Merge branch 'hot_search_query' into 'master'

Hot search query

See merge request !48

search/utils/auto_tips.py

@@ -64,7 +64,8 @@ def get_suggest_tips(query, lat, lng, offset=0, size=50, device_id=None):
             ret_list = equal_data
 
         if len(ret_list) >= 50:
-            return ret_list
+            logging.info("user_search_query:%s,get_sug_num:%s" % (query, len(ret_list)))
+            return ret_list[:50]
 
         else:
             query_ret_list = []
@@ -118,9 +119,11 @@ def get_suggest_tips(query, lat, lng, offset=0, size=50, device_id=None):
 
             ####
             if len(ret_list) >= 50:
+                logging.info("user_search_query:%s,get_sug_num:%s" % (query, len(ret_list)))
                 return ret_list[0:50]
 
             elif len(ret_list) < 50 and len(ret_list) > 3:
+                logging.info("user_search_query:%s,get_sug_num:%s" % (query, len(ret_list)))
                 return ret_list
 
             else:
@@ -137,13 +140,13 @@ def get_suggest_tips(query, lat, lng, offset=0, size=50, device_id=None):
                                                                                    have_read_tips_set=have_read_tips_set,
                                                                                    sub_index_name=sub_index_name)
                 ret_list.extend(pinyin_ret_list)
-
-                logging.info("get ret_list:%s" % ret_list)
+                logging.info("user_search_query:%s,get_sug_num:%s" % (query, len(ret_list)))
                 return ret_list
 
 
     except:
         logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+        logging.info("error_user_search_query:%s,get_sug_num:%s" % (query, 0))
         return list()
 
 
@@ -201,7 +204,6 @@ def get_query_by_es(query='', lat=0, lng=0, size=0, offset=0, highlight_query=No
         for tips_item in result_dict["suggest"]["tips-suggest"]:
             for hit_item in tips_item["options"]:
 
-                logging.info("get hit_item:%s" % hit_item)
                 if hit_item["_source"]["ori_name"] not in have_read_tips_set:
                     have_read_tips_set.add(hit_item["_source"]["ori_name"])
                     highlight_marks = u'<ems>%s</ems>' % query

search_tips/routers.py

+
+
+class DorisRouter(object):
+
+    def db_for_read(self, model, **hints):
+        if model._meta.app_label == "doris":
+            return 'doris'
+        return None
+
+    def db_for_write(self, model, **hints):
+        if model._meta.app_label == "doris":
+            return 'doris'
+        return None
+
+    def allow_relation(self, obj1, obj2, **hints):
+        if obj1._meta.app_label == 'doris' or \
+                obj2._meta.app_label == 'doris':
+            return True
+        return None
+
+    def allow_migrate(self, db, app_label, model_name=None, **hints):
+        if app_label == 'doris':
+            return db == 'doris'
+        return None

trans2es/commons/commons.py

@@ -20,7 +20,6 @@ def uuid4():
 
 def get_tips_suggest_list(instance_cn_name):
     try:
-        print(instance_cn_name)
         # ch_full_weight = 6.0 * 1000
         # py_full_weight = 3.0 * 1000
         full_weight = 3.0 * 1000

trans2es/commons/words_utils.py

@@ -6,6 +6,7 @@ import traceback
 import json
 from libs.cache import redis_client
 from trans2es.models.query_word_conversion import TagConversion, DoctorConversion, HospitalConversion
+from trans2es.models.strategy_search_word import StrategyHistoryQueryWords
 
 
 class TagTab(object):
@@ -96,6 +97,17 @@ class QueryWordAttr(object):
     doctab = DocTab()
     hostab = HosTab()
 
+    @classmethod
+    def get_hot_search_query_word_weight(cls, name):
+        try:
+            search_num = StrategyHistoryQueryWords.objects.filter(keyword=name, is_delete=0).values_list(
+                "search_num", flat=True).first()
+
+            return search_num * 0.01
+        except:
+            logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+            return 0.0
+
     @classmethod
     def get_project_query_word_weight(cls, name):
         try:
@@ -162,6 +174,20 @@ class QueryWordAttr(object):
             logging.error("catch exception,err_msg:%s" % traceback.format_exc())
             return 0
 
+    @classmethod
+    def get_search_query_results_num(cls, name):
+        try:
+            total_num = 0
+            data = StrategyHistoryQueryWords.objects.filter(keyword=name, is_delete=0).values("answer_num",
+                                                                                              "tractate_num",
+                                                                                              "diary_num").first()
+            for value in data.values():
+                total_num += value
+            return total_num
+        except:
+            logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+            return 0
+
 
 from gm_rpcd.all import bind
 from libs.cache import redis_client

trans2es/management/commands/trans2es_mapping2es.py

@@ -33,11 +33,8 @@ class Command(BaseCommand):
 
         try:
             es_cli = ESPerform.get_cli()
-
             type_name1 = "suggest-v1"
-
             type_name = "suggest"
-
             if len(options["type"]):
                 if options["type"] == "all" or type_name == options["type"] or type_name1 == options["type"]:
                     type_name = options["type"]

