stream.lisp 15.2 KB
Newer Older
1 2 3 4 5
;;;; ---------------------------------------------------------------------------
;;;; Utilities related to streams

(asdf/package:define-package :asdf/stream
  (:recycle :asdf/stream)
6
  (:use :cl :asdf/package :asdf/compatibility :asdf/utility :asdf/pathname)
7
  #+gcl<2.7 (:shadowing-import-from :asdf/compatibility #:with-standard-io-syntax)
8
  (:export
9
   #:*default-stream-element-type* #:*stderr* #:setup-stderr
10 11 12 13
   #:with-safe-io-syntax #:call-with-safe-io-syntax
   #:with-output #:output-string #:with-input
   #:with-input-file #:call-with-input-file
   #:finish-outputs #:format! #:safe-format!
14 15
   #:copy-stream-to-stream #:concatenate-files
   #:copy-stream-to-stream-line-by-line
16 17 18
   #:slurp-stream-string #:slurp-stream-lines #:slurp-stream-forms #:slurp-stream-form
   #:read-file-string #:read-file-lines #:read-file-forms #:read-file-form #:safe-read-file-form
   #:eval-input #:eval-thunk #:standard-eval-thunk
19 20 21
   #:detect-encoding #:*encoding-detection-hook* #:always-default-encoding
   #:encoding-external-format #:*encoding-external-format-hook* #:default-encoding-external-format
   #:*default-encoding* #:*utf-8-external-format*))
22 23 24 25 26
(in-package :asdf/stream)

(defvar *default-stream-element-type* (or #+(or abcl cmu cormanlisp scl xcl) 'character :default)
  "default element-type for open (depends on the current CL implementation)")

27
(defvar *stderr* *error-output*
28 29
  "the original error output stream at startup")

30
(defun setup-stderr ()
31 32 33 34 35
  (setf *stderr*
        #+allegro excl::*stderr*
        #+clozure ccl::*stderr*
        #-(or allegro clozure) *error-output*))
(setup-stderr)
36

37

38
;;; Safe syntax
39

40 41
(defvar *standard-readtable* (copy-readtable nil))

42 43 44
(defmacro with-safe-io-syntax ((&key (package :cl)) &body body)
  "Establish safe CL reader options around the evaluation of BODY"
  `(call-with-safe-io-syntax #'(lambda () (let ((*package* (find-package ,package))) ,@body))))
45

46 47 48
(defun* call-with-safe-io-syntax (thunk &key (package :cl))
  (with-standard-io-syntax ()
    (let ((*package* (find-package package))
49 50
          (*readtable* *standard-readtable*)
          (*read-default-float-format* 'double-float)
51 52 53
          (*print-readably* nil)
	  (*read-eval* nil))
      (funcall thunk))))
54 55 56 57


;;; Output to a stream or string, FORMAT-style

58 59 60 61 62 63 64
(defun* call-with-output (output function)
  "Calls FUNCTION with an actual stream argument,
behaving like FORMAT with respect to how stream designators are interpreted:
If OUTPUT is a stream, use it as the stream.
If OUTPUT is NIL, use a STRING-OUTPUT-STREAM as the stream, and return the resulting string.
If OUTPUT is T, use *STANDARD-OUTPUT* as the stream.
If OUTPUT is a string with a fill-pointer, use it as a string-output-stream.
65
Otherwise, signal an error."
66
  (etypecase output
67
    (null
68
     (with-output-to-string (stream) (funcall function stream)))
69
    ((eql t)
70
     (funcall function *standard-output*))
71
    (stream
72
     (funcall function output))
73
    (string
74 75
     (assert (fill-pointer output))
     (with-output-to-string (stream output) (funcall function stream)))))
76

77 78
(defmacro with-output ((output-var &optional (value output-var)) &body body)
  "Bind OUTPUT-VAR to an output stream, coercing VALUE (default: previous binding of OUTPUT-VAR)
79
as per FORMAT, and evaluate BODY within the scope of this binding."
80
  `(call-with-output ,value #'(lambda (,output-var) ,@body)))
81

