Commit | Line | Data |
---|---|---|
f4bc4e59 | 1 | /* Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, |
93c4fa21 | 2 | * 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013 Free Software Foundation, Inc. |
f4bc4e59 | 3 | * |
73be1d9e | 4 | * This library is free software; you can redistribute it and/or |
53befeb7 NJ |
5 | * modify it under the terms of the GNU Lesser General Public License |
6 | * as published by the Free Software Foundation; either version 3 of | |
7 | * the License, or (at your option) any later version. | |
0f2d19dd | 8 | * |
53befeb7 NJ |
9 | * This library is distributed in the hope that it will be useful, but |
10 | * WITHOUT ANY WARRANTY; without even the implied warranty of | |
73be1d9e MV |
11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
12 | * Lesser General Public License for more details. | |
0f2d19dd | 13 | * |
73be1d9e MV |
14 | * You should have received a copy of the GNU Lesser General Public |
15 | * License along with this library; if not, write to the Free Software | |
53befeb7 NJ |
16 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA |
17 | * 02110-1301 USA | |
73be1d9e | 18 | */ |
1bbd0b84 | 19 | |
1bbd0b84 | 20 | |
0f2d19dd | 21 | \f |
d68fee48 JB |
22 | /* Headers. */ |
23 | ||
2b829bbb KR |
24 | #define _LARGEFILE64_SOURCE /* ask for stat64 etc */ |
25 | ||
dbb605f5 | 26 | #ifdef HAVE_CONFIG_H |
bd515f37 RB |
27 | # include <config.h> |
28 | #endif | |
29 | ||
0f2d19dd | 30 | #include <stdio.h> |
e6e2e95a | 31 | #include <errno.h> |
8ab3d8a0 | 32 | #include <fcntl.h> /* for chsize on mingw */ |
b5cb4464 | 33 | #include <assert.h> |
f4bc4e59 | 34 | #include <iconv.h> |
889975e5 MG |
35 | #include <uniconv.h> |
36 | #include <unistr.h> | |
37 | #include <striconveh.h> | |
e6e2e95a | 38 | |
fca43887 LC |
39 | #include <assert.h> |
40 | ||
a0599745 | 41 | #include "libguile/_scm.h" |
4e047c3e | 42 | #include "libguile/async.h" |
8269ba5b | 43 | #include "libguile/deprecation.h" |
f0942910 | 44 | #include "libguile/eval.h" |
8ab3d8a0 | 45 | #include "libguile/fports.h" /* direct access for seek and truncate */ |
9511876f | 46 | #include "libguile/goops.h" |
a0599745 MD |
47 | #include "libguile/smob.h" |
48 | #include "libguile/chars.h" | |
185e369a | 49 | #include "libguile/dynwind.h" |
0f2d19dd | 50 | |
a0599745 | 51 | #include "libguile/keywords.h" |
5dbc6c06 | 52 | #include "libguile/hashtab.h" |
a0599745 MD |
53 | #include "libguile/root.h" |
54 | #include "libguile/strings.h" | |
b42170a4 | 55 | #include "libguile/mallocs.h" |
a0599745 MD |
56 | #include "libguile/validate.h" |
57 | #include "libguile/ports.h" | |
e4598559 | 58 | #include "libguile/ports-internal.h" |
3a5fb14d | 59 | #include "libguile/vectors.h" |
2721f918 | 60 | #include "libguile/weak-set.h" |
9de87eea | 61 | #include "libguile/fluids.h" |
889975e5 | 62 | #include "libguile/eq.h" |
a38024ba | 63 | #include "libguile/alist.h" |
0f2d19dd | 64 | |
bd9e24b3 GH |
65 | #ifdef HAVE_STRING_H |
66 | #include <string.h> | |
67 | #endif | |
68 | ||
ec65f5da MV |
69 | #ifdef HAVE_IO_H |
70 | #include <io.h> | |
71 | #endif | |
72 | ||
0f2d19dd JB |
73 | #ifdef HAVE_UNISTD_H |
74 | #include <unistd.h> | |
75 | #endif | |
76 | ||
95b88819 GH |
77 | #ifdef HAVE_SYS_IOCTL_H |
78 | #include <sys/ioctl.h> | |
79 | #endif | |
d68fee48 | 80 | |
8ab3d8a0 KR |
81 | /* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize |
82 | already, but have this code here in case that wasn't so in past versions, | |
83 | or perhaps to help other minimal DOS environments. | |
84 | ||
85 | gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which | |
86 | might be possibilities if we've got other systems without ftruncate. */ | |
87 | ||
56a3dcd4 | 88 | #if defined HAVE_CHSIZE && ! defined HAVE_FTRUNCATE |
82893676 | 89 | #define ftruncate(fd, size) chsize (fd, size) |
8ab3d8a0 KR |
90 | #undef HAVE_FTRUNCATE |
91 | #define HAVE_FTRUNCATE 1 | |
82893676 MG |
92 | #endif |
93 | ||
0f2d19dd | 94 | \f |
93c4fa21 AW |
95 | /* Port encodings are case-insensitive ASCII strings. */ |
96 | static char | |
97 | ascii_toupper (char c) | |
98 | { | |
99 | return (c < 'a' || c > 'z') ? c : ('A' + (c - 'a')); | |
100 | } | |
101 | ||
102 | /* It is only necessary to use this function on encodings that come from | |
103 | the user and have not been canonicalized yet. Encodings that are set | |
104 | on ports or in the default encoding fluid are in upper-case, and can | |
105 | be compared with strcmp. */ | |
106 | static int | |
107 | encoding_matches (const char *enc, const char *upper) | |
108 | { | |
109 | if (!enc) | |
110 | enc = "ISO-8859-1"; | |
111 | ||
112 | while (*enc) | |
113 | if (ascii_toupper (*enc++) != *upper++) | |
114 | return 0; | |
115 | ||
116 | return !*upper; | |
117 | } | |
118 | ||
119 | static char* | |
120 | canonicalize_encoding (const char *enc) | |
121 | { | |
122 | char *ret; | |
123 | int i; | |
124 | ||
125 | if (!enc) | |
126 | return "ISO-8859-1"; | |
127 | ||
128 | ret = scm_gc_strdup (enc, "port"); | |
129 | ||
130 | for (i = 0; ret[i]; i++) | |
131 | { | |
132 | if (ret[i] > 127) | |
133 | /* Restrict to ASCII. */ | |
134 | scm_misc_error (NULL, "invalid character encoding ~s", | |
135 | scm_list_1 (scm_from_latin1_string (enc))); | |
136 | else | |
137 | ret[i] = ascii_toupper (ret[i]); | |
138 | } | |
139 | ||
140 | return ret; | |
141 | } | |
142 | ||
143 | ||
144 | \f | |
d68fee48 | 145 | /* The port kind table --- a dynamically resized array of port types. */ |
0f2d19dd JB |
146 | |
147 | ||
148 | /* scm_ptobs scm_numptob | |
5dbc6c06 | 149 | * implement a dynamically resized array of ptob records. |
0f2d19dd JB |
150 | * Indexes into this table are used when generating type |
151 | * tags for smobjects (if you know a tag you can get an index and conversely). | |
152 | */ | |
62bd5d66 AW |
153 | static scm_t_ptob_descriptor **scm_ptobs = NULL; |
154 | static long scm_numptob = 0; /* Number of port types. */ | |
155 | static long scm_ptobs_size = 0; /* Number of slots in the port type | |
156 | table. */ | |
157 | static scm_i_pthread_mutex_t scm_ptobs_lock = SCM_I_PTHREAD_MUTEX_INITIALIZER; | |
158 | ||
159 | long | |
160 | scm_c_num_port_types (void) | |
161 | { | |
162 | long ret; | |
163 | ||
164 | scm_i_pthread_mutex_lock (&scm_ptobs_lock); | |
165 | ret = scm_numptob; | |
166 | scm_i_pthread_mutex_unlock (&scm_ptobs_lock); | |
167 | ||
168 | return ret; | |
169 | } | |
170 | ||
171 | scm_t_ptob_descriptor* | |
172 | scm_c_port_type_ref (long ptobnum) | |
173 | { | |
174 | scm_t_ptob_descriptor *ret = NULL; | |
175 | ||
176 | scm_i_pthread_mutex_lock (&scm_ptobs_lock); | |
177 | ||
178 | if (0 <= ptobnum && ptobnum < scm_numptob) | |
179 | ret = scm_ptobs[ptobnum]; | |
180 | ||
181 | scm_i_pthread_mutex_unlock (&scm_ptobs_lock); | |
182 | ||
183 | if (!ret) | |
184 | scm_out_of_range ("scm_c_port_type_ref", scm_from_long (ptobnum)); | |
185 | ||
186 | return ret; | |
187 | } | |
188 | ||
189 | long | |
190 | scm_c_port_type_add_x (scm_t_ptob_descriptor *desc) | |
191 | { | |
192 | long ret = -1; | |
193 | ||
194 | scm_i_pthread_mutex_lock (&scm_ptobs_lock); | |
195 | ||
196 | if (scm_numptob + 1 < SCM_I_MAX_PORT_TYPE_COUNT) | |
197 | { | |
198 | if (scm_numptob == scm_ptobs_size) | |
199 | { | |
200 | unsigned long old_size = scm_ptobs_size; | |
201 | scm_t_ptob_descriptor **old_ptobs = scm_ptobs; | |
202 | ||
203 | /* Currently there are only 9 predefined port types, so one | |
204 | resize will cover it. */ | |
205 | scm_ptobs_size = old_size + 10; | |
206 | ||
207 | if (scm_ptobs_size >= SCM_I_MAX_PORT_TYPE_COUNT) | |
208 | scm_ptobs_size = SCM_I_MAX_PORT_TYPE_COUNT; | |
209 | ||
210 | scm_ptobs = scm_gc_malloc (sizeof (*scm_ptobs) * scm_ptobs_size, | |
211 | "scm_ptobs"); | |
212 | ||
213 | memcpy (scm_ptobs, old_ptobs, sizeof (*scm_ptobs) * scm_numptob); | |
214 | } | |
215 | ||
216 | ret = scm_numptob++; | |
217 | scm_ptobs[ret] = desc; | |
218 | } | |
219 | ||
220 | scm_i_pthread_mutex_unlock (&scm_ptobs_lock); | |
221 | ||
222 | if (ret < 0) | |
223 | scm_out_of_range ("scm_c_port_type_add_x", scm_from_long (scm_numptob)); | |
224 | ||
225 | return ret; | |
226 | } | |
0f2d19dd | 227 | |
f12733c9 | 228 | /* |
f12733c9 | 229 | * We choose to use an interface similar to the smob interface with |
affc96b5 | 230 | * fill_input and write as standard fields, passed to the port |
f12733c9 MD |
231 | * type constructor, and optional fields set by setters. |
232 | */ | |
233 | ||
70df8af6 | 234 | static void |
e81d98ec | 235 | flush_port_default (SCM port SCM_UNUSED) |
70df8af6 GH |
236 | { |
237 | } | |
238 | ||
239 | static void | |
e81d98ec | 240 | end_input_default (SCM port SCM_UNUSED, int offset SCM_UNUSED) |
70df8af6 GH |
241 | { |
242 | } | |
0f2d19dd | 243 | |
92c2555f | 244 | scm_t_bits |
f12733c9 | 245 | scm_make_port_type (char *name, |
affc96b5 | 246 | int (*fill_input) (SCM port), |
8aa011a1 | 247 | void (*write) (SCM port, const void *data, size_t size)) |
0f2d19dd | 248 | { |
62bd5d66 AW |
249 | scm_t_ptob_descriptor *desc; |
250 | long ptobnum; | |
affc96b5 | 251 | |
62bd5d66 AW |
252 | desc = scm_gc_malloc_pointerless (sizeof (*desc), "port-type"); |
253 | memset (desc, 0, sizeof (*desc)); | |
affc96b5 | 254 | |
62bd5d66 AW |
255 | desc->name = name; |
256 | desc->print = scm_port_print; | |
257 | desc->write = write; | |
258 | desc->flush = flush_port_default; | |
259 | desc->end_input = end_input_default; | |
260 | desc->fill_input = fill_input; | |
affc96b5 | 261 | |
62bd5d66 | 262 | ptobnum = scm_c_port_type_add_x (desc); |
affc96b5 | 263 | |
62bd5d66 | 264 | /* Make a class object if GOOPS is present. */ |
63385df2 | 265 | if (SCM_UNPACK (scm_port_class[0]) != 0) |
62bd5d66 AW |
266 | scm_make_port_classes (ptobnum, name); |
267 | ||
268 | return scm_tc7_port + ptobnum * 256; | |
0f2d19dd JB |
269 | } |
270 | ||
f12733c9 | 271 | void |
23f2b9a3 | 272 | scm_set_port_mark (scm_t_bits tc, SCM (*mark) (SCM)) |
f12733c9 | 273 | { |
62bd5d66 | 274 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->mark = mark; |
f12733c9 MD |
275 | } |
276 | ||
277 | void | |
23f2b9a3 | 278 | scm_set_port_free (scm_t_bits tc, size_t (*free) (SCM)) |
f12733c9 | 279 | { |
62bd5d66 | 280 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->free = free; |
f12733c9 MD |
281 | } |
282 | ||
283 | void | |
23f2b9a3 | 284 | scm_set_port_print (scm_t_bits tc, int (*print) (SCM exp, SCM port, |
19b8d12b | 285 | scm_print_state *pstate)) |
f12733c9 | 286 | { |
62bd5d66 | 287 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->print = print; |
f12733c9 MD |
288 | } |
289 | ||
290 | void | |
23f2b9a3 | 291 | scm_set_port_equalp (scm_t_bits tc, SCM (*equalp) (SCM, SCM)) |
f12733c9 | 292 | { |
62bd5d66 | 293 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->equalp = equalp; |
f12733c9 MD |
294 | } |
295 | ||
31703ab8 | 296 | void |
19b8d12b | 297 | scm_set_port_close (scm_t_bits tc, int (*close) (SCM)) |
31703ab8 | 298 | { |
19b8d12b | 299 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->close = close; |
31703ab8 GH |
300 | } |
301 | ||
f12733c9 | 302 | void |
19b8d12b | 303 | scm_set_port_flush (scm_t_bits tc, void (*flush) (SCM port)) |
f12733c9 | 304 | { |
03a2eeb0 AW |
305 | scm_t_ptob_descriptor *ptob = scm_c_port_type_ref (SCM_TC2PTOBNUM (tc)); |
306 | ptob->flush = flush; | |
307 | ptob->flags |= SCM_PORT_TYPE_HAS_FLUSH; | |
f12733c9 MD |
308 | } |
309 | ||
310 | void | |
19b8d12b | 311 | scm_set_port_end_input (scm_t_bits tc, void (*end_input) (SCM port, int offset)) |
f12733c9 | 312 | { |
19b8d12b | 313 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->end_input = end_input; |
f12733c9 MD |
314 | } |
315 | ||
316 | void | |
19b8d12b | 317 | scm_set_port_seek (scm_t_bits tc, scm_t_off (*seek) (SCM, scm_t_off, int)) |
f12733c9 | 318 | { |
62bd5d66 | 319 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->seek = seek; |
f12733c9 MD |
320 | } |
321 | ||
322 | void | |
f1ce9199 | 323 | scm_set_port_truncate (scm_t_bits tc, void (*truncate) (SCM, scm_t_off)) |
f12733c9 | 324 | { |
62bd5d66 | 325 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->truncate = truncate; |
f12733c9 MD |
326 | } |
327 | ||
328 | void | |
23f2b9a3 | 329 | scm_set_port_input_waiting (scm_t_bits tc, int (*input_waiting) (SCM)) |
f12733c9 | 330 | { |
62bd5d66 | 331 | scm_c_port_type_ref (SCM_TC2PTOBNUM (tc))->input_waiting = input_waiting; |
f12733c9 MD |
332 | } |
333 | ||
45c0878b MW |
334 | static void |
335 | scm_i_set_pending_eof (SCM port) | |
336 | { | |
337 | SCM_PORT_GET_INTERNAL (port)->pending_eof = 1; | |
