Improve performance of get_files

CHANGELOG.md

@@ -2,112 +2,146 @@
 
 ## Unreleased
 
+- improve performance of `get_files`
+
 ## v1.7.0 - 5 April 2024
+
 - add `create_symlink` function to create a symbolic link
 - add `verify_is_symlink` function to check if a file is a symbolic link
 - add `describe_error` function to get human-readable descriptions out of errors
 
 ## v1.6.1 - 28 March 2024
+
 - fix bug I introduced to `clear_directory_all` which accidentally made absolute directories
   relative.
 
 ## v1.6.0 - 26 March 2024
-- add the `clear_directory` function to make it easy to delete the 
+
+- add the `clear_directory` function to make it easy to delete the
   contents of a directory while leaving the top level directory in place.
 - add the `filepath` dependency to clean up some code.
 
 ## v1.5.1 - 22 March 2024
+
 - use unicode module to convert filenames to utf8 on erlang
 
 ## v1.5.0 - 29 February 2024
+
 - add `file_info` function.
 
 ## v1.4.2 - 18 February 2024
+
 - Update msg in deprecated tag for `is_directory` to correctly point to `verify_is_directory`
 
 ## v1.4.1 - 1 February 2024
+
 - Update gleam_stdlib dependency to "~> 0.34 or ~> 1.0" in preparation for 1.0.
 
 ## v1.3.1 - 27 January 2024
+
 - Update msg in deprecated tag for `is_file` to correctly point to `verify_is_file`.
 
 ## v1.3.0 - 26 January 2024
+
 - Deprecate `is_file` and `is_directory` in favor of `verify_is_file` and `verify_is_directory`,
   which error on lack of permissions rather than silently returning false.
 
 ## v1.2.0 - 11 January 2024
+
 - Add the `current_directory` function
 
 ## v1.1.2 - 9 January 2024
+
 - Update stdlib dep to 0.33
 
 ## v1.1.1 - 5 January 2024
+
 - Fix bug with double "/" from `get_files`
 
 ## v1.1.0 - 25 December 2023
+
 - Added `set_permissions` and `set_permissions_octal` functions for setting
   permissions on a file
 
 ## v1.0.0 - 16 November 2023
+
 - No changes, just going v1 so somver starts working
 
 ## v0.4.0 - 15 November 2023
+
 - Renamed `list_contents` to `read_directory`
 - Added `get_files`
 
 ## v0.3.0 - 12 November 2023
+
 - Switched order of parameters for `write`, `append`, `write_bits`, and `append_bits` functions.
   They now take the filename first.
 
 ## v0.2.1 - 11 November 2023
+
 - Updated ffi js to use `BitArray`
 
 ## v0.2.0 - 6 November 2023
+
 - Updated for Gleam v0.32.
 
 ## v0.1.14 - 18 September 2023
+
 - added `delete_all` which takes a list of paths to call delete on
 
 ## v0.1.13 - 17 September 2023
+
 - copy and rename directory functions perform a create_directory_all
 
 ## v0.1.12 - 17 September 2023
+
 - Add `copy_directory` and `rename_directory` functions
 
 ## v0.1.11 - 11 September 2023
+
 - Add `copy_file` and `rename_file` functions
 
 ## v0.1.10 - 9 August 2023
+
 - Change `creat_dir_all` to `create_directory_all`
 
 ## v0.1.9 - 8 August 2023
+
 - Add functions `create_dir_all`, `create_file`, and `is_file`.
 - Remove function `delete_directory`. The `delete` function now deletes files
   and recursively deletes directories.
 - `make_directory` is now called `create_directory` to be consistent with `create_file`.
 
 ## v0.1.8 - 31 July 2023
+
 - Small refactor to remove js functions `writeFile` and `appendFile`
 
 ## v0.1.7 - 31 July 2023
-- Fix bug where `read` was incorrectly returning non utf8 content for string on 
-    javascript target. Now returns correct `NonUtf8` error
+
+- Fix bug where `read` was incorrectly returning non utf8 content for string on
+  javascript target. Now returns correct `NonUtf8` error
 
 ## v0.1.6 - 19 July 2023
+
 - Add `make_directory` and `delete_directory` functions for working with directories.
 
 ## v0.1.5 - 12 July 2023
+
 - Update to use new ffi syntax
 
 ## v0.1.4 - 21 June 2023
+
 - Add `is_directory` and `list_contents` functions for working with directories.
 
 ## v0.1.3 - 21 June 2023
+
 - Add ffi to remove gleam_erlang dependency
 
 ## v0.1.2 - 21 June 2023
+
 - Remove unnecessary dependency on `gleam_javascript`. Whoops.
 
 ## v0.1.1 - 31 May 2023
+
 - Refactored to match gleam idiom of ffi "do" functions and single signature.
-- Added documentation about the utf8 issue with the read function. 
+- Added documentation about the utf8 issue with the read function.

src/simplifile.gleam

@@ -182,10 +182,10 @@ pub type FileInfo {
   FileInfo(
     /// File size in bytes.
     size: Int,
-    /// File mode that indicates the file type and its permissions. 
-    /// For example, in Unix and Linux, a mode value of 33188 indicates 
-    /// a regular file and the permissions associated with it 
-    /// (read and write for the owner, and read-only for others, in 
+    /// File mode that indicates the file type and its permissions.
+    /// For example, in Unix and Linux, a mode value of 33188 indicates
+    /// a regular file and the permissions associated with it
+    /// (read and write for the owner, and read-only for others, in
     /// this case).
     mode: Int,
     /// Number of hard links that exist for the file.
@@ -265,9 +265,9 @@ pub fn delete(file_or_dir_at path: String) -> Result(Nil, FileError) {
 
