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 (do-external-formats (xf)
  25   (with-open-file (s #-win32 "/dev/null" #+win32 "nul" :direction :input :external-format xf)
  26     (assert (eq (read-char s nil s) s))))
  27
  28 ;;; Test standard character read-write equivalency over all external formats.
  29 (let ((standard-characters "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!$\"'(),_-./:;?+<=>#%&*@[\\]{|}`^~"))
  30   (do-external-formats (xf)
  31     (with-open-file (s "external-format-test.txt" :direction :output
  32                      :if-exists :supersede :external-format xf)
  33       (loop for character across standard-characters
  34             do (write-char character s)))
  35     (with-open-file (s "external-format-test.txt" :direction :input
  36                      :external-format xf)
  37       (loop for character across standard-characters
  38             do (assert (eql (read-char s) character))))))
  39
  40 (delete-file "external-format-test.txt")
  41 #-sb-unicode
  42 (progn
  43   (test-util:report-test-status)
  44   (sb-ext:quit :unix-status 104))
  45
  46 ;;; Test UTF-8 writing and reading of 1, 2, 3 and 4 octet characters with
  47 ;;; all possible offsets. Tests for buffer edge bugs. fd-stream buffers are
  48 ;;; 4096 wide.
  49 (dotimes (width-1 4)
  50   (let ((character (code-char (elt '(1 #x81 #x801 #x10001) width-1))))
  51     (dotimes (offset (+ width-1 1))
  52       (with-open-file (s "external-format-test.txt" :direction :output
  53                        :if-exists :supersede :external-format :utf-8)
  54         (dotimes (n offset)
  55           (write-char #\a s))
  56         (dotimes (n 4097)
  57           (write-char character s)))
  58       (with-open-file (s "external-format-test.txt" :direction :input
  59                        :external-format :utf-8)
  60         (dotimes (n offset)
  61           (assert (eql (read-char s) #\a)))
  62         (dotimes (n 4097)
  63           (assert (eql (read-char s) character)))
  64         (assert (eql (read-char s nil s) s))))))
  65
  66 ;;; Test character decode restarts.
  67 (with-open-file (s "external-format-test.txt" :direction :output
  68                  :if-exists :supersede :element-type '(unsigned-byte 8))
  69   (write-byte 65 s)
  70   (write-byte 66 s)
  71   (write-byte #xe0 s)
  72   (write-byte 67 s))
  73 (with-open-file (s "external-format-test.txt" :direction :input
  74                  :external-format :utf-8)
  75   (handler-bind
  76       ((sb-int:character-decoding-error #'(lambda (decoding-error)
  77                                             (declare (ignore decoding-error))
  78                                             (invoke-restart
  79                                              'sb-int:attempt-resync))))
  80     (assert (equal (read-line s nil s) "ABC"))
  81     (assert (equal (read-line s nil s) s))))
  82 (with-open-file (s "external-format-test.txt" :direction :input
  83                  :external-format :utf-8)
  84   (handler-bind
  85       ((sb-int:character-decoding-error #'(lambda (decoding-error)
  86                                             (declare (ignore decoding-error))
  87                                             (invoke-restart
  88                                              'sb-int:force-end-of-file))))
  89     (assert (equal (read-line s nil s) "AB"))
  90     (assert (equal (read-line s nil s) s))))
  91
  92 ;;; And again with more data to account for buffering (this was briefly)
  93 ;;; broken in early 0.9.6.
  94 (with-open-file (s "external-format-test.txt" :direction :output
  95                  :if-exists :supersede :element-type '(unsigned-byte 8))
  96   (let ((a (make-array 50
  97                        :element-type '(unsigned-byte 64)
  98                        :initial-contents (map 'list #'char-code
  99                                               "1234567890123456789012345678901234567890123456789."))))
 100     (setf (aref a 49) (char-code #\Newline))
 101     (dotimes (i 40)
 102       (write-sequence a s))
 103     (write-byte #xe0 s)
 104     (dotimes (i 40)
 105       (write-sequence a s))))
 106 (with-test (:name (:character-decode-large :attempt-resync))
 107   (with-open-file (s "external-format-test.txt" :direction :input
 108                      :external-format :utf-8)
 109     (handler-bind
 110         ((sb-int:character-decoding-error #'(lambda (decoding-error)
 111                                               (declare (ignore decoding-error))
 112                                               (invoke-restart
 113                                                'sb-int:attempt-resync)))
 114          ;; The failure mode is an infinite loop, add a timeout to detetct it.