338 | } | |
339 | ||
340 | static void | |
341 | scm_i_clear_pending_eof (SCM port) | |
342 | { | |
343 | SCM_PORT_GET_INTERNAL (port)->pending_eof = 0; | |
344 | } | |
345 | ||
a38024ba MW |
346 | SCM_DEFINE (scm_i_port_property, "%port-property", 2, 0, 0, |
347 | (SCM port, SCM key), | |
348 | "Return the property of @var{port} associated with @var{key}.") | |
349 | #define FUNC_NAME s_scm_i_port_property | |
05d7f762 | 350 | { |
79657fd3 MW |
351 | scm_i_pthread_mutex_t *lock; |
352 | SCM result; | |
353 | ||
a38024ba | 354 | SCM_VALIDATE_OPPORT (1, port); |
79657fd3 MW |
355 | scm_c_lock_port (port, &lock); |
356 | result = scm_assq_ref (SCM_PORT_GET_INTERNAL (port)->alist, key); | |
357 | if (lock) | |
358 | scm_i_pthread_mutex_unlock (lock); | |
359 | return result; | |
05d7f762 | 360 | } |
a38024ba | 361 | #undef FUNC_NAME |
05d7f762 | 362 | |
a38024ba MW |
363 | SCM_DEFINE (scm_i_set_port_property_x, "%set-port-property!", 3, 0, 0, |
364 | (SCM port, SCM key, SCM value), | |
365 | "Set the property of @var{port} associated with @var{key} to @var{value}.") | |
366 | #define FUNC_NAME s_scm_i_set_port_property_x | |
05d7f762 | 367 | { |
79657fd3 | 368 | scm_i_pthread_mutex_t *lock; |
a38024ba MW |
369 | scm_t_port_internal *pti; |
370 | ||
371 | SCM_VALIDATE_OPPORT (1, port); | |
79657fd3 | 372 | scm_c_lock_port (port, &lock); |
a38024ba MW |
373 | pti = SCM_PORT_GET_INTERNAL (port); |
374 | pti->alist = scm_assq_set_x (pti->alist, key, value); | |
79657fd3 MW |
375 | if (lock) |
376 | scm_i_pthread_mutex_unlock (lock); | |
a38024ba | 377 | return SCM_UNSPECIFIED; |
05d7f762 | 378 | } |
a38024ba | 379 | #undef FUNC_NAME |
05d7f762 | 380 | |
0f2d19dd | 381 | \f |
0f2d19dd | 382 | |
d68fee48 | 383 | /* Standard ports --- current input, output, error, and more(!). */ |
0f2d19dd | 384 | |
34297700 AW |
385 | static SCM cur_inport_fluid = SCM_BOOL_F; |
386 | static SCM cur_outport_fluid = SCM_BOOL_F; | |
387 | static SCM cur_errport_fluid = SCM_BOOL_F; | |
388 | static SCM cur_loadport_fluid = SCM_BOOL_F; | |
9de87eea | 389 | |
3b3b36dd | 390 | SCM_DEFINE (scm_current_input_port, "current-input-port", 0, 0, 0, |
e1546b65 MG |
391 | (), |
392 | "Return the current input port. This is the default port used\n" | |
393 | "by many input procedures. Initially, @code{current-input-port}\n" | |
394 | "returns the @dfn{standard input} in Unix and C terminology.") | |
1bbd0b84 | 395 | #define FUNC_NAME s_scm_current_input_port |
0f2d19dd | 396 | { |
34297700 | 397 | if (scm_is_true (cur_inport_fluid)) |
889975e5 MG |
398 | return scm_fluid_ref (cur_inport_fluid); |
399 | else | |
400 | return SCM_BOOL_F; | |
0f2d19dd | 401 | } |
1bbd0b84 | 402 | #undef FUNC_NAME |
0f2d19dd | 403 | |
3b3b36dd | 404 | SCM_DEFINE (scm_current_output_port, "current-output-port", 0, 0, 0, |
e1546b65 MG |
405 | (), |
406 | "Return the current output port. This is the default port used\n" | |
9401323e | 407 | "by many output procedures. Initially,\n" |
e1546b65 MG |
408 | "@code{current-output-port} returns the @dfn{standard output} in\n" |
409 | "Unix and C terminology.") | |
1bbd0b84 | 410 | #define FUNC_NAME s_scm_current_output_port |
0f2d19dd | 411 | { |
34297700 | 412 | if (scm_is_true (cur_outport_fluid)) |
889975e5 MG |
413 | return scm_fluid_ref (cur_outport_fluid); |
414 | else | |
415 | return SCM_BOOL_F; | |
0f2d19dd | 416 | } |
1bbd0b84 | 417 | #undef FUNC_NAME |
0f2d19dd | 418 | |
3b3b36dd | 419 | SCM_DEFINE (scm_current_error_port, "current-error-port", 0, 0, 0, |
1bbd0b84 | 420 | (), |
b380b885 MD |
421 | "Return the port to which errors and warnings should be sent (the\n" |
422 | "@dfn{standard error} in Unix and C terminology).") | |
1bbd0b84 | 423 | #define FUNC_NAME s_scm_current_error_port |
0f2d19dd | 424 | { |
34297700 | 425 | if (scm_is_true (cur_errport_fluid)) |
889975e5 MG |
426 | return scm_fluid_ref (cur_errport_fluid); |
427 | else | |
428 | return SCM_BOOL_F; | |
0f2d19dd | 429 | } |
1bbd0b84 | 430 | #undef FUNC_NAME |
0f2d19dd | 431 | |
3972de76 AW |
432 | SCM |
433 | scm_current_warning_port (void) | |
434 | { | |
f57ea23a MW |
435 | static SCM cwp_var = SCM_UNDEFINED; |
436 | static scm_i_pthread_mutex_t cwp_var_mutex | |
437 | = SCM_I_PTHREAD_MUTEX_INITIALIZER; | |
438 | ||
439 | scm_i_scm_pthread_mutex_lock (&cwp_var_mutex); | |
440 | if (SCM_UNBNDP (cwp_var)) | |
441 | cwp_var = scm_c_private_variable ("guile", "current-warning-port"); | |
442 | scm_i_pthread_mutex_unlock (&cwp_var_mutex); | |
3972de76 AW |
443 | |
444 | return scm_call_0 (scm_variable_ref (cwp_var)); | |
445 | } | |
446 | ||
3b3b36dd | 447 | SCM_DEFINE (scm_current_load_port, "current-load-port", 0, 0, 0, |
e1546b65 | 448 | (), |
b450f070 | 449 | "Return the current-load-port.\n" |
e1546b65 | 450 | "The load port is used internally by @code{primitive-load}.") |
1bbd0b84 | 451 | #define FUNC_NAME s_scm_current_load_port |
31614d8e | 452 | { |
9de87eea | 453 | return scm_fluid_ref (cur_loadport_fluid); |
31614d8e | 454 | } |
1bbd0b84 | 455 | #undef FUNC_NAME |
31614d8e | 456 | |
3b3b36dd | 457 | SCM_DEFINE (scm_set_current_input_port, "set-current-input-port", 1, 0, 0, |
1bbd0b84 | 458 | (SCM port), |
8f85c0c6 NJ |
459 | "@deffnx {Scheme Procedure} set-current-output-port port\n" |
460 | "@deffnx {Scheme Procedure} set-current-error-port port\n" | |
b380b885 MD |
461 | "Change the ports returned by @code{current-input-port},\n" |
462 | "@code{current-output-port} and @code{current-error-port}, respectively,\n" | |
463 | "so that they use the supplied @var{port} for input or output.") | |
1bbd0b84 | 464 | #define FUNC_NAME s_scm_set_current_input_port |
0f2d19dd | 465 | { |
9de87eea | 466 | SCM oinp = scm_fluid_ref (cur_inport_fluid); |
34d19ef6 | 467 | SCM_VALIDATE_OPINPORT (1, port); |
9de87eea | 468 | scm_fluid_set_x (cur_inport_fluid, port); |
0f2d19dd JB |
469 | return oinp; |
470 | } | |
1bbd0b84 | 471 | #undef FUNC_NAME |
0f2d19dd JB |
472 | |
473 | ||
3b3b36dd | 474 | SCM_DEFINE (scm_set_current_output_port, "set-current-output-port", 1, 0, 0, |
e1546b65 MG |
475 | (SCM port), |
476 | "Set the current default output port to @var{port}.") | |
1bbd0b84 | 477 | #define FUNC_NAME s_scm_set_current_output_port |
0f2d19dd | 478 | { |
9de87eea | 479 | SCM ooutp = scm_fluid_ref (cur_outport_fluid); |
78446828 | 480 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 481 | SCM_VALIDATE_OPOUTPORT (1, port); |
9de87eea | 482 | scm_fluid_set_x (cur_outport_fluid, port); |
0f2d19dd JB |
483 | return ooutp; |
484 | } | |
1bbd0b84 | 485 | #undef FUNC_NAME |
0f2d19dd JB |
486 | |
487 | ||
3b3b36dd | 488 | SCM_DEFINE (scm_set_current_error_port, "set-current-error-port", 1, 0, 0, |
e1546b65 MG |
489 | (SCM port), |
490 | "Set the current default error port to @var{port}.") | |
1bbd0b84 | 491 | #define FUNC_NAME s_scm_set_current_error_port |
0f2d19dd | 492 | { |
9de87eea | 493 | SCM oerrp = scm_fluid_ref (cur_errport_fluid); |
78446828 | 494 | port = SCM_COERCE_OUTPORT (port); |
34d19ef6 | 495 | SCM_VALIDATE_OPOUTPORT (1, port); |
9de87eea | 496 | scm_fluid_set_x (cur_errport_fluid, port); |
0f2d19dd JB |
497 | return oerrp; |
498 | } | |
1bbd0b84 | 499 | #undef FUNC_NAME |
0f2d19dd | 500 | |
3972de76 AW |
501 | |
502 | SCM | |
503 | scm_set_current_warning_port (SCM port) | |
504 | { | |
505 | static SCM cwp_var = SCM_BOOL_F; | |
506 | ||
507 | if (scm_is_false (cwp_var)) | |
508 | cwp_var = scm_c_private_lookup ("guile", "current-warning-port"); | |
509 | ||
510 | return scm_call_1 (scm_variable_ref (cwp_var), port); | |
511 | } | |
512 | ||
513 | ||
185e369a | 514 | void |
661ae7ab | 515 | scm_dynwind_current_input_port (SCM port) |
9de87eea | 516 | #define FUNC_NAME NULL |
185e369a | 517 | { |
9de87eea | 518 | SCM_VALIDATE_OPINPORT (1, port); |
661ae7ab | 519 | scm_dynwind_fluid (cur_inport_fluid, port); |
185e369a | 520 | } |
9de87eea | 521 | #undef FUNC_NAME |
185e369a MV |
522 | |
523 | void | |
661ae7ab | 524 | scm_dynwind_current_output_port (SCM port) |
9de87eea | 525 | #define FUNC_NAME NULL |
185e369a | 526 | { |
9de87eea MV |
527 | port = SCM_COERCE_OUTPORT (port); |
528 | SCM_VALIDATE_OPOUTPORT (1, port); | |
661ae7ab | 529 | scm_dynwind_fluid (cur_outport_fluid, port); |
185e369a | 530 | } |
9de87eea | 531 | #undef FUNC_NAME |
185e369a MV |
532 | |
533 | void | |
661ae7ab | 534 | scm_dynwind_current_error_port (SCM port) |
9de87eea MV |
535 | #define FUNC_NAME NULL |
536 | { | |
537 | port = SCM_COERCE_OUTPORT (port); | |
538 | SCM_VALIDATE_OPOUTPORT (1, port); | |
661ae7ab | 539 | scm_dynwind_fluid (cur_errport_fluid, port); |
9de87eea MV |
540 | } |
541 | #undef FUNC_NAME | |
542 | ||
543 | void | |
661ae7ab | 544 | scm_i_dynwind_current_load_port (SCM port) |
185e369a | 545 | { |
661ae7ab | 546 | scm_dynwind_fluid (cur_loadport_fluid, port); |
185e369a MV |
547 | } |
548 | ||
19b8d12b | 549 | |
0f2d19dd | 550 | \f |
0f2d19dd | 551 | |
19b8d12b AW |
552 | /* Retrieving a port's mode. */ |
553 | ||
554 | /* Return the flags that characterize a port based on the mode | |
555 | * string used to open a file for that port. | |
556 | * | |
557 | * See PORT FLAGS in scm.h | |
5dbc6c06 | 558 | */ |
19b8d12b AW |
559 | |
560 | static long | |
561 | scm_i_mode_bits_n (SCM modes) | |
562 | { | |
563 | return (SCM_OPN | |
564 | | (scm_i_string_contains_char (modes, 'r') | |
565 | || scm_i_string_contains_char (modes, '+') ? SCM_RDNG : 0) | |
566 | | (scm_i_string_contains_char (modes, 'w') | |
567 | || scm_i_string_contains_char (modes, 'a') | |
568 | || scm_i_string_contains_char (modes, '+') ? SCM_WRTNG : 0) | |
569 | | (scm_i_string_contains_char (modes, '0') ? SCM_BUF0 : 0) | |
570 | | (scm_i_string_contains_char (modes, 'l') ? SCM_BUFLINE : 0)); | |
571 | } | |
572 | ||
573 | long | |
574 | scm_mode_bits (char *modes) | |
575 | { | |
8ebd06c6 AW |
576 | /* Valid characters are rw+a0l. So, use latin1. */ |
577 | return scm_i_mode_bits (scm_from_latin1_string (modes)); | |
19b8d12b AW |
578 | } |
579 | ||
580 | long | |
581 | scm_i_mode_bits (SCM modes) | |
582 | { | |
583 | long bits; | |
584 | ||
585 | if (!scm_is_string (modes)) | |
586 | scm_wrong_type_arg_msg (NULL, 0, modes, "string"); | |
587 | ||
588 | bits = scm_i_mode_bits_n (modes); | |
589 | scm_remember_upto_here_1 (modes); | |
590 | return bits; | |
591 | } | |
592 | ||
593 | /* Return the mode flags from an open port. | |
594 | * Some modes such as "append" are only used when opening | |
595 | * a file and are not returned here. */ | |
596 | ||
597 | SCM_DEFINE (scm_port_mode, "port-mode", 1, 0, 0, | |
598 | (SCM port), | |
599 | "Return the port modes associated with the open port @var{port}.\n" | |
600 | "These will not necessarily be identical to the modes used when\n" | |
601 | "the port was opened, since modes such as \"append\" which are\n" | |
602 | "used only during port creation are not retained.") | |
603 | #define FUNC_NAME s_scm_port_mode | |
604 | { | |
605 | char modes[4]; | |
606 | modes[0] = '\0'; | |
607 | ||
608 | port = SCM_COERCE_OUTPORT (port); | |
609 | SCM_VALIDATE_OPPORT (1, port); | |
610 | if (SCM_CELL_WORD_0 (port) & SCM_RDNG) { | |
611 | if (SCM_CELL_WORD_0 (port) & SCM_WRTNG) | |
612 | strcpy (modes, "r+"); | |
613 | else | |
614 | strcpy (modes, "r"); | |
615 | } | |
616 | else if (SCM_CELL_WORD_0 (port) & SCM_WRTNG) | |
617 | strcpy (modes, "w"); | |
618 | if (SCM_CELL_WORD_0 (port) & SCM_BUF0) | |
619 | strcat (modes, "0"); | |
620 | ||
621 | return scm_from_latin1_string (modes); | |
622 | } | |
623 | #undef FUNC_NAME | |
624 | ||
625 | ||
626 | \f | |
627 | ||
628 | /* The port table --- a weak set of all ports. | |
629 | ||
630 | We need a global registry of ports to flush them all at exit, and to | |
631 | get all the ports matching a file descriptor. */ | |
2721f918 | 632 | SCM scm_i_port_weak_set; |
b9ad392e | 633 | |
19b8d12b | 634 | |
651a0735 | 635 | \f |
651a0735 | 636 | |
19b8d12b | 637 | /* Port finalization. */ |
1cc91f1b | 638 | |
5a771d5f AW |
639 | struct do_free_data |
640 | { | |
641 | scm_t_ptob_descriptor *ptob; | |
642 | SCM port; | |
643 | }; | |
644 | ||
645 | static SCM | |
646 | do_free (void *body_data) | |
647 | { | |
648 | struct do_free_data *data = body_data; | |
649 | ||
650 | /* `close' is for explicit `close-port' by user. `free' is for this | |
651 | purpose: ports collected by the GC. */ | |
652 | data->ptob->free (data->port); | |
653 | ||
654 | return SCM_BOOL_T; | |
655 | } | |
656 | ||
651a0735 LC |
657 | /* Finalize the object (a port) pointed to by PTR. */ |
658 | static void | |
