analise_curso_r_posts.R

library(tidyverse)
library(fs)

dir_path <- "projetos/site-master/content/blog/"

dir_info(dir_posts, recursive = TRUE) %>% 
  head() %>%
  knitr::kable()

rmd_or_r_file_paths_tbl <- dir_info(dir_path, recursive = T) %>%
  mutate(file_name = path_file(path)) %>%
  select(file_name, path) %>%
  filter(str_detect(file_name, "(\\.R|\\.Rmd)$")) 


build_function_names_tbl_from_file_path <- function(path, loaded_functions_tbl) {
  
  builder <- function(path, loaded_functions_tbl) {
    readLines(path) %>%
      parse_function_names() %>%
      left_join(loaded_functions_tbl) %>%
      filter(
        !(function_name == "filter" & !(package == "dplyr"))
      ) %>%
      mutate(package = ifelse(is.na(package), "Unknown", package))
  }
  
  safe_builder <- possibly(builder, otherwise = NA)
  
  safe_builder(path, loaded_functions_tbl)
}

parse_function_names <- function(text, stop_words = c("")) {
  
  parser <- function(text, stop_words) {
    ret <- text %>%
      str_c(collapse = " ") %>%
      str_split("\\(") %>%
      set_names("text") %>%
      as.tibble() %>%
      slice(-n()) %>%
      mutate(str_split = map(text, str_split, " ")) %>%
      select(-text) %>%
      unnest() %>%
      mutate(function_name = map_chr(str_split, ~ purrr::pluck(last(.x)))) %>%
      select(function_name) %>%
      separate(function_name, into = c("discard", "function_name"), 
               sep = "(:::|::|\n)", fill = "left") %>%
      select(-discard) %>%
      mutate(function_name = str_replace_all(function_name, 
                                             pattern = "[^[:alnum:]_\\.]", "")) %>%
      filter(!(function_name %in% stop_words)) 
    
    return(ret)
  }
  
  safe_parser <- possibly(parser, otherwise = NA)
  
  safe_parser(text, stop_words)
}


find_loaded_packages <- function() {
  
  ret <- search() %>%
    list() %>% 
    set_names("search") %>%
    as.tibble() %>%
    separate(search, into = c("discard", "keep"), sep = ":", fill = "right") %>%
    select(keep) %>%
    filter(!is.na(keep)) %>%
    rename(package = keep) %>%
    arrange(package)
  
  return(ret)
}

find_functions_in_package <- function(package) {
  
  pkg_text <- paste0("package:", package)
  
  safe_ls <- possibly(ls, otherwise = NA)
  
  package_functions <- safe_ls(pkg_text)
  
  if (is.na(package_functions[[1]])) return(package_functions)
  
  ret <- package_functions %>%
    as.tibble() %>%
    rename(function_name = value)
  
  return(ret)
}


map_loaded_package_functions <- function(data, col) {
  
  col_expr <- enquo(col)
  
  data %>%
    mutate(function_name = map(!! col_expr, find_functions_in_package)) %>%
    mutate(is_logical = map_dbl(function_name, is.logical)) %>%
    filter(is_logical != 1) %>%
    select(-is_logical) %>%
    unnest()
}


loaded_functions_tbl <- find_loaded_packages() %>%
  map_loaded_package_functions(package) 




file_path_1 <- rmd_or_r_file_paths_tbl$path[[1]]
file_path_1


build_function_names_tbl_from_file_path(rmd_or_r_file_paths_tbl$path[[4]],
                                        loaded_functions_tbl) %>%
  glimpse()


local_function_names_tbl <- rmd_or_r_file_paths_tbl %>%
  mutate(
    function_name = map(path, build_function_names_tbl_from_file_path, loaded_functions_tbl),
    is_logical    = map_dbl(function_name, is.logical)
  ) %>%
  filter(is_logical != 1) %>%
  select(file_name, function_name) %>%
  unnest() %>%
  left_join(loaded_functions_tbl)

local_function_names_tbl %>% glimpse()

count_to_pct <- function(data, ..., col = n) {
  
  grouping_vars_expr <- quos(...)
  col_expr <- enquo(col)
  
  data %>%
    group_by(!!! grouping_vars_expr) %>%
    mutate(pct = (!! col_expr) / sum(!! col_expr)) %>%
    ungroup()
}



# extrair autor de pacoite
x <- readLines(file_path_1)

extrair_autor <- function(path){
  x <- readLines(path)
  
  linha_autor <- x %>%
    str_subset("^author:") %>% 
    str_remove_all("author: ") %>% 
    str_remove_all("\\[|\\]") %>% 
    str_remove_all('"')
  
  linha_autor <- str_split(linha_autor, ",")[[1]] %>% str_trim()
  autor <- as.tibble(linha_autor)
  colnames(autor) <- "Autor"
  
  autor <- autor %>% mutate(path = path)
  autor
}

extrair_autor(file_path_1)


df_autores <- rmd_or_r_file_paths_tbl$path %>%
  map_df(extrair_autor)

df_autores <- left_join(df_autores, rmd_or_r_file_paths_tbl, by = "path")

head(df_autores)
head(rmd_or_r_file_paths_tbl)


df <- left_join(local_function_names_tbl, df_autores, by = "file_name")


head(df_autores, 1)
head(local_function_names_tbl, 1)


top_20 <- df %>%
  ungroup() %>% 
  count(Autor, package, function_name) %>%
  count_to_pct()

top_20 <- top_20 %>% 
  arrange(desc(n)) %>%
  group_by(Autor) %>% 
  slice(1:20) %>% 
  rowid_to_column(var = "rank")


library(tidyquant)
top_20 %>%
  ggplot(aes(x = n, y = fct_reorder(function_name, n), color = package)) +
  geom_segment(aes(xend = 0, yend = function_name), size = 2) +
  geom_point(size = 4) +
  # geom_label(aes(label = paste0(function_name, "(), ",
  #                               package, ", ",
  #                               scales::percent(pct))), 
  #            hjust = "inward", size = 3.5) +
  expand_limits(x = 0) +
  scale_color_tq() +
  theme_tq() +
  theme(legend.position = "none") + 
  facet_wrap(~ Autor, scales = "free", drop = TRUE)

ggsave("grafico-curso-r.png", height = 14, width = 11)