line_analysis.Rmd

---
title: "File Length Analysis"
author: "Alex Gold"
date: "3/27/2020"
output: 
  html_document:
    code_folding: hide
---
```{r, setup, message = FALSE}
library(dplyr)
library(ggplot2)
library(fs)
```

```{r, warning = FALSE}
dirs <- fs::dir_ls(".") %>%
  purrr::keep(function(x) stringr::str_detect(x, "^0"))

get_wc <- function(x) {
  system2("wc", c("-l", x), stdout = TRUE) %>%
    stringr::str_match("(\\d+) ") %>%
    magrittr::extract2(2) %>%
    as.numeric()
}

files <- dirs %>%
  purrr::map_dfr(
    function(x) tibble::tibble(
    file = fs::dir_ls(x, recurse = TRUE) %>%
      as.character(), 
    dir = x
    )
  ) %>%
  filter(
    stringr::str_detect(file, "\\.R$"), 
         !stringr::str_detect(file, "utils-pipe")
    ) %>%
  mutate(
    lines = purrr::map_dbl(file, ~ get_wc(.)), 
    is_app = stringr::str_detect(file, "app")
  ) %>%
  group_by(dir) %>%
  mutate(tot_lines = sum(lines), 
         mean_lines = mean(lines), 
         n_files = glue::glue("N Files: {n()}")) %>%
  ungroup()

```

```{r}
files %>%
  filter(is_app) %>%
ggplot(aes(x = dir, y = tot_lines)) +
  geom_col(fill = "#5782b2") +
  theme_minimal() +
  xlab("Version") +
  ylab("Total Lines") +
  ggtitle("Total Number of Lines Goes Up")
```

```{r}
files %>%
  ggplot(aes(x = dir)) +
  geom_line(data = filter(files, is_app), 
            aes(y = mean_lines, group = 1),
            color = "#5782b2") +
  geom_label(aes(y = 100, label = n_files)) +
  geom_point(aes(y = lines)) +
  ylab("Mean Lines Per File") +
  theme_minimal() +
  ggtitle("Lines Per File Goes Down a Lot")
```