6922d92f | 659 | finalize_port (void *ptr, void *data) |
651a0735 | 660 | { |
21041372 | 661 | SCM port = SCM_PACK_POINTER (ptr); |
651a0735 LC |
662 | |
663 | if (!SCM_PORTP (port)) | |
664 | abort (); | |
665 | ||
666 | if (SCM_OPENP (port)) | |
667 | { | |
3753e227 | 668 | struct do_free_data data; |
651a0735 | 669 | |
3753e227 | 670 | SCM_CLR_PORT_OPEN_FLAG (port); |
651a0735 | 671 | |
3753e227 AW |
672 | data.ptob = SCM_PORT_DESCRIPTOR (port); |
673 | data.port = port; | |
5a771d5f | 674 | |
3753e227 AW |
675 | scm_internal_catch (SCM_BOOL_T, do_free, &data, |
676 | scm_handle_by_message_noexit, NULL); | |
5a771d5f | 677 | |
3753e227 | 678 | scm_gc_ports_collected++; |
651a0735 LC |
679 | } |
680 | } | |
681 | ||
682 | ||
651a0735 LC |
683 | \f |
684 | ||
da220f27 | 685 | SCM |
2721f918 AW |
686 | scm_c_make_port_with_encoding (scm_t_bits tag, unsigned long mode_bits, |
687 | const char *encoding, | |
688 | scm_t_string_failed_conversion_handler handler, | |
689 | scm_t_bits stream) | |
0f2d19dd | 690 | { |
2721f918 AW |
691 | SCM ret; |
692 | scm_t_port *entry; | |
f6f4feb0 | 693 | scm_t_port_internal *pti; |
62bd5d66 | 694 | scm_t_ptob_descriptor *ptob; |
2721f918 | 695 | |
f6f4feb0 MW |
696 | entry = scm_gc_typed_calloc (scm_t_port); |
697 | pti = scm_gc_typed_calloc (scm_t_port_internal); | |
62bd5d66 AW |
698 | ptob = scm_c_port_type_ref (SCM_TC2PTOBNUM (tag)); |
699 | ||
700 | ret = scm_words (tag | mode_bits, 3); | |
701 | SCM_SET_CELL_WORD_1 (ret, (scm_t_bits) entry); | |
702 | SCM_SET_CELL_WORD_2 (ret, (scm_t_bits) ptob); | |
5f16b897 | 703 | |
92c0ebac AW |
704 | entry->lock = scm_gc_malloc_pointerless (sizeof (*entry->lock), "port lock"); |
705 | scm_i_pthread_mutex_init (entry->lock, scm_i_pthread_mutexattr_recursive); | |
30b126d2 | 706 | |
f6f4feb0 | 707 | entry->internal = pti; |
840ae05d | 708 | entry->file_name = SCM_BOOL_F; |
61e452ba | 709 | entry->rw_active = SCM_PORT_NEITHER; |
2721f918 AW |
710 | entry->port = ret; |
711 | entry->stream = stream; | |
93c4fa21 AW |
712 | |
713 | if (encoding_matches (encoding, "UTF-8")) | |
714 | { | |
f6f4feb0 | 715 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8; |
93c4fa21 AW |
716 | entry->encoding = "UTF-8"; |
717 | } | |
718 | else if (encoding_matches (encoding, "ISO-8859-1")) | |
719 | { | |
f6f4feb0 | 720 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_LATIN1; |
93c4fa21 AW |
721 | entry->encoding = "ISO-8859-1"; |
722 | } | |
6c98257f | 723 | else |
93c4fa21 | 724 | { |
f6f4feb0 | 725 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV; |
93c4fa21 AW |
726 | entry->encoding = canonicalize_encoding (encoding); |
727 | } | |
728 | ||
2721f918 | 729 | entry->ilseq_handler = handler; |
f6f4feb0 | 730 | pti->iconv_descriptors = NULL; |
0f2d19dd | 731 | |
f6f4feb0 MW |
732 | pti->at_stream_start_for_bom_read = 1; |
733 | pti->at_stream_start_for_bom_write = 1; | |
f4bc4e59 | 734 | |
f6f4feb0 MW |
735 | pti->pending_eof = 0; |
736 | pti->alist = SCM_EOL; | |
3e05fc04 | 737 | |
6978c673 AW |
738 | if (SCM_PORT_DESCRIPTOR (ret)->free) |
739 | scm_i_set_finalizer (SCM2PTR (ret), finalize_port, NULL); | |
740 | ||
03a2eeb0 AW |
741 | if (SCM_PORT_DESCRIPTOR (ret)->flags & SCM_PORT_TYPE_HAS_FLUSH) |
742 | scm_weak_set_add_x (scm_i_port_weak_set, ret); | |
651a0735 | 743 | |
2721f918 AW |
744 | return ret; |
745 | } | |
746 | ||
747 | SCM | |
748 | scm_c_make_port (scm_t_bits tag, unsigned long mode_bits, scm_t_bits stream) | |
749 | { | |
750 | return scm_c_make_port_with_encoding (tag, mode_bits, | |
751 | scm_i_default_port_encoding (), | |
0dd7c540 | 752 | scm_i_default_port_conversion_handler (), |
2721f918 AW |
753 | stream); |
754 | } | |
755 | ||
756 | SCM | |
757 | scm_new_port_table_entry (scm_t_bits tag) | |
758 | { | |
759 | return scm_c_make_port (tag, 0, 0); | |
0f2d19dd JB |
760 | } |
761 | ||
d68fee48 | 762 | \f |
d68fee48 | 763 | |
19b8d12b | 764 | /* Predicates. */ |
1cc91f1b | 765 | |
19b8d12b AW |
766 | SCM_DEFINE (scm_port_p, "port?", 1, 0, 0, |
767 | (SCM x), | |
768 | "Return a boolean indicating whether @var{x} is a port.\n" | |
769 | "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n" | |
770 | "@var{x}))}.") | |
771 | #define FUNC_NAME s_scm_port_p | |
0f2d19dd | 772 | { |
19b8d12b | 773 | return scm_from_bool (SCM_PORTP (x)); |
0f2d19dd | 774 | } |
19b8d12b | 775 | #undef FUNC_NAME |
0f2d19dd | 776 | |
19b8d12b AW |
777 | SCM_DEFINE (scm_input_port_p, "input-port?", 1, 0, 0, |
778 | (SCM x), | |
779 | "Return @code{#t} if @var{x} is an input port, otherwise return\n" | |
780 | "@code{#f}. Any object satisfying this predicate also satisfies\n" | |
781 | "@code{port?}.") | |
782 | #define FUNC_NAME s_scm_input_port_p | |
0f2d19dd | 783 | { |
19b8d12b | 784 | return scm_from_bool (SCM_INPUT_PORT_P (x)); |
0f2d19dd | 785 | } |
1bbd0b84 | 786 | #undef FUNC_NAME |
0f2d19dd | 787 | |
19b8d12b AW |
788 | SCM_DEFINE (scm_output_port_p, "output-port?", 1, 0, 0, |
789 | (SCM x), | |
790 | "Return @code{#t} if @var{x} is an output port, otherwise return\n" | |
791 | "@code{#f}. Any object satisfying this predicate also satisfies\n" | |
792 | "@code{port?}.") | |
793 | #define FUNC_NAME s_scm_output_port_p | |
0f2d19dd | 794 | { |
19b8d12b AW |
795 | x = SCM_COERCE_OUTPORT (x); |
796 | return scm_from_bool (SCM_OUTPUT_PORT_P (x)); | |
0f2d19dd | 797 | } |
1bbd0b84 | 798 | #undef FUNC_NAME |
0f2d19dd | 799 | |
19b8d12b AW |
800 | SCM_DEFINE (scm_port_closed_p, "port-closed?", 1, 0, 0, |
801 | (SCM port), | |
802 | "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n" | |
803 | "open.") | |
804 | #define FUNC_NAME s_scm_port_closed_p | |
d617ee18 | 805 | { |
19b8d12b AW |
806 | SCM_VALIDATE_PORT (1, port); |
807 | return scm_from_bool (!SCM_OPPORTP (port)); | |
d617ee18 | 808 | } |
19b8d12b | 809 | #undef FUNC_NAME |
eadd48de | 810 | |
19b8d12b AW |
811 | SCM_DEFINE (scm_eof_object_p, "eof-object?", 1, 0, 0, |
812 | (SCM x), | |
813 | "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n" | |
814 | "return @code{#f}.") | |
815 | #define FUNC_NAME s_scm_eof_object_p | |
eadd48de | 816 | { |
19b8d12b | 817 | return scm_from_bool (SCM_EOF_OBJECT_P (x)); |
eadd48de | 818 | } |
1bbd0b84 | 819 | #undef FUNC_NAME |
eadd48de GH |
820 | |
821 | ||
d68fee48 | 822 | \f |
19b8d12b | 823 | |
d68fee48 JB |
824 | /* Closing ports. */ |
825 | ||
03a2eeb0 AW |
826 | static void close_iconv_descriptors (scm_t_iconv_descriptors *id); |
827 | ||
0f2d19dd JB |
828 | /* scm_close_port |
829 | * Call the close operation on a port object. | |
eadd48de | 830 | * see also scm_close. |
0f2d19dd | 831 | */ |
3b3b36dd | 832 | SCM_DEFINE (scm_close_port, "close-port", 1, 0, 0, |
1bbd0b84 | 833 | (SCM port), |
1e6808ea MG |
834 | "Close the specified port object. Return @code{#t} if it\n" |
835 | "successfully closes a port or @code{#f} if it was already\n" | |
836 | "closed. An exception may be raised if an error occurs, for\n" | |
837 | "example when flushing buffered output. See also @ref{Ports and\n" | |
838 | "File Descriptors, close}, for a procedure which can close file\n" | |
839 | "descriptors.") | |
1bbd0b84 | 840 | #define FUNC_NAME s_scm_close_port |
0f2d19dd | 841 | { |
f6f4feb0 | 842 | scm_t_port_internal *pti; |
eadd48de GH |
843 | int rv; |
844 | ||
78446828 MV |
845 | port = SCM_COERCE_OUTPORT (port); |
846 | ||
7a754ca6 | 847 | SCM_VALIDATE_PORT (1, port); |
0f2d19dd | 848 | if (SCM_CLOSEDP (port)) |
eadd48de | 849 | return SCM_BOOL_F; |
03a2eeb0 | 850 | |
f6f4feb0 | 851 | pti = SCM_PORT_GET_INTERNAL (port); |
5a771d5f | 852 | SCM_CLR_PORT_OPEN_FLAG (port); |
03a2eeb0 AW |
853 | |
854 | if (SCM_PORT_DESCRIPTOR (port)->flags & SCM_PORT_TYPE_HAS_FLUSH) | |
855 | scm_weak_set_remove_x (scm_i_port_weak_set, port); | |
856 | ||
5a771d5f AW |
857 | if (SCM_PORT_DESCRIPTOR (port)->close) |
858 | /* Note! This may throw an exception. Anything after this point | |
859 | should be resilient to non-local exits. */ | |
860 | rv = SCM_PORT_DESCRIPTOR (port)->close (port); | |
861 | else | |
862 | rv = 0; | |
03a2eeb0 | 863 | |
f6f4feb0 | 864 | if (pti->iconv_descriptors) |
03a2eeb0 | 865 | { |
5a771d5f AW |
866 | /* If we don't get here, the iconv_descriptors finalizer will |
867 | clean up. */ | |
f6f4feb0 MW |
868 | close_iconv_descriptors (pti->iconv_descriptors); |
869 | pti->iconv_descriptors = NULL; | |
03a2eeb0 AW |
870 | } |
871 | ||
7888309b | 872 | return scm_from_bool (rv >= 0); |
7a754ca6 MD |
873 | } |
874 | #undef FUNC_NAME | |
875 | ||
876 | SCM_DEFINE (scm_close_input_port, "close-input-port", 1, 0, 0, | |
877 | (SCM port), | |
878 | "Close the specified input port object. The routine has no effect if\n" | |
879 | "the file has already been closed. An exception may be raised if an\n" | |
880 | "error occurs. The value returned is unspecified.\n\n" | |
881 | "See also @ref{Ports and File Descriptors, close}, for a procedure\n" | |
882 | "which can close file descriptors.") | |
883 | #define FUNC_NAME s_scm_close_input_port | |
884 | { | |
885 | SCM_VALIDATE_INPUT_PORT (1, port); | |
886 | scm_close_port (port); | |
887 | return SCM_UNSPECIFIED; | |
888 | } | |
889 | #undef FUNC_NAME | |
890 | ||
891 | SCM_DEFINE (scm_close_output_port, "close-output-port", 1, 0, 0, | |
892 | (SCM port), | |
893 | "Close the specified output port object. The routine has no effect if\n" | |
894 | "the file has already been closed. An exception may be raised if an\n" | |
895 | "error occurs. The value returned is unspecified.\n\n" | |
896 | "See also @ref{Ports and File Descriptors, close}, for a procedure\n" | |
897 | "which can close file descriptors.") | |
898 | #define FUNC_NAME s_scm_close_output_port | |
899 | { | |
900 | port = SCM_COERCE_OUTPORT (port); | |
901 | SCM_VALIDATE_OUTPUT_PORT (1, port); | |
902 | scm_close_port (port); | |
903 | return SCM_UNSPECIFIED; | |
0f2d19dd | 904 | } |
1bbd0b84 | 905 | #undef FUNC_NAME |
0f2d19dd | 906 | |
2721f918 | 907 | |
19b8d12b | 908 | \f |
2721f918 | 909 | |
19b8d12b AW |
910 | /* Encoding characters to byte streams, and decoding byte streams to |
911 | characters. */ | |
5dbc6c06 | 912 | |
19b8d12b AW |
913 | /* A fluid specifying the default encoding for newly created ports. If it is |
914 | a string, that is the encoding. If it is #f, it is in the "native" | |
915 | (Latin-1) encoding. */ | |
916 | SCM_VARIABLE (default_port_encoding_var, "%default-port-encoding"); | |
917 | ||
918 | static int scm_port_encoding_init = 0; | |
919 | ||
920 | /* Use ENCODING as the default encoding for future ports. */ | |
c536b4b3 | 921 | void |
19b8d12b | 922 | scm_i_set_default_port_encoding (const char *encoding) |
c2ca4493 | 923 | { |
19b8d12b AW |
924 | if (!scm_port_encoding_init |
925 | || !scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var))) | |
926 | scm_misc_error (NULL, "tried to set port encoding fluid before it is initialized", | |
927 | SCM_EOL); | |
fdfe6305 | 928 | |
449ca87b | 929 | if (encoding_matches (encoding, "ISO-8859-1")) |
19b8d12b AW |
930 | scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), SCM_BOOL_F); |
931 | else | |
93c4fa21 AW |
932 | scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), |
933 | scm_from_latin1_string (canonicalize_encoding (encoding))); | |
2721f918 | 934 | } |
3a5fb14d | 935 | |
93c4fa21 | 936 | /* Return the name of the default encoding for newly created ports. */ |
19b8d12b AW |
937 | const char * |
938 | scm_i_default_port_encoding (void) | |
2721f918 | 939 | { |
19b8d12b | 940 | if (!scm_port_encoding_init) |
93c4fa21 | 941 | return "ISO-8859-1"; |
19b8d12b | 942 | else if (!scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var))) |
93c4fa21 | 943 | return "ISO-8859-1"; |
19b8d12b AW |
944 | else |
945 | { | |
946 | SCM encoding; | |
947 | ||
948 | encoding = scm_fluid_ref (SCM_VARIABLE_REF (default_port_encoding_var)); | |
949 | if (!scm_is_string (encoding)) | |
93c4fa21 | 950 | return "ISO-8859-1"; |
19b8d12b AW |
951 | else |
952 | return scm_i_string_chars (encoding); | |
953 | } | |
c536b4b3 | 954 | } |
fdfe6305 | 955 | |
0dd7c540 AW |
956 | /* A fluid specifying the default conversion handler for newly created |
957 | ports. Its value should be one of the symbols below. */ | |
958 | SCM_VARIABLE (default_conversion_strategy_var, | |
959 | "%default-port-conversion-strategy"); | |
1b3daef0 | 960 | |
0dd7c540 AW |
961 | /* Whether the above fluid is initialized. */ |
962 | static int scm_conversion_strategy_init = 0; | |
c2ca4493 | 963 | |
0dd7c540 AW |
964 | /* The possible conversion strategies. */ |
965 | SCM_SYMBOL (sym_error, "error"); | |
