Raymond Toy pushed to branch issue-139-add-alias-local-external-format at cmucl / cmucl
Commits:
-
4a7207b6
by Raymond Toy at 2022-10-17T18:58:45+00:00
-
ba5c5d2a
by Raymond Toy at 2022-10-17T18:58:45+00:00
-
e8a0cc6c
by Raymond Toy at 2022-10-30T15:03:27+00:00
-
0dad5a1a
by Raymond Toy at 2022-10-30T15:03:28+00:00
-
1300830b
by Raymond Toy at 2022-10-31T17:12:48+00:00
-
649a4f1e
by Raymond Toy at 2022-10-31T17:12:49+00:00
-
69f2a990
by Raymond Toy at 2022-10-31T10:14:48-07:00
9 changed files:
- src/code/extfmts.lisp
- src/code/filesys.lisp
- src/general-info/release-21e.md
- src/lisp/os-common.c
- src/pcl/gray-streams.lisp
- + tests/.gitignore
- tests/issues.lisp
- + tests/utf8.txt
- + tests/안녕하십니까.txt
Changes:
| ... | ... | @@ -22,7 +22,7 @@ |
| 22 | 22 | describe-external-format))
|
| 23 | 23 | |
| 24 | 24 | (defvar *default-external-format*
|
| 25 | - :iso8859-1
|
|
| 25 | + :utf-8
|
|
| 26 | 26 | "The default external format to use if no other external format is
|
| 27 | 27 | specified")
|
| 28 | 28 |
| ... | ... | @@ -1079,13 +1079,21 @@ optionally keeping some of the most recent old versions." |
| 1079 | 1079 | :pathname file
|
| 1080 | 1080 | :format-control (intl:gettext "~S doesn't exist.")
|
| 1081 | 1081 | :format-arguments (list file)))
|
| 1082 | - (multiple-value-bind (winp dev ino mode nlink uid)
|
|
| 1083 | - (unix:unix-stat name)
|
|
| 1084 | - (declare (ignore dev ino mode nlink))
|
|
| 1085 | - (when winp
|
|
| 1086 | - (let ((user-info (unix:unix-getpwuid uid)))
|
|
| 1087 | - (when user-info
|
|
| 1088 | - (unix:user-info-name user-info))))))))
|
|
| 1082 | + ;; unix-namestring converts "." to "". Convert it back to
|
|
| 1083 | + ;; "." so we can stat the current directory. (Perhaps
|
|
| 1084 | + ;; that's a bug in unix-namestring?)
|
|
| 1085 | + (when (zerop (length name))
|
|
| 1086 | + (setf name "."))
|
|
| 1087 | + (let (author)
|
|
| 1088 | + (unwind-protect
|
|
| 1089 | + (progn
|
|
| 1090 | + (setf author (alien:alien-funcall
|
|
| 1091 | + (alien:extern-alien "os_file_author"
|
|
| 1092 | + (function (alien:* c-call:c-string) c-call:c-string))
|
|
| 1093 | + (unix::%name->file name)))
|
|
| 1094 | + (unless (alien:null-alien author)
|
|
| 1095 | + (alien:cast author c-call:c-string)))
|
|
| 1096 | + (alien:free-alien author))))))
|
|
| 1089 | 1097 | |
| 1090 | 1098 | |
| 1091 | 1099 | ;;;; DIRECTORY.
|
| ... | ... | @@ -22,6 +22,7 @@ public domain. |
| 22 | 22 | * Feature enhancements
|
| 23 | 23 | * Changes
|
| 24 | 24 | * Update to ASDF 3.3.6
|
| 25 | + * The default external format is `:utf-8` instead of `:iso8859-1`
|
|
| 25 | 26 | * ANSI compliance fixes:
|
| 26 | 27 | * Bug fixes:
|
| 27 | 28 | * ~~#97~~ Fixes stepping through the source forms in the debugger. This has been broken for quite some time, but it works now.
