'commit'
This commit is contained in:
@@ -4,6 +4,8 @@
|
|||||||
SCROLL_DISTANCE_RATIO = 0.5
|
SCROLL_DISTANCE_RATIO = 0.5
|
||||||
# 最大采集场站数量,达到此次数后停止采集
|
# 最大采集场站数量,达到此次数后停止采集
|
||||||
MAX_STATIONS_COUNT = 100
|
MAX_STATIONS_COUNT = 100
|
||||||
|
FIRST_RUN_ONLY_ONE_STATION = False
|
||||||
|
FIRST_RUN_ONLY_ONE_STATION = False
|
||||||
# 场站去重过期时间(秒),在此时间内重复出现的场站不会再次点击进入详情页
|
# 场站去重过期时间(秒),在此时间内重复出现的场站不会再次点击进入详情页
|
||||||
REDIS_STATION_EXPIRE = 120
|
REDIS_STATION_EXPIRE = 120
|
||||||
# 数据库数据保留时长(天),超过此时长的历史数据(is_current=0)将被删除
|
# 数据库数据保留时长(天),超过此时长的历史数据(is_current=0)将被删除
|
||||||
|
|||||||
@@ -26,7 +26,8 @@ from Apps.AiTeJiYiChong.Config.Setting import (
|
|||||||
WAIT_AFTER_SCROLL,
|
WAIT_AFTER_SCROLL,
|
||||||
SAFE_EXCLUDE_RATIO,
|
SAFE_EXCLUDE_RATIO,
|
||||||
BOTTOM_SAFE_EXCLUDE_RATIO,
|
BOTTOM_SAFE_EXCLUDE_RATIO,
|
||||||
TEST_CLEAR_REDIS
|
TEST_CLEAR_REDIS,
|
||||||
|
FIRST_RUN_ONLY_ONE_STATION
|
||||||
)
|
)
|
||||||
|
|
||||||
logger = logging.getLogger("AiTeJiYiChongCrawler")
|
logger = logging.getLogger("AiTeJiYiChongCrawler")
|
||||||
@@ -289,6 +290,10 @@ async def get_station_list(d, service, max_stations_count=MAX_STATIONS_COUNT):
|
|||||||
full_name_key = f"crawled:aite:{Kit.clean_station_name(station_name)}"
|
full_name_key = f"crawled:aite:{Kit.clean_station_name(station_name)}"
|
||||||
await redis_kit.set_data(full_name_key, "1", expire=REDIS_STATION_EXPIRE)
|
await redis_kit.set_data(full_name_key, "1", expire=REDIS_STATION_EXPIRE)
|
||||||
|
|
||||||
|
if FIRST_RUN_ONLY_ONE_STATION:
|
||||||
|
logger.info("已完成首个场站的全流程采集,根据配置退出艾特吉易充爬取任务。")
|
||||||
|
break
|
||||||
|
|
||||||
# 如果内层循环已达上限,外层循环也应退出,避免不必要的滑动
|
# 如果内层循环已达上限,外层循环也应退出,避免不必要的滑动
|
||||||
if total_encountered_count >= max_stations_count:
|
if total_encountered_count >= max_stations_count:
|
||||||
break
|
break
|
||||||
|
|||||||
@@ -5,6 +5,7 @@
|
|||||||
SCROLL_DISTANCE_RATIO = 0.5
|
SCROLL_DISTANCE_RATIO = 0.5
|
||||||
# 最大采集场站数量,达到此次数后停止采集
|
# 最大采集场站数量,达到此次数后停止采集
|
||||||
MAX_STATIONS_COUNT = 100
|
MAX_STATIONS_COUNT = 100
|
||||||
|
FIRST_RUN_ONLY_ONE_STATION = False
|
||||||
# 场站去重过期时间(秒),在此时间内重复出现的场站不会再次点击进入详情页
|
# 场站去重过期时间(秒),在此时间内重复出现的场站不会再次点击进入详情页
|
||||||
REDIS_STATION_EXPIRE = 120
|
REDIS_STATION_EXPIRE = 120
|
||||||
|
|
||||||
|
|||||||
