;;;; absolutely no warranty. See the COPYING and CREDITS files for
;;;; more information.
-#-sb-unicode
-(sb-ext:quit :unix-status 104)
-
(defmacro do-external-formats ((xf &optional result) &body body)
(let ((nxf (gensym)))
`(dolist (,nxf sb-impl::*external-formats* ,result)
(loop for character across standard-characters
do (assert (eql (read-char s) character))))))
+(delete-file "external-format-test.txt")
+#-sb-unicode
+(progn
+ (test-util:report-test-status)
+ (sb-ext:quit :unix-status 104))
+
;;; Test UTF-8 writing and reading of 1, 2, 3 and 4 octet characters with
;;; all possible offsets. Tests for buffer edge bugs. fd-stream buffers are
;;; 4096 wide.
(assert (equal (read-line s nil s) "AB"))
(assert (equal (read-line s nil s) s))))
+;;; And again with more data to account for buffering (this was briefly)
+;;; broken in early 0.9.6.
+(with-open-file (s "external-format-test.txt" :direction :output
+ :if-exists :supersede :element-type '(unsigned-byte 8))
+ (let ((a (make-array 50
+ :element-type '(unsigned-byte 64)
+ :initial-contents (map 'list #'char-code
+ "1234567890123456789012345678901234567890123456789."))))
+ (setf (aref a 49) (char-code #\Newline))
+ (dotimes (i 40)
+ (write-sequence a s))
+ (write-byte #xe0 s)
+ (dotimes (i 40)
+ (write-sequence a s))))
+(with-test (:name (:character-decode-large :attempt-resync))
+ (with-open-file (s "external-format-test.txt" :direction :input
+ :external-format :utf-8)
+ (handler-bind
+ ((sb-int:character-decoding-error #'(lambda (decoding-error)
+ (declare (ignore decoding-error))
+ (invoke-restart
+ 'sb-int:attempt-resync)))
+ ;; The failure mode is an infinite loop, add a timeout to detetct it.
+ (sb-ext:timeout (lambda () (error "Timeout"))))
+ (sb-ext:with-timeout 5
+ (dotimes (i 80)
+ (assert (equal (read-line s nil s)
+ "1234567890123456789012345678901234567890123456789")))))))
+
+(with-test (:name (:character-decode-large :force-end-of-file)
+ :fails-on :sbcl)
+ (error "We can't reliably test this due to WITH-TIMEOUT race condition")
+ ;; This test will currently fail. But sometimes it will fail in
+ ;; ungracefully due to the WITH-TIMEOUT race mentioned above. This
+ ;; rightfully confuses some people, so we'll skip running the code
+ ;; for now. -- JES, 2006-01-27
+ #+nil
+ (with-open-file (s "external-format-test.txt" :direction :input
+ :external-format :utf-8)
+ (handler-bind
+ ((sb-int:character-decoding-error #'(lambda (decoding-error)
+ (declare (ignore decoding-error))
+ (invoke-restart
+ 'sb-int:force-end-of-file)))
+ ;; The failure mode is an infinite loop, add a timeout to detetct it.
+ (sb-ext:timeout (lambda () (error "Timeout"))))
+ (sb-ext:with-timeout 5
+ (dotimes (i 80)
+ (assert (equal (read-line s nil s)
+ "1234567890123456789012345678901234567890123456789")))
+ (assert (equal (read-line s nil s) s))))))
+
;;; Test character encode restarts.