82 83 84 85
(defun* output-string (string &optional output)
  "If the desired OUTPUT is not NIL, print the string to the output; otherwise return the string"
  (if output
      (with-output (output) (princ string output))
86 87
      string))

88 89 90

;;; Input helpers

91 92 93 94 95 96 97
(defun* call-with-input (input function)
  "Calls FUNCTION with an actual stream argument, interpreting
stream designators like READ, but also coercing strings to STRING-INPUT-STREAM.
If INPUT is a STREAM, use it as the stream.
If INPUT is NIL, use a *STANDARD-INPUT* as the stream.
If INPUT is T, use *TERMINAL-IO* as the stream.
As an extension, if INPUT is a string, use it as a string-input-stream.
98
Otherwise, signal an error."
99 100 101 102 103
  (etypecase input
    (null (funcall function *standard-input*))
    ((eql t) (funcall function *terminal-io*))
    (stream (funcall function input))
    (string (with-input-from-string (stream input) (funcall function stream)))))
104

105 106 107 108
(defmacro with-input ((input-var &optional (value input-var)) &body body)
  "Bind INPUT-VAR to an input stream, coercing VALUE (default: previous binding of INPUT-VAR)
as per CALL-WITH-INPUT, and evaluate BODY within the scope of this binding."
  `(call-with-input ,value #'(lambda (,input-var) ,@body)))
109 110

(defun* call-with-input-file (pathname thunk
111 112 113 114 115 116
                                       &key
                                       (element-type *default-stream-element-type*)
                                       (external-format :default)
                                       (if-does-not-exist :error))
  "Open FILE for input with given recognizes options, call THUNK with the resulting stream.
Other keys are accepted but discarded."
117
  #+gcl<2.7 (declare (ignore external-format))
118
  (with-open-file (s pathname :direction :input
119 120
                     :element-type element-type
                     #-gcl<2.7 :external-format #-gcl<2.7 external-format
121
                     :if-does-not-exist if-does-not-exist)
122 123 124 125 126 127 128
    (funcall thunk s)))

(defmacro with-input-file ((var pathname &rest keys &key element-type external-format) &body body)
  (declare (ignore element-type external-format))
  `(call-with-input-file ,pathname #'(lambda (,var) ,@body) ,@keys))


129
;;; Ensure output buffers are flushed
130

131 132 133 134 135
(defun* finish-outputs (&rest streams)
  "Finish output on the main output streams as well as any specified one.
Useful for portably flushing I/O before user input or program exit."
  ;; CCL notably buffers its stream output by default.
  (dolist (s (append streams
136 137
                     (list *stderr* *error-output* *standard-output* *trace-output*
                           *debug-io* *terminal-io* *debug-io* *query-io*)))
138 139
    (ignore-errors (finish-output s)))
  (values))
140

141 142 143 144 145
(defun* format! (stream format &rest args)
  "Just like format, but call finish-outputs before and after the output."
  (finish-outputs stream)
  (apply 'format stream format args)
  (finish-output stream))
146

147 148 149 150
(defun* safe-format! (stream format &rest args)
  (with-safe-io-syntax ()
    (ignore-errors (apply 'format! stream format args))
    (finish-outputs stream))) ; just in case format failed
151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173


;;; Simple Whole-Stream processing

(defun* copy-stream-to-stream (input output &key (element-type 'character) (buffer-size 8192))
  "Copy the contents of the INPUT stream into the OUTPUT stream,
using WRITE-SEQUENCE and a sensibly sized buffer."
  (with-open-stream (input input)
    (loop
      :for buffer = (make-array (list buffer-size) :element-type element-type)
      :for end = (read-sequence buffer input)
      :until (zerop end)
      :do (write-sequence buffer output :end end)
          (when (< end buffer-size) (return)))))

(defun* concatenate-files (inputs output)
  (with-open-file (o output :element-type '(unsigned-byte 8)
                            :direction :output :if-exists :rename-and-delete)
    (dolist (input inputs)
      (with-open-file (i input :element-type '(unsigned-byte 8)
                               :direction :input :if-does-not-exist :error)
        (copy-stream-to-stream i o :element-type '(unsigned-byte 8))))))

