src/code/win32-pathname.lisp

   1 ;;;; pathname parsing for Win32 filesystems
   2
   3 ;;;; This software is part of the SBCL system. See the README file for
   4 ;;;; more information.
   5 ;;;;
   6 ;;;; This software is derived from the CMU CL system, which was
   7 ;;;; written at Carnegie Mellon University and released into the
   8 ;;;; public domain. The software is in the public domain and is
   9 ;;;; provided with absolutely no warranty. See the COPYING and CREDITS
  10 ;;;; files for more information.
  11
  12 (in-package "SB!IMPL")
  13
  14 (defun extract-device (namestr start end)
  15   (declare (type simple-string namestr)
  16            (type index start end))
  17   (if (and (>= end (+ start 2))
  18            (alpha-char-p (char namestr start))
  19            (eql (char namestr (1+ start)) #\:))
  20       (values (string (char namestr start)) (+ start 2))
  21       (values nil start)))
  22
  23 (defun split-at-slashes-and-backslashes (namestr start end)
  24   (declare (type simple-string namestr)
  25            (type index start end))
  26   (let ((absolute (and (/= start end)
  27                        (or (char= (schar namestr start) #\/)
  28                            (char= (schar namestr start) #\\)))))
  29     (when absolute
  30       (incf start))
  31     ;; Next, split the remainder into slash-separated chunks.
  32     (collect ((pieces))
  33       (loop
  34          (let ((slash (position-if (lambda (c)
  35                                      (or (char= c #\/)
  36                                          (char= c #\\)))
  37                                    namestr :start start :end end)))
  38            (pieces (cons start (or slash end)))
  39            (unless slash
  40              (return))
  41            (setf start (1+ slash))))
  42       (values absolute (pieces)))))
  43
  44 (defun parse-win32-namestring (namestring start end)
  45   (declare (type simple-string namestring)
  46            (type index start end))
  47   (setf namestring (coerce namestring 'simple-string))
  48   (multiple-value-bind (device new-start)
  49       (extract-device namestring start end)
  50     (multiple-value-bind (absolute pieces)
  51         (split-at-slashes-and-backslashes namestring new-start end)
  52       (multiple-value-bind (name type version)
  53           (let* ((tail (car (last pieces)))
  54                  (tail-start (car tail))
  55                  (tail-end (cdr tail)))
  56             (unless (= tail-start tail-end)
  57               (setf pieces (butlast pieces))
  58               (extract-name-type-and-version namestring tail-start tail-end)))
  59
  60         (when (stringp name)
  61           (let ((position (position-if (lambda (char)
  62                                          (or (char= char (code-char 0))
  63                                              (char= char #\/)))
  64                                        name)))
  65             (when position
  66               (error 'namestring-parse-error
  67                      :complaint "can't embed #\\Nul or #\\/ in Unix namestring"
  68                      :namestring namestring
  69                      :offset position))))
  70         ;; Now we have everything we want. So return it.
  71         (values nil ; no host for Win32 namestrings
  72                 device
  73                 (collect ((dirs))
  74                   (dolist (piece pieces)
  75                     (let ((piece-start (car piece))
  76                           (piece-end (cdr piece)))
  77                       (unless (= piece-start piece-end)
  78                         (cond ((string= namestring ".."
  79                                         :start1 piece-start
  80                                         :end1 piece-end)
  81                                (dirs :up))
  82                               ((string= namestring "**"
  83                                         :start1 piece-start
  84                                         :end1 piece-end)
  85                                (dirs :wild-inferiors))
  86                               (t
  87                                (dirs (maybe-make-pattern namestring
  88                                                          piece-start
  89                                                          piece-end)))))))
  90                   (cond (absolute
  91                          (cons :absolute (dirs)))
  92                         ((dirs)
  93                          (cons :relative (dirs)))
  94                         (t
  95                          nil)))
  96                 name
  97                 type
  98                 version)))))
  99
 100 (defun parse-native-win32-namestring (namestring start end)
 101   (declare (type simple-string namestring)
 102            (type index start end))
 103   (setf namestring (coerce namestring 'simple-string))
 104   (multiple-value-bind (device new-start)
 105       (extract-device namestring start end)
 106     (multiple-value-bind (absolute ranges)
 107         (split-at-slashes-and-backslashes namestring new-start end)
 108       (let* ((components (loop for ((start . end) . rest) on ranges
 109                                for piece = (subseq namestring start end)
 110                                collect (if (and (string= piece "..") rest)
 111                                            :up
 112                                            piece)))
 113              (name-and-type
 114               (let* ((end (first (last components)))
 115                      (dot (position #\. end :from-end t)))
 116                 ;; FIXME: can we get this dot-interpretation knowledge
 117                 ;; from existing code?  EXTRACT-NAME-TYPE-AND-VERSION
 118                 ;; does slightly more work than that.
