Commit | Line | Data |
---|---|---|
e4598559 | 1 | /* Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, 2006, |
bc8e6d7d MW |
2 | * 2007, 2008, 2009, 2010, 2011, 2012, 2013, |
3 | * 2014 Free Software Foundation, Inc. | |
f4bc4e59 | 4 | * |
73be1d9e | 5 | * This library is free software; you can redistribute it and/or |
53befeb7 NJ |
6 | * modify it under the terms of the GNU Lesser General Public License |
7 | * as published by the Free Software Foundation; either version 3 of | |
8 | * the License, or (at your option) any later version. | |
0f2d19dd | 9 | * |
53befeb7 NJ |
10 | * This library is distributed in the hope that it will be useful, but |
11 | * WITHOUT ANY WARRANTY; without even the implied warranty of | |
73be1d9e MV |
12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
13 | * Lesser General Public License for more details. | |
0f2d19dd | 14 | * |
73be1d9e MV |
15 | * You should have received a copy of the GNU Lesser General Public |
16 | * License along with this library; if not, write to the Free Software | |
53befeb7 NJ |
17 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA |
18 | * 02110-1301 USA | |
73be1d9e | 19 | */ |
1bbd0b84 | 20 | |
1bbd0b84 | 21 | |
0f2d19dd | 22 | \f |
d68fee48 JB |
23 | /* Headers. */ |
24 | ||
2b829bbb KR |
25 | #define _LARGEFILE64_SOURCE /* ask for stat64 etc */ |
26 | ||
dbb605f5 | 27 | #ifdef HAVE_CONFIG_H |
bd515f37 RB |
28 | # include <config.h> |
29 | #endif | |
30 | ||
0f2d19dd | 31 | #include <stdio.h> |
e6e2e95a | 32 | #include <errno.h> |
8ab3d8a0 | 33 | #include <fcntl.h> /* for chsize on mingw */ |
b5cb4464 | 34 | #include <assert.h> |
f4bc4e59 | 35 | #include <iconv.h> |
889975e5 MG |
36 | #include <uniconv.h> |
37 | #include <unistr.h> | |
38 | #include <striconveh.h> | |
96965a6e | 39 | #include <c-strcase.h> |
e6e2e95a | 40 | |
fca43887 LC |
41 | #include <assert.h> |
42 | ||
a0599745 | 43 | #include "libguile/_scm.h" |
4e047c3e | 44 | #include "libguile/async.h" |
8269ba5b | 45 | #include "libguile/deprecation.h" |
f0942910 | 46 | #include "libguile/eval.h" |
8ab3d8a0 | 47 | #include "libguile/fports.h" /* direct access for seek and truncate */ |
9511876f | 48 | #include "libguile/goops.h" |
a0599745 MD |
49 | #include "libguile/smob.h" |
50 | #include "libguile/chars.h" | |
185e369a | 51 | #include "libguile/dynwind.h" |
0f2d19dd | 52 | |
a0599745 | 53 | #include "libguile/keywords.h" |
5dbc6c06 | 54 | #include "libguile/hashtab.h" |
a0599745 MD |
55 | #include "libguile/root.h" |
56 | #include "libguile/strings.h" | |
b42170a4 | 57 | #include "libguile/mallocs.h" |
a0599745 MD |
58 | #include "libguile/validate.h" |
59 | #include "libguile/ports.h" | |
e4598559 | 60 | #include "libguile/ports-internal.h" |
3a5fb14d | 61 | #include "libguile/vectors.h" |
5dbc6c06 | 62 | #include "libguile/weaks.h" |
9de87eea | 63 | #include "libguile/fluids.h" |
889975e5 | 64 | #include "libguile/eq.h" |
a38024ba | 65 | #include "libguile/alist.h" |
0f2d19dd | 66 | |
bd9e24b3 GH |
67 | #ifdef HAVE_STRING_H |
68 | #include <string.h> | |
69 | #endif | |
70 | ||
ec65f5da MV |
71 | #ifdef HAVE_IO_H |
72 | #include <io.h> | |
73 | #endif | |
74 | ||
0f2d19dd | 75 | #include <unistd.h> |
0f2d19dd | 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 | ||
a38024ba MW |
257 | SCM_DEFINE (scm_i_port_property, "%port-property", 2, 0, 0, |
258 | (SCM port, SCM key), | |
259 | "Return the property of @var{port} associated with @var{key}.") | |
260 | #define FUNC_NAME s_scm_i_port_property | |
05d7f762 | 261 | { |
a38024ba MW |
262 | SCM_VALIDATE_OPPORT (1, port); |
263 | return scm_assq_ref (SCM_PORT_GET_INTERNAL (port)->alist, key); | |
05d7f762 | 264 | } |
a38024ba | 265 | #undef FUNC_NAME |
05d7f762 | 266 | |
a38024ba MW |
267 | SCM_DEFINE (scm_i_set_port_property_x, "%set-port-property!", 3, 0, 0, |
268 | (SCM port, SCM key, SCM value), | |
269 | "Set the property of @var{port} associated with @var{key} to @var{value}.") | |
270 | #define FUNC_NAME s_scm_i_set_port_property_x | |
05d7f762 | 271 | { |
a38024ba MW |
272 | scm_t_port_internal *pti; |
273 | ||
274 | SCM_VALIDATE_OPPORT (1, port); | |
275 | pti = SCM_PORT_GET_INTERNAL (port); | |
276 | pti->alist = scm_assq_set_x (pti->alist, key, value); | |
277 | return SCM_UNSPECIFIED; | |
05d7f762 | 278 | } |
a38024ba | 279 | #undef FUNC_NAME |
05d7f762 | 280 | |
0f2d19dd | 281 | \f |
0f2d19dd | 282 | |
3b3b36dd | 283 | SCM_DEFINE (scm_char_ready_p, "char-ready?", 0, 1, 0, |
1e6808ea MG |
284 | (SCM port), |
285 | "Return @code{#t} if a character is ready on input @var{port}\n" | |
286 | "and return @code{#f} otherwise. If @code{char-ready?} returns\n" | |
287 | "@code{#t} then the next @code{read-char} operation on\n" | |
288 | "@var{port} is guaranteed not to hang. If @var{port} is a file\n" | |
289 | "port at end of file then @code{char-ready?} returns @code{#t}.\n" | |
c2dfff19 KR |
290 | "\n" |
291 | "@code{char-ready?} exists to make it possible for a\n" | |
1e6808ea MG |
292 | "program to accept characters from interactive ports without\n" |
293 | "getting stuck waiting for input. Any input editors associated\n" | |
294 | "with such ports must make sure that characters whose existence\n" | |
295 | "has been asserted by @code{char-ready?} cannot be rubbed out.\n" | |
296 | "If @code{char-ready?} were to return @code{#f} at end of file,\n" | |
297 | "a port at end of file would be indistinguishable from an\n" | |
c2dfff19 | 298 | "interactive port that has no ready characters.") |
1bbd0b84 | 299 | #define FUNC_NAME s_scm_char_ready_p |
0f2d19dd | 300 | { |
92c2555f | 301 | scm_t_port *pt; |
6c951427 | 302 | |
0f2d19dd | 303 | if (SCM_UNBNDP (port)) |
9de87eea | 304 | port = scm_current_input_port (); |
b2456dd4 AW |
305 | /* It's possible to close the current input port, so validate even in |
306 | this case. */ | |
307 | SCM_VALIDATE_OPINPORT (1, port); | |
d68fee48 | 308 | |
ae4c4016 JB |
309 | pt = SCM_PTAB_ENTRY (port); |
310 | ||
6c951427 GH |
311 | /* if the current read buffer is filled, or the |
312 | last pushed-back char has been read and the saved buffer is | |
313 | filled, result is true. */ | |
314 | if (pt->read_pos < pt->read_end | |
315 | || (pt->read_buf == pt->putback_buf | |
316 | && pt->saved_read_pos < pt->saved_read_end)) | |
0f2d19dd | 317 | return SCM_BOOL_T; |
ee149d03 JB |
318 | else |
319 | { | |
92c2555f | 320 | scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)]; |
ee149d03 | 321 | |
affc96b5 | 322 | if (ptob->input_waiting) |
7888309b | 323 | return scm_from_bool(ptob->input_waiting (port)); |
ee149d03 | 324 | else |
6c951427 | 325 | return SCM_BOOL_T; |
ee149d03 | 326 | } |
0f2d19dd | 327 | } |
1bbd0b84 | 328 | #undef FUNC_NAME |
0f2d19dd | 329 | |
e8b21eec LC |
330 | /* Move up to READ_LEN bytes from PORT's putback and/or read buffers |
331 | into memory starting at DEST. Return the number of bytes moved. | |
332 | PORT's line/column numbers are left unchanged. */ | |
333 | size_t | |
334 | scm_take_from_input_buffers (SCM port, char *dest, size_t read_len) | |
c2da2648 | 335 | { |
92c2555f | 336 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
e8b21eec | 337 | size_t bytes_read = 0; |
c2da2648 GH |
338 | size_t from_buf = min (pt->read_end - pt->read_pos, read_len); |
339 | ||
340 | if (from_buf > 0) | |
341 | { | |
342 | memcpy (dest, pt->read_pos, from_buf); | |
343 | pt->read_pos += from_buf; | |
e8b21eec | 344 | bytes_read += from_buf; |
c2da2648 GH |
345 | read_len -= from_buf; |
346 | dest += from_buf; | |
347 | } | |
348 | ||
349 | /* if putback was active, try the real input buffer too. */ | |
350 | if (pt->read_buf == pt->putback_buf) | |
351 | { | |
352 | from_buf = min (pt->saved_read_end - pt->saved_read_pos, read_len); | |
353 | if (from_buf > 0) | |
354 | { | |
355 | memcpy (dest, pt->saved_read_pos, from_buf); | |
356 | pt->saved_read_pos += from_buf; | |
e8b21eec | 357 | bytes_read += from_buf; |
c2da2648 GH |
358 | } |
359 | } | |
e8b21eec LC |
360 | |
361 | return bytes_read; | |
c2da2648 GH |
362 | } |
363 | ||
6c951427 | 364 | /* Clear a port's read buffers, returning the contents. */ |
a1ec6916 | 365 | SCM_DEFINE (scm_drain_input, "drain-input", 1, 0, 0, |
1bbd0b84 | 366 | (SCM port), |
4a151b3d GH |
367 | "This procedure clears a port's input buffers, similar\n" |
368 | "to the way that force-output clears the output buffer. The\n" | |
369 | "contents of the buffers are returned as a single string, e.g.,\n" | |
370 | "\n" | |
371 | "@lisp\n" | |
372 | "(define p (open-input-file ...))\n" | |
373 | "(drain-input p) => empty string, nothing buffered yet.\n" | |
374 | "(unread-char (read-char p) p)\n" | |
375 | "(drain-input p) => initial chars from p, up to the buffer size.\n" | |
376 | "@end lisp\n\n" | |
377 | "Draining the buffers may be useful for cleanly finishing\n" | |
378 | "buffered I/O so that the file descriptor can be used directly\n" | |
379 | "for further input.") | |
1bbd0b84 | 380 | #define FUNC_NAME s_scm_drain_input |
ee149d03 | 381 | { |
840ae05d | 382 | SCM result; |
3a5fb14d | 383 | char *data; |
28a6e1b0 | 384 | scm_t_port *pt; |
c014a02e | 385 | long count; |
ee149d03 | 386 | |
34d19ef6 | 387 | SCM_VALIDATE_OPINPORT (1, port); |
28a6e1b0 | 388 | pt = SCM_PTAB_ENTRY (port); |
840ae05d | 389 | |
6c951427 GH |
390 | count = pt->read_end - pt->read_pos; |
391 | if (pt->read_buf == pt->putback_buf) | |
392 | count += pt->saved_read_end - pt->saved_read_pos; | |
840ae05d | 393 | |
67a72dc1 AW |
394 | if (count) |
395 | { | |
190d4b0d | 396 | result = scm_i_make_string (count, &data, 0); |
67a72dc1 AW |
397 | scm_take_from_input_buffers (port, data, count); |
398 | } | |
399 | else | |
400 | result = scm_nullstr; | |
401 | ||
840ae05d | 402 | return result; |
ee149d03 | 403 | } |
1bbd0b84 | 404 | #undef FUNC_NAME |
0f2d19dd JB |
405 | |
406 | \f | |
d68fee48 | 407 | /* Standard ports --- current input, output, error, and more(!). */ |
0f2d19dd | 408 | |
34297700 AW |
409 | static SCM cur_inport_fluid = SCM_BOOL_F; |
410 | static SCM cur_outport_fluid = SCM_BOOL_F; | |
411 | static SCM cur_errport_fluid = SCM_BOOL_F; | |
412 | static SCM cur_loadport_fluid = SCM_BOOL_F; | |
9de87eea | 413 | |
3b3b36dd | 414 | SCM_DEFINE (scm_current_input_port, "current-input-port", 0, 0, 0, |
e1546b65 MG |
415 | (), |
416 | "Return the current input port. This is the default port used\n" | |
417 | "by many input procedures. Initially, @code{current-input-port}\n" | |
418 | "returns the @dfn{standard input} in Unix and C terminology.") | |
1bbd0b84 | 419 | #define FUNC_NAME s_scm_current_input_port |
0f2d19dd | 420 | { |
34297700 | 421 | if (scm_is_true (cur_inport_fluid)) |
889975e5 MG |
422 | return scm_fluid_ref (cur_inport_fluid); |
423 | else | |
424 | return SCM_BOOL_F; | |
0f2d19dd | 425 | } |
1bbd0b84 | 426 | #undef FUNC_NAME |
0f2d19dd | 427 | |
3b3b36dd | 428 | SCM_DEFINE (scm_current_output_port, "current-output-port", 0, 0, 0, |
e1546b65 MG |
429 | (), |
430 | "Return the current output port. This is the default port used\n" | |
9401323e | 431 | "by many output procedures. Initially,\n" |
e1546b65 MG |
432 | "@code{current-output-port} returns the @dfn{standard output} in\n" |
433 | "Unix and C terminology.") | |
1bbd0b84 | 434 | #define FUNC_NAME s_scm_current_output_port |
0f2d19dd | 435 | { |
34297700 | 436 | if (scm_is_true (cur_outport_fluid)) |
889975e5 MG |
437 | return scm_fluid_ref (cur_outport_fluid); |
438 | else | |
439 | return SCM_BOOL_F; | |
0f2d19dd | 440 | } |
1bbd0b84 | 441 | #undef FUNC_NAME |
0f2d19dd | 442 | |
3b3b36dd | 443 | SCM_DEFINE (scm_current_error_port, "current-error-port", 0, 0, 0, |
1bbd0b84 | 444 | (), |
b380b885 MD |
445 | "Return the port to which errors and warnings should be sent (the\n" |
446 | "@dfn{standard error} in Unix and C terminology).") | |
1bbd0b84 | 447 | #define FUNC_NAME s_scm_current_error_port |
0f2d19dd | 448 | { |
34297700 | 449 | if (scm_is_true (cur_errport_fluid)) |
889975e5 MG |
450 | return scm_fluid_ref (cur_errport_fluid); |
451 | else | |
452 | return SCM_BOOL_F; | |
0f2d19dd | 453 | } |
1bbd0b84 | 454 | #undef FUNC_NAME |
0f2d19dd | 455 | |
60617d81 MW |
456 | static SCM current_warning_port_var; |
457 | static scm_i_pthread_once_t current_warning_port_once = SCM_I_PTHREAD_ONCE_INIT; | |
458 | ||
459 | static void | |
460 | init_current_warning_port_var (void) | |
461 | { | |
462 | current_warning_port_var | |
463 | = scm_c_private_variable ("guile", "current-warning-port"); | |
464 | } | |
465 | ||
3972de76 AW |
466 | SCM |
467 | scm_current_warning_port (void) | |
468 | { | |
60617d81 MW |
469 | scm_i_pthread_once (¤t_warning_port_once, |
470 | init_current_warning_port_var); | |
471 | return scm_call_0 (scm_variable_ref (current_warning_port_var)); | |
3972de76 AW |
472 | } |
473 | ||
3b3b36dd | 474 | SCM_DEFINE (scm_current_load_port, "current-load-port", 0, 0, 0, |
e1546b65 | 475 | (), |
b450f070 | 476 | "Return the current-load-port.\n" |
e1546b65 | 477 | "The load port is used internally by @code{primitive-load}.") |
1bbd0b84 | 478 | #define FUNC_NAME s_scm_current_load_port |
31614d8e | 479 | { |
9de87eea | 480 | return scm_fluid_ref (cur_loadport_fluid); |
31614d8e | 481 | } |
1bbd0b84 | 482 | #undef FUNC_NAME |
31614d8e | 483 | |
3b3b36dd | 484 | SCM_DEFINE (scm_set_current_input_port, "set-current-input-port", 1, 0, 0, |
1bbd0b84 | 485 | (SCM port), |
8f85c0c6 NJ |
486 | "@deffnx {Scheme Procedure} set-current-output-port port\n" |
487 | "@deffnx {Scheme Procedure} set-current-error-port port\n" | |
b380b885 MD |
488 | "Change the ports returned by @code{current-input-port},\n" |
489 | "@code{current-output-port} and @code{current-error-port}, respectively,\n" | |
490 | "so that they use the supplied @var{port} for input or output.") | |
1bbd0b84 | 491 | #define FUNC_NAME s_scm_set_current_input_port |
0f2d19dd | 492 | { |
9de87eea | 493 | SCM oinp = scm_fluid_ref (cur_inport_fluid); |
34d19ef6 | 494 | SCM_VALIDATE_OPINPORT (1, port); |
9de87eea | 495 | scm_fluid_set_x (cur_inport_fluid, port); |
0f2d19dd JB |
496 | return oinp; |
497 | } | |
1bbd0b84 | 498 | #undef FUNC_NAME |
0f2d19dd JB |
499 | |
500 | ||
