src/code/reader.lisp

   1 ;;;; READ and friends
   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 \f
  14 ;;;; miscellaneous global variables
  15
  16 ;;; ANSI: "the floating-point format that is to be used when reading a
  17 ;;; floating-point number that has no exponent marker or that has e or
  18 ;;; E for an exponent marker"
  19 (defvar *read-default-float-format* 'single-float)
  20 (declaim (type (member short-float single-float double-float long-float)
  21                *read-default-float-format*))
  22
  23 (defvar *readtable*)
  24 (declaim (type readtable *readtable*))
  25 #!+sb-doc
  26 (setf (fdocumentation '*readtable* 'variable)
  27        "Variable bound to current readtable.")
  28
  29 ;;; a standard Lisp readtable. This is for recovery from broken
  30 ;;; read-tables (and for WITH-STANDARD-IO-SYNTAX), and should not
  31 ;;; normally be user-visible.
  32 (defvar *standard-readtable*)
  33
  34 (defvar *old-package* nil
  35   #!+sb-doc
  36   "the value of *PACKAGE* at the start of the last read, or NIL")
  37
  38 ;;; In case we get an error trying to parse a symbol, we want to rebind the
  39 ;;; above stuff so it's cool.
  40
  41 ;;; FIXME: These forward declarations should be moved somewhere earlier,
  42 ;;; or discarded.
  43 (declaim (special *package* *keyword-package* *read-base*))
  44 \f
  45 ;;;; reader errors
  46
  47 (defun reader-eof-error (stream context)
  48   (error 'reader-eof-error
  49          :stream stream
  50          :context context))
  51
  52 (defun %reader-error (stream control &rest args)
  53   (error 'reader-error
  54          :stream stream
  55          :format-control control
  56          :format-arguments args))
  57 \f
  58 ;;;; macros and functions for character tables
  59
  60 ;;; FIXME: could be SB!XC:DEFMACRO inside EVAL-WHEN (COMPILE EVAL)
  61 (defmacro get-cat-entry (char rt)
  62   ;; KLUDGE: Only give this side-effect-free args.
  63   ;; FIXME: should probably become inline function
  64   `(elt (character-attribute-table ,rt)
  65         (char-code ,char)))
  66
  67 (defun set-cat-entry (char newvalue &optional (rt *readtable*))
  68   (setf (elt (character-attribute-table rt)
  69              (char-code char))
  70         newvalue))
  71
  72 ;;; the value actually stored in the character macro table. As per
  73 ;;; ANSI #'GET-MACRO-CHARACTER and #'SET-MACRO-CHARACTER, this can
  74 ;;; be either a function or NIL.
  75 (eval-when (:compile-toplevel :execute)
  76   (sb!xc:defmacro get-raw-cmt-entry (char readtable)
  77     `(svref (character-macro-table ,readtable)
  78             (char-code ,char))))
  79
  80 ;;; the value represented by whatever is stored in the character macro
  81 ;;; table. As per ANSI #'GET-MACRO-CHARACTER and #'SET-MACRO-CHARACTER,
  82 ;;; a function value represents itself, and a NIL value represents the
  83 ;;; default behavior.
  84 (defun get-coerced-cmt-entry (char readtable)
  85   (the function
  86     (or (get-raw-cmt-entry char readtable)
  87         #'read-token)))
  88
  89 (defun set-cmt-entry (char new-value-designator &optional (rt *readtable*))
  90   (setf (svref (character-macro-table rt)
  91                (char-code char))
  92         (and new-value-designator
  93              (%coerce-callable-to-fun new-value-designator))))
  94
  95 (defun undefined-macro-char (stream char)
  96   (unless *read-suppress*
  97     (%reader-error stream "undefined read-macro character ~S" char)))
  98
  99 ;;; The character attribute table is a CHAR-CODE-LIMIT vector of integers.
 100
 101 (defmacro test-attribute (char whichclass rt)
 102   `(= (the fixnum (get-cat-entry ,char ,rt)) ,whichclass))
 103
 104 ;;; predicates for testing character attributes
 105
 106 #!-sb-fluid (declaim (inline whitespacep))
 107 (defun whitespacep (char &optional (rt *readtable*))
 108   (test-attribute char +char-attr-whitespace+ rt))
 109
 110 (defmacro constituentp (char &optional (rt '*readtable*))
 111   `(>= (get-cat-entry ,char ,rt) +char-attr-constituent+))
 112
 113 (defmacro terminating-macrop (char &optional (rt '*readtable*))
 114   `(test-attribute ,char +char-attr-terminating-macro+ ,rt))
 115
 116 (defmacro escapep (char &optional (rt '*readtable*))
 117   `(test-attribute ,char +char-attr-escape+ ,rt))
 118
 119 (defmacro multiple-escape-p (char &optional (rt '*readtable*))
 120   `(test-attribute ,char +char-attr-multiple-escape+ ,rt))
 121
 122 (defmacro token-delimiterp (char &optional (rt '*readtable*))
 123   ;; depends on actual attribute numbering above.
 124   `(<= (get-cat-entry ,char ,rt) +char-attr-terminating-macro+))
 125 \f
 126 ;;;; secondary attribute table
 127
 128 ;;; There are a number of "secondary" attributes which are constant
 129 ;;; properties of characters (as long as they are constituents).
 130
 131 (defvar *secondary-attribute-table*)
 132 (declaim (type attribute-table *secondary-attribute-table*))
 133
 134 (defun !set-secondary-attribute (char attribute)
 135   (setf (elt *secondary-attribute-table* (char-code char))
 136         attribute))
 137
 138 (defun !cold-init-secondary-attribute-table ()
 139   (setq *secondary-attribute-table*
 140         (make-array char-code-limit :element-type '(unsigned-byte 8)
 141                     :initial-element +char-attr-constituent+))
 142   (!set-secondary-attribute #\: +char-attr-package-delimiter+)
 143   (!set-secondary-attribute #\| +char-attr-multiple-escape+) ; |) [for EMACS]
 144   (!set-secondary-attribute #\. +char-attr-constituent-dot+)
 145   (!set-secondary-attribute #\+ +char-attr-constituent-sign+)
 146   (!set-secondary-attribute #\- +char-attr-constituent-sign+)
 147   (!set-secondary-attribute #\/ +char-attr-constituent-slash+)
 148   (do ((i (char-code #\0) (1+ i)))
 149       ((> i (char-code #\9)))
 150     (!set-secondary-attribute (code-char i) +char-attr-constituent-digit+))
 151   (!set-secondary-attribute #\E +char-attr-constituent-expt+)
 152   (!set-secondary-attribute #\F +char-attr-constituent-expt+)
 153   (!set-secondary-attribute #\D +char-attr-constituent-expt+)
 154   (!set-secondary-attribute #\S +char-attr-constituent-expt+)
 155   (!set-secondary-attribute #\L +char-attr-constituent-expt+)
 156   (!set-secondary-attribute #\e +char-attr-constituent-expt+)
 157   (!set-secondary-attribute #\f +char-attr-constituent-expt+)
 158   (!set-secondary-attribute #\d +char-attr-constituent-expt+)
 159   (!set-secondary-attribute #\s +char-attr-constituent-expt+)
 160   (!set-secondary-attribute #\l +char-attr-constituent-expt+))
 161
 162 (defmacro get-secondary-attribute (char)
 163   `(elt *secondary-attribute-table*
 164         (char-code ,char)))
 165 \f
 166 ;;;; readtable operations
 167
 168 (defun copy-readtable (&optional (from-readtable *readtable*)
 169                                  to-readtable)
 170   (let ((really-from-readtable (or from-readtable *standard-readtable*))
 171         (really-to-readtable (or to-readtable (make-readtable))))
 172     (replace (character-attribute-table really-to-readtable)
 173              (character-attribute-table really-from-readtable))
 174     (replace (character-macro-table really-to-readtable)
 175              (character-macro-table really-from-readtable))
 176     (setf (dispatch-tables really-to-readtable)
 177           (mapcar (lambda (pair) (cons (car pair)
 178                                        (copy-seq (cdr pair))))
 179                   (dispatch-tables really-from-readtable)))
 180     (setf (readtable-case really-to-readtable)
 181           (readtable-case really-from-readtable))
 182     really-to-readtable))
 183
 184 (defun set-syntax-from-char (to-char from-char &optional
 185                                      (to-readtable *readtable*)
 186                                      (from-readtable ()))
 187   #!+sb-doc
 188   "Causes the syntax of TO-CHAR to be the same as FROM-CHAR in the
 189   optional readtable (defaults to the current readtable). The
 190   FROM-TABLE defaults to the standard Lisp readtable when NIL."
 191   (let ((really-from-readtable (or from-readtable *standard-readtable*)))
 192     ;; Copy FROM-CHAR entries to TO-CHAR entries, but make sure that if
 193     ;; FROM-CHAR is a constituent you don't copy non-movable secondary
 194     ;; attributes (constituent types), and that said attributes magically
 195     ;; appear if you transform a non-constituent to a constituent.
