R/oscn_refresh_sc.R

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

oscn_refresh_sc <- function() {

  six_mo <- Sys.Date() - 92

  connect_ojo()

  lu <- dbGetQuery(ojo_db, glue_sql(
    "SELECT * FROM oscn_civ_disps
                WHERE court = 'TULSA'
                AND file_date > {six_mo}
                AND casetype = 'SC'", .con = ojo_db)) %>%
    mutate(case_seq = casenum %>%
             str_sub(9,13) %>%
             as.numeric)

  disconnect_ojo()

  lu <- lu %>%
    group_by(casenum) %>%
    mutate(no_disp = all(is.na(disp))) %>%
    filter(no_disp == TRUE) %>%
    filter(!(all(is.na(defname)) & file_date < Sys.Date() - 30))

  today <- Sys.Date()

  ht <- read_html(paste0("http://www.oscn.net/applications/oscn/report.asp?report=DailyFilings&errorcheck=true&database=&db=Tulsa&StartDate=", month(today), "%2F", day(today), "%2F", year(today)))


  lastfiled <- oscn_lastcase("TULSA", "SC", year(Sys.Date()))

  lastscraped <- lu[ nrow(lu), "casenum"] %>%
    str_sub(9, 13) %>%
    as.numeric

  start_time <- Sys.time()

  oscn_scrape("TULSA",
              "SC",
              year(Sys.Date()),
              lastscraped:lastfiled,
              update_freq = 20)

  ccy <- lu %>%
    ungroup() %>%
    select(court, casetype, file_year) %>%
    distinct()

  for (i in 1:nrow(ccy)) {
    seqs <- lu %>%
      filter(court == ccy[i, "court"], casetype == ccy[i, "casetype"] & file_year == ccy[i, "file_year"])

    seqs <- seqs$case_seq

    oscn_scrape("TULSA",
                "SC",
                ccy[i, "file_year"],
                seqs,
                update_freq = 20)
  }

  elapsed <- as.numeric(Sys.time() - start_time) %>%
    round(digits = 2)

}