3b3b36dd | 501 | SCM_DEFINE (scm_set_current_output_port, "set-current-output-port", 1, 0, 0, |
e1546b65 MG |
502 | (SCM port), |
503 | "Set the current default output port to @var{port}.") | |
1bbd0b84 | 504 | #define FUNC_NAME s_scm_set_current_output_port |
0f2d19dd | 505 | { |
9de87eea | 506 | SCM ooutp = scm_fluid_ref (cur_outport_fluid); |
78446828 | 507 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 508 | SCM_VALIDATE_OPOUTPORT (1, port); |
9de87eea | 509 | scm_fluid_set_x (cur_outport_fluid, port); |
0f2d19dd JB |
510 | return ooutp; |
511 | } | |
1bbd0b84 | 512 | #undef FUNC_NAME |
0f2d19dd JB |
513 | |
514 | ||
3b3b36dd | 515 | SCM_DEFINE (scm_set_current_error_port, "set-current-error-port", 1, 0, 0, |
e1546b65 MG |
516 | (SCM port), |
517 | "Set the current default error port to @var{port}.") | |
1bbd0b84 | 518 | #define FUNC_NAME s_scm_set_current_error_port |
0f2d19dd | 519 | { |
9de87eea | 520 | SCM oerrp = scm_fluid_ref (cur_errport_fluid); |
78446828 | 521 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 522 | SCM_VALIDATE_OPOUTPORT (1, port); |
9de87eea | 523 | scm_fluid_set_x (cur_errport_fluid, port); |
0f2d19dd JB |
524 | return oerrp; |
525 | } | |
1bbd0b84 | 526 | #undef FUNC_NAME |
0f2d19dd | 527 | |
3972de76 AW |
528 | |
529 | SCM | |
530 | scm_set_current_warning_port (SCM port) | |
531 | { | |
60617d81 MW |
532 | scm_i_pthread_once (¤t_warning_port_once, |
533 | init_current_warning_port_var); | |
534 | return scm_call_1 (scm_variable_ref (current_warning_port_var), port); | |
3972de76 AW |
535 | } |
536 | ||
537 | ||
185e369a | 538 | void |
661ae7ab | 539 | scm_dynwind_current_input_port (SCM port) |
9de87eea | 540 | #define FUNC_NAME NULL |
185e369a | 541 | { |
9de87eea | 542 | SCM_VALIDATE_OPINPORT (1, port); |
661ae7ab | 543 | scm_dynwind_fluid (cur_inport_fluid, port); |
185e369a | 544 | } |
9de87eea | 545 | #undef FUNC_NAME |
185e369a MV |
546 | |
547 | void | |
661ae7ab | 548 | scm_dynwind_current_output_port (SCM port) |
9de87eea | 549 | #define FUNC_NAME NULL |
185e369a | 550 | { |
9de87eea MV |
551 | port = SCM_COERCE_OUTPORT (port); |
552 | SCM_VALIDATE_OPOUTPORT (1, port); | |
661ae7ab | 553 | scm_dynwind_fluid (cur_outport_fluid, port); |
185e369a | 554 | } |
9de87eea | 555 | #undef FUNC_NAME |
185e369a MV |
556 | |
557 | void | |
661ae7ab | 558 | scm_dynwind_current_error_port (SCM port) |
9de87eea MV |
559 | #define FUNC_NAME NULL |
560 | { | |
561 | port = SCM_COERCE_OUTPORT (port); | |
562 | SCM_VALIDATE_OPOUTPORT (1, port); | |
661ae7ab | 563 | scm_dynwind_fluid (cur_errport_fluid, port); |
9de87eea MV |
564 | } |
565 | #undef FUNC_NAME | |
566 | ||
567 | void | |
661ae7ab | 568 | scm_i_dynwind_current_load_port (SCM port) |
185e369a | 569 | { |
661ae7ab | 570 | scm_dynwind_fluid (cur_loadport_fluid, port); |
185e369a MV |
571 | } |
572 | ||
0f2d19dd | 573 | \f |
840ae05d | 574 | /* The port table --- an array of pointers to ports. */ |
0f2d19dd | 575 | |
5dbc6c06 HWN |
576 | /* |
577 | We need a global registry of ports to flush them all at exit, and to | |
05d7f762 | 578 | get all the ports matching a file descriptor. |
5dbc6c06 HWN |
579 | */ |
580 | SCM scm_i_port_weak_hash; | |
0f2d19dd | 581 | |
9de87eea | 582 | scm_i_pthread_mutex_t scm_i_port_table_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER; |
b9ad392e | 583 | |
651a0735 LC |
584 | \f |
585 | /* Port finalization. */ | |
586 | ||
1cc91f1b | 587 | |
6922d92f | 588 | static void finalize_port (void *, void *); |
651a0735 | 589 | |
f4bc4e59 | 590 | /* Register a finalizer for PORT. */ |
651a0735 LC |
591 | static SCM_C_INLINE_KEYWORD void |
592 | register_finalizer_for_port (SCM port) | |
593 | { | |
337edc59 MW |
594 | /* Register a finalizer for PORT so that its |
595 | type's `free' function gets called. */ | |
75ba64d6 | 596 | scm_i_set_finalizer (SCM2PTR (port), finalize_port, NULL); |
651a0735 LC |
597 | } |
598 | ||
599 | /* Finalize the object (a port) pointed to by PTR. */ | |
600 | static void | |
6922d92f | 601 | finalize_port (void *ptr, void *data) |
651a0735 LC |
602 | { |
603 | long port_type; | |
604 | SCM port = PTR2SCM (ptr); | |
605 | ||
606 | if (!SCM_PORTP (port)) | |
607 | abort (); | |
608 | ||
609 | if (SCM_OPENP (port)) | |
610 | { | |
611 | if (SCM_REVEALED (port) > 0) | |
612 | /* Keep "revealed" ports alive and re-register a finalizer. */ | |
613 | register_finalizer_for_port (port); | |
614 | else | |
615 | { | |
616 | port_type = SCM_TC2PTOBNUM (SCM_CELL_TYPE (port)); | |
617 | if (port_type >= scm_numptob) | |
618 | abort (); | |
619 | ||
620 | if (scm_ptobs[port_type].free) | |
621 | /* Yes, I really do mean `.free' rather than `.close'. `.close' | |
622 | is for explicit `close-port' by user. */ | |
623 | scm_ptobs[port_type].free (port); | |
624 | ||
625 | SCM_SETSTREAM (port, 0); | |
626 | SCM_CLR_PORT_OPEN_FLAG (port); | |
651a0735 LC |
627 | |
628 | scm_gc_ports_collected++; | |
629 | } | |
630 | } | |
631 | } | |
632 | ||
633 | ||
634 | ||
635 | \f | |
636 | ||
637 | /* This function is not and should not be thread safe. */ | |
da220f27 HWN |
638 | SCM |
639 | scm_new_port_table_entry (scm_t_bits tag) | |
402788a9 | 640 | #define FUNC_NAME "scm_new_port_table_entry" |
0f2d19dd | 641 | { |
85835e59 HWN |
642 | /* |
643 | We initialize the cell to empty, this is in case scm_gc_calloc | |
644 | triggers GC ; we don't want the GC to scan a half-finished Z. | |
645 | */ | |
646 | ||
67329a9e | 647 | SCM z = scm_cons (SCM_EOL, SCM_EOL); |
e4598559 MW |
648 | scm_t_port *entry = scm_gc_typed_calloc (scm_t_port); |
649 | scm_t_port_internal *pti = scm_gc_typed_calloc (scm_t_port_internal); | |
337edc59 | 650 | const char *encoding; |
5f16b897 | 651 | |
840ae05d | 652 | entry->file_name = SCM_BOOL_F; |
61e452ba | 653 | entry->rw_active = SCM_PORT_NEITHER; |
5dbc6c06 | 654 | entry->port = z; |
9d9c66ba | 655 | |
889975e5 MG |
656 | /* Initialize this port with the thread's current default |
657 | encoding. */ | |
337edc59 | 658 | encoding = scm_i_default_port_encoding (); |
b22e94db | 659 | entry->ilseq_handler = scm_i_default_port_conversion_handler (); |
337edc59 | 660 | entry->encoding = encoding ? scm_gc_strdup (encoding, "port") : NULL; |
96965a6e | 661 | if (encoding && c_strcasecmp (encoding, "UTF-8") == 0) |
337edc59 MW |
662 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8; |
663 | else | |
664 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV; | |
665 | pti->iconv_descriptors = NULL; | |
5f16b897 | 666 | |
cdd3d6c9 MW |
667 | pti->at_stream_start_for_bom_read = 1; |
668 | pti->at_stream_start_for_bom_write = 1; | |
669 | ||
e4598559 MW |
670 | /* XXX These fields are not what they seem. They have been |
671 | repurposed, but cannot safely be renamed in 2.0 without breaking | |
672 | ABI compatibility. This will be cleaned up in 2.2. */ | |
673 | entry->input_cd = pti; /* XXX pointer to the internal port structure */ | |
674 | entry->output_cd = NULL; /* XXX unused */ | |
675 | ||
45c0878b | 676 | pti->pending_eof = 0; |
05d7f762 MW |
677 | pti->alist = SCM_EOL; |
678 | ||
122f24cc LC |
679 | /* Until Guile 2.0.9 included, 'setvbuf' would only work on file |
680 | ports. Now all port types can be supported, but it's not clear | |
681 | that port types out in wild accept having someone else fiddle with | |
682 | their buffer. Thus, conservatively turn it off by default. */ | |
683 | pti->setvbuf = NULL; | |
684 | ||
5dbc6c06 HWN |
685 | SCM_SET_CELL_TYPE (z, tag); |
686 | SCM_SETPTAB_ENTRY (z, entry); | |
840ae05d | 687 | |
05d7f762 | 688 | scm_hashq_set_x (scm_i_port_weak_hash, z, SCM_BOOL_F); |
651a0735 LC |
689 | |
690 | /* For each new port, register a finalizer so that it port type's free | |
691 | function can be invoked eventually. */ | |
692 | register_finalizer_for_port (z); | |
693 | ||
da220f27 | 694 | return z; |
0f2d19dd | 695 | } |
c6c79933 | 696 | #undef FUNC_NAME |
0f2d19dd | 697 | |
67329a9e | 698 | #if SCM_ENABLE_DEPRECATED==1 |
8269ba5b | 699 | scm_t_port * |
67329a9e HWN |
700 | scm_add_to_port_table (SCM port) |
701 | { | |
8269ba5b AW |
702 | SCM z; |
703 | scm_t_port * pt; | |
67329a9e | 704 | |
8269ba5b AW |
705 | scm_c_issue_deprecation_warning ("scm_add_to_port_table is deprecated."); |
706 | ||
707 | scm_i_pthread_mutex_lock (&scm_i_port_table_mutex); | |
708 | z = scm_new_port_table_entry (scm_tc7_port); | |
709 | pt = SCM_PTAB_ENTRY(z); | |
67329a9e | 710 | pt->port = port; |
5dbc6c06 HWN |
711 | SCM_SETCAR (z, SCM_EOL); |
712 | SCM_SETCDR (z, SCM_EOL); | |
85835e59 | 713 | SCM_SETPTAB_ENTRY (port, pt); |
8269ba5b AW |
714 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); |
715 | ||
67329a9e HWN |
716 | return pt; |
717 | } | |
718 | #endif | |
719 | ||
720 | ||
6c951427 | 721 | /* Remove a port from the table and destroy it. */ |
b9ad392e | 722 | |
337edc59 MW |
723 | static void close_iconv_descriptors (scm_t_iconv_descriptors *id); |
724 | ||
8269ba5b | 725 | static void |
5dbc6c06 HWN |
726 | scm_i_remove_port (SCM port) |
727 | #define FUNC_NAME "scm_remove_port" | |
0f2d19dd | 728 | { |
8269ba5b | 729 | scm_t_port *p; |
e4598559 | 730 | scm_t_port_internal *pti; |
682d78d0 | 731 | |
8269ba5b | 732 | scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex); |
682d78d0 | 733 | |
8269ba5b | 734 | p = SCM_PTAB_ENTRY (port); |
e4598559 | 735 | pti = SCM_PORT_GET_INTERNAL (port); |
8269ba5b | 736 | scm_port_non_buffer (p); |
682d78d0 LC |
737 | p->putback_buf = NULL; |
738 | p->putback_buf_size = 0; | |
3e05fc04 | 739 | |
337edc59 | 740 | if (pti->iconv_descriptors) |
3e05fc04 | 741 | { |
337edc59 MW |
742 | close_iconv_descriptors (pti->iconv_descriptors); |
743 | pti->iconv_descriptors = NULL; | |
3e05fc04 AW |
744 | } |
745 | ||
0f2d19dd | 746 | SCM_SETPTAB_ENTRY (port, 0); |
8269ba5b | 747 | |
5dbc6c06 | 748 | scm_hashq_remove_x (scm_i_port_weak_hash, port); |
8269ba5b AW |
749 | |
750 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); | |
0f2d19dd | 751 | } |
db4b4ca6 DH |
752 | #undef FUNC_NAME |
753 | ||
0f2d19dd | 754 | |
b450f070 | 755 | /* Functions for debugging. */ |
5dbc6c06 | 756 | #ifdef GUILE_DEBUG |
3b3b36dd | 757 | SCM_DEFINE (scm_pt_size, "pt-size", 0, 0, 0, |
b450f070 | 758 | (), |
1e6808ea | 759 | "Return the number of ports in the port table. @code{pt-size}\n" |
5352393c | 760 | "is only included in @code{--enable-guile-debug} builds.") |
1bbd0b84 | 761 | #define FUNC_NAME s_scm_pt_size |
0f2d19dd | 762 | { |
5dbc6c06 | 763 | return scm_from_int (SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash)); |
0f2d19dd | 764 | } |
1bbd0b84 | 765 | #undef FUNC_NAME |
0f2d19dd JB |
766 | #endif |
767 | ||
70df8af6 | 768 | void |
92c2555f | 769 | scm_port_non_buffer (scm_t_port *pt) |
70df8af6 GH |
770 | { |
771 | pt->read_pos = pt->read_buf = pt->read_end = &pt->shortbuf; | |
772 | pt->write_buf = pt->write_pos = &pt->shortbuf; | |
773 | pt->read_buf_size = pt->write_buf_size = 1; | |
774 | pt->write_end = pt->write_buf + pt->write_buf_size; | |
775 | } | |
0f2d19dd | 776 | |
d68fee48 JB |
777 | \f |
778 | /* Revealed counts --- an oddity inherited from SCSH. */ | |
779 | ||
8b13c6b3 GH |
780 | /* Find a port in the table and return its revealed count. |
781 | Also used by the garbage collector. | |
0f2d19dd | 782 | */ |
1cc91f1b | 783 | |
0f2d19dd | 784 | int |
a284e297 | 785 | scm_revealed_count (SCM port) |
0f2d19dd JB |
786 | { |
787 | return SCM_REVEALED(port); | |
788 | } | |
789 | ||
790 | ||
791 | ||
792 | /* Return the revealed count for a port. */ | |
793 | ||
3b3b36dd | 794 | SCM_DEFINE (scm_port_revealed, "port-revealed", 1, 0, 0, |
1bbd0b84 | 795 | (SCM port), |
1e6808ea | 796 | "Return the revealed count for @var{port}.") |
1bbd0b84 | 797 | #define FUNC_NAME s_scm_port_revealed |
0f2d19dd | 798 | { |
78446828 | 799 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 800 | SCM_VALIDATE_OPENPORT (1, port); |
e11e83f3 | 801 | return scm_from_int (scm_revealed_count (port)); |
0f2d19dd | 802 | } |
1bbd0b84 | 803 | #undef FUNC_NAME |
0f2d19dd JB |
804 | |
805 | /* Set the revealed count for a port. */ | |
3b3b36dd | 806 | SCM_DEFINE (scm_set_port_revealed_x, "set-port-revealed!", 2, 0, 0, |
1bbd0b84 | 807 | (SCM port, SCM rcount), |
b450f070 | 808 | "Sets the revealed count for a port to a given value.\n" |
b380b885 | 809 | "The return value is unspecified.") |
1bbd0b84 | 810 | #define FUNC_NAME s_scm_set_port_revealed_x |
0f2d19dd | 811 | { |
78446828 | 812 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 813 | SCM_VALIDATE_OPENPORT (1, port); |
a55c2b68 | 814 | SCM_REVEALED (port) = scm_to_int (rcount); |
8b13c6b3 | 815 | return SCM_UNSPECIFIED; |
0f2d19dd | 816 | } |
1bbd0b84 | 817 | #undef FUNC_NAME |
0f2d19dd | 818 | |
d68fee48 JB |
819 | |
820 | \f | |
821 | /* Retrieving a port's mode. */ | |
822 | ||
eadd48de GH |
823 | /* Return the flags that characterize a port based on the mode |
824 | * string used to open a file for that port. | |
825 | * | |
826 | * See PORT FLAGS in scm.h | |
827 | */ | |
828 | ||
3a5fb14d | 829 | static long |
889975e5 | 830 | scm_i_mode_bits_n (SCM modes) |
3a5fb14d MV |
831 | { |
832 | return (SCM_OPN | |
889975e5 MG |
833 | | (scm_i_string_contains_char (modes, 'r') |
834 | || scm_i_string_contains_char (modes, '+') ? SCM_RDNG : 0) | |
835 | | (scm_i_string_contains_char (modes, 'w') | |
836 | || scm_i_string_contains_char (modes, 'a') | |
837 | || scm_i_string_contains_char (modes, '+') ? SCM_WRTNG : 0) | |
838 | | (scm_i_string_contains_char (modes, '0') ? SCM_BUF0 : 0) | |
839 | | (scm_i_string_contains_char (modes, 'l') ? SCM_BUFLINE : 0)); | |
3a5fb14d MV |
840 | } |
841 | ||
eadd48de | 842 | long |
a284e297 | 843 | scm_mode_bits (char *modes) |
eadd48de | 844 | { |
889975e5 | 845 | return scm_i_mode_bits (scm_from_locale_string (modes)); |
eadd48de GH |
846 | } |
847 | ||
d617ee18 MV |
848 | long |
849 | scm_i_mode_bits (SCM modes) | |
850 | { | |
851 | long bits; | |
852 | ||
853 | if (!scm_is_string (modes)) | |
854 | scm_wrong_type_arg_msg (NULL, 0, modes, "string"); | |
855 | ||
889975e5 | 856 | bits = scm_i_mode_bits_n (modes); |
d617ee18 MV |
857 | scm_remember_upto_here_1 (modes); |
858 | return bits; | |
859 | } | |
eadd48de GH |
860 | |
