library(tidyverse)
library(lubridate)

get_symptomonset2reporting <- function(start_date = as.Date("2020-03-13"), end_date = today(), IC = FALSE) {
  
  # get all case data files from Osiris
  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")]
  
  report_dates <- seq(start_date, end_date, by = "day") %>% as.character
  
  SO2rep <- rep(0,40)
  for(report_date in report_dates) {
    pattern_date <- report_date %>% str_split(pattern = "-") %>% unlist %>% paste0(collapse = "")
    tmpfiles <- files[grep(files, pattern = pattern_date)]
    file.name <- tmpfiles[tmpfiles %>% file.info %>% pull(ctime) %>% which.max]
    
    # if today's data is not yet available
    if(length(file.name)==0) {
      print(paste("only files up to ", report_date, "used"))
      return(cumsum(rev(SO2rep)/sum(SO2rep)))
    }
    
    newdata <- read_rds(file.name) %>% 
    {if(IC) filter(., NCOVopnameICU == "J") else filter(., NCOVpatZhs == "J")} %>% 
      filter(!is.na(ZIE1eZiekteDt) & ZIE1eZiekteDt <= end_date) %>%
      group_by(ZIE1eZiekteDt) %>%
      summarise(incidence = n()) %>% 
      full_join(tibble(ZIE1eZiekteDt = seq(min(min(.$ZIE1eZiekteDt), as.Date(report_date)-length(SO2rep)+1), as.Date(report_date), by = "day"))) %>%
      replace_na(list(incidence = 0)) %>%
      arrange(ZIE1eZiekteDt)
    
    if(as.Date(report_date) > start_date) {
      newdata <- full_join(prevdata, newdata, by = "ZIE1eZiekteDt") %>% 
        replace_na(list(incidence.x = 0, incidence.y = 0)) %>%
        arrange(ZIE1eZiekteDt) %>% 
        mutate(newentries = incidence.y - incidence.x) 
      SO2rep <- SO2rep + (newdata %>% pull(newentries) %>% tail(40))
      newdata <- newdata %>% rename(incidence = incidence.y)  %>% select(ZIE1eZiekteDt, incidence)
    }
    
    prevdata <- newdata
  }
  cdf <- cumsum(rev(SO2rep)/sum(SO2rep))
  cdf[cdf > 1] <- 1
  return(cdf)
}