Commit | Line | Data |
---|---|---|
e4598559 MW |
1 | /* Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, 2006, |
2 | * 2007, 2008, 2009, 2010, 2011, 2012, 2013 Free Software Foundation, Inc. | |
f4bc4e59 | 3 | * |
73be1d9e | 4 | * This library is free software; you can redistribute it and/or |
53befeb7 NJ |
5 | * modify it under the terms of the GNU Lesser General Public License |
6 | * as published by the Free Software Foundation; either version 3 of | |
7 | * the License, or (at your option) any later version. | |
0f2d19dd | 8 | * |
53befeb7 NJ |
9 | * This library is distributed in the hope that it will be useful, but |
10 | * WITHOUT ANY WARRANTY; without even the implied warranty of | |
73be1d9e MV |
11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
12 | * Lesser General Public License for more details. | |
0f2d19dd | 13 | * |
73be1d9e MV |
14 | * You should have received a copy of the GNU Lesser General Public |
15 | * License along with this library; if not, write to the Free Software | |
53befeb7 NJ |
16 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA |
17 | * 02110-1301 USA | |
73be1d9e | 18 | */ |
1bbd0b84 | 19 | |
1bbd0b84 | 20 | |
0f2d19dd | 21 | \f |
d68fee48 JB |
22 | /* Headers. */ |
23 | ||
2b829bbb KR |
24 | #define _LARGEFILE64_SOURCE /* ask for stat64 etc */ |
25 | ||
dbb605f5 | 26 | #ifdef HAVE_CONFIG_H |
bd515f37 RB |
27 | # include <config.h> |
28 | #endif | |
29 | ||
0f2d19dd | 30 | #include <stdio.h> |
e6e2e95a | 31 | #include <errno.h> |
8ab3d8a0 | 32 | #include <fcntl.h> /* for chsize on mingw */ |
b5cb4464 | 33 | #include <assert.h> |
f4bc4e59 | 34 | #include <iconv.h> |
889975e5 MG |
35 | #include <uniconv.h> |
36 | #include <unistr.h> | |
37 | #include <striconveh.h> | |
96965a6e | 38 | #include <c-strcase.h> |
e6e2e95a | 39 | |
fca43887 LC |
40 | #include <assert.h> |
41 | ||
a0599745 | 42 | #include "libguile/_scm.h" |
4e047c3e | 43 | #include "libguile/async.h" |
8269ba5b | 44 | #include "libguile/deprecation.h" |
f0942910 | 45 | #include "libguile/eval.h" |
8ab3d8a0 | 46 | #include "libguile/fports.h" /* direct access for seek and truncate */ |
9511876f | 47 | #include "libguile/goops.h" |
a0599745 MD |
48 | #include "libguile/smob.h" |
49 | #include "libguile/chars.h" | |
185e369a | 50 | #include "libguile/dynwind.h" |
0f2d19dd | 51 | |
a0599745 | 52 | #include "libguile/keywords.h" |
5dbc6c06 | 53 | #include "libguile/hashtab.h" |
a0599745 MD |
54 | #include "libguile/root.h" |
55 | #include "libguile/strings.h" | |
b42170a4 | 56 | #include "libguile/mallocs.h" |
a0599745 MD |
57 | #include "libguile/validate.h" |
58 | #include "libguile/ports.h" | |
e4598559 | 59 | #include "libguile/ports-internal.h" |
3a5fb14d | 60 | #include "libguile/vectors.h" |
5dbc6c06 | 61 | #include "libguile/weaks.h" |
9de87eea | 62 | #include "libguile/fluids.h" |
889975e5 | 63 | #include "libguile/eq.h" |
0f2d19dd | 64 | |
bd9e24b3 GH |
65 | #ifdef HAVE_STRING_H |
66 | #include <string.h> | |
67 | #endif | |
68 | ||
ec65f5da MV |
69 | #ifdef HAVE_IO_H |
70 | #include <io.h> | |
71 | #endif | |
72 | ||
0f2d19dd JB |
73 | #ifdef HAVE_UNISTD_H |
74 | #include <unistd.h> | |
75 | #endif | |
76 | ||
95b88819 GH |
77 | #ifdef HAVE_SYS_IOCTL_H |
78 | #include <sys/ioctl.h> | |
79 | #endif | |
d68fee48 | 80 | |
8ab3d8a0 KR |
81 | /* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize |
82 | already, but have this code here in case that wasn't so in past versions, | |
83 | or perhaps to help other minimal DOS environments. | |
84 | ||
85 | gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which | |
86 | might be possibilities if we've got other systems without ftruncate. */ | |
87 | ||
56a3dcd4 | 88 | #if defined HAVE_CHSIZE && ! defined HAVE_FTRUNCATE |
82893676 | 89 | #define ftruncate(fd, size) chsize (fd, size) |
8ab3d8a0 KR |
90 | #undef HAVE_FTRUNCATE |
91 | #define HAVE_FTRUNCATE 1 | |
82893676 MG |
92 | #endif |
93 | ||
0f2d19dd | 94 | \f |
d68fee48 | 95 | /* The port kind table --- a dynamically resized array of port types. */ |
0f2d19dd JB |
96 | |
97 | ||
98 | /* scm_ptobs scm_numptob | |
5dbc6c06 | 99 | * implement a dynamically resized array of ptob records. |
0f2d19dd JB |
100 | * Indexes into this table are used when generating type |
101 | * tags for smobjects (if you know a tag you can get an index and conversely). | |
102 | */ | |
dd3a26f3 AW |
103 | scm_t_ptob_descriptor *scm_ptobs = NULL; |
104 | long scm_numptob = 0; | |
0f2d19dd | 105 | |
ee149d03 | 106 | /* GC marker for a port with stream of SCM type. */ |
0f2d19dd | 107 | SCM |
a284e297 | 108 | scm_markstream (SCM ptr) |
0f2d19dd JB |
109 | { |
110 | int openp; | |
f9a64404 | 111 | openp = SCM_CELL_WORD_0 (ptr) & SCM_OPN; |
0f2d19dd | 112 | if (openp) |
74a16888 | 113 | return SCM_PACK (SCM_STREAM (ptr)); |
0f2d19dd JB |
114 | else |
115 | return SCM_BOOL_F; | |
116 | } | |
117 | ||
f12733c9 | 118 | /* |
f12733c9 | 119 | * We choose to use an interface similar to the smob interface with |
affc96b5 | 120 | * fill_input and write as standard fields, passed to the port |
f12733c9 MD |
121 | * type constructor, and optional fields set by setters. |
122 | */ | |
123 | ||
70df8af6 | 124 | static void |
e81d98ec | 125 | flush_port_default (SCM port SCM_UNUSED) |
70df8af6 GH |
126 | { |
127 | } | |
128 | ||
129 | static void | |
e81d98ec | 130 | end_input_default (SCM port SCM_UNUSED, int offset SCM_UNUSED) |
70df8af6 GH |
131 | { |
132 | } | |
0f2d19dd | 133 | |
92c2555f | 134 | scm_t_bits |
f12733c9 | 135 | scm_make_port_type (char *name, |
affc96b5 | 136 | int (*fill_input) (SCM port), |
8aa011a1 | 137 | void (*write) (SCM port, const void *data, size_t size)) |
0f2d19dd JB |
138 | { |
139 | char *tmp; | |
0953b549 | 140 | if (SCM_I_MAX_PORT_TYPE_COUNT - 1 <= scm_numptob) |
0f2d19dd | 141 | goto ptoberr; |
9de87eea | 142 | SCM_CRITICAL_SECTION_START; |
fca43887 LC |
143 | tmp = (char *) scm_gc_realloc ((char *) scm_ptobs, |
144 | scm_numptob * sizeof (scm_t_ptob_descriptor), | |
145 | (1 + scm_numptob) | |
146 | * sizeof (scm_t_ptob_descriptor), | |
147 | "port-type"); | |
0f2d19dd JB |
148 | if (tmp) |
149 | { | |
92c2555f | 150 | scm_ptobs = (scm_t_ptob_descriptor *) tmp; |
affc96b5 | 151 | |
f12733c9 MD |
152 | scm_ptobs[scm_numptob].name = name; |
153 | scm_ptobs[scm_numptob].mark = 0; | |
3051344b | 154 | scm_ptobs[scm_numptob].free = NULL; |
f12733c9 MD |
155 | scm_ptobs[scm_numptob].print = scm_port_print; |
156 | scm_ptobs[scm_numptob].equalp = 0; | |
affc96b5 GH |
157 | scm_ptobs[scm_numptob].close = 0; |
158 | ||
159 | scm_ptobs[scm_numptob].write = write; | |
70df8af6 | 160 | scm_ptobs[scm_numptob].flush = flush_port_default; |
affc96b5 | 161 | |
70df8af6 | 162 | scm_ptobs[scm_numptob].end_input = end_input_default; |
affc96b5 GH |
163 | scm_ptobs[scm_numptob].fill_input = fill_input; |
164 | scm_ptobs[scm_numptob].input_waiting = 0; | |
165 | ||
f12733c9 | 166 | scm_ptobs[scm_numptob].seek = 0; |
affc96b5 GH |
167 | scm_ptobs[scm_numptob].truncate = 0; |
168 | ||
0f2d19dd JB |
169 | scm_numptob++; |
170 | } | |
9de87eea | 171 | SCM_CRITICAL_SECTION_END; |
0f2d19dd | 172 | if (!tmp) |
2500356c DH |
173 | { |
174 | ptoberr: | |
175 | scm_memory_error ("scm_make_port_type"); | |
176 | } | |
f12733c9 | 177 | /* Make a class object if Goops is present */ |
63385df2 | 178 | if (SCM_UNPACK (scm_port_class[0]) != 0) |
f12733c9 | 179 | scm_make_port_classes (scm_numptob - 1, SCM_PTOBNAME (scm_numptob - 1)); |
0f2d19dd JB |
180 | return scm_tc7_port + (scm_numptob - 1) * 256; |
181 | } | |
182 | ||
f12733c9 | 183 | void |
23f2b9a3 | 184 | scm_set_port_mark (scm_t_bits tc, SCM (*mark) (SCM)) |
f12733c9 MD |
185 | { |
186 | scm_ptobs[SCM_TC2PTOBNUM (tc)].mark = mark; | |
187 | } | |
188 | ||
189 | void | |
23f2b9a3 | 190 | scm_set_port_free (scm_t_bits tc, size_t (*free) (SCM)) |
f12733c9 MD |
191 | { |
192 | scm_ptobs[SCM_TC2PTOBNUM (tc)].free = free; | |
193 | } | |
194 | ||
195 | void | |
23f2b9a3 | 196 | scm_set_port_print (scm_t_bits tc, int (*print) (SCM exp, SCM port, |
f12733c9 MD |
197 | scm_print_state *pstate)) |
198 | { | |
199 | scm_ptobs[SCM_TC2PTOBNUM (tc)].print = print; | |
200 | } | |
201 | ||
202 | void | |
23f2b9a3 | 203 | scm_set_port_equalp (scm_t_bits tc, SCM (*equalp) (SCM, SCM)) |
f12733c9 MD |
204 | { |
205 | scm_ptobs[SCM_TC2PTOBNUM (tc)].equalp = equalp; | |
206 | } | |
207 | ||
31703ab8 | 208 | void |
23f2b9a3 | 209 | scm_set_port_flush (scm_t_bits tc, void (*flush) (SCM port)) |
31703ab8 | 210 | { |
affc96b5 | 211 | scm_ptobs[SCM_TC2PTOBNUM (tc)].flush = flush; |
31703ab8 GH |
212 | } |
213 | ||
f12733c9 | 214 | void |
23f2b9a3 | 215 | scm_set_port_end_input (scm_t_bits tc, void (*end_input) (SCM port, int offset)) |
f12733c9 | 216 | { |
affc96b5 | 217 | scm_ptobs[SCM_TC2PTOBNUM (tc)].end_input = end_input; |
f12733c9 MD |
218 | } |
219 | ||
220 | void | |
23f2b9a3 | 221 | scm_set_port_close (scm_t_bits tc, int (*close) (SCM)) |
f12733c9 | 222 | { |
affc96b5 | 223 | scm_ptobs[SCM_TC2PTOBNUM (tc)].close = close; |
f12733c9 MD |
224 | } |
225 | ||
226 | void | |
f1ce9199 LC |
227 | scm_set_port_seek (scm_t_bits tc, |
228 | scm_t_off (*seek) (SCM, scm_t_off, int)) | |
f12733c9 MD |
229 | { |
230 | scm_ptobs[SCM_TC2PTOBNUM (tc)].seek = seek; | |
231 | } | |
232 | ||
233 | void | |
f1ce9199 | 234 | scm_set_port_truncate (scm_t_bits tc, void (*truncate) (SCM, scm_t_off)) |
f12733c9 | 235 | { |
affc96b5 | 236 | scm_ptobs[SCM_TC2PTOBNUM (tc)].truncate = truncate; |
f12733c9 MD |
237 | } |
238 | ||
239 | void | |
23f2b9a3 | 240 | scm_set_port_input_waiting (scm_t_bits tc, int (*input_waiting) (SCM)) |
f12733c9 | 241 | { |
affc96b5 | 242 | scm_ptobs[SCM_TC2PTOBNUM (tc)].input_waiting = input_waiting; |
f12733c9 MD |
243 | } |
244 | ||
45c0878b MW |
245 | static void |
246 | scm_i_set_pending_eof (SCM port) | |
247 | { | |
248 | SCM_PORT_GET_INTERNAL (port)->pending_eof = 1; | |
249 | } | |
250 | ||
251 | static void | |
252 | scm_i_clear_pending_eof (SCM port) | |
253 | { | |
254 | SCM_PORT_GET_INTERNAL (port)->pending_eof = 0; | |
255 | } | |
256 | ||
05d7f762 MW |
257 | SCM |
258 | scm_i_port_alist (SCM port) | |
259 | { | |
260 | return SCM_PORT_GET_INTERNAL (port)->alist; | |
261 | } | |
262 | ||
263 | void | |
264 | scm_i_set_port_alist_x (SCM port, SCM alist) | |
265 | { | |
266 | SCM_PORT_GET_INTERNAL (port)->alist = alist; | |
267 | } | |
268 | ||
0f2d19dd | 269 | \f |
0f2d19dd | 270 | |
3b3b36dd | 271 | SCM_DEFINE (scm_char_ready_p, "char-ready?", 0, 1, 0, |
1e6808ea MG |
272 | (SCM port), |
273 | "Return @code{#t} if a character is ready on input @var{port}\n" | |
274 | "and return @code{#f} otherwise. If @code{char-ready?} returns\n" | |
275 | "@code{#t} then the next @code{read-char} operation on\n" | |
276 | "@var{port} is guaranteed not to hang. If @var{port} is a file\n" | |
277 | "port at end of file then @code{char-ready?} returns @code{#t}.\n" | |
c2dfff19 KR |
278 | "\n" |
279 | "@code{char-ready?} exists to make it possible for a\n" | |
1e6808ea MG |
280 | "program to accept characters from interactive ports without\n" |
281 | "getting stuck waiting for input. Any input editors associated\n" | |
282 | "with such ports must make sure that characters whose existence\n" | |
283 | "has been asserted by @code{char-ready?} cannot be rubbed out.\n" | |
284 | "If @code{char-ready?} were to return @code{#f} at end of file,\n" | |
285 | "a port at end of file would be indistinguishable from an\n" | |
c2dfff19 | 286 | "interactive port that has no ready characters.") |
1bbd0b84 | 287 | #define FUNC_NAME s_scm_char_ready_p |
0f2d19dd | 288 | { |
92c2555f | 289 | scm_t_port *pt; |
6c951427 | 290 | |
0f2d19dd | 291 | if (SCM_UNBNDP (port)) |
9de87eea | 292 | port = scm_current_input_port (); |
b2456dd4 AW |
293 | /* It's possible to close the current input port, so validate even in |
294 | this case. */ | |
295 | SCM_VALIDATE_OPINPORT (1, port); | |
d68fee48 | 296 | |
ae4c4016 JB |
297 | pt = SCM_PTAB_ENTRY (port); |
298 | ||
6c951427 GH |
299 | /* if the current read buffer is filled, or the |
300 | last pushed-back char has been read and the saved buffer is | |
301 | filled, result is true. */ | |
302 | if (pt->read_pos < pt->read_end | |
303 | || (pt->read_buf == pt->putback_buf | |
304 | && pt->saved_read_pos < pt->saved_read_end)) | |
0f2d19dd | 305 | return SCM_BOOL_T; |
ee149d03 JB |
306 | else |
307 | { | |
92c2555f | 308 | scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)]; |
ee149d03 | 309 | |
affc96b5 | 310 | if (ptob->input_waiting) |
7888309b | 311 | return scm_from_bool(ptob->input_waiting (port)); |
ee149d03 | 312 | else |
6c951427 | 313 | return SCM_BOOL_T; |
ee149d03 | 314 | } |
0f2d19dd | 315 | } |
1bbd0b84 | 316 | #undef FUNC_NAME |
0f2d19dd | 317 | |
e8b21eec LC |
318 | /* Move up to READ_LEN bytes from PORT's putback and/or read buffers |
319 | into memory starting at DEST. Return the number of bytes moved. | |
320 | PORT's line/column numbers are left unchanged. */ | |
321 | size_t | |
322 | scm_take_from_input_buffers (SCM port, char *dest, size_t read_len) | |
c2da2648 | 323 | { |
92c2555f | 324 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
e8b21eec | 325 | size_t bytes_read = 0; |
c2da2648 GH |
326 | size_t from_buf = min (pt->read_end - pt->read_pos, read_len); |
327 | ||
328 | if (from_buf > 0) | |
329 | { | |
330 | memcpy (dest, pt->read_pos, from_buf); | |
331 | pt->read_pos += from_buf; | |
e8b21eec | 332 | bytes_read += from_buf; |
c2da2648 GH |
333 | read_len -= from_buf; |
334 | dest += from_buf; | |
335 | } | |
336 | ||
337 | /* if putback was active, try the real input buffer too. */ | |
338 | if (pt->read_buf == pt->putback_buf) | |
339 | { | |
340 | from_buf = min (pt->saved_read_end - pt->saved_read_pos, read_len); | |
341 | if (from_buf > 0) | |
342 | { | |
343 | memcpy (dest, pt->saved_read_pos, from_buf); | |
344 | pt->saved_read_pos += from_buf; | |
e8b21eec | 345 | bytes_read += from_buf; |
c2da2648 GH |
346 | } |
347 | } | |
e8b21eec LC |
348 | |
349 | return bytes_read; | |
c2da2648 GH |
350 | } |
351 | ||
6c951427 | 352 | /* Clear a port's read buffers, returning the contents. */ |
a1ec6916 | 353 | SCM_DEFINE (scm_drain_input, "drain-input", 1, 0, 0, |
1bbd0b84 | 354 | (SCM port), |
4a151b3d GH |
355 | "This procedure clears a port's input buffers, similar\n" |
356 | "to the way that force-output clears the output buffer. The\n" | |
357 | "contents of the buffers are returned as a single string, e.g.,\n" | |
358 | "\n" | |
359 | "@lisp\n" | |
360 | "(define p (open-input-file ...))\n" | |
361 | "(drain-input p) => empty string, nothing buffered yet.\n" | |
362 | "(unread-char (read-char p) p)\n" | |
363 | "(drain-input p) => initial chars from p, up to the buffer size.\n" | |
364 | "@end lisp\n\n" | |
365 | "Draining the buffers may be useful for cleanly finishing\n" | |
