tests/external-format.impure.lisp

   1 ;;;; This file is for testing external-format functionality, using
   2 ;;;; test machinery which might have side effects (e.g.  executing
   3 ;;;; DEFUN, writing files).  Note that the tests here reach into
   4 ;;;; unexported functionality, and should not be used as a guide for
   5 ;;;; users.
   6
   7 ;;;; This software is part of the SBCL system. See the README file for
   8 ;;;; more information.
   9 ;;;;
  10 ;;;; While most of SBCL is derived from the CMU CL system, the test
  11 ;;;; files (like this one) were written from scratch after the fork
  12 ;;;; from CMU CL.
  13 ;;;;
  14 ;;;; This software is in the public domain and is provided with
  15 ;;;; absolutely no warranty. See the COPYING and CREDITS files for
  16 ;;;; more information.
  17
  18 (defmacro do-external-formats ((xf &optional result) &body body)
  19   (let ((nxf (gensym)))
  20     `(dolist (,nxf sb-impl::*external-formats* ,result)
  21        (let ((,xf (first (first ,nxf))))
  22          ,@body))))
  23
  24 (defvar *test-path* "external-format-test.tmp")
  25
  26 (do-external-formats (xf)
  27   (with-open-file (s #-win32 "/dev/null" #+win32 "nul" :direction :input :external-format xf)
  28     (assert (eq (read-char s nil s) s))))
  29
  30 ;;; Test standard character read-write equivalency over all external formats.
  31 (let ((standard-characters "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!$\"'(),_-./:;?+<=>#%&*@[\\]{|}`^~"))
  32   (do-external-formats (xf)
  33     (with-open-file (s *test-path* :direction :output
  34                      :if-exists :supersede :external-format xf)
  35       (loop for character across standard-characters
  36             do (write-char character s)))
  37     (with-open-file (s *test-path* :direction :input
  38                      :external-format xf)
  39       (loop for character across standard-characters
  40             do (let ((got (read-char s)))
  41                  (unless (eql character got)
  42                    (error "wanted ~S, got ~S" character got)))))))
  43
  44 (delete-file *test-path*)
  45 #-sb-unicode
  46 (progn
  47   (test-util:report-test-status)
  48   (sb-ext:quit :unix-status 104))
  49
  50 ;;; Test UTF-8 writing and reading of 1, 2, 3 and 4 octet characters with
  51 ;;; all possible offsets. Tests for buffer edge bugs. fd-stream buffers are
  52 ;;; 4096 wide.
  53 (dotimes (width-1 4)
  54   (let ((character (code-char (elt '(1 #x81 #x801 #x10001) width-1))))
  55     (dotimes (offset (+ width-1 1))
  56       (with-open-file (s *test-path* :direction :output
  57                        :if-exists :supersede :external-format :utf-8)
  58         (dotimes (n offset)
  59           (write-char #\a s))
  60         (dotimes (n (+ 4 sb-impl::+bytes-per-buffer+))
  61           (write-char character s)))
  62       (with-open-file (s *test-path* :direction :input
  63                        :external-format :utf-8)
  64         (dotimes (n offset)
  65           (assert (eql (read-char s) #\a)))
  66         (dotimes (n (+ 4 sb-impl::+bytes-per-buffer+))
  67           (let ((got (read-char s)))
  68             (unless (eql got character)
  69               (error "wanted ~S, got ~S (~S)" character got n))))
  70         (assert (eql (read-char s nil s) s))))))
  71
  72 ;;; Test character decode restarts.
  73 (with-open-file (s *test-path* :direction :output
  74                  :if-exists :supersede :element-type '(unsigned-byte 8))
  75   (write-byte 65 s)
  76   (write-byte 66 s)
  77   (write-byte #xe0 s)
  78   (write-byte 67 s))
  79 (with-open-file (s *test-path* :direction :input
  80                  :external-format :utf-8)
  81   (handler-bind
  82       ((sb-int:character-decoding-error #'(lambda (decoding-error)
  83                                             (declare (ignore decoding-error))
  84                                             (invoke-restart
  85                                              'sb-int:attempt-resync))))
  86     (assert (equal (read-line s nil s) "ABC"))
  87     (assert (equal (read-line s nil s) s))))
  88 (with-open-file (s *test-path* :direction :input
  89                  :external-format :utf-8)
  90   (handler-bind
  91       ((sb-int:character-decoding-error #'(lambda (decoding-error)
  92                                             (declare (ignore decoding-error))
  93                                             (invoke-restart
  94                                              'sb-int:force-end-of-file))))
  95     (assert (equal (read-line s nil s) "AB"))
  96     (assert (equal (read-line s nil s) s))))
  97
  98 ;;; And again with more data to account for buffering (this was briefly)
  99 ;;; broken in early 0.9.6.