966 | SCM_SYMBOL (sym_substitute, "substitute"); | |
967 | SCM_SYMBOL (sym_escape, "escape"); | |
c536b4b3 | 968 | |
0dd7c540 AW |
969 | /* Return the default failed encoding conversion policy for new created |
970 | ports. */ | |
971 | scm_t_string_failed_conversion_handler | |
972 | scm_i_default_port_conversion_handler (void) | |
973 | { | |
974 | scm_t_string_failed_conversion_handler handler; | |
d68fee48 | 975 | |
0dd7c540 AW |
976 | if (!scm_conversion_strategy_init |
977 | || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var))) | |
978 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
979 | else | |
980 | { | |
981 | SCM fluid, value; | |
0f2d19dd | 982 | |
0dd7c540 AW |
983 | fluid = SCM_VARIABLE_REF (default_conversion_strategy_var); |
984 | value = scm_fluid_ref (fluid); | |
0f2d19dd | 985 | |
0dd7c540 AW |
986 | if (scm_is_eq (sym_substitute, value)) |
987 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
988 | else if (scm_is_eq (sym_escape, value)) | |
989 | handler = SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE; | |
990 | else | |
991 | /* Default to 'error also when the fluid's value is not one of | |
992 | the valid symbols. */ | |
993 | handler = SCM_FAILED_CONVERSION_ERROR; | |
994 | } | |
eb5c0a2a | 995 | |
0dd7c540 | 996 | return handler; |
60d0643d | 997 | } |
0f2d19dd | 998 | |
0dd7c540 AW |
999 | /* Use HANDLER as the default conversion strategy for future ports. */ |
1000 | void | |
1001 | scm_i_set_default_port_conversion_handler (scm_t_string_failed_conversion_handler | |
1002 | handler) | |
0f2d19dd | 1003 | { |
0dd7c540 | 1004 | SCM strategy; |
0f2d19dd | 1005 | |
0dd7c540 AW |
1006 | if (!scm_conversion_strategy_init |
1007 | || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var))) | |
1008 | scm_misc_error (NULL, "tried to set conversion strategy fluid before it is initialized", | |
1009 | SCM_EOL); | |
1010 | ||
1011 | switch (handler) | |
78446828 | 1012 | { |
0dd7c540 AW |
1013 | case SCM_FAILED_CONVERSION_ERROR: |
1014 | strategy = sym_error; | |
1015 | break; | |
1016 | ||
1017 | case SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE: | |
1018 | strategy = sym_escape; | |
1019 | break; | |
1020 | ||
1021 | case SCM_FAILED_CONVERSION_QUESTION_MARK: | |
1022 | strategy = sym_substitute; | |
1023 | break; | |
1024 | ||
1025 | default: | |
1026 | abort (); | |
78446828 | 1027 | } |
0f2d19dd | 1028 | |
0dd7c540 AW |
1029 | scm_fluid_set_x (SCM_VARIABLE_REF (default_conversion_strategy_var), |
1030 | strategy); | |
1031 | } | |
5dbc6c06 | 1032 | |
f6f4feb0 MW |
1033 | static void |
1034 | scm_i_unget_bytes_unlocked (const unsigned char *buf, size_t len, SCM port); | |
1035 | ||
1036 | /* If the next LEN bytes from PORT are equal to those in BYTES, then | |
1037 | return 1, else return 0. Leave the port position unchanged. */ | |
1038 | static int | |
1039 | looking_at_bytes (SCM port, const unsigned char *bytes, int len) | |
0f2d19dd | 1040 | { |
f6f4feb0 MW |
1041 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
1042 | int i = 0; | |
1043 | ||
1044 | while (i < len && scm_peek_byte_or_eof_unlocked (port) == bytes[i]) | |
78446828 | 1045 | { |
f6f4feb0 MW |
1046 | pt->read_pos++; |
1047 | i++; | |
78446828 | 1048 | } |
f6f4feb0 MW |
1049 | scm_i_unget_bytes_unlocked (bytes, i, port); |
1050 | return (i == len); | |
0f2d19dd JB |
1051 | } |
1052 | ||
f6f4feb0 MW |
1053 | static const unsigned char scm_utf8_bom[3] = {0xEF, 0xBB, 0xBF}; |
1054 | static const unsigned char scm_utf16be_bom[2] = {0xFE, 0xFF}; | |
1055 | static const unsigned char scm_utf16le_bom[2] = {0xFF, 0xFE}; | |
1056 | static const unsigned char scm_utf32be_bom[4] = {0x00, 0x00, 0xFE, 0xFF}; | |
1057 | static const unsigned char scm_utf32le_bom[4] = {0xFF, 0xFE, 0x00, 0x00}; | |
5dbc6c06 | 1058 | |
f6f4feb0 MW |
1059 | /* Decide what byte order to use for a UTF-16 port. Return "UTF-16BE" |
1060 | or "UTF-16LE". MODE must be either SCM_PORT_READ or SCM_PORT_WRITE, | |
1061 | and specifies which operation is about to be done. The MODE | |
1062 | determines how we will decide the byte order. We deliberately avoid | |
1063 | reading from the port unless the user is about to do so. If the user | |
1064 | is about to read, then we look for a BOM, and if present, we use it | |
1065 | to determine the byte order. Otherwise we choose big endian, as | |
1066 | recommended by the Unicode Standard. Note that the BOM (if any) is | |
1067 | not consumed here. */ | |
1068 | static const char * | |
1069 | decide_utf16_encoding (SCM port, scm_t_port_rw_active mode) | |
5dbc6c06 | 1070 | { |
f6f4feb0 MW |
1071 | if (mode == SCM_PORT_READ |
1072 | && SCM_PORT_GET_INTERNAL (port)->at_stream_start_for_bom_read | |
1073 | && looking_at_bytes (port, scm_utf16le_bom, sizeof scm_utf16le_bom)) | |
1074 | return "UTF-16LE"; | |
1075 | else | |
1076 | return "UTF-16BE"; | |
5dbc6c06 HWN |
1077 | } |
1078 | ||
f6f4feb0 MW |
1079 | /* Decide what byte order to use for a UTF-32 port. Return "UTF-32BE" |
1080 | or "UTF-32LE". See the comment above 'decide_utf16_encoding' for | |
1081 | details. */ | |
1082 | static const char * | |
1083 | decide_utf32_encoding (SCM port, scm_t_port_rw_active mode) | |
89ea5b7c | 1084 | { |
f6f4feb0 MW |
1085 | if (mode == SCM_PORT_READ |
1086 | && SCM_PORT_GET_INTERNAL (port)->at_stream_start_for_bom_read | |
1087 | && looking_at_bytes (port, scm_utf32le_bom, sizeof scm_utf32le_bom)) | |
1088 | return "UTF-32LE"; | |
1089 | else | |
1090 | return "UTF-32BE"; | |
89ea5b7c | 1091 | } |
0f2d19dd | 1092 | |
6c98257f | 1093 | static void |
2aed2667 | 1094 | finalize_iconv_descriptors (void *ptr, void *data) |
c536b4b3 | 1095 | { |
6c98257f AW |
1096 | close_iconv_descriptors (ptr); |
1097 | } | |
c2ca4493 | 1098 | |
6c98257f AW |
1099 | static scm_t_iconv_descriptors * |
1100 | open_iconv_descriptors (const char *encoding, int reading, int writing) | |
1101 | { | |
1102 | scm_t_iconv_descriptors *id; | |
1103 | iconv_t input_cd, output_cd; | |
8ebd06c6 | 1104 | size_t i; |
c536b4b3 | 1105 | |
6c98257f AW |
1106 | input_cd = (iconv_t) -1; |
1107 | output_cd = (iconv_t) -1; | |
d68fee48 | 1108 | |
8ebd06c6 AW |
1109 | for (i = 0; encoding[i]; i++) |
1110 | if (encoding[i] > 127) | |
1111 | goto invalid_encoding; | |
1112 | ||
6c98257f AW |
1113 | if (reading) |
1114 | { | |
1115 | /* Open an input iconv conversion descriptor, from ENCODING | |
1116 | to UTF-8. We choose UTF-8, not UTF-32, because iconv | |
1117 | implementations can typically convert from anything to | |
1118 | UTF-8, but not to UTF-32 (see | |
1119 | <http://lists.gnu.org/archive/html/bug-libunistring/2010-09/msg00007.html>). */ | |
1120 | ||
1121 | /* Assume opening an iconv descriptor causes about 16 KB of | |
1122 | allocation. */ | |
1123 | scm_gc_register_allocation (16 * 1024); | |
1124 | ||
1125 | input_cd = iconv_open ("UTF-8", encoding); | |
1126 | if (input_cd == (iconv_t) -1) | |
1127 | goto invalid_encoding; | |
1128 | } | |
0f2d19dd | 1129 | |
6c98257f | 1130 | if (writing) |
19b8d12b | 1131 | { |
6c98257f AW |
1132 | /* Assume opening an iconv descriptor causes about 16 KB of |
1133 | allocation. */ | |
1134 | scm_gc_register_allocation (16 * 1024); | |
0f2d19dd | 1135 | |
6c98257f AW |
1136 | output_cd = iconv_open (encoding, "UTF-8"); |
1137 | if (output_cd == (iconv_t) -1) | |
1138 | { | |
1139 | if (input_cd != (iconv_t) -1) | |
1140 | iconv_close (input_cd); | |
1141 | goto invalid_encoding; | |
1142 | } | |
19b8d12b | 1143 | } |
eb5c0a2a | 1144 | |
6c98257f AW |
1145 | id = scm_gc_malloc_pointerless (sizeof (*id), "iconv descriptors"); |
1146 | id->input_cd = input_cd; | |
1147 | id->output_cd = output_cd; | |
1148 | ||
6978c673 AW |
1149 | /* Register a finalizer to close the descriptors. */ |
1150 | scm_i_set_finalizer (id, finalize_iconv_descriptors, NULL); | |
19b8d12b | 1151 | |
6c98257f | 1152 | return id; |
19b8d12b AW |
1153 | |
1154 | invalid_encoding: | |
1155 | { | |
1156 | SCM err; | |
8ebd06c6 | 1157 | err = scm_from_latin1_string (encoding); |
6c98257f | 1158 | scm_misc_error ("open_iconv_descriptors", |
19b8d12b AW |
1159 | "invalid or unknown character encoding ~s", |
1160 | scm_list_1 (err)); | |
1161 | } | |
0f2d19dd JB |
1162 | } |
1163 | ||
6c98257f AW |
1164 | static void |
1165 | close_iconv_descriptors (scm_t_iconv_descriptors *id) | |
1166 | { | |
1167 | if (id->input_cd != (iconv_t) -1) | |
1168 | iconv_close (id->input_cd); | |
1169 | if (id->output_cd != (iconv_t) -1) | |
1170 | iconv_close (id->output_cd); | |
1171 | id->input_cd = (void *) -1; | |
1172 | id->output_cd = (void *) -1; | |
1173 | } | |
1174 | ||
1175 | scm_t_iconv_descriptors * | |
f6f4feb0 | 1176 | scm_i_port_iconv_descriptors (SCM port, scm_t_port_rw_active mode) |
6c98257f | 1177 | { |
f6f4feb0 | 1178 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
6c98257f | 1179 | |
f6f4feb0 | 1180 | assert (pti->encoding_mode == SCM_PORT_ENCODING_MODE_ICONV); |
6c98257f | 1181 | |
f6f4feb0 | 1182 | if (!pti->iconv_descriptors) |
6c98257f | 1183 | { |
f6f4feb0 MW |
1184 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
1185 | const char *precise_encoding; | |
7b292a9d | 1186 | |
6c98257f AW |
1187 | if (!pt->encoding) |
1188 | pt->encoding = "ISO-8859-1"; | |
7be1705d | 1189 | |
f6f4feb0 MW |
1190 | /* If the specified encoding is UTF-16 or UTF-32, then make |
1191 | that more precise by deciding what byte order to use. */ | |
1192 | if (strcmp (pt->encoding, "UTF-16") == 0) | |
1193 | precise_encoding = decide_utf16_encoding (port, mode); | |
1194 | else if (strcmp (pt->encoding, "UTF-32") == 0) | |
1195 | precise_encoding = decide_utf32_encoding (port, mode); | |
1196 | else | |
1197 | precise_encoding = pt->encoding; | |
7b292a9d | 1198 | |
f6f4feb0 MW |
1199 | pti->iconv_descriptors = |
1200 | open_iconv_descriptors (precise_encoding, | |
6c98257f AW |
1201 | SCM_INPUT_PORT_P (port), |
1202 | SCM_OUTPUT_PORT_P (port)); | |
1203 | } | |
1204 | ||
f6f4feb0 | 1205 | return pti->iconv_descriptors; |
6c98257f AW |
1206 | } |
1207 | ||
8ebd06c6 | 1208 | /* The name of the encoding is itself encoded in ASCII. */ |
6c98257f AW |
1209 | void |
1210 | scm_i_set_port_encoding_x (SCM port, const char *encoding) | |
1211 | { | |
1212 | scm_t_port *pt; | |
f6f4feb0 | 1213 | scm_t_port_internal *pti; |
6c98257f AW |
1214 | scm_t_iconv_descriptors *prev; |
1215 | ||
1216 | /* Set the character encoding for this port. */ | |
1217 | pt = SCM_PTAB_ENTRY (port); | |
f6f4feb0 MW |
1218 | pti = SCM_PORT_GET_INTERNAL (port); |
1219 | prev = pti->iconv_descriptors; | |
7b292a9d | 1220 | |
f6f4feb0 MW |
1221 | /* In order to handle cases where the encoding changes mid-stream |
1222 | (e.g. within an HTTP stream, or within a file that is composed of | |
1223 | segments with different encodings), we consider this to be "stream | |
1224 | start" for purposes of BOM handling, regardless of our actual file | |
1225 | position. */ | |
1226 | pti->at_stream_start_for_bom_read = 1; | |
1227 | pti->at_stream_start_for_bom_write = 1; | |
6c98257f | 1228 | |
93c4fa21 | 1229 | if (encoding_matches (encoding, "UTF-8")) |
6c98257f AW |
1230 | { |
1231 | pt->encoding = "UTF-8"; | |
f6f4feb0 | 1232 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_UTF8; |
6c98257f | 1233 | } |
93c4fa21 | 1234 | else if (encoding_matches (encoding, "ISO-8859-1")) |
79eb47ea AW |
1235 | { |
1236 | pt->encoding = "ISO-8859-1"; | |
f6f4feb0 | 1237 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_LATIN1; |
79eb47ea | 1238 | } |
6c98257f AW |
1239 | else |
1240 | { | |
f6f4feb0 MW |
1241 | pt->encoding = canonicalize_encoding (encoding); |
1242 | pti->encoding_mode = SCM_PORT_ENCODING_MODE_ICONV; | |
6c98257f AW |
1243 | } |
1244 | ||
f6f4feb0 | 1245 | pti->iconv_descriptors = NULL; |
6c98257f AW |
1246 | if (prev) |
1247 | close_iconv_descriptors (prev); | |
1248 | } | |
1249 | ||
19b8d12b AW |
1250 | SCM_DEFINE (scm_port_encoding, "port-encoding", 1, 0, 0, |
1251 | (SCM port), | |
1252 | "Returns, as a string, the character encoding that @var{port}\n" | |
1253 | "uses to interpret its input and output.\n") | |
1254 | #define FUNC_NAME s_scm_port_encoding | |
0f2d19dd | 1255 | { |
19b8d12b AW |
1256 | SCM_VALIDATE_PORT (1, port); |
1257 | ||
93c4fa21 | 1258 | return scm_from_latin1_string (SCM_PTAB_ENTRY (port)->encoding); |
0f2d19dd | 1259 | } |
1bbd0b84 | 1260 | #undef FUNC_NAME |
0f2d19dd | 1261 | |
19b8d12b AW |
1262 | SCM_DEFINE (scm_set_port_encoding_x, "set-port-encoding!", 2, 0, 0, |
1263 | (SCM port, SCM enc), | |
