Commit | Line | Data |
---|---|---|
fd26d811 FT |
1 | ;;;; CHAR-CODE -- Conversions between characters and byte |
2 | ;;;; representations thereof | |
3 | ||
4 | (defpackage :charcode | |
5 | (:use :cl #+sbcl :sb-gray #-sbcl :gray) | |
6 | (:export "MAKE-ENCODER" "MAKE-DECODER" "ENCODE-STRING" "DECODE-STRING" | |
7 | "CODING-ERROR" | |
8 | "LATIN-1" "LATIN1" "UTF-8" "UTF8")) | |
9 | (in-package :charcode) | |
10 | ||
11 | ;;; General stuff | |
12 | ||
13 | (define-condition coding-error (error) | |
14 | ((input :initarg :input) | |
15 | (position :initarg :position) | |
16 | (result :initarg :result))) | |
17 | ||
18 | (define-condition simple-coding-error (coding-error simple-error) ()) | |
19 | ||
20 | (defun coding-error (input position result format &rest format-args) | |
21 | (error 'simple-coding-error | |
22 | :input input :position position :result result | |
23 | :format-control format :format-arguments format-args)) | |
24 | ||
25 | (deftype decoder-fun () `(function ((array (unsigned-byte 8)) | |
26 | (array character) | |
27 | &key (start fixnum) (end fixnum)) | |
28 | (member t nil))) | |
29 | (deftype encoder-fun () `(function ((array character) | |
30 | (array (unsigned-byte 8)) | |
31 | &key (start fixnum) (end fixnum)) | |
32 | (member t nil))) | |
33 | ||
34 | (defmacro define-encoder ((name) &body make-encoder) | |
35 | `(setf (get ',name 'make-encoder) #'(lambda () ,@make-encoder))) | |
36 | ||
37 | (defmacro define-decoder ((name) &body make-decoder) | |
38 | `(setf (get ',name 'make-decoder) #'(lambda () ,@make-decoder))) | |
39 | ||
40 | (defmacro define-codec-synonyms (name &rest synonyms) | |
41 | `(eval-when (:load-toplevel :execute) | |
42 | ,@(mapcar #'(lambda (sym) | |
43 | `(setf (get ',sym 'make-encoder) (get ',name 'make-encoder) | |
44 | (get ',sym 'make-decoder) (get ',name 'make-decoder))) | |
45 | synonyms))) | |
46 | ||
47 | (defun make-encoder (name) | |
48 | (the encoder-fun (values (funcall (get name 'make-encoder))))) | |
49 | ||
50 | (defun make-decoder (name) | |
51 | (the decoder-fun (values (funcall (get name 'make-decoder))))) | |
52 | ||
53 | (defun encode-string (string coding) | |
54 | (declare (type string string)) | |
55 | (let ((encoder (make-encoder coding)) | |
56 | (buf (make-array (list (length string)) :element-type '(unsigned-byte 8) :adjustable t :fill-pointer 0))) | |
57 | (unless (funcall encoder string buf) | |
58 | (coding-error string (length string) buf "Encoding of string in ~A ended prematurely." coding)) | |
59 | buf)) | |
60 | ||
61 | (defun decode-string (buffer coding) | |
62 | (declare (type (array (unsigned-byte 8)) buffer)) | |
63 | (let ((decoder (make-decoder coding)) | |
64 | (buf (make-array (list (length buffer)) :element-type 'character :adjustable t :fill-pointer 0))) | |
65 | (unless (funcall decoder buffer buf) | |
66 | (coding-error buffer (length buffer) buf "~A byte sequence ended prematurely." coding)) | |
67 | buf)) | |
68 | ||
69 | ;;; Gray stream implementation | |
70 | ||
71 | (defclass codec-character-stream (fundamental-character-input-stream fundamental-character-output-stream) | |
72 | ((decoder :initarg :decoder) | |
73 | (encoder :initarg :encoder) | |
74 | (back :initarg :back) | |
75 | (read-pos :initform 0) | |