@@ -4,6 +4,7 @@
|
|||||||
SCROLL_DISTANCE_RATIO = 0.5
|
SCROLL_DISTANCE_RATIO = 0.5
|
||||||
# 最大采集场站数量,达到此次数后停止采集
|
# 最大采集场站数量,达到此次数后停止采集
|
||||||
MAX_STATIONS_COUNT = 100
|
MAX_STATIONS_COUNT = 100
|
||||||
|
FIRST_RUN_ONLY_ONE_STATION = False
|
||||||
# 场站去重过期时间(秒),在此时间内重复出现的场站不会再次点击进入详情页
|
# 场站去重过期时间(秒),在此时间内重复出现的场站不会再次点击进入详情页
|
||||||
REDIS_STATION_EXPIRE = 120
|
REDIS_STATION_EXPIRE = 120
|
||||||
# 数据库数据保留时长(天),超过此时长的历史数据(is_current=0)将被删除
|
# 数据库数据保留时长(天),超过此时长的历史数据(is_current=0)将被删除
|
||||||
@@ -36,3 +37,4 @@ BOTTOM_SAFE_EXCLUDE_RATIO = 0.10
|
|||||||
FALLBACK_WIDTH = 1080
|
FALLBACK_WIDTH = 1080
|
||||||
# 默认回退屏幕高度,当无法自动获取设备信息时使用
|
# 默认回退屏幕高度,当无法自动获取设备信息时使用
|
||||||
FALLBACK_HEIGHT = 2400
|
FALLBACK_HEIGHT = 2400
|
||||||
|
FIRST_RUN_ONLY_ONE_STATION = False
|
||||||
|
|||||||
@@ -12,7 +12,8 @@ from Apps.YeLiTe.Service import YiLaiTeService
|
|||||||
from Apps.YeLiTe.Config.Setting import (
|
from Apps.YeLiTe.Config.Setting import (
|
||||||
SCROLL_DISTANCE_RATIO, WAIT_AFTER_SCROLL, MAX_STATIONS_COUNT,
|
SCROLL_DISTANCE_RATIO, WAIT_AFTER_SCROLL, MAX_STATIONS_COUNT,
|
||||||
WAIT_DETAIL_PAGE_LOAD, WAIT_BACK_TO_LIST, TEST_CLEAR_REDIS,
|
WAIT_DETAIL_PAGE_LOAD, WAIT_BACK_TO_LIST, TEST_CLEAR_REDIS,
|
||||||
SAFE_EXCLUDE_RATIO, BOTTOM_SAFE_EXCLUDE_RATIO
|
SAFE_EXCLUDE_RATIO, BOTTOM_SAFE_EXCLUDE_RATIO,
|
||||||
|
FIRST_RUN_ONLY_ONE_STATION
|
||||||
)
|
)
|
||||||
from Util.RedisKit import RedisKit
|
from Util.RedisKit import RedisKit
|
||||||
from Core.BaseCrawler import BaseCrawler
|
from Core.BaseCrawler import BaseCrawler
|
||||||
@@ -238,6 +239,10 @@ class YiLaiTeCrawler(BaseCrawler):
|
|||||||
logger.info("似乎还在二级页面,尝试再次返回...")
|
logger.info("似乎还在二级页面,尝试再次返回...")
|
||||||
d.press("back")
|
d.press("back")
|
||||||
await asyncio.sleep(WAIT_BACK_TO_LIST)
|
await asyncio.sleep(WAIT_BACK_TO_LIST)
|
||||||
|
|
||||||
|
if FIRST_RUN_ONLY_ONE_STATION:
|
||||||
|
logger.info("已完成首个场站的全流程采集,根据配置退出驿来特爬取任务。")
|
||||||
|
return processed_count
|
||||||
else:
|
else:
|
||||||
logger.warning(f"点击场站 {name} 后页面似乎未跳转,跳过返回操作")
|
logger.warning(f"点击场站 {name} 后页面似乎未跳转,跳过返回操作")
|
||||||
if os.path.exists(detail_shot): os.remove(detail_shot)
|
if os.path.exists(detail_shot): os.remove(detail_shot)
|
||||||
|
|||||||
Reference in New Issue
Block a user