Crawler_airlinetiki_cola.R

library(httr)
library(rvest)
library(XML)
library(testit)
library(magrittr)
library(dplyr)

# origin="TPE"
# destination="SEA"

getAirlineCode <- function(origin="TPE", destination="SEA") {
  url = paste0("http://www.colatour.com.tw/C10C_AirTicket/C10C_00_Query.aspx?TxType=Airline&Destination=",destination,"&Origin=",origin,"&AirlineCode=")
  dat <- url %>% GET %>% content(as = "text",encoding = "utf8") %>% `Encoding<-`("UTF-8") %>% xmlToDataFrame
  dat %<>% mutate(AirlineName = iconv(AirlineName, "utf-8", "big5")) 
  dat
}

getAirlineCode("TPE","SFO")



getFlightsData <- function(fareType="C",  # C=business class &Ｙ=econemic class
                           startDate="2016/09/03",
                           endDate="2016/09/03",
                           journeyType="來回",
                           originCode="TN",
                           destinationCode="SEA",
                           bankName="全部") {
  journeyType %<>% `Encoding<-`("UTF-8")
  bankName %<>% `Encoding<-`("UTF-8")
  Sys.setlocale(category = "LC_ALL", locale = "C")
  
  url = paste0("http://www.colatour.com.tw/C10C_AirTicket/C10C_01_FareChoose.aspx?",
               "FareType=",fareType,
               "&JourneyType=",URLencode(URLencode(journeyType)),
               "&OriginCode=",originCode,"&DestinationCode=",destinationCode,
               "&ReturnCode=&ReturnName=&RouteSummary=",paste0(originCode,"-",destinationCode,"-",originCode),
               "&AirlineCode=&StartDate=",startDate,"&EndDate=",endDate,
               "&BankName=",URLencode(bankName),"&DirectFlightMark=False")
  url
  
  tables = url %>%
    GET %>%
    content(as = "text",encoding = "utf8") %>%
    `Encoding<-`("UTF-8") %>% 
    read_html %>%
    html_table(trim = F,fill = T)
  Sys.setlocale(category = "LC_ALL", locale = "Chinese (Traditional)_Taiwan.950")
  
  tables
  
  # tables[[8]] %>% View
}

tb <- getFlightsData(fareType="Y")

data_table_filter = sapply(tb, function(tab){
  tab[1,1] == "序號"
  # print(tab[1,1])
})
data_table_filter

df <- tb[data_table_filter][[1]]
df %>% View