Ejemplo n.º 1
0
def runSkyScannerHK():
    log.v("start to run")
    if data.returnDate.__eq__("") | data.departDate.__eq__("") | data.departCityCode.__eq__("") | data.arriveCityCode.__eq__(""):
        log.v("invalid param")
        return
    return_flight.setParamData(data)
    return_flight.runTask()
Ejemplo n.º 2
0
def runTask():
    filename = param.departCityCode + "-" + param.arriveCityCode + "-" + param.departDate + "-" + param.returnDate + "-" + str(int(time.mktime(datetime.datetime.now().timetuple())))
    csvUtil.filename = filename
    csvUtil.createReturnFile(filename)

    log.v("csv fire created, now waiting for internet response")
    # try:
    #
    #     PROXY = "127.0.0.1:1080"
    #     chrome_options = webdriver.ChromeOptions()
    #     chrome_options.add_argument('--proxy-server={0}'.format(PROXY))
    #     driver = webdriver.Chrome(chrome_options=chrome_options)
    #
    #     # driver = webdriver.Chrome()
    #
    #     driver.get(param.createReturnUrl())
    #
    #     # element = WebDriverWait(driver, 90).until(
    #     #     EC.presence_of_element_located((By.CSS_SELECTOR, "article")))
    #     time.sleep(20)
    #     # driver.implicitly_wait(30)
    #     articles = driver.find_elements_by_css_selector("article")
    #     log.v("found articles")
    #     log.v("end driver wait")
    #     time.sleep(5)
    #     log.v("end wait")
    #
    #     getInfo(articles)
    # finally:
    #     time.sleep(10)
    #     driver.quit()
    # #


    # PROXY = "127.0.0.1:1080"
    # chrome_options = webdriver.ChromeOptions()
    # chrome_options.add_argument('--proxy-server={0}'.format(PROXY))
    #
    # driver = webdriver.Chrome(chrome_options=chrome_options)

    # driver = webdriver.Chrome()


        # driver.delete_all_cookies()
    profile = webdriver.FirefoxProfile()
    profile.native_events_enabled = True
    driver = webdriver.Firefox(profile)

    driver.get(param.createReturnUrl())

        # element = WebDriverWait(driver, 90).until(
        #     EC.presence_of_element_located((By.CSS_SELECTOR, "article")))
    time.sleep(50)
        # driver.implicitly_wait(30)

    log.v("end driver wait")

    getInfo(driver)

    time.sleep(200)
Ejemplo n.º 3
0
def setReturnParams(departure, destination, departDate, returnDate):
    data.departCityCode = departure
    data.arriveCityCode = destination
    data.departDate = departDate
    data.returnDate = returnDate
    log.printLine()
    log.i("departure", departure)
    log.i("destination", destination)
    log.i("departDate", departDate)
    log.i("returnDate", returnDate)
    log.printLine()
    log.v("set params done")
Ejemplo n.º 4
0
def check_web(departure, arrival):

    log.v("start running")
    try:
        PROXY = ip_list.ips[0]
        chrome_options = webdriver.ChromeOptions()
        chrome_options.add_argument('--proxy-server={0}'.format(PROXY))
        driver = webdriver.Chrome(chrome_options=chrome_options)
        # driver.delete_all_cookies()
        driver.get(url)
        # element = WebDriverWait(driver, 90).until(
        #     EC.presence_of_element_located((By.CLASS_NAME, "day-list-item"))
        # )
        log.v("end driver waiting")
        time.sleep(30)
        # element = WebDriverWait(driver, 90).until(
        #     EC.presence_of_element_located((By.CLASS_NAME, "day-list-item"))
        # )
        # time.sleep(10)
        log.v("end waiting")

        getInfo(driver)
    # except Exception as e:
    #     print(e)
    #     driver.quit()
    finally:
        time.sleep(5)
        driver.quit()

        log.v("end")
Ejemplo n.º 5
0
def getInfo(driver):
    articles = driver.find_elements_by_class_name("day-list-item")
    log.v("start")
    for article in articles:
        log.v("get articles")
        # try:
        time.sleep(5)

        article.click()
        log.v("click open")
        time.sleep(5)

        popup_panel = driver.find_element_by_id("details-mobile-panel")
        log.v("found details-mobile-panel")
        # flight_content = popup_session.find_elements_by_class_name("fss-bp-itinerary")
        # print(">>>found fss-bp-itinerary")
        # count = 0
        # real_content = webdriver
        # for popup_session in popup_sessions:
        #     print(">>>found details-mobile-panel count")
        #     try:
        #         count += 1
        #         popup_session.find_element_by_class_name("fss-panel-content")
        #         real_content = popup_session
        #         break
        #     except:
        #         print(count)

