def get_results(self, id: str) -> (dict, tuple):
"""
Get a list of results crawled in a given task.
:param id: task_id
"""
args = self.parser.parse_args()
page_size = args.get('page_size') or 10
page_num = args.get('page_num') or 1
task = db_manager.get('tasks', id=id)
spider = db_manager.get('spiders', id=task['spider_id'])
col_name = spider.get('col')
if not col_name:
return []
fields = get_spider_col_fields(col_name)
items = db_manager.list(col_name, {'task_id': id},
skip=page_size * (page_num - 1),
limit=page_size)
return {
'status': 'ok',
'fields': jsonify(fields),
'total_count': db_manager.count(col_name, {'task_id': id}),
'page_num': page_num,
'page_size': page_size,
'items': jsonify(items)
}
def get_results(self, id):
task = db_manager.get('tasks', id=id)
spider = db_manager.get('spiders', id=task['spider_id'])
col_name = spider.get('col')
if not col_name:
return []
fields = get_spider_col_fields(col_name)
items = db_manager.list(col_name, {'task_id': id})
return jsonify({'status': 'ok', 'fields': fields, 'items': items})
def download_results(self, id: str):
task = db_manager.get('tasks', id=id)
spider = db_manager.get('spiders', id=task['spider_id'])
col_name = spider.get('col')
if not col_name:
return send_csv([], f'results_{col_name}_{round(time())}.csv')
items = db_manager.list(col_name, {'task_id': id}, limit=999999999)
fields = get_spider_col_fields(col_name, task_id=id, limit=999999999)
return send_csv(items,
filename=f'results_{col_name}_{round(time())}.csv',
fields=fields,
encoding='utf-8')
def get_results(self, id: str) -> (dict, tuple):
"""
Get a list of results crawled in a given task.
:param id: task_id
"""
args = self.parser.parse_args()
page_size = args.get('page_size') or 10
task = db_manager.get('tasks', id=id)
spider = db_manager.get('spiders', id=task['spider_id'])
col_name = spider.get('col')
if not col_name:
return []
fields = get_spider_col_fields(col_name)
fields = list(set(fields) - set(IGNORE_FIELD))
items = db_manager.list(col_name, {'task_id': id})
# 避免内容过长，做一下限制；同时剔除无用的字段不展示
adjust_items = []
for item in items:
adjust_item = {}
for key, value in item.items():
if isinstance(value, str) == False:
continue
if key in IGNORE_FIELD:
continue
if len(value) > 500:
value = value[:500] + '...'
adjust_item[key] = value
adjust_items += [adjust_item]
total_count = db_manager.count(col_name, {'task_id': id})
page_num = len(adjust_items) / page_size
if isinstance(page_num, float):
page_num = int(page_num) + 1
return {
'status': 'ok',
'fields': jsonify(fields),
'total_count': len(adjust_items),
'page_num': page_num,
'page_size': page_size,
'items': jsonify(adjust_items)
}
def get_results(self, id):
args = self.parser.parse_args()
page_size = args.get('page_size') or 10
page_num = args.get('page_num') or 1
task = db_manager.get('tasks', id=id)
spider = db_manager.get('spiders', id=task['spider_id'])
col_name = spider.get('col')
if not col_name:
return []
fields = get_spider_col_fields(col_name)
items = db_manager.list(col_name, {'task_id': id})
return {
'status': 'ok',
'fields': jsonify(fields),
'total_count': db_manager.count(col_name, {'task_id': id}),
'page_num': page_num,
'page_size': page_size,
'items': jsonify(items)
}