|
| ... | ... | @@ -56,6 +57,7 @@ public domain. |
| 56 | 57 | * ~~#132~~ Ansi test `RENAME-FILE.1` no fails
|
| 57 | 58 | * ~~#134~~ Handle the case of `(expt complex complex-rational)`
|
| 58 | 59 | * ~~#136~~ `ensure-directories-exist` should return the given pathspec
|
| 60 | + * #139 `*default-external-format*` defaults to `:utf-8`
|
|
| 59 | 61 | * ~~#142~~ `(random 0)` signals incorrect error
|
| 60 | 62 | * Other changes:
|
| 61 | 63 | * Improvements to the PCL implementation of CLOS:
|
| ... | ... | @@ -5,12 +5,16 @@ |
| 5 | 5 | |
| 6 | 6 | */
|
| 7 | 7 | |
| 8 | +#include <assert.h>
|
|
| 8 | 9 | #include <errno.h>
|
| 9 | 10 | #include <math.h>
|
| 10 | 11 | #include <netdb.h>
|
| 12 | +#include <pwd.h>
|
|
| 11 | 13 | #include <stdio.h>
|
| 14 | +#include <stdlib.h>
|
|
| 12 | 15 | #include <string.h>
|
| 13 | 16 | #include <sys/stat.h>
|
| 17 | +#include <unistd.h>
|
|
| 14 | 18 | #include <time.h>
|
| 15 | 19 | |
| 16 | 20 | #include "os.h"
|
| ... | ... | @@ -715,3 +719,57 @@ os_lstat(const char* path, u_int64_t *dev, u_int64_t *ino, unsigned int *mode, u |
| 715 | 719 | |
| 716 | 720 | return rc;
|
| 717 | 721 | }
|
| 722 | + |
|
| 723 | +/*
|
|
| 724 | + * Interface for file-author. Given a pathname, returns a new string
|
|
| 725 | + * holding the author of the file or NULL if some error occurred. The
|
|
| 726 | + * caller is responsible for freeing the memory used by the string.
|
|
| 727 | + */
|
|
| 728 | +char *
|
|
| 729 | +os_file_author(const char *path)
|
|
| 730 | +{
|
|
| 731 | + struct stat sb;
|
|
| 732 | + char initial[1024];
|
|
| 733 | + char *buffer, *obuffer;
|
|
| 734 | + size_t size;
|
|
| 735 | + struct passwd pwd;
|
|
| 736 | + struct passwd *ppwd;
|
|
| 737 | + char *result;
|
|
| 738 | + |
|
| 739 | + if (stat(path, &sb) != 0) {
|
|
| 740 | + return NULL;
|
|
| 741 | + }
|
|
| 742 | + |
|
| 743 | + result = NULL;
|
|
| 744 | + buffer = initial;
|
|
| 745 | + obuffer = NULL;
|
|
| 746 | + size = sizeof(initial) / sizeof(initial[0]);
|
|
| 747 | + |
|
| 748 | + /*
|
|
| 749 | + * Keep trying with larger buffers until a maximum is reached. We
|
|
| 750 | + * assume (1 << 20) is large enough for any OS.
|
|
| 751 | + */
|
|
| 752 | + while (size <= (1 << 20)) {
|
|
| 753 | + switch (getpwuid_r(sb.st_uid, &pwd, buffer, size, &ppwd)) {
|
|
| 754 | + case 0:
|
|
| 755 | + /* Success, though we might not have a matching entry */
|
|
| 756 | + result = (ppwd == NULL) ? NULL : strdup(pwd.pw_name);
|
|
| 757 | + goto exit;
|
|
| 758 | + case ERANGE:
|
|
| 759 | + /* Buffer is too small, double its size and try again */
|
|
| 760 | + size *= 2;
|
|
| 761 | + obuffer = (buffer == initial) ? NULL : buffer;
|
|
| 762 | + if ((buffer = realloc(obuffer, size)) == NULL) {
|
|
| 763 | + goto exit;
|
|
| 764 | + }
|
|
| 765 | + continue;
|
|
| 766 | + default:
|
|
| 767 | + /* All other errors */
|
|
| 768 | + goto exit;
|
|
| 769 | + }
|
|
| 770 | + }
|
|
| 771 | +exit:
|
|
| 772 | + free(obuffer);
|
|
| 773 | +
|
|
| 774 | + return result;
|
|
| 775 | +} |
| ... | ... | @@ -235,6 +235,9 @@ |
| 235 | 235 | defined for this function, although it is permissible for it to
|
| 236 | 236 | always return NIL."))
|
| 237 | 237 | |
| 238 | +(defmethod stream-line-column ((stream fundamental-character-output-stream))
|
|
| 239 | + nil)
|
|
| 240 | + |
|
| 238 | 241 | ;;; Stream-line-length is a CMUCL extension to Gray streams.
|
| 239 | 242 | (defgeneric stream-line-length (stream)
|
| 240 | 243 | (:documentation _N"Return the stream line length or Nil."))
|
| 1 | +/out-utf8.txt |
| ... | ... | @@ -5,6 +5,12 @@ |
| 5 | 5 | |
| 6 | 6 | (in-package "ISSUES-TESTS")
|
| 7 | 7 | |
| 8 | +(defparameter *test-path*
|
|
| 9 | + (merge-pathnames (make-pathname :name :unspecific :type :unspecific
|
|
| 10 | + :version :unspecific)
|
|
| 11 | + *load-truename*)
|
|
| 12 | + "Path to where this file is.")