 100 (with-open-file (s *test-path* :direction :output
 101                  :if-exists :supersede :element-type '(unsigned-byte 8))
 102   (let ((a (make-array 50
 103                        :element-type '(unsigned-byte 64)
 104                        :initial-contents (map 'list #'char-code
 105                                               "1234567890123456789012345678901234567890123456789."))))
 106     (setf (aref a 49) (char-code #\Newline))
 107     (dotimes (i 40)
 108       (write-sequence a s))
 109     (write-byte #xe0 s)
 110     (dotimes (i 40)
 111       (write-sequence a s))))
 112 (with-test (:name (:character-decode-large :attempt-resync))
 113   (with-open-file (s *test-path* :direction :input
 114                      :external-format :utf-8)
 115     (handler-bind
 116         ((sb-int:character-decoding-error #'(lambda (decoding-error)
 117                                               (declare (ignore decoding-error))
 118                                               (invoke-restart
 119                                                'sb-int:attempt-resync)))
 120          ;; The failure mode is an infinite loop, add a timeout to detetct it.
 121          (sb-ext:timeout (lambda () (error "Timeout"))))
 122       (sb-ext:with-timeout 5
 123         (dotimes (i 80)
 124           (assert (equal (read-line s nil s)
 125                          "1234567890123456789012345678901234567890123456789")))))))
 126
 127 (with-test (:name (:character-decode-large :force-end-of-file)
 128             :fails-on :sbcl)
 129   (error "We can't reliably test this due to WITH-TIMEOUT race condition")
 130   ;; This test will currently fail. But sometimes it will fail in
 131   ;; ungracefully due to the WITH-TIMEOUT race mentioned above. This
 132   ;; rightfully confuses some people, so we'll skip running the code
 133   ;; for now. -- JES, 2006-01-27
 134   #+nil
 135   (with-open-file (s *test-path* :direction :input
 136                      :external-format :utf-8)
 137     (handler-bind
 138         ((sb-int:character-decoding-error #'(lambda (decoding-error)
 139                                               (declare (ignore decoding-error))
 140                                               (invoke-restart
 141                                                'sb-int:force-end-of-file)))
 142          ;; The failure mode is an infinite loop, add a timeout to detetct it.
 143          (sb-ext:timeout (lambda () (error "Timeout"))))
 144       (sb-ext:with-timeout 5
 145         (dotimes (i 80)
 146           (assert (equal (read-line s nil s)
 147                          "1234567890123456789012345678901234567890123456789")))
 148         (assert (equal (read-line s nil s) s))))))
 149
 150 ;;; Test character encode restarts.
 151 (with-open-file (s *test-path* :direction :output
 152                  :if-exists :supersede :external-format :latin-1)
 153   (handler-bind
 154       ((sb-int:character-encoding-error #'(lambda (encoding-error)
 155                                             (declare (ignore encoding-error))
 156                                             (invoke-restart
 157                                              'sb-impl::output-nothing))))
 158     (write-char #\A s)
 159     (write-char #\B s)
 160     (write-char (code-char 322) s)
 161     (write-char #\C s)))
 162 (with-open-file (s *test-path* :direction :input
 163                  :external-format :latin-1)
 164   (assert (equal (read-line s nil s) "ABC"))
 165   (assert (equal (read-line s nil s) s)))
 166
 167 (with-open-file (s *test-path* :direction :output
 168                  :if-exists :supersede :external-format :latin-1)
 169   (handler-bind
 170       ((sb-int:character-encoding-error #'(lambda (encoding-error)
 171                                             (declare (ignore encoding-error))
 172                                             (invoke-restart
 173                                              'sb-impl::output-nothing))))
 174     (let ((string (make-array 4 :element-type 'character
 175                               :initial-contents `(#\A #\B ,(code-char 322)
 176                                                       #\C))))
 177       (write-string string s))))
 178 (with-open-file (s *test-path* :direction :input
 179                  :external-format :latin-1)
 180   (assert (equal (read-line s nil s) "ABC"))
 181   (assert (equal (read-line s nil s) s)))
 182
 183 ;;; Test skipping character-decode-errors in comments.