1264 | "Sets the character encoding that will be used to interpret all\n" | |
1265 | "port I/O. New ports are created with the encoding\n" | |
1266 | "appropriate for the current locale if @code{setlocale} has \n" | |
1267 | "been called or ISO-8859-1 otherwise\n" | |
1268 | "and this procedure can be used to modify that encoding.\n") | |
1269 | #define FUNC_NAME s_scm_set_port_encoding_x | |
5dbc6c06 | 1270 | { |
19b8d12b | 1271 | char *enc_str; |
5dbc6c06 | 1272 | |
19b8d12b AW |
1273 | SCM_VALIDATE_PORT (1, port); |
1274 | SCM_VALIDATE_STRING (2, enc); | |
0f2d19dd | 1275 | |
8ebd06c6 | 1276 | enc_str = scm_to_latin1_string (enc); |
19b8d12b AW |
1277 | scm_i_set_port_encoding_x (port, enc_str); |
1278 | free (enc_str); | |
1279 | ||
1280 | return SCM_UNSPECIFIED; | |
0f2d19dd | 1281 | } |
1bbd0b84 | 1282 | #undef FUNC_NAME |
0f2d19dd | 1283 | |
19b8d12b AW |
1284 | SCM_DEFINE (scm_port_conversion_strategy, "port-conversion-strategy", |
1285 | 1, 0, 0, (SCM port), | |
1286 | "Returns the behavior of the port when handling a character that\n" | |
1287 | "is not representable in the port's current encoding.\n" | |
1288 | "It returns the symbol @code{error} if unrepresentable characters\n" | |
1289 | "should cause exceptions, @code{substitute} if the port should\n" | |
1290 | "try to replace unrepresentable characters with question marks or\n" | |
1291 | "approximate characters, or @code{escape} if unrepresentable\n" | |
1292 | "characters should be converted to string escapes.\n" | |
1293 | "\n" | |
1294 | "If @var{port} is @code{#f}, then the current default behavior\n" | |
1295 | "will be returned. New ports will have this default behavior\n" | |
1296 | "when they are created.\n") | |
1297 | #define FUNC_NAME s_scm_port_conversion_strategy | |
889975e5 | 1298 | { |
19b8d12b | 1299 | scm_t_string_failed_conversion_handler h; |
7b292a9d | 1300 | |
0dd7c540 AW |
1301 | if (scm_is_false (port)) |
1302 | h = scm_i_default_port_conversion_handler (); | |
1303 | else | |
7b292a9d | 1304 | { |
0dd7c540 AW |
1305 | scm_t_port *pt; |
1306 | ||
19b8d12b | 1307 | SCM_VALIDATE_OPPORT (1, port); |
0dd7c540 AW |
1308 | pt = SCM_PTAB_ENTRY (port); |
1309 | ||
1310 | h = pt->ilseq_handler; | |
7b292a9d LC |
1311 | } |
1312 | ||
19b8d12b AW |
1313 | if (h == SCM_FAILED_CONVERSION_ERROR) |
1314 | return scm_from_latin1_symbol ("error"); | |
1315 | else if (h == SCM_FAILED_CONVERSION_QUESTION_MARK) | |
1316 | return scm_from_latin1_symbol ("substitute"); | |
1317 | else if (h == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE) | |
1318 | return scm_from_latin1_symbol ("escape"); | |
1319 | else | |
1320 | abort (); | |
7b292a9d | 1321 | |
19b8d12b AW |
1322 | /* Never gets here. */ |
1323 | return SCM_UNDEFINED; | |
1324 | } | |
1325 | #undef FUNC_NAME | |
7b292a9d | 1326 | |
19b8d12b AW |
1327 | SCM_DEFINE (scm_set_port_conversion_strategy_x, "set-port-conversion-strategy!", |
1328 | 2, 0, 0, | |
1329 | (SCM port, SCM sym), | |
1330 | "Sets the behavior of the interpreter when outputting a character\n" | |
1331 | "that is not representable in the port's current encoding.\n" | |
1332 | "@var{sym} can be either @code{'error}, @code{'substitute}, or\n" | |
1333 | "@code{'escape}. If it is @code{'error}, an error will be thrown\n" | |
1334 | "when an unconvertible character is encountered. If it is\n" | |
1335 | "@code{'substitute}, then unconvertible characters will \n" | |
1336 | "be replaced with approximate characters, or with question marks\n" | |
1337 | "if no approximately correct character is available.\n" | |
1338 | "If it is @code{'escape},\n" | |
1339 | "it will appear as a hex escape when output.\n" | |
1340 | "\n" | |
1341 | "If @var{port} is an open port, the conversion error behavior\n" | |
1342 | "is set for that port. If it is @code{#f}, it is set as the\n" | |
1343 | "default behavior for any future ports that get created in\n" | |
1344 | "this thread.\n") | |
1345 | #define FUNC_NAME s_scm_set_port_conversion_strategy_x | |
1346 | { | |
0dd7c540 | 1347 | scm_t_string_failed_conversion_handler handler; |
7b292a9d | 1348 | |
0dd7c540 AW |
1349 | if (scm_is_eq (sym, sym_error)) |
1350 | handler = SCM_FAILED_CONVERSION_ERROR; | |
1351 | else if (scm_is_eq (sym, sym_substitute)) | |
1352 | handler = SCM_FAILED_CONVERSION_QUESTION_MARK; | |
1353 | else if (scm_is_eq (sym, sym_escape)) | |
1354 | handler = SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE; | |
1355 | else | |
1356 | SCM_MISC_ERROR ("unknown conversion strategy ~s", scm_list_1 (sym)); | |
7b292a9d | 1357 | |
0dd7c540 AW |
1358 | if (scm_is_false (port)) |
1359 | scm_i_set_default_port_conversion_handler (handler); | |
1360 | else | |
19b8d12b | 1361 | { |
0dd7c540 AW |
1362 | SCM_VALIDATE_OPPORT (1, port); |
1363 | SCM_PTAB_ENTRY (port)->ilseq_handler = handler; | |
19b8d12b | 1364 | } |
7b292a9d | 1365 | |
19b8d12b AW |
1366 | return SCM_UNSPECIFIED; |
1367 | } | |
1368 | #undef FUNC_NAME | |
7be1705d | 1369 | |
7b292a9d | 1370 | |
19b8d12b | 1371 | \f |
7b292a9d | 1372 | |
14dcb5cc AW |
1373 | /* The port lock. */ |
1374 | ||
1375 | static void | |
92c0ebac | 1376 | lock_port (void *mutex) |
14dcb5cc | 1377 | { |
789dd40b | 1378 | scm_i_pthread_mutex_lock ((scm_i_pthread_mutex_t *) mutex); |
14dcb5cc AW |
1379 | } |
1380 | ||
1381 | static void | |
92c0ebac | 1382 | unlock_port (void *mutex) |
14dcb5cc | 1383 | { |
789dd40b | 1384 | scm_i_pthread_mutex_unlock ((scm_i_pthread_mutex_t *) mutex); |
14dcb5cc AW |
1385 | } |
1386 | ||
1387 | void | |
1388 | scm_dynwind_lock_port (SCM port) | |
92c0ebac | 1389 | #define FUNC_NAME "dynwind-lock-port" |
14dcb5cc | 1390 | { |
92c0ebac AW |
1391 | scm_i_pthread_mutex_t *lock; |
1392 | SCM_VALIDATE_OPPORT (SCM_ARG1, port); | |
1393 | scm_c_lock_port (port, &lock); | |
1394 | if (lock) | |
1395 | { | |
1396 | scm_dynwind_unwind_handler (unlock_port, lock, SCM_F_WIND_EXPLICITLY); | |
1397 | scm_dynwind_rewind_handler (lock_port, lock, 0); | |
1398 | } | |
14dcb5cc | 1399 | } |
92c0ebac | 1400 | #undef FUNC_NAME |
14dcb5cc AW |
1401 | |
1402 | ||
1403 | \f | |
1404 | ||
19b8d12b | 1405 | /* Input. */ |
7be1705d | 1406 | |
0d959103 AW |
1407 | int |
1408 | scm_get_byte_or_eof (SCM port) | |
1409 | { | |
92c0ebac | 1410 | scm_i_pthread_mutex_t *lock; |
0d959103 AW |
1411 | int ret; |
1412 | ||
92c0ebac | 1413 | scm_c_lock_port (port, &lock); |
0d959103 | 1414 | ret = scm_get_byte_or_eof_unlocked (port); |
92c0ebac AW |
1415 | if (lock) |
1416 | scm_i_pthread_mutex_unlock (lock); | |
0d959103 AW |
1417 | |
1418 | return ret; | |
1419 | } | |
1420 | ||
1421 | int | |
1422 | scm_peek_byte_or_eof (SCM port) | |
1423 | { | |
92c0ebac | 1424 | scm_i_pthread_mutex_t *lock; |
0d959103 AW |
1425 | int ret; |
1426 | ||
92c0ebac | 1427 | scm_c_lock_port (port, &lock); |
0d959103 | 1428 | ret = scm_peek_byte_or_eof_unlocked (port); |
92c0ebac AW |
1429 | if (lock) |
1430 | scm_i_pthread_mutex_unlock (lock); | |
0d959103 AW |
1431 | |
1432 | return ret; | |
1433 | } | |
1434 | ||
19b8d12b AW |
1435 | /* scm_c_read |
1436 | * | |
1437 | * Used by an application to read arbitrary number of bytes from an | |
1438 | * SCM port. Same semantics as libc read, except that scm_c_read only | |
1439 | * returns less than SIZE bytes if at end-of-file. | |
1440 | * | |
1441 | * Warning: Doesn't update port line and column counts! */ | |
7b292a9d | 1442 | |
19b8d12b AW |
1443 | /* This structure, and the following swap_buffer function, are used |
1444 | for temporarily swapping a port's own read buffer, and the buffer | |
1445 | that the caller of scm_c_read provides. */ | |
1446 | struct port_and_swap_buffer | |
1447 | { | |
1448 | scm_t_port *pt; | |
1449 | unsigned char *buffer; | |
1450 | size_t size; | |
1451 | }; | |
7b292a9d | 1452 | |
19b8d12b AW |
1453 | static void |
1454 | swap_buffer (void *data) | |
1455 | { | |
1456 | struct port_and_swap_buffer *psb = (struct port_and_swap_buffer *) data; | |
1457 | unsigned char *old_buf = psb->pt->read_buf; | |
1458 | size_t old_size = psb->pt->read_buf_size; | |
7be1705d | 1459 | |
19b8d12b AW |
1460 | /* Make the port use (buffer, size) from the struct. */ |
1461 | psb->pt->read_pos = psb->pt->read_buf = psb->pt->read_end = psb->buffer; | |
1462 | psb->pt->read_buf_size = psb->size; | |
7b292a9d | 1463 | |
19b8d12b AW |
1464 | /* Save the port's old (buffer, size) in the struct. */ |
1465 | psb->buffer = old_buf; | |
1466 | psb->size = old_size; | |
7b292a9d LC |
1467 | } |
1468 | ||
f6f4feb0 MW |
1469 | static int scm_i_fill_input_unlocked (SCM port); |
1470 | ||
19b8d12b | 1471 | size_t |
be632904 | 1472 | scm_c_read_unlocked (SCM port, void *buffer, size_t size) |
19b8d12b | 1473 | #define FUNC_NAME "scm_c_read" |
7b292a9d LC |
1474 | { |
1475 | scm_t_port *pt; | |
f6f4feb0 | 1476 | scm_t_port_internal *pti; |
19b8d12b AW |
1477 | size_t n_read = 0, n_available; |
1478 | struct port_and_swap_buffer psb; | |
889975e5 | 1479 | |
19b8d12b | 1480 | SCM_VALIDATE_OPINPORT (1, port); |
889975e5 | 1481 | |
19b8d12b | 1482 | pt = SCM_PTAB_ENTRY (port); |
f6f4feb0 | 1483 | pti = SCM_PORT_GET_INTERNAL (port); |
19b8d12b AW |
1484 | if (pt->rw_active == SCM_PORT_WRITE) |
1485 | SCM_PORT_DESCRIPTOR (port)->flush (port); | |
889975e5 | 1486 | |
19b8d12b AW |
1487 | if (pt->rw_random) |
1488 | pt->rw_active = SCM_PORT_READ; | |
889975e5 | 1489 | |
19b8d12b AW |
1490 | /* Take bytes first from the port's read buffer. */ |
1491 | if (pt->read_pos < pt->read_end) | |
1492 | { | |
1493 | n_available = min (size, pt->read_end - pt->read_pos); | |
1494 | memcpy (buffer, pt->read_pos, n_available); | |
1495 | buffer = (char *) buffer + n_available; | |
1496 | pt->read_pos += n_available; | |
1497 | n_read += n_available; | |
1498 | size -= n_available; | |
1499 | } | |
889975e5 | 1500 | |
19b8d12b AW |
1501 | /* Avoid the scm_dynwind_* costs if we now have enough data. */ |
1502 | if (size == 0) | |
1503 | return n_read; | |
b5cb4464 | 1504 | |
f6f4feb0 MW |
1505 | /* Now we will call scm_i_fill_input_unlocked repeatedly until we have |
1506 | read the requested number of bytes. (Note that a single | |
1507 | scm_i_fill_input_unlocked call does not guarantee to fill the whole | |
1508 | of the port's read buffer.) */ | |
d1b9f8ac | 1509 | if (pt->read_buf_size <= 1 |
f6f4feb0 | 1510 | && pti->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1) |
b5cb4464 | 1511 | { |
f6f4feb0 MW |
1512 | /* The port that we are reading from is unbuffered - i.e. does not |
1513 | have its own persistent buffer - but we have a buffer, provided | |
1514 | by our caller, that is the right size for the data that is | |
1515 | wanted. For the following scm_i_fill_input_unlocked calls, | |
1516 | therefore, we use the buffer in hand as the port's read buffer. | |
1517 | ||
1518 | We need to make sure that the port's normal (1 byte) buffer is | |
1519 | reinstated in case one of the scm_i_fill_input_unlocked () | |
1520 | calls throws an exception; we use the scm_dynwind_* API to | |
1521 | achieve that. | |
75192345 MG |
1522 | |
1523 | A consequence of this optimization is that the fill_input | |
1524 | functions can't unget characters. That'll push data to the | |
1525 | pushback buffer instead of this psb buffer. */ | |
1526 | #if SCM_DEBUG == 1 | |
1527 | unsigned char *pback = pt->putback_buf; | |
1528 | #endif | |
6d227556 NJ |
1529 | psb.pt = pt; |
1530 | psb.buffer = buffer; | |
1531 | psb.size = size; | |
1532 | scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE); | |
1533 | scm_dynwind_rewind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY); | |
1534 | scm_dynwind_unwind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY); | |
1535 | ||
f6f4feb0 MW |
1536 | /* Call scm_i_fill_input_unlocked until we have all the bytes that |
1537 | we need, or we hit EOF. */ | |
1538 | while (pt->read_buf_size && (scm_i_fill_input_unlocked (port) != EOF)) | |
6d227556 NJ |
1539 | { |
1540 | pt->read_buf_size -= (pt->read_end - pt->read_pos); | |
1541 | pt->read_pos = pt->read_buf = pt->read_end; | |
1542 | } | |
75192345 MG |
1543 | #if SCM_DEBUG == 1 |
1544 | if (pback != pt->putback_buf | |
1545 | || pt->read_buf - (unsigned char *) buffer < 0) | |
1546 | scm_misc_error (FUNC_NAME, | |
1547 | "scm_c_read must not call a fill function that pushes " | |
1548 | "back characters onto an unbuffered port", SCM_EOL); | |
1549 | #endif | |
6d227556 | 1550 | n_read += pt->read_buf - (unsigned char *) buffer; |
75192345 | 1551 | |
6d227556 NJ |
1552 | /* Reinstate the port's normal buffer. */ |
1553 | scm_dynwind_end (); | |
1554 | } | |
1555 | else | |
1556 | { | |
