kitty-gridlock/src/Decompression.cake

(comptime-define-symbol 'Unix)
(set-cakelisp-option cakelisp-src-dir "Dependencies/cakelisp/src")
(add-cakelisp-search-directory "test/src" "Dependencies/cakelisp/runtime")
(import &comptime-only "Macros.cake")

(add-c-search-directory-module "../../src") ;; Building from Dependencies/gamelib
(c-import "<stdio.h>" "<stdlib.h>" "<fcntl.h>" "<unistd.h>"
          "bunzip.h")

(add-c-build-dependency "bunzip-4.1.c") ;; LGPLv2

(var g-known-num-puzzles (const int) 2577412)
;; This controls how many should be exported!
(var g-num-puzzles-to-read (const int) 50000)

(var-global g-bunzip-debug bool false)

(def-function-signature-global bunzip-decompress-callback (buffer (* char) num-bytes-this-read int
                                                                  user-data (* void) &return bool))

(defun bunzip-decompress (bz2-file (* (const char))
                          receive-data-callback bunzip-decompress-callback
                          user-data (* void)
                          &return bool)
  (var file-descriptor int (open bz2-file O_RDONLY))
  (when (= -1 file-descriptor)
    (printf "Failed to load file %s\n" bz2-file)
    (return false))

  ;; Prints all of it, which is a lot!
  ;; (unless (= 0 (uncompressStream file-descriptor 1))
  ;; (return false))

  (var header-data (* bunzip_data) null)
  (var error-code int 0)
  (set error-code (start_bunzip (addr header-data) file-descriptor
                                null 0)) ;; Unused: Input buffer
  (when error-code
    (printf "Bunzip error %d\n" error-code)
    (free header-data)
    (close file-descriptor)
    (return false))

  ;; Add room for null terminator. Not necessary for binary data
  (var output ([] 4097 char) (array 0))
  (var num-bytes-read int 0)
  (while true
    (var return-code int (read_bunzip header-data output (- (array-size output) 1)))
    (when (< return-code RETVAL_LAST_BLOCK) ;; On success, return code = num bytes written
      (printf "Bunzip error %d\n" return-code)
      (free header-data)
      (close file-descriptor)
      (return false))

    (var num-bytes-this-read int 0)
    (when (> return-code 0)
      (set num-bytes-this-read return-code)
      ;; Set null terminator
      (set (at num-bytes-this-read output) 0)
      (set num-bytes-read (+ num-bytes-read num-bytes-this-read)))

    (when receive-data-callback
      (unless (receive-data-callback output num-bytes-this-read user-data)
        (free header-data)
        (close file-descriptor)
        (return false)))

    ;; (printf "Output: %d bytes\n%s[end]\n" return-code output)
    (when (= return-code RETVAL_LAST_BLOCK)
      (when g-bunzip-debug (printf "Last Output: %s[end]\n" output))
      (break)))

  (when g-bunzip-debug (printf "Output: %d bytes\n" num-bytes-read))
  (free header-data)
  (close file-descriptor)
  (return true))

(comptime-cond
 ('Kitty-Main)
 (true ;; Generate puzzles.txt
  ;; Note that this doesn't change the distribution of puzzles
  (var g-puzzle-skip-count (const int) (/ g-known-num-puzzles g-num-puzzles-to-read))

  (defstruct-local puzzle-line-buffer
      buffer (* char)
      buffer-size int
      buffer-write (* char)
      num-puzzles-read int
      file-out (* FILE))

  (defun-local process-puzzle-data (buffer (* char) num-bytes-this-read int
                                           user-data (* void) &return bool)
    (var work-in-progress-line (* puzzle-line-buffer) (type-cast user-data (* puzzle-line-buffer)))

    (var current-char (* char) buffer)
    (while (< (- current-char buffer) num-bytes-this-read)
      (cond
        ((= '\n' (deref current-char)) ;; Found puzzle delimiter
         (set (deref (path work-in-progress-line > buffer-write)) 0)

         ;; Output puzzle
         (when (= 0 (mod (path work-in-progress-line > num-puzzles-read) g-puzzle-skip-count))
           (printf "%s\n" (path work-in-progress-line > buffer))
           (when (path work-in-progress-line > file-out)
             (fprintf (path work-in-progress-line > file-out) "%s\n"
                      (path work-in-progress-line > buffer))))

         (set (path work-in-progress-line > buffer-write)
              (path work-in-progress-line > buffer))
         (incr (path work-in-progress-line > num-puzzles-read)))

        (true ;; Reading a puzzle in
         (set (deref (path work-in-progress-line > buffer-write)) (deref current-char))
         (incr (path work-in-progress-line > buffer-write))
         (when (>= (- (path work-in-progress-line > buffer-write)
                      (path work-in-progress-line > buffer))
                   (- (path work-in-progress-line > buffer-size) 1))
           (printf "Error: didn't encounter newline before end of buffer\n")
           (return false))))
      (incr current-char))
    (return true))

  (defun main (&return int)
    (var bz2-file (* (const char)) "assets/rush.txt.bz2")

    (var file-out (* FILE) (fopen "data/puzzles.txt" "w"))
    (unless file-out
      (printf "Could not open output file\n")
      (return 1))
    (fprintf file-out "%d\n" g-num-puzzles-to-read)

    ;; For when the read buffer straddles a single puzzle
    ;; 46 chars max per puzzle, plus newline, plus room for null terminator for easy printing
    (var work-in-progress-buffer ([] 48 char) (array 0))
    (var work-in-progress-line puzzle-line-buffer
         (array work-in-progress-buffer (array-size work-in-progress-buffer)
                work-in-progress-buffer 0 file-out))
    (unless (bunzip-decompress bz2-file process-puzzle-data (addr work-in-progress-line))
      (fclose file-out)
      (return 1))
    (fclose file-out)
    (return 0))

  (set-cakelisp-option executable-output "../../decompression-test")))