 184 (let ((s (open "external-format-test.lisp" :direction :output
 185                :if-exists :supersede :external-format :latin-1)))
 186   (unwind-protect
 187        (progn
 188          (write-string ";;; ABCD" s)
 189          (write-char (code-char 233) s)
 190          (terpri s)
 191          (close s)
 192          (compile-file "external-format-test.lisp" :external-format :utf-8))
 193     (delete-file s)
 194     (let ((p (probe-file (compile-file-pathname "external-format-test.lisp"))))
 195       (when p
 196         (delete-file p)))))
 197
 198 \f
 199 ;;;; KOI8-R external format
 200 (with-open-file (s *test-path* :direction :output
 201                  :if-exists :supersede :external-format :koi8-r)
 202   (write-char (code-char #xB0) s)
 203   (assert (eq
 204            (handler-case
 205                (progn
 206                  (write-char (code-char #xBAAD) s)
 207                  :bad)
 208              (sb-int:character-encoding-error ()
 209                :good))
 210            :good)))
 211 (with-open-file (s *test-path* :direction :input
 212                  :element-type '(unsigned-byte 8))
 213   (let ((byte (read-byte s)))
 214     (assert (= (eval byte) #x9C))))
 215 (with-open-file (s *test-path* :direction :input
 216                  :external-format :koi8-r)
 217   (let ((char (read-char s)))
 218     (assert (= (char-code (eval char)) #xB0))))
 219 (delete-file *test-path*)
 220
 221 (let* ((koi8-r-codes (coerce '(240 210 201 215 197 212 33) '(vector (unsigned-byte 8))))
 222        (uni-codes #(1055 1088 1080 1074 1077 1090 33))
 223
 224        (string (octets-to-string koi8-r-codes :external-format :koi8-r))
 225        (uni-decoded (map 'vector #'char-code string)))
 226   (assert (equalp (map 'vector #'char-code (octets-to-string koi8-r-codes :external-format :koi8-r))
 227                   uni-codes))
 228   (assert (equalp (string-to-octets (map 'string #'code-char uni-codes) :external-format :koi8-r)
 229                   koi8-r-codes)))
 230 \f
 231 ;;; tests of FILE-STRING-LENGTH
 232 (let ((standard-characters "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!$\"'(),_-./:;?+<=>#%&*@[\\]{|}`^~"))
 233   (do-external-formats (xf)
 234     (with-open-file (s *test-path* :direction :output
 235                        :external-format xf)
 236       (loop for x across standard-characters
 237             for position = (file-position s)
 238             for char-length = (file-string-length s x)
 239             do (write-char x s)
 240             do (assert (= (file-position s) (+ position char-length))))
 241       (let ((position (file-position s))
 242             (string-length (file-string-length s standard-characters)))
 243         (write-string standard-characters s)
 244         (assert (= (file-position s) (+ position string-length)))))
 245     (delete-file *test-path*)))
 246
 247 (let ((char-codes '(0 1 255 256 511 512 1023 1024 2047 2048 4095 4096
 248                     8191 8192 16383 16384 32767 32768 65535 65536 131071
 249                     131072 262143 262144)))
 250   (with-open-file (s *test-path* :direction :output
 251                      :external-format :utf-8)
 252     (dolist (code char-codes)
 253       (let* ((char (code-char code))
 254              (position (file-position s))
 255              (char-length (file-string-length s char)))
 256         (write-char char s)
 257         (assert (= (file-position s) (+ position char-length)))))
 258     (let* ((string (map 'string #'code-char char-codes))
 259            (position (file-position s))
 260            (string-length (file-string-length s string)))
 261       (write-string string s)
 262       (assert (= (file-position s) (+ position string-length))))))
 263 \f
 264
 265 ;;; See sbcl-devel "Subject: Bug in FILE-POSITION on UTF-8-encoded files"
 266 ;;; by Lutz Euler on 2006-03-05 for more details.