 196     (let ((att (get-cat-entry from-char really-from-readtable)))
 197       (if (constituentp from-char really-from-readtable)
 198           (setq att (get-secondary-attribute to-char)))
 199       (set-cat-entry to-char att to-readtable)
 200       (set-cmt-entry to-char
 201                      (get-raw-cmt-entry from-char really-from-readtable)
 202                      to-readtable)))
 203   t)
 204
 205 (defun set-macro-character (char function &optional
 206                                  (non-terminatingp nil)
 207                                  (readtable *readtable*))
 208   #!+sb-doc
 209   "Causes CHAR to be a macro character which invokes FUNCTION when seen
 210    by the reader. The NON-TERMINATINGP flag can be used to make the macro
 211    character non-terminating, i.e. embeddable in a symbol name."
 212   (let ((designated-readtable (or readtable *standard-readtable*)))
 213     (set-cat-entry char
 214                    (if non-terminatingp
 215                        (get-secondary-attribute char)
 216                        +char-attr-terminating-macro+)
 217                    designated-readtable)
 218     (set-cmt-entry char function designated-readtable)
 219     t)) ; (ANSI-specified return value)
 220
 221 (defun get-macro-character (char &optional (readtable *readtable*))
 222   #!+sb-doc
 223   "Return the function associated with the specified CHAR which is a macro
 224   character, or NIL if there is no such function. As a second value, return
 225   T if CHAR is a macro character which is non-terminating, i.e. which can
 226   be embedded in a symbol name."
 227   (let* ((designated-readtable (or readtable *standard-readtable*))
 228          ;; the first return value: a FUNCTION if CHAR is a macro
 229          ;; character, or NIL otherwise
 230          (fun-value (get-raw-cmt-entry char designated-readtable)))
 231     (values fun-value
 232             ;; NON-TERMINATING-P return value:
 233             (if fun-value
 234                 (or (constituentp char)
 235                     (not (terminating-macrop char)))
 236                 ;; ANSI's definition of GET-MACRO-CHARACTER says this
 237                 ;; value is NIL when CHAR is not a macro character.
 238                 ;; I.e. this value means not just "non-terminating
 239                 ;; character?" but "non-terminating macro character?".
 240                 nil))))
 241 \f
 242 ;;;; definitions to support internal programming conventions
 243
 244 (defmacro eofp (char)
 245   `(eq ,char *eof-object*))
 246
 247 (defun flush-whitespace (stream)
 248   ;; This flushes whitespace chars, returning the last char it read (a
 249   ;; non-white one). It always gets an error on end-of-file.
 250   (let ((stream (in-synonym-of stream)))
 251     (if (ansi-stream-p stream)
 252         (prepare-for-fast-read-char stream
 253           (do ((attribute-table (character-attribute-table *readtable*))
 254                (char (fast-read-char t) (fast-read-char t)))
 255               ((/= (the fixnum (aref attribute-table (char-code char)))
 256                    +char-attr-whitespace+)
 257                (done-with-fast-read-char)
 258                char)))
 259         ;; fundamental-stream
 260         (do ((attribute-table (character-attribute-table *readtable*))
 261              (char (stream-read-char stream) (stream-read-char stream)))
 262             ((or (eq char :eof)
 263                  (/= (the fixnum (aref attribute-table (char-code char)))
 264                      +char-attr-whitespace+))
 265              (if (eq char :eof)
 266                  (error 'end-of-file :stream stream)
 267                  char))))))
 268 \f
 269 ;;;; temporary initialization hack
 270
 271 (defun !cold-init-standard-readtable ()
 272   (setq *standard-readtable* (make-readtable))
 273   ;; All characters get boring defaults in MAKE-READTABLE. Now we
 274   ;; override the boring defaults on characters which need more
 275   ;; interesting behavior.
 276   (let ((*readtable* *standard-readtable*))
 277
 278     (flet ((whitespaceify (char)
 279              (set-cat-entry char +char-attr-whitespace+)))
 280       (whitespaceify (code-char tab-char-code))
 281       (whitespaceify #\linefeed)
 282       (whitespaceify #\space)
 283       (whitespaceify (code-char form-feed-char-code))
 284       (whitespaceify (code-char return-char-code)))
 285
 286     (set-cat-entry #\\ +char-attr-escape+)
 287     (set-cmt-entry #\\ #'read-token)
 288
 289     ;; Easy macro-character definitions are in this source file.
 290     (set-macro-character #\" #'read-string)
 291     (set-macro-character #\' #'read-quote)
 292     (set-macro-character #\( #'read-list)
 293     (set-macro-character #\) #'read-right-paren)
 294     (set-macro-character #\; #'read-comment)
 295     ;; (The hairier macro-character definitions, for #\# and #\`, are
 296     ;; defined elsewhere, in their own source files.)
 297
 298     ;; all constituents
 299     (do ((ichar 0 (1+ ichar))
 300          (char))
 301         ((= ichar #O200))
 302       (setq char (code-char ichar))
 303       (when (constituentp char *standard-readtable*)
 304             (set-cat-entry char (get-secondary-attribute char))
 305             (set-cmt-entry char nil)))))
 306 \f
 307 ;;;; implementation of the read buffer
 308
 309 (defvar *read-buffer*)
 310 (defvar *read-buffer-length*)
 311 ;;; FIXME: Is it really helpful to have *READ-BUFFER-LENGTH* be a
 312 ;;; separate variable instead of just calculating it on the fly as
 313 ;;; (LENGTH *READ-BUFFER*)?
 314
 315 (defvar *inch-ptr*)
 316 (defvar *ouch-ptr*)
 317
 318 (declaim (type index *read-buffer-length* *inch-ptr* *ouch-ptr*))
 319 (declaim (simple-string *read-buffer*))
 320
 321 (defmacro reset-read-buffer ()
 322   ;; Turn *READ-BUFFER* into an empty read buffer.
 323   `(progn
 324      ;; *OUCH-PTR* always points to next char to write.
 325      (setq *ouch-ptr* 0)
 326      ;; *INCH-PTR* always points to next char to read.
 327      (setq *inch-ptr* 0)))
 328
 329 (defun !cold-init-read-buffer ()
 330   (setq *read-buffer* (make-string 512)) ; initial bufsize
 331   (setq *read-buffer-length* 512)
 332   (reset-read-buffer))
 333
 334 ;;; FIXME I removed "THE FIXNUM"'s from OUCH-READ-BUFFER and
 335 ;;; OUCH-UNREAD-BUFFER, check to make sure that Python really is smart
 336 ;;; enough to make good code without them. And while I'm at it,
 337 ;;; converting them from macros to inline functions might be good,
 338 ;;; too.
 339
 340 (defmacro ouch-read-buffer (char)
 341   `(progn
 342      ;; When buffer overflow
 343      (when (>= *ouch-ptr* *read-buffer-length*)
 344        ;; Size should be doubled.
 345        (grow-read-buffer))
 346      (setf (elt (the simple-string *read-buffer*) *ouch-ptr*) ,char)
 347      (setq *ouch-ptr* (1+ *ouch-ptr*))))
 348
 349 ;;; macro to move *ouch-ptr* back one.
 350 (defmacro ouch-unread-buffer ()
 351   '(when (> *ouch-ptr* *inch-ptr*)
 352      (setq *ouch-ptr* (1- (the fixnum *ouch-ptr*)))))
 353
 354 (defun grow-read-buffer ()
 355   (let ((rbl (length (the simple-string *read-buffer*))))
 356     (setq *read-buffer*
 357           (concatenate 'simple-string
 358                        *read-buffer*
 359                        (make-string rbl)))
 360     (setq *read-buffer-length* (* 2 rbl))))
 361
 362 (defun inchpeek-read-buffer ()
 363   (if (>= (the fixnum *inch-ptr*) (the fixnum *ouch-ptr*))
 364       *eof-object*
 365       (elt *read-buffer* *inch-ptr*)))
 366
 367 (defun inch-read-buffer ()
 368   (if (>= *inch-ptr* *ouch-ptr*)
 369       *eof-object*
 370       (prog1
 371           (elt *read-buffer* *inch-ptr*)
 372         (incf *inch-ptr*))))
 373
 374 (defmacro unread-buffer ()
 375   `(decf *inch-ptr*))
 376
 377 (defun read-unwind-read-buffer ()
 378   ;; Keep contents, but make next (INCH..) return first character.
