R/oscn_scrape_dates.R

In openjusticeok/ojo: Access and analyze data from the Open Justice Oklahoma database

oscn_scrape_dates <- function(courts, casetypes, from_mdy, to_mdy, updatedb = FALSE) {
  ctlist <- casetype_codes %>%
    filter(casetype %in% casetypes)
  
  from_m <- str_extract(from_mdy, "\\d{1,2}(?=(/|-))")
  from_d <- str_extract(from_mdy, "(?<=(/|-))\\d{1,2}(?=(/|-))")
  from_y <- str_extract(from_mdy, "(?<=(/|-))\\d{4}$")
  
  to_m <- str_extract(to_mdy, "\\d{1,2}(?=(/|-))")
  to_d <- str_extract(to_mdy, "(?<=(/|-))\\d{1,2}(?=(/|-))")
  to_y <- str_extract(to_mdy, "(?<=(/|-))\\d{4}$")
  
  for (l in courts) {
    court_tmp <- l
    
    for (k in ctlist$casetype_id) {
      casetype_id <- k
      
      d <- read_html(paste0("http://www.oscn.net/dockets/Results.aspx?db=",
                            str_remove_all(court_tmp, " "),
                            "&number=&lname=&fname=&mname=&DoBMin=&DoBMax=&partytype=&apct=&dcct=",
                            casetype_id,
                            "&FiledDateL=",
                            from_m, "%2F", from_d, "%2F", from_y,
                            "&FiledDateH=&ClosedDateL=&ClosedDateH=&iLC=&iLCType=&iYear=&iNumber=&citation="))
      
      if (d %>% html_node("body") %>%
          html_text() %>%
          str_detect("Found No Records") == FALSE) {
        
        d <- d %>%
          html_node("table") %>%
          html_table()
        
        d$seq <- str_extract(d$`Case Number`, "(?<=-)\\d{1,5}($|A)") %>% str_remove_all("[[:alpha:]]") %>% as.numeric()
        d$casetype <- str_extract(d$`Case Number`, "\\w{2}(?=-)")
        
        from_seq <- min(d$seq)
      }
      
      d <- read_html(paste0("http://www.oscn.net/dockets/Results.aspx?db=",
                            str_remove_all(court_tmp, " "),
                            "&number=&lname=&fname=&mname=&DoBMin=&DoBMax=&partytype=&apct=&dcct=",
                            casetype_id,
                            "&FiledDateL=",
                            to_m, "%2F", to_d, "%2F", to_y,
                            "&FiledDateH=&ClosedDateL=&ClosedDateH=&iLC=&iLCType=&iYear=&iNumber=&citation="))
      
      if (d %>% html_node("body") %>%
          html_text() %>%
          str_detect("Found No Records") == FALSE) {
        
        d <- d %>%
          html_node("table") %>%
          html_table()
        
        d$seq <- str_extract(d$`Case Number`, "(?<=-)\\d{1,5}($|A)") %>% str_remove_all("[[:alpha:]]") %>% as.numeric()
        d$casetype <- str_extract(d$`Case Number`, "\\w{2}(?=-)")
        
        to_seq <- max(d$seq)
      }
      
      oscn_scrape(courts = court_tmp, casetype = as.character(ctlist[ctlist$casetype_id == casetype_id, 1]),
                  years = as.numeric(from_y), case_seqs = from_seq:to_seq, updatedb = FALSE)
    }
  }
}