1557 | /* The port has its own buffer. It is important that we use it, | |
1558 | even if it happens to be smaller than our caller's buffer, so | |
1559 | that a custom port implementation's entry points (in | |
1560 | particular, fill_input) can rely on the buffer always being | |
1561 | the same as they first set up. */ | |
f6f4feb0 | 1562 | while (size && (scm_i_fill_input_unlocked (port) != EOF)) |
6d227556 NJ |
1563 | { |
1564 | n_available = min (size, pt->read_end - pt->read_pos); | |
1565 | memcpy (buffer, pt->read_pos, n_available); | |
1566 | buffer = (char *) buffer + n_available; | |
1567 | pt->read_pos += n_available; | |
1568 | n_read += n_available; | |
1569 | size -= n_available; | |
1570 | } | |
1571 | } | |
6fe692e9 | 1572 | |
b5cb4464 | 1573 | return n_read; |
6fe692e9 | 1574 | } |
693758d5 | 1575 | #undef FUNC_NAME |
6fe692e9 | 1576 | |
be632904 AW |
1577 | size_t |
1578 | scm_c_read (SCM port, void *buffer, size_t size) | |
1579 | { | |
92c0ebac | 1580 | scm_i_pthread_mutex_t *lock; |
be632904 AW |
1581 | size_t ret; |
1582 | ||
92c0ebac | 1583 | scm_c_lock_port (port, &lock); |
be632904 | 1584 | ret = scm_c_read_unlocked (port, buffer, size); |
92c0ebac AW |
1585 | if (lock) |
1586 | scm_i_pthread_mutex_unlock (lock); | |
1587 | ||
be632904 AW |
1588 | |
1589 | return ret; | |
1590 | } | |
1591 | ||
19b8d12b AW |
1592 | /* Update the line and column number of PORT after consumption of C. */ |
1593 | static inline void | |
1594 | update_port_lf (scm_t_wchar c, SCM port) | |
6fe692e9 | 1595 | { |
19b8d12b AW |
1596 | switch (c) |
1597 | { | |
1598 | case '\a': | |
1599 | case EOF: | |
1600 | break; | |
1601 | case '\b': | |
1602 | SCM_DECCOL (port); | |
1603 | break; | |
1604 | case '\n': | |
1605 | SCM_INCLINE (port); | |
1606 | break; | |
1607 | case '\r': | |
1608 | SCM_ZEROCOL (port); | |
1609 | break; | |
1610 | case '\t': | |
1611 | SCM_TABCOL (port); | |
1612 | break; | |
1613 | default: | |
1614 | SCM_INCCOL (port); | |
1615 | break; | |
1616 | } | |
1617 | } | |
6fe692e9 | 1618 | |
19b8d12b | 1619 | #define SCM_MBCHAR_BUF_SIZE (4) |
6fe692e9 | 1620 | |
19b8d12b AW |
1621 | /* Convert the SIZE-byte UTF-8 sequence in UTF8_BUF to a codepoint. |
1622 | UTF8_BUF is assumed to contain a valid UTF-8 sequence. */ | |
1623 | static scm_t_wchar | |
1624 | utf8_to_codepoint (const scm_t_uint8 *utf8_buf, size_t size) | |
283a1a0e | 1625 | { |
19b8d12b | 1626 | scm_t_wchar codepoint; |
283a1a0e | 1627 | |
19b8d12b | 1628 | if (utf8_buf[0] <= 0x7f) |
283a1a0e | 1629 | { |
19b8d12b AW |
1630 | assert (size == 1); |
1631 | codepoint = utf8_buf[0]; | |
1632 | } | |
1633 | else if ((utf8_buf[0] & 0xe0) == 0xc0) | |
1634 | { | |
1635 | assert (size == 2); | |
1636 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x1f) << 6UL | |
1637 | | (utf8_buf[1] & 0x3f); | |
1638 | } | |
1639 | else if ((utf8_buf[0] & 0xf0) == 0xe0) | |
1640 | { | |
1641 | assert (size == 3); | |
1642 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x0f) << 12UL | |
1643 | | ((scm_t_wchar) utf8_buf[1] & 0x3f) << 6UL | |
1644 | | (utf8_buf[2] & 0x3f); | |
283a1a0e GH |
1645 | } |
1646 | else | |
19b8d12b AW |
1647 | { |
1648 | assert (size == 4); | |
1649 | codepoint = ((scm_t_wchar) utf8_buf[0] & 0x07) << 18UL | |
1650 | | ((scm_t_wchar) utf8_buf[1] & 0x3f) << 12UL | |
1651 | | ((scm_t_wchar) utf8_buf[2] & 0x3f) << 6UL | |
1652 | | (utf8_buf[3] & 0x3f); | |
1653 | } | |
283a1a0e | 1654 | |
19b8d12b | 1655 | return codepoint; |
283a1a0e GH |
1656 | } |
1657 | ||
19b8d12b AW |
1658 | /* Read a UTF-8 sequence from PORT. On success, return 0 and set |
1659 | *CODEPOINT to the codepoint that was read, fill BUF with its UTF-8 | |
1660 | representation, and set *LEN to the length in bytes. Return | |
1661 | `EILSEQ' on error. */ | |
1662 | static int | |
1663 | get_utf8_codepoint (SCM port, scm_t_wchar *codepoint, | |
1664 | scm_t_uint8 buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
1665 | { | |
1666 | #define ASSERT_NOT_EOF(b) \ | |
1667 | if (SCM_UNLIKELY ((b) == EOF)) \ | |
1668 | goto invalid_seq | |
1669 | #define CONSUME_PEEKED_BYTE() \ | |
1670 | pt->read_pos++ | |
ee149d03 | 1671 | |
19b8d12b AW |
1672 | int byte; |
1673 | scm_t_port *pt; | |
ee149d03 | 1674 | |
19b8d12b AW |
1675 | *len = 0; |
1676 | pt = SCM_PTAB_ENTRY (port); | |
840ae05d | 1677 | |
0d959103 | 1678 | byte = scm_get_byte_or_eof_unlocked (port); |
19b8d12b | 1679 | if (byte == EOF) |
6c951427 | 1680 | { |
19b8d12b AW |
1681 | *codepoint = EOF; |
1682 | return 0; | |
1683 | } | |
6c951427 | 1684 | |
19b8d12b AW |
1685 | buf[0] = (scm_t_uint8) byte; |
1686 | *len = 1; | |
6c951427 | 1687 | |
19b8d12b AW |
1688 | if (buf[0] <= 0x7f) |
1689 | /* 1-byte form. */ | |
1690 | *codepoint = buf[0]; | |
1691 | else if (buf[0] >= 0xc2 && buf[0] <= 0xdf) | |
1692 | { | |
1693 | /* 2-byte form. */ | |
0d959103 | 1694 | byte = scm_peek_byte_or_eof_unlocked (port); |
19b8d12b | 1695 | ASSERT_NOT_EOF (byte); |
6c951427 | 1696 | |
19b8d12b AW |
1697 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1698 | goto invalid_seq; | |
6c951427 | 1699 | |
19b8d12b AW |
1700 | CONSUME_PEEKED_BYTE (); |
1701 | buf[1] = (scm_t_uint8) byte; | |
1702 | *len = 2; | |
1703 | ||
1704 | *codepoint = ((scm_t_wchar) buf[0] & 0x1f) << 6UL | |
1705 | | (buf[1] & 0x3f); | |
6c951427 | 1706 | } |
19b8d12b | 1707 | else if ((buf[0] & 0xf0) == 0xe0) |
6c951427 | 1708 | { |
19b8d12b | 1709 | /* 3-byte form. */ |
0d959103 | 1710 | byte = scm_peek_byte_or_eof_unlocked (port); |
19b8d12b | 1711 | ASSERT_NOT_EOF (byte); |
6c951427 | 1712 | |
19b8d12b AW |
1713 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80 |
1714 | || (buf[0] == 0xe0 && byte < 0xa0) | |
1715 | || (buf[0] == 0xed && byte > 0x9f))) | |
1716 | goto invalid_seq; | |
6c951427 | 1717 | |
19b8d12b AW |
1718 | CONSUME_PEEKED_BYTE (); |
1719 | buf[1] = (scm_t_uint8) byte; | |
1720 | *len = 2; | |
1721 | ||
0d959103 | 1722 | byte = scm_peek_byte_or_eof_unlocked (port); |
19b8d12b AW |
1723 | ASSERT_NOT_EOF (byte); |
1724 | ||
1725 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) | |
1726 | goto invalid_seq; | |
1727 | ||
1728 | CONSUME_PEEKED_BYTE (); | |
1729 | buf[2] = (scm_t_uint8) byte; | |
1730 | *len = 3; | |
1731 | ||
1732 | *codepoint = ((scm_t_wchar) buf[0] & 0x0f) << 12UL | |
1733 | | ((scm_t_wchar) buf[1] & 0x3f) << 6UL | |
1734 | | (buf[2] & 0x3f); | |
6c951427 | 1735 | } |
19b8d12b AW |
1736 | else if (buf[0] >= 0xf0 && buf[0] <= 0xf4) |
1737 | { | |
1738 | /* 4-byte form. */ | |
0d959103 | 1739 | byte = scm_peek_byte_or_eof_unlocked (port); |
19b8d12b | 1740 | ASSERT_NOT_EOF (byte); |
6c951427 | 1741 | |
19b8d12b AW |
1742 | if (SCM_UNLIKELY (((byte & 0xc0) != 0x80) |
1743 | || (buf[0] == 0xf0 && byte < 0x90) | |
1744 | || (buf[0] == 0xf4 && byte > 0x8f))) | |
1745 | goto invalid_seq; | |
ee149d03 | 1746 | |
19b8d12b AW |
1747 | CONSUME_PEEKED_BYTE (); |
1748 | buf[1] = (scm_t_uint8) byte; | |
1749 | *len = 2; | |
889975e5 | 1750 | |
0d959103 | 1751 | byte = scm_peek_byte_or_eof_unlocked (port); |
19b8d12b | 1752 | ASSERT_NOT_EOF (byte); |
889975e5 | 1753 | |
19b8d12b AW |
1754 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1755 | goto invalid_seq; | |
63479112 | 1756 | |
19b8d12b AW |
1757 | CONSUME_PEEKED_BYTE (); |
1758 | buf[2] = (scm_t_uint8) byte; | |
1759 | *len = 3; | |
63479112 | 1760 | |
0d959103 | 1761 | byte = scm_peek_byte_or_eof_unlocked (port); |
19b8d12b | 1762 | ASSERT_NOT_EOF (byte); |
63479112 | 1763 | |
19b8d12b AW |
1764 | if (SCM_UNLIKELY ((byte & 0xc0) != 0x80)) |
1765 | goto invalid_seq; | |
63479112 | 1766 | |
19b8d12b AW |
1767 | CONSUME_PEEKED_BYTE (); |
1768 | buf[3] = (scm_t_uint8) byte; | |
1769 | *len = 4; | |
840ae05d | 1770 | |
19b8d12b AW |
1771 | *codepoint = ((scm_t_wchar) buf[0] & 0x07) << 18UL |
1772 | | ((scm_t_wchar) buf[1] & 0x3f) << 12UL | |
1773 | | ((scm_t_wchar) buf[2] & 0x3f) << 6UL | |
1774 | | (buf[3] & 0x3f); | |
ee149d03 JB |
1775 | } |
1776 | else | |
19b8d12b | 1777 | goto invalid_seq; |
ee149d03 | 1778 | |
19b8d12b | 1779 | return 0; |
ee149d03 | 1780 | |
19b8d12b AW |
1781 | invalid_seq: |
1782 | /* Here we could choose the consume the faulty byte when it's not a | |
1783 | valid starting byte, but it's not a requirement. What Section 3.9 | |
1784 | of Unicode 6.0.0 mandates, though, is to not consume a byte that | |
1785 | would otherwise be a valid starting byte. */ | |
ee149d03 | 1786 | |
19b8d12b | 1787 | return EILSEQ; |
ee149d03 | 1788 | |
19b8d12b AW |
1789 | #undef CONSUME_PEEKED_BYTE |
1790 | #undef ASSERT_NOT_EOF | |
1791 | } | |
1792 | ||
79eb47ea | 1793 | /* Read an ISO-8859-1 codepoint (a byte) from PORT. On success, return |
05b4d910 AW |
1794 | 0 and set *CODEPOINT to the codepoint that was read, fill BUF with |
1795 | its UTF-8 representation, and set *LEN to the length in bytes. | |
1796 | Return `EILSEQ' on error. */ | |
79eb47ea AW |
1797 | static int |
1798 | get_latin1_codepoint (SCM port, scm_t_wchar *codepoint, | |
1799 | char buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
1800 | { | |
1801 | *codepoint = scm_get_byte_or_eof_unlocked (port); | |
1802 | ||
1803 | if (*codepoint == EOF) | |
1804 | *len = 0; | |
1805 | else | |
1806 | { | |
1807 | *len = 1; | |
1808 | buf[0] = *codepoint; | |
1809 | } | |
1810 | return 0; | |
1811 | } | |
1812 | ||
19b8d12b AW |
1813 | /* Likewise, read a byte sequence from PORT, passing it through its |
1814 | input conversion descriptor. */ | |
1815 | static int | |
1816 | get_iconv_codepoint (SCM port, scm_t_wchar *codepoint, | |
1817 | char buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
1818 | { | |
6c98257f | 1819 | scm_t_iconv_descriptors *id; |
19b8d12b | 1820 | scm_t_uint8 utf8_buf[SCM_MBCHAR_BUF_SIZE]; |
f6f4feb0 | 1821 | size_t input_size = 0; |
19b8d12b | 1822 | |
f6f4feb0 | 1823 | id = scm_i_port_iconv_descriptors (port, SCM_PORT_READ); |
19b8d12b | 1824 | |
f6f4feb0 | 1825 | for (;;) |
19b8d12b | 1826 | { |
f6f4feb0 MW |
1827 | int byte_read; |
1828 | char *input, *output; | |
19b8d12b AW |
1829 | size_t input_left, output_left, done; |
1830 | ||
0d959103 | 1831 | byte_read = scm_get_byte_or_eof_unlocked (port); |
f6f4feb0 | 1832 | if (SCM_UNLIKELY (byte_read == EOF)) |
19b8d12b | 1833 | { |
f6f4feb0 MW |
1834 | if (SCM_LIKELY (input_size == 0)) |
1835 | { | |
1836 | *codepoint = (scm_t_wchar) EOF; | |
1837 | *len = input_size; | |
1838 | return 0; | |
1839 | } | |
1840 | else | |
1841 | { | |
1842 | /* EOF found in the middle of a multibyte character. */ | |
1843 | scm_i_set_pending_eof (port); | |
1844 | return EILSEQ; | |
1845 | } | |
19b8d12b AW |
1846 | } |
1847 | ||
f6f4feb0 | 1848 | buf[input_size++] = byte_read; |
19b8d12b AW |
1849 | |
1850 | input = buf; | |
f6f4feb0 MW |
1851 | input_left = input_size; |
1852 | output = (char *) utf8_buf; | |
19b8d12b AW |
1853 | output_left = sizeof (utf8_buf); |
1854 | ||
6c98257f | 1855 | done = iconv (id->input_cd, &input, &input_left, &output, &output_left); |
f6f4feb0 | 1856 | |
19b8d12b AW |
1857 | if (done == (size_t) -1) |
1858 | { | |
f6f4feb0 MW |
1859 | int err = errno; |
1860 | if (SCM_LIKELY (err == EINVAL)) | |
1861 | /* The input byte sequence did not form a complete | |
1862 | character. Read another byte and try again. */ | |
1863 | continue; | |
1864 | else | |
1865 | return err; | |
19b8d12b AW |
1866 | } |
1867 | else | |
f6f4feb0 MW |
1868 | { |
1869 | size_t output_size = sizeof (utf8_buf) - output_left; | |
1870 | if (SCM_LIKELY (output_size > 0)) | |
1871 | { | |
1872 | /* iconv generated output. Convert the UTF8_BUF sequence | |
1873 | to a Unicode code point. */ | |
1874 | *codepoint = utf8_to_codepoint (utf8_buf, output_size); | |
1875 | *len = input_size; | |
1876 | return 0; | |
1877 | } | |
1878 | else | |
1879 | { | |
1880 | /* iconv consumed some bytes without producing any output. | |
1881 | Most likely this means that a Unicode byte-order mark | |
1882 | (BOM) was consumed, which should not be included in the | |
1883 | returned buf. Shift any remaining bytes to the beginning | |
1884 | of buf, and continue the loop. */ | |
1885 | memmove (buf, input, input_left); | |
1886 | input_size = input_left; | |
1887 | continue; | |
1888 | } | |
1889 | } | |
19b8d12b | 1890 | } |
19b8d12b AW |
1891 | } |
1892 | ||
1893 | /* Read a codepoint from PORT and return it in *CODEPOINT. Fill BUF | |
1894 | with the byte representation of the codepoint in PORT's encoding, and | |
1895 | set *LEN to the length in bytes of that representation. Return 0 on | |
1896 | success and an errno value on error. */ | |