76 | (buffer :initform (make-array '(64) :element-type 'character :adjustable t :fill-pointer 0)))) | |
77 | ||
78 | (defmethod close ((stream codec-character-stream) &key abort) | |
79 | (with-slots (back) stream | |
80 | (close back :abort abort)) | |
81 | (call-next-method)) | |
82 | ||
83 | (defmethod open-stream-p ((stream codec-character-stream)) | |
84 | (with-slots (back) stream | |
85 | (open-stream-p stream))) | |
86 | ||
87 | (defun ccs-ensure-buffer (stream len) | |
88 | (declare (type codec-character-stream stream) | |
89 | (type integer len)) | |
90 | (with-slots (decoder back buffer read-pos) stream | |
91 | (let ((readbuf (make-array (list len) :element-type '(unsigned-byte 8)))) | |
92 | (loop (unless (< (- (length buffer) read-pos) len) (return t)) | |
93 | (let ((readlen (read-sequence readbuf back))) | |
94 | (when (= readlen 0) | |
95 | (return-from ccs-ensure-buffer nil)) | |
96 | (funcall decoder readbuf buffer :end readlen)))))) | |
97 | ||
98 | (defun ccs-clear-buffer (stream) | |
99 | (declare (type codec-character-stream stream)) | |
100 | (with-slots (read-pos buffer) stream | |
101 | (replace buffer buffer :start2 read-pos) | |
102 | (setf (fill-pointer buffer) (- (fill-pointer buffer) read-pos) | |
103 | read-pos 0))) | |
104 | ||
105 | (defmethod stream-read-char ((stream codec-character-stream)) | |
106 | (unless (ccs-ensure-buffer stream 1) | |
107 | (return-from stream-read-char :eof)) | |
108 | (with-slots (read-pos buffer) stream | |
109 | (prog1 (aref buffer read-pos) | |
110 | (when (>= (incf read-pos) 16) | |
111 | (ccs-clear-buffer stream))))) | |
112 | ||
113 | (defmethod stream-unread-char ((stream codec-character-stream) char) | |
114 | (with-slots (read-pos buffer) stream | |
115 | (when (= read-pos 0) | |
116 | (let ((len (length buffer))) | |
117 | (when (< (array-dimension buffer 0) (+ len 16)) | |
118 | (adjust-array buffer (list (setf (fill-pointer buffer) | |
119 | (+ len 16))))) | |
120 | (replace buffer buffer :start1 16 :end2 len))) | |
121 | (setf (aref buffer read-pos) char) | |
122 | (decf read-pos) | |
123 | nil)) | |
124 | ||
125 | (defun ccs-wont-hang-p (stream) | |
126 | (declare (type codec-character-stream stream)) | |
127 | (with-slots (read-pos back buffer) stream | |
128 | (or (and (< read-pos (length buffer)) (aref buffer read-pos)) | |
129 | (listen back)))) | |
130 | ||
131 | (defmethod stream-read-char-no-hang ((stream codec-character-stream)) | |
132 | (if (ccs-wont-hang-p stream) | |
133 | (stream-read-char stream) | |
134 | nil)) | |
135 | ||
136 | (defmethod stream-peek-char ((stream codec-character-stream)) | |
137 | (unless (ccs-ensure-buffer stream 1) | |
138 | (return-from stream-peek-char :eof)) | |
139 | (with-slots (read-pos buffer) stream | |
140 | (aref buffer read-pos))) | |
141 | ||
142 | (defmethod stream-listen ((stream codec-character-stream)) | |
143 | (if (ccs-wont-hang-p stream) | |
144 | (let ((peek (stream-peek-char stream))) | |
145 | (if (eq peek :eof) | |
146 | nil | |
147 | peek)) | |
148 | nil)) | |
149 | ||
150 | (defmethod stream-write-char ((stream codec-character-stream) char) | |
151 | (with-slots (encoder back) stream | |
152 | (let ((seq (make-array '(1) :element-type 'character :initial-element char)) | |