174
(defun* copy-stream-to-stream-line-by-line (input output &key prefix)
175 176 177 178 179 180 181 182 183 184 185
  "Copy the contents of the INPUT stream into the OUTPUT stream,
reading contents line by line."
  (with-open-stream (input input)
    (loop :for (line eof) = (multiple-value-list (read-line input nil nil))
      :while line :do
      (when prefix (princ prefix output))
      (princ line output)
      (unless eof (terpri output))
      (finish-output output)
      (when eof (return)))))

186
(defun* slurp-stream-string (input &key (element-type 'character))
187 188 189 190 191
  "Read the contents of the INPUT stream as a string"
  (with-open-stream (input input)
    (with-output-to-string (output)
      (copy-stream-to-stream input output :element-type element-type))))

192 193 194 195 196
(defun* slurp-stream-lines (input &key count)
  "Read the contents of the INPUT stream as a list of lines, return those lines.

Read no more than COUNT lines."
  (check-type count (or null integer))
197
  (with-open-stream (input input)
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219
    (loop :for n :from 0
          :for l = (and (or (not count) (< n count))
                        (read-line input nil nil))
          :while l :collect l)))

(defun* slurp-stream-line (input &key (path 0))
  "Read the contents of the INPUT stream as a list of lines,
then return the SUB-OBJECT of that list of lines following the PATH.
PATH defaults to 0, i.e. return the first line.
PATH is typically an integer, or a list of an integer and a function.
If PATH is NIL, it will return all the lines in the file.

The stream will not be read beyond the Nth lines,
where N is the index specified by path
if path is either an integer or a list that starts with an integer."
  (let* ((count (cond
                  ((integerp path)
                   (1+ path))
                  ((and (consp path) (integerp (first path)))
                   (1+ (first path)))))
         (forms (slurp-stream-lines input :count count)))
    (sub-object forms path)))
220

221 222 223 224 225 226 227
(defun* slurp-stream-forms (input &key count)
"Read the contents of the INPUT stream as a list of forms,
and return those forms.

If COUNT is null, read to the end of the stream;
if COUNT is an integer, stop after COUNT forms were read.

228
BEWARE: be sure to use WITH-SAFE-IO-SYNTAX, or some variant thereof"
229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255
  (check-type count (or null integer))
  (loop :with eof = '#:eof
        :for n :from 0
        :for form = (if (and count (>= n count))
                        eof
                        (read-preserving-whitespace input nil eof))
        :until (eq form eof) :collect form))

(defun* slurp-stream-form (input &key (path 0))
"Read the contents of the INPUT stream as a list of forms,
then return the SUB-OBJECT of these forms following the PATH.
PATH defaults to 0, i.e. return the first form.
PATH is typically a list of integers.
If PATH is NIL, it will return all the forms in the file.

The stream will not be read beyond the Nth form,
where N is the index specified by path,
if path is either an integer or a list that starts with an integer.

BEWARE: be sure to use WITH-SAFE-IO-SYNTAX, or some variant thereof"
  (let* ((count (cond
                  ((integerp path)
                   (1+ path))
                  ((and (consp path) (integerp (first path)))
                   (1+ (first path)))))
         (forms (slurp-stream-forms input :count count)))
    (sub-object forms path)))
256

257
(defun* read-file-string (file &rest keys)
258 259 260
  "Open FILE with option KEYS, read its contents as a string"
  (apply 'call-with-input-file file 'slurp-stream-string keys))

261 262 263
(defun* read-file-lines (file &rest keys)
  "Open FILE with option KEYS, read its contents as a list of lines
BEWARE: be sure to use WITH-SAFE-IO-SYNTAX, or some variant thereof"
264 265
  (apply 'call-with-input-file file 'slurp-stream-lines keys))

266 267 268
(defun* read-file-forms (file &rest keys &key count &allow-other-keys)
  "Open input FILE with option KEYS (except COUNT),
and read its contents as per SLURP-STREAM-FORMS with given COUNT.
269
BEWARE: be sure to use WITH-SAFE-IO-SYNTAX, or some variant thereof"
270 271
  (apply 'call-with-input-file file
         #'(lambda (input) (slurp-stream-forms input :count count))
272
         (remove-plist-key :count keys)))
273