 119                 (cond
 120                   ((string= end "")
 121                    (list nil nil))
 122                   ((and dot (> dot 0))
 123                    (list (subseq end 0 dot) (subseq end (1+ dot))))
 124                   (t
 125                    (list end nil))))))
 126         (values nil
 127                 device
 128                 (cons (if absolute :absolute :relative) (butlast components))
 129                 (first name-and-type)
 130                 (second name-and-type)
 131                 nil)))))
 132
 133
 134
 135 (defun unparse-win32-host (pathname)
 136   (declare (type pathname pathname)
 137            (ignore pathname))
 138   ;; FIXME: same as UNPARSE-UNIX-HOST.  That's probably not good.
 139   "")
 140
 141 (defun unparse-win32-device (pathname)
 142   (declare (type pathname pathname))
 143   (let ((device (pathname-device pathname)))
 144     (if (or (null device) (eq device :unspecific))
 145         ""
 146         (concatenate 'simple-string (string device) ":"))))
 147
 148 (defun unparse-win32-piece (thing)
 149   (etypecase thing
 150     ((member :wild) "*")
 151     (simple-string
 152      (let* ((srclen (length thing))
 153             (dstlen srclen))
 154        (dotimes (i srclen)
 155          (case (schar thing i)
 156            ((#\* #\? #\[)
 157             (incf dstlen))))
 158        (let ((result (make-string dstlen))
 159              (dst 0))
 160          (dotimes (src srclen)
 161            (let ((char (schar thing src)))
 162              (case char
 163                ((#\* #\? #\[)
 164                 (setf (schar result dst) #\\)
 165                 (incf dst)))
 166              (setf (schar result dst) char)
 167              (incf dst)))
 168          result)))
 169     (pattern
 170      (collect ((strings))
 171        (dolist (piece (pattern-pieces thing))
 172          (etypecase piece
 173            (simple-string
 174             (strings piece))
 175            (symbol
 176             (ecase piece
 177               (:multi-char-wild
 178                (strings "*"))
 179               (:single-char-wild
 180                (strings "?"))))
 181            (cons
 182             (case (car piece)
 183               (:character-set
 184                (strings "[")
 185                (strings (cdr piece))
 186                (strings "]"))
 187               (t
 188                (error "invalid pattern piece: ~S" piece))))))
 189        (apply #'concatenate
 190               'simple-string
 191               (strings))))))
 192
 193 (defun unparse-win32-directory-list (directory)
 194   (declare (type list directory))
 195   (collect ((pieces))
 196     (when directory
 197       (ecase (pop directory)
 198         (:absolute
 199          (pieces "\\"))
 200         (:relative
 201          ;; nothing special
 202          ))
 203       (dolist (dir directory)
 204         (typecase dir
 205           ((member :up)
 206            (pieces "..\\"))
 207           ((member :back)
 208            (error ":BACK cannot be represented in namestrings."))
 209           ((member :wild-inferiors)
 210            (pieces "**\\"))
 211           ((or simple-string pattern (member :wild))
 212            (pieces (unparse-unix-piece dir))
 213            (pieces "\\"))
 214           (t
 215            (error "invalid directory component: ~S" dir)))))
 216     (apply #'concatenate 'simple-string (pieces))))
 217
 218 (defun unparse-win32-directory (pathname)
 219   (declare (type pathname pathname))
 220   (unparse-win32-directory-list (%pathname-directory pathname)))
 221
 222 (defun unparse-win32-file (pathname)
 223   (declare (type pathname pathname))
 224   (collect ((strings))
 225     (let* ((name (%pathname-name pathname))
 226            (type (%pathname-type pathname))
 227            (type-supplied (not (or (null type) (eq type :unspecific)))))
 228       ;; Note: by ANSI 19.3.1.1.5, we ignore the version slot when
 229       ;; translating logical pathnames to a filesystem without
 230       ;; versions (like Win32).
