;;;; miscellaneous tests of pathname-related stuff ;;;; This file is naturally impure because we mess with ;;;; LOGICAL-PATHNAME-TRANSLATIONS. ;;;; This software is part of the SBCL system. See the README file for ;;;; more information. ;;;; ;;;; While most of SBCL is derived from the CMU CL system, the test ;;;; files (like this one) were written from scratch after the fork ;;;; from CMU CL. ;;;; ;;;; This software is in the public domain and is provided with ;;;; absolutely no warranty. See the COPYING and CREDITS files for ;;;; more information. (load "assertoid.lisp") (use-package "ASSERTOID") (setf (logical-pathname-translations "demo0") '(("**;*.*.*" "/tmp/"))) ;;; In case of a parse error we want to get a condition of type ;;; CL:PARSE-ERROR (or more specifically, of type ;;; SB-KERNEL:NAMESTRING-PARSE-ERROR). (assert (typep (grab-condition (logical-pathname "demo0::bla;file.lisp")) 'parse-error)) ;;; some things SBCL-0.6.9 used not to parse correctly: ;;; ;;; SBCL used to throw an error saying there's no translation. (assert (equal (namestring (translate-logical-pathname "demo0:file.lisp")) "/tmp/file.lisp")) ;;; We do not match a null directory to every wild path: (assert (not (pathname-match-p "demo0:file.lisp" (logical-pathname "demo0:tmp;**;*.*.*")))) ;;; Remove "**" from our resulting pathname when the source-dir is NIL: (setf (logical-pathname-translations "demo1") '(("**;*.*.*" "/tmp/**/*.*") (";**;*.*.*" "/tmp/rel/**/*.*"))) (assert (not (equal (namestring (translate-logical-pathname "demo1:foo.lisp")) "/tmp/**/foo.lisp"))) ;;; That should be correct: (assert (equal (namestring (translate-logical-pathname "demo1:foo.lisp")) "/tmp/foo.lisp")) ;;; Check for absolute/relative path confusion: (assert (not (equal (namestring (translate-logical-pathname "demo1:;foo.lisp")) "tmp/rel/foo.lisp"))) (assert (equal (namestring (translate-logical-pathname "demo1:;foo.lisp")) "/tmp/rel/foo.lisp")) ;;; Under SBCL: new function #'UNPARSE-ENOUGH-NAMESTRING, to ;;; handle the following case exactly (otherwise we get an error: ;;; "#'IDENTITY CALLED WITH 2 ARGS." (setf (logical-pathname-translations "demo2") '(("test;**;*.*" "/tmp/demo2/test"))) (enough-namestring "demo2:test;foo.lisp") ;;; When a pathname comes from a logical host, it should be in upper ;;; case. (This doesn't seem to be specifically required in the ANSI ;;; spec, but it's left up to the implementors, and the arguments made ;;; in the cleanup issue PATHNAME-LOGICAL:ADD seem to be a pretty ;;; compelling reason for the implementors to choose case ;;; insensitivity and a canonical case.) (setf (logical-pathname-translations "FOO") '(("**;*.*.*" "/full/path/to/foo/**/*.*"))) (let* ((pn1 (make-pathname :host "FOO" :directory "etc" :name "INETD" :type "conf")) (pn2 (make-pathname :host "foo" :directory "ETC" :name "inetd" :type "CONF")) (pn3 (read-from-string (prin1-to-string pn1)))) (assert (equal pn1 pn2)) (assert (equal pn1 pn3))) ;;; In addition to the upper-case constraint above, if the logical-pathname ;;; contains a string component in e.g. the directory, name and type slot, ;;; these should be valid "WORDS", according to CLHS 19.3.1. ;;; FIXME: currently SBCL throws NAMESTRING-PARSE-ERROR: should this be ;;; a TYPE-ERROR? (locally ;; MAKE-PATHNAME is UNSAFELY-FLUSHABLE (declare (optimize safety)) (assert (not (ignore-errors (make-pathname :host "FOO" :directory "!bla" :name "bar")))) ;; error: name-component not valid (assert (not (ignore-errors (make-pathname :host "FOO" :directory "bla" :name "!bar")))) ;; error: type-component not valid. (assert (not (ignore-errors (make-pathname :host "FOO" :directory "bla" :name "bar" :type "&baz"))))) ;;; We may need to parse the host as a LOGICAL-NAMESTRING HOST. The ;;; HOST in PARSE-NAMESTRING can be either a string or :UNSPECIFIC ;;; without actually requiring the system to signal an error (apart ;;; from host mismatches). (assert (equal (namestring (parse-namestring "" "FOO")) "FOO:")) (assert (equal (namestring (parse-namestring "" :unspecific)) "")) ;;; The third would work if the call were (and it should continue to ;;; work ...) (parse-namestring "" (pathname-host (translate-logical-pathname "FOO:"))) ;;; ANSI says PARSE-NAMESTRING returns TYPE-ERROR on host mismatch. (let ((cond (grab-condition (parse-namestring "foo:jeamland" "demo2")))) (assert (typep cond 'type-error))) ;;; turning one logical pathname into another: (setf (logical-pathname-translations "foo") '(("todemo;*.*.*" "demo0:*.*.*"))) (assert (equal (namestring (translate-logical-pathname "foo:todemo;x.y")) (namestring (translate-logical-pathname "demo0:x.y")))) ;;; ANSI, in its wisdom, specifies that it's an error (specifically a ;;; TYPE-ERROR) to query the system about the translations of a string ;;; which doesn't have any translations. It's not clear why we don't ;;; just return NIL in that case, but they make the rules.. (let ((cond (grab-condition (logical-pathname-translations "unregistered-host")))) (assert (typep cond 'type-error))) (assert (not (string-equal (host-namestring (parse-namestring "OTHER-HOST:ILLEGAL/LPN")) "OTHER-HOST"))) (assert (string-equal (pathname-name (parse-namestring "OTHER-HOST:ILLEGAL/LPN")) "LPN")) ;;; FIXME: A comment on this section up to sbcl-0.6.11.30 or so said ;;; examples from CLHS: Section 19.4, LOGICAL-PATHNAME-TRANSLATIONS ;;; (sometimes converted to the Un*x way of things) ;;; but when I looked it up I didn't see the connection. Presumably ;;; there's some code in this section which should be attributed ;;; to something in the ANSI spec, but I don't know what code it is ;;; or what section of the specification has the related code. (setf (logical-pathname-translations "test0") '(("**;*.*.*" "/library/foo/**/"))) (assert (equal (namestring (translate-logical-pathname "test0:foo;bar;baz;mum.quux")) "/library/foo/foo/bar/baz/mum.quux")) (setf (logical-pathname-translations "prog") '(("RELEASED;*.*.*" "MY-UNIX:/sys/bin/my-prog/") ("RELEASED;*;*.*.*" "MY-UNIX:/sys/bin/my-prog/*/") ("EXPERIMENTAL;*.*.*" "MY-UNIX:/usr/Joe/development/prog/") ("EXPERIMENTAL;*;*.*.*" "MY-UNIX:/usr/Joe/development/prog/*/"))) (setf (logical-pathname-translations "prog") '(("CODE;*.*.*" "/lib/prog/"))) (assert (equal (namestring (translate-logical-pathname "prog:code;documentation.lisp")) "/lib/prog/documentation.lisp")) (setf (logical-pathname-translations "prog") '(("CODE;DOCUMENTATION.*.*" "/lib/prog/docum.*") ("CODE;*.*.*" "/lib/prog/"))) (assert (equal (namestring (translate-logical-pathname "prog:code;documentation.lisp")) "/lib/prog/docum.lisp")) ;;; ANSI section 19.3.1.1.5 specifies that translation to a filesystem ;;; which doesn't have versions should ignore the version slot. CMU CL ;;; didn't ignore this as it should, but we do. (assert (equal (namestring (translate-logical-pathname "test0:foo;bar;baz;mum.quux.3")) "/library/foo/foo/bar/baz/mum.quux")) ;;;; MERGE-PATHNAME tests ;;;; ;;;; There are some things we don't bother testing, just because they're ;;;; not meaningful on the underlying filesystem anyway. ;;;; ;;;; Mostly that means that we don't do devices, we don't do versions ;;;; except minimally in LPNs (they get lost in the translation to ;;;; physical hosts, so it's not much of an issue), and we don't do ;;;; hosts