 379   (setq *inch-ptr* 0))
 380
 381 (defun read-buffer-to-string ()
 382   (subseq *read-buffer* 0 *ouch-ptr*))
 383 \f
 384 ;;;; READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST, and READ
 385
 386 ;;; an alist for #=, used to keep track of objects with labels assigned that
 387 ;;; have been completely read. Each entry is (integer-tag gensym-tag value).
 388 ;;;
 389 ;;; KLUDGE: Should this really be an alist? It seems as though users
 390 ;;; could reasonably expect N log N performance for large datasets.
 391 ;;; On the other hand, it's probably very very seldom a problem in practice.
 392 ;;; On the third hand, it might be just as easy to use a hash table
 393 ;;; as an alist, so maybe we should. -- WHN 19991202
 394 (defvar *sharp-equal-alist* ())
 395
 396 (declaim (special *standard-input*))
 397
 398 ;;; READ-PRESERVING-WHITESPACE behaves just like READ, only it makes
 399 ;;; sure to leave terminating whitespace in the stream. (This is a
 400 ;;; COMMON-LISP exported symbol.)
 401 (defun read-preserving-whitespace (&optional (stream *standard-input*)
 402                                              (eof-error-p t)
 403                                              (eof-value nil)
 404                                              (recursivep nil))
 405   #!+sb-doc
 406   "Read from STREAM and return the value read, preserving any whitespace
 407    that followed the object."
 408   (if recursivep
 409     ;; a loop for repeating when a macro returns nothing
 410     (loop
 411       (let ((char (read-char stream eof-error-p *eof-object*)))
 412         (cond ((eofp char) (return eof-value))
 413               ((whitespacep char))
 414               (t
 415                (let* ((macrofun (get-coerced-cmt-entry char *readtable*))
 416                       (result (multiple-value-list
 417                                (funcall macrofun stream char))))
 418                  ;; Repeat if macro returned nothing.
 419                   (if result (return (car result))))))))
 420     (let ((*sharp-equal-alist* nil))
 421         (read-preserving-whitespace stream eof-error-p eof-value t))))
 422
 423 ;;; Return NIL or a list with one thing, depending.
 424 ;;;
 425 ;;; for functions that want comments to return so that they can look
 426 ;;; past them. We assume CHAR is not whitespace.
 427 (defun read-maybe-nothing (stream char)
 428   (let ((retval (multiple-value-list
 429                  (funcall (get-coerced-cmt-entry char *readtable*)
 430                           stream
 431                           char))))
 432     (if retval (rplacd retval nil))))
 433
 434 (defun read (&optional (stream *standard-input*)
 435                        (eof-error-p t)
 436                        (eof-value ())
 437                        (recursivep ()))
 438   #!+sb-doc
 439   "Read the next Lisp value from STREAM, and return it."
 440   (let ((result (read-preserving-whitespace stream
 441                                             eof-error-p
 442                                             eof-value
 443                                             recursivep)))
 444     ;; (This function generally discards trailing whitespace. If you
 445     ;; don't want to discard trailing whitespace, call
 446     ;; CL:READ-PRESERVING-WHITESPACE instead.)
 447     (unless (or (eql result eof-value) recursivep)
 448       (let ((next-char (read-char stream nil nil)))
 449         (unless (or (null next-char)
 450                     (whitespacep next-char))
 451           (unread-char next-char stream))))
 452     result))
 453
 454 ;;; (This is a COMMON-LISP exported symbol.)
 455 (defun read-delimited-list (endchar &optional
 456                                     (input-stream *standard-input*)
 457                                     recursive-p)
 458   #!+sb-doc
 459   "Read Lisp values from INPUT-STREAM until the next character after a
 460    value's representation is ENDCHAR, and return the objects as a list."
 461   (declare (ignore recursive-p))
 462   (do ((char (flush-whitespace input-stream)
 463              (flush-whitespace input-stream))
 464        (retlist ()))
 465       ((char= char endchar) (nreverse retlist))
 466     (setq retlist (nconc (read-maybe-nothing input-stream char) retlist))))
 467 \f
 468 ;;;; basic readmacro definitions
 469 ;;;;
 470 ;;;; Some large, hairy subsets of readmacro definitions (backquotes
 471 ;;;; and sharp macros) are not here, but in their own source files.
 472
 473 (defun read-quote (stream ignore)
 474   (declare (ignore ignore))
 475   (list 'quote (read stream t nil t)))
 476
 477 (defun read-comment (stream ignore)
 478   (declare (ignore ignore))
 479   (let ((stream (in-synonym-of stream)))
 480     (if (ansi-stream-p stream)
 481         (prepare-for-fast-read-char stream
 482           (do ((char (fast-read-char nil nil)
 483                      (fast-read-char nil nil)))
 484               ((or (not char) (char= char #\newline))
 485                (done-with-fast-read-char))))
 486         ;; FUNDAMENTAL-STREAM
 487         (do ((char (stream-read-char stream) (stream-read-char stream)))
 488             ((or (eq char :eof) (char= char #\newline))))))
 489   ;; Don't return anything.
 490   (values))
 491
 492 (defun read-list (stream ignore)
 493   (declare (ignore ignore))
 494   (let* ((thelist (list nil))
 495          (listtail thelist))
 496     (do ((firstchar (flush-whitespace stream) (flush-whitespace stream)))
 497         ((char= firstchar #\) ) (cdr thelist))
 498       (when (char= firstchar #\.)
 499             (let ((nextchar (read-char stream t)))
 500               (cond ((token-delimiterp nextchar)
 501                      (cond ((eq listtail thelist)
 502                             (%reader-error
 503                              stream
 504                              "Nothing appears before . in list."))
 505                            ((whitespacep nextchar)
 506                             (setq nextchar (flush-whitespace stream))))
 507                      (rplacd listtail
 508                              ;; Return list containing last thing.
 509                              (car (read-after-dot stream nextchar)))
 510                      (return (cdr thelist)))
 511                     ;; Put back NEXTCHAR so that we can read it normally.
 512                     (t (unread-char nextchar stream)))))
 513       ;; Next thing is not an isolated dot.
 514       (let ((listobj (read-maybe-nothing stream firstchar)))
 515         ;; allows the possibility that a comment was read
 516         (when listobj
 517               (rplacd listtail listobj)
 518               (setq listtail listobj))))))
 519
 520 (defun read-after-dot (stream firstchar)
 521   ;; FIRSTCHAR is non-whitespace!
 522   (let ((lastobj ()))
 523     (do ((char firstchar (flush-whitespace stream)))
 524         ((char= char #\) )
 525          (%reader-error stream "Nothing appears after . in list."))
 526       ;; See whether there's something there.
 527       (setq lastobj (read-maybe-nothing stream char))
 528       (when lastobj (return t)))
 529     ;; At least one thing appears after the dot.
 530     ;; Check for more than one thing following dot.
 531     (do ((lastchar (flush-whitespace stream)
 532                    (flush-whitespace stream)))
 533         ((char= lastchar #\) ) lastobj) ;success!
 534       ;; Try reading virtual whitespace.
 535       (if (read-maybe-nothing stream lastchar)
 536           (%reader-error stream "More than one object follows . in list.")))))
 537
 538 (defun read-string (stream closech)
 539   ;; This accumulates chars until it sees same char that invoked it.
 540   ;; For a very long string, this could end up bloating the read buffer.
 541   (reset-read-buffer)
 542   (let ((stream (in-synonym-of stream)))
 543     (if (ansi-stream-p stream)
 544         (prepare-for-fast-read-char stream
 545           (do ((char (fast-read-char t) (fast-read-char t)))
 546               ((char= char closech)
 547                (done-with-fast-read-char))
 548             (if (escapep char) (setq char (fast-read-char t)))
 549             (ouch-read-buffer char)))
 550         ;; FUNDAMENTAL-STREAM
 551         (do ((char (stream-read-char stream) (stream-read-char stream)))
 552             ((or (eq char :eof) (char= char closech))
 553              (if (eq char :eof)
 554                  (error 'end-of-file :stream stream)))
 555           (when (escapep char)
 556             (setq char (stream-read-char stream))
 557             (if (eq char :eof)
 558                 (error 'end-of-file :stream stream)))
 559           (ouch-read-buffer char))))
 560   (read-buffer-to-string))
 561
 562 (defun read-right-paren (stream ignore)
 563   (declare (ignore ignore))
 564   (%reader-error stream "unmatched close parenthesis"))
 565
 566 ;;; Read from the stream up to the next delimiter. Leave the resulting
 567 ;;; token in *READ-BUFFER*, and return two values:
 568 ;;; -- a list of the escaped character positions, and
 569 ;;; -- The position of the first package delimiter (or NIL).