 231       (when name
 232         (when (and (null type)
 233                    (typep name 'string)
 234                    (> (length name) 0)
 235                    (position #\. name :start 1))
 236           (error "too many dots in the name: ~S" pathname))
 237         (when (and (typep name 'string)
 238                    (string= name ""))
 239           (error "name is of length 0: ~S" pathname))
 240         (strings (unparse-unix-piece name)))
 241       (when type-supplied
 242         (unless name
 243           (error "cannot specify the type without a file: ~S" pathname))
 244         (when (typep type 'simple-string)
 245           (when (position #\. type)
 246             (error "type component can't have a #\. inside: ~S" pathname)))
 247         (strings ".")
 248         (strings (unparse-unix-piece type))))
 249     (apply #'concatenate 'simple-string (strings))))
 250
 251 (defun unparse-win32-namestring (pathname)
 252   (declare (type pathname pathname))
 253   (concatenate 'simple-string
 254                (unparse-win32-device pathname)
 255                (unparse-win32-directory pathname)
 256                (unparse-win32-file pathname)))
 257
 258 (defun unparse-native-win32-namestring (pathname)
 259   (declare (type pathname pathname))
 260   (let ((device (pathname-device pathname))
 261         (directory (pathname-directory pathname))
 262         (name (pathname-name pathname))
 263         (type (pathname-type pathname)))
 264     (coerce
 265      (with-output-to-string (s)
 266        (when device
 267          (write-string device s)
 268          (write-char #\: s))
 269        (tagbody
 270           (ecase (pop directory)
 271             (:absolute (write-char #\\ s))
 272             (:relative))
 273           (unless directory (go :done))
 274         :subdir
 275           (let ((piece (pop directory)))
 276             (typecase piece
 277               ((member :up) (write-string ".." s))
 278               (string (write-string piece s))
 279               (t (error "ungood piece in NATIVE-NAMESTRING: ~S" piece)))
 280             (when (or directory name type)
 281               (write-char #\\ s)))
 282           (when directory
 283             (go :subdir))
 284         :done)
 285        (when name
 286          (unless (stringp name)
 287            (error "non-STRING name in NATIVE-NAMESTRING: ~S" name))
 288          (write-string name s)
 289          (when type
 290            (unless (stringp type)
 291              (error "non-STRING type in NATIVE-NAMESTRING: ~S" name))
 292            (write-char #\. s)
 293            (write-string type s))))
 294      'simple-string)))
 295
 296 ;;; FIXME.
 297 (defun unparse-win32-enough (pathname defaults)
 298   (declare (type pathname pathname defaults))
 299   (flet ((lose ()
 300            (error "~S cannot be represented relative to ~S."
 301                   pathname defaults)))
 302     (collect ((strings))
 303       (let* ((pathname-directory (%pathname-directory pathname))
 304              (defaults-directory (%pathname-directory defaults))
 305              (prefix-len (length defaults-directory))
 306              (result-directory
 307               (cond ((null pathname-directory) '(:relative))
 308                     ((eq (car pathname-directory) :relative)
 309                      pathname-directory)
 310                     ((and (> prefix-len 0)
 311                           (>= (length pathname-directory) prefix-len)
 312                           (compare-component (subseq pathname-directory
 313                                                      0 prefix-len)
 314                                              defaults-directory))
 315                      ;; Pathname starts with a prefix of default. So
 316                      ;; just use a relative directory from then on out.
 317                      (cons :relative (nthcdr prefix-len pathname-directory)))
 318                     ((eq (car pathname-directory) :absolute)
 319                      ;; We are an absolute pathname, so we can just use it.