366 | "buffered I/O so that the file descriptor can be used directly\n" | |
367 | "for further input.") | |
1bbd0b84 | 368 | #define FUNC_NAME s_scm_drain_input |
ee149d03 | 369 | { |
840ae05d | 370 | SCM result; |
3a5fb14d | 371 | char *data; |
28a6e1b0 | 372 | scm_t_port *pt; |
c014a02e | 373 | long count; |
ee149d03 | 374 | |
34d19ef6 | 375 | SCM_VALIDATE_OPINPORT (1, port); |
28a6e1b0 | 376 | pt = SCM_PTAB_ENTRY (port); |
840ae05d | 377 | |
6c951427 GH |
378 | count = pt->read_end - pt->read_pos; |
379 | if (pt->read_buf == pt->putback_buf) | |
380 | count += pt->saved_read_end - pt->saved_read_pos; | |
840ae05d | 381 | |
67a72dc1 AW |
382 | if (count) |
383 | { | |
190d4b0d | 384 | result = scm_i_make_string (count, &data, 0); |
67a72dc1 AW |
385 | scm_take_from_input_buffers (port, data, count); |
386 | } | |
387 | else | |
388 | result = scm_nullstr; | |
389 | ||
840ae05d | 390 | return result; |
ee149d03 | 391 | } |
1bbd0b84 | 392 | #undef FUNC_NAME |
0f2d19dd JB |
393 | |
394 | \f | |
d68fee48 | 395 | /* Standard ports --- current input, output, error, and more(!). */ |
0f2d19dd | 396 | |
34297700 AW |
397 | static SCM cur_inport_fluid = SCM_BOOL_F; |
398 | static SCM cur_outport_fluid = SCM_BOOL_F; | |
399 | static SCM cur_errport_fluid = SCM_BOOL_F; | |
400 | static SCM cur_loadport_fluid = SCM_BOOL_F; | |
9de87eea | 401 | |
3b3b36dd | 402 | SCM_DEFINE (scm_current_input_port, "current-input-port", 0, 0, 0, |
e1546b65 MG |
403 | (), |
404 | "Return the current input port. This is the default port used\n" | |
405 | "by many input procedures. Initially, @code{current-input-port}\n" | |
406 | "returns the @dfn{standard input} in Unix and C terminology.") | |
1bbd0b84 | 407 | #define FUNC_NAME s_scm_current_input_port |
0f2d19dd | 408 | { |
34297700 | 409 | if (scm_is_true (cur_inport_fluid)) |
889975e5 MG |
410 | return scm_fluid_ref (cur_inport_fluid); |
411 | else | |
412 | return SCM_BOOL_F; | |
0f2d19dd | 413 | } |
1bbd0b84 | 414 | #undef FUNC_NAME |
0f2d19dd | 415 | |
3b3b36dd | 416 | SCM_DEFINE (scm_current_output_port, "current-output-port", 0, 0, 0, |
e1546b65 MG |
417 | (), |
418 | "Return the current output port. This is the default port used\n" | |
9401323e | 419 | "by many output procedures. Initially,\n" |
e1546b65 MG |
420 | "@code{current-output-port} returns the @dfn{standard output} in\n" |
421 | "Unix and C terminology.") | |
1bbd0b84 | 422 | #define FUNC_NAME s_scm_current_output_port |
0f2d19dd | 423 | { |
34297700 | 424 | if (scm_is_true (cur_outport_fluid)) |
889975e5 MG |
425 | return scm_fluid_ref (cur_outport_fluid); |
426 | else | |
427 | return SCM_BOOL_F; | |
0f2d19dd | 428 | } |
1bbd0b84 | 429 | #undef FUNC_NAME |
0f2d19dd | 430 | |
3b3b36dd | 431 | SCM_DEFINE (scm_current_error_port, "current-error-port", 0, 0, 0, |
1bbd0b84 | 432 | (), |
b380b885 MD |
433 | "Return the port to which errors and warnings should be sent (the\n" |
434 | "@dfn{standard error} in Unix and C terminology).") | |
1bbd0b84 | 435 | #define FUNC_NAME s_scm_current_error_port |
0f2d19dd | 436 | { |
34297700 | 437 | if (scm_is_true (cur_errport_fluid)) |
889975e5 MG |
438 | return scm_fluid_ref (cur_errport_fluid); |
439 | else | |
440 | return SCM_BOOL_F; | |
0f2d19dd | 441 | } |
1bbd0b84 | 442 | #undef FUNC_NAME |
0f2d19dd | 443 | |
3972de76 AW |
444 | SCM |
445 | scm_current_warning_port (void) | |
446 | { | |
f57ea23a MW |
447 | static SCM cwp_var = SCM_UNDEFINED; |
448 | static scm_i_pthread_mutex_t cwp_var_mutex | |
449 | = SCM_I_PTHREAD_MUTEX_INITIALIZER; | |
450 | ||
451 | scm_i_scm_pthread_mutex_lock (&cwp_var_mutex); | |
452 | if (SCM_UNBNDP (cwp_var)) | |
453 | cwp_var = scm_c_private_variable ("guile", "current-warning-port"); | |
454 | scm_i_pthread_mutex_unlock (&cwp_var_mutex); | |
3972de76 AW |
455 | |
456 | return scm_call_0 (scm_variable_ref (cwp_var)); | |
457 | } | |
458 | ||
3b3b36dd | 459 | SCM_DEFINE (scm_current_load_port, "current-load-port", 0, 0, 0, |
e1546b65 | 460 | (), |
b450f070 | 461 | "Return the current-load-port.\n" |
e1546b65 | 462 | "The load port is used internally by @code{primitive-load}.") |
1bbd0b84 | 463 | #define FUNC_NAME s_scm_current_load_port |
31614d8e | 464 | { |
9de87eea | 465 | return scm_fluid_ref (cur_loadport_fluid); |
31614d8e | 466 | } |
1bbd0b84 | 467 | #undef FUNC_NAME |
31614d8e | 468 | |
3b3b36dd | 469 | SCM_DEFINE (scm_set_current_input_port, "set-current-input-port", 1, 0, 0, |
1bbd0b84 | 470 | (SCM port), |
8f85c0c6 NJ |
471 | "@deffnx {Scheme Procedure} set-current-output-port port\n" |
472 | "@deffnx {Scheme Procedure} set-current-error-port port\n" | |
b380b885 MD |
473 | "Change the ports returned by @code{current-input-port},\n" |
474 | "@code{current-output-port} and @code{current-error-port}, respectively,\n" | |
475 | "so that they use the supplied @var{port} for input or output.") | |
1bbd0b84 | 476 | #define FUNC_NAME s_scm_set_current_input_port |
0f2d19dd | 477 | { |
9de87eea | 478 | SCM oinp = scm_fluid_ref (cur_inport_fluid); |
34d19ef6 | 479 | SCM_VALIDATE_OPINPORT (1, port); |
9de87eea | 480 | scm_fluid_set_x (cur_inport_fluid, port); |
0f2d19dd JB |
481 | return oinp; |
482 | } | |
1bbd0b84 | 483 | #undef FUNC_NAME |
0f2d19dd JB |
484 | |
485 | ||
3b3b36dd | 486 | SCM_DEFINE (scm_set_current_output_port, "set-current-output-port", 1, 0, 0, |
e1546b65 MG |
487 | (SCM port), |
488 | "Set the current default output port to @var{port}.") | |
1bbd0b84 | 489 | #define FUNC_NAME s_scm_set_current_output_port |
0f2d19dd | 490 | { |
9de87eea | 491 | SCM ooutp = scm_fluid_ref (cur_outport_fluid); |
78446828 | 492 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 493 | SCM_VALIDATE_OPOUTPORT (1, port); |
9de87eea | 494 | scm_fluid_set_x (cur_outport_fluid, port); |
0f2d19dd JB |
495 | return ooutp; |
496 | } | |
1bbd0b84 | 497 | #undef FUNC_NAME |
0f2d19dd JB |
498 | |
499 | ||
3b3b36dd | 500 | SCM_DEFINE (scm_set_current_error_port, "set-current-error-port", 1, 0, 0, |
e1546b65 MG |
501 | (SCM port), |
502 | "Set the current default error port to @var{port}.") | |
1bbd0b84 | 503 | #define FUNC_NAME s_scm_set_current_error_port |
0f2d19dd | 504 | { |
9de87eea | 505 | SCM oerrp = scm_fluid_ref (cur_errport_fluid); |
78446828 | 506 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 507 | SCM_VALIDATE_OPOUTPORT (1, port); |
9de87eea | 508 | scm_fluid_set_x (cur_errport_fluid, port); |
0f2d19dd JB |
509 | return oerrp; |
510 | } | |
1bbd0b84 | 511 | #undef FUNC_NAME |
0f2d19dd | 512 | |
3972de76 AW |
513 | |
514 | SCM | |
515 | scm_set_current_warning_port (SCM port) | |
516 | { | |
517 | static SCM cwp_var = SCM_BOOL_F; | |
518 | ||
519 | if (scm_is_false (cwp_var)) | |
520 | cwp_var = scm_c_private_lookup ("guile", "current-warning-port"); | |
521 | ||
522 | return scm_call_1 (scm_variable_ref (cwp_var), port); | |
523 | } | |
524 | ||
525 | ||
185e369a | 526 | void |
661ae7ab | 527 | scm_dynwind_current_input_port (SCM port) |
9de87eea | 528 | #define FUNC_NAME NULL |
185e369a | 529 | { |
9de87eea | 530 | SCM_VALIDATE_OPINPORT (1, port); |
661ae7ab | 531 | scm_dynwind_fluid (cur_inport_fluid, port); |
185e369a | 532 | } |
9de87eea | 533 | #undef FUNC_NAME |
185e369a MV |
534 | |
535 | void | |
661ae7ab | 536 | scm_dynwind_current_output_port (SCM port) |
9de87eea | 537 | #define FUNC_NAME NULL |
185e369a | 538 | { |
9de87eea MV |
539 | port = SCM_COERCE_OUTPORT (port); |
540 | SCM_VALIDATE_OPOUTPORT (1, port); | |
661ae7ab | 541 | scm_dynwind_fluid (cur_outport_fluid, port); |
185e369a | 542 | } |
9de87eea | 543 | #undef FUNC_NAME |
185e369a MV |
544 | |
545 | void | |
661ae7ab | 546 | scm_dynwind_current_error_port (SCM port) |
9de87eea MV |
547 | #define FUNC_NAME NULL |
548 | { | |
549 | port = SCM_COERCE_OUTPORT (port); | |
550 | SCM_VALIDATE_OPOUTPORT (1, port); | |
661ae7ab | 551 | scm_dynwind_fluid (cur_errport_fluid, port); |
9de87eea MV |
552 | } |
553 | #undef FUNC_NAME | |
554 | ||
555 | void | |
661ae7ab | 556 | scm_i_dynwind_current_load_port (SCM port) |
185e369a | 557 | { |
661ae7ab | 558 | scm_dynwind_fluid (cur_loadport_fluid, port); |
185e369a MV |
559 | } |
560 | ||
0f2d19dd | 561 | \f |
840ae05d | 562 | /* The port table --- an array of pointers to ports. */ |
0f2d19dd | 563 | |
5dbc6c06 HWN |
564 | /* |
565 | We need a global registry of ports to flush them all at exit, and to | |
05d7f762 | 566 | get all the ports matching a file descriptor. |
5dbc6c06 HWN |
567 | */ |
568 | SCM scm_i_port_weak_hash; | |
0f2d19dd | 569 | |
9de87eea | 570 | scm_i_pthread_mutex_t scm_i_port_table_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER; |
b9ad392e | 571 | |
651a0735 LC |
572 | \f |
573 | /* Port finalization. */ | |
574 | ||
1cc91f1b | 575 | |
6922d92f | 576 | static void finalize_port (void *, void *); |
651a0735 | 577 | |
f4bc4e59 | 578 | /* Register a finalizer for PORT. */ |
651a0735 LC |
579 | static SCM_C_INLINE_KEYWORD void |
580 | register_finalizer_for_port (SCM port) | |
581 | { | |
337edc59 MW |
582 | /* Register a finalizer for PORT so that its |
583 | type's `free' function gets called. */ | |
75ba64d6 | 584 | scm_i_set_finalizer (SCM2PTR (port), finalize_port, NULL); |
651a0735 LC |
585 | } |
586 | ||
587 | /* Finalize the object (a port) pointed to by PTR. */ | |
588 | static void | |
6922d92f | 589 | finalize_port (void *ptr, void *data) |
651a0735 LC |
590 | { |
591 | long port_type; | |
592 | SCM port = PTR2SCM (ptr); | |
593 | ||
594 | if (!SCM_PORTP (port)) | |
595 | abort (); | |
596 | ||
597 | if (SCM_OPENP (port)) | |
598 | { | |
599 | if (SCM_REVEALED (port) > 0) | |
600 | /* Keep "revealed" ports alive and re-register a finalizer. */ | |
601 | register_finalizer_for_port (port); | |
602 | else | |
603 | { | |
604 | port_type = SCM_TC2PTOBNUM (SCM_CELL_TYPE (port)); | |
605 | if (port_type >= scm_numptob) | |
606 | abort (); | |
607 | ||
608 | if (scm_ptobs[port_type].free) | |
609 | /* Yes, I really do mean `.free' rather than `.close'. `.close' | |
610 | is for explicit `close-port' by user. */ | |
611 | scm_ptobs[port_type].free (port); | |
612 | ||
613 | SCM_SETSTREAM (port, 0); | |
614 | SCM_CLR_PORT_OPEN_FLAG (port); | |
651a0735 LC |
615 | |
616 | scm_gc_ports_collected++; | |
617 | } | |
618 | } | |
619 | } | |
620 | ||
621 | ||
622 | ||
623 | \f | |
624 | ||
625 | /* This function is not and should not be thread safe. */ | |
da220f27 HWN |
626 | SCM |
627 | scm_new_port_table_entry (scm_t_bits tag) | |
402788a9 | 628 | #define FUNC_NAME "scm_new_port_table_entry" |
0f2d19dd | 629 | { |
85835e59 HWN |
630 | /* |
631 | We initialize the cell to empty, this is in case scm_gc_calloc | |
632 | triggers GC ; we don't want the GC to scan a half-finished Z. | |
633 | */ | |
634 | ||
67329a9e | 635 | SCM z = scm_cons (SCM_EOL, SCM_EOL); |
e4598559 MW |
636 | scm_t_port *entry = scm_gc_typed_calloc (scm_t_port); |
637 | scm_t_port_internal *pti = scm_gc_typed_calloc (scm_t_port_internal); | |
337edc59 | 638 | const char *encoding; |
5f16b897 | 639 | |
840ae05d | 640 | entry->file_name = SCM_BOOL_F; |
61e452ba | 641 | entry->rw_active = SCM_PORT_NEITHER; |
5dbc6c06 | 642 | entry->port = z; |
9d9c66ba | 643 | |
889975e5 MG |
644 | /* Initialize this port with the thread's current default |
645 | encoding. */ | |
337edc59 | 646 | encoding = scm_i_default_port_encoding (); |
b22e94db | 647 | entry->ilseq_handler = scm_i_default_port_conversion_handler (); |
337edc59 | 648 | entry->encoding = encoding ? scm_gc_strdup (encoding, "port") : NULL; |
96965a6e | 649 | if (encoding && c_strcasecmp (encoding, "UTF-8") == 0) |
337edc59 MW |
650 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8; |
651 | else | |
652 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV; | |
653 | pti->iconv_descriptors = NULL; | |
5f16b897 | 654 | |
cdd3d6c9 MW |
655 | pti->at_stream_start_for_bom_read = 1; |
656 | pti->at_stream_start_for_bom_write = 1; | |
657 | ||
e4598559 MW |
658 | /* XXX These fields are not what they seem. They have been |
659 | repurposed, but cannot safely be renamed in 2.0 without breaking | |
660 | ABI compatibility. This will be cleaned up in 2.2. */ | |
661 | entry->input_cd = pti; /* XXX pointer to the internal port structure */ | |
662 | entry->output_cd = NULL; /* XXX unused */ | |
663 | ||
45c0878b | 664 | pti->pending_eof = 0; |
05d7f762 MW |
665 | pti->alist = SCM_EOL; |
666 | ||
5dbc6c06 HWN |
667 | SCM_SET_CELL_TYPE (z, tag); |
668 | SCM_SETPTAB_ENTRY (z, entry); | |
840ae05d | 669 | |
05d7f762 | 670 | scm_hashq_set_x (scm_i_port_weak_hash, z, SCM_BOOL_F); |
651a0735 LC |
671 | |
672 | /* For each new port, register a finalizer so that it port type's free | |
673 | function can be invoked eventually. */ | |
674 | register_finalizer_for_port (z); | |
675 | ||
da220f27 | 676 | return z; |
0f2d19dd | 677 | } |
c6c79933 | 678 | #undef FUNC_NAME |
0f2d19dd | 679 | |
67329a9e | 680 | #if SCM_ENABLE_DEPRECATED==1 |
8269ba5b | 681 | scm_t_port * |
67329a9e HWN |
682 | scm_add_to_port_table (SCM port) |
683 | { | |
8269ba5b AW |
684 | SCM z; |
685 | scm_t_port * pt; | |
67329a9e | 686 | |
8269ba5b AW |
687 | scm_c_issue_deprecation_warning ("scm_add_to_port_table is deprecated."); |
688 | ||
689 | scm_i_pthread_mutex_lock (&scm_i_port_table_mutex); | |
690 | z = scm_new_port_table_entry (scm_tc7_port); | |
691 | pt = SCM_PTAB_ENTRY(z); | |
67329a9e | 692 | pt->port = port; |
5dbc6c06 HWN |
693 | SCM_SETCAR (z, SCM_EOL); |
694 | SCM_SETCDR (z, SCM_EOL); | |
85835e59 | 695 | SCM_SETPTAB_ENTRY (port, pt); |
8269ba5b AW |
696 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); |
697 | ||
67329a9e HWN |
698 | return pt; |
699 | } | |
700 | #endif | |
701 | ||
702 | ||
6c951427 | 703 | /* Remove a port from the table and destroy it. */ |
b9ad392e | 704 | |
337edc59 MW |
705 | static void close_iconv_descriptors (scm_t_iconv_descriptors *id); |
706 | ||
8269ba5b | 707 | static void |
5dbc6c06 HWN |
708 | scm_i_remove_port (SCM port) |
709 | #define FUNC_NAME "scm_remove_port" | |
0f2d19dd | 710 | { |
8269ba5b | 711 | scm_t_port *p; |
e4598559 | 712 | scm_t_port_internal *pti; |
682d78d0 | 713 | |
8269ba5b | 714 | scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex); |
682d78d0 | 715 | |
8269ba5b | 716 | p = SCM_PTAB_ENTRY (port); |
e4598559 | 717 | pti = SCM_PORT_GET_INTERNAL (port); |
8269ba5b | 718 | scm_port_non_buffer (p); |
682d78d0 LC |
719 | p->putback_buf = NULL; |
720 | p->putback_buf_size = 0; | |
3e05fc04 | 721 | |
337edc59 | 722 | if (pti->iconv_descriptors) |
3e05fc04 | 723 | { |
337edc59 MW |
724 | close_iconv_descriptors (pti->iconv_descriptors); |
725 | pti->iconv_descriptors = NULL; | |
3e05fc04 AW |
726 | } |
727 | ||
0f2d19dd | 728 | SCM_SETPTAB_ENTRY (port, 0); |
8269ba5b | 729 | |
5dbc6c06 | 730 | scm_hashq_remove_x (scm_i_port_weak_hash, port); |
8269ba5b AW |
731 | |
732 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); | |
0f2d19dd | 733 | } |