        legs = popup_panel.find_elements_by_class_name("itinerary-leg")
        for leg in legs:
            time.sleep(5)
            leg.click()
            log.v("click leg")
            airlines = leg.find_elements_by_class_name("operated-by")
            for airline in airlines:
                log.i("airline", airline.text)
                depart = leg.find_element_by_class_name("departure")
                log.i("depart time",
                      depart.find_element_by_class_name("times").text)
                log.i("depart airport",
                      depart.find_element_by_class_name("route").text)
                destination = leg.find_element_by_class_name("destination")

                log.i("arrive time",
                      destination.find_element_by_class_name("times").text)
                log.i("arrive airport",
                      destination.find_element_by_class_name("route").text)

        prices = popup_panel.find_elements_by_class_name("price")
        for price in prices:
            log.i("price", price.text)
            # a = popup_session.find_element_by_class_name("fss-panel-content")

            log.v("click legs")

        time.sleep(5)

        driver.find_element_by_id("fss-overlay").click()
        log.v("click close")
Ejemplo n.º 6
0
from webspider.src.spider.wsdomestic import check_web
from webspider.src.spider import skyscanner
from webspider.src.logger import log

from webspider.src import run_spider

# this is the main entrance of the program

# check_web("")

log.w("test Warning")
log.e("test Error")
log.v("test Verbose")
log.i("test Info", "info")

# skyscanner.check_web("", "")

departure = "can"  # departure airport code
destination = "sha"  # destination airport code

departDate = "170613"  # depart date (format: YYMMDD)
returnDate = "170701"  # arrive date (format: YYMMDD)

run_spider.setReturnParams(departure, destination, departDate, returnDate)

run_spider.runSkyScannerHK()
Ejemplo n.º 7
0
def main():

    log.v("program is starting and waiting for web respond")

    return_flight.setParamData(data)
    return_flight.runTask()
Ejemplo n.º 8
0
def getInfo(driver):
    articles = driver.find_elements_by_css_selector("article")
    log.v("found articles")

    time.sleep(5)
    log.v("end wait")
    # articles = driver.find_elements_by_class_name("day-list-item")
    log.v("start")
    log.v("get articles")
    for article in articles:
        entity = return_entity

        log.v("article start")
        time.sleep(2)
        sections = article.find_elements_by_css_selector("section")
        log.v("found sections")

        count = 0
        for section in sections:
            # time.sleep(3)

            try:
                bigairline = section.find_element_by_class_name("big-airline")
                log.v("found big-airline")
            except:
                continue

            # time.sleep(1)

            try:
                airline = bigairline.find_element_by_class_name("text-sm")
                log.v(airline.text)
                if count == 0:
                    entity.departairline = airline.text
                else:
                    entity.returnairline = airline.text
                log.v("found text-sm")
            except:
                try:
                    airline = bigairline.find_element_by_class_name("big")
                    log.v(airline.get_attribute("alt"))
                    if count == 0:
                        entity.departairline = airline.get_attribute("alt")
                    else:
                        entity.returnairline = airline.get_attribute("alt")
                    log.v("found img")
                except:
                    continue
            finally:
                log.v("count:" + str(count))

            try:
                # time.sleep(1)
                stations = section.find_elements_by_class_name("station-tooltip")
                log.v("found station-tooltip")
                for i in range(2):
                    times = stations.__getitem__(i).find_element_by_class_name("times")
                    airport = stations.__getitem__(i).find_element_by_class_name("stop-station")
                    log.v("time:" + times.text)
                    log.v("airport:" + airport.text)
                    if count == 0:
                        if i == 0 :
                            entity.departtime = times.text
                            entity.departairport = airport.text
                        else:
                            entity.arrivetime = times.text
                            entity.arriveairport = airport.text
                    else:
                        if i == 0 :
                            entity.returndeparttime = times.text
                            entity.returndepartairport = airport.text
                        else:
                            entity.returnarrivetime = times.text
                            entity.returnarriveairport = airport.text

            except:
                continue
            finally:
                count = count + 1
                log.v("section end")


        price = article.find_element_by_class_name("mainquote-price")
        log.v("price:"+price.text)

        entity.cheapestprice1 = price.text

        entity.print()
        csvUtil.writeReturnEntity(entity)
        log.v("article end")

    log.v("page end")
    try:
        next = driver.find_element_by_class_name("day-list-pagination").find_element_by_class_name("next").find_element_by_css_selector("button")
        log.v("found next button end")
        next.click()
        log.v("go to next page")
        time.sleep(2)
        getInfo(driver)
    except:
        log.v("finish")
        driver.quit()
Ejemplo n.º 9
0
def setParamData(data):
    param = data
    mUrl = data.createReturnUrl()
    log.v(param.createReturnUrl())