274 275 276
(defun* read-file-form (file &rest keys &key (path 0) &allow-other-keys)
  "Open input FILE with option KEYS (except path),
and read its contents as per SLURP-STREAM-FORM with given PATH.
277
BEWARE: be sure to use WITH-SAFE-IO-SYNTAX, or some variant thereof"
278 279
  (apply 'call-with-input-file file
         #'(lambda (input) (slurp-stream-form input :path path))
280
         (remove-plist-key :path keys)))
281 282 283 284 285

(defun* safe-read-file-form (pathname &rest keys &key (package :cl) &allow-other-keys)
  "Reads the specified form from the top of a file using a safe standardized syntax.
Extracts the form using READ-FILE-FORM,
within an WITH-SAFE-IO-SYNTAX using the specified PACKAGE."
286
  (with-safe-io-syntax (:package package)
287
    (apply 'read-file-form pathname (remove-plist-key :package keys))))
288 289 290 291 292 293 294 295 296 297

(defun* eval-input (input)
  "Portably read and evaluate forms from INPUT, return the last values."
  (with-input (input)
    (loop :with results :with eof ='#:eof
          :for form = (read input nil eof)
          :until (eq form eof)
          :do (setf results (multiple-value-list (eval form)))
          :finally (return (apply 'values results)))))

298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316
(defun* eval-thunk (thunk)
  "Evaluate a THUNK of code:
If a function, FUNCALL it without arguments.
If a constant literal and not a sequence, return it.
If a cons or a symbol, EVAL it.
If a string, repeatedly read and evaluate from it, returning the last values."
  (etypecase thunk
    ((or boolean keyword number character pathname) thunk)
    ((or cons symbol) (eval thunk))
    (function (funcall thunk))
    (string (eval-input thunk))))

(defun* standard-eval-thunk (thunk &key (package :cl))
  "Like EVAL-THUNK, but in a more standardized evaluation context."
  ;; Note: it's "standard-" not "safe-", because evaluation is never safe.
  (when thunk
    (with-safe-io-syntax (:package package)
      (let ((*read-eval* t))
        (eval-thunk thunk)))))
317

318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366

;;; Encodings

(defvar *default-encoding* :default
  "Default encoding for source files.
The default value :default preserves the legacy behavior.
A future default might be :utf-8 or :autodetect
reading emacs-style -*- coding: utf-8 -*- specifications,
and falling back to utf-8 or latin1 if nothing is specified.")

(defparameter *utf-8-external-format*
  #+(and asdf-unicode (not clisp)) :utf-8
  #+(and asdf-unicode clisp) charset:utf-8
  #-asdf-unicode :default
  "Default :external-format argument to pass to CL:OPEN and also
CL:LOAD or CL:COMPILE-FILE to best process a UTF-8 encoded file.
On modern implementations, this will decode UTF-8 code points as CL characters.
On legacy implementations, it may fall back on some 8-bit encoding,
with non-ASCII code points being read as several CL characters;
hopefully, if done consistently, that won't affect program behavior too much.")

(defun* always-default-encoding (pathname)
  (declare (ignore pathname))
  *default-encoding*)

(defvar *encoding-detection-hook* #'always-default-encoding
  "Hook for an extension to define a function to automatically detect a file's encoding")

(defun* detect-encoding (pathname)
  (if (and pathname (not (directory-pathname-p pathname)) (probe-file pathname))
      (funcall *encoding-detection-hook* pathname)
      *default-encoding*))

(defun* default-encoding-external-format (encoding)
  (case encoding
    (:default :default) ;; for backward-compatibility only. Explicit usage discouraged.
    (:utf-8 *utf-8-external-format*)
    (otherwise
     (cerror "Continue using :external-format :default" (compatfmt "~@<Your ASDF component is using encoding ~S but it isn't recognized. Your system should :defsystem-depends-on (:asdf-encodings).~:>") encoding)
     :default)))

(defvar *encoding-external-format-hook*
  #'default-encoding-external-format
  "Hook for an extension to define a mapping between non-default encodings
and implementation-defined external-format's")

(defun* encoding-external-format (encoding)
  (funcall *encoding-external-format-hook* encoding))