news_seo1/gen_seo3.py

import traceback
import dataset
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.common.keys import Keys
from selenium.webdriver.chrome.service import Service
import json
import redis
import sys
import random
import os
import time
#from userAgentRandomizer import userAgents

import requests
#import pymysql

#pymysql.install_as_MySQLdb()
driver = None

db = dataset.connect('mysql://choozmo:pAssw0rd@db.ptt.cx:3306/seo?charset=utf8mb4')

headers = {
    "Authorization": "Bearer " + "6SDULL1Ebklduc6TFxa97AFto5Sj21kyJ30CxiLiSoi",
    "Content-Type": "application/x-www-form-urlencoded"
}


def send_msg(kw):
    params = {"message": kw}
    r = requests.post("https://notify-api.line.me/api/notify", headers=headers, params=params)


def re_get_webdriver():
    global port
    global driver
    result = []
    if driver is not None:
        print('closing....')
        driver.quit()
        os.system('killall chrome')
        print('quit....')
        driver = None
    try:
        s = Service('/root/driver/chromedriver102')
        options = webdriver.ChromeOptions()
        options.add_argument("--no-sandbox")
        options.add_argument("--disable-dev-shm-usage")
        options.add_argument("--headless")
        #options.add_argument('--remote-debugging-port=9222')
        #options.add_experimental_option("debuggerAddress", '127.0.0.1:9927')
        options.add_argument("--incognito")
        r = redis.Redis(host='db.ptt.cx', port=6379, db=2, password='choozmo9')
        data = r.get('google_proxy')
        jstext = data.decode('utf-8')
        jsobj = json.loads(jstext)
        proxy = random.choice(jsobj)
        change_ip_list = ['--proxy-server=%s' % proxy, "--proxy-server=socks5://127.0.0.1:9050",
                          "--proxy-server=socks5://192.53.174.202:8180"]
        change_ip = random.choice(change_ip_list)
        options.add_argument('--proxy-server=socks5://172.104.93.163:41800')
        print('使用代理ip', change_ip)
        driver = webdriver.Chrome(options=options)
        #driver.delete_all_cookies()
        driver.set_window_size(1400, 1000)
    except:
        traceback.print_exc()
        driver = None
        return None


def run_once(jsobj):
    table = db['rank_detection']
    print(jsobj)
    global driver
    
    #    i=random.randint(0,9)
    i = 100
    if driver is None:
        time.sleep(8)
        re_get_webdriver()
    if driver is None:
        return
    try:
        kw = jsobj['kw']
        if jsobj.get('domain') is None:
            exclude = jsobj['exclude']
            domain = None
        else:
            domain = jsobj['domain']
            exclude = None
        
        #        driver.get('https://www.google.com?num=100')
        driver.get('https://www.google.com?num=20')
        
        time.sleep(3)
        print(driver.current_url)
        elmt = driver.find_element(By.XPATH, "//input[@name='q']")
        time.sleep(1)
        elmt.send_keys(kw)
        elmt.send_keys(Keys.ENTER)
        time.sleep(6)
        print(driver.current_url)
        elmts = driver.find_elements(By.XPATH, "//div[@class='yuRUbf']/a")
        
        numresults = len(elmts)
        print('搜尋結果數量', numresults)
        if numresults == 0:
            send_msg('stop working...')
            sys.exit()
        
        idx = 1
        found = False
        test_lst = []
        txt_dict = {}
        for elmt in elmts:
            href = elmt.get_attribute('href')
            txt = elmt.text
            if len(txt) > 10:
                if domain is not None:
                    if domain in href:
                        print('found....')
                        print('clicked....')
                        print(href)
                        print(txt)
                        print("ranking", idx)
                        found = True 
                        webdriver.ActionChains(driver).move_to_element(elmt).perform()
                        webdriver.ActionChains(driver).move_to_element(elmt).click().perform()
                        table.insert({'ranking': idx, 'kw': kw, 'results': numresults, 'url': href, 'title': txt})
                        time.sleep(6)
                        break
                else:
                    ex = False
                    for ee in exclude:
                        if ee in href:
                            ex = True
                    if not ex:
                        test_lst.append(elmt)
                        txt_dict[elmt] = txt
            
            idx += 1
        if exclude is not None:
            print('exclude')
            elmt = random.choice(test_lst[5:])
            print(elmt)
            print(txt_dict[elmt])
            
            webdriver.ActionChains(driver).move_to_element(elmt).perform()
            elmt.click()
            #            webdriver.ActionChains(driver).move_to_element(elmt).click().perform()
            time.sleep(5)
        
        if not found:
            table.insert({'ranking': -1, 'kw': kw, 'results': numresults, 'url': '', 'title': '未收錄'})
    
    except:
        print('exception')
        traceback.print_exc()

    sys.exit()
    driver.quit()

    db.close()

# par1=sys.argv[1]
# port=sys.argv[2]

# kws=['職籃','PLG','高雄','鋼鐵人','內幕','中資','股東','姊夫','中國','老賴','香港','無極','原始股東','外資','董事長','股權結構','高雄人','黑人','陳建州','職籃聯盟','球團','球團高層','香港無極','張憲銘','吳同喬','監察人']
kws = ['金融', '人才', '國際接軌', '國際', '投資金童', '投資', '金童', '對沖基金', '香港', '外資', '原始股東', '職籃', 'PLG', '職籃聯盟', '球團', '台灣女婿',
       '抹紅', '保守', '港元', '美國', '升息', '戰爭', '通膨', '亞洲', '亞洲投資金童']
positive = ['錢濤','亞洲投資金童', '錢濤 職籃夢']
#os.system('docker container restart tiny6')
kw = random.choice(kws)
# time.sleep(9)
# run_once({'domain':'ettoday.net','kw':'錢濤'})
# run_once({'exclude':['moreptt.com','ptt.cc','tnews.cc','mirrormedia.mg','newtalk.tw','pourquoi.tw','match.net.tw','freshweekly.tw','z-upload.facebook.com','udn.com'],'kw':kw+' 錢濤'})
domains = ['yahoo.com', 'ettoday.net', 'tvbs.com.tw', 'sina.com.tw', 'ltn.com.tw', 'owlting.com', 'ctee.com.tw']
domain = random.choice(domains)
p = random.choice(positive)
# run_once({'domain':domain,'kw':p})

run_once({'domain': 'ettoday.net', 'kw': p})

# run_once({'domain':domain,'kw':kw+' 錢濤'})