153 | (outbuf (make-array '(16) :element-type '(unsigned-byte 8) :adjustable t :fill-pointer 0))) | |
154 | (funcall encoder seq outbuf) | |
155 | (write-sequence outbuf back)))) | |
156 | ||
157 | (defmethod stream-finish-output ((stream codec-character-stream)) | |
158 | (finish-output (slot-value stream 'back))) | |
159 | ||
160 | (defmethod stream-force-output ((stream codec-character-stream)) | |
161 | (force-output (slot-value stream 'back))) | |
162 | ||
163 | (defmethod stream-read-sequence ((stream codec-character-stream) seq &optional (start 0) (end (length seq))) | |
164 | (ccs-ensure-buffer stream (- end start)) | |
165 | (with-slots (read-pos buffer) stream | |
166 | (replace seq buffer :start1 start :end1 end :start2 read-pos :end2 (length buffer)) | |
167 | (let ((len (min (- end start) (- (length buffer) read-pos)))) | |
168 | (when (>= (incf read-pos len) 128) | |
169 | (ccs-clear-buffer stream))))) | |
170 | ||
171 | (defmethod stream-write-sequence ((stream codec-character-stream) seq &optional (start 0) (end (length seq))) | |
172 | (with-slots (encoder back) stream | |
173 | (let ((outbuf (make-array (list (- end start)) :element-type '(unsigned-byte 8) :adjustable t :fill-pointer 0))) | |
174 | (funcall encoder seq outbuf) | |
175 | (write-sequence outbuf back)))) | |
176 | ||
177 | ;;; Implementation-specific functions | |
178 | ||
179 | #+(or (and clisp unicode) sbcl) | |
180 | (defun unicode->char (unicode) | |
181 | (declare (type (unsigned-byte 24) unicode)) | |
182 | (code-char unicode)) | |
183 | ||
184 | #+(or (and clisp unicode) sbcl) | |
185 | (defun char->unicode (char) | |
186 | (declare (type character char)) | |
187 | (char-code char)) | |
188 | ||
189 | ;;; Latin-1 | |
190 | ||
191 | (defun decode-latin-1 (byteseq charseq &key (start 0) (end (length byteseq))) | |
192 | (declare (type (array (unsigned-byte 8)) byteseq) | |
193 | (type (array character) charseq) | |
194 | (type fixnum start end)) | |
195 | (do ((i start (1+ i))) | |
196 | ((>= i end)) | |
197 | (vector-push-extend (unicode->char (aref byteseq i)) charseq)) | |
198 | t) | |
199 | ||
200 | (defun encode-latin-1 (charseq byteseq &key (start 0) (end (length charseq))) | |
201 | (declare (type (array (unsigned-byte 8)) byteseq) | |
202 | (type (array character) charseq) | |
203 | (type fixnum start end)) | |
204 | (loop | |
205 | (restart-case | |
206 | (loop | |
207 | (unless (< start end) (return-from encode-latin-1 t)) | |
208 | (vector-push-extend (let ((cp (char->unicode (aref charseq (prog1 start (incf start)))))) | |
209 | (unless (< cp 256) | |
210 | (coding-error charseq start byteseq "ISO-8859-1 cannot encode code-points higher than 256.")) | |
211 | cp) | |
212 | byteseq)) | |
213 | (:replace-char (&optional (replacement #\?)) | |
214 | :report "Replace this character with another." | |
215 | (vector-push-extend (char->unicode replacement) byteseq)) | |
216 | (:skip-char () | |
217 | :report "Ignore this character." | |
218 | nil)))) | |
219 | ||
220 | (define-decoder (latin-1) | |
221 | #'decode-latin-1) | |
222 | ||
223 | (define-encoder (latin-1) | |
224 | #'encode-latin-1) | |
225 | ||
226 | (define-codec-synonyms latin-1 latin1 iso-8859-1) | |
227 | ||
228 | ;;; UTF-8 | |
229 | ||
