match the table and request to es

main.py

@@ -79,17 +79,34 @@ def main():
     df = pd.read_excel('table.xlsx')
     df = pd.read_excel('table.xlsx')
     data = df.iloc[:, [0]].values
     data = df.iloc[:, [0]].values
     ans = []
     ans = []
+    result = []
     for i in range(len(data)):
     for i in range(len(data)):
+        if i > 10:
+           break   # for debug, don't forget to delete this line.
         sku_name_str = data[i][0]
         sku_name_str = data[i][0]
         res_tmp = match_sku(sku_name_str)
         res_tmp = match_sku(sku_name_str)
         ans.append(res_tmp)
         ans.append(res_tmp)
 
 
     for i in range(len(ans)):
     for i in range(len(ans)):
+        score_in_one_row = []
         for j in range(len(ans[i])):
         for j in range(len(ans[i])):
             tmp_goods_name = ans[i][j]
             tmp_goods_name = ans[i][j]
             tmp_query = query
             tmp_query = query
             tmp_query["query"]["bool"]["must"][0]["match"]["goods_name"] = tmp_goods_name
             tmp_query["query"]["bool"]["must"][0]["match"]["goods_name"] = tmp_goods_name
-
+            tmp_map = process(tmp_query)
+            if len(tmp_map) != 0:
+                score_in_one_row.append(tmp_map)
+        if len(score_in_one_row) == 0:
+            continue
+        else:
+            max_score = score_in_one_row[0].get("score")
+            index = 0
+        for k in range(len(score_in_one_row)):
+            if score_in_one_row[k].get("score") > max_score:
+                max_score = score_in_one_row[k].get("score")
+                index = k
+        result.append(score_in_one_row[index])
+    return result
 
 
 def match_sku(s: str) -> []:
 def match_sku(s: str) -> []:
     # s = "RG58/U射频连接线SMA公头转BNC公头延长线 SMA/BNC-JJ Q9/SMA-JJ RG58/U-SMA/BNC-JJ 2m"
     # s = "RG58/U射频连接线SMA公头转BNC公头延长线 SMA/BNC-JJ Q9/SMA-JJ RG58/U-SMA/BNC-JJ 2m"
@@ -124,6 +141,19 @@ def post_to_elasticsearch(q: dict):
     ans = response.json()
     ans = response.json()
     return ans
     return ans
 
 
+def process(query_str: dict):
+    res = post_to_elasticsearch(query_str)
+    if  res["hits"]["max_score"] is not None and res["hits"]["max_score"] > 80:
+        for i in range(len(res["hits"]["hits"])):
+            if res["hits"]["hits"][i]["_score"] == res["hits"]["max_score"]:
+                _score = res["hits"]["hits"][i]["_score"]
+                _goods_name = res["hits"]["hits"][i]["_source"]["goods_name"]
+                _goods_id =  res["hits"]["hits"][i]["_source"]["goods_id"]
+                _brand_name = res["hits"]["hits"][i]["_source"]["brand_name"]
+                return {"goods_name": _goods_name, "goods_id": _goods_id, "brand_name": _brand_name, "score": _score}
+    return {}
+
 
 
 if __name__ == '__main__':
 if __name__ == '__main__':
-    main()
+    res = main()
+    print(res)
\ No newline at end of file