except for LPN hosts ;;;; ;;;; Although these tests could conceivably be useful in principle for ;;;; other implementations, they depend quite heavily on the rules for ;;;; namestring parsing, which are implementation-specific. So, success ;;;; or failure in these tests doesn't tell you anything about ;;;; ANSI-compliance unless your PARSE-NAMESTRING works like ours. ;;; Needs to be done at compile time, so that the #p"" read-macro ;;; correctly parses things as logical pathnames. This is not a ;;; problem as was, as this is an impure file and so gets loaded in, ;;; but just for future proofing... (eval-when (:compile-toplevel :load-toplevel :execute) (setf (logical-pathname-translations "scratch") '(("**;*.*.*" "/usr/local/doc/**/*")))) (loop for (expected-result . params) in `(;; trivial merge (#P"/usr/local/doc/foo" #p"foo" #p"/usr/local/doc/") ;; If pathname does not specify a host, device, directory, ;; name, or type, each such component is copied from ;; default-pathname. ;; 1) no name, no type (#p"/supplied-dir/name.type" #p"/supplied-dir/" #p"/dir/name.type") ;; 2) no directory, no type (#p"/dir/supplied-name.type" #p"supplied-name" #p"/dir/name.type") ;; 3) no name, no dir (must use make-pathname as ".foo" is parsed ;; as a name) (#p"/dir/name.supplied-type" ,(make-pathname :type "supplied-type") #p"/dir/name.type") ;; If (pathname-directory pathname) is a list whose car is ;; :relative, and (pathname-directory default-pathname) is a ;; list, then the merged directory is [...] (#p"/aaa/bbb/ccc/ddd/qqq/www" #p"qqq/www" #p"/aaa/bbb/ccc/ddd/eee") ;; except that if the resulting list contains a string or ;; :wild immediately followed by :back, both of them are ;; removed. (#P"/aaa/bbb/ccc/blah/eee" ;; "../" in a namestring is parsed as :up not :back, so make-pathname ,(make-pathname :directory '(:relative :back "blah")) #p"/aaa/bbb/ccc/ddd/eee") ;; If (pathname-directory default-pathname) is not a list or ;; (pathname-directory pathname) is not a list whose car is ;; :relative, the merged directory is (or (pathname-directory ;; pathname) (pathname-directory default-pathname)) (#P"/absolute/path/name.type" #p"/absolute/path/name" #p"/dir/default-name.type") ;; === logical pathnames === ;; recognizes a logical pathname namestring when ;; default-pathname is a logical pathname ;; FIXME: 0.6.12.23 fails this one. ;; ;; And, as it happens, it's right to fail it. Because ;; #p"name1" is read in with the ambient *d-p-d* value, which ;; has a physical (Unix) host; therefore, the host of the ;; default-pathname argument to merge-pathnames is ;; irrelevant. The result is (correctly) different if ;; '#p"name1"' is replaced by "name1", below, though it's ;; still not what one might expect... -- CSR, 2002-05-09 #+nil (#P"scratch:foo;name1" #p"name1" #p"scratch:foo;") ;; or when the namestring begins with the name of a defined ;; logical host followed by a colon [I assume that refers to pathname ;; rather than default-pathname] (#p"SCRATCH:FOO;NAME2" #p"scratch:;name2" #p"scratch:foo;") ;; conduct the previous set of tests again, with a lpn first argument (#P"SCRATCH:USR;LOCAL;DOC;FOO" #p"scratch:;foo" #p"/usr/local/doc/") (#p"SCRATCH:SUPPLIED-DIR;NAME.TYPE" #p"scratch:supplied-dir;" #p"/dir/name.type") (#p"SCRATCH:DIR;SUPPLIED-NAME.TYPE" #p"scratch:;supplied-name" #p"/dir/name.type") (#p"SCRATCH:DIR;NAME.SUPPLIED-TYPE" ,(make-pathname :host "scratch" :type "supplied-type") #p"/dir/name.type") (#p"SCRATCH:AAA;BBB;CCC;DDD;FOO;BAR" ,(make-pathname :host "scratch" :directory '(:relative "foo") :name "bar") #p"/aaa/bbb/ccc/ddd/eee") (#p"SCRATCH:AAA;BBB;CCC;FOO;BAR" ,(make-pathname :host "scratch" :directory '(:relative :back "foo") :name "bar") #p"/aaa/bbb/ccc/ddd/eee") (#p"SCRATCH:ABSOLUTE;PATH;NAME.TYPE" #p"scratch:absolute;path;name" #p"/dir/default-name.type") ;; FIXME: test version handling in LPNs ) do (let ((result (apply #'merge-pathnames params))) (macrolet ((frob (op) `(assert (equal (,op result) (,op expected-result))))) (frob pathname-host) (frob pathname-directory) (frob pathname-name) (frob pathname-type)))) ;;; host-namestring testing (assert (string= (namestring (parse-namestring "/foo" (host-namestring #p"/bar"))) "/foo")) (assert (string= (namestring (parse-namestring "FOO" (host-namestring #p"SCRATCH:BAR"))) "SCRATCH:FOO")) (assert (raises-error? (setf (logical-pathname-translations "") (list '("**;*.*.*" "/**/*.*"))))) ;;; Bug 200: translate-logical-pathname is according to the spec supposed ;;; not to give errors if asked to translate a namestring for a valid ;;; physical pathname. Failed in 0.7.7.28 and before (assert (string= (namestring (translate-logical-pathname "/")) "/")) ;;; Not strictly pathname logic testing, but until sbcl-0.7.6.19 we ;;; had difficulty with non-FILE-STREAM stream arguments to pathname ;;; functions (they would cause memory protection errors). Make sure ;;; that those errors are gone: (assert (raises-error? (pathname (make-string-input-stream "FOO")) type-error)) (assert (raises-error? (merge-pathnames (make-string-output-stream)) type-error)) ;;; ensure read/print consistency (or print-not-readable-error) on ;;; pathnames: (let ((pathnames (list (make-pathname :name "foo" :type "txt" :version :newest) (make-pathname :name "foo" :type "txt" :version 1) (make-pathname :name "foo" :type ".txt") (make-pathname :name "foo." :type "txt") (parse-namestring "SCRATCH:FOO.TXT.1") (parse-namestring "SCRATCH:FOO.TXT.NEWEST") (parse-namestring "SCRATCH:FOO.TXT")))) (dolist (p pathnames) (print p) (handler-case (let ((*print-readably* t)) (assert (equal (read-from-string (format nil "~S" p)) p))) (print-not-readable () nil)))) ;;; BUG 330: "PARSE-NAMESTRING should accept namestrings as the ;;; default argument" ...and streams as well (assert (equal (parse-namestring "foo" nil "/") (parse-namestring "foo" nil #P"/"))) (let ((test "parse-namestring-test.tmp")) (unwind-protect (with-open-file (f test :direction :output) ;; FIXME: This test is a bit flaky, since we only check that ;; no error is signalled. The dilemma here is "what is the ;; correct result when defaults is a _file_, not a ;; directory". Currently (0.8.10.73) we get #P"foo" here (as ;; opposed to eg. #P"/path/to/current/foo"), which is ;; possibly mildly surprising but probably conformant. (assert (parse-namestring "foo" nil f))) (when (probe-file test) (delete-file test)))) ;;; ENOUGH-NAMESTRING should probably not fail when the namestring in ;;; question has a :RELATIVE pathname. (assert (equal (enough-namestring #p"foo" #p"./") "foo")) ;;; bug reported by Artem V. Andreev: :WILD not handled in unparsing ;;; directory lists. (assert (equal (namestring #p"/tmp/*/") "/tmp/*/")) ;;; Printing of pathnames; see CLHS 22.1.3.1. This section was started ;;; to confirm that pathnames are printed as their namestrings under ;;; :escape nil :readably nil. (loop for (pathname expected . vars) in `((#p"/foo" "#P\"/foo\"") (#p"/foo" "#P\"/foo\"" :readably nil) (#p"/foo" "#P\"/foo\"" :escape nil) (#p"/foo" "/foo" :readably nil :escape nil)) for actual = (with-standard-io-syntax (apply #'write-to-string pathname vars)) do (assert (string= expected actual) () "~S should be ~S, was ~S" (list* 'write-to-string pathname vars) expected actual)) ;;; we got (truename "/") wrong for about 6 months. Check that it's ;;; still right. (let ((pathname (truename "/"))) (assert (equalp pathname #p"/")) (assert (equal (pathname-directory pathname) '(:absolute)))) ;;; we failed to unparse logical pathnames with :NAME :WILD :TYPE NIL. ;;; (Reported by Pascal Bourguignon. (let ((pathname (make-pathname :host "SYS" :directory '(:absolute :wild-inferiors) :name :wild :type nil))) (assert (string= (namestring pathname) "SYS:**;*")) (assert (string= (write-to-string pathname :readably t) "#P\"SYS:**;*\""))) ;;; reported by James Y Knight on sbcl-devel 2006-05-17 (let ((p1 (make-pathname :directory '(:relative "bar"))) (p2 (make-pathname :directory '(:relative :back "foo")))) (assert (equal (merge-pathnames p1 p2) (make-pathname :directory '(:relative :back "foo" "bar"))))) ;;; construct native namestrings even if the directory is empty (means ;;; that same as if (:relative)) (assert (equal (sb-ext:native-namestring (make-pathname :directory '(:relative) :name "foo" :type "txt")) (sb-ext:native-namestring (let ((p (make-pathname :directory nil :name "foo" :type "txt"))) (assert (not (pathname-directory p))) p)))) ;;; reported by Richard Kreuter: PATHNAME and MERGE-PATHNAMES used to ;;; be unsafely-flushable. Since they are known to return non-nil values ;;; only, the test-node of the IF is flushed, and since the function ;;; is unsafely-flushable, out it goes, and bad pathname designators ;;; breeze through. ;;; ;;; These tests rely on using a stream that appears as a file-stream ;;; but isn't a valid pathname-designator. (assert (eq :false (if (ignore-errors (pathname sb-sys::*tty*)) :true :false))) (assert (eq :false (if (ignore-errors (merge-pathnames sb-sys::*tty*)) :true :false))) ;;; This used to return "quux/bar.lisp" (assert (equal #p"quux/bar.fasl" (let ((*default-pathname-defaults* #p"quux/")) (compile-file-pathname "foo.lisp" :output-file "bar")))) (assert (equal #p"quux/bar.fasl" (let ((*default-pathname-defaults* #p"quux/")) (compile-file-pathname "bar.lisp")))) (enough-namestring #p".a*") (assert (eq 99 (pathname-version (translate-pathname (make-pathname :name "foo" :type "bar" :version 99) (make-pathname :name :wild :type :wild :version :wild) (make-pathname :name :wild :type :wild :version :wild))))) (assert (eq 99 (pathname-version (translate-pathname (make-pathname :name "foo" :type "bar" :version 99) (make-pathname :name :wild :type :wild :version :wild) (make-pathname :name :wild :type :wild :version nil))))) ;;; enough-namestring relative to root (assert (equal "foo" (enough-namestring "/foo" "/"))) ;;; Check the handling of NIL, :UNSPECIFIC, the empty string, and ;;; non-NIL strings in NATIVE-NAMESTRING implementations. Revised by ;;; RMK 2007-11-28, attempting to preserve the apparent intended ;;; denotation of SBCL's then-current pathname implementation. (assert (equal (loop with components = (list nil :unspecific "" "a") for name in components appending (loop for type in components as pathname = (make-pathname #+win32 "C" :directory '(:absolute "tmp") :name name :type type) collect (ignore-errors (sb-ext:native-namestring pathname)))) #-win32 #|type NIL :UNSPECIFIC "" "a" |# #|name |# #|NIL |# '("/tmp/" "/tmp/" NIL NIL #|:UNSPECIFIC|# "/tmp/" "/tmp/" NIL NIL #|"" |# "/tmp/" "/tmp/" "/tmp/." "/tmp/.a" #|"a" |# "/tmp/a" "/tmp/a" "/tmp/a." "/tmp/a.a") #+win32 #|type NIL :UNSPECIFIC "" "a" |# #|name |# #|NIL |# '("C:\\tmp\\" "C:\\tmp\\" NIL NIL #|:UNSPECIFIC|# "C:\\tmp\\" "C:\\tmp\\" NIL NIL #|"" |# "C:\\tmp\\" "C:\\tmp\\" "C:\\tmp\\." "C:\\tmp\\.a" #|"a" |# "C:\\tmp\\a" "C:\\tmp\\a" "C:\\tmp\\a." "C:\\tmp\\a.a"))) ;;;; success