861 | /* Return the mode flags from an open port. | |
862 | * Some modes such as "append" are only used when opening | |
863 | * a file and are not returned here. */ | |
864 | ||
3b3b36dd | 865 | SCM_DEFINE (scm_port_mode, "port-mode", 1, 0, 0, |
1bbd0b84 | 866 | (SCM port), |
1e6808ea MG |
867 | "Return the port modes associated with the open port @var{port}.\n" |
868 | "These will not necessarily be identical to the modes used when\n" | |
869 | "the port was opened, since modes such as \"append\" which are\n" | |
870 | "used only during port creation are not retained.") | |
1bbd0b84 | 871 | #define FUNC_NAME s_scm_port_mode |
eadd48de | 872 | { |
26a3038d | 873 | char modes[4]; |
eadd48de | 874 | modes[0] = '\0'; |
78446828 MV |
875 | |
876 | port = SCM_COERCE_OUTPORT (port); | |
34d19ef6 | 877 | SCM_VALIDATE_OPPORT (1, port); |
f9a64404 DH |
878 | if (SCM_CELL_WORD_0 (port) & SCM_RDNG) { |
879 | if (SCM_CELL_WORD_0 (port) & SCM_WRTNG) | |
eadd48de GH |
880 | strcpy (modes, "r+"); |
881 | else | |
882 | strcpy (modes, "r"); | |
883 | } | |
f9a64404 | 884 | else if (SCM_CELL_WORD_0 (port) & SCM_WRTNG) |
eadd48de | 885 | strcpy (modes, "w"); |
f9a64404 | 886 | if (SCM_CELL_WORD_0 (port) & SCM_BUF0) |
eadd48de | 887 | strcat (modes, "0"); |
3a5fb14d | 888 | return scm_from_locale_string (modes); |
eadd48de | 889 | } |
1bbd0b84 | 890 | #undef FUNC_NAME |
eadd48de GH |
891 | |
892 | ||
d68fee48 JB |
893 | \f |
894 | /* Closing ports. */ | |
895 | ||
0f2d19dd JB |
896 | /* scm_close_port |
897 | * Call the close operation on a port object. | |
eadd48de | 898 | * see also scm_close. |
0f2d19dd | 899 | */ |
3b3b36dd | 900 | SCM_DEFINE (scm_close_port, "close-port", 1, 0, 0, |
1bbd0b84 | 901 | (SCM port), |
1e6808ea MG |
902 | "Close the specified port object. Return @code{#t} if it\n" |
903 | "successfully closes a port or @code{#f} if it was already\n" | |
904 | "closed. An exception may be raised if an error occurs, for\n" | |
905 | "example when flushing buffered output. See also @ref{Ports and\n" | |
906 | "File Descriptors, close}, for a procedure which can close file\n" | |
907 | "descriptors.") | |
1bbd0b84 | 908 | #define FUNC_NAME s_scm_close_port |
0f2d19dd | 909 | { |
1be6b49c | 910 | size_t i; |
eadd48de GH |
911 | int rv; |
912 | ||
78446828 MV |
913 | port = SCM_COERCE_OUTPORT (port); |
914 | ||
7a754ca6 | 915 | SCM_VALIDATE_PORT (1, port); |
0f2d19dd | 916 | if (SCM_CLOSEDP (port)) |
eadd48de | 917 | return SCM_BOOL_F; |
0f2d19dd | 918 | i = SCM_PTOBNUM (port); |
affc96b5 GH |
919 | if (scm_ptobs[i].close) |
920 | rv = (scm_ptobs[i].close) (port); | |
eadd48de GH |
921 | else |
922 | rv = 0; | |
5dbc6c06 | 923 | scm_i_remove_port (port); |
22a52da1 | 924 | SCM_CLR_PORT_OPEN_FLAG (port); |
7888309b | 925 | return scm_from_bool (rv >= 0); |
7a754ca6 MD |
926 | } |
927 | #undef FUNC_NAME | |
928 | ||
929 | SCM_DEFINE (scm_close_input_port, "close-input-port", 1, 0, 0, | |
930 | (SCM port), | |
931 | "Close the specified input port object. The routine has no effect if\n" | |
932 | "the file has already been closed. An exception may be raised if an\n" | |
933 | "error occurs. The value returned is unspecified.\n\n" | |
934 | "See also @ref{Ports and File Descriptors, close}, for a procedure\n" | |
935 | "which can close file descriptors.") | |
936 | #define FUNC_NAME s_scm_close_input_port | |
937 | { | |
938 | SCM_VALIDATE_INPUT_PORT (1, port); | |
939 | scm_close_port (port); | |
940 | return SCM_UNSPECIFIED; | |
941 | } | |
942 | #undef FUNC_NAME | |
943 | ||
944 | SCM_DEFINE (scm_close_output_port, "close-output-port", 1, 0, 0, | |
945 | (SCM port), | |
946 | "Close the specified output port object. The routine has no effect if\n" | |
947 | "the file has already been closed. An exception may be raised if an\n" | |
948 | "error occurs. The value returned is unspecified.\n\n" | |
949 | "See also @ref{Ports and File Descriptors, close}, for a procedure\n" | |
950 | "which can close file descriptors.") | |
951 | #define FUNC_NAME s_scm_close_output_port | |
952 | { | |
953 | port = SCM_COERCE_OUTPORT (port); | |
954 | SCM_VALIDATE_OUTPUT_PORT (1, port); | |
955 | scm_close_port (port); | |
956 | return SCM_UNSPECIFIED; | |
0f2d19dd | 957 | } |
1bbd0b84 | 958 | #undef FUNC_NAME |
0f2d19dd | 959 | |
5dbc6c06 | 960 | static SCM |
b7b4aef9 | 961 | collect_keys (void *unused, SCM key, SCM value, SCM result) |
5dbc6c06 | 962 | { |
b7b4aef9 | 963 | return scm_cons (key, result); |
5dbc6c06 HWN |
964 | } |
965 | ||
c536b4b3 MV |
966 | void |
967 | scm_c_port_for_each (void (*proc)(void *data, SCM p), void *data) | |
c2ca4493 | 968 | { |
fdfe6305 MV |
969 | SCM ports; |
970 | ||
b7b4aef9 AW |
971 | /* Copy out the port table as a list so that we get strong references |
972 | to all the values. */ | |
5dbc6c06 | 973 | scm_i_pthread_mutex_lock (&scm_i_port_table_mutex); |
b7b4aef9 AW |
974 | ports = scm_internal_hash_fold (collect_keys, NULL, |
975 | SCM_EOL, scm_i_port_weak_hash); | |
9de87eea | 976 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); |
3a5fb14d | 977 | |
b7b4aef9 AW |
978 | for (; scm_is_pair (ports); ports = scm_cdr (ports)) |
979 | { | |
980 | SCM p = scm_car (ports); | |
981 | if (SCM_PORTP (p)) | |
982 | proc (data, p); | |
983 | } | |
c536b4b3 | 984 | } |
fdfe6305 | 985 | |
c536b4b3 MV |
986 | SCM_DEFINE (scm_port_for_each, "port-for-each", 1, 0, 0, |
987 | (SCM proc), | |
988 | "Apply @var{proc} to each port in the Guile port table\n" | |
989 | "in turn. The return value is unspecified. More specifically,\n" | |
990 | "@var{proc} is applied exactly once to every port that exists\n" | |
b7e64f8b BT |
991 | "in the system at the time @code{port-for-each} is invoked.\n" |
992 | "Changes to the port table while @code{port-for-each} is running\n" | |
993 | "have no effect as far as @code{port-for-each} is concerned.") | |
c536b4b3 MV |
994 | #define FUNC_NAME s_scm_port_for_each |
995 | { | |
1b3daef0 AW |
996 | SCM ports; |
997 | ||
c536b4b3 MV |
998 | SCM_VALIDATE_PROC (1, proc); |
999 | ||
1b3daef0 AW |
1000 | /* Copy out the port table as a list so that we get strong references |
1001 | to all the values. */ | |
1002 | scm_i_pthread_mutex_lock (&scm_i_port_table_mutex); | |
1003 | ports = scm_internal_hash_fold (collect_keys, NULL, | |
1004 | SCM_EOL, scm_i_port_weak_hash); | |
1005 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); | |
1006 | ||
1007 | for (; scm_is_pair (ports); ports = scm_cdr (ports)) | |
1008 | if (SCM_PORTP (SCM_CAR (ports))) | |
1009 | scm_call_1 (proc, SCM_CAR (ports)); | |
1010 | ||
c2ca4493 GH |
1011 | return SCM_UNSPECIFIED; |
1012 | } | |
1013 | #undef FUNC_NAME | |
1014 | ||
c536b4b3 | 1015 | |
d68fee48 JB |
1016 | \f |
1017 | /* Utter miscellany. Gosh, we should clean this up some time. */ | |
1018 | ||
3b3b36dd | 1019 | SCM_DEFINE (scm_input_port_p, "input-port?", 1, 0, 0, |
1bbd0b84 | 1020 | (SCM x), |
1e6808ea MG |
1021 | "Return @code{#t} if @var{x} is an input port, otherwise return\n" |
1022 | "@code{#f}. Any object satisfying this predicate also satisfies\n" | |
1023 | "@code{port?}.") | |
1bbd0b84 | 1024 | #define FUNC_NAME s_scm_input_port_p |
0f2d19dd | 1025 | { |
7888309b | 1026 | return scm_from_bool (SCM_INPUT_PORT_P (x)); |
0f2d19dd | 1027 | } |
1bbd0b84 | 1028 | #undef FUNC_NAME |
0f2d19dd | 1029 | |
3b3b36dd | 1030 | SCM_DEFINE (scm_output_port_p, "output-port?", 1, 0, 0, |
1bbd0b84 | 1031 | (SCM x), |
1e6808ea MG |
1032 | "Return @code{#t} if @var{x} is an output port, otherwise return\n" |
1033 | "@code{#f}. Any object satisfying this predicate also satisfies\n" | |
1034 | "@code{port?}.") | |
1bbd0b84 | 1035 | #define FUNC_NAME s_scm_output_port_p |
0f2d19dd | 1036 | { |
82893676 | 1037 | x = SCM_COERCE_OUTPORT (x); |
7888309b | 1038 | return scm_from_bool (SCM_OUTPUT_PORT_P (x)); |
0f2d19dd | 1039 | } |
1bbd0b84 | 1040 | #undef FUNC_NAME |
0f2d19dd | 1041 | |
eb5c0a2a GH |
1042 | SCM_DEFINE (scm_port_p, "port?", 1, 0, 0, |
1043 | (SCM x), | |
1e6808ea | 1044 | "Return a boolean indicating whether @var{x} is a port.\n" |
5352393c MG |
1045 | "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n" |
1046 | "@var{x}))}.") | |
eb5c0a2a GH |
1047 | #define FUNC_NAME s_scm_port_p |
1048 | { | |
7888309b | 1049 | return scm_from_bool (SCM_PORTP (x)); |
eb5c0a2a GH |
1050 | } |
1051 | #undef FUNC_NAME | |
1052 | ||
3b3b36dd | 1053 | SCM_DEFINE (scm_port_closed_p, "port-closed?", 1, 0, 0, |
1bbd0b84 | 1054 | (SCM port), |
1e6808ea MG |
1055 | "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n" |
1056 | "open.") | |
1bbd0b84 | 1057 | #define FUNC_NAME s_scm_port_closed_p |
60d0643d | 1058 | { |
34d19ef6 | 1059 | SCM_VALIDATE_PORT (1, port); |
7888309b | 1060 | return scm_from_bool (!SCM_OPPORTP (port)); |
60d0643d | 1061 | } |
1bbd0b84 | 1062 | #undef FUNC_NAME |
0f2d19dd | 1063 | |
3b3b36dd | 1064 | SCM_DEFINE (scm_eof_object_p, "eof-object?", 1, 0, 0, |
1bbd0b84 | 1065 | (SCM x), |
1e6808ea MG |
1066 | "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n" |
1067 | "return @code{#f}.") | |
1bbd0b84 | 1068 | #define FUNC_NAME s_scm_eof_object_p |
0f2d19dd | 1069 | { |
7888309b | 1070 | return scm_from_bool(SCM_EOF_OBJECT_P (x)); |
0f2d19dd | 1071 | } |
1bbd0b84 | 1072 | #undef FUNC_NAME |
0f2d19dd | 1073 | |
3b3b36dd | 1074 | SCM_DEFINE (scm_force_output, "force-output", 0, 1, 0, |
1bbd0b84 | 1075 | (SCM port), |
b380b885 | 1076 | "Flush the specified output port, or the current output port if @var{port}\n" |
9401323e | 1077 | "is omitted. The current output buffer contents are passed to the\n" |
b380b885 MD |
1078 | "underlying port implementation (e.g., in the case of fports, the\n" |
1079 | "data will be written to the file and the output buffer will be cleared.)\n" | |
1080 | "It has no effect on an unbuffered port.\n\n" | |
1081 | "The return value is unspecified.") | |
1bbd0b84 | 1082 | #define FUNC_NAME s_scm_force_output |
0f2d19dd JB |
1083 | { |
1084 | if (SCM_UNBNDP (port)) | |
9de87eea | 1085 | port = scm_current_output_port (); |
0f2d19dd | 1086 | else |
78446828 MV |
1087 | { |
1088 | port = SCM_COERCE_OUTPORT (port); | |
34d19ef6 | 1089 | SCM_VALIDATE_OPOUTPORT (1, port); |
78446828 | 1090 | } |
affc96b5 | 1091 | scm_flush (port); |
ee149d03 | 1092 | return SCM_UNSPECIFIED; |
0f2d19dd | 1093 | } |
1bbd0b84 | 1094 | #undef FUNC_NAME |
0f2d19dd | 1095 | |
5dbc6c06 HWN |
1096 | |
1097 | static void | |
61d3568b | 1098 | flush_output_port (void *closure, SCM port) |
5dbc6c06 | 1099 | { |
5dbc6c06 HWN |
1100 | if (SCM_OPOUTPORTP (port)) |
1101 | scm_flush (port); | |
1102 | } | |
1103 | ||
a1ec6916 | 1104 | SCM_DEFINE (scm_flush_all_ports, "flush-all-ports", 0, 0, 0, |
1bbd0b84 | 1105 | (), |
b380b885 MD |
1106 | "Equivalent to calling @code{force-output} on\n" |
1107 | "all open output ports. The return value is unspecified.") | |
1bbd0b84 | 1108 | #define FUNC_NAME s_scm_flush_all_ports |
89ea5b7c | 1109 | { |
5dbc6c06 | 1110 | scm_c_port_for_each (&flush_output_port, NULL); |
89ea5b7c GH |
1111 | return SCM_UNSPECIFIED; |
1112 | } | |
1bbd0b84 | 1113 | #undef FUNC_NAME |
0f2d19dd | 1114 | |
3b3b36dd | 1115 | SCM_DEFINE (scm_read_char, "read-char", 0, 1, 0, |
1bbd0b84 | 1116 | (SCM port), |
1e6808ea MG |
1117 | "Return the next character available from @var{port}, updating\n" |
1118 | "@var{port} to point to the following character. If no more\n" | |
c62da8f8 LC |
1119 | "characters are available, the end-of-file object is returned.\n" |
1120 | "\n" | |
1121 | "When @var{port}'s data cannot be decoded according to its\n" | |
1122 | "character encoding, a @code{decoding-error} is raised and\n" | |
1123 | "@var{port} points past the erroneous byte sequence.\n") | |
1bbd0b84 | 1124 | #define FUNC_NAME s_scm_read_char |
0f2d19dd | 1125 | { |
889975e5 | 1126 | scm_t_wchar c; |
0f2d19dd | 1127 | if (SCM_UNBNDP (port)) |
9de87eea | 1128 | port = scm_current_input_port (); |
34d19ef6 | 1129 | SCM_VALIDATE_OPINPORT (1, port); |
b7f3516f | 1130 | c = scm_getc (port); |
0f2d19dd JB |
1131 | if (EOF == c) |
1132 | return SCM_EOF_VAL; | |
7866a09b | 1133 | return SCM_MAKE_CHAR (c); |
0f2d19dd | 1134 | } |
1bbd0b84 | 1135 | #undef FUNC_NAME |
0f2d19dd | 1136 | |
f4bc4e59 LC |
1137 | /* Update the line and column number of PORT after consumption of C. */ |
1138 | static inline void | |
1139 | update_port_lf (scm_t_wchar c, SCM port) | |
1140 | { | |
1141 | switch (c) | |
1142 | { | |
1143 | case '\a': | |
7b292a9d | 1144 | case EOF: |
f4bc4e59 LC |
1145 | break; |
1146 | case '\b': | |
1147 | SCM_DECCOL (port); | |
1148 | break; | |
1149 | case '\n': | |
1150 | SCM_INCLINE (port); | |
1151 | break; | |
1152 | case '\r': | |
1153 | SCM_ZEROCOL (port); | |
1154 | break; | |
1155 | case '\t': | |
1156 | SCM_TABCOL (port); | |
1157 | break; | |
1158 | default: | |
1159 | SCM_INCCOL (port); | |
1160 | break; | |
1161 | } | |
1162 | } | |
1163 | ||
889975e5 MG |
1164 | #define SCM_MBCHAR_BUF_SIZE (4) |
1165 | ||
f4bc4e59 LC |
1166 | /* Convert the SIZE-byte UTF-8 sequence in UTF8_BUF to a codepoint. |
1167 | UTF8_BUF is assumed to contain a valid UTF-8 sequence. */ | |
1168 | static scm_t_wchar | |
1169 | utf8_to_codepoint (const scm_t_uint8 *utf8_buf, size_t size) | |
1170 | { | |
1171 | scm_t_wchar codepoint; | |
1172 | ||
1173 | if (utf8_buf[0] <= 0x7f) | |
1174 | { | |
1175 | assert (size == 1); | |
1176 | codepoint = utf8_buf[0]; | |
1177 | } | |
1178 | else if ((utf8_buf[0] & 0xe0) == 0xc0) | |
1179 | { | |
1180 | assert (size == 2); | |
1181 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x1f) << 6UL | |
1182 | | (utf8_buf[1] & 0x3f); | |
1183 | } | |
1184 | else if ((utf8_buf[0] & 0xf0) == 0xe0) | |
1185 | { | |
1186 | assert (size == 3); | |
1187 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x0f) << 12UL | |
1188 | | ((scm_t_wchar) utf8_buf[1] & 0x3f) << 6UL | |
1189 | | (utf8_buf[2] & 0x3f); | |
1190 | } | |
1191 | else | |
1192 | { | |
1193 | assert (size == 4); | |
1194 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x07) << 18UL | |
1195 | | ((scm_t_wchar) utf8_buf[1] & 0x3f) << 12UL | |
1196 | | ((scm_t_wchar) utf8_buf[2] & 0x3f) << 6UL | |
1197 | | (utf8_buf[3] & 0x3f); | |
1198 | } | |
1199 | ||
1200 | return codepoint; | |