 570 (defun internal-read-extended-token (stream firstchar escape-firstchar)
 571   (reset-read-buffer)
 572   (let ((escapes '()))
 573     (when escape-firstchar
 574       (push *ouch-ptr* escapes)
 575       (ouch-read-buffer firstchar)
 576       (setq firstchar (read-char stream nil *eof-object*)))
 577   (do ((char firstchar (read-char stream nil *eof-object*))
 578        (colon nil))
 579       ((cond ((eofp char) t)
 580              ((token-delimiterp char)
 581               (unread-char char stream)
 582               t)
 583              (t nil))
 584        (values escapes colon))
 585     (cond ((escapep char)
 586            ;; It can't be a number, even if it's 1\23.
 587            ;; Read next char here, so it won't be casified.
 588            (push *ouch-ptr* escapes)
 589            (let ((nextchar (read-char stream nil *eof-object*)))
 590              (if (eofp nextchar)
 591                  (reader-eof-error stream "after escape character")
 592                  (ouch-read-buffer nextchar))))
 593           ((multiple-escape-p char)
 594            ;; Read to next multiple-escape, escaping single chars
 595            ;; along the way.
 596            (loop
 597              (let ((ch (read-char stream nil *eof-object*)))
 598                (cond
 599                 ((eofp ch)
 600                  (reader-eof-error stream "inside extended token"))
 601                 ((multiple-escape-p ch) (return))
 602                 ((escapep ch)
 603                  (let ((nextchar (read-char stream nil *eof-object*)))
 604                    (cond ((eofp nextchar)
 605                           (reader-eof-error stream "after escape character"))
 606                          (t
 607                           (push *ouch-ptr* escapes)
 608                           (ouch-read-buffer nextchar)))))
 609                 (t
 610                  (push *ouch-ptr* escapes)
 611                  (ouch-read-buffer ch))))))
 612           (t
 613            (when (and (constituentp char)
 614                         (eql (get-secondary-attribute char)
 615                              +char-attr-package-delimiter+)
 616                       (not colon))
 617              (setq colon *ouch-ptr*))
 618            (ouch-read-buffer char))))))
 619 \f
 620 ;;;; character classes
 621
 622 ;;; Return the character class for CHAR.
 623 (defmacro char-class (char attable)
 624   `(let ((att (aref ,attable (char-code ,char))))
 625      (declare (fixnum att))
 626      (if (<= att +char-attr-terminating-macro+)
 627          +char-attr-delimiter+
 628          att)))
 629
 630 ;;; Return the character class for CHAR, which might be part of a
 631 ;;; rational number.
 632 (defmacro char-class2 (char attable)
 633   `(let ((att (aref ,attable (char-code ,char))))
 634      (declare (fixnum att))
 635      (if (<= att +char-attr-terminating-macro+)
 636          +char-attr-delimiter+
 637          (if (digit-char-p ,char *read-base*)
 638              +char-attr-constituent-digit+
 639              (if (= att +char-attr-constituent-digit+)
 640                  +char-attr-constituent+
 641                  att)))))
 642
 643 ;;; Return the character class for a char which might be part of a
 644 ;;; rational or floating number. (Assume that it is a digit if it
 645 ;;; could be.)
 646 (defmacro char-class3 (char attable)
 647   `(let ((att (aref ,attable (char-code ,char))))
 648      (declare (fixnum att))
 649      (if possibly-rational
 650          (setq possibly-rational
 651                (or (digit-char-p ,char *read-base*)
 652                    (= att +char-attr-constituent-slash+))))
 653      (if possibly-float
 654          (setq possibly-float
 655                (or (digit-char-p ,char 10)
 656                    (= att +char-attr-constituent-dot+))))
 657      (if (<= att +char-attr-terminating-macro+)
 658          +char-attr-delimiter+
 659          (if (digit-char-p ,char (max *read-base* 10))
 660              (if (digit-char-p ,char *read-base*)
 661                  +char-attr-constituent-digit+
 662                  +char-attr-constituent+)
 663              att))))
 664 \f
 665 ;;;; token fetching
 666
 667 (defvar *read-suppress* nil
 668   #!+sb-doc
 669   "Suppress most interpreting in the reader when T.")
 670
 671 (defvar *read-base* 10
 672   #!+sb-doc
 673   "the radix that Lisp reads numbers in")
 674 (declaim (type (integer 2 36) *read-base*))
 675
 676 ;;; Modify the read buffer according to READTABLE-CASE, ignoring
 677 ;;; ESCAPES. ESCAPES is a list of the escaped indices, in reverse
 678 ;;; order.
 679 (defun casify-read-buffer (escapes)
 680   (let ((case (readtable-case *readtable*)))
 681     (cond
 682      ((and (null escapes) (eq case :upcase))
 683       (dotimes (i *ouch-ptr*)
 684         (setf (schar *read-buffer* i)
 685               (char-upcase (schar *read-buffer* i)))))
 686      ((eq case :preserve))
 687      (t
 688       (macrolet ((skip-esc (&body body)
 689                    `(do ((i (1- *ouch-ptr*) (1- i))
 690                          (escapes escapes))
 691                         ((minusp i))
 692                       (declare (fixnum i))
 693                       (when (or (null escapes)
 694                                 (let ((esc (first escapes)))
 695                                   (declare (fixnum esc))
 696                                   (cond ((< esc i) t)
 697                                         (t
 698                                          (aver (= esc i))
 699                                          (pop escapes)
 700                                          nil))))
 701                         (let ((ch (schar *read-buffer* i)))
 702                           ,@body)))))
 703         (flet ((lower-em ()
 704                  (skip-esc (setf (schar *read-buffer* i) (char-downcase ch))))
 705                (raise-em ()
 706                  (skip-esc (setf (schar *read-buffer* i) (char-upcase ch)))))
 707           (ecase case
 708             (:upcase (raise-em))
 709             (:downcase (lower-em))
 710             (:invert
 711              (let ((all-upper t)
 712                    (all-lower t))
 713                (skip-esc
 714                  (when (both-case-p ch)
 715                    (if (upper-case-p ch)
 716                        (setq all-lower nil)
 717                        (setq all-upper nil))))
 718                (cond (all-lower (raise-em))
 719                      (all-upper (lower-em))))))))))))
 720
 721 (defun read-token (stream firstchar)
 722   #!+sb-doc
 723   "This function is just an fsm that recognizes numbers and symbols."
 724   ;; Check explicitly whether FIRSTCHAR has an entry for
 725   ;; NON-TERMINATING in CHARACTER-ATTRIBUTE-TABLE and
 726   ;; READ-DOT-NUMBER-SYMBOL in CMT. Report an error if these are
 727   ;; violated. (If we called this, we want something that is a
 728   ;; legitimate token!) Read in the longest possible string satisfying
 729   ;; the Backus-Naur form for "unqualified-token". Leave the result in
 730   ;; the *READ-BUFFER*. Return next char after token (last char read).
 731   (when *read-suppress*
 732     (internal-read-extended-token stream firstchar nil)
 733     (return-from read-token nil))
 734   (let ((attribute-table (character-attribute-table *readtable*))
 735         (package-designator nil)
 736         (colons 0)
 737         (possibly-rational t)
 738         (possibly-float t)
 739         (escapes ()))
 740     (reset-read-buffer)
 741     (prog ((char firstchar))
 742       (case (char-class3 char attribute-table)
 743         (#.+char-attr-constituent-sign+ (go SIGN))
 744         (#.+char-attr-constituent-digit+ (go LEFTDIGIT))
 745         (#.+char-attr-constituent-dot+ (go FRONTDOT))
 746         (#.+char-attr-escape+ (go ESCAPE))
 747         (#.+char-attr-package-delimiter+ (go COLON))
 748         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 749         ;; can't have eof, whitespace, or terminating macro as first char!
