Merge branch 'haow/pin_crawl' into 'test'

Haow/pin crawl

See merge request alpha/saturn!60

api/views/topic.py

@@ -7,7 +7,7 @@ from api.cache.cache import ins_cache
 from libs.user import get_user_by_ids
 from alpha_types.venus import ERROR as CODES
 from alpha_types.venus import GRAP_PLATFORM
-from engine.logger import info_logger
+from engine.logger import info_logger, error_logger
 
 
 topic_id_cache = "ins_account_cache"
@@ -18,6 +18,99 @@ class CreateTopicForBatch(BaseView):
     内部使用，批量建帖
     """
 
+    def batch_create_tags(self, tags, is_location=None, is_own=None):
+
+        info_logger.info({
+            'api': 'venus/community/tag/batch_create_tag_by_name',
+            'tags': tags,
+            'is_own': 1,
+        })
+        try:
+            _tag_error, _tag_data = self.call_rpc(
+                "venus/community/tag/batch_create_tag_by_name",
+                tags=tags, is_own=is_own, is_location=is_location,
+            )
+            if _tag_error:
+                error_logger.error({'api': 'venus/community/tag/batch_create_tag_by_name',
+                                    'error': _tag_error})
+                _tag_data = {}
+        except Exception as e:
+            error_logger.error({'api': 'venus/community/tag/batch_create_tag_by_name',
+                                'information': e})
+            _tag_data = {}
+        info_logger.info({
+            'api': 'venus/community/tag/batch_create_tag_by_name',
+            '_tag_data': _tag_data,
+        })
+        return _tag_error, _tag_data
+
+    def replace_tag_info(self, topic_list, tag_data):
+        # 处理标签，将文本中的标签处理成现有标签
+        for item in topic_list:
+            tags = item.get("tags") or []
+            tags = [tag.replace("#", '').strip() for tag in tags]
+
+            #添加地域标签
+            if item.get("location") and item.get("location").get("name"):
+                tags.append(item.get("location").get("name"))
+            content = item.get("content")
+            for tag_name, tag_id  in tag_data.items():
+                if tag_name in tags:
+                    alpha_tag = '<topic>{' + '"id":{},"name":"{}"'.format(tag_id, tag_name) + '}</topic>'
+                    content = content.replace('#' + tag_name, alpha_tag)
+            item["content"] = content.replace('#', '')
+
+            item["tag_ids"] = [
+                tag_data[tag_name]
+                for tag_name in tags if tag_data.get(tag_name)
+            ]
+        return topic_list
+
+    def check_exist_ids(self, topics, platform):
+        # 帖子去重 redis层面
+        not_exists_ids = []
+        topic_list = []
+        if platform == GRAP_PLATFORM.INS:
+            cache_key = topic_id_cache
+        elif platform == GRAP_PLATFORM.PIN:
+            cache_key = topic_id_cache + ":2"
+        elif platform == GRAP_PLATFORM.FASHION:
+            cache_key = topic_id_cache + ":3"
+        else:
+            return
+
+        if platform in (GRAP_PLATFORM.INS, GRAP_PLATFORM.PIN, GRAP_PLATFORM.FASHION):
+            for item in topics:
+                _id = item.get("id")
+                if not _id:
+                    continue
+
+                exists = ins_cache.sismember(cache_key, _id)
+                if exists:
+                    continue
+
+                item["platform"] = platform
+                item["platform_id"] = _id
+                topic_list.append(item)
+                not_exists_ids.append(_id)
+
+        return not_exists_ids, topic_list, cache_key
+
+    def check_sensitive_content(self, topic_list):
+        # 敏感词检测，获取可用的帖子
+        topics = []
+    
+        check_info = Sensitive.check([topic["content"] for topic in topic_list if topic.get("content")])
+        for topic in topic_list:
+            if topic.get('content'):
+                succ = check_info.get(topic.get("content"))
+                if not succ:
+                    topics.append(topic)
+            else:
+                if topic.get('images') or topic.get('video'):
+                    topics.append(topic)
+        return topics
+
     def post(self, request):
 
         user_id = request.POST.get("user_id", 0)
@@ -48,18 +141,12 @@ class CreateTopicForBatch(BaseView):
         if not topic_list:
             return self.ok()
 