1201 | } | |
1202 | ||
7b292a9d LC |
1203 | /* Read a UTF-8 sequence from PORT. On success, return 0 and set |
1204 | *CODEPOINT to the codepoint that was read, fill BUF with its UTF-8 | |
1205 | representation, and set *LEN to the length in bytes. Return | |
1206 | `EILSEQ' on error. */ | |
c62da8f8 | 1207 | static int |
7b292a9d LC |
1208 | get_utf8_codepoint (SCM port, scm_t_wchar *codepoint, |
1209 | scm_t_uint8 buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
889975e5 | 1210 | { |
7b292a9d LC |
1211 | #define ASSERT_NOT_EOF(b) \ |
1212 | if (SCM_UNLIKELY ((b) == EOF)) \ | |
1213 | goto invalid_seq | |
7be1705d LC |
1214 | #define CONSUME_PEEKED_BYTE() \ |
1215 | pt->read_pos++ | |
7b292a9d LC |
1216 | |
1217 | int byte; | |
7be1705d | 1218 | scm_t_port *pt; |
7b292a9d LC |
1219 | |
1220 | *len = 0; | |
7be1705d | 1221 | pt = SCM_PTAB_ENTRY (port); |
7b292a9d LC |
1222 | |
1223 | byte = scm_get_byte_or_eof (port); | |
1224 | if (byte == EOF) | |
1225 | { | |
1226 | *codepoint = EOF; | |
1227 | return 0; | |
1228 | } | |
1229 | ||
1230 | buf[0] = (scm_t_uint8) byte; | |
1231 | *len = 1; | |
1232 | ||
1233 | if (buf[0] <= 0x7f) | |
1234 | /* 1-byte form. */ | |
1235 | *codepoint = buf[0]; | |
1236 | else if (buf[0] >= 0xc2 && buf[0] <= 0xdf) | |
1237 | { | |
1238 | /* 2-byte form. */ | |
7be1705d | 1239 | byte = scm_peek_byte_or_eof (port); |
7b292a9d LC |
1240 | ASSERT_NOT_EOF (byte); |
1241 | ||
7b292a9d LC |
1242 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1243 | goto invalid_seq; | |
1244 | ||
7be1705d LC |
1245 | CONSUME_PEEKED_BYTE (); |
1246 | buf[1] = (scm_t_uint8) byte; | |
1247 | *len = 2; | |
1248 | ||
7b292a9d LC |
1249 | *codepoint = ((scm_t_wchar) buf[0] & 0x1f) << 6UL |
1250 | | (buf[1] & 0x3f); | |
1251 | } | |
1252 | else if ((buf[0] & 0xf0) == 0xe0) | |
1253 | { | |
1254 | /* 3-byte form. */ | |
7be1705d LC |
1255 | byte = scm_peek_byte_or_eof (port); |
1256 | ASSERT_NOT_EOF (byte); | |
7b292a9d LC |
1257 | |
1258 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80 | |
1259 | || (buf[0] == 0xe0 && byte < 0xa0) | |
1260 | || (buf[0] == 0xed && byte > 0x9f))) | |
7be1705d | 1261 | goto invalid_seq; |
7b292a9d | 1262 | |
7be1705d LC |
1263 | CONSUME_PEEKED_BYTE (); |
1264 | buf[1] = (scm_t_uint8) byte; | |
1265 | *len = 2; | |
7b292a9d | 1266 | |
7be1705d | 1267 | byte = scm_peek_byte_or_eof (port); |
7b292a9d LC |
1268 | ASSERT_NOT_EOF (byte); |
1269 | ||
7b292a9d LC |
1270 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1271 | goto invalid_seq; | |
1272 | ||
7be1705d LC |
1273 | CONSUME_PEEKED_BYTE (); |
1274 | buf[2] = (scm_t_uint8) byte; | |
1275 | *len = 3; | |
1276 | ||
7b292a9d LC |
1277 | *codepoint = ((scm_t_wchar) buf[0] & 0x0f) << 12UL |
1278 | | ((scm_t_wchar) buf[1] & 0x3f) << 6UL | |
1279 | | (buf[2] & 0x3f); | |
1280 | } | |
1281 | else if (buf[0] >= 0xf0 && buf[0] <= 0xf4) | |
1282 | { | |
1283 | /* 4-byte form. */ | |
7be1705d | 1284 | byte = scm_peek_byte_or_eof (port); |
7b292a9d LC |
1285 | ASSERT_NOT_EOF (byte); |
1286 | ||
7b292a9d LC |
1287 | if (SCM_UNLIKELY (((byte & 0xc0) != 0x80) |
1288 | || (buf[0] == 0xf0 && byte < 0x90) | |
1289 | || (buf[0] == 0xf4 && byte > 0x8f))) | |
7be1705d | 1290 | goto invalid_seq; |
7b292a9d | 1291 | |
7be1705d LC |
1292 | CONSUME_PEEKED_BYTE (); |
1293 | buf[1] = (scm_t_uint8) byte; | |
1294 | *len = 2; | |
1295 | ||
1296 | byte = scm_peek_byte_or_eof (port); | |
7b292a9d LC |
1297 | ASSERT_NOT_EOF (byte); |
1298 | ||
7be1705d LC |
1299 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1300 | goto invalid_seq; | |
1301 | ||
1302 | CONSUME_PEEKED_BYTE (); | |
7b292a9d LC |
1303 | buf[2] = (scm_t_uint8) byte; |
1304 | *len = 3; | |
1305 | ||
7be1705d | 1306 | byte = scm_peek_byte_or_eof (port); |
7b292a9d LC |
1307 | ASSERT_NOT_EOF (byte); |
1308 | ||
7b292a9d LC |
1309 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1310 | goto invalid_seq; | |
1311 | ||
7be1705d LC |
1312 | CONSUME_PEEKED_BYTE (); |
1313 | buf[3] = (scm_t_uint8) byte; | |
1314 | *len = 4; | |
1315 | ||
7b292a9d LC |
1316 | *codepoint = ((scm_t_wchar) buf[0] & 0x07) << 18UL |
1317 | | ((scm_t_wchar) buf[1] & 0x3f) << 12UL | |
1318 | | ((scm_t_wchar) buf[2] & 0x3f) << 6UL | |
1319 | | (buf[3] & 0x3f); | |
1320 | } | |
1321 | else | |
1322 | goto invalid_seq; | |
1323 | ||
1324 | return 0; | |
1325 | ||
1326 | invalid_seq: | |
7be1705d LC |
1327 | /* Here we could choose the consume the faulty byte when it's not a |
1328 | valid starting byte, but it's not a requirement. What Section 3.9 | |
1329 | of Unicode 6.0.0 mandates, though, is to not consume a byte that | |
1330 | would otherwise be a valid starting byte. */ | |
1331 | ||
7b292a9d LC |
1332 | return EILSEQ; |
1333 | ||
7be1705d | 1334 | #undef CONSUME_PEEKED_BYTE |
7b292a9d LC |
1335 | #undef ASSERT_NOT_EOF |
1336 | } | |
1337 | ||
1338 | /* Likewise, read a byte sequence from PORT, passing it through its | |
1339 | input conversion descriptor. */ | |
1340 | static int | |
1341 | get_iconv_codepoint (SCM port, scm_t_wchar *codepoint, | |
1342 | char buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
1343 | { | |
cdd3d6c9 | 1344 | scm_t_iconv_descriptors *id; |
f4bc4e59 | 1345 | scm_t_uint8 utf8_buf[SCM_MBCHAR_BUF_SIZE]; |
1ee237d9 | 1346 | size_t input_size = 0; |
889975e5 | 1347 | |
cdd3d6c9 MW |
1348 | id = scm_i_port_iconv_descriptors (port, SCM_PORT_READ); |
1349 | ||
1ee237d9 | 1350 | for (;;) |
63479112 | 1351 | { |
1ee237d9 MW |
1352 | int byte_read; |
1353 | char *input, *output; | |
f4bc4e59 | 1354 | size_t input_left, output_left, done; |
889975e5 | 1355 | |
f4bc4e59 | 1356 | byte_read = scm_get_byte_or_eof (port); |
1ee237d9 | 1357 | if (SCM_UNLIKELY (byte_read == EOF)) |
889975e5 | 1358 | { |
1ee237d9 MW |
1359 | if (SCM_LIKELY (input_size == 0)) |
1360 | { | |
1361 | *codepoint = (scm_t_wchar) EOF; | |
1362 | *len = input_size; | |
1363 | return 0; | |
1364 | } | |
1365 | else | |
45c0878b MW |
1366 | { |
1367 | /* EOF found in the middle of a multibyte character. */ | |
1368 | scm_i_set_pending_eof (port); | |
1369 | return EILSEQ; | |
1370 | } | |
889975e5 MG |
1371 | } |
1372 | ||
1ee237d9 | 1373 | buf[input_size++] = byte_read; |
889975e5 | 1374 | |
f4bc4e59 | 1375 | input = buf; |
1ee237d9 MW |
1376 | input_left = input_size; |
1377 | output = (char *) utf8_buf; | |
f4bc4e59 | 1378 | output_left = sizeof (utf8_buf); |
889975e5 | 1379 | |
337edc59 | 1380 | done = iconv (id->input_cd, &input, &input_left, &output, &output_left); |
1ee237d9 | 1381 | |
f4bc4e59 | 1382 | if (done == (size_t) -1) |
889975e5 | 1383 | { |
1ee237d9 MW |
1384 | int err = errno; |
1385 | if (SCM_LIKELY (err == EINVAL)) | |
1386 | /* The input byte sequence did not form a complete | |
1387 | character. Read another byte and try again. */ | |
1388 | continue; | |
1389 | else | |
1390 | return err; | |
889975e5 | 1391 | } |
f4bc4e59 | 1392 | else |
1ee237d9 MW |
1393 | { |
1394 | size_t output_size = sizeof (utf8_buf) - output_left; | |
1395 | if (SCM_LIKELY (output_size > 0)) | |
1396 | { | |
1397 | /* iconv generated output. Convert the UTF8_BUF sequence | |
1398 | to a Unicode code point. */ | |
1399 | *codepoint = utf8_to_codepoint (utf8_buf, output_size); | |
1400 | *len = input_size; | |
1401 | return 0; | |
1402 | } | |
1403 | else | |
1404 | { | |
1405 | /* iconv consumed some bytes without producing any output. | |
1406 | Most likely this means that a Unicode byte-order mark | |
1407 | (BOM) was consumed, which should not be included in the | |
1408 | returned buf. Shift any remaining bytes to the beginning | |
1409 | of buf, and continue the loop. */ | |
1410 | memmove (buf, input, input_left); | |
1411 | input_size = input_left; | |
1412 | continue; | |
1413 | } | |
1414 | } | |
7b292a9d | 1415 | } |
7b292a9d | 1416 | } |
c62da8f8 | 1417 | |
7b292a9d LC |
1418 | /* Read a codepoint from PORT and return it in *CODEPOINT. Fill BUF |
1419 | with the byte representation of the codepoint in PORT's encoding, and | |
1420 | set *LEN to the length in bytes of that representation. Return 0 on | |
1421 | success and an errno value on error. */ | |
1422 | static int | |
1423 | get_codepoint (SCM port, scm_t_wchar *codepoint, | |
1424 | char buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
1425 | { | |
1426 | int err; | |
1427 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
e4598559 | 1428 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
7b292a9d | 1429 | |
337edc59 | 1430 | if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8) |
7b292a9d | 1431 | err = get_utf8_codepoint (port, codepoint, (scm_t_uint8 *) buf, len); |
cc540d0b | 1432 | else |
7b292a9d LC |
1433 | err = get_iconv_codepoint (port, codepoint, buf, len); |
1434 | ||
1435 | if (SCM_LIKELY (err == 0)) | |
cdd3d6c9 MW |
1436 | { |
1437 | if (SCM_UNLIKELY (pti->at_stream_start_for_bom_read)) | |
1438 | { | |
1439 | /* Record that we're no longer at stream start. */ | |
1440 | pti->at_stream_start_for_bom_read = 0; | |
1441 | if (pt->rw_random) | |
1442 | pti->at_stream_start_for_bom_write = 0; | |
1443 | ||
1444 | /* If we just read a BOM in an encoding that recognizes them, | |
1445 | then silently consume it and read another code point. */ | |
1446 | if (SCM_UNLIKELY | |
1447 | (*codepoint == SCM_UNICODE_BOM | |
1448 | && (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8 | |
96965a6e MW |
1449 | || c_strcasecmp (pt->encoding, "UTF-16") == 0 |
1450 | || c_strcasecmp (pt->encoding, "UTF-32") == 0))) | |
cdd3d6c9 MW |
1451 | return get_codepoint (port, codepoint, buf, len); |
1452 | } | |
1453 | update_port_lf (*codepoint, port); | |
1454 | } | |
7b292a9d | 1455 | else if (pt->ilseq_handler == SCM_ICONVEH_QUESTION_MARK) |
a42d7971 | 1456 | { |
7b292a9d LC |
1457 | *codepoint = '?'; |
1458 | err = 0; | |
a42d7971 LC |
1459 | update_port_lf (*codepoint, port); |
1460 | } | |
889975e5 | 1461 | |
c62da8f8 | 1462 | return err; |
889975e5 MG |
1463 | } |
1464 | ||
fd5eec2b LC |
1465 | /* Read a codepoint from PORT and return it. */ |
1466 | scm_t_wchar | |
1467 | scm_getc (SCM port) | |
c62da8f8 | 1468 | #define FUNC_NAME "scm_getc" |
fd5eec2b | 1469 | { |
c62da8f8 | 1470 | int err; |
fd5eec2b | 1471 | size_t len; |
c62da8f8 | 1472 | scm_t_wchar codepoint; |
fd5eec2b LC |
1473 | char buf[SCM_MBCHAR_BUF_SIZE]; |
1474 | ||
c62da8f8 LC |
1475 | err = get_codepoint (port, &codepoint, buf, &len); |
1476 | if (SCM_UNLIKELY (err != 0)) | |
1477 | /* At this point PORT should point past the invalid encoding, as per | |
1478 | R6RS-lib Section 8.2.4. */ | |
1479 | scm_decoding_error (FUNC_NAME, err, "input decoding error", port); | |
1480 | ||
1481 | return codepoint; | |
fd5eec2b | 1482 | } |
c62da8f8 | 1483 | #undef FUNC_NAME |
889975e5 | 1484 | |
5c070ca7 | 1485 | /* this should only be called when the read buffer is empty. it |
affc96b5 | 1486 | tries to refill the read buffer. it returns the first char from |
5c070ca7 | 1487 | the port, which is either EOF or *(pt->read_pos). */ |
187fa0b9 MW |
1488 | static int |
1489 | scm_i_fill_input (SCM port) | |
6c951427 | 1490 | { |
92c2555f | 1491 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
45c0878b | 1492 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
283a1a0e | 1493 | |
b5cb4464 NJ |
1494 | assert (pt->read_pos == pt->read_end); |
1495 | ||
45c0878b MW |
1496 | if (pti->pending_eof) |
1497 | { | |
1498 | pti->pending_eof = 0; | |
1499 | return EOF; | |
1500 | } | |
1501 | ||
6c951427 GH |
1502 | if (pt->read_buf == pt->putback_buf) |
1503 | { | |
1504 | /* finished reading put-back chars. */ | |
1505 | pt->read_buf = pt->saved_read_buf; | |
1506 | pt->read_pos = pt->saved_read_pos; | |
1507 | pt->read_end = pt->saved_read_end; | |
1508 | pt->read_buf_size = pt->saved_read_buf_size; | |
1509 | if (pt->read_pos < pt->read_end) | |
5c070ca7 | 1510 | return *(pt->read_pos); |
6c951427 | 1511 | } |
affc96b5 | 1512 | return scm_ptobs[SCM_PTOBNUM (port)].fill_input (port); |
6c951427 GH |
1513 | } |
1514 | ||
187fa0b9 MW |
1515 | int |
1516 | scm_fill_input (SCM port) | |
1517 | { | |
1518 | return scm_i_fill_input (port); | |
1519 | } | |
1520 | ||
8a2b5965 MW |
1521 | /* Slow-path fallback for 'scm_get_byte_or_eof' in inline.h */ |
1522 | int | |
e00793d7 | 1523 | scm_slow_get_byte_or_eof (SCM port) |
8a2b5965 MW |
1524 | { |
1525 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
1526 | ||
1527 | if (pt->rw_active == SCM_PORT_WRITE) | |
1528 | scm_flush (port); | |
1529 | ||
1530 | if (pt->rw_random) | |
1531 | pt->rw_active = SCM_PORT_READ; | |
1532 | ||
1533 | if (pt->read_pos >= pt->read_end) | |
1534 | { | |
1535 | if (SCM_UNLIKELY (scm_i_fill_input (port) == EOF)) | |
1536 | return EOF; | |
1537 | } | |
1538 | ||
1539 | return *pt->read_pos++; | |
1540 | } | |
1541 | ||
1542 | /* Slow-path fallback for 'scm_peek_byte_or_eof' in inline.h */ | |
1543 | int | |
e00793d7 | 1544 | scm_slow_peek_byte_or_eof (SCM port) |
8a2b5965 MW |
1545 | { |
1546 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
1547 | ||
1548 | if (pt->rw_active == SCM_PORT_WRITE) | |
1549 | scm_flush (port); | |
1550 | ||
1551 | if (pt->rw_random) | |
1552 | pt->rw_active = SCM_PORT_READ; | |
1553 | ||
1554 | if (pt->read_pos >= pt->read_end) | |
1555 | { | |
1556 | if (SCM_UNLIKELY (scm_i_fill_input (port) == EOF)) | |
45c0878b MW |
1557 | { |
1558 | scm_i_set_pending_eof (port); | |
1559 | return EOF; | |
1560 | } | |
8a2b5965 MW |
1561 | } |
1562 | ||
1563 | return *pt->read_pos; | |
1564 | } | |
1565 | ||
3cb988bd | 1566 | |
6fe692e9 MD |
1567 | /* scm_lfwrite |
1568 | * | |
53802ea8 MV |
1569 | * This function differs from scm_c_write; it updates port line and |
1570 | * column. */ | |
6fe692e9 | 1571 | |
9c44cd45 | 1572 | void |
1be6b49c | 1573 | scm_lfwrite (const char *ptr, size_t size, SCM port) |
ee149d03 | 1574 | { |
92c2555f MV |
1575 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
1576 | scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)]; | |
3e2043c4 | 1577 | |
840ae05d | 1578 | if (pt->rw_active == SCM_PORT_READ) |
affc96b5 | 1579 | scm_end_input (port); |
283a1a0e | 1580 | |
31703ab8 | 1581 | ptob->write (port, ptr, size); |
840ae05d | 1582 | |