db4b4ca6 DH |
734 | #undef FUNC_NAME |
735 | ||
0f2d19dd | 736 | |
b450f070 | 737 | /* Functions for debugging. */ |
5dbc6c06 | 738 | #ifdef GUILE_DEBUG |
3b3b36dd | 739 | SCM_DEFINE (scm_pt_size, "pt-size", 0, 0, 0, |
b450f070 | 740 | (), |
1e6808ea | 741 | "Return the number of ports in the port table. @code{pt-size}\n" |
5352393c | 742 | "is only included in @code{--enable-guile-debug} builds.") |
1bbd0b84 | 743 | #define FUNC_NAME s_scm_pt_size |
0f2d19dd | 744 | { |
5dbc6c06 | 745 | return scm_from_int (SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash)); |
0f2d19dd | 746 | } |
1bbd0b84 | 747 | #undef FUNC_NAME |
0f2d19dd JB |
748 | #endif |
749 | ||
70df8af6 | 750 | void |
92c2555f | 751 | scm_port_non_buffer (scm_t_port *pt) |
70df8af6 GH |
752 | { |
753 | pt->read_pos = pt->read_buf = pt->read_end = &pt->shortbuf; | |
754 | pt->write_buf = pt->write_pos = &pt->shortbuf; | |
755 | pt->read_buf_size = pt->write_buf_size = 1; | |
756 | pt->write_end = pt->write_buf + pt->write_buf_size; | |
757 | } | |
0f2d19dd | 758 | |
d68fee48 JB |
759 | \f |
760 | /* Revealed counts --- an oddity inherited from SCSH. */ | |
761 | ||
8b13c6b3 GH |
762 | /* Find a port in the table and return its revealed count. |
763 | Also used by the garbage collector. | |
0f2d19dd | 764 | */ |
1cc91f1b | 765 | |
0f2d19dd | 766 | int |
a284e297 | 767 | scm_revealed_count (SCM port) |
0f2d19dd JB |
768 | { |
769 | return SCM_REVEALED(port); | |
770 | } | |
771 | ||
772 | ||
773 | ||
774 | /* Return the revealed count for a port. */ | |
775 | ||
3b3b36dd | 776 | SCM_DEFINE (scm_port_revealed, "port-revealed", 1, 0, 0, |
1bbd0b84 | 777 | (SCM port), |
1e6808ea | 778 | "Return the revealed count for @var{port}.") |
1bbd0b84 | 779 | #define FUNC_NAME s_scm_port_revealed |
0f2d19dd | 780 | { |
78446828 | 781 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 782 | SCM_VALIDATE_OPENPORT (1, port); |
e11e83f3 | 783 | return scm_from_int (scm_revealed_count (port)); |
0f2d19dd | 784 | } |
1bbd0b84 | 785 | #undef FUNC_NAME |
0f2d19dd JB |
786 | |
787 | /* Set the revealed count for a port. */ | |
3b3b36dd | 788 | SCM_DEFINE (scm_set_port_revealed_x, "set-port-revealed!", 2, 0, 0, |
1bbd0b84 | 789 | (SCM port, SCM rcount), |
b450f070 | 790 | "Sets the revealed count for a port to a given value.\n" |
b380b885 | 791 | "The return value is unspecified.") |
1bbd0b84 | 792 | #define FUNC_NAME s_scm_set_port_revealed_x |
0f2d19dd | 793 | { |
78446828 | 794 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 795 | SCM_VALIDATE_OPENPORT (1, port); |
a55c2b68 | 796 | SCM_REVEALED (port) = scm_to_int (rcount); |
8b13c6b3 | 797 | return SCM_UNSPECIFIED; |
0f2d19dd | 798 | } |
1bbd0b84 | 799 | #undef FUNC_NAME |
0f2d19dd | 800 | |
d68fee48 JB |
801 | |
802 | \f | |
803 | /* Retrieving a port's mode. */ | |
804 | ||
eadd48de GH |
805 | /* Return the flags that characterize a port based on the mode |
806 | * string used to open a file for that port. | |
807 | * | |
808 | * See PORT FLAGS in scm.h | |
809 | */ | |
810 | ||
3a5fb14d | 811 | static long |
889975e5 | 812 | scm_i_mode_bits_n (SCM modes) |
3a5fb14d MV |
813 | { |
814 | return (SCM_OPN | |
889975e5 MG |
815 | | (scm_i_string_contains_char (modes, 'r') |
816 | || scm_i_string_contains_char (modes, '+') ? SCM_RDNG : 0) | |
817 | | (scm_i_string_contains_char (modes, 'w') | |
818 | || scm_i_string_contains_char (modes, 'a') | |
819 | || scm_i_string_contains_char (modes, '+') ? SCM_WRTNG : 0) | |
820 | | (scm_i_string_contains_char (modes, '0') ? SCM_BUF0 : 0) | |
821 | | (scm_i_string_contains_char (modes, 'l') ? SCM_BUFLINE : 0)); | |
3a5fb14d MV |
822 | } |
823 | ||
eadd48de | 824 | long |
a284e297 | 825 | scm_mode_bits (char *modes) |
eadd48de | 826 | { |
889975e5 | 827 | return scm_i_mode_bits (scm_from_locale_string (modes)); |
eadd48de GH |
828 | } |
829 | ||
d617ee18 MV |
830 | long |
831 | scm_i_mode_bits (SCM modes) | |
832 | { | |
833 | long bits; | |
834 | ||
835 | if (!scm_is_string (modes)) | |
836 | scm_wrong_type_arg_msg (NULL, 0, modes, "string"); | |
837 | ||
889975e5 | 838 | bits = scm_i_mode_bits_n (modes); |
d617ee18 MV |
839 | scm_remember_upto_here_1 (modes); |
840 | return bits; | |
841 | } | |
eadd48de GH |
842 | |
843 | /* Return the mode flags from an open port. | |
844 | * Some modes such as "append" are only used when opening | |
845 | * a file and are not returned here. */ | |
846 | ||
3b3b36dd | 847 | SCM_DEFINE (scm_port_mode, "port-mode", 1, 0, 0, |
1bbd0b84 | 848 | (SCM port), |
1e6808ea MG |
849 | "Return the port modes associated with the open port @var{port}.\n" |
850 | "These will not necessarily be identical to the modes used when\n" | |
851 | "the port was opened, since modes such as \"append\" which are\n" | |
852 | "used only during port creation are not retained.") | |
1bbd0b84 | 853 | #define FUNC_NAME s_scm_port_mode |
eadd48de | 854 | { |
26a3038d | 855 | char modes[4]; |
eadd48de | 856 | modes[0] = '\0'; |
78446828 MV |
857 | |
858 | port = SCM_COERCE_OUTPORT (port); | |
34d19ef6 | 859 | SCM_VALIDATE_OPPORT (1, port); |
f9a64404 DH |
860 | if (SCM_CELL_WORD_0 (port) & SCM_RDNG) { |
861 | if (SCM_CELL_WORD_0 (port) & SCM_WRTNG) | |
eadd48de GH |
862 | strcpy (modes, "r+"); |
863 | else | |
864 | strcpy (modes, "r"); | |
865 | } | |
f9a64404 | 866 | else if (SCM_CELL_WORD_0 (port) & SCM_WRTNG) |
eadd48de | 867 | strcpy (modes, "w"); |
f9a64404 | 868 | if (SCM_CELL_WORD_0 (port) & SCM_BUF0) |
eadd48de | 869 | strcat (modes, "0"); |
3a5fb14d | 870 | return scm_from_locale_string (modes); |
eadd48de | 871 | } |
1bbd0b84 | 872 | #undef FUNC_NAME |
eadd48de GH |
873 | |
874 | ||
d68fee48 JB |
875 | \f |
876 | /* Closing ports. */ | |
877 | ||
0f2d19dd JB |
878 | /* scm_close_port |
879 | * Call the close operation on a port object. | |
eadd48de | 880 | * see also scm_close. |
0f2d19dd | 881 | */ |
3b3b36dd | 882 | SCM_DEFINE (scm_close_port, "close-port", 1, 0, 0, |
1bbd0b84 | 883 | (SCM port), |
1e6808ea MG |
884 | "Close the specified port object. Return @code{#t} if it\n" |
885 | "successfully closes a port or @code{#f} if it was already\n" | |
886 | "closed. An exception may be raised if an error occurs, for\n" | |
887 | "example when flushing buffered output. See also @ref{Ports and\n" | |
888 | "File Descriptors, close}, for a procedure which can close file\n" | |
889 | "descriptors.") | |
1bbd0b84 | 890 | #define FUNC_NAME s_scm_close_port |
0f2d19dd | 891 | { |
1be6b49c | 892 | size_t i; |
eadd48de GH |
893 | int rv; |
894 | ||
78446828 MV |
895 | port = SCM_COERCE_OUTPORT (port); |
896 | ||
7a754ca6 | 897 | SCM_VALIDATE_PORT (1, port); |
0f2d19dd | 898 | if (SCM_CLOSEDP (port)) |
eadd48de | 899 | return SCM_BOOL_F; |
0f2d19dd | 900 | i = SCM_PTOBNUM (port); |
affc96b5 GH |
901 | if (scm_ptobs[i].close) |
902 | rv = (scm_ptobs[i].close) (port); | |
eadd48de GH |
903 | else |
904 | rv = 0; | |
5dbc6c06 | 905 | scm_i_remove_port (port); |
22a52da1 | 906 | SCM_CLR_PORT_OPEN_FLAG (port); |
7888309b | 907 | return scm_from_bool (rv >= 0); |
7a754ca6 MD |
908 | } |
909 | #undef FUNC_NAME | |
910 | ||
911 | SCM_DEFINE (scm_close_input_port, "close-input-port", 1, 0, 0, | |
912 | (SCM port), | |
913 | "Close the specified input port object. The routine has no effect if\n" | |
914 | "the file has already been closed. An exception may be raised if an\n" | |
915 | "error occurs. The value returned is unspecified.\n\n" | |
916 | "See also @ref{Ports and File Descriptors, close}, for a procedure\n" | |
917 | "which can close file descriptors.") | |
918 | #define FUNC_NAME s_scm_close_input_port | |
919 | { | |
920 | SCM_VALIDATE_INPUT_PORT (1, port); | |
921 | scm_close_port (port); | |
922 | return SCM_UNSPECIFIED; | |
923 | } | |
924 | #undef FUNC_NAME | |
925 | ||
926 | SCM_DEFINE (scm_close_output_port, "close-output-port", 1, 0, 0, | |
927 | (SCM port), | |
928 | "Close the specified output port object. The routine has no effect if\n" | |
929 | "the file has already been closed. An exception may be raised if an\n" | |
930 | "error occurs. The value returned is unspecified.\n\n" | |
931 | "See also @ref{Ports and File Descriptors, close}, for a procedure\n" | |
932 | "which can close file descriptors.") | |
933 | #define FUNC_NAME s_scm_close_output_port | |
934 | { | |
935 | port = SCM_COERCE_OUTPORT (port); | |
936 | SCM_VALIDATE_OUTPUT_PORT (1, port); | |
937 | scm_close_port (port); | |
938 | return SCM_UNSPECIFIED; | |
0f2d19dd | 939 | } |
1bbd0b84 | 940 | #undef FUNC_NAME |
0f2d19dd | 941 | |
5dbc6c06 | 942 | static SCM |
b7b4aef9 | 943 | collect_keys (void *unused, SCM key, SCM value, SCM result) |
5dbc6c06 | 944 | { |
b7b4aef9 | 945 | return scm_cons (key, result); |
5dbc6c06 HWN |
946 | } |
947 | ||
c536b4b3 MV |
948 | void |
949 | scm_c_port_for_each (void (*proc)(void *data, SCM p), void *data) | |
c2ca4493 | 950 | { |
fdfe6305 MV |
951 | SCM ports; |
952 | ||
b7b4aef9 AW |
953 | /* Copy out the port table as a list so that we get strong references |
954 | to all the values. */ | |
5dbc6c06 | 955 | scm_i_pthread_mutex_lock (&scm_i_port_table_mutex); |
b7b4aef9 AW |
956 | ports = scm_internal_hash_fold (collect_keys, NULL, |
957 | SCM_EOL, scm_i_port_weak_hash); | |
9de87eea | 958 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); |
3a5fb14d | 959 | |
b7b4aef9 AW |
960 | for (; scm_is_pair (ports); ports = scm_cdr (ports)) |
961 | { | |
962 | SCM p = scm_car (ports); | |
963 | if (SCM_PORTP (p)) | |
964 | proc (data, p); | |
965 | } | |
c536b4b3 | 966 | } |
fdfe6305 | 967 | |
c536b4b3 MV |
968 | SCM_DEFINE (scm_port_for_each, "port-for-each", 1, 0, 0, |
969 | (SCM proc), | |
970 | "Apply @var{proc} to each port in the Guile port table\n" | |
971 | "in turn. The return value is unspecified. More specifically,\n" | |
972 | "@var{proc} is applied exactly once to every port that exists\n" | |
b7e64f8b BT |
973 | "in the system at the time @code{port-for-each} is invoked.\n" |
974 | "Changes to the port table while @code{port-for-each} is running\n" | |
975 | "have no effect as far as @code{port-for-each} is concerned.") | |
c536b4b3 MV |
976 | #define FUNC_NAME s_scm_port_for_each |
977 | { | |
1b3daef0 AW |
978 | SCM ports; |
979 | ||
c536b4b3 MV |
980 | SCM_VALIDATE_PROC (1, proc); |
981 | ||
1b3daef0 AW |
982 | /* Copy out the port table as a list so that we get strong references |
983 | to all the values. */ | |
984 | scm_i_pthread_mutex_lock (&scm_i_port_table_mutex); | |
985 | ports = scm_internal_hash_fold (collect_keys, NULL, | |
986 | SCM_EOL, scm_i_port_weak_hash); | |
987 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); | |
988 | ||
989 | for (; scm_is_pair (ports); ports = scm_cdr (ports)) | |
990 | if (SCM_PORTP (SCM_CAR (ports))) | |
991 | scm_call_1 (proc, SCM_CAR (ports)); | |
992 | ||
c2ca4493 GH |
993 | return SCM_UNSPECIFIED; |
994 | } | |
995 | #undef FUNC_NAME | |
996 | ||
c536b4b3 | 997 | |
d68fee48 JB |
998 | \f |
999 | /* Utter miscellany. Gosh, we should clean this up some time. */ | |
1000 | ||
3b3b36dd | 1001 | SCM_DEFINE (scm_input_port_p, "input-port?", 1, 0, 0, |
1bbd0b84 | 1002 | (SCM x), |
1e6808ea MG |
1003 | "Return @code{#t} if @var{x} is an input port, otherwise return\n" |
1004 | "@code{#f}. Any object satisfying this predicate also satisfies\n" | |
1005 | "@code{port?}.") | |
1bbd0b84 | 1006 | #define FUNC_NAME s_scm_input_port_p |
0f2d19dd | 1007 | { |
7888309b | 1008 | return scm_from_bool (SCM_INPUT_PORT_P (x)); |
0f2d19dd | 1009 | } |
1bbd0b84 | 1010 | #undef FUNC_NAME |
0f2d19dd | 1011 | |
3b3b36dd | 1012 | SCM_DEFINE (scm_output_port_p, "output-port?", 1, 0, 0, |
1bbd0b84 | 1013 | (SCM x), |
1e6808ea MG |
1014 | "Return @code{#t} if @var{x} is an output port, otherwise return\n" |
1015 | "@code{#f}. Any object satisfying this predicate also satisfies\n" | |
1016 | "@code{port?}.") | |
1bbd0b84 | 1017 | #define FUNC_NAME s_scm_output_port_p |
0f2d19dd | 1018 | { |
82893676 | 1019 | x = SCM_COERCE_OUTPORT (x); |
7888309b | 1020 | return scm_from_bool (SCM_OUTPUT_PORT_P (x)); |
0f2d19dd | 1021 | } |
1bbd0b84 | 1022 | #undef FUNC_NAME |
0f2d19dd | 1023 | |
eb5c0a2a GH |
1024 | SCM_DEFINE (scm_port_p, "port?", 1, 0, 0, |
1025 | (SCM x), | |
1e6808ea | 1026 | "Return a boolean indicating whether @var{x} is a port.\n" |
5352393c MG |
1027 | "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n" |
1028 | "@var{x}))}.") | |
eb5c0a2a GH |
1029 | #define FUNC_NAME s_scm_port_p |
1030 | { | |
7888309b | 1031 | return scm_from_bool (SCM_PORTP (x)); |
eb5c0a2a GH |
1032 | } |
1033 | #undef FUNC_NAME | |
1034 | ||
3b3b36dd | 1035 | SCM_DEFINE (scm_port_closed_p, "port-closed?", 1, 0, 0, |
1bbd0b84 | 1036 | (SCM port), |
1e6808ea MG |
1037 | "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n" |
1038 | "open.") | |
1bbd0b84 | 1039 | #define FUNC_NAME s_scm_port_closed_p |
60d0643d | 1040 | { |
34d19ef6 | 1041 | SCM_VALIDATE_PORT (1, port); |
7888309b | 1042 | return scm_from_bool (!SCM_OPPORTP (port)); |
60d0643d | 1043 | } |
1bbd0b84 | 1044 | #undef FUNC_NAME |
0f2d19dd | 1045 | |
3b3b36dd | 1046 | SCM_DEFINE (scm_eof_object_p, "eof-object?", 1, 0, 0, |
1bbd0b84 | 1047 | (SCM x), |
1e6808ea MG |
1048 | "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n" |
1049 | "return @code{#f}.") | |
1bbd0b84 | 1050 | #define FUNC_NAME s_scm_eof_object_p |
0f2d19dd | 1051 | { |
7888309b | 1052 | return scm_from_bool(SCM_EOF_OBJECT_P (x)); |
0f2d19dd | 1053 | } |
1bbd0b84 | 1054 | #undef FUNC_NAME |
0f2d19dd | 1055 | |
3b3b36dd | 1056 | SCM_DEFINE (scm_force_output, "force-output", 0, 1, 0, |
1bbd0b84 | 1057 | (SCM port), |
b380b885 | 1058 | "Flush the specified output port, or the current output port if @var{port}\n" |
9401323e | 1059 | "is omitted. The current output buffer contents are passed to the\n" |
b380b885 MD |
1060 | "underlying port implementation (e.g., in the case of fports, the\n" |
1061 | "data will be written to the file and the output buffer will be cleared.)\n" | |
1062 | "It has no effect on an unbuffered port.\n\n" | |
1063 | "The return value is unspecified.") | |
1bbd0b84 | 1064 | #define FUNC_NAME s_scm_force_output |
0f2d19dd JB |
1065 | { |
1066 | if (SCM_UNBNDP (port)) | |
9de87eea | 1067 | port = scm_current_output_port (); |
0f2d19dd | 1068 | else |
78446828 MV |
1069 | { |
1070 | port = SCM_COERCE_OUTPORT (port); | |
34d19ef6 | 1071 | SCM_VALIDATE_OPOUTPORT (1, port); |
78446828 | 1072 | } |
affc96b5 | 1073 | scm_flush (port); |
ee149d03 | 1074 | return SCM_UNSPECIFIED; |
0f2d19dd | 1075 | } |
1bbd0b84 | 1076 | #undef FUNC_NAME |
0f2d19dd | 1077 | |
5dbc6c06 HWN |
1078 | |
1079 | static void | |
61d3568b | 1080 | flush_output_port (void *closure, SCM port) |
5dbc6c06 | 1081 | { |
5dbc6c06 HWN |
1082 | if (SCM_OPOUTPORTP (port)) |
1083 | scm_flush (port); | |
1084 | } | |
1085 | ||
a1ec6916 | 1086 | SCM_DEFINE (scm_flush_all_ports, "flush-all-ports", 0, 0, 0, |
