From 85f29d8b49c6f8bce5883fcefc2ce662b0457f47 Mon Sep 17 00:00:00 2001
From: William Tso <vitalitymailg@gmail.com>
Date: Thu, 10 Apr 2025 18:31:24 +0800
Subject: [PATCH] click subpath match

---
 app/api/events/utm/route.ts | 29 +++++++++++++++++------------
 lib/clickhouse.ts           | 26 ++++++++++++++++----------
 2 files changed, 33 insertions(+), 22 deletions(-)

diff --git a/app/api/events/utm/route.ts b/app/api/events/utm/route.ts
index aa1cd51..052fda8 100644
--- a/app/api/events/utm/route.ts
+++ b/app/api/events/utm/route.ts
@@ -66,23 +66,28 @@ export async function GET(request: NextRequest) {
       conditions.push(`link_id = '${linkId}'`);
     }
     
-    // 添加子路径筛选
-    if (subpath) {
+    // 添加子路径筛选 - 使用更精确的匹配方式
+    if (subpath && subpath.trim() !== '') {
       console.log('====== UTM API SUBPATH DEBUG ======');
       console.log('Raw subpath param:', subpath);
-      console.log('Subpath type:', typeof subpath);
-      console.log('Subpath length:', subpath.length);
-      console.log('Subpath chars:', Array.from(subpath).map(c => c.charCodeAt(0)));
       
-      // 确保没有前导斜杠，避免双斜杠问题
-      const cleanSubpath = subpath.startsWith('/') 
-        ? subpath.substring(1) 
-        : subpath;
-        
+      // 清理并准备subpath值
+      let cleanSubpath = subpath.trim();
+      // 移除开头的斜杠以便匹配
+      if (cleanSubpath.startsWith('/')) {
+        cleanSubpath = cleanSubpath.substring(1);
+      }
+      // 移除结尾的斜杠以便匹配
+      if (cleanSubpath.endsWith('/')) {
+        cleanSubpath = cleanSubpath.substring(0, cleanSubpath.length - 1);
+      }
+      
       console.log('Cleaned subpath:', cleanSubpath);
       
-      // 在event_attributes JSON的full_url字段中查找subpath  
-      const condition = `JSONExtractString(event_attributes, 'full_url') LIKE '%${cleanSubpath}%'`;
+      // 使用正则表达式匹配URL中的第二个路径部分
+      // 示例: 在 "https://abc.com/slug/subpath/" 中匹配 "subpath"
+      const condition = `match(JSONExtractString(event_attributes, 'full_url'), '/[^/]+/${cleanSubpath}(/|\\\\?|$)')`;
+      
       console.log('Final SQL condition:', condition);
       console.log('==================================');
       
diff --git a/lib/clickhouse.ts b/lib/clickhouse.ts
index 1ff5e1b..bdde8c2 100644
--- a/lib/clickhouse.ts
+++ b/lib/clickhouse.ts
@@ -58,22 +58,28 @@ export function buildFilter(params: Partial<EventsQueryParams>): string {
     filters.push(`user_id = '${params.userId}'`);
   }
   
-  // 添加子路径过滤条件
-  if (params.subpath) {
+  // 添加子路径过滤条件 - 使用更精确的匹配方式
+  if (params.subpath && params.subpath.trim() !== '') {
     console.log('====== SUBPATH DEBUG ======');
     console.log('Raw subpath param:', params.subpath);
-    console.log('Subpath type:', typeof params.subpath);
-    console.log('Subpath length:', params.subpath.length);
     
-    // 确保子路径没有前导斜杠，避免双斜杠问题
-    const cleanSubpath = params.subpath.startsWith('/') 
-      ? params.subpath.substring(1) 
-      : params.subpath;
+    // 清理并准备subpath值
+    let cleanSubpath = params.subpath.trim();
+    // 移除开头的斜杠以便匹配
+    if (cleanSubpath.startsWith('/')) {
+      cleanSubpath = cleanSubpath.substring(1);
+    }
+    // 移除结尾的斜杠以便匹配
+    if (cleanSubpath.endsWith('/')) {
+      cleanSubpath = cleanSubpath.substring(0, cleanSubpath.length - 1);
+    }
     
     console.log('Cleaned subpath:', cleanSubpath);
     
-    // 在event_attributes JSON的full_url字段中查找subpath
-    const condition = `JSONExtractString(event_attributes, 'full_url') LIKE '%${cleanSubpath}%'`;
+    // 使用正则表达式匹配URL中的第二个路径部分
+    // 示例: 在 "https://abc.com/slug/subpath/" 中匹配 "subpath"
+    const condition = `match(JSONExtractString(event_attributes, 'full_url'), '/[^/]+/${cleanSubpath}(/|\\\\?|$)')`;
+    
     console.log('Final SQL condition:', condition);
     console.log('==========================');