9c44cd45 MG |
1583 | for (; size; ptr++, size--) |
1584 | update_port_lf ((scm_t_wchar) (unsigned char) *ptr, port); | |
1585 | ||
1586 | if (pt->rw_random) | |
1587 | pt->rw_active = SCM_PORT_WRITE; | |
1588 | } | |
1589 | ||
4325620f | 1590 | /* Write STR to PORT from START inclusive to END exclusive. */ |
9c44cd45 MG |
1591 | void |
1592 | scm_lfwrite_substr (SCM str, size_t start, size_t end, SCM port) | |
1593 | { | |
9c44cd45 | 1594 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
9c44cd45 MG |
1595 | |
1596 | if (pt->rw_active == SCM_PORT_READ) | |
1597 | scm_end_input (port); | |
1598 | ||
4325620f LC |
1599 | if (end == (size_t) -1) |
1600 | end = scm_i_string_length (str); | |
9c44cd45 | 1601 | |
b908768a | 1602 | scm_i_display_substring (str, start, end, port); |
53802ea8 | 1603 | |
840ae05d JB |
1604 | if (pt->rw_random) |
1605 | pt->rw_active = SCM_PORT_WRITE; | |
ee149d03 | 1606 | } |
3cb988bd | 1607 | |
6fe692e9 MD |
1608 | /* scm_c_read |
1609 | * | |
1610 | * Used by an application to read arbitrary number of bytes from an | |
1611 | * SCM port. Same semantics as libc read, except that scm_c_read only | |
1612 | * returns less than SIZE bytes if at end-of-file. | |
1613 | * | |
1614 | * Warning: Doesn't update port line and column counts! */ | |
1615 | ||
b5cb4464 NJ |
1616 | /* This structure, and the following swap_buffer function, are used |
1617 | for temporarily swapping a port's own read buffer, and the buffer | |
1618 | that the caller of scm_c_read provides. */ | |
1619 | struct port_and_swap_buffer | |
1620 | { | |
1621 | scm_t_port *pt; | |
1622 | unsigned char *buffer; | |
1623 | size_t size; | |
1624 | }; | |
1625 | ||
1626 | static void | |
1627 | swap_buffer (void *data) | |
1628 | { | |
1629 | struct port_and_swap_buffer *psb = (struct port_and_swap_buffer *) data; | |
1630 | unsigned char *old_buf = psb->pt->read_buf; | |
1631 | size_t old_size = psb->pt->read_buf_size; | |
1632 | ||
1633 | /* Make the port use (buffer, size) from the struct. */ | |
1634 | psb->pt->read_pos = psb->pt->read_buf = psb->pt->read_end = psb->buffer; | |
1635 | psb->pt->read_buf_size = psb->size; | |
1636 | ||
1637 | /* Save the port's old (buffer, size) in the struct. */ | |
1638 | psb->buffer = old_buf; | |
1639 | psb->size = old_size; | |
1640 | } | |
1641 | ||
1be6b49c ML |
1642 | size_t |
1643 | scm_c_read (SCM port, void *buffer, size_t size) | |
693758d5 | 1644 | #define FUNC_NAME "scm_c_read" |
6fe692e9 | 1645 | { |
693758d5 | 1646 | scm_t_port *pt; |
1be6b49c | 1647 | size_t n_read = 0, n_available; |
b5cb4464 | 1648 | struct port_and_swap_buffer psb; |
6fe692e9 | 1649 | |
693758d5 LC |
1650 | SCM_VALIDATE_OPINPORT (1, port); |
1651 | ||
1652 | pt = SCM_PTAB_ENTRY (port); | |
6fe692e9 MD |
1653 | if (pt->rw_active == SCM_PORT_WRITE) |
1654 | scm_ptobs[SCM_PTOBNUM (port)].flush (port); | |
1655 | ||
1656 | if (pt->rw_random) | |
1657 | pt->rw_active = SCM_PORT_READ; | |
1658 | ||
b5cb4464 NJ |
1659 | /* Take bytes first from the port's read buffer. */ |
1660 | if (pt->read_pos < pt->read_end) | |
6fe692e9 | 1661 | { |
b5cb4464 | 1662 | n_available = min (size, pt->read_end - pt->read_pos); |
6fe692e9 | 1663 | memcpy (buffer, pt->read_pos, n_available); |
910d1e40 | 1664 | buffer = (char *) buffer + n_available; |
6fe692e9 MD |
1665 | pt->read_pos += n_available; |
1666 | n_read += n_available; | |
6fe692e9 | 1667 | size -= n_available; |
6fe692e9 MD |
1668 | } |
1669 | ||
b5cb4464 NJ |
1670 | /* Avoid the scm_dynwind_* costs if we now have enough data. */ |
1671 | if (size == 0) | |
1672 | return n_read; | |
1673 | ||
187fa0b9 MW |
1674 | /* Now we will call scm_i_fill_input repeatedly until we have read the |
1675 | requested number of bytes. (Note that a single scm_i_fill_input | |
b5cb4464 | 1676 | call does not guarantee to fill the whole of the port's read |
6d227556 | 1677 | buffer.) */ |
c61be450 LC |
1678 | if (pt->read_buf_size <= 1 && |
1679 | (pt->encoding == NULL | |
1680 | || c_strcasecmp (pt->encoding, "ISO-8859-1") == 0)) | |
b5cb4464 | 1681 | { |
6d227556 NJ |
1682 | /* The port that we are reading from is unbuffered - i.e. does |
1683 | not have its own persistent buffer - but we have a buffer, | |
1684 | provided by our caller, that is the right size for the data | |
187fa0b9 | 1685 | that is wanted. For the following scm_i_fill_input calls, |
6d227556 NJ |
1686 | therefore, we use the buffer in hand as the port's read |
1687 | buffer. | |
1688 | ||
1689 | We need to make sure that the port's normal (1 byte) buffer | |
187fa0b9 | 1690 | is reinstated in case one of the scm_i_fill_input () calls |
6d227556 | 1691 | throws an exception; we use the scm_dynwind_* API to achieve |
75192345 MG |
1692 | that. |
1693 | ||
1694 | A consequence of this optimization is that the fill_input | |
1695 | functions can't unget characters. That'll push data to the | |
1696 | pushback buffer instead of this psb buffer. */ | |
1697 | #if SCM_DEBUG == 1 | |
1698 | unsigned char *pback = pt->putback_buf; | |
1699 | #endif | |
6d227556 NJ |
1700 | psb.pt = pt; |
1701 | psb.buffer = buffer; | |
1702 | psb.size = size; | |
1703 | scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE); | |
1704 | scm_dynwind_rewind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY); | |
1705 | scm_dynwind_unwind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY); | |
1706 | ||
187fa0b9 | 1707 | /* Call scm_i_fill_input until we have all the bytes that we need, |
6d227556 | 1708 | or we hit EOF. */ |
187fa0b9 | 1709 | while (pt->read_buf_size && (scm_i_fill_input (port) != EOF)) |
6d227556 NJ |
1710 | { |
1711 | pt->read_buf_size -= (pt->read_end - pt->read_pos); | |
1712 | pt->read_pos = pt->read_buf = pt->read_end; | |
1713 | } | |
75192345 MG |
1714 | #if SCM_DEBUG == 1 |
1715 | if (pback != pt->putback_buf | |
1716 | || pt->read_buf - (unsigned char *) buffer < 0) | |
1717 | scm_misc_error (FUNC_NAME, | |
1718 | "scm_c_read must not call a fill function that pushes " | |
1719 | "back characters onto an unbuffered port", SCM_EOL); | |
1720 | #endif | |
6d227556 | 1721 | n_read += pt->read_buf - (unsigned char *) buffer; |
75192345 | 1722 | |
6d227556 NJ |
1723 | /* Reinstate the port's normal buffer. */ |
1724 | scm_dynwind_end (); | |
1725 | } | |
1726 | else | |
1727 | { | |
1728 | /* The port has its own buffer. It is important that we use it, | |
1729 | even if it happens to be smaller than our caller's buffer, so | |
1730 | that a custom port implementation's entry points (in | |
1731 | particular, fill_input) can rely on the buffer always being | |
1732 | the same as they first set up. */ | |
187fa0b9 | 1733 | while (size && (scm_i_fill_input (port) != EOF)) |
6d227556 NJ |
1734 | { |
1735 | n_available = min (size, pt->read_end - pt->read_pos); | |
1736 | memcpy (buffer, pt->read_pos, n_available); | |
1737 | buffer = (char *) buffer + n_available; | |
1738 | pt->read_pos += n_available; | |
1739 | n_read += n_available; | |
1740 | size -= n_available; | |
1741 | } | |
1742 | } | |
6fe692e9 | 1743 | |
b5cb4464 | 1744 | return n_read; |
6fe692e9 | 1745 | } |
693758d5 | 1746 | #undef FUNC_NAME |
6fe692e9 MD |
1747 | |
1748 | /* scm_c_write | |
1749 | * | |
1750 | * Used by an application to write arbitrary number of bytes to an SCM | |
1751 | * port. Similar semantics as libc write. However, unlike libc | |
1752 | * write, scm_c_write writes the requested number of bytes and has no | |
1753 | * return value. | |
1754 | * | |
1755 | * Warning: Doesn't update port line and column counts! | |
1756 | */ | |
1757 | ||
693758d5 | 1758 | void |
1be6b49c | 1759 | scm_c_write (SCM port, const void *ptr, size_t size) |
693758d5 | 1760 | #define FUNC_NAME "scm_c_write" |
6fe692e9 | 1761 | { |
693758d5 LC |
1762 | scm_t_port *pt; |
1763 | scm_t_ptob_descriptor *ptob; | |
1764 | ||
1765 | SCM_VALIDATE_OPOUTPORT (1, port); | |
1766 | ||
1767 | pt = SCM_PTAB_ENTRY (port); | |
1768 | ptob = &scm_ptobs[SCM_PTOBNUM (port)]; | |
6fe692e9 MD |
1769 | |
1770 | if (pt->rw_active == SCM_PORT_READ) | |
1771 | scm_end_input (port); | |
1772 | ||
1773 | ptob->write (port, ptr, size); | |
1774 | ||
1775 | if (pt->rw_random) | |
1776 | pt->rw_active = SCM_PORT_WRITE; | |
1777 | } | |
693758d5 | 1778 | #undef FUNC_NAME |
3cb988bd | 1779 | |
fca43887 | 1780 | void |
a284e297 | 1781 | scm_flush (SCM port) |
ee149d03 | 1782 | { |
c014a02e | 1783 | long i = SCM_PTOBNUM (port); |
7f2a6c38 | 1784 | assert (i >= 0); |
affc96b5 | 1785 | (scm_ptobs[i].flush) (port); |
ee149d03 JB |
1786 | } |
1787 | ||
283a1a0e | 1788 | void |
a284e297 | 1789 | scm_end_input (SCM port) |
283a1a0e | 1790 | { |
c014a02e | 1791 | long offset; |
92c2555f | 1792 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
283a1a0e | 1793 | |
45c0878b | 1794 | scm_i_clear_pending_eof (port); |
283a1a0e GH |
1795 | if (pt->read_buf == pt->putback_buf) |
1796 | { | |
1797 | offset = pt->read_end - pt->read_pos; | |
1798 | pt->read_buf = pt->saved_read_buf; | |
1799 | pt->read_pos = pt->saved_read_pos; | |
1800 | pt->read_end = pt->saved_read_end; | |
1801 | pt->read_buf_size = pt->saved_read_buf_size; | |
1802 | } | |
1803 | else | |
1804 | offset = 0; | |
1805 | ||
affc96b5 | 1806 | scm_ptobs[SCM_PTOBNUM (port)].end_input (port, offset); |
283a1a0e GH |
1807 | } |
1808 | ||
ee149d03 JB |
1809 | \f |
1810 | ||
1811 | ||
7f6c3f8f MW |
1812 | static void |
1813 | scm_i_unget_bytes (const unsigned char *buf, size_t len, SCM port) | |
1814 | #define FUNC_NAME "scm_unget_bytes" | |
ee149d03 | 1815 | { |
92c2555f | 1816 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
7f6c3f8f | 1817 | size_t old_len, new_len; |
840ae05d | 1818 | |
45c0878b | 1819 | scm_i_clear_pending_eof (port); |
6c951427 | 1820 | |
7f6c3f8f | 1821 | if (pt->read_buf != pt->putback_buf) |
6c951427 GH |
1822 | /* switch to the put-back buffer. */ |
1823 | { | |
1824 | if (pt->putback_buf == NULL) | |
1825 | { | |
7f6c3f8f MW |
1826 | pt->putback_buf_size = (len > SCM_INITIAL_PUTBACK_BUF_SIZE |
1827 | ? len : SCM_INITIAL_PUTBACK_BUF_SIZE); | |
c357d546 | 1828 | pt->putback_buf |
92d8fd32 | 1829 | = (unsigned char *) scm_gc_malloc_pointerless |
7f6c3f8f | 1830 | (pt->putback_buf_size, "putback buffer"); |
6c951427 GH |
1831 | } |
1832 | ||
1833 | pt->saved_read_buf = pt->read_buf; | |
1834 | pt->saved_read_pos = pt->read_pos; | |
1835 | pt->saved_read_end = pt->read_end; | |
1836 | pt->saved_read_buf_size = pt->read_buf_size; | |
1837 | ||
7f6c3f8f MW |
1838 | /* Put read_pos at the end of the buffer, so that ungets will not |
1839 | have to shift the buffer contents each time. */ | |
1840 | pt->read_buf = pt->putback_buf; | |
1841 | pt->read_pos = pt->read_end = pt->putback_buf + pt->putback_buf_size; | |
6c951427 GH |
1842 | pt->read_buf_size = pt->putback_buf_size; |
1843 | } | |
1844 | ||
7f6c3f8f MW |
1845 | old_len = pt->read_end - pt->read_pos; |
1846 | new_len = old_len + len; | |
1847 | ||
1848 | if (new_len > pt->read_buf_size) | |
1849 | /* The putback buffer needs to be enlarged. */ | |
1850 | { | |
1851 | size_t new_buf_size; | |
1852 | unsigned char *new_buf, *new_end, *new_pos; | |
1853 | ||
1854 | new_buf_size = pt->read_buf_size * 2; | |
1855 | if (new_buf_size < new_len) | |
1856 | new_buf_size = new_len; | |
1857 | ||
1858 | new_buf = (unsigned char *) | |
1859 | scm_gc_malloc_pointerless (new_buf_size, "putback buffer"); | |
1860 | ||
1861 | /* Put the bytes at the end of the buffer, so that future | |
1862 | ungets won't need to shift the buffer. */ | |
1863 | new_end = new_buf + new_buf_size; | |
1864 | new_pos = new_end - old_len; | |
1865 | memcpy (new_pos, pt->read_pos, old_len); | |
1866 | ||
1867 | pt->read_buf = pt->putback_buf = new_buf; | |
1868 | pt->read_pos = new_pos; | |
1869 | pt->read_end = new_end; | |
1870 | pt->read_buf_size = pt->putback_buf_size = new_buf_size; | |
1871 | } | |
1872 | else if (pt->read_buf + len < pt->read_pos) | |
1873 | /* If needed, shift the existing buffer contents up. | |
1874 | This should not happen unless some external code | |
1875 | manipulates the putback buffer pointers. */ | |
1876 | { | |
1877 | unsigned char *new_end = pt->read_buf + pt->read_buf_size; | |
1878 | unsigned char *new_pos = new_end - old_len; | |
1879 | ||
1880 | memmove (new_pos, pt->read_pos, old_len); | |
1881 | pt->read_pos = new_pos; | |
1882 | pt->read_end = new_end; | |
1883 | } | |
1884 | ||
1885 | /* Move read_pos back and copy the bytes there. */ | |
1886 | pt->read_pos -= len; | |
1887 | memcpy (pt->read_buf + (pt->read_pos - pt->read_buf), buf, len); | |
1888 | ||
1889 | if (pt->rw_active == SCM_PORT_WRITE) | |
1890 | scm_flush (port); | |
ee149d03 | 1891 | |
840ae05d JB |
1892 | if (pt->rw_random) |
1893 | pt->rw_active = SCM_PORT_READ; | |
889975e5 MG |
1894 | } |
1895 | #undef FUNC_NAME | |
1896 | ||
7f6c3f8f MW |
1897 | void |
1898 | scm_unget_bytes (const unsigned char *buf, size_t len, SCM port) | |
1899 | { | |
1900 | scm_i_unget_bytes (buf, len, port); | |
1901 | } | |
1902 | ||
1903 | void | |
1904 | scm_unget_byte (int c, SCM port) | |
1905 | { | |
1906 | unsigned char byte; | |
1907 | ||
1908 | byte = c; | |
1909 | scm_i_unget_bytes (&byte, 1, port); | |
1910 | } | |
1911 | ||
63479112 | 1912 | void |
889975e5 MG |
1913 | scm_ungetc (scm_t_wchar c, SCM port) |
1914 | #define FUNC_NAME "scm_ungetc" | |
1915 | { | |
1916 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
63479112 LC |
1917 | char *result; |
1918 | char result_buf[10]; | |
1919 | const char *encoding; | |
889975e5 | 1920 | size_t len; |
889975e5 | 1921 | |
63479112 LC |
1922 | if (pt->encoding != NULL) |
1923 | encoding = pt->encoding; | |
1924 | else | |
1925 | encoding = "ISO-8859-1"; | |
1926 | ||
1927 | len = sizeof (result_buf); | |
1928 | result = u32_conv_to_encoding (encoding, | |
1929 | (enum iconv_ilseq_handler) pt->ilseq_handler, | |
1930 | (uint32_t *) &c, 1, NULL, | |
1931 | result_buf, &len); | |
1932 | ||
1933 | if (SCM_UNLIKELY (result == NULL || len == 0)) | |
6851d3be LC |
1934 | scm_encoding_error (FUNC_NAME, errno, |
1935 | "conversion to port encoding failed", | |
1936 | SCM_BOOL_F, SCM_MAKE_CHAR (c)); | |
63479112 | 1937 | |