 320                      pathname-directory)
 321                     (t
 322                      (bug "Bad fallthrough in ~S" 'unparse-unix-enough)))))
 323         (strings (unparse-unix-directory-list result-directory)))
 324       (let* ((pathname-type (%pathname-type pathname))
 325              (type-needed (and pathname-type
 326                                (not (eq pathname-type :unspecific))))
 327              (pathname-name (%pathname-name pathname))
 328              (name-needed (or type-needed
 329                               (and pathname-name
 330                                    (not (compare-component pathname-name
 331                                                            (%pathname-name
 332                                                             defaults)))))))
 333         (when name-needed
 334           (unless pathname-name (lose))
 335           (when (and (null pathname-type)
 336                      (typep pathname-name 'simple-string)
 337                      (position #\. pathname-name :start 1))
 338             (error "too many dots in the name: ~S" pathname))
 339           (strings (unparse-unix-piece pathname-name)))
 340         (when type-needed
 341           (when (or (null pathname-type) (eq pathname-type :unspecific))
 342             (lose))
 343           (when (typep pathname-type 'simple-string)
 344             (when (position #\. pathname-type)
 345               (error "type component can't have a #\. inside: ~S" pathname)))
 346           (strings ".")
 347           (strings (unparse-unix-piece pathname-type))))
 348       (apply #'concatenate 'simple-string (strings)))))
 349
 350 ;; FIXME: This has been converted rather blindly from the Unix
 351 ;; version, with no reference to any Windows docs what so ever.
 352 (defun simplify-win32-namestring (src)
 353   (declare (type simple-string src))
 354   (let* ((src-len (length src))
 355          (dst (make-string src-len :element-type 'character))
 356          (dst-len 0)
 357          (dots 0)
 358          (last-slash nil))
 359     (flet ((deposit (char)
 360              (setf (schar dst dst-len) char)
 361              (incf dst-len))
 362            (slashp (char)
 363              (find char "\\/")))
 364       (dotimes (src-index src-len)
 365         (let ((char (schar src src-index)))
 366           (cond ((char= char #\.)
 367                  (when dots
 368                    (incf dots))
 369                  (deposit char))
 370                 ((slashp char)
 371                  (case dots
 372                    (0
 373                     ;; either ``/...' or ``...//...'
 374                     (unless last-slash
 375                       (setf last-slash dst-len)
 376                       (deposit char)))
 377                    (1
 378                     ;; either ``./...'' or ``..././...''
 379                     (decf dst-len))
 380                    (2
 381                     ;; We've found ..
 382                     (cond
 383                       ((and last-slash (not (zerop last-slash)))
 384                        ;; There is something before this ..
 385                        (let ((prev-prev-slash
 386                               (position-if #'slashp dst :end last-slash :from-end t)))
 387                          (cond ((and (= (+ (or prev-prev-slash 0) 2)
 388                                         last-slash)
 389                                      (char= (schar dst (- last-slash 2)) #\.)
 390                                      (char= (schar dst (1- last-slash)) #\.))
 391                                 ;; The something before this .. is another ..
 392                                 (deposit char)
 393                                 (setf last-slash dst-len))
 394                                (t
 395                                 ;; The something is some directory or other.
 396                                 (setf dst-len
 397                                       (if prev-prev-slash
 398                                           (1+ prev-prev-slash)
 399                                           0))
 400                                 (setf last-slash prev-prev-slash)))))
 401                       (t
 402                        ;; There is nothing before this .., so we need to keep it
 403                        (setf last-slash dst-len)
 404                        (deposit char))))
 405                    (t
 406                     ;; something other than a dot between slashes
 407                     (setf last-slash dst-len)
 408                     (deposit char)))
 409                  (setf dots 0))
 410                 (t
 411                  (setf dots nil)
 412                  (setf (schar dst dst-len) char)
 413                  (incf dst-len)))))
 414       ;; ...finish off
 415       (when (and last-slash (not (zerop last-slash)))
 416         (case dots
 417           (1
 418            ;; We've got  ``foobar/.''
 419            (decf dst-len))
 420           (2
 421            ;; We've got ``foobar/..''
 422            (unless (and (>= last-slash 2)
 423                         (char= (schar dst (1- last-slash)) #\.)
 424                         (char= (schar dst (- last-slash 2)) #\.)
 425                         (or (= last-slash 2)
 426                             (slashp (schar dst (- last-slash 3)))))
 427              (let ((prev-prev-slash
 428                     (position-if #'slashp dst :end last-slash :from-end t)))
 429                (if prev-prev-slash
 430                    (setf dst-len (1+ prev-prev-slash))
 431                    (return-from simplify-win32-namestring
 432                      (coerce ".\\" 'simple-string)))))))))
 433     (cond ((zerop dst-len)
 434            ".\\")
 435           ((= dst-len src-len)
 436            dst)
 437           (t
 438            (subseq dst 0 dst-len)))))