230 | (defun encode-utf-8 (charseq byteseq &key (start 0) (end (length charseq))) | |
231 | (declare (type (array (unsigned-byte 8)) byteseq) | |
232 | (type (array character) charseq) | |
233 | (type fixnum start end)) | |
234 | (do ((i start (1+ i))) | |
235 | ((>= i end)) | |
236 | (let ((cp (char->unicode (aref charseq i)))) | |
237 | (if (< cp 128) | |
238 | (vector-push-extend cp byteseq) | |
239 | (let ((nbytes 0) | |
240 | (bytes '())) | |
241 | (loop | |
242 | (push (logior (ldb (byte 6 0) cp) #x80) bytes) | |
243 | (setf cp (truncate cp 64)) | |
244 | (incf nbytes) | |
245 | (when (< cp (expt 2 (- 6 nbytes))) | |
246 | (push (logior (logand #xff (lognot (1- (expt 2 (- 7 nbytes))))) | |
247 | cp) | |
248 | bytes) | |
249 | (return))) | |
250 | (dolist (byte bytes) | |
251 | (vector-push-extend byte byteseq)))))) | |
252 | t) | |
253 | ||
254 | (define-encoder (utf-8) | |
255 | #'encode-utf-8) | |
256 | ||
257 | (define-decoder (utf-8) | |
258 | (let ((mbuf 0) | |
259 | (mlen 0)) | |
260 | (flet ((decode (byteseq charseq &key (start 0) (end (length byteseq))) | |
261 | (declare (type (array (unsigned-byte 8)) byteseq) | |
262 | (type (array character) charseq) | |
263 | (type fixnum start end)) | |
264 | (let ((i start)) | |
265 | (flet ((failure (format &rest args) | |
266 | (error 'simple-coding-error | |
267 | :input byteseq :position i :result charseq | |
268 | :format-control format :format-arguments args))) | |
269 | (loop | |
270 | (restart-case | |
271 | (progn | |
272 | (loop | |
273 | (unless (< i end) (return)) | |
274 | (let ((byte (aref byteseq (prog1 i (incf i))))) | |
275 | (if (= mlen 0) | |
276 | (if (< byte 128) | |
277 | (vector-push-extend (unicode->char byte) charseq) | |
278 | (setf mlen (block zero | |
279 | (dotimes (i 7) | |
280 | (when (= (ldb (byte 1 (- 7 i)) byte) 0) | |
281 | (when (< i 2) | |
282 | (failure "UTF-8 sequence started with continuation byte: ~D" byte)) | |
283 | (return-from zero (1- i)))) | |
284 | (failure "Invalid UTF-8 sequence start byte: ~D" byte)) | |
285 | mbuf (ldb (byte (- 6 mlen) 0) byte))) | |
286 | (progn (when (not (= (ldb (byte 2 6) byte) 2)) | |
287 | (failure "Invalid UTF-8 continuation byte: ~D" byte)) | |
288 | (setf mbuf (+ (* mbuf 64) (ldb (byte 6 0) byte))) | |
289 | (when (= (decf mlen) 0) | |
290 | (when (< mbuf 128) | |
291 | (with-simple-restart (:accept "Accept anyway.") | |
292 | (failure "UTF-8 multibyte sequence denoted an ASCII character ~S (either an encoding error or an attempt at breaking security)." (unicode->char mbuf)))) | |
293 | (vector-push-extend (unicode->char mbuf) charseq)))))) | |
294 | (return-from decode (= mlen 0))) | |
295 | (:replace-char (&optional (replacement (unicode->char #xfffd))) | |
296 | :report "Replace the invalid bytes with a character." | |
297 | (vector-push-extend replacement charseq) | |
298 | (loop (unless (and (< i end) (= (ldb (byte 2 6) (aref byteseq i)) 2)) | |
299 | (return)) | |
300 | (incf i)) | |
301 | (setf mlen 0)) | |
302 | (:skip-char () | |
303 | :report "Ignore the invalid byte sequence." | |
304 | (loop (unless (and (< i end) (= (ldb (byte 2 6) (aref byteseq i)) 2)) | |
305 | (return)) | |
306 | (incf i)) | |
307 | (setf mlen 0)))))))) | |
308 | #'decode))) | |
309 | ||
310 | (define-codec-synonyms utf-8 utf8) |