##################################################################
# Generate case data file for hospitalised patients
# - date of symptom onset
# - date of admission
# - date admission reported
# - infected abroad (yes/no)
##################################################################

library(tidyverse)
library(lubridate)

generate_case_data_reports <- function(date = today()) {
  
  files <- list.files("/___UITZONDERINGSGROND_6___Previous", full.names = TRUE)
  files <- c(files, list.files("/___UITZONDERINGSGROND_6___Geschoond", full.names = TRUE))
  files <- files[grepl(files, pattern = "rds")]
  files <- tibble(file_name = files, 
                  date = files %>% str_extract_all(pattern = "(Data_[0-9]+)") %>% unlist %>% gsub(pattern = "Data_", replacement = "") %>% as.Date(format = "%Y%m%d"),
                  time = files %>% file.info %>% pull(ctime))
  
  case_data_filename <- "/___UITZONDERINGSGROND_6___case_data_reports.rds"
  
  case_data <- if(file.exists(case_data_filename)) {
    readRDS(case_data_filename) 
    } else {
      tibble(admission_date = min(files$date), admission_date_reported = min(files$date), admission_reported = min(files$date))
    }
  
  if(date > max(case_data$admission_reported)) {
    report_dates <- seq(max(case_data$admission_reported), date, by = "day") %>% as.character
  } else {
    return(case_data) # remember to cut-off the case data after date when using it in main scrips
  }
  
  for(report_date in report_dates) {
    file_name <- files %>% 
      filter(date == report_date) %>%
      filter(time == max(time)) %>% # get most recent file (if multiple files exist) 
      pull(file_name)
    
    # if data is not available
    if(length(file_name)==0) {
      print(paste("file ", report_date, "not available"))
      return(0)
    }
    
    case_data <- read_rds(file_name) %>%
    {if("EIGENAARDesc" %in% names(.)) filter(., !(EIGENAARDesc %in% c("Sint Eustatius", "Bonaire", "Saba", "Sint Maarten", "Aruba", "Curaçao", "Curacao"))) else . } %>% 
      mutate(admission_date = MELGGDOntvDt) %>%
      mutate(admission_date = as.Date(admission_date, tryFormats = c("%d-%m-%Y", "%Y-%m-%d"))) %>% 
      #filter(is.na(admission_date) | (admission_date >= as.Date("2020-02-01") & admission_date <= date)) %>%
      select(OSIRISNR, admission_date) %>% 
      full_join(case_data) %>%
      replace_na(replace = list(admission_date_reported = report_date, admission_reported = report_date)) %>%
      mutate(admission_date = if_else(admission_date >= as.Date("2020-02-01") & admission_date <= date, admission_date, ymd(NA)),
             admission_date_reported = if_else(is.na(admission_date), ymd(NA), admission_date_reported))
    
  }
  
  case_data <- case_data  %>% 
    group_by(OSIRISNR) %>%
    summarise(admission_date = admission_date[which.max(admission_reported)],
              admission_date_reported = min(admission_date_reported, na.rm = TRUE),
              admission_reported = min(admission_reported))
  
  file_name <- files %>% 
    filter(date == last(report_dates)) %>%
    filter(time == max(time)) %>% # get most recent file (if multiple files exist) 
    pull(file_name)
  
  case_data <- read_rds(file_name) %>% 
  {if("EIGENAARDesc" %in% names(.)) filter(., !(EIGENAARDesc %in% c("Sint Eustatius", "Bonaire", "Saba", "Sint Maarten", "Aruba", "Curaçao", "Curacao"))) else . } %>% 
    mutate(# infected_abroad = ifelse(Bron_land == "Buitenland", "Yes", "No"),
           infected_abroad = ifelse(Bron_buitenland %in% c("Waarschijnlijk buitenland", "Mogelijk buitenland"), "Yes", "No"),
           symptom_onset = ZIE1eZiekteDt,
           no_symptoms = if_else(NCOVVast1eziektedag == "NVT", 1, 0, 0)) %>%
    select(OSIRISNR, symptom_onset, no_symptoms, infected_abroad, Provincie, Veiligheidsregio) %>% 
    left_join(case_data) %>% 
    #select(symptom_onset, admission_date, admission_date_reported, admission_reported, infected_abroad) %>% 
    filter(!(is.na(admission_date) & is.na(symptom_onset)),
           is.na(symptom_onset) | symptom_onset >= as.Date("2020-02-01"),
           is.na(admission_date) | admission_date >= as.Date("2020-02-01"))
  
  saveRDS(case_data, case_data_filename)
  
  return(case_data)
}