get_colatour_data.R

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

# 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=")
  url %>% GET %>% content(as = "text",encoding = "utf8") %>% `Encoding<-`("UTF-8") %>% xmlToDataFrame
}

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")
  
  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)

  # tables[[8]] %>% View

  data_table_filter = sapply(tables, function(tab){
    tab[1,1] == "序號"
  })

  tables[data_table_filter][[1]]
}


df = getFlightsData(fareType="Y")
df %>% View