1bbd0b84 | 1087 | (), |
b380b885 MD |
1088 | "Equivalent to calling @code{force-output} on\n" |
1089 | "all open output ports. The return value is unspecified.") | |
1bbd0b84 | 1090 | #define FUNC_NAME s_scm_flush_all_ports |
89ea5b7c | 1091 | { |
5dbc6c06 | 1092 | scm_c_port_for_each (&flush_output_port, NULL); |
89ea5b7c GH |
1093 | return SCM_UNSPECIFIED; |
1094 | } | |
1bbd0b84 | 1095 | #undef FUNC_NAME |
0f2d19dd | 1096 | |
3b3b36dd | 1097 | SCM_DEFINE (scm_read_char, "read-char", 0, 1, 0, |
1bbd0b84 | 1098 | (SCM port), |
1e6808ea MG |
1099 | "Return the next character available from @var{port}, updating\n" |
1100 | "@var{port} to point to the following character. If no more\n" | |
c62da8f8 LC |
1101 | "characters are available, the end-of-file object is returned.\n" |
1102 | "\n" | |
1103 | "When @var{port}'s data cannot be decoded according to its\n" | |
1104 | "character encoding, a @code{decoding-error} is raised and\n" | |
1105 | "@var{port} points past the erroneous byte sequence.\n") | |
1bbd0b84 | 1106 | #define FUNC_NAME s_scm_read_char |
0f2d19dd | 1107 | { |
889975e5 | 1108 | scm_t_wchar c; |
0f2d19dd | 1109 | if (SCM_UNBNDP (port)) |
9de87eea | 1110 | port = scm_current_input_port (); |
34d19ef6 | 1111 | SCM_VALIDATE_OPINPORT (1, port); |
b7f3516f | 1112 | c = scm_getc (port); |
0f2d19dd JB |
1113 | if (EOF == c) |
1114 | return SCM_EOF_VAL; | |
7866a09b | 1115 | return SCM_MAKE_CHAR (c); |
0f2d19dd | 1116 | } |
1bbd0b84 | 1117 | #undef FUNC_NAME |
0f2d19dd | 1118 | |
f4bc4e59 LC |
1119 | /* Update the line and column number of PORT after consumption of C. */ |
1120 | static inline void | |
1121 | update_port_lf (scm_t_wchar c, SCM port) | |
1122 | { | |
1123 | switch (c) | |
1124 | { | |
1125 | case '\a': | |
7b292a9d | 1126 | case EOF: |
f4bc4e59 LC |
1127 | break; |
1128 | case '\b': | |
1129 | SCM_DECCOL (port); | |
1130 | break; | |
1131 | case '\n': | |
1132 | SCM_INCLINE (port); | |
1133 | break; | |
1134 | case '\r': | |
1135 | SCM_ZEROCOL (port); | |
1136 | break; | |
1137 | case '\t': | |
1138 | SCM_TABCOL (port); | |
1139 | break; | |
1140 | default: | |
1141 | SCM_INCCOL (port); | |
1142 | break; | |
1143 | } | |
1144 | } | |
1145 | ||
889975e5 MG |
1146 | #define SCM_MBCHAR_BUF_SIZE (4) |
1147 | ||
f4bc4e59 LC |
1148 | /* Convert the SIZE-byte UTF-8 sequence in UTF8_BUF to a codepoint. |
1149 | UTF8_BUF is assumed to contain a valid UTF-8 sequence. */ | |
1150 | static scm_t_wchar | |
1151 | utf8_to_codepoint (const scm_t_uint8 *utf8_buf, size_t size) | |
1152 | { | |
1153 | scm_t_wchar codepoint; | |
1154 | ||
1155 | if (utf8_buf[0] <= 0x7f) | |
1156 | { | |
1157 | assert (size == 1); | |
1158 | codepoint = utf8_buf[0]; | |
1159 | } | |
1160 | else if ((utf8_buf[0] & 0xe0) == 0xc0) | |
1161 | { | |
1162 | assert (size == 2); | |
1163 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x1f) << 6UL | |
1164 | | (utf8_buf[1] & 0x3f); | |
1165 | } | |
1166 | else if ((utf8_buf[0] & 0xf0) == 0xe0) | |
1167 | { | |
1168 | assert (size == 3); | |
1169 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x0f) << 12UL | |
1170 | | ((scm_t_wchar) utf8_buf[1] & 0x3f) << 6UL | |
1171 | | (utf8_buf[2] & 0x3f); | |
1172 | } | |
1173 | else | |
1174 | { | |
1175 | assert (size == 4); | |
1176 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x07) << 18UL | |
1177 | | ((scm_t_wchar) utf8_buf[1] & 0x3f) << 12UL | |
1178 | | ((scm_t_wchar) utf8_buf[2] & 0x3f) << 6UL | |
1179 | | (utf8_buf[3] & 0x3f); | |
1180 | } | |
1181 | ||
1182 | return codepoint; | |
1183 | } | |
1184 | ||
7b292a9d LC |
1185 | /* Read a UTF-8 sequence from PORT. On success, return 0 and set |
1186 | *CODEPOINT to the codepoint that was read, fill BUF with its UTF-8 | |
1187 | representation, and set *LEN to the length in bytes. Return | |
1188 | `EILSEQ' on error. */ | |
c62da8f8 | 1189 | static int |
7b292a9d LC |
1190 | get_utf8_codepoint (SCM port, scm_t_wchar *codepoint, |
1191 | scm_t_uint8 buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
889975e5 | 1192 | { |
7b292a9d LC |
1193 | #define ASSERT_NOT_EOF(b) \ |
1194 | if (SCM_UNLIKELY ((b) == EOF)) \ | |
1195 | goto invalid_seq | |
7be1705d LC |
1196 | #define CONSUME_PEEKED_BYTE() \ |
1197 | pt->read_pos++ | |
7b292a9d LC |
1198 | |
1199 | int byte; | |
7be1705d | 1200 | scm_t_port *pt; |
7b292a9d LC |
1201 | |
1202 | *len = 0; | |
7be1705d | 1203 | pt = SCM_PTAB_ENTRY (port); |
7b292a9d LC |
1204 | |
1205 | byte = scm_get_byte_or_eof (port); | |
1206 | if (byte == EOF) | |
1207 | { | |
1208 | *codepoint = EOF; | |
1209 | return 0; | |
1210 | } | |
1211 | ||
1212 | buf[0] = (scm_t_uint8) byte; | |
1213 | *len = 1; | |
1214 | ||
1215 | if (buf[0] <= 0x7f) | |
1216 | /* 1-byte form. */ | |
1217 | *codepoint = buf[0]; | |
1218 | else if (buf[0] >= 0xc2 && buf[0] <= 0xdf) | |
1219 | { | |
1220 | /* 2-byte form. */ | |
7be1705d | 1221 | byte = scm_peek_byte_or_eof (port); |
7b292a9d LC |
1222 | ASSERT_NOT_EOF (byte); |
1223 | ||
7b292a9d LC |
1224 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1225 | goto invalid_seq; | |
1226 | ||
7be1705d LC |
1227 | CONSUME_PEEKED_BYTE (); |
1228 | buf[1] = (scm_t_uint8) byte; | |
1229 | *len = 2; | |
1230 | ||
7b292a9d LC |
1231 | *codepoint = ((scm_t_wchar) buf[0] & 0x1f) << 6UL |
1232 | | (buf[1] & 0x3f); | |
1233 | } | |
1234 | else if ((buf[0] & 0xf0) == 0xe0) | |
1235 | { | |
1236 | /* 3-byte form. */ | |
7be1705d LC |
1237 | byte = scm_peek_byte_or_eof (port); |
1238 | ASSERT_NOT_EOF (byte); | |
7b292a9d LC |
1239 | |
1240 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80 | |
1241 | || (buf[0] == 0xe0 && byte < 0xa0) | |
1242 | || (buf[0] == 0xed && byte > 0x9f))) | |
7be1705d | 1243 | goto invalid_seq; |
7b292a9d | 1244 | |
7be1705d LC |
1245 | CONSUME_PEEKED_BYTE (); |
1246 | buf[1] = (scm_t_uint8) byte; | |
1247 | *len = 2; | |
7b292a9d | 1248 | |
7be1705d | 1249 | byte = scm_peek_byte_or_eof (port); |
7b292a9d LC |
1250 | ASSERT_NOT_EOF (byte); |
1251 | ||
7b292a9d LC |
1252 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1253 | goto invalid_seq; | |
1254 | ||
7be1705d LC |
1255 | CONSUME_PEEKED_BYTE (); |
1256 | buf[2] = (scm_t_uint8) byte; | |
1257 | *len = 3; | |
1258 | ||
7b292a9d LC |
1259 | *codepoint = ((scm_t_wchar) buf[0] & 0x0f) << 12UL |
1260 | | ((scm_t_wchar) buf[1] & 0x3f) << 6UL | |
1261 | | (buf[2] & 0x3f); | |
1262 | } | |
1263 | else if (buf[0] >= 0xf0 && buf[0] <= 0xf4) | |
1264 | { | |
1265 | /* 4-byte form. */ | |
7be1705d | 1266 | byte = scm_peek_byte_or_eof (port); |
7b292a9d LC |
1267 | ASSERT_NOT_EOF (byte); |
1268 | ||
7b292a9d LC |
1269 | if (SCM_UNLIKELY (((byte & 0xc0) != 0x80) |
1270 | || (buf[0] == 0xf0 && byte < 0x90) | |
1271 | || (buf[0] == 0xf4 && byte > 0x8f))) | |
7be1705d | 1272 | goto invalid_seq; |
7b292a9d | 1273 | |
7be1705d LC |
1274 | CONSUME_PEEKED_BYTE (); |
1275 | buf[1] = (scm_t_uint8) byte; | |
1276 | *len = 2; | |
1277 | ||
1278 | byte = scm_peek_byte_or_eof (port); | |
7b292a9d LC |
1279 | ASSERT_NOT_EOF (byte); |
1280 | ||
7be1705d LC |
1281 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1282 | goto invalid_seq; | |
1283 | ||
1284 | CONSUME_PEEKED_BYTE (); | |
7b292a9d LC |
1285 | buf[2] = (scm_t_uint8) byte; |
1286 | *len = 3; | |
1287 | ||
7be1705d | 1288 | byte = scm_peek_byte_or_eof (port); |
7b292a9d LC |
1289 | ASSERT_NOT_EOF (byte); |
1290 | ||
7b292a9d LC |
1291 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1292 | goto invalid_seq; | |
1293 | ||
7be1705d LC |
1294 | CONSUME_PEEKED_BYTE (); |
1295 | buf[3] = (scm_t_uint8) byte; | |
1296 | *len = 4; | |
1297 | ||
7b292a9d LC |
1298 | *codepoint = ((scm_t_wchar) buf[0] & 0x07) << 18UL |
1299 | | ((scm_t_wchar) buf[1] & 0x3f) << 12UL | |
1300 | | ((scm_t_wchar) buf[2] & 0x3f) << 6UL | |
1301 | | (buf[3] & 0x3f); | |
1302 | } | |
1303 | else | |
1304 | goto invalid_seq; | |
1305 | ||
1306 | return 0; | |
1307 | ||
1308 | invalid_seq: | |
7be1705d LC |
1309 | /* Here we could choose the consume the faulty byte when it's not a |
1310 | valid starting byte, but it's not a requirement. What Section 3.9 | |
1311 | of Unicode 6.0.0 mandates, though, is to not consume a byte that | |
1312 | would otherwise be a valid starting byte. */ | |
1313 | ||
7b292a9d LC |
1314 | return EILSEQ; |
1315 | ||
7be1705d | 1316 | #undef CONSUME_PEEKED_BYTE |
7b292a9d LC |
1317 | #undef ASSERT_NOT_EOF |
1318 | } | |
1319 | ||
1320 | /* Likewise, read a byte sequence from PORT, passing it through its | |
1321 | input conversion descriptor. */ | |
1322 | static int | |
1323 | get_iconv_codepoint (SCM port, scm_t_wchar *codepoint, | |
1324 | char buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
1325 | { | |
cdd3d6c9 | 1326 | scm_t_iconv_descriptors *id; |
f4bc4e59 | 1327 | scm_t_uint8 utf8_buf[SCM_MBCHAR_BUF_SIZE]; |
1ee237d9 | 1328 | size_t input_size = 0; |
889975e5 | 1329 | |
cdd3d6c9 MW |
1330 | id = scm_i_port_iconv_descriptors (port, SCM_PORT_READ); |
1331 | ||
1ee237d9 | 1332 | for (;;) |
63479112 | 1333 | { |
1ee237d9 MW |
1334 | int byte_read; |
1335 | char *input, *output; | |
f4bc4e59 | 1336 | size_t input_left, output_left, done; |
889975e5 | 1337 | |
f4bc4e59 | 1338 | byte_read = scm_get_byte_or_eof (port); |
1ee237d9 | 1339 | if (SCM_UNLIKELY (byte_read == EOF)) |
889975e5 | 1340 | { |
1ee237d9 MW |
1341 | if (SCM_LIKELY (input_size == 0)) |
1342 | { | |
1343 | *codepoint = (scm_t_wchar) EOF; | |
1344 | *len = input_size; | |
1345 | return 0; | |
1346 | } | |
1347 | else | |
45c0878b MW |
1348 | { |
1349 | /* EOF found in the middle of a multibyte character. */ | |
1350 | scm_i_set_pending_eof (port); | |
1351 | return EILSEQ; | |
1352 | } | |
889975e5 MG |
1353 | } |
1354 | ||
1ee237d9 | 1355 | buf[input_size++] = byte_read; |
889975e5 | 1356 | |
f4bc4e59 | 1357 | input = buf; |
1ee237d9 MW |
1358 | input_left = input_size; |
1359 | output = (char *) utf8_buf; | |
f4bc4e59 | 1360 | output_left = sizeof (utf8_buf); |
889975e5 | 1361 | |
337edc59 | 1362 | done = iconv (id->input_cd, &input, &input_left, &output, &output_left); |
1ee237d9 | 1363 | |
f4bc4e59 | 1364 | if (done == (size_t) -1) |
889975e5 | 1365 | { |
1ee237d9 MW |
1366 | int err = errno; |
1367 | if (SCM_LIKELY (err == EINVAL)) | |
1368 | /* The input byte sequence did not form a complete | |
1369 | character. Read another byte and try again. */ | |
1370 | continue; | |
1371 | else | |
1372 | return err; | |
889975e5 | 1373 | } |
f4bc4e59 | 1374 | else |
1ee237d9 MW |
1375 | { |
1376 | size_t output_size = sizeof (utf8_buf) - output_left; | |
1377 | if (SCM_LIKELY (output_size > 0)) | |
1378 | { | |
1379 | /* iconv generated output. Convert the UTF8_BUF sequence | |
1380 | to a Unicode code point. */ | |
1381 | *codepoint = utf8_to_codepoint (utf8_buf, output_size); | |
1382 | *len = input_size; | |
1383 | return 0; | |
1384 | } | |
1385 | else | |
1386 | { | |
1387 | /* iconv consumed some bytes without producing any output. | |
1388 | Most likely this means that a Unicode byte-order mark | |
1389 | (BOM) was consumed, which should not be included in the | |
1390 | returned buf. Shift any remaining bytes to the beginning | |
1391 | of buf, and continue the loop. */ | |
1392 | memmove (buf, input, input_left); | |
1393 | input_size = input_left; | |
1394 | continue; | |
1395 | } | |
1396 | } | |
7b292a9d | 1397 | } |
7b292a9d | 1398 | } |
c62da8f8 | 1399 | |
7b292a9d LC |
1400 | /* Read a codepoint from PORT and return it in *CODEPOINT. Fill BUF |
1401 | with the byte representation of the codepoint in PORT's encoding, and | |
1402 | set *LEN to the length in bytes of that representation. Return 0 on | |
1403 | success and an errno value on error. */ | |
1404 | static int | |
1405 | get_codepoint (SCM port, scm_t_wchar *codepoint, | |
1406 | char buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
1407 | { | |
1408 | int err; | |
1409 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
e4598559 | 1410 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
7b292a9d | 1411 | |
337edc59 | 1412 | if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8) |
7b292a9d | 1413 | err = get_utf8_codepoint (port, codepoint, (scm_t_uint8 *) buf, len); |
cc540d0b | 1414 | else |
7b292a9d LC |
1415 | err = get_iconv_codepoint (port, codepoint, buf, len); |
1416 | ||
1417 | if (SCM_LIKELY (err == 0)) | |
cdd3d6c9 MW |
1418 | { |
1419 | if (SCM_UNLIKELY (pti->at_stream_start_for_bom_read)) | |
1420 | { | |
1421 | /* Record that we're no longer at stream start. */ | |
1422 | pti->at_stream_start_for_bom_read = 0; | |
1423 | if (pt->rw_random) | |
1424 | pti->at_stream_start_for_bom_write = 0; | |
1425 | ||
1426 | /* If we just read a BOM in an encoding that recognizes them, | |
1427 | then silently consume it and read another code point. */ | |
1428 | if (SCM_UNLIKELY | |
1429 | (*codepoint == SCM_UNICODE_BOM | |
1430 | && (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8 | |
96965a6e MW |
1431 | || c_strcasecmp (pt->encoding, "UTF-16") == 0 |
1432 | || c_strcasecmp (pt->encoding, "UTF-32") == 0))) | |
cdd3d6c9 MW |
1433 | return get_codepoint (port, codepoint, buf, len); |
1434 | } | |
1435 | update_port_lf (*codepoint, port); | |
1436 | } | |
7b292a9d | 1437 | else if (pt->ilseq_handler == SCM_ICONVEH_QUESTION_MARK) |
a42d7971 | 1438 | { |
7b292a9d LC |
1439 | *codepoint = '?'; |
1440 | err = 0; | |
a42d7971 LC |
1441 | update_port_lf (*codepoint, port); |
1442 | } | |
889975e5 | 1443 | |
c62da8f8 | 1444 | return err; |
889975e5 MG |
1445 | } |
1446 | ||
fd5eec2b LC |
1447 | /* Read a codepoint from PORT and return it. */ |
1448 | scm_t_wchar | |
1449 | scm_getc (SCM port) | |
c62da8f8 | 1450 | #define FUNC_NAME "scm_getc" |
fd5eec2b | 1451 | { |
c62da8f8 | 1452 | int err; |
fd5eec2b | 1453 | size_t len; |
c62da8f8 | 1454 | scm_t_wchar codepoint; |
fd5eec2b LC |
1455 | char buf[SCM_MBCHAR_BUF_SIZE]; |
1456 | ||
c62da8f8 LC |
1457 | err = get_codepoint (port, &codepoint, buf, &len); |
1458 | if (SCM_UNLIKELY (err != 0)) | |
1459 | /* At this point PORT should point past the invalid encoding, as per | |
1460 | R6RS-lib Section 8.2.4. */ | |
1461 | scm_decoding_error (FUNC_NAME, err, "input decoding error", port); | |
1462 | ||
1463 | return codepoint; | |
fd5eec2b | 1464 | } |
c62da8f8 | 1465 | #undef FUNC_NAME |
889975e5 | 1466 | |
5c070ca7 | 1467 | /* this should only be called when the read buffer is empty. it |
affc96b5 | 1468 | tries to refill the read buffer. it returns the first char from |
5c070ca7 | 1469 | the port, which is either EOF or *(pt->read_pos). */ |
187fa0b9 MW |
1470 | static int |
1471 | scm_i_fill_input (SCM port) | |
6c951427 | 1472 | { |
92c2555f | 1473 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
45c0878b | 1474 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
283a1a0e | 1475 | |
b5cb4464 NJ |
1476 | assert (pt->read_pos == pt->read_end); |
1477 | ||
45c0878b MW |
