Parses YS order emails and brings them under a single data frame

parse_ys_gmail_threads.R

options(stringsAsFactors=FALSE)
library(tidyverse)
library(rvest)
library(gmailr)

## FOLLOW AUTH INSTRUCTIONS FROM HERE https://gmailr.r-lib.org/articles/gmailr.html

parse_order_table <-function(my_msg,full_info_list=FALSE){
  my_info <- read_html(gm_body(my_msg)) %>% html_nodes(xpath="/html/body/table/tr/td/center/table[1]/tr[2]/td") %>% html_children() %>% `[[`(2)

  info_data <- tibble(info_raw = str_trim(my_info %>% html_nodes("td.padding-right-20") %>% html_text()))

  info_data <- info_data %>% separate(info_raw, into=c("key","value"), sep=":",extra="merge") %>% mutate_all(str_trim)

  info_list <- info_data %>% spread(key,value) %>% as.list()

  my_order <- read_html(gm_body(my_msg)) %>% html_nodes(xpath="/html/body/table/tr/td/center/table[1]/tr[2]/td") %>% html_children() %>% `[[`(3)

  the_place <- my_order %>% html_nodes("td.sub-columns.twelve.padding-left-20") %>% html_text()

  order_items <- my_order %>% html_nodes("table.basket-item.full-width.border-bottom-e5e5e5.nbs") %>% html_nodes("td.padding-top-10.sub-columns.six.left-text-pad.nls") %>% html_text()

  item_unit_costs_and_units <- my_order %>% html_nodes("table.basket-item.full-width.border-bottom-e5e5e5.nbs") %>% html_nodes("td.padding-top-10.sub-columns.two.center.last") %>% html_text()

  item_unit_costs <- as.numeric(gsub(",",".",gsub("\\.","",gsub(" TL","",item_unit_costs_and_units[(1:length(item_unit_costs_and_units)) %% 2 == 1]))))
  item_units <- as.numeric(item_unit_costs_and_units[(1:length(item_unit_costs_and_units)) %% 2 == 0])

  item_total_costs <- my_order %>% html_nodes("table.basket-item.full-width.border-bottom-e5e5e5.nbs") %>% html_nodes("td.padding-top-10.sub-columns.two.right.last") %>% html_text()
  item_total_costs <- as.numeric(gsub(",",".",gsub("\\.","",gsub(" TL","",item_total_costs))))

  order_df <- tibble(date=parse_date(info_list$`Gönderim Tarihi`,format="%d.%m.%Y"),city=info_list$`Şehir`,neighborhood=info_list$`Semt`,place=the_place,item=order_items,unit_cost=item_unit_costs,units=item_units,total_cost=item_total_costs)

  if(full_info_list){
    return(list(order_df=order_df,info_list=info_list))
  }

  return(order_df)
}


ys_threads <- gm_threads("from:siparis@news.yemeksepeti.com",num_results = 500)

the_vec <- c()

for(i in 1:length(ys_threads)){
  the_sublist <- ys_threads[[i]]$threads
  for(j in 1:length(the_sublist)){
    the_vec <- c(the_vec, ys_threads[[i]]$threads[[j]]$id)
  }

}

all_orders_df <- tibble()

for(i in 1:length(the_vec)){
  print(i)
  latest_thread <- gm_thread(the_vec[i])
  my_msg <- latest_thread$messages[[1]]
  if(grepl("Yemeksepeti Sipariş Onay",gm_subject(my_msg))){
    tryCatch({all_orders_df <- bind_rows(all_orders_df,parse_order_table(my_msg))},error=function(e){print(e)})
  }else{
    print("This is not an order. Order titles start with Yemeksepeti Sipariş Onay")
  }
  print("======")
  Sys.sleep(0.5)
}


print(all_orders_df)