def get(self, id=None, action=None):
# action by id
if action is not None:
if not hasattr(self, action):
return {
'status': 'ok',
'code': 400,
'error': 'action "%s" invalid' % action
}, 400
return getattr(self, action)(id)
# get one node
elif id is not None:
return jsonify(db_manager.get('spiders', id=id))
# get a list of items
else:
items = []
dirs = os.listdir(PROJECT_SOURCE_FILE_FOLDER)
for _dir in dirs:
if _dir in IGNORE_DIRS:
continue
dir_path = os.path.join(PROJECT_SOURCE_FILE_FOLDER, _dir)
dir_name = _dir
spider = db_manager.get_one_by_key('spiders', key='src', value=dir_path)
# new spider
if spider is None:
stats = get_file_suffix_stats(dir_path)
lang = get_lang_by_stats(stats)
db_manager.save('spiders', {
'name': dir_name,
'src': dir_path,
'lang': lang,
'suffix_stats': stats,
})
# existing spider
else:
stats = get_file_suffix_stats(dir_path)
lang = get_lang_by_stats(stats)
db_manager.update_one('spiders', id=str(spider['_id']), values={
'lang': lang,
'suffix_stats': stats,
})
# append spider
items.append(spider)
return jsonify({
'status': 'ok',
'items': items
})
def get(self, id=None, action=None):
"""
GET method of SpiderAPI.
:param id: spider_id
:param action: action
"""
# action by id
if action is not None:
if not hasattr(self, action):
return {
'status': 'ok',
'code': 400,
'error': 'action "%s" invalid' % action
}, 400
return getattr(self, action)(id)
# get one node
elif id is not None:
spider = db_manager.get('spiders', id=id)
# get deploy
last_deploy = db_manager.get_last_deploy(spider_id=spider['_id'])
if last_deploy is not None:
spider['deploy_ts'] = last_deploy['finish_ts']
return jsonify(spider)
# get a list of items
else:
items = []
# get customized spiders
dirs = os.listdir(PROJECT_SOURCE_FILE_FOLDER)
for _dir in dirs:
if _dir in IGNORE_DIRS:
continue
dir_path = os.path.join(PROJECT_SOURCE_FILE_FOLDER, _dir)
dir_name = _dir
spider = db_manager.get_one_by_key('spiders',
key='src',
value=dir_path)
# new spider
if spider is None:
stats = get_file_suffix_stats(dir_path)
lang = get_lang_by_stats(stats)
spider = db_manager.save(
'spiders', {
'name': dir_name,
'src': dir_path,
'lang': lang,
'suffix_stats': stats,
'type': SpiderType.CUSTOMIZED
})
# existing spider
else:
# get last deploy
last_deploy = db_manager.get_last_deploy(
spider_id=spider['_id'])
if last_deploy is not None:
spider['deploy_ts'] = last_deploy['finish_ts']
# file stats
stats = get_file_suffix_stats(dir_path)
# language
lang = get_lang_by_stats(stats)
# spider type
type_ = SpiderType.CUSTOMIZED
# update spider data
db_manager.update_one('spiders',
id=str(spider['_id']),
values={
'lang': lang,
'type': type_,
'suffix_stats': stats,
})
# append spider
items.append(spider)
# get configurable spiders
for spider in db_manager.list('spiders',
{'type': SpiderType.CONFIGURABLE}):
# append spider
items.append(spider)
# get other info
for i in range(len(items)):
spider = items[i]
# get site
if spider.get('site') is not None:
site = db_manager.get('sites', spider['site'])
if site is not None:
items[i]['site_name'] = site['name']
# get last task
last_task = db_manager.get_last_task(spider_id=spider['_id'])
if last_task is not None:
items[i]['task_ts'] = last_task['create_ts']
# ---------
# stats
# ---------
# last 5-run errors
items[i]['last_5_errors'] = get_last_n_run_errors_count(
spider_id=spider['_id'], n=5)
items[i]['last_7d_tasks'] = get_last_n_day_tasks_count(
spider_id=spider['_id'], n=5)
return {'status': 'ok', 'items': jsonify(items)}