7f6c3f8f | 1938 | scm_i_unget_bytes ((unsigned char *) result, len, port); |
63479112 LC |
1939 | |
1940 | if (SCM_UNLIKELY (result != result_buf)) | |
1941 | free (result); | |
840ae05d | 1942 | |
ee149d03 JB |
1943 | if (c == '\n') |
1944 | { | |
1945 | /* What should col be in this case? | |
1946 | * We'll leave it at -1. | |
1947 | */ | |
1948 | SCM_LINUM (port) -= 1; | |
1949 | } | |
1950 | else | |
1951 | SCM_COL(port) -= 1; | |
1952 | } | |
c6c79933 | 1953 | #undef FUNC_NAME |
ee149d03 JB |
1954 | |
1955 | ||
1956 | void | |
70d63753 | 1957 | scm_ungets (const char *s, int n, SCM port) |
ee149d03 JB |
1958 | { |
1959 | /* This is simple minded and inefficient, but unreading strings is | |
1960 | * probably not a common operation, and remember that line and | |
1961 | * column numbers have to be handled... | |
1962 | * | |
1963 | * Please feel free to write an optimized version! | |
1964 | */ | |
1965 | while (n--) | |
1966 | scm_ungetc (s[n], port); | |
1967 | } | |
1968 | ||
1969 | ||
3b3b36dd | 1970 | SCM_DEFINE (scm_peek_char, "peek-char", 0, 1, 0, |
1bbd0b84 | 1971 | (SCM port), |
1e6808ea MG |
1972 | "Return the next character available from @var{port},\n" |
1973 | "@emph{without} updating @var{port} to point to the following\n" | |
1974 | "character. If no more characters are available, the\n" | |
c2dfff19 KR |
1975 | "end-of-file object is returned.\n" |
1976 | "\n" | |
1977 | "The value returned by\n" | |
1e6808ea MG |
1978 | "a call to @code{peek-char} is the same as the value that would\n" |
1979 | "have been returned by a call to @code{read-char} on the same\n" | |
1980 | "port. The only difference is that the very next call to\n" | |
1981 | "@code{read-char} or @code{peek-char} on that @var{port} will\n" | |
1982 | "return the value returned by the preceding call to\n" | |
1983 | "@code{peek-char}. In particular, a call to @code{peek-char} on\n" | |
1984 | "an interactive port will hang waiting for input whenever a call\n" | |
c62da8f8 LC |
1985 | "to @code{read-char} would have hung.\n" |
1986 | "\n" | |
1987 | "As for @code{read-char}, a @code{decoding-error} may be raised\n" | |
1988 | "if such a situation occurs. However, unlike with @code{read-char},\n" | |
1989 | "@var{port} still points at the beginning of the erroneous byte\n" | |
1990 | "sequence when the error is raised.\n") | |
1bbd0b84 | 1991 | #define FUNC_NAME s_scm_peek_char |
ee149d03 | 1992 | { |
c62da8f8 | 1993 | int err; |
fd5eec2b LC |
1994 | SCM result; |
1995 | scm_t_wchar c; | |
1996 | char bytes[SCM_MBCHAR_BUF_SIZE]; | |
7f6c3f8f | 1997 | long column, line; |
fd5eec2b LC |
1998 | size_t len; |
1999 | ||
ee149d03 | 2000 | if (SCM_UNBNDP (port)) |
9de87eea | 2001 | port = scm_current_input_port (); |
b2456dd4 | 2002 | SCM_VALIDATE_OPINPORT (1, port); |
fd5eec2b LC |
2003 | |
2004 | column = SCM_COL (port); | |
2005 | line = SCM_LINUM (port); | |
2006 | ||
c62da8f8 | 2007 | err = get_codepoint (port, &c, bytes, &len); |
fd5eec2b | 2008 | |
7f6c3f8f | 2009 | scm_i_unget_bytes ((unsigned char *) bytes, len, port); |
fd5eec2b | 2010 | |
c62da8f8 LC |
2011 | SCM_COL (port) = column; |
2012 | SCM_LINUM (port) = line; | |
fd5eec2b | 2013 | |
c62da8f8 LC |
2014 | if (SCM_UNLIKELY (err != 0)) |
2015 | { | |
2016 | scm_decoding_error (FUNC_NAME, err, "input decoding error", port); | |
2017 | ||
2018 | /* Shouldn't happen since `catch' always aborts to prompt. */ | |
2019 | result = SCM_BOOL_F; | |
fd5eec2b | 2020 | } |
c62da8f8 | 2021 | else if (c == EOF) |
45c0878b MW |
2022 | { |
2023 | scm_i_set_pending_eof (port); | |
2024 | result = SCM_EOF_VAL; | |
2025 | } | |
c62da8f8 LC |
2026 | else |
2027 | result = SCM_MAKE_CHAR (c); | |
fd5eec2b LC |
2028 | |
2029 | return result; | |
3cb988bd | 2030 | } |
1bbd0b84 | 2031 | #undef FUNC_NAME |
3cb988bd | 2032 | |
1be4270a | 2033 | SCM_DEFINE (scm_unread_char, "unread-char", 1, 1, 0, |
1bbd0b84 | 2034 | (SCM cobj, SCM port), |
b7e64f8b BT |
2035 | "Place character @var{cobj} in @var{port} so that it will be\n" |
2036 | "read by the next read operation. If called multiple times, the\n" | |
2037 | "unread characters will be read again in last-in first-out\n" | |
2038 | "order. If @var{port} is not supplied, the current input port\n" | |
2039 | "is used.") | |
1bbd0b84 | 2040 | #define FUNC_NAME s_scm_unread_char |
0f2d19dd JB |
2041 | { |
2042 | int c; | |
2043 | ||
34d19ef6 | 2044 | SCM_VALIDATE_CHAR (1, cobj); |
0f2d19dd | 2045 | if (SCM_UNBNDP (port)) |
9de87eea | 2046 | port = scm_current_input_port (); |
b2456dd4 | 2047 | SCM_VALIDATE_OPINPORT (2, port); |
0f2d19dd | 2048 | |
7866a09b | 2049 | c = SCM_CHAR (cobj); |
0f2d19dd | 2050 | |
b7f3516f | 2051 | scm_ungetc (c, port); |
0f2d19dd JB |
2052 | return cobj; |
2053 | } | |
1bbd0b84 | 2054 | #undef FUNC_NAME |
0f2d19dd | 2055 | |
a1ec6916 | 2056 | SCM_DEFINE (scm_unread_string, "unread-string", 2, 0, 0, |
1bbd0b84 | 2057 | (SCM str, SCM port), |
b380b885 MD |
2058 | "Place the string @var{str} in @var{port} so that its characters will be\n" |
2059 | "read in subsequent read operations. If called multiple times, the\n" | |
2060 | "unread characters will be read again in last-in first-out order. If\n" | |
2061 | "@var{port} is not supplied, the current-input-port is used.") | |
1bbd0b84 | 2062 | #define FUNC_NAME s_scm_unread_string |
ee1e7e13 | 2063 | { |
889975e5 | 2064 | int n; |
34d19ef6 | 2065 | SCM_VALIDATE_STRING (1, str); |
ee1e7e13 | 2066 | if (SCM_UNBNDP (port)) |
9de87eea | 2067 | port = scm_current_input_port (); |
b2456dd4 | 2068 | SCM_VALIDATE_OPINPORT (2, port); |
ee1e7e13 | 2069 | |
889975e5 MG |
2070 | n = scm_i_string_length (str); |
2071 | ||
2072 | while (n--) | |
2073 | scm_ungetc (scm_i_string_ref (str, n), port); | |
ee1e7e13 MD |
2074 | |
2075 | return str; | |
2076 | } | |
1bbd0b84 | 2077 | #undef FUNC_NAME |
ee1e7e13 | 2078 | |
a1ec6916 | 2079 | SCM_DEFINE (scm_seek, "seek", 3, 0, 0, |
1e6808ea | 2080 | (SCM fd_port, SCM offset, SCM whence), |
b7e64f8b | 2081 | "Sets the current position of @var{fd_port} to the integer\n" |
1e6808ea MG |
2082 | "@var{offset}, which is interpreted according to the value of\n" |
2083 | "@var{whence}.\n" | |
2084 | "\n" | |
2085 | "One of the following variables should be supplied for\n" | |
2086 | "@var{whence}:\n" | |
b380b885 MD |
2087 | "@defvar SEEK_SET\n" |
2088 | "Seek from the beginning of the file.\n" | |
2089 | "@end defvar\n" | |
2090 | "@defvar SEEK_CUR\n" | |
2091 | "Seek from the current position.\n" | |
2092 | "@end defvar\n" | |
2093 | "@defvar SEEK_END\n" | |
2094 | "Seek from the end of the file.\n" | |
1e6808ea | 2095 | "@end defvar\n" |
b7e64f8b | 2096 | "If @var{fd_port} is a file descriptor, the underlying system\n" |
1e6808ea MG |
2097 | "call is @code{lseek}. @var{port} may be a string port.\n" |
2098 | "\n" | |
2099 | "The value returned is the new position in the file. This means\n" | |
2100 | "that the current position of a port can be obtained using:\n" | |
2101 | "@lisp\n" | |
b380b885 | 2102 | "(seek port 0 SEEK_CUR)\n" |
1e6808ea | 2103 | "@end lisp") |
1bbd0b84 | 2104 | #define FUNC_NAME s_scm_seek |
840ae05d | 2105 | { |
840ae05d JB |
2106 | int how; |
2107 | ||
1e6808ea | 2108 | fd_port = SCM_COERCE_OUTPORT (fd_port); |
840ae05d | 2109 | |
a55c2b68 | 2110 | how = scm_to_int (whence); |
840ae05d | 2111 | if (how != SEEK_SET && how != SEEK_CUR && how != SEEK_END) |
1bbd0b84 | 2112 | SCM_OUT_OF_RANGE (3, whence); |
23f2b9a3 | 2113 | |
0a94eb00 | 2114 | if (SCM_OPPORTP (fd_port)) |
840ae05d | 2115 | { |
cdd3d6c9 | 2116 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (fd_port); |
92c2555f | 2117 | scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (fd_port); |
f1ce9199 LC |
2118 | off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset); |
2119 | off_t_or_off64_t rv; | |
840ae05d JB |
2120 | |
2121 | if (!ptob->seek) | |
1bbd0b84 | 2122 | SCM_MISC_ERROR ("port is not seekable", |
1e6808ea | 2123 | scm_cons (fd_port, SCM_EOL)); |
840ae05d | 2124 | else |
cdd3d6c9 MW |
2125 | rv = ptob->seek (fd_port, off, how); |
2126 | ||
2127 | /* Set stream-start flags according to new position. */ | |
2128 | pti->at_stream_start_for_bom_read = (rv == 0); | |
2129 | pti->at_stream_start_for_bom_write = (rv == 0); | |
2130 | ||
2131 | scm_i_clear_pending_eof (fd_port); | |
2132 | ||
f1ce9199 | 2133 | return scm_from_off_t_or_off64_t (rv); |
840ae05d JB |
2134 | } |
2135 | else /* file descriptor?. */ | |
2136 | { | |
23f2b9a3 KR |
2137 | off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset); |
2138 | off_t_or_off64_t rv; | |
2139 | rv = lseek_or_lseek64 (scm_to_int (fd_port), off, how); | |
840ae05d | 2140 | if (rv == -1) |
1bbd0b84 | 2141 | SCM_SYSERROR; |
23f2b9a3 | 2142 | return scm_from_off_t_or_off64_t (rv); |
840ae05d | 2143 | } |
840ae05d | 2144 | } |
1bbd0b84 | 2145 | #undef FUNC_NAME |
840ae05d | 2146 | |
8ab3d8a0 KR |
2147 | #ifndef O_BINARY |
2148 | #define O_BINARY 0 | |
2149 | #endif | |
2150 | ||
2151 | /* Mingw has ftruncate(), perhaps implemented above using chsize, but | |
2152 | doesn't have the filename version truncate(), hence this code. */ | |
2153 | #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE | |
2154 | static int | |
2155 | truncate (const char *file, off_t length) | |
82893676 | 2156 | { |
8ab3d8a0 KR |
2157 | int ret, fdes; |
2158 | ||
2159 | fdes = open (file, O_BINARY | O_WRONLY); | |
2160 | if (fdes == -1) | |
2161 | return -1; | |
2162 | ||
2163 | ret = ftruncate (fdes, length); | |
2164 | if (ret == -1) | |
82893676 | 2165 | { |
8ab3d8a0 | 2166 | int save_errno = errno; |
82893676 | 2167 | close (fdes); |
8ab3d8a0 KR |
2168 | errno = save_errno; |
2169 | return -1; | |
82893676 | 2170 | } |
8ab3d8a0 KR |
2171 | |
2172 | return close (fdes); | |
82893676 | 2173 | } |
8ab3d8a0 | 2174 | #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */ |
82893676 | 2175 | |
a1ec6916 | 2176 | SCM_DEFINE (scm_truncate_file, "truncate-file", 1, 1, 0, |
1bbd0b84 | 2177 | (SCM object, SCM length), |
b7e64f8b BT |
2178 | "Truncate file @var{object} to @var{length} bytes. @var{object}\n" |
2179 | "can be a filename string, a port object, or an integer file\n" | |
2180 | "descriptor.\n" | |
8ab3d8a0 KR |
2181 | "The return value is unspecified.\n" |
2182 | "\n" | |
2183 | "For a port or file descriptor @var{length} can be omitted, in\n" | |
2184 | "which case the file is truncated at the current position (per\n" | |
2185 | "@code{ftell} above).\n" | |
2186 | "\n" | |
2187 | "On most systems a file can be extended by giving a length\n" | |
2188 | "greater than the current size, but this is not mandatory in the\n" | |
2189 | "POSIX standard.") | |
1bbd0b84 | 2190 | #define FUNC_NAME s_scm_truncate_file |
840ae05d | 2191 | { |
69bc9ff3 | 2192 | int rv; |
69bc9ff3 | 2193 | |
2b829bbb KR |
2194 | /* "object" can be a port, fdes or filename. |
2195 | ||
2196 | Negative "length" makes no sense, but it's left to truncate() or | |
2197 | ftruncate() to give back an error for that (normally EINVAL). | |
2198 | */ | |
840ae05d | 2199 | |
840ae05d JB |
2200 | if (SCM_UNBNDP (length)) |
2201 | { | |
69bc9ff3 | 2202 | /* must supply length if object is a filename. */ |
7f9994d9 | 2203 | if (scm_is_string (object)) |
34d19ef6 | 2204 | SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL); |
69bc9ff3 | 2205 | |
e11e83f3 | 2206 | length = scm_seek (object, SCM_INUM0, scm_from_int (SEEK_CUR)); |
840ae05d | 2207 | } |
3fe6190f | 2208 | |
69bc9ff3 | 2209 | object = SCM_COERCE_OUTPORT (object); |
e11e83f3 | 2210 | if (scm_is_integer (object)) |
69bc9ff3 | 2211 | { |
23f2b9a3 KR |
2212 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); |
2213 | SCM_SYSCALL (rv = ftruncate_or_ftruncate64 (scm_to_int (object), | |
2214 | c_length)); | |
69bc9ff3 | 2215 | } |
0c95b57d | 2216 | else if (SCM_OPOUTPORTP (object)) |
69bc9ff3 | 2217 | { |
f1ce9199 | 2218 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); |
92c2555f MV |
2219 | scm_t_port *pt = SCM_PTAB_ENTRY (object); |
2220 | scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (object); | |
45c0878b | 2221 | |
affc96b5 | 2222 | if (!ptob->truncate) |
1bbd0b84 | 2223 | SCM_MISC_ERROR ("port is not truncatable", SCM_EOL); |
45c0878b MW |
2224 | |
2225 | scm_i_clear_pending_eof (object); | |
69bc9ff3 | 2226 | if (pt->rw_active == SCM_PORT_READ) |
affc96b5 | 2227 | scm_end_input (object); |
69bc9ff3 | 2228 | else if (pt->rw_active == SCM_PORT_WRITE) |
affc96b5 | 2229 | ptob->flush (object); |
45c0878b | 2230 | |
affc96b5 | 2231 | ptob->truncate (object, c_length); |
69bc9ff3 GH |
2232 | rv = 0; |
2233 | } | |
2234 | else | |
2235 | { | |
2b829bbb | 2236 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); |
7f9994d9 MV |
2237 | char *str = scm_to_locale_string (object); |
2238 | int eno; | |
2b829bbb | 2239 | SCM_SYSCALL (rv = truncate_or_truncate64 (str, c_length)); |
7f9994d9 MV |
2240 | eno = errno; |
2241 | free (str); | |
2242 | errno = eno; | |
69bc9ff3 GH |
2243 | } |
2244 | if (rv == -1) | |
1bbd0b84 | 2245 | SCM_SYSERROR; |
840ae05d JB |
2246 | return SCM_UNSPECIFIED; |
2247 | } | |
1bbd0b84 | 2248 | #undef FUNC_NAME |
840ae05d | 2249 | |
a1ec6916 | 2250 | SCM_DEFINE (scm_port_line, "port-line", 1, 0, 0, |
1bbd0b84 | 2251 | (SCM port), |
a150979d KR |
2252 | "Return the current line number for @var{port}.\n" |
2253 | "\n" | |
2254 | "The first line of a file is 0. But you might want to add 1\n" | |
2255 | "when printing line numbers, since starting from 1 is\n" | |
2256 | "traditional in error messages, and likely to be more natural to\n" | |
2257 | "non-programmers.") | |
1bbd0b84 | 2258 | #define FUNC_NAME s_scm_port_line |
0f2d19dd | 2259 | { |
78446828 | 2260 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2261 | SCM_VALIDATE_OPENPORT (1, port); |
651f2cd2 | 2262 | return scm_from_long (SCM_LINUM (port)); |
0f2d19dd | 2263 | } |
1bbd0b84 | 2264 | #undef FUNC_NAME |
0f2d19dd | 2265 | |
a1ec6916 | 2266 | SCM_DEFINE (scm_set_port_line_x, "set-port-line!", 2, 0, 0, |
1bbd0b84 | 2267 | (SCM port, SCM line), |
a150979d KR |
2268 | "Set the current line number for @var{port} to @var{line}. The\n" |
2269 | "first line of a file is 0.") | |
1bbd0b84 | 2270 | #define FUNC_NAME s_scm_set_port_line_x |