|
|
| 13 | + |
|
| 8 | 14 | (defun square (x)
|
| 9 | 15 | (expt x 2))
|
| 10 | 16 | |
| ... | ... | @@ -670,3 +676,72 @@ |
| 670 | 676 | (err (relerr value answer)))
|
| 671 | 677 | (assert-true (<= err eps) base err eps)))))))
|
| 672 | 678 | |
| 679 | +(define-test issue.130
|
|
| 680 | + (:tag :issues)
|
|
| 681 | + ;; Just verify that file-author works. In particular "." should
|
|
| 682 | + ;; work and not return NIL.
|
|
| 683 | + (assert-true (file-author "."))
|
|
| 684 | + (assert-true (file-author "bin/build.sh"))
|
|
| 685 | + (let ((unix::*filename-encoding* :utf-8))
|
|
| 686 | + ;; Set filename encoding to utf-8 so that we can encode the
|
|
| 687 | + ;; filename properly.
|
|
| 688 | + (assert-true
|
|
| 689 | + (file-author
|
|
| 690 | + (merge-pathnames
|
|
| 691 | + (concatenate 'string
|
|
| 692 | + ;; Write the test file name this way so
|
|
| 693 | + ;; that it's independent of the encoding
|
|
| 694 | + ;; used to load this file. The name is
|
|
| 695 | + ;; "안녕하십니까".
|
|
| 696 | + '(#\Hangul_Syllable_An #\Hangul_Syllable_Nyeong #\Hangul_Syllable_Ha
|
|
| 697 | + #\Hangul_Syllable_Sib #\Hangul_Syllable_Ni #\Hangul_Syllable_Gga)
|
|
| 698 | + ".txt")
|
|
| 699 | + *test-path*)))))
|
|
| 700 | + |
|
| 701 | +(define-test issue.139-default-external-format
|
|
| 702 | + (:tag :issues)
|
|
| 703 | + (assert-eq :utf-8 stream:*default-external-format*))
|
|
| 704 | + |
|
| 705 | +(define-test issue.139-default-external-format-read-file
|
|
| 706 | + (:tag :issues)
|
|
| 707 | + (let ((string (concatenate 'string
|
|
| 708 | + ;; This is "hello" in Korean
|
|
| 709 | + '(#\Hangul_syllable_an
|
|
| 710 | + #\Hangul_Syllable_Nyeong
|
|
| 711 | + #\Hangul_Syllable_Ha
|
|
| 712 | + #\Hangul_Syllable_Se
|
|
| 713 | + #\Hangul_Syllable_Yo))))
|
|
| 714 | + ;; Test that opening a file for reading uses the the default :utf8
|
|
| 715 | + ;; encoding.
|
|
| 716 | + (with-open-file (s (merge-pathnames "utf8.txt"
|
|
| 717 | + *test-path*)
|
|
| 718 | + :direction :input)
|
|
| 719 | + ;; The first line should be "hello" in Hangul.
|
|
| 720 | + (assert-equal (map 'list #'char-name string)
|
|
| 721 | + (map 'list #'char-name (read-line s))))))
|
|
| 722 | + |
|
| 723 | +(define-test issue.139-default-external-format-write-file
|
|
| 724 | + (:tag :issues)
|
|
| 725 | + ;; Test that opening a file for writing uses the default :utf8.
|
|
| 726 | + ;; First write something out to the file. Then read it back in
|
|
| 727 | + ;; using an explicit format of utf8 and verifying that we got the
|
|
| 728 | + ;; right contents.
|
|
| 729 | + (let ((string (concatenate 'string
|
|
| 730 | + ;; This is "hello" in Korean
|
|
| 731 | + '(#\Hangul_syllable_an
|
|
| 732 | + #\Hangul_Syllable_Nyeong
|
|
| 733 | + #\Hangul_Syllable_Ha
|
|
| 734 | + #\Hangul_Syllable_Se
|
|
| 735 | + #\Hangul_Syllable_Yo))))
|
|
| 736 | + (with-open-file (s (merge-pathnames "out-utf8.txt"
|
|
| 737 | + *test-path*)
|
|
| 738 | + :direction :output
|
|
| 739 | + :if-exists :supersede)
|
|
| 740 | + (write-line string s))
|
|
| 741 | + (with-open-file (s (merge-pathnames "out-utf8.txt"
|
|
| 742 | + *test-path*)
|
|
| 743 | + :direction :input
|
|
| 744 | + :external-format :utf-8)
|
|
| 745 | + (assert-equal (map 'list #'char-name string)
|
|
| 746 | + (map 'list #'char-name (read-line s))))))
|
|
| 747 | + |
| 1 | +안녕하세요
|
|
| 2 | +UTF8 test. The above line is "Hello" in Hangul. |
| 1 | +The file name of this file is "안녕하십니까.txt" ("Hello" in Korean.)
|
|
| 2 | + |
|
| 3 | + |