo.es_response.took_ms}ms</div>`;
+        html += `<div>总命中数: ${debugInfo.es_response.total_hits}</div>`;
+        html += `<div>最高分: ${debugInfo.es_response.max_score?.toFixed(3) || 0}</div>`;
+        html += '</div>';
+    }
  
-    if (queryInfo.translations && Object.keys(queryInfo.translations).length > 0) {
-        for (const [lang, translation] of Object.entries(queryInfo.translations)) {
-            if (translation) {
-                html += `
-                    <div class="info-item">
-                        <strong>Translation (${getLanguageName(lang)})</strong>
-                        <span>${escapeHtml(translation)}</span>
-                    </div>
-                `;
-            }
+    // Stage Timings
+    if (debugInfo.stage_timings) {
+        html += '<div style="margin-bottom: 15px;"><strong style="font-size: 14px;">各阶段耗时:</strong>';
+        for (const [stage, duration] of Object.entries(debugInfo.stage_timings)) {
+            html += `<div>${stage}: ${duration.toFixed(2)}ms</div>`;
         }
+        html += '</div>';
     }
  
-    if (queryInfo.has_vector) {
-        html += `
-            <div class="info-item">
-                <strong>Semantic Search</strong>
-                <span style="color: #27ae60;">✓ Enabled (Vector)</span>
-            </div>
-        `;
+    // ES Query
+    if (debugInfo.es_query) {
+        html += '<div style="margin-bottom: 15px;"><strong style="font-size: 14px;">ES查询DSL:</strong>';
+        html += `<pre style="background: #f5f5f5; padding: 10px; overflow: auto; max-height: 400px;">${escapeHtml(JSON.stringify(debugInfo.es_query, null, 2))}</pre>`;
+        html += '</div>';
     }
  
     html += '</div>';
-    
-    queryInfoDiv.innerHTML = html;
+    debugInfoDiv.innerHTML = html;
 }
  
 // Helper functions
@@ -28,7 +28,8 @@ class SearchResult:
         max_score: float,
         took_ms: int,
         aggregations: Optional[Dict[str, Any]] = None,
-        query_info: Optional[Dict[str, Any]] = None
+        query_info: Optional[Dict[str, Any]] = None,
+        debug_info: Optional[Dict[str, Any]] = None
     ):
         self.hits = hits
         self.total = total
@@ -36,10 +37,11 @@ class SearchResult:
         self.took_ms = took_ms
         self.aggregations = aggregations or {}
         self.query_info = query_info or {}
+        self.debug_info = debug_info
  
     def to_dict(self) -> Dict[str, Any]:
         """Convert to dictionary representation."""
-        return {
+        result = {
             "hits": self.hits,
             "total": self.total,
             "max_score": self.max_score,
@@ -47,6 +49,9 @@ class SearchResult:
             "aggregations": self.aggregations,
             "query_info": self.query_info
         }
+        if self.debug_info is not None:
+            result["debug_info"] = self.debug_info
+        return result
  
  
 class Searcher:
@@ -106,7 +111,8 @@ class Searcher:
         context: Optional[RequestContext] = None,
         aggregations: Optional[Dict[str, Any]] = None,
         sort_by: Optional[str] = None,
-        sort_order: Optional[str] = "desc"
+        sort_order: Optional[str] = "desc",
+        debug: bool = False
     ) -> SearchResult:
         """
         Execute search query.
@@ -121,6 +127,7 @@ class Searcher:
             aggregations: Aggregation specifications for faceted search
             sort_by: Field name for sorting
             sort_order: Sort order: 'asc' or 'desc'
+            debug: Enable debug information output
  
         Returns:
             SearchResult object
@@ -408,6 +415,35 @@ class Searcher:
         total_duration = context.end_stage(RequestContextStage.TOTAL)
         context.performance_metrics.total_duration = total_duration
  
+        # Collect debug information if requested
+        debug_info = None
+        if debug:
+            debug_info = {
+                "query_analysis": {
+                    "original_query": context.query_analysis.original_query,
+                    "normalized_query": context.query_analysis.normalized_query,
+                    "rewritten_query": context.query_analysis.rewritten_query,
+                    "detected_language": context.query_analysis.detected_language,
+                    "translations": context.query_analysis.translations,
+                    "has_vector": context.query_analysis.query_vector is not None,
+                    "is_simple_query": context.query_analysis.is_simple_query,
+                    "boolean_ast": context.get_intermediate_result('boolean_ast'),
+                    "domain": context.query_analysis.domain
+                },
+                "es_query": context.get_intermediate_result('es_query', {}),
+                "es_response": {
+                    "took_ms": es_response.get('took', 0),
+                    "total_hits": total_value,
+                    "max_score": max_score,
+                    "shards": es_response.get('_shards', {})
+                },
+                "feature_flags": context.metadata.get('feature_flags', {}),
+                "stage_timings": {
+                    k: round(v, 2) for k, v in context.performance_metrics.stage_timings.items()
+                },
+                "search_params": context.metadata.get('search_params', {})
+            }
+
         # Build result
         result = SearchResult(
             hits=hits,
@@ -415,7 +451,8 @@ class Searcher:
             max_score=max_score,
             took_ms=int(total_duration),
             aggregations=aggregations,
-            query_info=parsed_query.to_dict()
+            query_info=parsed_query.to_dict(),
+            debug_info=debug_info
         )
  
         # Log complete performance summary