1478 | if (pti->pending_eof) |
1479 | { | |
1480 | pti->pending_eof = 0; | |
1481 | return EOF; | |
1482 | } | |
1483 | ||
6c951427 GH |
1484 | if (pt->read_buf == pt->putback_buf) |
1485 | { | |
1486 | /* finished reading put-back chars. */ | |
1487 | pt->read_buf = pt->saved_read_buf; | |
1488 | pt->read_pos = pt->saved_read_pos; | |
1489 | pt->read_end = pt->saved_read_end; | |
1490 | pt->read_buf_size = pt->saved_read_buf_size; | |
1491 | if (pt->read_pos < pt->read_end) | |
5c070ca7 | 1492 | return *(pt->read_pos); |
6c951427 | 1493 | } |
affc96b5 | 1494 | return scm_ptobs[SCM_PTOBNUM (port)].fill_input (port); |
6c951427 GH |
1495 | } |
1496 | ||
187fa0b9 MW |
1497 | int |
1498 | scm_fill_input (SCM port) | |
1499 | { | |
1500 | return scm_i_fill_input (port); | |
1501 | } | |
1502 | ||
8a2b5965 MW |
1503 | /* Slow-path fallback for 'scm_get_byte_or_eof' in inline.h */ |
1504 | int | |
e00793d7 | 1505 | scm_slow_get_byte_or_eof (SCM port) |
8a2b5965 MW |
1506 | { |
1507 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
1508 | ||
1509 | if (pt->rw_active == SCM_PORT_WRITE) | |
1510 | scm_flush (port); | |
1511 | ||
1512 | if (pt->rw_random) | |
1513 | pt->rw_active = SCM_PORT_READ; | |
1514 | ||
1515 | if (pt->read_pos >= pt->read_end) | |
1516 | { | |
1517 | if (SCM_UNLIKELY (scm_i_fill_input (port) == EOF)) | |
1518 | return EOF; | |
1519 | } | |
1520 | ||
1521 | return *pt->read_pos++; | |
1522 | } | |
1523 | ||
1524 | /* Slow-path fallback for 'scm_peek_byte_or_eof' in inline.h */ | |
1525 | int | |
e00793d7 | 1526 | scm_slow_peek_byte_or_eof (SCM port) |
8a2b5965 MW |
1527 | { |
1528 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
1529 | ||
1530 | if (pt->rw_active == SCM_PORT_WRITE) | |
1531 | scm_flush (port); | |
1532 | ||
1533 | if (pt->rw_random) | |
1534 | pt->rw_active = SCM_PORT_READ; | |
1535 | ||
1536 | if (pt->read_pos >= pt->read_end) | |
1537 | { | |
1538 | if (SCM_UNLIKELY (scm_i_fill_input (port) == EOF)) | |
45c0878b MW |
1539 | { |
1540 | scm_i_set_pending_eof (port); | |
1541 | return EOF; | |
1542 | } | |
8a2b5965 MW |
1543 | } |
1544 | ||
1545 | return *pt->read_pos; | |
1546 | } | |
1547 | ||
3cb988bd | 1548 | |
6fe692e9 MD |
1549 | /* scm_lfwrite |
1550 | * | |
53802ea8 MV |
1551 | * This function differs from scm_c_write; it updates port line and |
1552 | * column. */ | |
6fe692e9 | 1553 | |
9c44cd45 | 1554 | void |
1be6b49c | 1555 | scm_lfwrite (const char *ptr, size_t size, SCM port) |
ee149d03 | 1556 | { |
92c2555f MV |
1557 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
1558 | scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)]; | |
3e2043c4 | 1559 | |
840ae05d | 1560 | if (pt->rw_active == SCM_PORT_READ) |
affc96b5 | 1561 | scm_end_input (port); |
283a1a0e | 1562 | |
31703ab8 | 1563 | ptob->write (port, ptr, size); |
840ae05d | 1564 | |
9c44cd45 MG |
1565 | for (; size; ptr++, size--) |
1566 | update_port_lf ((scm_t_wchar) (unsigned char) *ptr, port); | |
1567 | ||
1568 | if (pt->rw_random) | |
1569 | pt->rw_active = SCM_PORT_WRITE; | |
1570 | } | |
1571 | ||
4325620f | 1572 | /* Write STR to PORT from START inclusive to END exclusive. */ |
9c44cd45 MG |
1573 | void |
1574 | scm_lfwrite_substr (SCM str, size_t start, size_t end, SCM port) | |
1575 | { | |
9c44cd45 | 1576 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
9c44cd45 MG |
1577 | |
1578 | if (pt->rw_active == SCM_PORT_READ) | |
1579 | scm_end_input (port); | |
1580 | ||
4325620f LC |
1581 | if (end == (size_t) -1) |
1582 | end = scm_i_string_length (str); | |
9c44cd45 | 1583 | |
b908768a | 1584 | scm_i_display_substring (str, start, end, port); |
53802ea8 | 1585 | |
840ae05d JB |
1586 | if (pt->rw_random) |
1587 | pt->rw_active = SCM_PORT_WRITE; | |
ee149d03 | 1588 | } |
3cb988bd | 1589 | |
6fe692e9 MD |
1590 | /* scm_c_read |
1591 | * | |
1592 | * Used by an application to read arbitrary number of bytes from an | |
1593 | * SCM port. Same semantics as libc read, except that scm_c_read only | |
1594 | * returns less than SIZE bytes if at end-of-file. | |
1595 | * | |
1596 | * Warning: Doesn't update port line and column counts! */ | |
1597 | ||
b5cb4464 NJ |
1598 | /* This structure, and the following swap_buffer function, are used |
1599 | for temporarily swapping a port's own read buffer, and the buffer | |
1600 | that the caller of scm_c_read provides. */ | |
1601 | struct port_and_swap_buffer | |
1602 | { | |
1603 | scm_t_port *pt; | |
1604 | unsigned char *buffer; | |
1605 | size_t size; | |
1606 | }; | |
1607 | ||
1608 | static void | |
1609 | swap_buffer (void *data) | |
1610 | { | |
1611 | struct port_and_swap_buffer *psb = (struct port_and_swap_buffer *) data; | |
1612 | unsigned char *old_buf = psb->pt->read_buf; | |
1613 | size_t old_size = psb->pt->read_buf_size; | |
1614 | ||
1615 | /* Make the port use (buffer, size) from the struct. */ | |
1616 | psb->pt->read_pos = psb->pt->read_buf = psb->pt->read_end = psb->buffer; | |
1617 | psb->pt->read_buf_size = psb->size; | |
1618 | ||
1619 | /* Save the port's old (buffer, size) in the struct. */ | |
1620 | psb->buffer = old_buf; | |
1621 | psb->size = old_size; | |
1622 | } | |
1623 | ||
1be6b49c ML |
1624 | size_t |
1625 | scm_c_read (SCM port, void *buffer, size_t size) | |
693758d5 | 1626 | #define FUNC_NAME "scm_c_read" |
6fe692e9 | 1627 | { |
693758d5 | 1628 | scm_t_port *pt; |
1be6b49c | 1629 | size_t n_read = 0, n_available; |
b5cb4464 | 1630 | struct port_and_swap_buffer psb; |
6fe692e9 | 1631 | |
693758d5 LC |
1632 | SCM_VALIDATE_OPINPORT (1, port); |
1633 | ||
1634 | pt = SCM_PTAB_ENTRY (port); | |
6fe692e9 MD |
1635 | if (pt->rw_active == SCM_PORT_WRITE) |
1636 | scm_ptobs[SCM_PTOBNUM (port)].flush (port); | |
1637 | ||
1638 | if (pt->rw_random) | |
1639 | pt->rw_active = SCM_PORT_READ; | |
1640 | ||
b5cb4464 NJ |
1641 | /* Take bytes first from the port's read buffer. */ |
1642 | if (pt->read_pos < pt->read_end) | |
6fe692e9 | 1643 | { |
b5cb4464 | 1644 | n_available = min (size, pt->read_end - pt->read_pos); |
6fe692e9 | 1645 | memcpy (buffer, pt->read_pos, n_available); |
910d1e40 | 1646 | buffer = (char *) buffer + n_available; |
6fe692e9 MD |
1647 | pt->read_pos += n_available; |
1648 | n_read += n_available; | |
6fe692e9 | 1649 | size -= n_available; |
6fe692e9 MD |
1650 | } |
1651 | ||
b5cb4464 NJ |
1652 | /* Avoid the scm_dynwind_* costs if we now have enough data. */ |
1653 | if (size == 0) | |
1654 | return n_read; | |
1655 | ||
187fa0b9 MW |
1656 | /* Now we will call scm_i_fill_input repeatedly until we have read the |
1657 | requested number of bytes. (Note that a single scm_i_fill_input | |
b5cb4464 | 1658 | call does not guarantee to fill the whole of the port's read |
6d227556 | 1659 | buffer.) */ |
75192345 | 1660 | if (pt->read_buf_size <= 1 && pt->encoding == NULL) |
b5cb4464 | 1661 | { |
6d227556 NJ |
1662 | /* The port that we are reading from is unbuffered - i.e. does |
1663 | not have its own persistent buffer - but we have a buffer, | |
1664 | provided by our caller, that is the right size for the data | |
187fa0b9 | 1665 | that is wanted. For the following scm_i_fill_input calls, |
6d227556 NJ |
1666 | therefore, we use the buffer in hand as the port's read |
1667 | buffer. | |
1668 | ||
1669 | We need to make sure that the port's normal (1 byte) buffer | |
187fa0b9 | 1670 | is reinstated in case one of the scm_i_fill_input () calls |
6d227556 | 1671 | throws an exception; we use the scm_dynwind_* API to achieve |
75192345 MG |
1672 | that. |
1673 | ||
1674 | A consequence of this optimization is that the fill_input | |
1675 | functions can't unget characters. That'll push data to the | |
1676 | pushback buffer instead of this psb buffer. */ | |
1677 | #if SCM_DEBUG == 1 | |
1678 | unsigned char *pback = pt->putback_buf; | |
1679 | #endif | |
6d227556 NJ |
1680 | psb.pt = pt; |
1681 | psb.buffer = buffer; | |
1682 | psb.size = size; | |
1683 | scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE); | |
1684 | scm_dynwind_rewind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY); | |
1685 | scm_dynwind_unwind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY); | |
1686 | ||
187fa0b9 | 1687 | /* Call scm_i_fill_input until we have all the bytes that we need, |
6d227556 | 1688 | or we hit EOF. */ |
187fa0b9 | 1689 | while (pt->read_buf_size && (scm_i_fill_input (port) != EOF)) |
6d227556 NJ |
1690 | { |
1691 | pt->read_buf_size -= (pt->read_end - pt->read_pos); | |
1692 | pt->read_pos = pt->read_buf = pt->read_end; | |
1693 | } | |
75192345 MG |
1694 | #if SCM_DEBUG == 1 |
1695 | if (pback != pt->putback_buf | |
1696 | || pt->read_buf - (unsigned char *) buffer < 0) | |
1697 | scm_misc_error (FUNC_NAME, | |
1698 | "scm_c_read must not call a fill function that pushes " | |
1699 | "back characters onto an unbuffered port", SCM_EOL); | |
1700 | #endif | |
6d227556 | 1701 | n_read += pt->read_buf - (unsigned char *) buffer; |
75192345 | 1702 | |
6d227556 NJ |
1703 | /* Reinstate the port's normal buffer. */ |
1704 | scm_dynwind_end (); | |
1705 | } | |
1706 | else | |
1707 | { | |
1708 | /* The port has its own buffer. It is important that we use it, | |
1709 | even if it happens to be smaller than our caller's buffer, so | |
1710 | that a custom port implementation's entry points (in | |
1711 | particular, fill_input) can rely on the buffer always being | |
1712 | the same as they first set up. */ | |
187fa0b9 | 1713 | while (size && (scm_i_fill_input (port) != EOF)) |
6d227556 NJ |
1714 | { |
1715 | n_available = min (size, pt->read_end - pt->read_pos); | |
1716 | memcpy (buffer, pt->read_pos, n_available); | |
1717 | buffer = (char *) buffer + n_available; | |
1718 | pt->read_pos += n_available; | |
1719 | n_read += n_available; | |
1720 | size -= n_available; | |
1721 | } | |
1722 | } | |
6fe692e9 | 1723 | |
b5cb4464 | 1724 | return n_read; |
6fe692e9 | 1725 | } |
693758d5 | 1726 | #undef FUNC_NAME |
6fe692e9 MD |
1727 | |
1728 | /* scm_c_write | |
1729 | * | |
1730 | * Used by an application to write arbitrary number of bytes to an SCM | |
1731 | * port. Similar semantics as libc write. However, unlike libc | |
1732 | * write, scm_c_write writes the requested number of bytes and has no | |
1733 | * return value. | |
1734 | * | |
1735 | * Warning: Doesn't update port line and column counts! | |
1736 | */ | |
1737 | ||
693758d5 | 1738 | void |
1be6b49c | 1739 | scm_c_write (SCM port, const void *ptr, size_t size) |
693758d5 | 1740 | #define FUNC_NAME "scm_c_write" |
6fe692e9 | 1741 | { |
693758d5 LC |
1742 | scm_t_port *pt; |
1743 | scm_t_ptob_descriptor *ptob; | |
1744 | ||
1745 | SCM_VALIDATE_OPOUTPORT (1, port); | |
1746 | ||
1747 | pt = SCM_PTAB_ENTRY (port); | |
1748 | ptob = &scm_ptobs[SCM_PTOBNUM (port)]; | |
6fe692e9 MD |
1749 | |
1750 | if (pt->rw_active == SCM_PORT_READ) | |
1751 | scm_end_input (port); | |
1752 | ||
1753 | ptob->write (port, ptr, size); | |
1754 | ||
1755 | if (pt->rw_random) | |
1756 | pt->rw_active = SCM_PORT_WRITE; | |
1757 | } | |
693758d5 | 1758 | #undef FUNC_NAME |
3cb988bd | 1759 | |
fca43887 | 1760 | void |
a284e297 | 1761 | scm_flush (SCM port) |
ee149d03 | 1762 | { |
c014a02e | 1763 | long i = SCM_PTOBNUM (port); |
7f2a6c38 | 1764 | assert (i >= 0); |
affc96b5 | 1765 | (scm_ptobs[i].flush) (port); |
ee149d03 JB |
1766 | } |
1767 | ||
283a1a0e | 1768 | void |
a284e297 | 1769 | scm_end_input (SCM port) |
283a1a0e | 1770 | { |
c014a02e | 1771 | long offset; |
92c2555f | 1772 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
283a1a0e | 1773 | |
45c0878b | 1774 | scm_i_clear_pending_eof (port); |
283a1a0e GH |
1775 | if (pt->read_buf == pt->putback_buf) |
1776 | { | |
1777 | offset = pt->read_end - pt->read_pos; | |
1778 | pt->read_buf = pt->saved_read_buf; | |
1779 | pt->read_pos = pt->saved_read_pos; | |
1780 | pt->read_end = pt->saved_read_end; | |
1781 | pt->read_buf_size = pt->saved_read_buf_size; | |
1782 | } | |
1783 | else | |
1784 | offset = 0; | |
1785 | ||
affc96b5 | 1786 | scm_ptobs[SCM_PTOBNUM (port)].end_input (port, offset); |
283a1a0e GH |
1787 | } |
1788 | ||
ee149d03 JB |
1789 | \f |
1790 | ||
1791 | ||
7f6c3f8f MW |
1792 | static void |
1793 | scm_i_unget_bytes (const unsigned char *buf, size_t len, SCM port) | |
1794 | #define FUNC_NAME "scm_unget_bytes" | |
ee149d03 | 1795 | { |
92c2555f | 1796 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
7f6c3f8f | 1797 | size_t old_len, new_len; |
840ae05d | 1798 | |
45c0878b | 1799 | scm_i_clear_pending_eof (port); |
6c951427 | 1800 | |
7f6c3f8f | 1801 | if (pt->read_buf != pt->putback_buf) |
6c951427 GH |
1802 | /* switch to the put-back buffer. */ |
1803 | { | |
1804 | if (pt->putback_buf == NULL) | |
1805 | { | |
7f6c3f8f MW |
1806 | pt->putback_buf_size = (len > SCM_INITIAL_PUTBACK_BUF_SIZE |
1807 | ? len : SCM_INITIAL_PUTBACK_BUF_SIZE); | |
c357d546 | 1808 | pt->putback_buf |
92d8fd32 | 1809 | = (unsigned char *) scm_gc_malloc_pointerless |
7f6c3f8f | 1810 | (pt->putback_buf_size, "putback buffer"); |
6c951427 GH |
1811 | } |
1812 | ||
1813 | pt->saved_read_buf = pt->read_buf; | |
1814 | pt->saved_read_pos = pt->read_pos; | |
1815 | pt->saved_read_end = pt->read_end; | |
1816 | pt->saved_read_buf_size = pt->read_buf_size; | |
1817 | ||
7f6c3f8f MW |
1818 | /* Put read_pos at the end of the buffer, so that ungets will not |
1819 | have to shift the buffer contents each time. */ | |
1820 | pt->read_buf = pt->putback_buf; | |
1821 | pt->read_pos = pt->read_end = pt->putback_buf + pt->putback_buf_size; | |
6c951427 GH |
1822 | pt->read_buf_size = pt->putback_buf_size; |
1823 | } | |
1824 | ||
7f6c3f8f MW |
1825 | old_len = pt->read_end - pt->read_pos; |
1826 | new_len = old_len + len; | |
1827 | ||
1828 | if (new_len > pt->read_buf_size) | |
1829 | /* The putback buffer needs to be enlarged. */ | |
1830 | { | |
1831 | size_t new_buf_size; | |
1832 | unsigned char *new_buf, *new_end, *new_pos; | |
1833 | ||
1834 | new_buf_size = pt->read_buf_size * 2; | |
1835 | if (new_buf_size < new_len) | |
1836 | new_buf_size = new_len; | |
1837 | ||
1838 | new_buf = (unsigned char *) | |
1839 | scm_gc_malloc_pointerless (new_buf_size, "putback buffer"); | |
1840 | ||
1841 | /* Put the bytes at the end of the buffer, so that future | |
1842 | ungets won't need to shift the buffer. */ | |
1843 | new_end = new_buf + new_buf_size; | |
1844 | new_pos = new_end - old_len; | |
1845 | memcpy (new_pos, pt->read_pos, old_len); | |
1846 | ||
1847 | pt->read_buf = pt->putback_buf = new_buf; | |
1848 | pt->read_pos = new_pos; | |
1849 | pt->read_end = new_end; | |
1850 | pt->read_buf_size = pt->putback_buf_size = new_buf_size; | |
1851 | } | |
1852 | else if (pt->read_buf + len < pt->read_pos) | |
1853 | /* If needed, shift the existing buffer contents up. | |
1854 | This should not happen unless some external code | |
1855 | manipulates the putback buffer pointers. */ | |
1856 | { | |