24ea9f9c | 1897 | static SCM_C_INLINE int |
19b8d12b AW |
1898 | get_codepoint (SCM port, scm_t_wchar *codepoint, |
1899 | char buf[SCM_MBCHAR_BUF_SIZE], size_t *len) | |
1900 | { | |
1901 | int err; | |
1902 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
f6f4feb0 | 1903 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
19b8d12b | 1904 | |
f6f4feb0 | 1905 | if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8) |
19b8d12b | 1906 | err = get_utf8_codepoint (port, codepoint, (scm_t_uint8 *) buf, len); |
f6f4feb0 | 1907 | else if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1) |
79eb47ea | 1908 | err = get_latin1_codepoint (port, codepoint, buf, len); |
19b8d12b AW |
1909 | else |
1910 | err = get_iconv_codepoint (port, codepoint, buf, len); | |
1911 | ||
1912 | if (SCM_LIKELY (err == 0)) | |
f6f4feb0 MW |
1913 | { |
1914 | if (SCM_UNLIKELY (pti->at_stream_start_for_bom_read)) | |
1915 | { | |
1916 | /* Record that we're no longer at stream start. */ | |
1917 | pti->at_stream_start_for_bom_read = 0; | |
1918 | if (pt->rw_random) | |
1919 | pti->at_stream_start_for_bom_write = 0; | |
1920 | ||
1921 | /* If we just read a BOM in an encoding that recognizes them, | |
1922 | then silently consume it and read another code point. */ | |
1923 | if (SCM_UNLIKELY | |
1924 | (*codepoint == SCM_UNICODE_BOM | |
1925 | && (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8 | |
1926 | || strcmp (pt->encoding, "UTF-16") == 0 | |
1927 | || strcmp (pt->encoding, "UTF-32") == 0))) | |
1928 | return get_codepoint (port, codepoint, buf, len); | |
1929 | } | |
1930 | update_port_lf (*codepoint, port); | |
1931 | } | |
19b8d12b AW |
1932 | else if (pt->ilseq_handler == SCM_ICONVEH_QUESTION_MARK) |
1933 | { | |
1934 | *codepoint = '?'; | |
1935 | err = 0; | |
1936 | update_port_lf (*codepoint, port); | |
1937 | } | |
1938 | ||
1939 | return err; | |
1940 | } | |
1941 | ||
1942 | /* Read a codepoint from PORT and return it. */ | |
1943 | scm_t_wchar | |
be632904 | 1944 | scm_getc_unlocked (SCM port) |
19b8d12b AW |
1945 | #define FUNC_NAME "scm_getc" |
1946 | { | |
1947 | int err; | |
1948 | size_t len; | |
1949 | scm_t_wchar codepoint; | |
1950 | char buf[SCM_MBCHAR_BUF_SIZE]; | |
1951 | ||
1952 | err = get_codepoint (port, &codepoint, buf, &len); | |
1953 | if (SCM_UNLIKELY (err != 0)) | |
1954 | /* At this point PORT should point past the invalid encoding, as per | |
1955 | R6RS-lib Section 8.2.4. */ | |
1956 | scm_decoding_error (FUNC_NAME, err, "input decoding error", port); | |
1957 | ||
1958 | return codepoint; | |
1959 | } | |
1960 | #undef FUNC_NAME | |
1961 | ||
be632904 AW |
1962 | scm_t_wchar |
1963 | scm_getc (SCM port) | |
1964 | { | |
92c0ebac | 1965 | scm_i_pthread_mutex_t *lock; |
be632904 AW |
1966 | scm_t_wchar ret; |
1967 | ||
92c0ebac | 1968 | scm_c_lock_port (port, &lock); |
be632904 | 1969 | ret = scm_getc_unlocked (port); |
92c0ebac AW |
1970 | if (lock) |
1971 | scm_i_pthread_mutex_unlock (lock); | |
1972 | ||
be632904 AW |
1973 | |
1974 | return ret; | |
1975 | } | |
1976 | ||
19b8d12b AW |
1977 | SCM_DEFINE (scm_read_char, "read-char", 0, 1, 0, |
1978 | (SCM port), | |
1979 | "Return the next character available from @var{port}, updating\n" | |
1980 | "@var{port} to point to the following character. If no more\n" | |
1981 | "characters are available, the end-of-file object is returned.\n" | |
1982 | "\n" | |
1983 | "When @var{port}'s data cannot be decoded according to its\n" | |
1984 | "character encoding, a @code{decoding-error} is raised and\n" | |
1985 | "@var{port} points past the erroneous byte sequence.\n") | |
1986 | #define FUNC_NAME s_scm_read_char | |
1987 | { | |
1988 | scm_t_wchar c; | |
1989 | if (SCM_UNBNDP (port)) | |
1990 | port = scm_current_input_port (); | |
1991 | SCM_VALIDATE_OPINPORT (1, port); | |
be632904 | 1992 | c = scm_getc_unlocked (port); |
19b8d12b AW |
1993 | if (EOF == c) |
1994 | return SCM_EOF_VAL; | |
1995 | return SCM_MAKE_CHAR (c); | |
1996 | } | |
1997 | #undef FUNC_NAME | |
1998 | ||
1999 | ||
2000 | \f | |
2001 | ||
2002 | /* Pushback. */ | |
f6f4feb0 | 2003 | \f |
19b8d12b | 2004 | |
f6f4feb0 MW |
2005 | |
2006 | static void | |
2007 | scm_i_unget_bytes_unlocked (const unsigned char *buf, size_t len, SCM port) | |
2008 | #define FUNC_NAME "scm_unget_bytes" | |
19b8d12b AW |
2009 | { |
2010 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
f6f4feb0 | 2011 | size_t old_len, new_len; |
19b8d12b | 2012 | |
f6f4feb0 | 2013 | scm_i_clear_pending_eof (port); |
19b8d12b | 2014 | |
f6f4feb0 | 2015 | if (pt->read_buf != pt->putback_buf) |
19b8d12b AW |
2016 | /* switch to the put-back buffer. */ |
2017 | { | |
2018 | if (pt->putback_buf == NULL) | |
2019 | { | |
f6f4feb0 MW |
2020 | pt->putback_buf_size = (len > SCM_INITIAL_PUTBACK_BUF_SIZE |
2021 | ? len : SCM_INITIAL_PUTBACK_BUF_SIZE); | |
19b8d12b AW |
2022 | pt->putback_buf |
2023 | = (unsigned char *) scm_gc_malloc_pointerless | |
f6f4feb0 | 2024 | (pt->putback_buf_size, "putback buffer"); |
19b8d12b AW |
2025 | } |
2026 | ||
2027 | pt->saved_read_buf = pt->read_buf; | |
2028 | pt->saved_read_pos = pt->read_pos; | |
2029 | pt->saved_read_end = pt->read_end; | |
2030 | pt->saved_read_buf_size = pt->read_buf_size; | |
2031 | ||
f6f4feb0 MW |
2032 | /* Put read_pos at the end of the buffer, so that ungets will not |
2033 | have to shift the buffer contents each time. */ | |
2034 | pt->read_buf = pt->putback_buf; | |
2035 | pt->read_pos = pt->read_end = pt->putback_buf + pt->putback_buf_size; | |
19b8d12b AW |
2036 | pt->read_buf_size = pt->putback_buf_size; |
2037 | } | |
2038 | ||
f6f4feb0 MW |
2039 | old_len = pt->read_end - pt->read_pos; |
2040 | new_len = old_len + len; | |
2041 | ||
2042 | if (new_len > pt->read_buf_size) | |
2043 | /* The putback buffer needs to be enlarged. */ | |
2044 | { | |
2045 | size_t new_buf_size; | |
2046 | unsigned char *new_buf, *new_end, *new_pos; | |
2047 | ||
2048 | new_buf_size = pt->read_buf_size * 2; | |
2049 | if (new_buf_size < new_len) | |
2050 | new_buf_size = new_len; | |
2051 | ||
2052 | new_buf = (unsigned char *) | |
2053 | scm_gc_malloc_pointerless (new_buf_size, "putback buffer"); | |
2054 | ||
2055 | /* Put the bytes at the end of the buffer, so that future | |
2056 | ungets won't need to shift the buffer. */ | |
2057 | new_end = new_buf + new_buf_size; | |
2058 | new_pos = new_end - old_len; | |
2059 | memcpy (new_pos, pt->read_pos, old_len); | |
2060 | ||
2061 | pt->read_buf = pt->putback_buf = new_buf; | |
2062 | pt->read_pos = new_pos; | |
2063 | pt->read_end = new_end; | |
2064 | pt->read_buf_size = pt->putback_buf_size = new_buf_size; | |
2065 | } | |
2066 | else if (pt->read_buf + len < pt->read_pos) | |
2067 | /* If needed, shift the existing buffer contents up. | |
2068 | This should not happen unless some external code | |
2069 | manipulates the putback buffer pointers. */ | |
2070 | { | |
2071 | unsigned char *new_end = pt->read_buf + pt->read_buf_size; | |
2072 | unsigned char *new_pos = new_end - old_len; | |
2073 | ||
2074 | memmove (new_pos, pt->read_pos, old_len); | |
2075 | pt->read_pos = new_pos; | |
2076 | pt->read_end = new_end; | |
2077 | } | |
2078 | ||
2079 | /* Move read_pos back and copy the bytes there. */ | |
2080 | pt->read_pos -= len; | |
2081 | memcpy (pt->read_buf + (pt->read_pos - pt->read_buf), buf, len); | |
2082 | ||
2083 | if (pt->rw_active == SCM_PORT_WRITE) | |
2084 | scm_flush (port); | |
19b8d12b AW |
2085 | |
2086 | if (pt->rw_random) | |
2087 | pt->rw_active = SCM_PORT_READ; | |
2088 | } | |
2089 | #undef FUNC_NAME | |
2090 | ||
f6f4feb0 MW |
2091 | void |
2092 | scm_unget_bytes_unlocked (const unsigned char *buf, size_t len, SCM port) | |
7f6c3f8f | 2093 | { |
f6f4feb0 | 2094 | scm_i_unget_bytes_unlocked (buf, len, port); |
7f6c3f8f MW |
2095 | } |
2096 | ||
2097 | void | |
f6f4feb0 MW |
2098 | scm_unget_byte_unlocked (int c, SCM port) |
2099 | { | |
2100 | unsigned char byte = c; | |
2101 | scm_i_unget_bytes_unlocked (&byte, 1, port); | |
2102 | } | |
2103 | ||
2104 | void | |
2105 | scm_unget_bytes (const unsigned char *buf, size_t len, SCM port) | |
7f6c3f8f | 2106 | { |
f6f4feb0 MW |
2107 | scm_i_pthread_mutex_t *lock; |
2108 | scm_c_lock_port (port, &lock); | |
2109 | scm_i_unget_bytes_unlocked (buf, len, port); | |
2110 | if (lock) | |
2111 | scm_i_pthread_mutex_unlock (lock); | |
2112 | } | |
7f6c3f8f | 2113 | |
c932ce0b AW |
2114 | void |
2115 | scm_unget_byte (int c, SCM port) | |
2116 | { | |
f6f4feb0 | 2117 | unsigned char byte = c; |
92c0ebac AW |
2118 | scm_i_pthread_mutex_t *lock; |
2119 | scm_c_lock_port (port, &lock); | |
f6f4feb0 | 2120 | scm_i_unget_bytes_unlocked (&byte, 1, port); |
92c0ebac AW |
2121 | if (lock) |
2122 | scm_i_pthread_mutex_unlock (lock); | |
c932ce0b AW |
2123 | } |
2124 | ||
19b8d12b | 2125 | void |
c932ce0b | 2126 | scm_ungetc_unlocked (scm_t_wchar c, SCM port) |
19b8d12b AW |
2127 | #define FUNC_NAME "scm_ungetc" |
2128 | { | |
2129 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
be7ecef0 | 2130 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
19b8d12b AW |
2131 | char *result; |
2132 | char result_buf[10]; | |
19b8d12b | 2133 | size_t len; |
19b8d12b | 2134 | |
19b8d12b | 2135 | len = sizeof (result_buf); |
be7ecef0 AW |
2136 | |
2137 | if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8) | |
2138 | { | |
2139 | if (c < 0xf0) | |
2140 | { | |
2141 | result_buf[0] = (char) c; | |
2142 | result = result_buf; | |
2143 | len = 1; | |
2144 | } | |
2145 | else | |
2146 | result = | |
2147 | (char *) u32_to_u8 ((uint32_t *) &c, 1, (uint8_t *) result_buf, &len); | |
2148 | } | |
2149 | else if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1 && c <= 0xff) | |
2150 | { | |
2151 | result_buf[0] = (char) c; | |
2152 | result = result_buf; | |
2153 | len = 1; | |
2154 | } | |
2155 | else | |
2156 | result = u32_conv_to_encoding (pt->encoding, | |
2157 | (enum iconv_ilseq_handler) pt->ilseq_handler, | |
2158 | (uint32_t *) &c, 1, NULL, | |
2159 | result_buf, &len); | |
19b8d12b AW |
2160 | |
2161 | if (SCM_UNLIKELY (result == NULL || len == 0)) | |
2162 | scm_encoding_error (FUNC_NAME, errno, | |
2163 | "conversion to port encoding failed", | |
2164 | SCM_BOOL_F, SCM_MAKE_CHAR (c)); | |
2165 | ||
f6f4feb0 | 2166 | scm_i_unget_bytes_unlocked ((unsigned char *) result, len, port); |
19b8d12b AW |
2167 | |
2168 | if (SCM_UNLIKELY (result != result_buf)) | |
2169 | free (result); | |
2170 | ||
2171 | if (c == '\n') | |
2172 | { | |
2173 | /* What should col be in this case? | |
2174 | * We'll leave it at -1. | |
2175 | */ | |
2176 | SCM_LINUM (port) -= 1; | |
2177 | } | |
2178 | else | |
2179 | SCM_COL(port) -= 1; | |
2180 | } | |
2181 | #undef FUNC_NAME | |
2182 | ||
c932ce0b AW |
2183 | void |
2184 | scm_ungetc (scm_t_wchar c, SCM port) | |
2185 | { | |
92c0ebac AW |
2186 | scm_i_pthread_mutex_t *lock; |
2187 | scm_c_lock_port (port, &lock); | |
c932ce0b | 2188 | scm_ungetc_unlocked (c, port); |
92c0ebac AW |
2189 | if (lock) |
2190 | scm_i_pthread_mutex_unlock (lock); | |
2191 | ||
c932ce0b | 2192 | } |
19b8d12b AW |
2193 | |
2194 | void | |
c932ce0b | 2195 | scm_ungets_unlocked (const char *s, int n, SCM port) |
19b8d12b AW |
2196 | { |
2197 | /* This is simple minded and inefficient, but unreading strings is | |
2198 | * probably not a common operation, and remember that line and | |
2199 | * column numbers have to be handled... | |
2200 | * | |
2201 | * Please feel free to write an optimized version! | |
2202 | */ | |
2203 | while (n--) | |
c932ce0b | 2204 | scm_ungetc_unlocked (s[n], port); |
19b8d12b AW |
2205 | } |
2206 | ||
c932ce0b AW |
2207 | void |
2208 | scm_ungets (const char *s, int n, SCM port) | |
2209 | { | |
92c0ebac AW |
2210 | scm_i_pthread_mutex_t *lock; |
2211 | scm_c_lock_port (port, &lock); | |
c932ce0b | 2212 | scm_ungets_unlocked (s, n, port); |
92c0ebac AW |
2213 | if (lock) |
2214 | scm_i_pthread_mutex_unlock (lock); | |
2215 | ||
c932ce0b | 2216 | } |
19b8d12b AW |
2217 | |
2218 | SCM_DEFINE (scm_peek_char, "peek-char", 0, 1, 0, | |
1bbd0b84 | 2219 | (SCM port), |
1e6808ea MG |
2220 | "Return the next character available from @var{port},\n" |
2221 | "@emph{without} updating @var{port} to point to the following\n" | |
2222 | "character. If no more characters are available, the\n" | |
c2dfff19 KR |
2223 | "end-of-file object is returned.\n" |
2224 | "\n" | |
2225 | "The value returned by\n" | |
1e6808ea MG |
2226 | "a call to @code{peek-char} is the same as the value that would\n" |
2227 | "have been returned by a call to @code{read-char} on the same\n" | |