trans2es/models/strategy_search_word.py

+# coding=utf8
+from __future__ import unicode_literals, absolute_import, print_function
+from django.db import models
+from django.conf import settings
+
+
+class StrategyHistoryQueryWords(models.Model):
+    """
+    历史搜索词
+    """
+
+    class Meta:
+        app_label = 'doris'
+        db_table = 'strategy_history_query_words'
+
+    keyword = models.CharField(u'项目介绍', max_length=48)
+    is_delete = models.BooleanField(u'是否上线', default=True)
+    search_num = models.IntegerField(verbose_name=u'搜索次数')
+    answer_num = models.IntegerField(verbose_name=u'回答搜索结果次数')
+    tractate_num = models.IntegerField(verbose_name=u'帖子搜索结果次数')
+    diary_num = models.IntegerField(verbose_name=u'日记搜索结果次数')

trans2es/type_info.py

@@ -13,7 +13,8 @@ import elasticsearch.helpers
 import sys
 import copy
 
-from trans2es.models import doctor, itemwiki, collectwiki, brandwiki, productwiki, tag, wordresemble
+from trans2es.models import doctor, itemwiki, collectwiki, brandwiki, productwiki, tag, wordresemble, \
+    strategy_search_word
 from trans2es.utils.doctor_transfer import DoctorTransfer, DoctorTransferV1
 from trans2es.utils.hospital_transfer import HospitalTransfer, HospitalTransferV1
 from trans2es.utils.itemwiki_transfer import ItemWikiTransfer, ItemWikiTransferV1
@@ -22,6 +23,7 @@ from trans2es.utils.brandwiki_transfer import BrandWikiTransfer, BrandWikiTransf
 from trans2es.utils.productwiki_transfer import ProduceWikiTransfer, ProduceWikiTransferV1
 from trans2es.utils.tag_transfer import TagTransfer, TagTransferV1
 from trans2es.utils.wordresemble import WordResemble, WordResembleV1
+from trans2es.utils.search_query import SearchWord
 from libs.es import ESPerform
 from libs.tools import tzlc, getMd5Digest
 from trans2es.commons.words_utils import QueryWordAttr
@@ -139,7 +141,7 @@ class TypeInfo(object):
                     (item_dict, suggest_list) = data
 
                     if item_dict:
-                        if item_dict["tips_name_type"] == 4:
+                        if item_dict["tips_name_type"] == 4 or item_dict["tips_name_type"] == 5:
                             instance.name = instance.keyword
 
                         resemble_list = WordResemble.get_word_resemble_list(str(instance.name))
@@ -366,6 +368,28 @@ def get_type_info_map():
             round_insert_chunk_size=5,
             round_insert_period=2,
         ),
+        TypeInfo(
+            name='suggest',
+            type='search_query',  # tag
+            model=strategy_search_word.StrategyHistoryQueryWords,
+            query_deferred=lambda: strategy_search_word.StrategyHistoryQueryWords.objects.using(
+                settings.DORIS_DB_NAME).all().query,
+            get_data_func=SearchWord.get_search_query,
+            bulk_insert_chunk_size=100,
+            round_insert_chunk_size=5,
+            round_insert_period=2
+        ),
+        TypeInfo(
+            name='suggest-v1',
+            type='search_query-v1',  # tag
+            model=strategy_search_word.StrategyHistoryQueryWords,
+            query_deferred=lambda: strategy_search_word.StrategyHistoryQueryWords.objects.using(
+                settings.DORIS_DB_NAME).all().query,
+            get_data_func=SearchWord.get_search_query,
+            bulk_insert_chunk_size=100,
+            round_insert_chunk_size=5,
+            round_insert_period=2
+        ),
         TypeInfo(
             name='suggest-v1',
             type='doctor_tips-v1',  # doctor

trans2es/utils/search_query.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+import logging
+import traceback
+from libs.tools import getMd5Digest
+from trans2es.commons.words_utils import QueryWordAttr, get_tips_word_type
+from trans2es.commons.commons import get_tips_suggest_list_v1
+
+
+class SearchWord(object):
+
+    @classmethod
+    def get_search_query(self, instance):
+        try:
+            ret_list = list()
+            item_dict = dict()
+            item_dict["id"] = getMd5Digest(str(instance.keyword))
+            item_dict["ori_name"] = instance.keyword
+            item_dict["is_online"] = False if instance.is_delete else True
+            item_dict["order_weight"] = QueryWordAttr.get_hot_search_query_word_weight(instance.keyword)
+            item_dict["results_num"] = QueryWordAttr.get_search_query_results_num(instance.keyword)
+            item_dict["type_flag"] = get_tips_word_type(instance.keyword)
+            item_dict["offline_score"] = 0.0
+            item_dict["tips_name_type"] = 5
+            ret_list.append(item_dict)
+            suggest_list = get_tips_suggest_list_v1(str(instance.keyword).lower())
+            return (item_dict, suggest_list)
+
+        except:
+            logging.error("catch exception,err_msg:%s" % traceback.format_exc())
+            return ([], [])