d043d8c2 | 2271 | { |
360fc44c | 2272 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2273 | SCM_VALIDATE_OPENPORT (1, port); |
651f2cd2 | 2274 | SCM_PTAB_ENTRY (port)->line_number = scm_to_long (line); |
564478fd | 2275 | return SCM_UNSPECIFIED; |
d043d8c2 | 2276 | } |
1bbd0b84 | 2277 | #undef FUNC_NAME |
d043d8c2 | 2278 | |
a1ec6916 | 2279 | SCM_DEFINE (scm_port_column, "port-column", 1, 0, 0, |
1bbd0b84 | 2280 | (SCM port), |
a150979d KR |
2281 | "Return the current column number of @var{port}.\n" |
2282 | "If the number is\n" | |
b380b885 MD |
2283 | "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n" |
2284 | "- i.e. the first character of the first line is line 0, column 0.\n" | |
2285 | "(However, when you display a file position, for example in an error\n" | |
650a1cf9 | 2286 | "message, we recommend you add 1 to get 1-origin integers. This is\n" |
b380b885 MD |
2287 | "because lines and column numbers traditionally start with 1, and that is\n" |
2288 | "what non-programmers will find most natural.)") | |
1bbd0b84 | 2289 | #define FUNC_NAME s_scm_port_column |
0f2d19dd | 2290 | { |
78446828 | 2291 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2292 | SCM_VALIDATE_OPENPORT (1, port); |
e11e83f3 | 2293 | return scm_from_int (SCM_COL (port)); |
0f2d19dd | 2294 | } |
1bbd0b84 | 2295 | #undef FUNC_NAME |
0f2d19dd | 2296 | |
a1ec6916 | 2297 | SCM_DEFINE (scm_set_port_column_x, "set-port-column!", 2, 0, 0, |
1bbd0b84 | 2298 | (SCM port, SCM column), |
a150979d KR |
2299 | "Set the current column of @var{port}. Before reading the first\n" |
2300 | "character on a line the column should be 0.") | |
1bbd0b84 | 2301 | #define FUNC_NAME s_scm_set_port_column_x |
d043d8c2 | 2302 | { |
360fc44c | 2303 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2304 | SCM_VALIDATE_OPENPORT (1, port); |
a55c2b68 | 2305 | SCM_PTAB_ENTRY (port)->column_number = scm_to_int (column); |
564478fd | 2306 | return SCM_UNSPECIFIED; |
d043d8c2 | 2307 | } |
1bbd0b84 | 2308 | #undef FUNC_NAME |
d043d8c2 | 2309 | |
a1ec6916 | 2310 | SCM_DEFINE (scm_port_filename, "port-filename", 1, 0, 0, |
1bbd0b84 | 2311 | (SCM port), |
ac012a27 AW |
2312 | "Return the filename associated with @var{port}, or @code{#f}\n" |
2313 | "if no filename is associated with the port.") | |
1bbd0b84 | 2314 | #define FUNC_NAME s_scm_port_filename |
0f2d19dd | 2315 | { |
78446828 | 2316 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2317 | SCM_VALIDATE_OPENPORT (1, port); |
b24b5e13 | 2318 | return SCM_FILENAME (port); |
0f2d19dd | 2319 | } |
1bbd0b84 | 2320 | #undef FUNC_NAME |
0f2d19dd | 2321 | |
a1ec6916 | 2322 | SCM_DEFINE (scm_set_port_filename_x, "set-port-filename!", 2, 0, 0, |
1bbd0b84 | 2323 | (SCM port, SCM filename), |
b380b885 MD |
2324 | "Change the filename associated with @var{port}, using the current input\n" |
2325 | "port if none is specified. Note that this does not change the port's\n" | |
2326 | "source of data, but only the value that is returned by\n" | |
2327 | "@code{port-filename} and reported in diagnostic output.") | |
1bbd0b84 | 2328 | #define FUNC_NAME s_scm_set_port_filename_x |
d14af9f2 | 2329 | { |
360fc44c | 2330 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 2331 | SCM_VALIDATE_OPENPORT (1, port); |
360fc44c | 2332 | /* We allow the user to set the filename to whatever he likes. */ |
b24b5e13 DH |
2333 | SCM_SET_FILENAME (port, filename); |
2334 | return SCM_UNSPECIFIED; | |
d14af9f2 | 2335 | } |
1bbd0b84 | 2336 | #undef FUNC_NAME |
d14af9f2 | 2337 | |
d6a6989e LC |
2338 | /* A fluid specifying the default encoding for newly created ports. If it is |
2339 | a string, that is the encoding. If it is #f, it is in the "native" | |
2340 | (Latin-1) encoding. */ | |
2341 | SCM_VARIABLE (default_port_encoding_var, "%default-port-encoding"); | |
2342 | ||
889975e5 MG |
2343 | static int scm_port_encoding_init = 0; |
2344 | ||
9d9c66ba LC |
2345 | /* Use ENCODING as the default encoding for future ports. */ |
2346 | void | |
2347 | scm_i_set_default_port_encoding (const char *encoding) | |
2348 | { | |
2349 | if (!scm_port_encoding_init | |
2350 | || !scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var))) | |
2351 | scm_misc_error (NULL, "tried to set port encoding fluid before it is initialized", | |
2352 | SCM_EOL); | |
2353 | ||
2354 | if (encoding == NULL | |
96965a6e MW |
2355 | || c_strcasecmp (encoding, "ASCII") == 0 |
2356 | || c_strcasecmp (encoding, "ANSI_X3.4-1968") == 0 | |
2357 | || c_strcasecmp (encoding, "ISO-8859-1") == 0) | |
9d9c66ba LC |
2358 | scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), SCM_BOOL_F); |
2359 | else | |
2360 | scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), | |
2361 | scm_from_locale_string (encoding)); | |
2362 | } | |
2363 | ||
2364 | /* Return the name of the default encoding for newly created ports; a | |
2365 | return value of NULL means "ISO-8859-1". */ | |
889975e5 | 2366 | const char * |
9d9c66ba | 2367 | scm_i_default_port_encoding (void) |
889975e5 | 2368 | { |
9d9c66ba LC |
2369 | if (!scm_port_encoding_init) |
2370 | return NULL; | |
2371 | else if (!scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var))) | |
2372 | return NULL; | |
2373 | else | |
889975e5 | 2374 | { |
9d9c66ba LC |
2375 | SCM encoding; |
2376 | ||
2377 | encoding = scm_fluid_ref (SCM_VARIABLE_REF (default_port_encoding_var)); | |
2378 | if (!scm_is_string (encoding)) | |
889975e5 MG |
2379 | return NULL; |
2380 | else | |
9d9c66ba | 2381 | return scm_i_string_chars (encoding); |
889975e5 MG |
2382 | } |
2383 | } | |
2384 | ||
cdd3d6c9 MW |
2385 | /* If the next LEN bytes from PORT are equal to those in BYTES, then |
2386 | return 1, else return 0. Leave the port position unchanged. */ | |
2387 | static int | |
2388 | looking_at_bytes (SCM port, const unsigned char *bytes, int len) | |
2389 | { | |
2390 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
cdd3d6c9 MW |
2391 | int i = 0; |
2392 | ||
2393 | while (i < len && scm_peek_byte_or_eof (port) == bytes[i]) | |
2394 | { | |
2395 | pt->read_pos++; | |
2396 | i++; | |
2397 | } | |
7f6c3f8f MW |
2398 | scm_i_unget_bytes (bytes, i, port); |
2399 | return (i == len); | |
cdd3d6c9 MW |
2400 | } |
2401 | ||
2402 | static const unsigned char scm_utf8_bom[3] = {0xEF, 0xBB, 0xBF}; | |
2403 | static const unsigned char scm_utf16be_bom[2] = {0xFE, 0xFF}; | |
2404 | static const unsigned char scm_utf16le_bom[2] = {0xFF, 0xFE}; | |
2405 | static const unsigned char scm_utf32be_bom[4] = {0x00, 0x00, 0xFE, 0xFF}; | |
2406 | static const unsigned char scm_utf32le_bom[4] = {0xFF, 0xFE, 0x00, 0x00}; | |
2407 | ||
2408 | /* Decide what byte order to use for a UTF-16 port. Return "UTF-16BE" | |
2409 | or "UTF-16LE". MODE must be either SCM_PORT_READ or SCM_PORT_WRITE, | |
2410 | and specifies which operation is about to be done. The MODE | |
2411 | determines how we will decide the byte order. We deliberately avoid | |
2412 | reading from the port unless the user is about to do so. If the user | |
2413 | is about to read, then we look for a BOM, and if present, we use it | |
2414 | to determine the byte order. Otherwise we choose big endian, as | |
2415 | recommended by the Unicode Standard. Note that the BOM (if any) is | |
2416 | not consumed here. */ | |
2417 | static const char * | |
2418 | decide_utf16_encoding (SCM port, scm_t_port_rw_active mode) | |
2419 | { | |
2420 | if (mode == SCM_PORT_READ | |
2421 | && SCM_PORT_GET_INTERNAL (port)->at_stream_start_for_bom_read | |
2422 | && looking_at_bytes (port, scm_utf16le_bom, sizeof scm_utf16le_bom)) | |
2423 | return "UTF-16LE"; | |
2424 | else | |
2425 | return "UTF-16BE"; | |
2426 | } | |
2427 | ||
2428 | /* Decide what byte order to use for a UTF-32 port. Return "UTF-32BE" | |
2429 | or "UTF-32LE". See the comment above 'decide_utf16_encoding' for | |
2430 | details. */ | |
2431 | static const char * | |
2432 | decide_utf32_encoding (SCM port, scm_t_port_rw_active mode) | |
2433 | { | |
2434 | if (mode == SCM_PORT_READ | |
2435 | && SCM_PORT_GET_INTERNAL (port)->at_stream_start_for_bom_read | |
2436 | && looking_at_bytes (port, scm_utf32le_bom, sizeof scm_utf32le_bom)) | |
2437 | return "UTF-32LE"; | |
2438 | else | |
2439 | return "UTF-32BE"; | |
2440 | } | |
2441 | ||
337edc59 | 2442 | static void |
1fa89aaf | 2443 | finalize_iconv_descriptors (void *ptr, void *data) |
337edc59 MW |
2444 | { |
2445 | close_iconv_descriptors (ptr); | |
2446 | } | |
2447 | ||
2448 | static scm_t_iconv_descriptors * | |
2449 | open_iconv_descriptors (const char *encoding, int reading, int writing) | |
2450 | { | |
2451 | scm_t_iconv_descriptors *id; | |
2452 | iconv_t input_cd, output_cd; | |
2453 | ||
2454 | input_cd = (iconv_t) -1; | |
2455 | output_cd = (iconv_t) -1; | |
2456 | if (reading) | |
2457 | { | |
2458 | /* Open an input iconv conversion descriptor, from ENCODING | |
2459 | to UTF-8. We choose UTF-8, not UTF-32, because iconv | |
2460 | implementations can typically convert from anything to | |
2461 | UTF-8, but not to UTF-32 (see | |
2462 | <http://lists.gnu.org/archive/html/bug-libunistring/2010-09/msg00007.html>). */ | |
2463 | ||
2464 | /* Assume opening an iconv descriptor causes about 16 KB of | |
2465 | allocation. */ | |
2466 | scm_gc_register_allocation (16 * 1024); | |
2467 | ||
2468 | input_cd = iconv_open ("UTF-8", encoding); | |
2469 | if (input_cd == (iconv_t) -1) | |
2470 | goto invalid_encoding; | |
2471 | } | |
2472 | ||
2473 | if (writing) | |
2474 | { | |
2475 | /* Assume opening an iconv descriptor causes about 16 KB of | |
2476 | allocation. */ | |
2477 | scm_gc_register_allocation (16 * 1024); | |
2478 | ||
2479 | output_cd = iconv_open (encoding, "UTF-8"); | |
2480 | if (output_cd == (iconv_t) -1) | |
2481 | { | |
2482 | if (input_cd != (iconv_t) -1) | |
2483 | iconv_close (input_cd); | |
2484 | goto invalid_encoding; | |
2485 | } | |
2486 | } | |
2487 | ||
2488 | id = scm_gc_malloc_pointerless (sizeof (*id), "iconv descriptors"); | |
2489 | id->input_cd = input_cd; | |
2490 | id->output_cd = output_cd; | |
2491 | ||
2492 | /* Register a finalizer to close the descriptors. */ | |
2493 | scm_i_set_finalizer (id, finalize_iconv_descriptors, NULL); | |
2494 | ||
2495 | return id; | |
2496 | ||
2497 | invalid_encoding: | |
2498 | { | |
2499 | SCM err; | |
2500 | err = scm_from_locale_string (encoding); | |
2501 | scm_misc_error ("open_iconv_descriptors", | |
2502 | "invalid or unknown character encoding ~s", | |
2503 | scm_list_1 (err)); | |
2504 | } | |
2505 | } | |
2506 | ||
2507 | static void | |
2508 | close_iconv_descriptors (scm_t_iconv_descriptors *id) | |
2509 | { | |
2510 | if (id->input_cd != (iconv_t) -1) | |
2511 | iconv_close (id->input_cd); | |
2512 | if (id->output_cd != (iconv_t) -1) | |
2513 | iconv_close (id->output_cd); | |
2514 | id->input_cd = (void *) -1; | |
2515 | id->output_cd = (void *) -1; | |
2516 | } | |
2517 | ||
cdd3d6c9 MW |
2518 | /* Return the iconv_descriptors, initializing them if necessary. MODE |
2519 | must be either SCM_PORT_READ or SCM_PORT_WRITE, and specifies which | |
2520 | operation is about to be done. We deliberately avoid reading from | |
2521 | the port unless the user was about to do so. */ | |
337edc59 | 2522 | scm_t_iconv_descriptors * |
cdd3d6c9 | 2523 | scm_i_port_iconv_descriptors (SCM port, scm_t_port_rw_active mode) |
337edc59 | 2524 | { |
cdd3d6c9 | 2525 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
337edc59 MW |
2526 | |
2527 | assert (pti->encoding_mode == SCM_PORT_ENCODING_MODE_ICONV); | |
2528 | ||
2529 | if (!pti->iconv_descriptors) | |
2530 | { | |
cdd3d6c9 MW |
2531 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
2532 | const char *precise_encoding; | |
2533 | ||
337edc59 MW |
2534 | if (!pt->encoding) |
2535 | pt->encoding = "ISO-8859-1"; | |
cdd3d6c9 MW |
2536 | |
2537 | /* If the specified encoding is UTF-16 or UTF-32, then make | |
2538 | that more precise by deciding what byte order to use. */ | |
96965a6e | 2539 | if (c_strcasecmp (pt->encoding, "UTF-16") == 0) |
cdd3d6c9 | 2540 | precise_encoding = decide_utf16_encoding (port, mode); |
96965a6e | 2541 | else if (c_strcasecmp (pt->encoding, "UTF-32") == 0) |
cdd3d6c9 MW |
2542 | precise_encoding = decide_utf32_encoding (port, mode); |
2543 | else | |
2544 | precise_encoding = pt->encoding; | |
2545 | ||
337edc59 | 2546 | pti->iconv_descriptors = |
cdd3d6c9 | 2547 | open_iconv_descriptors (precise_encoding, |
337edc59 MW |
2548 | SCM_INPUT_PORT_P (port), |
2549 | SCM_OUTPUT_PORT_P (port)); | |
2550 | } | |
2551 | ||
2552 | return pti->iconv_descriptors; | |
2553 | } | |
2554 | ||
889975e5 | 2555 | void |
064c27c4 | 2556 | scm_i_set_port_encoding_x (SCM port, const char *encoding) |
889975e5 | 2557 | { |
889975e5 | 2558 | scm_t_port *pt; |
e4598559 | 2559 | scm_t_port_internal *pti; |
337edc59 | 2560 | scm_t_iconv_descriptors *prev; |
f4bc4e59 | 2561 | |
9d9c66ba LC |
2562 | /* Set the character encoding for this port. */ |
2563 | pt = SCM_PTAB_ENTRY (port); | |
e4598559 | 2564 | pti = SCM_PORT_GET_INTERNAL (port); |
337edc59 | 2565 | prev = pti->iconv_descriptors; |
f4bc4e59 | 2566 | |
cdd3d6c9 MW |
2567 | /* In order to handle cases where the encoding changes mid-stream |
2568 | (e.g. within an HTTP stream, or within a file that is composed of | |
2569 | segments with different encodings), we consider this to be "stream | |
2570 | start" for purposes of BOM handling, regardless of our actual file | |
2571 | position. */ | |
2572 | pti->at_stream_start_for_bom_read = 1; | |
2573 | pti->at_stream_start_for_bom_write = 1; | |
2574 | ||
9d9c66ba LC |
2575 | if (encoding == NULL) |
2576 | encoding = "ISO-8859-1"; | |
f4bc4e59 | 2577 | |
7b292a9d LC |
2578 | /* If ENCODING is UTF-8, then no conversion descriptor is opened |
2579 | because we do I/O ourselves. This saves 100+ KiB for each | |
2580 | descriptor. */ | |
cdd3d6c9 | 2581 | pt->encoding = scm_gc_strdup (encoding, "port"); |
96965a6e | 2582 | if (c_strcasecmp (encoding, "UTF-8") == 0) |
cdd3d6c9 | 2583 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8; |
337edc59 | 2584 | else |
cdd3d6c9 | 2585 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV; |