1857 | unsigned char *new_end = pt->read_buf + pt->read_buf_size; | |
1858 | unsigned char *new_pos = new_end - old_len; | |
1859 | ||
1860 | memmove (new_pos, pt->read_pos, old_len); | |
1861 | pt->read_pos = new_pos; | |
1862 | pt->read_end = new_end; | |
1863 | } | |
1864 | ||
1865 | /* Move read_pos back and copy the bytes there. */ | |
1866 | pt->read_pos -= len; | |
1867 | memcpy (pt->read_buf + (pt->read_pos - pt->read_buf), buf, len); | |
1868 | ||
1869 | if (pt->rw_active == SCM_PORT_WRITE) | |
1870 | scm_flush (port); | |
ee149d03 | 1871 | |
840ae05d JB |
1872 | if (pt->rw_random) |
1873 | pt->rw_active = SCM_PORT_READ; | |
889975e5 MG |
1874 | } |
1875 | #undef FUNC_NAME | |
1876 | ||
7f6c3f8f MW |
1877 | void |
1878 | scm_unget_bytes (const unsigned char *buf, size_t len, SCM port) | |
1879 | { | |
1880 | scm_i_unget_bytes (buf, len, port); | |
1881 | } | |
1882 | ||
1883 | void | |
1884 | scm_unget_byte (int c, SCM port) | |
1885 | { | |
1886 | unsigned char byte; | |
1887 | ||
1888 | byte = c; | |
1889 | scm_i_unget_bytes (&byte, 1, port); | |
1890 | } | |
1891 | ||
63479112 | 1892 | void |
889975e5 MG |
1893 | scm_ungetc (scm_t_wchar c, SCM port) |
1894 | #define FUNC_NAME "scm_ungetc" | |
1895 | { | |
1896 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
63479112 LC |
1897 | char *result; |
1898 | char result_buf[10]; | |
1899 | const char *encoding; | |
889975e5 | 1900 | size_t len; |
889975e5 | 1901 | |
63479112 LC |
1902 | if (pt->encoding != NULL) |
1903 | encoding = pt->encoding; | |
1904 | else | |
1905 | encoding = "ISO-8859-1"; | |
1906 | ||
1907 | len = sizeof (result_buf); | |
1908 | result = u32_conv_to_encoding (encoding, | |
1909 | (enum iconv_ilseq_handler) pt->ilseq_handler, | |
1910 | (uint32_t *) &c, 1, NULL, | |
1911 | result_buf, &len); | |
1912 | ||
1913 | if (SCM_UNLIKELY (result == NULL || len == 0)) | |
6851d3be LC |
1914 | scm_encoding_error (FUNC_NAME, errno, |
1915 | "conversion to port encoding failed", | |
1916 | SCM_BOOL_F, SCM_MAKE_CHAR (c)); | |
63479112 | 1917 | |
7f6c3f8f | 1918 | scm_i_unget_bytes ((unsigned char *) result, len, port); |
63479112 LC |
1919 | |
1920 | if (SCM_UNLIKELY (result != result_buf)) | |
1921 | free (result); | |
840ae05d | 1922 | |
ee149d03 JB |
1923 | if (c == '\n') |
1924 | { | |
1925 | /* What should col be in this case? | |
1926 | * We'll leave it at -1. | |
1927 | */ | |
1928 | SCM_LINUM (port) -= 1; | |
1929 | } | |
1930 | else | |
1931 | SCM_COL(port) -= 1; | |
1932 | } | |
c6c79933 | 1933 | #undef FUNC_NAME |
ee149d03 JB |
1934 | |
1935 | ||
1936 | void | |
70d63753 | 1937 | scm_ungets (const char *s, int n, SCM port) |
ee149d03 JB |
1938 | { |
1939 | /* This is simple minded and inefficient, but unreading strings is | |
1940 | * probably not a common operation, and remember that line and | |
1941 | * column numbers have to be handled... | |
1942 | * | |
1943 | * Please feel free to write an optimized version! | |
1944 | */ | |
1945 | while (n--) | |
1946 | scm_ungetc (s[n], port); | |
1947 | } | |
1948 | ||
1949 | ||
3b3b36dd | 1950 | SCM_DEFINE (scm_peek_char, "peek-char", 0, 1, 0, |
1bbd0b84 | 1951 | (SCM port), |
1e6808ea MG |
1952 | "Return the next character available from @var{port},\n" |
1953 | "@emph{without} updating @var{port} to point to the following\n" | |
1954 | "character. If no more characters are available, the\n" | |
c2dfff19 KR |
1955 | "end-of-file object is returned.\n" |
1956 | "\n" | |
1957 | "The value returned by\n" | |
1e6808ea MG |
1958 | "a call to @code{peek-char} is the same as the value that would\n" |
1959 | "have been returned by a call to @code{read-char} on the same\n" | |
1960 | "port. The only difference is that the very next call to\n" | |
1961 | "@code{read-char} or @code{peek-char} on that @var{port} will\n" | |
1962 | "return the value returned by the preceding call to\n" | |
1963 | "@code{peek-char}. In particular, a call to @code{peek-char} on\n" | |
1964 | "an interactive port will hang waiting for input whenever a call\n" | |
c62da8f8 LC |
1965 | "to @code{read-char} would have hung.\n" |
1966 | "\n" | |
1967 | "As for @code{read-char}, a @code{decoding-error} may be raised\n" | |
1968 | "if such a situation occurs. However, unlike with @code{read-char},\n" | |
1969 | "@var{port} still points at the beginning of the erroneous byte\n" | |
1970 | "sequence when the error is raised.\n") | |
1bbd0b84 | 1971 | #define FUNC_NAME s_scm_peek_char |
ee149d03 | 1972 | { |
c62da8f8 | 1973 | int err; |
fd5eec2b LC |
1974 | SCM result; |
1975 | scm_t_wchar c; | |
1976 | char bytes[SCM_MBCHAR_BUF_SIZE]; | |
7f6c3f8f | 1977 | long column, line; |
fd5eec2b LC |
1978 | size_t len; |
1979 | ||
ee149d03 | 1980 | if (SCM_UNBNDP (port)) |
9de87eea | 1981 | port = scm_current_input_port (); |
b2456dd4 | 1982 | SCM_VALIDATE_OPINPORT (1, port); |
fd5eec2b LC |
1983 | |
1984 | column = SCM_COL (port); | |
1985 | line = SCM_LINUM (port); | |
1986 | ||
c62da8f8 | 1987 | err = get_codepoint (port, &c, bytes, &len); |
fd5eec2b | 1988 | |
7f6c3f8f | 1989 | scm_i_unget_bytes ((unsigned char *) bytes, len, port); |
fd5eec2b | 1990 | |
c62da8f8 LC |
1991 | SCM_COL (port) = column; |
1992 | SCM_LINUM (port) = line; | |
fd5eec2b | 1993 | |
c62da8f8 LC |
1994 | if (SCM_UNLIKELY (err != 0)) |
1995 | { | |
1996 | scm_decoding_error (FUNC_NAME, err, "input decoding error", port); | |
1997 | ||
1998 | /* Shouldn't happen since `catch' always aborts to prompt. */ | |
1999 | result = SCM_BOOL_F; | |
fd5eec2b | 2000 | } |
c62da8f8 | 2001 | else if (c == EOF) |
45c0878b MW |
2002 | { |
2003 | scm_i_set_pending_eof (port); | |
2004 | result = SCM_EOF_VAL; | |
2005 | } | |
c62da8f8 LC |
2006 | else |
2007 | result = SCM_MAKE_CHAR (c); | |
fd5eec2b LC |
2008 | |
2009 | return result; | |
3cb988bd | 2010 | } |
1bbd0b84 | 2011 | #undef FUNC_NAME |
3cb988bd | 2012 | |
1be4270a | 2013 | SCM_DEFINE (scm_unread_char, "unread-char", 1, 1, 0, |
1bbd0b84 | 2014 | (SCM cobj, SCM port), |
b7e64f8b BT |
2015 | "Place character @var{cobj} in @var{port} so that it will be\n" |
2016 | "read by the next read operation. If called multiple times, the\n" | |
2017 | "unread characters will be read again in last-in first-out\n" | |
2018 | "order. If @var{port} is not supplied, the current input port\n" | |
2019 | "is used.") | |
1bbd0b84 | 2020 | #define FUNC_NAME s_scm_unread_char |
0f2d19dd JB |
2021 | { |
2022 | int c; | |
2023 | ||
34d19ef6 | 2024 | SCM_VALIDATE_CHAR (1, cobj); |
0f2d19dd | 2025 | if (SCM_UNBNDP (port)) |
9de87eea | 2026 | port = scm_current_input_port (); |
b2456dd4 | 2027 | SCM_VALIDATE_OPINPORT (2, port); |
0f2d19dd | 2028 | |
7866a09b | 2029 | c = SCM_CHAR (cobj); |
0f2d19dd | 2030 | |
b7f3516f | 2031 | scm_ungetc (c, port); |
0f2d19dd JB |
2032 | return cobj; |
2033 | } | |
1bbd0b84 | 2034 | #undef FUNC_NAME |
0f2d19dd | 2035 | |
a1ec6916 | 2036 | SCM_DEFINE (scm_unread_string, "unread-string", 2, 0, 0, |
1bbd0b84 | 2037 | (SCM str, SCM port), |
b380b885 MD |
2038 | "Place the string @var{str} in @var{port} so that its characters will be\n" |
2039 | "read in subsequent read operations. If called multiple times, the\n" | |
2040 | "unread characters will be read again in last-in first-out order. If\n" | |
2041 | "@var{port} is not supplied, the current-input-port is used.") | |
1bbd0b84 | 2042 | #define FUNC_NAME s_scm_unread_string |
ee1e7e13 | 2043 | { |
889975e5 | 2044 | int n; |
34d19ef6 | 2045 | SCM_VALIDATE_STRING (1, str); |
ee1e7e13 | 2046 | if (SCM_UNBNDP (port)) |
9de87eea | 2047 | port = scm_current_input_port (); |
b2456dd4 | 2048 | SCM_VALIDATE_OPINPORT (2, port); |
ee1e7e13 | 2049 | |
889975e5 MG |
2050 | n = scm_i_string_length (str); |
2051 | ||
2052 | while (n--) | |
2053 | scm_ungetc (scm_i_string_ref (str, n), port); | |
ee1e7e13 MD |
2054 | |
2055 | return str; | |
2056 | } | |
1bbd0b84 | 2057 | #undef FUNC_NAME |
ee1e7e13 | 2058 | |
a1ec6916 | 2059 | SCM_DEFINE (scm_seek, "seek", 3, 0, 0, |
1e6808ea | 2060 | (SCM fd_port, SCM offset, SCM whence), |
b7e64f8b | 2061 | "Sets the current position of @var{fd_port} to the integer\n" |
1e6808ea MG |
2062 | "@var{offset}, which is interpreted according to the value of\n" |
2063 | "@var{whence}.\n" | |
2064 | "\n" | |
2065 | "One of the following variables should be supplied for\n" | |
2066 | "@var{whence}:\n" | |
b380b885 MD |
2067 | "@defvar SEEK_SET\n" |
2068 | "Seek from the beginning of the file.\n" | |
2069 | "@end defvar\n" | |
2070 | "@defvar SEEK_CUR\n" | |
2071 | "Seek from the current position.\n" | |
2072 | "@end defvar\n" | |
2073 | "@defvar SEEK_END\n" | |
2074 | "Seek from the end of the file.\n" | |
1e6808ea | 2075 | "@end defvar\n" |
b7e64f8b | 2076 | "If @var{fd_port} is a file descriptor, the underlying system\n" |
1e6808ea MG |
2077 | "call is @code{lseek}. @var{port} may be a string port.\n" |
2078 | "\n" | |
2079 | "The value returned is the new position in the file. This means\n" | |
2080 | "that the current position of a port can be obtained using:\n" | |
2081 | "@lisp\n" | |
b380b885 | 2082 | "(seek port 0 SEEK_CUR)\n" |
1e6808ea | 2083 | "@end lisp") |
1bbd0b84 | 2084 | #define FUNC_NAME s_scm_seek |
840ae05d | 2085 | { |
840ae05d JB |
2086 | int how; |
2087 | ||
1e6808ea | 2088 | fd_port = SCM_COERCE_OUTPORT (fd_port); |
840ae05d | 2089 | |
a55c2b68 | 2090 | how = scm_to_int (whence); |
840ae05d | 2091 | if (how != SEEK_SET && how != SEEK_CUR && how != SEEK_END) |
1bbd0b84 | 2092 | SCM_OUT_OF_RANGE (3, whence); |
23f2b9a3 | 2093 | |
0a94eb00 | 2094 | if (SCM_OPPORTP (fd_port)) |
840ae05d | 2095 | { |
cdd3d6c9 | 2096 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (fd_port); |
92c2555f | 2097 | scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (fd_port); |
f1ce9199 LC |
2098 | off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset); |
2099 | off_t_or_off64_t rv; | |
840ae05d JB |
2100 | |
2101 | if (!ptob->seek) | |
1bbd0b84 | 2102 | SCM_MISC_ERROR ("port is not seekable", |
1e6808ea | 2103 | scm_cons (fd_port, SCM_EOL)); |
840ae05d | 2104 | else |
cdd3d6c9 MW |
2105 | rv = ptob->seek (fd_port, off, how); |
2106 | ||
2107 | /* Set stream-start flags according to new position. */ | |
2108 | pti->at_stream_start_for_bom_read = (rv == 0); | |
2109 | pti->at_stream_start_for_bom_write = (rv == 0); | |
2110 | ||
2111 | scm_i_clear_pending_eof (fd_port); | |
2112 | ||
f1ce9199 | 2113 | return scm_from_off_t_or_off64_t (rv); |
840ae05d JB |
2114 | } |
2115 | else /* file descriptor?. */ | |
2116 | { | |
23f2b9a3 KR |
2117 | off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset); |
2118 | off_t_or_off64_t rv; | |
2119 | rv = lseek_or_lseek64 (scm_to_int (fd_port), off, how); | |
840ae05d | 2120 | if (rv == -1) |
1bbd0b84 | 2121 | SCM_SYSERROR; |
23f2b9a3 | 2122 | return scm_from_off_t_or_off64_t (rv); |
840ae05d | 2123 | } |
840ae05d | 2124 | } |
1bbd0b84 | 2125 | #undef FUNC_NAME |
840ae05d | 2126 | |
8ab3d8a0 KR |
2127 | #ifndef O_BINARY |
2128 | #define O_BINARY 0 | |
2129 | #endif | |
2130 | ||
2131 | /* Mingw has ftruncate(), perhaps implemented above using chsize, but | |
2132 | doesn't have the filename version truncate(), hence this code. */ | |
2133 | #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE | |
2134 | static int | |
2135 | truncate (const char *file, off_t length) | |
82893676 | 2136 | { |
8ab3d8a0 KR |
2137 | int ret, fdes; |
2138 | ||
2139 | fdes = open (file, O_BINARY | O_WRONLY); | |
2140 | if (fdes == -1) | |
2141 | return -1; | |
2142 | ||
2143 | ret = ftruncate (fdes, length); | |
2144 | if (ret == -1) | |
82893676 | 2145 | { |
8ab3d8a0 | 2146 | int save_errno = errno; |
82893676 | 2147 | close (fdes); |
8ab3d8a0 KR |
2148 | errno = save_errno; |
2149 | return -1; | |
82893676 | 2150 | } |
8ab3d8a0 KR |
2151 | |
2152 | return close (fdes); | |
82893676 | 2153 | } |
8ab3d8a0 | 2154 | #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */ |
82893676 | 2155 | |
a1ec6916 | 2156 | SCM_DEFINE (scm_truncate_file, "truncate-file", 1, 1, 0, |
1bbd0b84 | 2157 | (SCM object, SCM length), |
b7e64f8b BT |
2158 | "Truncate file @var{object} to @var{length} bytes. @var{object}\n" |
2159 | "can be a filename string, a port object, or an integer file\n" | |
2160 | "descriptor.\n" | |
8ab3d8a0 KR |
2161 | "The return value is unspecified.\n" |
2162 | "\n" | |
2163 | "For a port or file descriptor @var{length} can be omitted, in\n" | |
2164 | "which case the file is truncated at the current position (per\n" | |
2165 | "@code{ftell} above).\n" | |
2166 | "\n" | |
2167 | "On most systems a file can be extended by giving a length\n" | |
2168 | "greater than the current size, but this is not mandatory in the\n" | |
2169 | "POSIX standard.") | |
1bbd0b84 | 2170 | #define FUNC_NAME s_scm_truncate_file |
840ae05d | 2171 | { |
69bc9ff3 | 2172 | int rv; |
69bc9ff3 | 2173 | |
2b829bbb KR |
2174 | /* "object" can be a port, fdes or filename. |
2175 | ||
2176 | Negative "length" makes no sense, but it's left to truncate() or | |
2177 | ftruncate() to give back an error for that (normally EINVAL). | |
2178 | */ | |
840ae05d | 2179 | |
840ae05d JB |
2180 | if (SCM_UNBNDP (length)) |
2181 | { | |
69bc9ff3 | 2182 | /* must supply length if object is a filename. */ |
7f9994d9 | 2183 | if (scm_is_string (object)) |
34d19ef6 | 2184 | SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL); |
69bc9ff3 | 2185 | |
e11e83f3 | 2186 | length = scm_seek (object, SCM_INUM0, scm_from_int (SEEK_CUR)); |
840ae05d | 2187 | } |
3fe6190f | 2188 | |
69bc9ff3 | 2189 | object = SCM_COERCE_OUTPORT (object); |
e11e83f3 | 2190 | if (scm_is_integer (object)) |
69bc9ff3 | 2191 | { |
23f2b9a3 KR |
2192 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); |
2193 | SCM_SYSCALL (rv = ftruncate_or_ftruncate64 (scm_to_int (object), | |
2194 | c_length)); | |
69bc9ff3 | 2195 | } |
0c95b57d | 2196 | else if (SCM_OPOUTPORTP (object)) |
69bc9ff3 | 2197 | { |
f1ce9199 | 2198 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); |
92c2555f MV |
2199 | scm_t_port *pt = SCM_PTAB_ENTRY (object); |
2200 | scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (object); | |
45c0878b | 2201 | |
affc96b5 | 2202 | if (!ptob->truncate) |
1bbd0b84 | 2203 | SCM_MISC_ERROR ("port is not truncatable", SCM_EOL); |
45c0878b MW |
2204 | |
2205 | scm_i_clear_pending_eof (object); | |
69bc9ff3 | 2206 | if (pt->rw_active == SCM_PORT_READ) |
affc96b5 | 2207 | scm_end_input (object); |
69bc9ff3 | 2208 | else if (pt->rw_active == SCM_PORT_WRITE) |
affc96b5 | 2209 | ptob->flush (object); |
45c0878b | 2210 | |
affc96b5 | 2211 | ptob->truncate (object, c_length); |
69bc9ff3 GH |
2212 | rv = 0; |
2213 | } | |
2214 | else | |
2215 | { | |
2b829bbb | 2216 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); |
7f9994d9 MV |
2217 | char *str = scm_to_locale_string (object); |
2218 | int eno; | |
2b829bbb | 2219 | SCM_SYSCALL (rv = truncate_or_truncate64 (str, c_length)); |
