def process_request(self, request, spider):
save_path = ToCrawlUrl.settings.get(
"ROOT_PATH_LOG") + ctime.get_today() + "_to_crawl.log"
log_file(save_path,
str(ctime.get_now_time()) + "\t" + spider.name + "\t" +
request.url + "\n",
method="a")
def process_request(self, request, spider):
save_path = self.settings.get(
"ROOT_PATH_LOG") + Dt.get_today() + "_to_crawl.log"
# TODO:
logging.info(save_path,
str(Dt.get_now_time()) + "\t" + spider.name + "\t" +
request.url + "\n",
method="a")
def process_response(self, request, response, spider):
save_path = CrawledUrl.settings.get(
"ROOT_PATH_LOG") + ctime.get_today() + "_carwled.log"
log_file(save_path,
str(ctime.get_now_time()) + "\t" + spider.name + "\t" +
str(response.status) + "\t" + request.url + "\n",
method="a")
return response
def process_response(self, request, response, spider):
save_path = self.settings.get(
"ROOT_PATH_LOG") + Dt.get_today() + "_carwled.log"
# TODO:
logging.info(save_path,
str(Dt.get_now_time()) + "\t" + spider.name + "\t" +
str(response.status) + "\t" + request.url + "\n",
method="a")
return response
def get_date_time(s):
if not s or not isinstance(s, str):
return ""
r = r"(\d{2,4}[-/年]\d{1,2}[-/月]\d{1,2}日? \d{2}:\d{2}(:\d{2})?)"
r = r"\d{2,4}[-/年]\d{1,2}[-/月]\d{1,2}日? \d{2}:\d{2}(:\d{2})?"
p = re.compile(r)
result = re.findall(p, s)
try:
result[0] = result[0].replace('年', '-').replace('月', '-').replace(
'日', '').replace('号', '')
if result[0] == "今天":
return DT.get_today()
if len(result[0]) == 4 or len(result[0]) == 5:
print(result[0])
return str(DT.get_today().split("-")[0]) + result[0]
return result[0]
except Exception as e:
print(e)
def get_date(s):
if not s or not isinstance(s, str):
return ""
p = re.compile(
r'\d{4}[-|/|.]\d{1,2}[-|/|.]\d{1,2}|\d{4}年\d{1,2}月\d{1,2}[日号]|今天|\d{1,2}月\d{1,2}[日号]'
)
result = re.findall(p, s)
try:
print(result[0])
result[0] = result[0].replace('年', '-').replace('月', '-').replace(
'日', '').replace('号', '')
if result[0] == "今天":
return DT.get_today()
if len(result[0]) == 4 or len(result[0]) == 5:
print(result[0])
return str(DT.get_today().split("-")[0]) + result[0]
return result[0]
except Exception as e:
print(e)
def inner(*args, **kwargs):
if len(args) < 2:
json_data = kwargs.get("json_data", None)
spider = kwargs.get("spider", None)
fp = kwargs.get("fp", None)
else:
json_data = args[0]
spider = args[1]
fp = args[3]
if json_data:
today = dt.get_today().replace("-", "_")
uf.FileHelper.mkdir(fp + spider)
json.dump(json_data, open(fp + spider + os.sep + today + ".json", "a", encoding="utf-8"), ensure_ascii=False)
return func(*args, **kwargs) # 2
def json_2_redis(*args, **kw):
rcfg = kw.get("rcfg")
if not rcfg:
print("No rcfg" + "===" * 10)
return
rename = kw.get("rename", 0)
conn_redis = dr.RedisHelper.get_redis_connect_by_cfg(rcfg)
fp = kw.get("fp", "")
ts = kw.get("ts", 1)
spider = kw.get("spider")
if rename and str(fp).startswith(dt.get_today().replace("-", "_")):
return
for line in open(fp, encoding="utf-8"):
length = conn_redis.llen(spider + ":items")
if length > 50000:
bf.print_from_head(fp + "\t Too much,Please customer\t" +
str(length) + "\t\t")
time.sleep(ts)
bf.print_blank_end(conn_redis.lpush(spider + ":items", line))
if rename:
uf.FileHelper.rename_file(fp, str(fp) + "1")
print("=====File Over\t" + fp + "=====")
conn_redis.connection_pool.disconnect()
