---
title: "Shiny-app"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Shiny-app}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

To make the easiest possible transition from spreadsheet/dataset to REDCap, I have created a small app, which adds a graphical interface to the casting of a data dictionary and data upload. Install the package and launch the app as follows:

```{r eval=FALSE}
REDCapCAST::shiny_cast()
```

The app primarily wraps one function: `ds2dd_detailed()`.

```{r}
library(REDCapCAST)
ds <- REDCap_split(
    records = redcapcast_data,
    metadata = redcapcast_meta,
    forms = "all"
  ) |> 
  sanitize_split() |>
  redcap_wider() 
str(ds)
```

```{r}
ds|> 
  ds2dd_detailed()|>  
  purrr::pluck("data") |> 
  str()
```

```{r}
ds|> 
  ds2dd_detailed()|>  
  purrr::pluck("meta") |> 
  head(10)
```

Different data formats are accepted, which all mostly implements the `readr::col_guess()` functionality to parse column classes.

To ensure uniformity in data import this parsing has been implemented on its own to use with `ds2dd_detailed()` or any other data set for that matter:

```{r}
ds_parsed <- redcapcast_data |> 
  dplyr::mutate(dplyr::across(dplyr::everything(),as.character)) |> 
  parse_data()
str(ds_parsed)
```

It will ignore specified columns, which is neat for numeric-looking strings like cpr-with a leading 0:

```{r}
redcapcast_data |> 
  dplyr::mutate(dplyr::across(dplyr::everything(),as.character)) |> 
  parse_data(ignore.vars = c("record_id","cpr")) |> 
  str()
```


```{r}

```


Column classes can be passed to `parse_data()`.

Making a few crude assumption for factorising data, `numchar2fct()` factorises numerical and character vectors based on a set threshold for unique values:

```{r}
mtcars |> str()
mtcars |>
  numchar2fct(numeric.threshold = 6) |>
  str()
```

```{r}
ds_parsed|>
  numchar2fct(numeric.threshold = 2) |>
  str()
```