 267 (with-test (:name (:file-position :utf-8))
 268   (let ((path *test-path*))
 269     (with-open-file (s path
 270                        :direction :output
 271                        :if-exists :supersede
 272                        :element-type '(unsigned-byte 8))
 273       ;; Write #\*, encoded in UTF-8, to the file.
 274       (write-byte 42 s)
 275       ;; Append #\adiaeresis, encoded in UTF-8, to the file.
 276       (write-sequence '(195 164) s))
 277     (with-open-file (s path :external-format :utf-8)
 278       (read-char s)
 279       (let ((pos (file-position s))
 280             (char (read-char s)))
 281         (format t "read character with code ~a successfully from file position ~a~%"
 282                 (char-code char) pos)
 283         (file-position s pos)
 284         (format t "set file position back to ~a, trying to read-char again~%" pos)
 285         (let ((new-char (read-char s)))
 286           (assert (char= char new-char)))))
 287     (values)))
 288 (delete-file *test-path*)
 289
 290 ;;; We used to call STREAM-EXTERNAL-FORMAT on the stream in the error
 291 ;;; when printing a coding error, but that didn't work if the stream
 292 ;;; was closed by the time the error was printed.  See sbcl-devel
 293 ;;; "Subject: Printing coding errors for closed streams" by Zach Beane
 294 ;;; on 2008-10-16 for more info.
 295 (with-test (:name (:character-coding-error-stream-external-format))
 296   (flet ((first-file-character ()
 297            (with-open-file (stream *test-path* :external-format :utf-8)
 298              (read-char stream))))
 299     (with-open-file (stream *test-path*
 300                             :direction :output
 301                             :if-exists :supersede
 302                             :element-type '(unsigned-byte 8))
 303       (write-byte 192 stream))
 304     (princ-to-string (nth-value 1 (ignore-errors (first-file-character))))))
 305 (delete-file *test-path*)
 306
 307 ;;; External format support in SB-ALIEN
 308
 309 (with-test (:name (:sb-alien :vanilla))
 310   (define-alien-routine strdup c-string (str c-string))
 311   (assert (equal "foo" (strdup "foo"))))
 312
 313 (with-test (:name (:sb-alien :utf-8 :utf-8))
 314   (define-alien-routine strdup (c-string :external-format :utf-8)
 315     (str (c-string :external-format :utf-8)))
 316   (assert (equal "foo" (strdup "foo"))))
 317
 318 (with-test (:name (:sb-alien :latin-1 :utf-8))
 319   (define-alien-routine strdup (c-string :external-format :latin-1)
 320     (str (c-string :external-format :utf-8)))
 321   (assert (= (length (strdup (string (code-char 246))))
 322              2)))
 323
 324 (with-test (:name (:sb-alien :utf-8 :latin-1))
 325   (define-alien-routine strdup (c-string :external-format :utf-8)
 326     (str (c-string :external-format :latin-1)))
 327   (assert (equal (string (code-char 228))
 328                  (strdup (concatenate 'string
 329                                       (list (code-char 195))
 330                                       (list (code-char 164)))))))
 331
 332 (with-test (:name (:sb-alien :ebcdic :ebcdic))
 333   (define-alien-routine strdup (c-string :external-format :ebcdic-us)
 334     (str (c-string :external-format :ebcdic-us)))
 335   (assert (equal "foo" (strdup "foo"))))
 336
 337 (with-test (:name (:sb-alien :latin-1 :ebcdic))
 338   (define-alien-routine strdup (c-string :external-format :latin-1)
 339     (str (c-string :external-format :ebcdic-us)))
 340   (assert (not (equal "foo" (strdup "foo")))))
 341
 342 (with-test (:name (:sb-alien :simple-base-string))
 343   (define-alien-routine strdup (c-string :external-format :ebcdic-us
 344                                          :element-type base-char)
 345     (str (c-string :external-format :ebcdic-us)))
 346   (assert (typep (strdup "foo") 'simple-base-string)))
 347
 348 ;;;; success