(with-open-file (s "external-format-test.txt" :direction :output
:if-exists :supersede :external-format :latin-1)
(when p
(delete-file p)))))
+\f
+;;;; KOI8-R external format
+(with-open-file (s "external-format-test.txt" :direction :output
+ :if-exists :supersede :external-format :koi8-r)
+ (write-char (code-char #xB0) s)
+ (assert (eq
+ (handler-case
+ (progn
+ (write-char (code-char #xBAAD) s)
+ :bad)
+ (sb-int:character-encoding-error ()
+ :good))
+ :good)))
+(with-open-file (s "external-format-test.txt" :direction :input
+ :element-type '(unsigned-byte 8))
+ (let ((byte (read-byte s)))
+ (assert (= (eval byte) #x9C))))
+(with-open-file (s "external-format-test.txt" :direction :input
+ :external-format :koi8-r)
+ (let ((char (read-char s)))
+ (assert (= (char-code (eval char)) #xB0))))
(delete-file "external-format-test.txt")
-(sb-ext:quit :unix-status 104)
+(let* ((koi8-r-codes (coerce '(240 210 201 215 197 212 33) '(vector (unsigned-byte 8))))
+ (uni-codes #(1055 1088 1080 1074 1077 1090 33))
+
+ (string (octets-to-string koi8-r-codes :external-format :koi8-r))
+ (uni-decoded (map 'vector #'char-code string)))
+ (assert (equalp (map 'vector #'char-code (octets-to-string koi8-r-codes :external-format :koi8-r))
+ uni-codes))
+ (assert (equalp (string-to-octets (map 'string #'code-char uni-codes) :external-format :koi8-r)
+ koi8-r-codes)))
+\f
+;;; tests of FILE-STRING-LENGTH
+(let ((standard-characters "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789!$\"'(),_-./:;?+<=>#%&*@[\\]{|}`^~"))
+ (do-external-formats (xf)
+ (with-open-file (s "external-format-test.txt" :direction :output
+ :external-format xf)
+ (loop for x across standard-characters
+ for position = (file-position s)
+ for char-length = (file-string-length s x)
+ do (write-char x s)
+ do (assert (= (file-position s) (+ position char-length))))
+ (let ((position (file-position s))
+ (string-length (file-string-length s standard-characters)))
+ (write-string standard-characters s)
+ (assert (= (file-position s) (+ position string-length)))))
+ (delete-file "external-format-test.txt")))
+
+(let ((char-codes '(0 1 255 256 511 512 1023 1024 2047 2048 4095 4096
+ 8191 8192 16383 16384 32767 32768 65535 65536 131071
+ 131072 262143 262144)))
+ (with-open-file (s "external-format-test.txt" :direction :output
+ :external-format :utf-8)
+ (dolist (code char-codes)
+ (let* ((char (code-char code))
+ (position (file-position s))
+ (char-length (file-string-length s char)))
+ (write-char char s)
+ (assert (= (file-position s) (+ position char-length)))))
+ (let* ((string (map 'string #'code-char char-codes))
+ (position (file-position s))
+ (string-length (file-string-length s string)))
+ (write-string string s)
+ (assert (= (file-position s) (+ position string-length))))))
+\f
+
+;;; See sbcl-devel "Subject: Bug in FILE-POSITION on UTF-8-encoded files"
+;;; by Lutz Euler on 2006-03-05 for more details.
+(with-test (:name (:file-position :utf-8)
+ :fails-on :sbcl)
+ (let ((path "external-format-test.txt"))
+ (with-open-file (s path
+ :direction :output
+ :if-exists :supersede
+ :element-type '(unsigned-byte 8))
+ ;; Write #\*, encoded in UTF-8, to the file.
+ (write-byte 42 s)
+ ;; Append #\adiaeresis, encoded in UTF-8, to the file.
+ (write-sequence '(195 164) s))
+ (with-open-file (s path :external-format :utf-8)
+ (read-char s)
+ (let ((pos (file-position s))
+ (char (read-char s)))
+ (format t "read character with code ~a successfully from file position ~a~%"
+ (char-code char) pos)
+ (file-position s pos)
+ (format t "set file position back to ~a, trying to read-char again~%" pos)
+ (let ((new-char (read-char s)))
+ (assert (char= char new-char)))))
+ (values)))
+
+;;;; success
\ No newline at end of file
projects / sbcl.git / blobdiff