try all

eda/smart_rank/new_tag3_cold_start.py

@@ -10,6 +10,24 @@ def merge_lists(*lol):
     return [i for i in chain(*a) if i is not None]
 
 
+def get_result(hits):
+    res_dict = {}
+    for info in hits:
+        second_demands = info["_source"].get("second_demands", [])
+        if second_demands and len(second_demands) > 0:
+            id = info["_source"]["id"]
+            sd = second_demands[0]
+            if sd not in res_dict:
+                res_dict[sd] = [id]
+            else:
+                res_dict[sd].append(id)
+    res_pair = sorted(res_dict.items(), key=lambda x: len(x[1]), reverse=True)
+    res_list = []
+    for _, v in res_pair:
+        res_list.append(v)
+    return merge_lists(res_list)
+
+
 keywords = [
     "瘦脸", "双眼皮", "补水", "美白嫩肤", "瘦小腿", "除皱", "隆鼻", "牙齿矫正", "脱毛", "祛斑", "控油", "面部祛脂", "隆胸", "垫鼻基底", "缩鼻翼", "生头发", "开眼角", "瘦肩", "祛痘",
     "缩毛孔", "轮廓改善", "除眼底细纹", "眼部修复", "垫下巴", "颧骨内推", "除法令纹", "缩咬肌", "瘦臀部", "丰眼窝", "瘦手臂", "丰唇", "瘦腰腹", "瘦大腿", "填充泪沟", "洁面", "除面部细纹",
@@ -132,23 +150,141 @@ def get_es_diary(keywords, city_tag_id=-1, version=False):
     q["sort"] = sort_list
     q["_source"] = {"includes": ["id", "second_demands"]}
     es_res = es_query("diary", q, offset=0, size=5000)
-    diary_dict = {}
-    for diary_info in es_res["hits"]["hits"]:
-        second_demands = diary_info["_source"].get("second_demands", [])
-        if second_demands and len(second_demands) > 0:
-            id = diary_info["_source"]["id"]
-            sd = second_demands[0]
-            if sd not in diary_dict:
-                diary_dict[sd] = [id]
-            else:
-                diary_dict[sd].append(id)
-    res_pair = sorted(diary_dict.items(), key=lambda x: len(x[1]), reverse=True)
-    res_list = []
-    for _, v in res_pair:
-        res_list.append(v)
-    return merge_lists(res_list)
+    return get_result(es_res["hits"]["hits"])
+
+
+def get_es_tractate(keywords, version=False):
+    q = {}
+    should_list = [{
+        "terms": {
+            "first_demands": keywords
+        }
+    }, {
+        "terms": {
+            "second_demands": keywords
+        }
+    }, {
+        "terms": {
+            "first_solutions": keywords
+        }
+    }, {
+        "terms": {
+            "second_solutions": keywords
+        }
+    }, {
+        "terms": {
+            "positions": keywords
+        }
+    }, {
+        "terms": {
+            "second_positions": keywords
+        }
+    }, {
+        "terms": {
+            "tags_v3": keywords
+        }
+    }]
+    q["query"] = {
+        "bool": {
+            "filter": [{
+                "term": {
+                    "is_online": True
+                }
+            }, {
+                "terms": {
+                    "content_level": [6, 5, 4, 3.5, 3]
+                }
+            }],
+            "should": should_list,
+            "minimum_should_match": 1
+        }
+    }
+    q["_source"] = {"includes": ["id", "second_demands"]}
+    if version:
+        q["sort"] = [{
+            "is_video": {
+                "order": "asc"
+            }
+        }, {
+            "good_click_tractate_score": {
+                "order": "desc"
+            }
+        }, {
+            "good_click": {
+                "order": "desc"
+            }
+        }]
+    else:
+        q["sort"] = [{"is_video": {"order": "asc"}}, {"tractate_score": {"order": "desc"}}, {"good_click": {"order": "desc"}}]
+    es_res = es_query("tractate", q, offset=0, size=5000)
+    return get_result(es_res["hits"]["hits"])
+
+
+def get_es_answer(keywords, version=False):
+    q = {}
+    if version:
+        sort_list = [{"has_picture": {"order": "desc"}}, {"smart_rank_v3": {"order": "desc"}}, {"good_click": {"order": "desc"}}]
+    else:
+        sort_list = [{"has_picture": {"order": "desc"}}, {"smart_rank_v2": {"order": "desc"}}, {"good_click": {"order": "desc"}}]
+    should_list = [{
+        "terms": {
+            "first_demands": keywords
+        }
+    }, {
+        "terms": {
+            "second_demands": keywords
+        }
+    }, {
+        "terms": {
+            "first_solutions": keywords
+        }
+    }, {
+        "terms": {
+            "second_solutions": keywords
+        }
+    }, {
+        "terms": {
+            "positions": keywords
+        }
+    }, {
+        "terms": {
+            "second_positions": keywords
+        }
+    }, {
+        "terms": {
+            "tags_v3": keywords
+        }
+    }]
+    q["query"] = {
+        "bool": {
+            "filter": [{
+                "range": {
+                    "content_length": {
+                        "gte": 30
+                    }
+                }
+            }, {
+                "term": {
+                    "is_online": True
+                }
+            }, {
+                "terms": {
+                    "content_level": ["6", "5", "4", "3.5", "3"]
+                }
+            }],
+            "should":
+            should_list,
+            "minimum_should_match":
+            1
+        }
+    }
+    q["_source"] = {"includes": ["id", "second_demands"]}
+    q["sort"] = sort_list
+    es_res = es_query("answer", q, offset=0, size=5000)
+    return get_result(es_res["hits"]["hits"])
 
 
 if __name__ == "__main__":
-    ids = get_es_diary(keywords)
-    print(ids)
+    print(get_es_diary(keywords)[:10])
+    print(get_es_tractate(keywords)[:10])
+    print(get_es_answer(keywords)[:10])