 115          (sb-ext:timeout (lambda () (error "Timeout"))))
 116       (sb-ext:with-timeout 5
 117         (dotimes (i 80)
 118           (assert (equal (read-line s nil s)
 119                          "1234567890123456789012345678901234567890123456789")))))))
 120
 121 (with-test (:name (:character-decode-large :force-end-of-file)
 122             :fails-on :sbcl)
 123   (error "We can't reliably test this due to WITH-TIMEOUT race condition")
 124   ;; This test will currently fail. But sometimes it will fail in
 125   ;; ungracefully due to the WITH-TIMEOUT race mentioned above. This
 126   ;; rightfully confuses some people, so we'll skip running the code
 127   ;; for now. -- JES, 2006-01-27
 128   #+nil
 129   (with-open-file (s "external-format-test.txt" :direction :input
 130                      :external-format :utf-8)
 131     (handler-bind
 132         ((sb-int:character-decoding-error #'(lambda (decoding-error)
 133                                               (declare (ignore decoding-error))
 134                                               (invoke-restart
 135                                                'sb-int:force-end-of-file)))
 136          ;; The failure mode is an infinite loop, add a timeout to detetct it.
 137          (sb-ext:timeout (lambda () (error "Timeout"))))
 138       (sb-ext:with-timeout 5
 139         (dotimes (i 80)
 140           (assert (equal (read-line s nil s)
 141                          "1234567890123456789012345678901234567890123456789")))
 142         (assert (equal (read-line s nil s) s))))))
 143
 144 ;;; Test character encode restarts.
 145 (with-open-file (s "external-format-test.txt" :direction :output
 146                  :if-exists :supersede :external-format :latin-1)
 147   (handler-bind
 148       ((sb-int:character-encoding-error #'(lambda (encoding-error)
 149                                             (declare (ignore encoding-error))
 150                                             (invoke-restart
 151                                              'sb-impl::output-nothing))))
 152     (write-char #\A s)
 153     (write-char #\B s)
 154     (write-char (code-char 322) s)
 155     (write-char #\C s)))
 156 (with-open-file (s "external-format-test.txt" :direction :input
 157                  :external-format :latin-1)
 158   (assert (equal (read-line s nil s) "ABC"))
 159   (assert (equal (read-line s nil s) s)))
 160
 161 (with-open-file (s "external-format-test.txt" :direction :output
 162                  :if-exists :supersede :external-format :latin-1)
 163   (handler-bind
 164       ((sb-int:character-encoding-error #'(lambda (encoding-error)
 165                                             (declare (ignore encoding-error))
 166                                             (invoke-restart
 167                                              'sb-impl::output-nothing))))
 168     (let ((string (make-array 4 :element-type 'character
 169                               :initial-contents `(#\A #\B ,(code-char 322)
 170                                                       #\C))))
 171       (write-string string s))))
 172 (with-open-file (s "external-format-test.txt" :direction :input
 173                  :external-format :latin-1)
 174   (assert (equal (read-line s nil s) "ABC"))
 175   (assert (equal (read-line s nil s) s)))
 176
 177 ;;; Test skipping character-decode-errors in comments.
 178 (let ((s (open "external-format-test.lisp" :direction :output
 179                :if-exists :supersede :external-format :latin-1)))
 180   (unwind-protect
 181        (progn
 182          (write-string ";;; ABCD" s)
 183          (write-char (code-char 233) s)
 184          (terpri s)
 185          (close s)
 186          (compile-file "external-format-test.lisp" :external-format :utf-8))
 187     (delete-file s)
 188     (let ((p (probe-file (compile-file-pathname "external-format-test.lisp"))))
 189       (when p
 190         (delete-file p)))))
 191
 192 \f
 193 ;;;; KOI8-R external format
 194 (with-open-file (s "external-format-test.txt" :direction :output
 195                  :if-exists :supersede :external-format :koi8-r)
 196   (write-char (code-char #xB0) s)
 197   (assert (eq
 198            (handler-case
 199                (progn
 200                  (write-char (code-char #xBAAD) s)
 201                  :bad)
 202              (sb-int:character-encoding-error ()
 203                :good))
 204            :good)))
 205 (with-open-file (s "external-format-test.txt" :direction :input
 206                  :element-type '(unsigned-byte 8))
 207   (let ((byte (read-byte s)))
 208     (assert (= (eval byte) #x9C))))
 209 (with-open-file (s "external-format-test.txt" :direction :input
 210                  :external-format :koi8-r)
 211   (let ((char (read-char s)))
 212     (assert (= (char-code (eval char)) #xB0))))
 213 (delete-file "external-format-test.txt")
 214
 215 (let* ((koi8-r-codes (coerce '(240 210 201 215 197 212 33) '(vector (unsigned-byte 8))))
 216        (uni-codes #(1055 1088 1080 1074 1077 1090 33))
 217
 218        (string (octets-to-string koi8-r-codes :external-format :koi8-r))
 219        (uni-decoded (map 'vector #'char-code string)))
