sloc.lisp

(ql:quickload '(cl-fad cl-ppcre))

(defpackage :sloc
  (:use :cl)
  (:import-from :cl-fad #:walk-directory))

(in-package :sloc)

(defvar *ext->syntax*
  '(("factor" "^!.*"))
  "A mapping from pathname-types to regexes to ignore.")

(defun count-file-lines (file regex)
  "Count the lines in FILE not matching REGEX."
  (with-open-file (in file)
    (loop for line = (read-line in nil) while line
       unless (or (cl-ppcre:scan regex line)
                  (string= "" line))
       count line)))

(defun count-lines (filetype dir &optional constraints)
  "Count all the valid lines under DIR for the given FILETYPE."
  (let ((regex (second (assoc filetype *ext->syntax* :test #'equal)))
        (count 0))
    (unless regex
      (error "No such filetype in ~A" *ext->syntax*))
    (walk-directory dir (lambda (f) (incf count (count-file-lines f regex)))
                    :test (lambda (f)
                            (and (string= (pathname-type f) filetype)
                                 (not (search "-docs" (pathname-name f))))))
    count))

;; A few interesting stats:
;; Factor Compiler      ->  25,505 lines Factor
;; Factor CPU           ->   8,286 lines Factor
;; Factor Bootstrap     ->     841 lines Factor
;; Factor Tools         ->   4,404 lines Factor
;; Factor Help          ->   2,651 lines Factor
;; Factor Classes       ->   2,470 lines Factor
;; Factor Generics      ->     968 lines Factor
;; Factor Stack-xer     ->   1,913 lines Factor
;; Factor Alien/FFI     ->   1,710 lines Factor
;; Factor PEG Parser    ->   1,775 lines Factor
;; Factor Prettyprinter ->   1,048 lines Factor
;; Factor Vocabs        ->     658 lines Factor
;; Factor Kernel        ->     343 lines Factor
;; Factor Debugger      ->     354 lines Factor
;; Factor Listener      ->     225 lines Factor
;; Factor Inspector     ->     110 lines Factor
;; Factor Macros        ->     118 lines Factor
;; TOTAL                ->  53,379 lines Factor

;; Other stats:
;; Factor VM            -> ~14,500 lines C/C++ (via CLOC)
;; Factor Basis         -> 151,446 lines Factor
;; Factor Core          ->  14,435 lines Factor
;; Factor Extra         ->  73,884 lines Factor
;; Unmaintained Code    ->  16,360 lines Factor

;; TODO: Rough comparisons to CCL, LuaJIT?

;; and some interesting posts by littledan:
;; http://useless-factor.blogspot.com/2010/01/type-feedback-in-factor.html
;; http://useless-factor.blogspot.com/2010/04/couple-language-design-ideas.html
;; http://useless-factor.blogspot.com/2010/04/guarded-method-inlining-for-factor.html

;; and the DLS paper:
;; http://factorcode.org/littledan/dls.pdf