7f9994d9 MV |
2220 | eno = errno; |
2221 | free (str); | |
2222 | errno = eno; | |
69bc9ff3 GH |
2223 | } |
2224 | if (rv == -1) | |
1bbd0b84 | 2225 | SCM_SYSERROR; |
840ae05d JB |
2226 | return SCM_UNSPECIFIED; |
2227 | } | |
1bbd0b84 | 2228 | #undef FUNC_NAME |
840ae05d | 2229 | |
a1ec6916 | 2230 | SCM_DEFINE (scm_port_line, "port-line", 1, 0, 0, |
1bbd0b84 | 2231 | (SCM port), |
a150979d KR |
2232 | "Return the current line number for @var{port}.\n" |
2233 | "\n" | |
2234 | "The first line of a file is 0. But you might want to add 1\n" | |
2235 | "when printing line numbers, since starting from 1 is\n" | |
2236 | "traditional in error messages, and likely to be more natural to\n" | |
2237 | "non-programmers.") | |
1bbd0b84 | 2238 | #define FUNC_NAME s_scm_port_line |
0f2d19dd | 2239 | { |
78446828 | 2240 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2241 | SCM_VALIDATE_OPENPORT (1, port); |
651f2cd2 | 2242 | return scm_from_long (SCM_LINUM (port)); |
0f2d19dd | 2243 | } |
1bbd0b84 | 2244 | #undef FUNC_NAME |
0f2d19dd | 2245 | |
a1ec6916 | 2246 | SCM_DEFINE (scm_set_port_line_x, "set-port-line!", 2, 0, 0, |
1bbd0b84 | 2247 | (SCM port, SCM line), |
a150979d KR |
2248 | "Set the current line number for @var{port} to @var{line}. The\n" |
2249 | "first line of a file is 0.") | |
1bbd0b84 | 2250 | #define FUNC_NAME s_scm_set_port_line_x |
d043d8c2 | 2251 | { |
360fc44c | 2252 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2253 | SCM_VALIDATE_OPENPORT (1, port); |
651f2cd2 | 2254 | SCM_PTAB_ENTRY (port)->line_number = scm_to_long (line); |
564478fd | 2255 | return SCM_UNSPECIFIED; |
d043d8c2 | 2256 | } |
1bbd0b84 | 2257 | #undef FUNC_NAME |
d043d8c2 | 2258 | |
a1ec6916 | 2259 | SCM_DEFINE (scm_port_column, "port-column", 1, 0, 0, |
1bbd0b84 | 2260 | (SCM port), |
a150979d KR |
2261 | "Return the current column number of @var{port}.\n" |
2262 | "If the number is\n" | |
b380b885 MD |
2263 | "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n" |
2264 | "- i.e. the first character of the first line is line 0, column 0.\n" | |
2265 | "(However, when you display a file position, for example in an error\n" | |
650a1cf9 | 2266 | "message, we recommend you add 1 to get 1-origin integers. This is\n" |
b380b885 MD |
2267 | "because lines and column numbers traditionally start with 1, and that is\n" |
2268 | "what non-programmers will find most natural.)") | |
1bbd0b84 | 2269 | #define FUNC_NAME s_scm_port_column |
0f2d19dd | 2270 | { |
78446828 | 2271 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2272 | SCM_VALIDATE_OPENPORT (1, port); |
e11e83f3 | 2273 | return scm_from_int (SCM_COL (port)); |
0f2d19dd | 2274 | } |
1bbd0b84 | 2275 | #undef FUNC_NAME |
0f2d19dd | 2276 | |
a1ec6916 | 2277 | SCM_DEFINE (scm_set_port_column_x, "set-port-column!", 2, 0, 0, |
1bbd0b84 | 2278 | (SCM port, SCM column), |
a150979d KR |
2279 | "Set the current column of @var{port}. Before reading the first\n" |
2280 | "character on a line the column should be 0.") | |
1bbd0b84 | 2281 | #define FUNC_NAME s_scm_set_port_column_x |
d043d8c2 | 2282 | { |
360fc44c | 2283 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2284 | SCM_VALIDATE_OPENPORT (1, port); |
a55c2b68 | 2285 | SCM_PTAB_ENTRY (port)->column_number = scm_to_int (column); |
564478fd | 2286 | return SCM_UNSPECIFIED; |
d043d8c2 | 2287 | } |
1bbd0b84 | 2288 | #undef FUNC_NAME |
d043d8c2 | 2289 | |
a1ec6916 | 2290 | SCM_DEFINE (scm_port_filename, "port-filename", 1, 0, 0, |
1bbd0b84 | 2291 | (SCM port), |
ac012a27 AW |
2292 | "Return the filename associated with @var{port}, or @code{#f}\n" |
2293 | "if no filename is associated with the port.") | |
1bbd0b84 | 2294 | #define FUNC_NAME s_scm_port_filename |
0f2d19dd | 2295 | { |
78446828 | 2296 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2297 | SCM_VALIDATE_OPENPORT (1, port); |
b24b5e13 | 2298 | return SCM_FILENAME (port); |
0f2d19dd | 2299 | } |
1bbd0b84 | 2300 | #undef FUNC_NAME |
0f2d19dd | 2301 | |
a1ec6916 | 2302 | SCM_DEFINE (scm_set_port_filename_x, "set-port-filename!", 2, 0, 0, |
1bbd0b84 | 2303 | (SCM port, SCM filename), |
b380b885 MD |
2304 | "Change the filename associated with @var{port}, using the current input\n" |
2305 | "port if none is specified. Note that this does not change the port's\n" | |
2306 | "source of data, but only the value that is returned by\n" | |
2307 | "@code{port-filename} and reported in diagnostic output.") | |
1bbd0b84 | 2308 | #define FUNC_NAME s_scm_set_port_filename_x |
d14af9f2 | 2309 | { |
360fc44c | 2310 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2311 | SCM_VALIDATE_OPENPORT (1, port); |
360fc44c | 2312 | /* We allow the user to set the filename to whatever he likes. */ |
b24b5e13 DH |
2313 | SCM_SET_FILENAME (port, filename); |
2314 | return SCM_UNSPECIFIED; | |
d14af9f2 | 2315 | } |
1bbd0b84 | 2316 | #undef FUNC_NAME |
d14af9f2 | 2317 | |
d6a6989e LC |
2318 | /* A fluid specifying the default encoding for newly created ports. If it is |
2319 | a string, that is the encoding. If it is #f, it is in the "native" | |
2320 | (Latin-1) encoding. */ | |
2321 | SCM_VARIABLE (default_port_encoding_var, "%default-port-encoding"); | |
2322 | ||
889975e5 MG |
2323 | static int scm_port_encoding_init = 0; |
2324 | ||
9d9c66ba LC |
2325 | /* Use ENCODING as the default encoding for future ports. */ |
2326 | void | |
2327 | scm_i_set_default_port_encoding (const char *encoding) | |
2328 | { | |
2329 | if (!scm_port_encoding_init | |
2330 | || !scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var))) | |
2331 | scm_misc_error (NULL, "tried to set port encoding fluid before it is initialized", | |
2332 | SCM_EOL); | |
2333 | ||
2334 | if (encoding == NULL | |
96965a6e MW |
2335 | || c_strcasecmp (encoding, "ASCII") == 0 |
2336 | || c_strcasecmp (encoding, "ANSI_X3.4-1968") == 0 | |
2337 | || c_strcasecmp (encoding, "ISO-8859-1") == 0) | |
9d9c66ba LC |
2338 | scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), SCM_BOOL_F); |
2339 | else | |
2340 | scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), | |
2341 | scm_from_locale_string (encoding)); | |
2342 | } | |
2343 | ||
2344 | /* Return the name of the default encoding for newly created ports; a | |
2345 | return value of NULL means "ISO-8859-1". */ | |
889975e5 | 2346 | const char * |
9d9c66ba | 2347 | scm_i_default_port_encoding (void) |
889975e5 | 2348 | { |
9d9c66ba LC |
2349 | if (!scm_port_encoding_init) |
2350 | return NULL; | |
2351 | else if (!scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var))) | |
2352 | return NULL; | |
2353 | else | |
889975e5 | 2354 | { |
9d9c66ba LC |
2355 | SCM encoding; |
2356 | ||
2357 | encoding = scm_fluid_ref (SCM_VARIABLE_REF (default_port_encoding_var)); | |
2358 | if (!scm_is_string (encoding)) | |
889975e5 MG |
2359 | return NULL; |
2360 | else | |
9d9c66ba | 2361 | return scm_i_string_chars (encoding); |
889975e5 MG |
2362 | } |
2363 | } | |
2364 | ||
cdd3d6c9 MW |
2365 | /* If the next LEN bytes from PORT are equal to those in BYTES, then |
2366 | return 1, else return 0. Leave the port position unchanged. */ | |
2367 | static int | |
2368 | looking_at_bytes (SCM port, const unsigned char *bytes, int len) | |
2369 | { | |
2370 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
cdd3d6c9 MW |
2371 | int i = 0; |
2372 | ||
2373 | while (i < len && scm_peek_byte_or_eof (port) == bytes[i]) | |
2374 | { | |
2375 | pt->read_pos++; | |
2376 | i++; | |
2377 | } | |
7f6c3f8f MW |
2378 | scm_i_unget_bytes (bytes, i, port); |
2379 | return (i == len); | |
cdd3d6c9 MW |
2380 | } |
2381 | ||
2382 | static const unsigned char scm_utf8_bom[3] = {0xEF, 0xBB, 0xBF}; | |
2383 | static const unsigned char scm_utf16be_bom[2] = {0xFE, 0xFF}; | |
2384 | static const unsigned char scm_utf16le_bom[2] = {0xFF, 0xFE}; | |
2385 | static const unsigned char scm_utf32be_bom[4] = {0x00, 0x00, 0xFE, 0xFF}; | |
2386 | static const unsigned char scm_utf32le_bom[4] = {0xFF, 0xFE, 0x00, 0x00}; | |
2387 | ||
2388 | /* Decide what byte order to use for a UTF-16 port. Return "UTF-16BE" | |
2389 | or "UTF-16LE". MODE must be either SCM_PORT_READ or SCM_PORT_WRITE, | |
2390 | and specifies which operation is about to be done. The MODE | |
2391 | determines how we will decide the byte order. We deliberately avoid | |
2392 | reading from the port unless the user is about to do so. If the user | |
2393 | is about to read, then we look for a BOM, and if present, we use it | |
2394 | to determine the byte order. Otherwise we choose big endian, as | |
2395 | recommended by the Unicode Standard. Note that the BOM (if any) is | |
2396 | not consumed here. */ | |
2397 | static const char * | |
2398 | decide_utf16_encoding (SCM port, scm_t_port_rw_active mode) | |
2399 | { | |
2400 | if (mode == SCM_PORT_READ | |
2401 | && SCM_PORT_GET_INTERNAL (port)->at_stream_start_for_bom_read | |
2402 | && looking_at_bytes (port, scm_utf16le_bom, sizeof scm_utf16le_bom)) | |
2403 | return "UTF-16LE"; | |
2404 | else | |
2405 | return "UTF-16BE"; | |
2406 | } | |
2407 | ||
2408 | /* Decide what byte order to use for a UTF-32 port. Return "UTF-32BE" | |
2409 | or "UTF-32LE". See the comment above 'decide_utf16_encoding' for | |
2410 | details. */ | |
2411 | static const char * | |
2412 | decide_utf32_encoding (SCM port, scm_t_port_rw_active mode) | |
2413 | { | |
2414 | if (mode == SCM_PORT_READ | |
2415 | && SCM_PORT_GET_INTERNAL (port)->at_stream_start_for_bom_read | |
2416 | && looking_at_bytes (port, scm_utf32le_bom, sizeof scm_utf32le_bom)) | |
2417 | return "UTF-32LE"; | |
2418 | else | |
2419 | return "UTF-32BE"; | |
2420 | } | |
2421 | ||
337edc59 | 2422 | static void |
1fa89aaf | 2423 | finalize_iconv_descriptors (void *ptr, void *data) |
337edc59 MW |
2424 | { |
2425 | close_iconv_descriptors (ptr); | |
2426 | } | |
2427 | ||
2428 | static scm_t_iconv_descriptors * | |
2429 | open_iconv_descriptors (const char *encoding, int reading, int writing) | |
2430 | { | |
2431 | scm_t_iconv_descriptors *id; | |
2432 | iconv_t input_cd, output_cd; | |
2433 | ||
2434 | input_cd = (iconv_t) -1; | |
2435 | output_cd = (iconv_t) -1; | |
2436 | if (reading) | |
2437 | { | |
2438 | /* Open an input iconv conversion descriptor, from ENCODING | |
2439 | to UTF-8. We choose UTF-8, not UTF-32, because iconv | |
2440 | implementations can typically convert from anything to | |
2441 | UTF-8, but not to UTF-32 (see | |
2442 | <http://lists.gnu.org/archive/html/bug-libunistring/2010-09/msg00007.html>). */ | |
2443 | ||
2444 | /* Assume opening an iconv descriptor causes about 16 KB of | |
2445 | allocation. */ | |
2446 | scm_gc_register_allocation (16 * 1024); | |
2447 | ||
2448 | input_cd = iconv_open ("UTF-8", encoding); | |
2449 | if (input_cd == (iconv_t) -1) | |
2450 | goto invalid_encoding; | |
2451 | } | |
2452 | ||
2453 | if (writing) | |
2454 | { | |
2455 | /* Assume opening an iconv descriptor causes about 16 KB of | |
2456 | allocation. */ | |
2457 | scm_gc_register_allocation (16 * 1024); | |
2458 | ||
2459 | output_cd = iconv_open (encoding, "UTF-8"); | |
2460 | if (output_cd == (iconv_t) -1) | |
2461 | { | |
2462 | if (input_cd != (iconv_t) -1) | |
2463 | iconv_close (input_cd); | |
2464 | goto invalid_encoding; | |
2465 | } | |
2466 | } | |
2467 | ||
2468 | id = scm_gc_malloc_pointerless (sizeof (*id), "iconv descriptors"); | |
2469 | id->input_cd = input_cd; | |
2470 | id->output_cd = output_cd; | |
2471 | ||
2472 | /* Register a finalizer to close the descriptors. */ | |
2473 | scm_i_set_finalizer (id, finalize_iconv_descriptors, NULL); | |
2474 | ||
2475 | return id; | |
2476 | ||
2477 | invalid_encoding: | |
2478 | { | |
2479 | SCM err; | |
2480 | err = scm_from_locale_string (encoding); | |
2481 | scm_misc_error ("open_iconv_descriptors", | |
2482 | "invalid or unknown character encoding ~s", | |
2483 | scm_list_1 (err)); | |
2484 | } | |
2485 | } | |
2486 | ||
2487 | static void | |
2488 | close_iconv_descriptors (scm_t_iconv_descriptors *id) | |
2489 | { | |
2490 | if (id->input_cd != (iconv_t) -1) | |
2491 | iconv_close (id->input_cd); | |
2492 | if (id->output_cd != (iconv_t) -1) | |
2493 | iconv_close (id->output_cd); | |
2494 | id->input_cd = (void *) -1; | |
2495 | id->output_cd = (void *) -1; | |
2496 | } | |
2497 | ||
cdd3d6c9 MW |
2498 | /* Return the iconv_descriptors, initializing them if necessary. MODE |
2499 | must be either SCM_PORT_READ or SCM_PORT_WRITE, and specifies which | |
2500 | operation is about to be done. We deliberately avoid reading from | |
2501 | the port unless the user was about to do so. */ | |
337edc59 | 2502 | scm_t_iconv_descriptors * |
cdd3d6c9 | 2503 | scm_i_port_iconv_descriptors (SCM port, scm_t_port_rw_active mode) |
337edc59 | 2504 | { |
cdd3d6c9 | 2505 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
337edc59 MW |
2506 | |
2507 | assert (pti->encoding_mode == SCM_PORT_ENCODING_MODE_ICONV); | |
2508 | ||
2509 | if (!pti->iconv_descriptors) | |
2510 | { | |
cdd3d6c9 MW |
2511 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
2512 | const char *precise_encoding; | |
2513 | ||
337edc59 MW |
2514 | if (!pt->encoding) |
2515 | pt->encoding = "ISO-8859-1"; | |
cdd3d6c9 MW |
2516 | |
2517 | /* If the specified encoding is UTF-16 or UTF-32, then make | |
2518 | that more precise by deciding what byte order to use. */ | |
96965a6e | 2519 | if (c_strcasecmp (pt->encoding, "UTF-16") == 0) |
cdd3d6c9 | 2520 | precise_encoding = decide_utf16_encoding (port, mode); |
96965a6e | 2521 | else if (c_strcasecmp (pt->encoding, "UTF-32") == 0) |
cdd3d6c9 MW |
2522 | precise_encoding = decide_utf32_encoding (port, mode); |
2523 | else | |
2524 | precise_encoding = pt->encoding; | |
2525 | ||
337edc59 | 2526 | pti->iconv_descriptors = |
cdd3d6c9 | 2527 | open_iconv_descriptors (precise_encoding, |
337edc59 MW |
2528 | SCM_INPUT_PORT_P (port), |
2529 | SCM_OUTPUT_PORT_P (port)); | |
2530 | } | |
2531 | ||
2532 | return pti->iconv_descriptors; | |
2533 | } | |
2534 | ||
889975e5 | 2535 | void |
064c27c4 | 2536 | scm_i_set_port_encoding_x (SCM port, const char *encoding) |
889975e5 | 2537 | { |
889975e5 | 2538 | scm_t_port *pt; |
e4598559 | 2539 | scm_t_port_internal *pti; |
337edc59 | 2540 | scm_t_iconv_descriptors *prev; |
f4bc4e59 | 2541 | |
9d9c66ba LC |
2542 | /* Set the character encoding for this port. */ |
2543 | pt = SCM_PTAB_ENTRY (port); | |
e4598559 | 2544 | pti = SCM_PORT_GET_INTERNAL (port); |
337edc59 | 2545 | prev = pti->iconv_descriptors; |
f4bc4e59 | 2546 | |
cdd3d6c9 MW |
2547 | /* In order to handle cases where the encoding changes mid-stream |
2548 | (e.g. within an HTTP stream, or within a file that is composed of | |
2549 | segments with different encodings), we consider this to be "stream | |
2550 | start" for purposes of BOM handling, regardless of our actual file | |
2551 | position. */ | |