 750         (t (go SYMBOL)))
 751      SIGN ; saw "sign"
 752       (ouch-read-buffer char)
 753       (setq char (read-char stream nil nil))
 754       (unless char (go RETURN-SYMBOL))
 755       (setq possibly-rational t
 756             possibly-float t)
 757       (case (char-class3 char attribute-table)
 758         (#.+char-attr-constituent-digit+ (go LEFTDIGIT))
 759         (#.+char-attr-constituent-dot+ (go SIGNDOT))
 760         (#.+char-attr-escape+ (go ESCAPE))
 761         (#.+char-attr-package-delimiter+ (go COLON))
 762         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 763         (#.+char-attr-delimiter+ (unread-char char stream) (go RETURN-SYMBOL))
 764         (t (go SYMBOL)))
 765      LEFTDIGIT ; saw "[sign] {digit}+"
 766       (ouch-read-buffer char)
 767       (setq char (read-char stream nil nil))
 768       (unless char (return (make-integer)))
 769       (case (char-class3 char attribute-table)
 770         (#.+char-attr-constituent-digit+ (go LEFTDIGIT))
 771         (#.+char-attr-constituent-dot+ (if possibly-float
 772                                            (go MIDDLEDOT)
 773                                            (go SYMBOL)))
 774         (#.+char-attr-constituent-expt+ (go EXPONENT))
 775         (#.+char-attr-constituent-slash+ (if possibly-rational
 776                                              (go RATIO)
 777                                              (go SYMBOL)))
 778         (#.+char-attr-delimiter+ (unread-char char stream)
 779                                  (return (make-integer)))
 780         (#.+char-attr-escape+ (go ESCAPE))
 781         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 782         (#.+char-attr-package-delimiter+ (go COLON))
 783         (t (go SYMBOL)))
 784      MIDDLEDOT ; saw "[sign] {digit}+ dot"
 785       (ouch-read-buffer char)
 786       (setq char (read-char stream nil nil))
 787       (unless char (return (let ((*read-base* 10))
 788                              (make-integer))))
 789       (case (char-class char attribute-table)
 790         (#.+char-attr-constituent-digit+ (go RIGHTDIGIT))
 791         (#.+char-attr-constituent-expt+ (go EXPONENT))
 792         (#.+char-attr-delimiter+
 793          (unread-char char stream)
 794          (return (let ((*read-base* 10))
 795                    (make-integer))))
 796         (#.+char-attr-escape+ (go ESCAPE))
 797         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 798         (#.+char-attr-package-delimiter+ (go COLON))
 799         (t (go SYMBOL)))
 800      RIGHTDIGIT ; saw "[sign] {digit}* dot {digit}+"
 801       (ouch-read-buffer char)
 802       (setq char (read-char stream nil nil))
 803       (unless char (return (make-float stream)))
 804       (case (char-class char attribute-table)
 805         (#.+char-attr-constituent-digit+ (go RIGHTDIGIT))
 806         (#.+char-attr-constituent-expt+ (go EXPONENT))
 807         (#.+char-attr-delimiter+
 808          (unread-char char stream)
 809          (return (make-float stream)))
 810         (#.+char-attr-escape+ (go ESCAPE))
 811         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 812         (#.+char-attr-package-delimiter+ (go COLON))
 813         (t (go SYMBOL)))
 814      SIGNDOT ; saw "[sign] dot"
 815       (ouch-read-buffer char)
 816       (setq char (read-char stream nil nil))
 817       (unless char (go RETURN-SYMBOL))
 818       (case (char-class char attribute-table)
 819         (#.+char-attr-constituent-digit+ (go RIGHTDIGIT))
 820         (#.+char-attr-delimiter+ (unread-char char stream) (go RETURN-SYMBOL))
 821         (#.+char-attr-escape+ (go ESCAPE))
 822         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 823         (t (go SYMBOL)))
 824      FRONTDOT ; saw "dot"
 825       (ouch-read-buffer char)
 826       (setq char (read-char stream nil nil))
 827       (unless char (%reader-error stream "dot context error"))
 828       (case (char-class char attribute-table)
 829         (#.+char-attr-constituent-digit+ (go RIGHTDIGIT))
 830         (#.+char-attr-constituent-dot+ (go DOTS))
 831         (#.+char-attr-delimiter+  (%reader-error stream "dot context error"))
 832         (#.+char-attr-escape+ (go ESCAPE))
 833         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 834         (#.+char-attr-package-delimiter+ (go COLON))
 835         (t (go SYMBOL)))
 836      EXPONENT
 837       (ouch-read-buffer char)
 838       (setq char (read-char stream nil nil))
 839       (unless char (go RETURN-SYMBOL))
 840       (case (char-class char attribute-table)
 841         (#.+char-attr-constituent-sign+ (go EXPTSIGN))
 842         (#.+char-attr-constituent-digit+ (go EXPTDIGIT))
 843         (#.+char-attr-delimiter+ (unread-char char stream) (go RETURN-SYMBOL))
 844         (#.+char-attr-escape+ (go ESCAPE))
 845         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 846         (#.+char-attr-package-delimiter+ (go COLON))
 847         (t (go SYMBOL)))
 848      EXPTSIGN ; got to EXPONENT, and saw a sign character
 849       (ouch-read-buffer char)
 850       (setq char (read-char stream nil nil))
 851       (unless char (go RETURN-SYMBOL))
 852       (case (char-class char attribute-table)
 853         (#.+char-attr-constituent-digit+ (go EXPTDIGIT))
 854         (#.+char-attr-delimiter+ (unread-char char stream) (go RETURN-SYMBOL))
 855         (#.+char-attr-escape+ (go ESCAPE))
 856         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 857         (#.+char-attr-package-delimiter+ (go COLON))
 858         (t (go SYMBOL)))
 859      EXPTDIGIT ; got to EXPONENT, saw "[sign] {digit}+"
 860       (ouch-read-buffer char)
 861       (setq char (read-char stream nil nil))
 862       (unless char (return (make-float stream)))
 863       (case (char-class char attribute-table)
 864         (#.+char-attr-constituent-digit+ (go EXPTDIGIT))
 865         (#.+char-attr-delimiter+
 866          (unread-char char stream)
 867          (return (make-float stream)))
 868         (#.+char-attr-escape+ (go ESCAPE))
 869         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 870         (#.+char-attr-package-delimiter+ (go COLON))
 871         (t (go SYMBOL)))
 872      RATIO ; saw "[sign] {digit}+ slash"
 873       (ouch-read-buffer char)
 874       (setq char (read-char stream nil nil))
 875       (unless char (go RETURN-SYMBOL))
 876       (case (char-class2 char attribute-table)
 877         (#.+char-attr-constituent-digit+ (go RATIODIGIT))
 878         (#.+char-attr-delimiter+ (unread-char char stream) (go RETURN-SYMBOL))
 879         (#.+char-attr-escape+ (go ESCAPE))
 880         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 881         (#.+char-attr-package-delimiter+ (go COLON))
 882         (t (go SYMBOL)))
 883      RATIODIGIT ; saw "[sign] {digit}+ slash {digit}+"
 884       (ouch-read-buffer char)
 885       (setq char (read-char stream nil nil))
 886       (unless char (return (make-ratio stream)))
 887       (case (char-class2 char attribute-table)
 888         (#.+char-attr-constituent-digit+ (go RATIODIGIT))
 889         (#.+char-attr-delimiter+
 890          (unread-char char stream)
 891          (return (make-ratio stream)))
 892         (#.+char-attr-escape+ (go ESCAPE))
 893         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 894         (#.+char-attr-package-delimiter+ (go COLON))
 895         (t (go SYMBOL)))
 896      DOTS ; saw "dot {dot}+"
 897       (ouch-read-buffer char)
 898       (setq char (read-char stream nil nil))
 899       (unless char (%reader-error stream "too many dots"))
 900       (case (char-class char attribute-table)
 901         (#.+char-attr-constituent-dot+ (go DOTS))
 902         (#.+char-attr-delimiter+
 903          (unread-char char stream)
 904          (%reader-error stream "too many dots"))
 905         (#.+char-attr-escape+ (go ESCAPE))
 906         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 907         (#.+char-attr-package-delimiter+ (go COLON))
 908         (t (go SYMBOL)))
 909      SYMBOL ; not a dot, dots, or number
 910       (let ((stream (in-synonym-of stream)))
 911         (if (ansi-stream-p stream)
 912             (prepare-for-fast-read-char stream
 913               (prog ()
 914                SYMBOL-LOOP
 915                (ouch-read-buffer char)
 916                (setq char (fast-read-char nil nil))
 917                (unless char (go RETURN-SYMBOL))
 918                (case (char-class char attribute-table)
 919                  (#.+char-attr-escape+ (done-with-fast-read-char)
 920                                        (go ESCAPE))
 921                  (#.+char-attr-delimiter+ (done-with-fast-read-char)
 922                                           (unread-char char stream)
 923                                           (go RETURN-SYMBOL))
 924                  (#.+char-attr-multiple-escape+ (done-with-fast-read-char)
 925                                                 (go MULT-ESCAPE))
 926                  (#.+char-attr-package-delimiter+ (done-with-fast-read-char)
 927                                                   (go COLON))
 928                  (t (go SYMBOL-LOOP)))))
 929             ;; fundamental-stream
 930             (prog ()
 931              SYMBOL-LOOP
 932              (ouch-read-buffer char)
 933              (setq char (stream-read-char stream))
 934              (when (eq char :eof) (go RETURN-SYMBOL))
 935              (case (char-class char attribute-table)
 936                (#.+char-attr-escape+ (go ESCAPE))
 937                (#.+char-attr-delimiter+ (stream-unread-char stream char)
 938                             (go RETURN-SYMBOL))
 939                (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 940                (#.+char-attr-package-delimiter+ (go COLON))
 941                (t (go SYMBOL-LOOP))))))
 942      ESCAPE ; saw an escape
 943       ;; Don't put the escape in the read buffer.