 220   (assert (equalp (map 'vector #'char-code (octets-to-string koi8-r-codes :external-format :koi8-r))
 221                   uni-codes))
 222   (assert (equalp (string-to-octets (map 'string #'code-char uni-codes) :external-format :koi8-r)
 223                   koi8-r-codes)))
 224 \f
 225 ;;; tests of FILE-STRING-LENGTH
 226 (let ((standard-characters "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!$\"'(),_-./:;?+<=>#%&*@[\\]{|}`^~"))
 227   (do-external-formats (xf)
 228     (with-open-file (s "external-format-test.txt" :direction :output
 229                        :external-format xf)
 230       (loop for x across standard-characters
 231             for position = (file-position s)
 232             for char-length = (file-string-length s x)
 233             do (write-char x s)
 234             do (assert (= (file-position s) (+ position char-length))))
 235       (let ((position (file-position s))
 236             (string-length (file-string-length s standard-characters)))
 237         (write-string standard-characters s)
 238         (assert (= (file-position s) (+ position string-length)))))
 239     (delete-file "external-format-test.txt")))
 240
 241 (let ((char-codes '(0 1 255 256 511 512 1023 1024 2047 2048 4095 4096
 242                     8191 8192 16383 16384 32767 32768 65535 65536 131071
 243                     131072 262143 262144)))
 244   (with-open-file (s "external-format-test.txt" :direction :output
 245                      :external-format :utf-8)
 246     (dolist (code char-codes)
 247       (let* ((char (code-char code))
 248              (position (file-position s))
 249              (char-length (file-string-length s char)))
 250         (write-char char s)
 251         (assert (= (file-position s) (+ position char-length)))))
 252     (let* ((string (map 'string #'code-char char-codes))
 253            (position (file-position s))
 254            (string-length (file-string-length s string)))
 255       (write-string string s)
 256       (assert (= (file-position s) (+ position string-length))))))
 257 \f
 258
 259 ;;; See sbcl-devel "Subject: Bug in FILE-POSITION on UTF-8-encoded files"
 260 ;;; by Lutz Euler on 2006-03-05 for more details.
 261 (with-test (:name (:file-position :utf-8))
 262   (let ((path "external-format-test.txt"))
 263     (with-open-file (s path
 264                        :direction :output
 265                        :if-exists :supersede
 266                        :element-type '(unsigned-byte 8))
 267       ;; Write #\*, encoded in UTF-8, to the file.
 268       (write-byte 42 s)
 269       ;; Append #\adiaeresis, encoded in UTF-8, to the file.
 270       (write-sequence '(195 164) s))
 271     (with-open-file (s path :external-format :utf-8)
 272       (read-char s)
 273       (let ((pos (file-position s))
 274             (char (read-char s)))
 275         (format t "read character with code ~a successfully from file position ~a~%"
 276                 (char-code char) pos)
 277         (file-position s pos)
 278         (format t "set file position back to ~a, trying to read-char again~%" pos)
 279         (let ((new-char (read-char s)))
 280           (assert (char= char new-char)))))
 281     (values)))
 282
 283 ;;; External format support in SB-ALIEN
 284
 285 (with-test (:name (:sb-alien :vanilla))
 286   (define-alien-routine strdup c-string (str c-string))
 287   (assert (equal "foo" (strdup "foo"))))
 288
 289 (with-test (:name (:sb-alien :utf-8 :utf-8))
 290   (define-alien-routine strdup (c-string :external-format :utf-8)
 291     (str (c-string :external-format :utf-8)))
 292   (assert (equal "foo" (strdup "foo"))))
 293
 294 (with-test (:name (:sb-alien :latin-1 :utf-8))
 295   (define-alien-routine strdup (c-string :external-format :latin-1)
 296     (str (c-string :external-format :utf-8)))
 297   (assert (= (length (strdup (string (code-char 246))))
 298              2)))
 299
 300 (with-test (:name (:sb-alien :utf-8 :latin-1))
 301   (define-alien-routine strdup (c-string :external-format :utf-8)
 302     (str (c-string :external-format :latin-1)))
 303   (assert (equal (string (code-char 228))
 304                  (strdup (concatenate 'string
 305                                       (list (code-char 195))
 306                                       (list (code-char 164)))))))
 307
 308 (with-test (:name (:sb-alien :ebcdic :ebcdic))
 309   (define-alien-routine strdup (c-string :external-format :ebcdic-us)
 310     (str (c-string :external-format :ebcdic-us)))
 311   (assert (equal "foo" (strdup "foo"))))
 312
 313 (with-test (:name (:sb-alien :latin-1 :ebcdic))
 314   (define-alien-routine strdup (c-string :external-format :latin-1)
 315     (str (c-string :external-format :ebcdic-us)))
 316   (assert (not (equal "foo" (strdup "foo")))))
 317
 318 (with-test (:name (:sb-alien :simple-base-string))
 319   (define-alien-routine strdup (c-string :external-format :ebcdic-us
 320                                          :element-type base-char)
 321     (str (c-string :external-format :ebcdic-us)))
 322   (assert (typep (strdup "foo") 'simple-base-string)))
 323
 324 ;;;; success