#--------------------------
processnum = 3
#function3
description = '執行完第三步驟時錯誤'
logtext = '一般log紀錄 第三步'
putintoSql(items)
monitor.filewrite(setlogfile, logtext, processnum)
#--------------------------
#.........
state = "success"
errnum = 0
except Exception as e:
state = "fail"
errnum = 1
#文件寫入點1 寫入錯誤
monitor.toErrFile(errlogfilename, daytime, filename, e, processnum,
description)
#=============================================================
# 比對 應放入資料庫筆數 / 已入資料庫筆數
# note:此處因為沒有真實資料先使用log資料，上線後請改成爬蟲或清洗資料
#-------------------------------------------------------------
try:
#請放入此次執行應進SQL行數（改成df行數）
sSQL = len(item_list)
#請select出此次執行已進入SQL行數(改成已輸入的DATA數量)
sql = "select count(*) as cou from web1_test;"
#SQL query 請更改 ip account password databases query
db, cursor, accounts = monitor.pymysqlcon(logip, loguser, logpw, logdb,
sql)
aSQL = accounts[0]['cou']
except:
for keyword in keyword_list:
#process1
processnum = 1
try:
search_engine = driver.find_element_by_id('kwds')
search_engine.send_keys(keyword)
driver.find_element_by_id('gtm_search_job_web').click()
state = "success"
errnum = 0
print('輸入關鍵字成功,現在正在爬的關鍵字是：',keyword)
except Exception as e:
state = "fail"
errnum = 1
serialnum = monitor.serialnum(serialnumfile,codenum,errnum)
monitor.toErrFile(errlogfilename,daytime,filename,e,serialnum,processnum,keyword)
try:
driver.find_element_by_css_selector('#linkpage > span.sum > em')
except:
url_counts = 0
print('應進入行數:',url_counts)
else:
url_counts = int(driver.find_element_by_css_selector('#linkpage > span.sum > em').text.replace(',',''))
print('應進入行數:',url_counts)
link_url = []
job_details_all = []
#process2