 944       ;; READ-NEXT CHAR, put in buffer (no case conversion).
 945       (let ((nextchar (read-char stream nil nil)))
 946         (unless nextchar
 947           (reader-eof-error stream "after escape character"))
 948         (push *ouch-ptr* escapes)
 949         (ouch-read-buffer nextchar))
 950       (setq char (read-char stream nil nil))
 951       (unless char (go RETURN-SYMBOL))
 952       (case (char-class char attribute-table)
 953         (#.+char-attr-delimiter+ (unread-char char stream) (go RETURN-SYMBOL))
 954         (#.+char-attr-escape+ (go ESCAPE))
 955         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 956         (#.+char-attr-package-delimiter+ (go COLON))
 957         (t (go SYMBOL)))
 958       MULT-ESCAPE
 959       (do ((char (read-char stream t) (read-char stream t)))
 960           ((multiple-escape-p char))
 961         (if (escapep char) (setq char (read-char stream t)))
 962         (push *ouch-ptr* escapes)
 963         (ouch-read-buffer char))
 964       (setq char (read-char stream nil nil))
 965       (unless char (go RETURN-SYMBOL))
 966       (case (char-class char attribute-table)
 967         (#.+char-attr-delimiter+ (unread-char char stream) (go RETURN-SYMBOL))
 968         (#.+char-attr-escape+ (go ESCAPE))
 969         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 970         (#.+char-attr-package-delimiter+ (go COLON))
 971         (t (go SYMBOL)))
 972       COLON
 973       (casify-read-buffer escapes)
 974       (unless (zerop colons)
 975         (%reader-error stream "too many colons in ~S"
 976                       (read-buffer-to-string)))
 977       (setq colons 1)
 978       (setq package-designator
 979             (if (plusp *ouch-ptr*)
 980                 ;; FIXME: It seems inefficient to cons up a package
 981                 ;; designator string every time we read a symbol with an
 982                 ;; explicit package prefix. Perhaps we could implement
 983                 ;; a FIND-PACKAGE* function analogous to INTERN*
 984                 ;; and friends?
 985                 (read-buffer-to-string)
 986                 *keyword-package*))
 987       (reset-read-buffer)
 988       (setq escapes ())
 989       (setq char (read-char stream nil nil))
 990       (unless char (reader-eof-error stream "after reading a colon"))
 991       (case (char-class char attribute-table)
 992         (#.+char-attr-delimiter+
 993          (unread-char char stream)
 994          (%reader-error stream
 995                         "illegal terminating character after a colon: ~S"
 996                         char))
 997         (#.+char-attr-escape+ (go ESCAPE))
 998         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
 999         (#.+char-attr-package-delimiter+ (go INTERN))
1000         (t (go SYMBOL)))
1001       INTERN
1002       (setq colons 2)
1003       (setq char (read-char stream nil nil))
1004       (unless char
1005         (reader-eof-error stream "after reading a colon"))
1006       (case (char-class char attribute-table)
1007         (#.+char-attr-delimiter+
1008          (unread-char char stream)
1009          (%reader-error stream
1010                         "illegal terminating character after a colon: ~S"
1011                         char))
1012         (#.+char-attr-escape+ (go ESCAPE))
1013         (#.+char-attr-multiple-escape+ (go MULT-ESCAPE))
1014         (#.+char-attr-package-delimiter+
1015          (%reader-error stream
1016                         "too many colons after ~S name"
1017                         package-designator))
1018         (t (go SYMBOL)))
1019       RETURN-SYMBOL
1020       (casify-read-buffer escapes)
1021       (let ((found (if package-designator
1022                        (find-package package-designator)
1023                        (sane-package))))
1024         (unless found
1025           (error 'reader-package-error :stream stream
1026                  :format-arguments (list package-designator)
1027                  :format-control "package ~S not found"))
1028
1029         (if (or (zerop colons) (= colons 2) (eq found *keyword-package*))
1030             (return (intern* *read-buffer* *ouch-ptr* found))
1031             (multiple-value-bind (symbol test)
1032                 (find-symbol* *read-buffer* *ouch-ptr* found)
1033               (when (eq test :external) (return symbol))
1034               (let ((name (read-buffer-to-string)))
1035                 (with-simple-restart (continue "Use symbol anyway.")
1036                   (error 'reader-package-error :stream stream
1037                          :format-arguments (list name (package-name found))
1038                          :format-control
1039                          (if test
1040                              "The symbol ~S is not external in the ~A package."
1041                              "Symbol ~S not found in the ~A package.")))
1042                 (return (intern name found)))))))))
1043
1044 ;;; for semi-external use:
1045 ;;;
1046 ;;; For semi-external use: Return 3 values: the string for the token,
1047 ;;; a flag for whether there was an escape char, and the position of
1048 ;;; any package delimiter.
1049 (defun read-extended-token (stream &optional (*readtable* *readtable*))
1050   (let ((first-char (read-char stream nil nil t)))
1051     (cond (first-char
1052            (multiple-value-bind (escapes colon)
1053                (internal-read-extended-token stream first-char nil)
1054              (casify-read-buffer escapes)
1055              (values (read-buffer-to-string) (not (null escapes)) colon)))
1056           (t
1057            (values "" nil nil)))))
1058
1059 ;;; for semi-external use:
1060 ;;;
1061 ;;; Read an extended token with the first character escaped. Return
1062 ;;; the string for the token.
1063 (defun read-extended-token-escaped (stream &optional (*readtable* *readtable*))
1064   (let ((first-char (read-char stream nil nil)))
1065     (cond (first-char
1066             (let ((escapes (internal-read-extended-token stream first-char t)))
1067               (casify-read-buffer escapes)
1068               (read-buffer-to-string)))
1069           (t
1070             (reader-eof-error stream "after escape")))))
1071 \f
1072 ;;;; number-reading functions
1073
1074 (defmacro digit* nil
1075   `(do ((ch char (inch-read-buffer)))
1076        ((or (eofp ch) (not (digit-char-p ch))) (setq char ch))
1077      ;; Report if at least one digit is seen.
1078      (setq one-digit t)))
1079
1080 (defmacro exponent-letterp (letter)
1081   `(memq ,letter '(#\E #\S #\F #\L #\D #\e #\s #\f #\l #\d)))
1082
1083 ;;; FIXME: It would be cleaner to have these generated automatically
1084 ;;; by compile-time code instead of having them hand-created like
1085 ;;; this. The !COLD-INIT-INTEGER-READER code below should be resurrected
1086 ;;; and tested.