2552 | pti->at_stream_start_for_bom_read = 1; | |
2553 | pti->at_stream_start_for_bom_write = 1; | |
2554 | ||
9d9c66ba LC |
2555 | if (encoding == NULL) |
2556 | encoding = "ISO-8859-1"; | |
f4bc4e59 | 2557 | |
7b292a9d LC |
2558 | /* If ENCODING is UTF-8, then no conversion descriptor is opened |
2559 | because we do I/O ourselves. This saves 100+ KiB for each | |
2560 | descriptor. */ | |
cdd3d6c9 | 2561 | pt->encoding = scm_gc_strdup (encoding, "port"); |
96965a6e | 2562 | if (c_strcasecmp (encoding, "UTF-8") == 0) |
cdd3d6c9 | 2563 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8; |
337edc59 | 2564 | else |
cdd3d6c9 | 2565 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV; |
f4bc4e59 | 2566 | |
cdd3d6c9 | 2567 | pti->iconv_descriptors = NULL; |
337edc59 MW |
2568 | if (prev) |
2569 | close_iconv_descriptors (prev); | |
889975e5 MG |
2570 | } |
2571 | ||
2572 | SCM_DEFINE (scm_port_encoding, "port-encoding", 1, 0, 0, | |
2573 | (SCM port), | |
2574 | "Returns, as a string, the character encoding that @var{port}\n" | |
2575 | "uses to interpret its input and output.\n") | |
2576 | #define FUNC_NAME s_scm_port_encoding | |
2577 | { | |
2578 | scm_t_port *pt; | |
2579 | const char *enc; | |
2580 | ||
2581 | SCM_VALIDATE_PORT (1, port); | |
2582 | ||
2583 | pt = SCM_PTAB_ENTRY (port); | |
9d9c66ba | 2584 | enc = pt->encoding; |
889975e5 MG |
2585 | if (enc) |
2586 | return scm_from_locale_string (pt->encoding); | |
2587 | else | |
0af34a3f | 2588 | return SCM_BOOL_F; |
889975e5 MG |
2589 | } |
2590 | #undef FUNC_NAME | |
d6a6989e | 2591 | |
889975e5 MG |
2592 | SCM_DEFINE (scm_set_port_encoding_x, "set-port-encoding!", 2, 0, 0, |
2593 | (SCM port, SCM enc), | |
2594 | "Sets the character encoding that will be used to interpret all\n" | |
2595 | "port I/O. New ports are created with the encoding\n" | |
2596 | "appropriate for the current locale if @code{setlocale} has \n" | |
2597 | "been called or ISO-8859-1 otherwise\n" | |
2598 | "and this procedure can be used to modify that encoding.\n") | |
889975e5 MG |
2599 | #define FUNC_NAME s_scm_set_port_encoding_x |
2600 | { | |
2601 | char *enc_str; | |
889975e5 MG |
2602 | |
2603 | SCM_VALIDATE_PORT (1, port); | |
2604 | SCM_VALIDATE_STRING (2, enc); | |
2605 | ||
2606 | enc_str = scm_to_locale_string (enc); | |
da288f50 | 2607 | scm_i_set_port_encoding_x (port, enc_str); |
3e05fc04 | 2608 | free (enc_str); |
da288f50 | 2609 | |
889975e5 MG |
2610 | return SCM_UNSPECIFIED; |
2611 | } | |
2612 | #undef FUNC_NAME | |
2613 | ||
2614 | ||
b22e94db LC |
2615 | /* A fluid specifying the default conversion handler for newly created |
2616 | ports. Its value should be one of the symbols below. */ | |
2617 | SCM_VARIABLE (default_conversion_strategy_var, | |
2618 | "%default-port-conversion-strategy"); | |
2619 | ||
2620 | /* Whether the above fluid is initialized. */ | |
889975e5 MG |
2621 | static int scm_conversion_strategy_init = 0; |
2622 | ||
b22e94db LC |
2623 | /* The possible conversion strategies. */ |
2624 | SCM_SYMBOL (sym_error, "error"); | |
2625 | SCM_SYMBOL (sym_substitute, "substitute"); | |
2626 | SCM_SYMBOL (sym_escape, "escape"); | |
2627 | ||
2628 | /* Return the default failed encoding conversion policy for new created | |
2629 | ports. */ | |
889975e5 | 2630 | scm_t_string_failed_conversion_handler |
b22e94db | 2631 | scm_i_default_port_conversion_handler (void) |
889975e5 | 2632 | { |
b22e94db LC |
2633 | scm_t_string_failed_conversion_handler handler; |
2634 | ||
2635 | if (!scm_conversion_strategy_init | |
2636 | || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var))) | |
2637 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
889975e5 MG |
2638 | else |
2639 | { | |
b22e94db LC |
2640 | SCM fluid, value; |
2641 | ||
2642 | fluid = SCM_VARIABLE_REF (default_conversion_strategy_var); | |
2643 | value = scm_fluid_ref (fluid); | |
2644 | ||
2645 | if (scm_is_eq (sym_substitute, value)) | |
2646 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
2647 | else if (scm_is_eq (sym_escape, value)) | |
2648 | handler = SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE; | |
2649 | else | |
2650 | /* Default to 'error also when the fluid's value is not one of | |
2651 | the valid symbols. */ | |
2652 | handler = SCM_FAILED_CONVERSION_ERROR; | |
889975e5 | 2653 | } |
b22e94db LC |
2654 | |
2655 | return handler; | |
889975e5 MG |
2656 | } |
2657 | ||
b22e94db | 2658 | /* Use HANDLER as the default conversion strategy for future ports. */ |
889975e5 | 2659 | void |
b22e94db LC |
2660 | scm_i_set_default_port_conversion_handler (scm_t_string_failed_conversion_handler |
2661 | handler) | |
889975e5 MG |
2662 | { |
2663 | SCM strategy; | |
b22e94db LC |
2664 | |
2665 | if (!scm_conversion_strategy_init | |
2666 | || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var))) | |
2667 | scm_misc_error (NULL, "tried to set conversion strategy fluid before it is initialized", | |
2668 | SCM_EOL); | |
2669 | ||
2670 | switch (handler) | |
889975e5 | 2671 | { |
b22e94db LC |
2672 | case SCM_FAILED_CONVERSION_ERROR: |
2673 | strategy = sym_error; | |
2674 | break; | |
2675 | ||
2676 | case SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE: | |
2677 | strategy = sym_escape; | |
2678 | break; | |
2679 | ||
2680 | case SCM_FAILED_CONVERSION_QUESTION_MARK: | |
2681 | strategy = sym_substitute; | |
2682 | break; | |
2683 | ||
2684 | default: | |
2685 | abort (); | |
889975e5 | 2686 | } |
b22e94db LC |
2687 | |
2688 | scm_fluid_set_x (SCM_VARIABLE_REF (default_conversion_strategy_var), | |
2689 | strategy); | |
889975e5 MG |
2690 | } |
2691 | ||
2692 | SCM_DEFINE (scm_port_conversion_strategy, "port-conversion-strategy", | |
2693 | 1, 0, 0, (SCM port), | |
2694 | "Returns the behavior of the port when handling a character that\n" | |
2695 | "is not representable in the port's current encoding.\n" | |
2696 | "It returns the symbol @code{error} if unrepresentable characters\n" | |
2697 | "should cause exceptions, @code{substitute} if the port should\n" | |
2698 | "try to replace unrepresentable characters with question marks or\n" | |
2699 | "approximate characters, or @code{escape} if unrepresentable\n" | |
2700 | "characters should be converted to string escapes.\n" | |
2701 | "\n" | |
2702 | "If @var{port} is @code{#f}, then the current default behavior\n" | |
2703 | "will be returned. New ports will have this default behavior\n" | |
2704 | "when they are created.\n") | |
2705 | #define FUNC_NAME s_scm_port_conversion_strategy | |
2706 | { | |
2707 | scm_t_string_failed_conversion_handler h; | |
2708 | ||
b22e94db LC |
2709 | if (scm_is_false (port)) |
2710 | h = scm_i_default_port_conversion_handler (); | |
2711 | else | |
889975e5 | 2712 | { |
b22e94db LC |
2713 | scm_t_port *pt; |
2714 | ||
889975e5 | 2715 | SCM_VALIDATE_OPPORT (1, port); |
b22e94db LC |
2716 | pt = SCM_PTAB_ENTRY (port); |
2717 | ||
2718 | h = pt->ilseq_handler; | |
889975e5 MG |
2719 | } |
2720 | ||
889975e5 | 2721 | if (h == SCM_FAILED_CONVERSION_ERROR) |
4a655e50 | 2722 | return scm_from_latin1_symbol ("error"); |
889975e5 | 2723 | else if (h == SCM_FAILED_CONVERSION_QUESTION_MARK) |
4a655e50 | 2724 | return scm_from_latin1_symbol ("substitute"); |
889975e5 | 2725 | else if (h == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE) |
4a655e50 | 2726 | return scm_from_latin1_symbol ("escape"); |
889975e5 MG |
2727 | else |
2728 | abort (); | |
2729 | ||
2730 | /* Never gets here. */ | |
2731 | return SCM_UNDEFINED; | |
2732 | } | |
2733 | #undef FUNC_NAME | |
2734 | ||
2735 | SCM_DEFINE (scm_set_port_conversion_strategy_x, "set-port-conversion-strategy!", | |
2736 | 2, 0, 0, | |
2737 | (SCM port, SCM sym), | |
2738 | "Sets the behavior of the interpreter when outputting a character\n" | |
2739 | "that is not representable in the port's current encoding.\n" | |
2740 | "@var{sym} can be either @code{'error}, @code{'substitute}, or\n" | |
2741 | "@code{'escape}. If it is @code{'error}, an error will be thrown\n" | |
2742 | "when an unconvertible character is encountered. If it is\n" | |
2743 | "@code{'substitute}, then unconvertible characters will \n" | |
2744 | "be replaced with approximate characters, or with question marks\n" | |
2745 | "if no approximately correct character is available.\n" | |
2746 | "If it is @code{'escape},\n" | |
2747 | "it will appear as a hex escape when output.\n" | |
2748 | "\n" | |
2749 | "If @var{port} is an open port, the conversion error behavior\n" | |
2750 | "is set for that port. If it is @code{#f}, it is set as the\n" | |
2751 | "default behavior for any future ports that get created in\n" | |
2752 | "this thread.\n") | |
2753 | #define FUNC_NAME s_scm_set_port_conversion_strategy_x | |
2754 | { | |
b22e94db | 2755 | scm_t_string_failed_conversion_handler handler; |
889975e5 | 2756 | |
b22e94db LC |
2757 | if (scm_is_eq (sym, sym_error)) |
2758 | handler = SCM_FAILED_CONVERSION_ERROR; | |
2759 | else if (scm_is_eq (sym, sym_substitute)) | |
2760 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
2761 | else if (scm_is_eq (sym, sym_escape)) | |
2762 | handler = SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE; | |
2763 | else | |
2764 | SCM_MISC_ERROR ("unknown conversion strategy ~s", scm_list_1 (sym)); | |
889975e5 | 2765 | |
b22e94db LC |
2766 | if (scm_is_false (port)) |
2767 | scm_i_set_default_port_conversion_handler (handler); | |
2768 | else | |
889975e5 | 2769 | { |
b22e94db LC |
2770 | SCM_VALIDATE_OPPORT (1, port); |
2771 | SCM_PTAB_ENTRY (port)->ilseq_handler = handler; | |
889975e5 MG |
2772 | } |
2773 | ||
889975e5 MG |
2774 | return SCM_UNSPECIFIED; |
2775 | } | |
2776 | #undef FUNC_NAME | |
2777 | ||
2778 | ||
2779 | ||
f12733c9 MD |
2780 | void |
2781 | scm_print_port_mode (SCM exp, SCM port) | |
2782 | { | |
2783 | scm_puts (SCM_CLOSEDP (exp) | |
2784 | ? "closed: " | |
f9a64404 DH |
2785 | : (SCM_RDNG & SCM_CELL_WORD_0 (exp) |
2786 | ? (SCM_WRTNG & SCM_CELL_WORD_0 (exp) | |
f12733c9 MD |
2787 | ? "input-output: " |
2788 | : "input: ") | |
f9a64404 | 2789 | : (SCM_WRTNG & SCM_CELL_WORD_0 (exp) |
f12733c9 MD |
2790 | ? "output: " |
2791 | : "bogus: ")), | |
2792 | port); | |
2793 | } | |
1cc91f1b | 2794 | |
f12733c9 | 2795 | int |
e81d98ec | 2796 | scm_port_print (SCM exp, SCM port, scm_print_state *pstate SCM_UNUSED) |
0f2d19dd | 2797 | { |
f12733c9 MD |
2798 | char *type = SCM_PTOBNAME (SCM_PTOBNUM (exp)); |
2799 | if (!type) | |
2800 | type = "port"; | |
b7f3516f | 2801 | scm_puts ("#<", port); |
f12733c9 | 2802 | scm_print_port_mode (exp, port); |
b7f3516f TT |
2803 | scm_puts (type, port); |
2804 | scm_putc (' ', port); | |
0345e278 | 2805 | scm_uintprint (SCM_CELL_WORD_1 (exp), 16, port); |
b7f3516f | 2806 | scm_putc ('>', port); |
f12733c9 | 2807 | return 1; |
0f2d19dd JB |
2808 | } |
2809 | ||
0f2d19dd | 2810 | \f |
ee149d03 | 2811 | |
d68fee48 | 2812 | /* Void ports. */ |
0f2d19dd | 2813 | |
92c2555f | 2814 | scm_t_bits scm_tc16_void_port = 0; |
0f2d19dd | 2815 | |
e81d98ec | 2816 | static int fill_input_void_port (SCM port SCM_UNUSED) |
283a1a0e | 2817 | { |
70df8af6 | 2818 | return EOF; |
283a1a0e GH |
2819 | } |
2820 | ||
31703ab8 | 2821 | static void |
e81d98ec DH |
2822 | write_void_port (SCM port SCM_UNUSED, |
2823 | const void *data SCM_UNUSED, | |
2824 | size_t size SCM_UNUSED) | |
31703ab8 GH |
2825 | { |
2826 | } | |
2827 | ||
d617ee18 MV |
2828 | static SCM |
2829 | scm_i_void_port (long mode_bits) | |
0f2d19dd | 2830 | { |
9de87eea | 2831 | scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex); |
402788a9 | 2832 | { |
da220f27 HWN |
2833 | SCM answer = scm_new_port_table_entry (scm_tc16_void_port); |
2834 | scm_t_port * pt = SCM_PTAB_ENTRY(answer); | |
2835 | ||
402788a9 | 2836 | scm_port_non_buffer (pt); |
402788a9 HWN |
2837 | |
2838 | SCM_SETSTREAM (answer, 0); | |
2839 | SCM_SET_CELL_TYPE (answer, scm_tc16_void_port | mode_bits); | |
9de87eea | 2840 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); |
402788a9 HWN |
2841 | return answer; |
2842 | } | |
0f2d19dd JB |
2843 | } |
2844 | ||
d617ee18 MV |
2845 | SCM |
2846 | scm_void_port (char *mode_str) | |
2847 | { | |
2848 | return scm_i_void_port (scm_mode_bits (mode_str)); | |
2849 | } | |
2850 | ||
a1ec6916 | 2851 | SCM_DEFINE (scm_sys_make_void_port, "%make-void-port", 1, 0, 0, |
1bbd0b84 | 2852 | (SCM mode), |
70df8af6 | 2853 | "Create and return a new void port. A void port acts like\n" |
bb2c02f2 | 2854 | "@file{/dev/null}. The @var{mode} argument\n" |
70df8af6 | 2855 | "specifies the input/output modes for this port: see the\n" |
b380b885 | 2856 | "documentation for @code{open-file} in @ref{File Ports}.") |
1bbd0b84 | 2857 | #define FUNC_NAME s_scm_sys_make_void_port |
0f2d19dd | 2858 | { |
d617ee18 | 2859 | return scm_i_void_port (scm_i_mode_bits (mode)); |
0f2d19dd | 2860 | } |
1bbd0b84 | 2861 | #undef FUNC_NAME |
0f2d19dd | 2862 | |
0f2d19dd | 2863 | \f |
89545eba | 2864 | /* Initialization. */ |
1cc91f1b | 2865 | |
0f2d19dd JB |
2866 | void |
2867 | scm_init_ports () | |
0f2d19dd | 2868 | { |
840ae05d | 2869 | /* lseek() symbols. */ |
e11e83f3 MV |
2870 | scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET)); |
2871 | scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR)); | |
2872 | scm_c_define ("SEEK_END", scm_from_int (SEEK_END)); | |
840ae05d | 2873 | |
70df8af6 GH |
2874 | scm_tc16_void_port = scm_make_port_type ("void", fill_input_void_port, |
2875 | write_void_port); | |
9de87eea | 2876 | |
f39448c5 AW |
2877 | cur_inport_fluid = scm_make_fluid (); |
2878 | cur_outport_fluid = scm_make_fluid (); | |
2879 | cur_errport_fluid = scm_make_fluid (); | |
2880 | cur_loadport_fluid = scm_make_fluid (); | |
9de87eea | 2881 | |
f39448c5 | 2882 | scm_i_port_weak_hash = scm_make_weak_key_hash_table (SCM_I_MAKINUM(31)); |
d6a6989e | 2883 | |
a0599745 | 2884 | #include "libguile/ports.x" |
889975e5 | 2885 | |
d6a6989e | 2886 | /* Use Latin-1 as the default port encoding. */ |
c81c2ad3 AW |
2887 | SCM_VARIABLE_SET (default_port_encoding_var, |
2888 | scm_make_fluid_with_default (SCM_BOOL_F)); | |
889975e5 | 2889 | scm_port_encoding_init = 1; |
d6a6989e | 2890 | |
b22e94db LC |
2891 | SCM_VARIABLE_SET (default_conversion_strategy_var, |
2892 | scm_make_fluid_with_default (sym_substitute)); | |
889975e5 | 2893 | scm_conversion_strategy_init = 1; |
b22e94db | 2894 | |
9670f238 AW |
2895 | /* These bindings are used when boot-9 turns `current-input-port' et |
2896 | al into parameters. They are then removed from the guile module. */ | |
2897 | scm_c_define ("%current-input-port-fluid", cur_inport_fluid); | |
2898 | scm_c_define ("%current-output-port-fluid", cur_outport_fluid); | |
2899 | scm_c_define ("%current-error-port-fluid", cur_errport_fluid); | |
0f2d19dd | 2900 | } |
89e00824 ML |
2901 | |
2902 | /* | |
2903 | Local Variables: | |
2904 | c-file-style: "gnu" | |
2905 | End: | |
2906 | */ |