f4bc4e59 | 2586 | |
cdd3d6c9 | 2587 | pti->iconv_descriptors = NULL; |
337edc59 MW |
2588 | if (prev) |
2589 | close_iconv_descriptors (prev); | |
889975e5 MG |
2590 | } |
2591 | ||
2592 | SCM_DEFINE (scm_port_encoding, "port-encoding", 1, 0, 0, | |
2593 | (SCM port), | |
2594 | "Returns, as a string, the character encoding that @var{port}\n" | |
2595 | "uses to interpret its input and output.\n") | |
2596 | #define FUNC_NAME s_scm_port_encoding | |
2597 | { | |
2598 | scm_t_port *pt; | |
2599 | const char *enc; | |
2600 | ||
2601 | SCM_VALIDATE_PORT (1, port); | |
2602 | ||
2603 | pt = SCM_PTAB_ENTRY (port); | |
9d9c66ba | 2604 | enc = pt->encoding; |
889975e5 MG |
2605 | if (enc) |
2606 | return scm_from_locale_string (pt->encoding); | |
2607 | else | |
0af34a3f | 2608 | return SCM_BOOL_F; |
889975e5 MG |
2609 | } |
2610 | #undef FUNC_NAME | |
d6a6989e | 2611 | |
889975e5 MG |
2612 | SCM_DEFINE (scm_set_port_encoding_x, "set-port-encoding!", 2, 0, 0, |
2613 | (SCM port, SCM enc), | |
2614 | "Sets the character encoding that will be used to interpret all\n" | |
2615 | "port I/O. New ports are created with the encoding\n" | |
2616 | "appropriate for the current locale if @code{setlocale} has \n" | |
2617 | "been called or ISO-8859-1 otherwise\n" | |
2618 | "and this procedure can be used to modify that encoding.\n") | |
889975e5 MG |
2619 | #define FUNC_NAME s_scm_set_port_encoding_x |
2620 | { | |
2621 | char *enc_str; | |
889975e5 MG |
2622 | |
2623 | SCM_VALIDATE_PORT (1, port); | |
2624 | SCM_VALIDATE_STRING (2, enc); | |
2625 | ||
2626 | enc_str = scm_to_locale_string (enc); | |
da288f50 | 2627 | scm_i_set_port_encoding_x (port, enc_str); |
3e05fc04 | 2628 | free (enc_str); |
da288f50 | 2629 | |
889975e5 MG |
2630 | return SCM_UNSPECIFIED; |
2631 | } | |
2632 | #undef FUNC_NAME | |
2633 | ||
2634 | ||
b22e94db LC |
2635 | /* A fluid specifying the default conversion handler for newly created |
2636 | ports. Its value should be one of the symbols below. */ | |
2637 | SCM_VARIABLE (default_conversion_strategy_var, | |
2638 | "%default-port-conversion-strategy"); | |
2639 | ||
2640 | /* Whether the above fluid is initialized. */ | |
889975e5 MG |
2641 | static int scm_conversion_strategy_init = 0; |
2642 | ||
b22e94db LC |
2643 | /* The possible conversion strategies. */ |
2644 | SCM_SYMBOL (sym_error, "error"); | |
2645 | SCM_SYMBOL (sym_substitute, "substitute"); | |
2646 | SCM_SYMBOL (sym_escape, "escape"); | |
2647 | ||
2648 | /* Return the default failed encoding conversion policy for new created | |
2649 | ports. */ | |
889975e5 | 2650 | scm_t_string_failed_conversion_handler |
b22e94db | 2651 | scm_i_default_port_conversion_handler (void) |
889975e5 | 2652 | { |
b22e94db LC |
2653 | scm_t_string_failed_conversion_handler handler; |
2654 | ||
2655 | if (!scm_conversion_strategy_init | |
2656 | || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var))) | |
2657 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
889975e5 MG |
2658 | else |
2659 | { | |
b22e94db LC |
2660 | SCM fluid, value; |
2661 | ||
2662 | fluid = SCM_VARIABLE_REF (default_conversion_strategy_var); | |
2663 | value = scm_fluid_ref (fluid); | |
2664 | ||
2665 | if (scm_is_eq (sym_substitute, value)) | |
2666 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
2667 | else if (scm_is_eq (sym_escape, value)) | |
2668 | handler = SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE; | |
2669 | else | |
2670 | /* Default to 'error also when the fluid's value is not one of | |
2671 | the valid symbols. */ | |
2672 | handler = SCM_FAILED_CONVERSION_ERROR; | |
889975e5 | 2673 | } |
b22e94db LC |
2674 | |
2675 | return handler; | |
889975e5 MG |
2676 | } |
2677 | ||
b22e94db | 2678 | /* Use HANDLER as the default conversion strategy for future ports. */ |
889975e5 | 2679 | void |
b22e94db LC |
2680 | scm_i_set_default_port_conversion_handler (scm_t_string_failed_conversion_handler |
2681 | handler) | |
889975e5 MG |
2682 | { |
2683 | SCM strategy; | |
b22e94db LC |
2684 | |
2685 | if (!scm_conversion_strategy_init | |
2686 | || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var))) | |
2687 | scm_misc_error (NULL, "tried to set conversion strategy fluid before it is initialized", | |
2688 | SCM_EOL); | |
2689 | ||
2690 | switch (handler) | |
889975e5 | 2691 | { |
b22e94db LC |
2692 | case SCM_FAILED_CONVERSION_ERROR: |
2693 | strategy = sym_error; | |
2694 | break; | |
2695 | ||
2696 | case SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE: | |
2697 | strategy = sym_escape; | |
2698 | break; | |
2699 | ||
2700 | case SCM_FAILED_CONVERSION_QUESTION_MARK: | |
2701 | strategy = sym_substitute; | |
2702 | break; | |
2703 | ||
2704 | default: | |
2705 | abort (); | |
889975e5 | 2706 | } |
b22e94db LC |
2707 | |
2708 | scm_fluid_set_x (SCM_VARIABLE_REF (default_conversion_strategy_var), | |
2709 | strategy); | |
889975e5 MG |
2710 | } |
2711 | ||
2712 | SCM_DEFINE (scm_port_conversion_strategy, "port-conversion-strategy", | |
2713 | 1, 0, 0, (SCM port), | |
2714 | "Returns the behavior of the port when handling a character that\n" | |
2715 | "is not representable in the port's current encoding.\n" | |
2716 | "It returns the symbol @code{error} if unrepresentable characters\n" | |
2717 | "should cause exceptions, @code{substitute} if the port should\n" | |
2718 | "try to replace unrepresentable characters with question marks or\n" | |
2719 | "approximate characters, or @code{escape} if unrepresentable\n" | |
2720 | "characters should be converted to string escapes.\n" | |
2721 | "\n" | |
2722 | "If @var{port} is @code{#f}, then the current default behavior\n" | |
2723 | "will be returned. New ports will have this default behavior\n" | |
2724 | "when they are created.\n") | |
2725 | #define FUNC_NAME s_scm_port_conversion_strategy | |
2726 | { | |
2727 | scm_t_string_failed_conversion_handler h; | |
2728 | ||
b22e94db LC |
2729 | if (scm_is_false (port)) |
2730 | h = scm_i_default_port_conversion_handler (); | |
2731 | else | |
889975e5 | 2732 | { |
b22e94db LC |
2733 | scm_t_port *pt; |
2734 | ||
889975e5 | 2735 | SCM_VALIDATE_OPPORT (1, port); |
b22e94db LC |
2736 | pt = SCM_PTAB_ENTRY (port); |
2737 | ||
2738 | h = pt->ilseq_handler; | |
889975e5 MG |
2739 | } |
2740 | ||
889975e5 | 2741 | if (h == SCM_FAILED_CONVERSION_ERROR) |
4a655e50 | 2742 | return scm_from_latin1_symbol ("error"); |
889975e5 | 2743 | else if (h == SCM_FAILED_CONVERSION_QUESTION_MARK) |
4a655e50 | 2744 | return scm_from_latin1_symbol ("substitute"); |
889975e5 | 2745 | else if (h == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE) |
4a655e50 | 2746 | return scm_from_latin1_symbol ("escape"); |
889975e5 MG |
2747 | else |
2748 | abort (); | |
2749 | ||
2750 | /* Never gets here. */ | |
2751 | return SCM_UNDEFINED; | |
2752 | } | |
2753 | #undef FUNC_NAME | |
2754 | ||
2755 | SCM_DEFINE (scm_set_port_conversion_strategy_x, "set-port-conversion-strategy!", | |
2756 | 2, 0, 0, | |
2757 | (SCM port, SCM sym), | |
2758 | "Sets the behavior of the interpreter when outputting a character\n" | |
2759 | "that is not representable in the port's current encoding.\n" | |
2760 | "@var{sym} can be either @code{'error}, @code{'substitute}, or\n" | |
2761 | "@code{'escape}. If it is @code{'error}, an error will be thrown\n" | |
2762 | "when an unconvertible character is encountered. If it is\n" | |
2763 | "@code{'substitute}, then unconvertible characters will \n" | |
2764 | "be replaced with approximate characters, or with question marks\n" | |
2765 | "if no approximately correct character is available.\n" | |
2766 | "If it is @code{'escape},\n" | |
2767 | "it will appear as a hex escape when output.\n" | |
2768 | "\n" | |
2769 | "If @var{port} is an open port, the conversion error behavior\n" | |
2770 | "is set for that port. If it is @code{#f}, it is set as the\n" | |
2771 | "default behavior for any future ports that get created in\n" | |
2772 | "this thread.\n") | |
2773 | #define FUNC_NAME s_scm_set_port_conversion_strategy_x | |
2774 | { | |
b22e94db | 2775 | scm_t_string_failed_conversion_handler handler; |
889975e5 | 2776 | |
b22e94db LC |
2777 | if (scm_is_eq (sym, sym_error)) |
2778 | handler = SCM_FAILED_CONVERSION_ERROR; | |
2779 | else if (scm_is_eq (sym, sym_substitute)) | |
2780 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
2781 | else if (scm_is_eq (sym, sym_escape)) | |
2782 | handler = SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE; | |
2783 | else | |
2784 | SCM_MISC_ERROR ("unknown conversion strategy ~s", scm_list_1 (sym)); | |
889975e5 | 2785 | |
b22e94db LC |
2786 | if (scm_is_false (port)) |
2787 | scm_i_set_default_port_conversion_handler (handler); | |
2788 | else | |
889975e5 | 2789 | { |
b22e94db LC |
2790 | SCM_VALIDATE_OPPORT (1, port); |
2791 | SCM_PTAB_ENTRY (port)->ilseq_handler = handler; | |
889975e5 MG |
2792 | } |
2793 | ||
889975e5 MG |
2794 | return SCM_UNSPECIFIED; |
2795 | } | |
2796 | #undef FUNC_NAME | |
2797 | ||
2798 | ||
2799 | ||
f12733c9 MD |
2800 | void |
2801 | scm_print_port_mode (SCM exp, SCM port) | |
2802 | { | |
2803 | scm_puts (SCM_CLOSEDP (exp) | |
2804 | ? "closed: " | |
f9a64404 DH |
2805 | : (SCM_RDNG & SCM_CELL_WORD_0 (exp) |
2806 | ? (SCM_WRTNG & SCM_CELL_WORD_0 (exp) | |
f12733c9 MD |
2807 | ? "input-output: " |
2808 | : "input: ") | |
f9a64404 | 2809 | : (SCM_WRTNG & SCM_CELL_WORD_0 (exp) |
f12733c9 MD |
2810 | ? "output: " |
2811 | : "bogus: ")), | |
2812 | port); | |
2813 | } | |
1cc91f1b | 2814 | |
f12733c9 | 2815 | int |
e81d98ec | 2816 | scm_port_print (SCM exp, SCM port, scm_print_state *pstate SCM_UNUSED) |
0f2d19dd | 2817 | { |
f12733c9 MD |
2818 | char *type = SCM_PTOBNAME (SCM_PTOBNUM (exp)); |
2819 | if (!type) | |
2820 | type = "port"; | |
b7f3516f | 2821 | scm_puts ("#<", port); |
f12733c9 | 2822 | scm_print_port_mode (exp, port); |
b7f3516f TT |
2823 | scm_puts (type, port); |
2824 | scm_putc (' ', port); | |
0345e278 | 2825 | scm_uintprint (SCM_CELL_WORD_1 (exp), 16, port); |
b7f3516f | 2826 | scm_putc ('>', port); |
f12733c9 | 2827 | return 1; |
0f2d19dd JB |
2828 | } |
2829 | ||
0f2d19dd | 2830 | \f |
ee149d03 | 2831 | |
d68fee48 | 2832 | /* Void ports. */ |
0f2d19dd | 2833 | |
92c2555f | 2834 | scm_t_bits scm_tc16_void_port = 0; |
0f2d19dd | 2835 | |
e81d98ec | 2836 | static int fill_input_void_port (SCM port SCM_UNUSED) |
283a1a0e | 2837 | { |
70df8af6 | 2838 | return EOF; |
283a1a0e GH |
2839 | } |
2840 | ||
31703ab8 | 2841 | static void |
e81d98ec DH |
2842 | write_void_port (SCM port SCM_UNUSED, |
2843 | const void *data SCM_UNUSED, | |
2844 | size_t size SCM_UNUSED) | |
31703ab8 GH |
2845 | { |
2846 | } | |
2847 | ||
d617ee18 MV |
2848 | static SCM |
2849 | scm_i_void_port (long mode_bits) | |
0f2d19dd | 2850 | { |
9de87eea | 2851 | scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex); |
402788a9 | 2852 | { |
da220f27 HWN |
2853 | SCM answer = scm_new_port_table_entry (scm_tc16_void_port); |
2854 | scm_t_port * pt = SCM_PTAB_ENTRY(answer); | |
2855 | ||
402788a9 | 2856 | scm_port_non_buffer (pt); |
402788a9 HWN |
2857 | |
2858 | SCM_SETSTREAM (answer, 0); | |
2859 | SCM_SET_CELL_TYPE (answer, scm_tc16_void_port | mode_bits); | |
9de87eea | 2860 | scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex); |
402788a9 HWN |
2861 | return answer; |
2862 | } | |
0f2d19dd JB |
2863 | } |
2864 | ||
d617ee18 MV |
2865 | SCM |
2866 | scm_void_port (char *mode_str) | |
2867 | { | |
2868 | return scm_i_void_port (scm_mode_bits (mode_str)); | |
2869 | } | |
2870 | ||
a1ec6916 | 2871 | SCM_DEFINE (scm_sys_make_void_port, "%make-void-port", 1, 0, 0, |
1bbd0b84 | 2872 | (SCM mode), |
70df8af6 | 2873 | "Create and return a new void port. A void port acts like\n" |
bb2c02f2 | 2874 | "@file{/dev/null}. The @var{mode} argument\n" |
70df8af6 | 2875 | "specifies the input/output modes for this port: see the\n" |
b380b885 | 2876 | "documentation for @code{open-file} in @ref{File Ports}.") |
1bbd0b84 | 2877 | #define FUNC_NAME s_scm_sys_make_void_port |
0f2d19dd | 2878 | { |
d617ee18 | 2879 | return scm_i_void_port (scm_i_mode_bits (mode)); |
0f2d19dd | 2880 | } |
1bbd0b84 | 2881 | #undef FUNC_NAME |
0f2d19dd | 2882 | |
0f2d19dd | 2883 | \f |
89545eba | 2884 | /* Initialization. */ |
1cc91f1b | 2885 | |
0f2d19dd JB |
2886 | void |
2887 | scm_init_ports () | |
0f2d19dd | 2888 | { |
840ae05d | 2889 | /* lseek() symbols. */ |
e11e83f3 MV |
2890 | scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET)); |
2891 | scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR)); | |
2892 | scm_c_define ("SEEK_END", scm_from_int (SEEK_END)); | |
840ae05d | 2893 | |
70df8af6 GH |
2894 | scm_tc16_void_port = scm_make_port_type ("void", fill_input_void_port, |
2895 | write_void_port); | |
9de87eea | 2896 | |
f39448c5 AW |
2897 | cur_inport_fluid = scm_make_fluid (); |
2898 | cur_outport_fluid = scm_make_fluid (); | |
2899 | cur_errport_fluid = scm_make_fluid (); | |
2900 | cur_loadport_fluid = scm_make_fluid (); | |
9de87eea | 2901 | |
f39448c5 | 2902 | scm_i_port_weak_hash = scm_make_weak_key_hash_table (SCM_I_MAKINUM(31)); |
d6a6989e | 2903 | |
a0599745 | 2904 | #include "libguile/ports.x" |
889975e5 | 2905 | |
d6a6989e | 2906 | /* Use Latin-1 as the default port encoding. */ |
c81c2ad3 AW |
2907 | SCM_VARIABLE_SET (default_port_encoding_var, |
2908 | scm_make_fluid_with_default (SCM_BOOL_F)); | |
889975e5 | 2909 | scm_port_encoding_init = 1; |
d6a6989e | 2910 | |
b22e94db LC |
2911 | SCM_VARIABLE_SET (default_conversion_strategy_var, |
2912 | scm_make_fluid_with_default (sym_substitute)); | |
889975e5 | 2913 | scm_conversion_strategy_init = 1; |
b22e94db | 2914 | |
9670f238 AW |
2915 | /* These bindings are used when boot-9 turns `current-input-port' et |
2916 | al into parameters. They are then removed from the guile module. */ | |
2917 | scm_c_define ("%current-input-port-fluid", cur_inport_fluid); | |
2918 | scm_c_define ("%current-output-port-fluid", cur_outport_fluid); | |
2919 | scm_c_define ("%current-error-port-fluid", cur_errport_fluid); | |
0f2d19dd | 2920 | } |
89e00824 ML |
2921 | |
2922 | /* | |
2923 | Local Variables: | |
2924 | c-file-style: "gnu" | |
2925 | End: | |
2926 | */ |