1087 (defvar *integer-reader-safe-digits*
1088   #(nil nil
1089     26 17 13 11 10 9 8 8 8 7 7 7 7 6 6 6 6 6 6 6 6 5 5 5 5 5 5 5 5 5 5 5 5 5 5)
1090   #!+sb-doc
1091   "the mapping of base to 'safe' number of digits to read for a fixnum")
1092 (defvar *integer-reader-base-power*
1093   #(nil nil
1094     67108864 129140163 67108864 48828125 60466176 40353607
1095     16777216 43046721 100000000 19487171 35831808 62748517 105413504 11390625
1096     16777216 24137569 34012224 47045881 64000000 85766121 113379904 6436343
1097     7962624 9765625 11881376 14348907 17210368 20511149 24300000 28629151
1098     33554432 39135393 45435424 52521875 60466176)
1099   #!+sb-doc
1100   "the largest fixnum power of the base for MAKE-INTEGER")
1101 (declaim (simple-vector *integer-reader-safe-digits*
1102                         *integer-reader-base-power*))
1103 #|
1104 (defun !cold-init-integer-reader ()
1105   (do ((base 2 (1+ base)))
1106       ((> base 36))
1107     (let ((digits
1108           (do ((fix (truncate most-positive-fixnum base)
1109                     (truncate fix base))
1110                (digits 0 (1+ digits)))
1111               ((zerop fix) digits))))
1112       (setf (aref *integer-reader-safe-digits* base)
1113             digits
1114             (aref *integer-reader-base-power* base)
1115             (expt base digits)))))
1116 |#
1117
1118 (defun make-integer ()
1119   #!+sb-doc
1120   "Minimizes bignum-fixnum multiplies by reading a 'safe' number of digits,
1121   then multiplying by a power of the base and adding."
1122   (let* ((base *read-base*)
1123          (digits-per (aref *integer-reader-safe-digits* base))
1124          (base-power (aref *integer-reader-base-power* base))
1125          (negativep nil)
1126          (number 0))
1127     (declare (type index digits-per base-power))
1128     (read-unwind-read-buffer)
1129     (let ((char (inch-read-buffer)))
1130       (cond ((char= char #\-)
1131              (setq negativep t))
1132             ((char= char #\+))
1133             (t (unread-buffer))))
1134     (loop
1135      (let ((num 0))
1136        (declare (type index num))
1137        (dotimes (digit digits-per)
1138          (let* ((ch (inch-read-buffer)))
1139            (cond ((or (eofp ch) (char= ch #\.))
1140                   (return-from make-integer
1141                                (let ((res
1142                                       (if (zerop number) num
1143                                           (+ num (* number
1144                                                     (expt base digit))))))
1145                                  (if negativep (- res) res))))
1146                  (t (setq num (+ (digit-char-p ch base)
1147                                  (the index (* num base))))))))
1148        (setq number (+ num (* number base-power)))))))
1149
1150 (defun make-float (stream)
1151   ;; Assume that the contents of *read-buffer* are a legal float, with nothing
1152   ;; else after it.
1153   (read-unwind-read-buffer)
1154   (let ((negative-fraction nil)
1155         (number 0)
1156         (divisor 1)
1157         (negative-exponent nil)
1158         (exponent 0)
1159         (float-char ())
1160         (char (inch-read-buffer)))
1161     (if (cond ((char= char #\+) t)
1162               ((char= char #\-) (setq negative-fraction t)))
1163         ;; Flush it.
1164         (setq char (inch-read-buffer)))
1165     ;; Read digits before the dot.
1166     (do* ((ch char (inch-read-buffer))
1167           (dig (digit-char-p ch) (digit-char-p ch)))
1168          ((not dig) (setq char ch))
1169       (setq number (+ (* number 10) dig)))
1170     ;; Deal with the dot, if it's there.
1171     (when (char= char #\.)
1172       (setq char (inch-read-buffer))
1173       ;; Read digits after the dot.
1174       (do* ((ch char (inch-read-buffer))
1175             (dig (and (not (eofp ch)) (digit-char-p ch))
1176                  (and (not (eofp ch)) (digit-char-p ch))))
1177            ((not dig) (setq char ch))
1178         (setq divisor (* divisor 10))
1179         (setq number (+ (* number 10) dig))))
1180     ;; Is there an exponent letter?
1181     (cond ((eofp char)
1182            ;; If not, we've read the whole number.
1183            (let ((num (make-float-aux number divisor
1184                                       *read-default-float-format*
1185                                       stream)))
1186              (return-from make-float (if negative-fraction (- num) num))))
1187           ((exponent-letterp char)
1188            (setq float-char char)
1189            ;; Build exponent.
1190            (setq char (inch-read-buffer))
1191            ;; Check leading sign.
1192            (if (cond ((char= char #\+) t)
1193                      ((char= char #\-) (setq negative-exponent t)))
1194                ;; Flush sign.
1195                (setq char (inch-read-buffer)))
1196            ;; Read digits for exponent.
1197            (do* ((ch char (inch-read-buffer))
1198                  (dig (and (not (eofp ch)) (digit-char-p ch))
1199                       (and (not (eofp ch)) (digit-char-p ch))))
1200                 ((not dig)
1201                  (setq exponent (if negative-exponent (- exponent) exponent)))
1202              (setq exponent (+ (* exponent 10) dig)))
1203            ;; Generate and return the float, depending on FLOAT-CHAR:
1204            (let* ((float-format (case (char-upcase float-char)
1205                                   (#\E *read-default-float-format*)
1206                                   (#\S 'short-float)
1207                                   (#\F 'single-float)
1208                                   (#\D 'double-float)
1209                                   (#\L 'long-float)))
1210                   num)
1211              ;; Raymond Toy writes: We need to watch out if the
1212              ;; exponent is too small or too large. We add enough to
1213              ;; EXPONENT to make it within range and scale NUMBER
1214              ;; appropriately. This should avoid any unnecessary
1215              ;; underflow or overflow problems.
1216              (multiple-value-bind (min-expo max-expo)
1217                  ;; FIXME: These forms are broken w.r.t.
1218                  ;; cross-compilation portability, as the
1219                  ;; cross-compiler will call the host's LOG function
1220                  ;; while attempting to constant-fold. Maybe some sort
1221                  ;; of load-time-form magic could be used instead?
1222                  (case float-format
1223                    (short-float
1224                     (values
1225                      (log sb!xc:least-positive-normalized-short-float 10s0)
1226                      (log sb!xc:most-positive-short-float 10s0)))
1227                    (single-float
1228                     (values
1229                      (log sb!xc:least-positive-normalized-single-float 10f0)
1230                      (log sb!xc:most-positive-single-float 10f0)))
1231                    (double-float
1232                     (values
1233                      (log sb!xc:least-positive-normalized-double-float 10d0)
1234                      (log sb!xc:most-positive-double-float 10d0)))
1235                    (long-float
1236                     (values
1237                      (log sb!xc:least-positive-normalized-long-float 10L0)
1238                      (log sb!xc:most-positive-long-float 10L0))))
1239                (let ((correction (cond ((<= exponent min-expo)
1240                                         (ceiling (- min-expo exponent)))
1241                                        ((>= exponent max-expo)
1242                                         (floor (- max-expo exponent)))
1243                                        (t
1244                                         0))))
1245                  (incf exponent correction)
1246                  (setf number (/ number (expt 10 correction)))
1247                  (setq num (make-float-aux number divisor float-format stream))
1248                  (setq num (* num (expt 10 exponent)))
1249                  (return-from make-float (if negative-fraction
1250                                              (- num)
1251                                              num))))))
1252           ;; should never happen
1253           (t (bug "bad fallthrough in floating point reader")))))
1254
1255 (defun make-float-aux (number divisor float-format stream)
1256   (handler-case
1257       (coerce (/ number divisor) float-format)
1258     (type-error (c)
1259       (error 'reader-impossible-number-error
1260              :error c :stream stream
1261              :format-control "failed to build float"))))
1262
1263 (defun make-ratio (stream)
1264   ;; Assume *READ-BUFFER* contains a legal ratio. Build the number from
1265   ;; the string.
1266   ;;
1267   ;; Look for optional "+" or "-".
1268   (let ((numerator 0) (denominator 0) (char ()) (negative-number nil))
1269     (read-unwind-read-buffer)
1270     (setq char (inch-read-buffer))
1271     (cond ((char= char #\+)
1272            (setq char (inch-read-buffer)))
1273           ((char= char #\-)
1274            (setq char (inch-read-buffer))
1275            (setq negative-number t)))
1276     ;; Get numerator.
1277     (do* ((ch char (inch-read-buffer))
1278           (dig (digit-char-p ch *read-base*)
1279                (digit-char-p ch *read-base*)))
1280          ((not dig))
1281          (setq numerator (+ (* numerator *read-base*) dig)))
1282     ;; Get denominator.