 /// Delete all files/directories specified in a list of paths.
 /// Recursively deletes provided directories.
-/// Does not return an error if one or more of the provided paths 
-/// do not exist. 
-/// 
+/// Does not return an error if one or more of the provided paths
+/// do not exist.
+///
 pub fn delete_all(paths paths: List(String)) -> Result(Nil, FileError) {
   case paths {
     [] -> Ok(Nil)
@@ -338,15 +338,15 @@ pub fn append_bits(
 /// ```gleam
 /// let assert True = is_directory("./test")
 /// ```
-/// 
+///
 @deprecated("Use `verify_is_directory` instead")
 pub fn is_directory(filepath: String) -> Bool {
   do_is_directory(filepath)
 }
 
 /// Checks if the provided filepath exists and is a directory.
 /// Returns an error if it lacks permissions to read the directory.
-/// 
+///
 /// ## Example
 /// ```gleam
 /// let assert Ok(True) = verify_is_directory("./test")
@@ -392,27 +392,27 @@ pub fn create_symlink(
 
 /// Lists the contents of a directory.
 /// The list contains directory and file names, and is not recursive.
-/// 
+///
 /// ## Example
 /// ```gleam
 /// let assert Ok(files_and_folders) = read_directory(at: "./Folder1")
 /// ```
-/// 
+///
 pub fn read_directory(at path: String) -> Result(List(String), FileError) {
   do_read_directory(path)
   |> cast_error
 }
 
 /// Returns `True` if there is a file at the given path, false otherwise.
-/// 
+///
 @deprecated("Use `verify_is_file` instead")
 pub fn is_file(filepath: String) -> Bool {
   do_is_file(filepath)
 }
 
 /// Checks if the file at the provided filepath exists and is a file.
 /// Returns an Error if it lacks permissions to read the file.
-/// 
+///
 /// ## Example
 /// ```gleam
 /// let assert Ok(True) = verify_is_file("./test.txt")
@@ -454,7 +454,7 @@ fn do_verify_is_symlink(filepath: String) -> Result(Bool, FileError)
 
 /// Creates an empty file at the given filepath. Returns an `Error(Eexist)`
 /// if the file already exists.
-/// 
+///
 pub fn create_file(at filepath: String) -> Result(Nil, FileError) {
   case
     filepath
@@ -502,8 +502,8 @@ pub fn rename_file(at src: String, to dest: String) -> Result(Nil, FileError) {
 
 /// Copy a directory recursively
 pub fn copy_directory(at src: String, to dest: String) -> Result(Nil, FileError) {
-  // Erlang does not provide a built in `copy_dir` function, 
-  // and Deno doesn't support Node's `fs.cpSync`, so we'll just roll 
+  // Erlang does not provide a built in `copy_dir` function,
+  // and Deno doesn't support Node's `fs.cpSync`, so we'll just roll
   // our own for now.
   use _ <- result.try(create_directory_all(dest))
   do_copy_directory(src, dest)
@@ -561,19 +561,24 @@ pub fn clear_directory(at path: String) -> Result(Nil, FileError) {
 
 /// Returns a list of filepaths for every file in the directory, including nested
 /// files.
-/// 
+///
 pub fn get_files(in directory: String) -> Result(List(String), FileError) {
   use contents <- result.try(read_directory(directory))
-  let paths =
-    contents
-    |> list.map(filepath.join(directory, _))
-  let files = list.filter(paths, fn(path) { verify_is_file(path) == Ok(True) })
-  case list.filter(paths, fn(path) { verify_is_directory(path) == Ok(True) }) {
-    [] -> Ok(files)
-    directories -> {
-      use nested_files <- result.try(list.try_map(directories, get_files))
-      Ok(list.append(files, list.flatten(nested_files)))
-    }
+  use acc, content <- list.try_fold(over: contents, from: [])
+  let path = filepath.join(directory, content)
+
+  case verify_is_file(path) {
+    Error(e) -> Error(e)
+    Ok(True) -> Ok([path, ..acc])
+    Ok(False) ->
+      case verify_is_directory(path) {
+        Error(e) -> Error(e)
+        Ok(False) -> Ok(acc)
+        Ok(True) -> {
+          use nested_files <- result.try(get_files(path))
+          Ok(list.append(acc, nested_files))
+        }
+      }
   }
 }
 
@@ -617,7 +622,7 @@ pub fn file_permissions_to_octal(permissions: FilePermissions) -> Int {
 }
 
 /// Sets the permissions for a given file
-/// 
+///
 /// # Example
 /// ```gleam
 /// let all = set.from_list([Read, Write, Execute])
@@ -632,7 +637,7 @@ pub fn set_permissions(
 }
 
 /// Sets the permissions for a given file using an octal representation
-/// 
+///
 /// # Example
 /// ```gleam
 /// set_permissions_octal("./script.sh", 0o777)
@@ -646,7 +651,7 @@ pub fn set_permissions_octal(
 }
 
 /// Returns the current working directory
-/// 
+///
 pub fn current_directory() -> Result(String, FileError) {
   do_current_directory()
   |> cast_error