add ps_read_vrl family?

[sebdalgarno]

Vrl output csv come in a standard (messy) format and are therefore easy to get into a format that we want.
similar to other read functions, ps_read_vrl takes a single file. and ps_read_vrls takes a directory. A tz argument allows automatic conversion of datetime column into desired tz.

ps_read_vrl <- function(path = '~/Dropbox/Data/quesnel-exploitation/20170208/test.csv', tz = getOption("ps.tz", "UTC"),
                        datetime_column = "DateTimeDetection", receiver_column = "Receiver", transmitter_column = "Transmitter") {
  # need to check that the selected csv fits standard format of vrl output
  data <- readr::read_csv(path) %>% 
    select(1:3)
  colnames(data) <- c(datetime_column, receiver_column, transmitter_column)
  data[[datetime_column]] %<>% with_tz(tz = tz)
  data
}

ps_read_vrls <- function(dir = ".", tz = getOption("ps.tz", "UTC")) {
  files <- list.files(path = dir, pattern = "[.]csv$", full.names = TRUE, recursive = TRUE)
  data <- purrr::map_dfr(files, ~ ps_read_vrl(., tz = tz))
  data
}

[joethorley]

Good idea although I'm thinking we only need a ps_read_vrl and it can be combined with the general function ps_load_files - thoughts?

[sebdalgarno]

I'm not sure ps_load_files is what we want for this. I want to take a directory and combine all files within that directory into a single data.frame which can be named...more like what ps_read_tracks_gpxs does.

[joethorley]

if we were to use ps_load_files to pull into a clean environment and then ps_bind_datas this would get us there without having to write a specific multi file function for each file type.

an alternative option might be to use a function factory to generate them http://adv-r.had.co.nz/Functional-programming.html

My issue with manually writing functions with identical code save the bit to read in an individual file is that its at odds with the do not repeat yourself (DRY) principle

[sebdalgarno]

good points.
Perhaps a good option is a simple wrapper of list.files with useful default args, in conjunction with map_dfr:

vrl_df <- map_dfr(ps_list_files("~/Poisson/..."), ~ ps_read_vrl(.))

where defaults args are:
pattern = "[.]csv$", full.names = TRUE, recursive = TRUE

I might not be seeing the full utility of the ps_load_files/ps_bind_datas combo...

[joethorley]

let's chat about more in person when we meet to discuss the ssdca shiny app tomorrow