1283     (do* ((ch (inch-read-buffer) (inch-read-buffer))
1284           (dig ()))
1285          ((or (eofp ch) (not (setq dig (digit-char-p ch *read-base*)))))
1286          (setq denominator (+ (* denominator *read-base*) dig)))
1287     (let ((num (handler-case
1288                    (/ numerator denominator)
1289                  (arithmetic-error (c)
1290                    (error 'reader-impossible-number-error
1291                           :error c :stream stream
1292                           :format-control "failed to build ratio")))))
1293       (if negative-number (- num) num))))
1294 \f
1295 ;;;; cruft for dispatch macros
1296
1297 (defun make-char-dispatch-table ()
1298   (make-array char-code-limit :initial-element #'dispatch-char-error))
1299
1300 (defun dispatch-char-error (stream sub-char ignore)
1301   (declare (ignore ignore))
1302   (if *read-suppress*
1303       (values)
1304       (%reader-error stream "no dispatch function defined for ~S" sub-char)))
1305
1306 (defun make-dispatch-macro-character (char &optional
1307                                            (non-terminating-p nil)
1308                                            (rt *readtable*))
1309   #!+sb-doc
1310   "Cause CHAR to become a dispatching macro character in readtable (which
1311    defaults to the current readtable). If NON-TERMINATING-P, the char will
1312    be non-terminating."
1313   (set-macro-character char #'read-dispatch-char non-terminating-p rt)
1314   (let* ((dalist (dispatch-tables rt))
1315          (dtable (cdr (find char dalist :test #'char= :key #'car))))
1316     (cond (dtable
1317            (error "The dispatch character ~S already exists." char))
1318           (t
1319            (setf (dispatch-tables rt)
1320                  (push (cons char (make-char-dispatch-table)) dalist)))))
1321   t)
1322
1323 (defun set-dispatch-macro-character (disp-char sub-char function
1324                                                &optional (rt *readtable*))
1325   #!+sb-doc
1326   "Cause FUNCTION to be called whenever the reader reads DISP-CHAR
1327    followed by SUB-CHAR."
1328   ;; Get the dispatch char for macro (error if not there), diddle
1329   ;; entry for sub-char.
1330   (when (digit-char-p sub-char)
1331     (error "SUB-CHAR must not be a decimal digit: ~S" sub-char))
1332   (let* ((sub-char (char-upcase sub-char))
1333          (rt (or rt *standard-readtable*))
1334          (dpair (find disp-char (dispatch-tables rt)
1335                       :test #'char= :key #'car)))
1336     (if dpair
1337         (setf (elt (the simple-vector (cdr dpair))
1338                    (char-code sub-char))
1339               (coerce function 'function))
1340         (error "~S is not a dispatch char." disp-char))))
1341
1342 (defun get-dispatch-macro-character (disp-char sub-char
1343                                      &optional (rt *readtable*))
1344   #!+sb-doc
1345   "Return the macro character function for SUB-CHAR under DISP-CHAR
1346    or NIL if there is no associated function."
1347   (let* ((sub-char (char-upcase sub-char))
1348          (rt (or rt *standard-readtable*))
1349          (dpair (find disp-char (dispatch-tables rt)
1350                       :test #'char= :key #'car)))
1351     (if dpair
1352         (let ((dispatch-fun (elt (the simple-vector (cdr dpair))
1353                                  (char-code sub-char))))
1354           ;; Digits are also initialized in a dispatch table to
1355           ;; #'dispatch-char-error; READ-DISPATCH-CHAR handles them
1356           ;; separately. - CSR, 2002-04-12
1357           (if (eq dispatch-fun #'dispatch-char-error)
1358               nil
1359               dispatch-fun))
1360         (error "~S is not a dispatch char." disp-char))))
1361
1362 (defun read-dispatch-char (stream char)
1363   ;; Read some digits.
1364   (let ((numargp nil)
1365         (numarg 0)
1366         (sub-char ()))
1367     (do* ((ch (read-char stream nil *eof-object*)
1368               (read-char stream nil *eof-object*))
1369           (dig ()))
1370          ((or (eofp ch)
1371               (not (setq dig (digit-char-p ch))))
1372           ;; Take care of the extra char.
1373           (if (eofp ch)
1374               (reader-eof-error stream "inside dispatch character")
1375               (setq sub-char (char-upcase ch))))
1376       (setq numargp t)
1377       (setq numarg (+ (* numarg 10) dig)))
1378     ;; Look up the function and call it.
1379     (let ((dpair (find char (dispatch-tables *readtable*)
1380                        :test #'char= :key #'car)))
1381       (if dpair
1382           (funcall (the function
1383                         (elt (the simple-vector (cdr dpair))
1384                              (char-code sub-char)))
1385                    stream sub-char (if numargp numarg nil))
1386           (%reader-error stream "no dispatch table for dispatch char")))))
1387 \f
1388 ;;;; READ-FROM-STRING
1389
1390 ;;; FIXME: Is it really worth keeping this pool?
1391 (defvar *read-from-string-spares* ()
1392   #!+sb-doc
1393   "A resource of string streams for Read-From-String.")
1394
1395 (defun read-from-string (string &optional eof-error-p eof-value
1396                                 &key (start 0) end
1397                                 preserve-whitespace)
1398   #!+sb-doc
1399   "The characters of string are successively given to the lisp reader
1400    and the lisp object built by the reader is returned. Macro chars
1401    will take effect."
1402   (declare (string string))
1403
1404   (with-array-data ((string string)
1405                     (start start)
1406                     (end (%check-vector-sequence-bounds string start end)))
1407     (unless *read-from-string-spares*
1408       (push (internal-make-string-input-stream "" 0 0)
1409             *read-from-string-spares*))
1410     (let ((stream (pop *read-from-string-spares*)))
1411       (setf (string-input-stream-string stream) string)
1412       (setf (string-input-stream-current stream) start)
1413       (setf (string-input-stream-end stream) end)
1414       (unwind-protect
1415           (values (if preserve-whitespace
1416                       (read-preserving-whitespace stream eof-error-p eof-value)
1417                       (read stream eof-error-p eof-value))
1418                   (string-input-stream-current stream))
1419         (push stream *read-from-string-spares*)))))
1420 \f
1421 ;;;; PARSE-INTEGER
1422
1423 (defun parse-integer (string &key (start 0) end (radix 10) junk-allowed)
1424   #!+sb-doc
1425   "Examine the substring of string delimited by start and end
1426   (default to the beginning and end of the string)  It skips over
1427   whitespace characters and then tries to parse an integer. The
1428   radix parameter must be between 2 and 36."
1429   (macrolet ((parse-error (format-control)
1430                `(error 'simple-parse-error
1431                        :format-control ,format-control
1432                        :format-arguments (list string))))
1433     (with-array-data ((string string)
1434                       (start start)
1435                       (end (%check-vector-sequence-bounds string start end)))
1436       (let ((index (do ((i start (1+ i)))
1437                        ((= i end)
1438                         (if junk-allowed
1439                             (return-from parse-integer (values nil end))
1440                             (parse-error "no non-whitespace characters in string ~S.")))
1441                      (declare (fixnum i))
1442                      (unless (whitespacep (char string i)) (return i))))
1443             (minusp nil)
1444             (found-digit nil)
1445             (result 0))
1446         (declare (fixnum index))
1447         (let ((char (char string index)))
1448           (cond ((char= char #\-)
1449                  (setq minusp t)
1450                  (incf index))
1451                 ((char= char #\+)
1452                  (incf index))))
1453         (loop
1454          (when (= index end) (return nil))
1455          (let* ((char (char string index))
1456                 (weight (digit-char-p char radix)))
1457            (cond (weight
1458                   (setq result (+ weight (* result radix))
1459                         found-digit t))
1460                  (junk-allowed (return nil))
1461                  ((whitespacep char)
1462                   (do ((jndex (1+ index) (1+ jndex)))
1463                       ((= jndex end))
1464                     (declare (fixnum jndex))
1465                     (unless (whitespacep (char string jndex))
1466                       (parse-error "junk in string ~S")))
1467                   (return nil))
1468                  (t
1469                   (parse-error "junk in string ~S"))))
1470          (incf index))
1471         (values
1472          (if found-digit
1473              (if minusp (- result) result)
1474              (if junk-allowed
1475                  nil
1476                  (parse-error "no digits in string ~S")))
1477          index)))))
1478 \f
1479 ;;;; reader initialization code
1480
1481 (defun !reader-cold-init ()
1482   (!cold-init-read-buffer)
1483   (!cold-init-secondary-attribute-table)
1484   (!cold-init-standard-readtable)
1485   ;; FIXME: This was commented out, but should probably be restored.
1486   #+nil (!cold-init-integer-reader))
1487 \f
1488 (def!method print-object ((readtable readtable) stream)
1489   (print-unreadable-object (readtable stream :identity t :type t)))