-        topics = []
         # 敏感词检测，获取可用的帖子
-        # check_info = Sensitive.check([topic["content"] for topic in topic_list if topic.get("content")])
-        # for topic in topic_list:
-        #     if topic.get('content'):
-        #         succ = check_info.get(topic.get("content"))
-        #         if not succ:
-        #             topics.append(topic)
-        #     else:
-        #         if topic.get('images') or topic.get('video'):
-        #             topics.append(topic)
-        for topic in topic_list:
+        # checked_topics = self.check_sensitive_content(topic_list)
+        checked_topics = topic_list
+
+        topics = []
+        for topic in checked_topics:
             if topic.get('content'):
                 topics.append(topic)
             else:
@@ -80,82 +167,25 @@ class CreateTopicForBatch(BaseView):
             item["is_online"] = is_online
             item["pictorial_tag_ids"] = pictorial_tag_ids
             if item.get("location") and item.get("location").get("name"):
-                _tag_error, _location_tag_data = self.call_rpc(
-                    "venus/community/tag/batch_create_tag_by_name",
-                    tags=tags, is_location=1, is_own=1,
-                )
-                if _tag_error:
-                    return self.error(_tag_error)
+                _tag_error, _location_tag_data = self.batch_create_tags(tags=tags, is_location=1, is_own=1)
                 location_tags.append(_location_tag_data)
 
-        not_exists_ids = []
-        topic_list = []
-        if platform == GRAP_PLATFORM.INS:
-            cache_key = topic_id_cache
-        elif platform == GRAP_PLATFORM.PIN:
-            cache_key = topic_id_cache + ":2"
-        elif platform == GRAP_PLATFORM.FASHION:
-            cache_key = topic_id_cache + ":3"
-        else:
-            return
-
-        if platform in (GRAP_PLATFORM.INS, GRAP_PLATFORM.PIN, GRAP_PLATFORM.FASHION):
-            for item in topics:
-                _id = item.get("id")
-                if not _id:
-                    continue
-
-                exists = ins_cache.sismember(cache_key, _id)
-                if exists:
-                    continue
-
-                item["platform"] = platform
-                item["platform_id"] = _id
-                topic_list.append(item)
-                not_exists_ids.append(_id)
+        not_exists_ids, topic_list, cache_key = self.check_exist_ids(topics=topics, platform=platform)
 
         # check_info = Sensitive.check(tag_names)
         # tags  = [tag_name for tag_name, succ in check_info.items() if not succ]
         tags = tag_names
 
         # 先创建标签
-        info_logger.info({
-            'api': 'venus/community/tag/batch_create_tag_by_name',
-            'tags': tags,
-            'is_own': 1,
-        })
-        _tag_error, _tag_data = self.call_rpc(
-            "venus/community/tag/batch_create_tag_by_name",
-            tags=tags, is_own=1
-        )
-        if _tag_error:
-            return self.error(_tag_error)
-        info_logger.info({
-            'api': 'venus/community/tag/batch_create_tag_by_name',
-            '_tag_data': _tag_data,
-        })
-        #_tag_data.update(location_tags)
+        _tag_error, _tag_data = self.batch_create_tags(tags=tags, is_own=1)
+
+        # 地域标签
+        if location_tags:
+            _tag_data = _tag_data.extend(location_tags)
 
         # 更新发帖
         # 处理标签，将文本中的标签处理成现有标签
-        for item in topic_list:
-            tags = item.get("tags") or []
-            tags = [tag.replace("#", '').strip() for tag in tags]
-
-            #添加地域标签
-            if item.get("location") and item.get("location").get("name"):
-                tags.append(item.get("location").get("name"))
-            content = item.get("content")
-            for tag_name, tag_id  in _tag_data.items():
-                if tag_name in tags:
-                    alpha_tag = '<topic>{' + '"id":{},"name":"{}"'.format(tag_id, tag_name) + '}</topic>'
-                    content = content.replace('#' + tag_name, alpha_tag)
-            item["content"] = content.replace('#', '')
-
-            item["tag_ids"] = [
-                _tag_data[tag_name]
-                for tag_name in tags if _tag_data.get(tag_name)
-            ]
+        topic_list = self.replace_tag_info(topic_list=topic_list, tag_data=_tag_data)
 
         info_logger.info({
             'api': 'venus/community/topic/batch_create_for_inner',