2228 | "port. The only difference is that the very next call to\n" | |
2229 | "@code{read-char} or @code{peek-char} on that @var{port} will\n" | |
2230 | "return the value returned by the preceding call to\n" | |
2231 | "@code{peek-char}. In particular, a call to @code{peek-char} on\n" | |
2232 | "an interactive port will hang waiting for input whenever a call\n" | |
c62da8f8 LC |
2233 | "to @code{read-char} would have hung.\n" |
2234 | "\n" | |
2235 | "As for @code{read-char}, a @code{decoding-error} may be raised\n" | |
2236 | "if such a situation occurs. However, unlike with @code{read-char},\n" | |
2237 | "@var{port} still points at the beginning of the erroneous byte\n" | |
2238 | "sequence when the error is raised.\n") | |
1bbd0b84 | 2239 | #define FUNC_NAME s_scm_peek_char |
ee149d03 | 2240 | { |
c62da8f8 | 2241 | int err; |
fd5eec2b LC |
2242 | SCM result; |
2243 | scm_t_wchar c; | |
2244 | char bytes[SCM_MBCHAR_BUF_SIZE]; | |
7f6c3f8f | 2245 | long column, line; |
fe8935d4 | 2246 | size_t len = 0; |
fd5eec2b | 2247 | |
ee149d03 | 2248 | if (SCM_UNBNDP (port)) |
9de87eea | 2249 | port = scm_current_input_port (); |
b2456dd4 | 2250 | SCM_VALIDATE_OPINPORT (1, port); |
fd5eec2b LC |
2251 | |
2252 | column = SCM_COL (port); | |
2253 | line = SCM_LINUM (port); | |
2254 | ||
c62da8f8 | 2255 | err = get_codepoint (port, &c, bytes, &len); |
fd5eec2b | 2256 | |
f6f4feb0 | 2257 | scm_i_unget_bytes_unlocked ((unsigned char *) bytes, len, port); |
fd5eec2b | 2258 | |
c62da8f8 LC |
2259 | SCM_COL (port) = column; |
2260 | SCM_LINUM (port) = line; | |
fd5eec2b | 2261 | |
c62da8f8 LC |
2262 | if (SCM_UNLIKELY (err != 0)) |
2263 | { | |
2264 | scm_decoding_error (FUNC_NAME, err, "input decoding error", port); | |
2265 | ||
2266 | /* Shouldn't happen since `catch' always aborts to prompt. */ | |
2267 | result = SCM_BOOL_F; | |
fd5eec2b | 2268 | } |
c62da8f8 | 2269 | else if (c == EOF) |
69bc9ff3 | 2270 | { |
f6f4feb0 MW |
2271 | scm_i_set_pending_eof (port); |
2272 | result = SCM_EOF_VAL; | |
69bc9ff3 | 2273 | } |
c62da8f8 LC |
2274 | else |
2275 | result = SCM_MAKE_CHAR (c); | |
fd5eec2b LC |
2276 | |
2277 | return result; | |
3cb988bd | 2278 | } |
1bbd0b84 | 2279 | #undef FUNC_NAME |
3cb988bd | 2280 | |
1be4270a | 2281 | SCM_DEFINE (scm_unread_char, "unread-char", 1, 1, 0, |
1bbd0b84 | 2282 | (SCM cobj, SCM port), |
b7e64f8b BT |
2283 | "Place character @var{cobj} in @var{port} so that it will be\n" |
2284 | "read by the next read operation. If called multiple times, the\n" | |
2285 | "unread characters will be read again in last-in first-out\n" | |
2286 | "order. If @var{port} is not supplied, the current input port\n" | |
2287 | "is used.") | |
1bbd0b84 | 2288 | #define FUNC_NAME s_scm_unread_char |
0f2d19dd JB |
2289 | { |
2290 | int c; | |
2291 | ||
34d19ef6 | 2292 | SCM_VALIDATE_CHAR (1, cobj); |
0f2d19dd | 2293 | if (SCM_UNBNDP (port)) |
9de87eea | 2294 | port = scm_current_input_port (); |
b2456dd4 | 2295 | SCM_VALIDATE_OPINPORT (2, port); |
0f2d19dd | 2296 | |
7866a09b | 2297 | c = SCM_CHAR (cobj); |
0f2d19dd | 2298 | |
c932ce0b | 2299 | scm_ungetc_unlocked (c, port); |
0f2d19dd JB |
2300 | return cobj; |
2301 | } | |
1bbd0b84 | 2302 | #undef FUNC_NAME |
0f2d19dd | 2303 | |
a1ec6916 | 2304 | SCM_DEFINE (scm_unread_string, "unread-string", 2, 0, 0, |
1bbd0b84 | 2305 | (SCM str, SCM port), |
b380b885 MD |
2306 | "Place the string @var{str} in @var{port} so that its characters will be\n" |
2307 | "read in subsequent read operations. If called multiple times, the\n" | |
2308 | "unread characters will be read again in last-in first-out order. If\n" | |
2309 | "@var{port} is not supplied, the current-input-port is used.") | |
1bbd0b84 | 2310 | #define FUNC_NAME s_scm_unread_string |
ee1e7e13 | 2311 | { |
889975e5 | 2312 | int n; |
34d19ef6 | 2313 | SCM_VALIDATE_STRING (1, str); |
ee1e7e13 | 2314 | if (SCM_UNBNDP (port)) |
9de87eea | 2315 | port = scm_current_input_port (); |
b2456dd4 | 2316 | SCM_VALIDATE_OPINPORT (2, port); |
ee1e7e13 | 2317 | |
889975e5 MG |
2318 | n = scm_i_string_length (str); |
2319 | ||
2320 | while (n--) | |
c932ce0b | 2321 | scm_ungetc_unlocked (scm_i_string_ref (str, n), port); |
ee1e7e13 MD |
2322 | |
2323 | return str; | |
2324 | } | |
1bbd0b84 | 2325 | #undef FUNC_NAME |
ee1e7e13 | 2326 | |
840ae05d | 2327 | |
19b8d12b | 2328 | \f |
23f2b9a3 | 2329 | |
19b8d12b | 2330 | /* Manipulating the buffers. */ |
840ae05d | 2331 | |
4251ae2e AW |
2332 | /* This routine does not take any locks, as it is usually called as part |
2333 | of a port implementation. */ | |
19b8d12b AW |
2334 | void |
2335 | scm_port_non_buffer (scm_t_port *pt) | |
2336 | { | |
2337 | pt->read_pos = pt->read_buf = pt->read_end = &pt->shortbuf; | |
2338 | pt->write_buf = pt->write_pos = &pt->shortbuf; | |
2339 | pt->read_buf_size = pt->write_buf_size = 1; | |
2340 | pt->write_end = pt->write_buf + pt->write_buf_size; | |
840ae05d | 2341 | } |
8ab3d8a0 | 2342 | |
19b8d12b AW |
2343 | /* this should only be called when the read buffer is empty. it |
2344 | tries to refill the read buffer. it returns the first char from | |
2345 | the port, which is either EOF or *(pt->read_pos). */ | |
f6f4feb0 MW |
2346 | static int |
2347 | scm_i_fill_input_unlocked (SCM port) | |
82893676 | 2348 | { |
19b8d12b | 2349 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
f6f4feb0 | 2350 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port); |
8ab3d8a0 | 2351 | |
19b8d12b | 2352 | assert (pt->read_pos == pt->read_end); |
8ab3d8a0 | 2353 | |
f6f4feb0 MW |
2354 | if (pti->pending_eof) |
2355 | { | |
2356 | pti->pending_eof = 0; | |
2357 | return EOF; | |
2358 | } | |
2359 | ||
19b8d12b | 2360 | if (pt->read_buf == pt->putback_buf) |
82893676 | 2361 | { |
19b8d12b AW |
2362 | /* finished reading put-back chars. */ |
2363 | pt->read_buf = pt->saved_read_buf; | |
2364 | pt->read_pos = pt->saved_read_pos; | |
2365 | pt->read_end = pt->saved_read_end; | |
2366 | pt->read_buf_size = pt->saved_read_buf_size; | |
2367 | if (pt->read_pos < pt->read_end) | |
2368 | return *(pt->read_pos); | |
82893676 | 2369 | } |
19b8d12b | 2370 | return SCM_PORT_DESCRIPTOR (port)->fill_input (port); |
82893676 | 2371 | } |
82893676 | 2372 | |
4251ae2e AW |
2373 | int |
2374 | scm_fill_input (SCM port) | |
2375 | { | |
92c0ebac | 2376 | scm_i_pthread_mutex_t *lock; |
4251ae2e AW |
2377 | int ret; |
2378 | ||
92c0ebac | 2379 | scm_c_lock_port (port, &lock); |
4251ae2e | 2380 | ret = scm_fill_input_unlocked (port); |
92c0ebac AW |
2381 | if (lock) |
2382 | scm_i_pthread_mutex_unlock (lock); | |
2383 | ||
4251ae2e AW |
2384 | |
2385 | return ret; | |
2386 | } | |
2387 | ||
f6f4feb0 MW |
2388 | /* Slow-path fallback for 'scm_get_byte_or_eof_unlocked' */ |
2389 | int | |
2390 | scm_slow_get_byte_or_eof_unlocked (SCM port) | |
d14af9f2 | 2391 | { |
f6f4feb0 MW |
2392 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
2393 | ||
2394 | if (pt->rw_active == SCM_PORT_WRITE) | |
2395 | scm_flush_unlocked (port); | |
2396 | ||
2397 | if (pt->rw_random) | |
2398 | pt->rw_active = SCM_PORT_READ; | |
2399 | ||
2400 | if (pt->read_pos >= pt->read_end) | |
2401 | { | |
2402 | if (SCM_UNLIKELY (scm_i_fill_input_unlocked (port) == EOF)) | |
2403 | return EOF; | |
2404 | } | |
2405 | ||
2406 | return *pt->read_pos++; | |
d14af9f2 MD |
2407 | } |
2408 | ||
f6f4feb0 MW |
2409 | /* Slow-path fallback for 'scm_peek_byte_or_eof_unlocked' */ |
2410 | int | |
2411 | scm_slow_peek_byte_or_eof_unlocked (SCM port) | |
2412 | { | |
2413 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
d6a6989e | 2414 | |
f6f4feb0 MW |
2415 | if (pt->rw_active == SCM_PORT_WRITE) |
2416 | scm_flush_unlocked (port); | |
889975e5 | 2417 | |
f6f4feb0 MW |
2418 | if (pt->rw_random) |
2419 | pt->rw_active = SCM_PORT_READ; | |
2420 | ||
2421 | if (pt->read_pos >= pt->read_end) | |
2422 | { | |
2423 | if (SCM_UNLIKELY (scm_i_fill_input_unlocked (port) == EOF)) | |
2424 | { | |
2425 | scm_i_set_pending_eof (port); | |
2426 | return EOF; | |
2427 | } | |
2428 | } | |
2429 | ||
2430 | return *pt->read_pos; | |
2431 | } | |
2432 | ||
a3ded465 AW |
2433 | /* Move up to READ_LEN bytes from PORT's putback and/or read buffers |
2434 | into memory starting at DEST. Return the number of bytes moved. | |
2435 | PORT's line/column numbers are left unchanged. */ | |
19b8d12b AW |
2436 | size_t |
2437 | scm_take_from_input_buffers (SCM port, char *dest, size_t read_len) | |
840ae05d | 2438 | { |
19b8d12b | 2439 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
a3ded465 | 2440 | size_t bytes_read = 0; |
19b8d12b | 2441 | size_t from_buf = min (pt->read_end - pt->read_pos, read_len); |
840ae05d | 2442 | |
19b8d12b | 2443 | if (from_buf > 0) |
840ae05d | 2444 | { |
19b8d12b AW |
2445 | memcpy (dest, pt->read_pos, from_buf); |
2446 | pt->read_pos += from_buf; | |
a3ded465 | 2447 | bytes_read += from_buf; |
19b8d12b AW |
2448 | read_len -= from_buf; |
2449 | dest += from_buf; | |
840ae05d | 2450 | } |
3fe6190f | 2451 | |
19b8d12b AW |
2452 | /* if putback was active, try the real input buffer too. */ |
2453 | if (pt->read_buf == pt->putback_buf) | |
69bc9ff3 | 2454 | { |
19b8d12b AW |
2455 | from_buf = min (pt->saved_read_end - pt->saved_read_pos, read_len); |
2456 | if (from_buf > 0) | |
2457 | { | |
2458 | memcpy (dest, pt->saved_read_pos, from_buf); | |
2459 | pt->saved_read_pos += from_buf; | |
a3ded465 | 2460 | bytes_read += from_buf; |
19b8d12b | 2461 | } |
69bc9ff3 | 2462 | } |
8ab3d8a0 | 2463 | |
a3ded465 | 2464 | return bytes_read; |
840ae05d JB |
2465 | } |
2466 | ||
19b8d12b AW |
2467 | /* Clear a port's read buffers, returning the contents. */ |
2468 | SCM_DEFINE (scm_drain_input, "drain-input", 1, 0, 0, | |
1bbd0b84 | 2469 | (SCM port), |
19b8d12b AW |
2470 | "This procedure clears a port's input buffers, similar\n" |
2471 | "to the way that force-output clears the output buffer. The\n" | |
2472 | "contents of the buffers are returned as a single string, e.g.,\n" | |
a150979d | 2473 | "\n" |
19b8d12b AW |
2474 | "@lisp\n" |
2475 | "(define p (open-input-file ...))\n" | |
2476 | "(drain-input p) => empty string, nothing buffered yet.\n" | |
2477 | "(unread-char (read-char p) p)\n" | |
2478 | "(drain-input p) => initial chars from p, up to the buffer size.\n" | |
2479 | "@end lisp\n\n" | |
2480 | "Draining the buffers may be useful for cleanly finishing\n" | |
2481 | "buffered I/O so that the file descriptor can be used directly\n" | |
2482 | "for further input.") | |
2483 | #define FUNC_NAME s_scm_drain_input | |
0f2d19dd | 2484 | { |
19b8d12b AW |
2485 | SCM result; |
2486 | char *data; | |
2487 | scm_t_port *pt; | |
2488 | long count; | |
0f2d19dd | 2489 | |
19b8d12b AW |
2490 | SCM_VALIDATE_OPINPORT (1, port); |
2491 | pt = SCM_PTAB_ENTRY (port); | |
2492 | ||
2493 | count = pt->read_end - pt->read_pos; | |
2494 | if (pt->read_buf == pt->putback_buf) | |
2495 | count += pt->saved_read_end - pt->saved_read_pos; | |
2496 | ||
2497 | if (count) | |
2498 | { | |
2499 | result = scm_i_make_string (count, &data, 0); | |
2500 | scm_take_from_input_buffers (port, data, count); | |
2501 | } | |
2502 | else | |
2503 | result = scm_nullstr; | |
2504 | ||
2505 | return result; | |
d043d8c2 | 2506 | } |
1bbd0b84 | 2507 | #undef FUNC_NAME |
d043d8c2 | 2508 | |
19b8d12b | 2509 | void |
4251ae2e | 2510 | scm_end_input_unlocked (SCM port) |
0f2d19dd | 2511 | { |
19b8d12b AW |
2512 | long offset; |
2513 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
2514 | ||
f6f4feb0 | 2515 | scm_i_clear_pending_eof (port); |
19b8d12b AW |
2516 | if (pt->read_buf == pt->putback_buf) |
2517 | { | |
2518 | offset = pt->read_end - pt->read_pos; | |
2519 | pt->read_buf = pt->saved_read_buf; | |
2520 | pt->read_pos = pt->saved_read_pos; | |
2521 | pt->read_end = pt->saved_read_end; | |
2522 | pt->read_buf_size = pt->saved_read_buf_size; | |
2523 | } | |
2524 | else | |
2525 | offset = 0; | |
2526 | ||
2527 | SCM_PORT_DESCRIPTOR (port)->end_input (port, offset); | |
0f2d19dd JB |
2528 | } |
2529 | ||
4251ae2e AW |
2530 | void |
2531 | scm_end_input (SCM port) | |
2532 | { | |
92c0ebac AW |
2533 | scm_i_pthread_mutex_t *lock; |
2534 | scm_c_lock_port (port, &lock); | |
4251ae2e | 2535 | scm_end_input_unlocked (port); |
92c0ebac AW |
2536 | if (lock) |
2537 | scm_i_pthread_mutex_unlock (lock); | |
2538 | ||
4251ae2e AW |
2539 | } |
2540 | ||
19b8d12b AW |
2541 | SCM_DEFINE (scm_force_output, "force-output", 0, 1, 0, |
2542 | (SCM port), | |
2543 | "Flush the specified output port, or the current output port if @var{port}\n" | |
2544 | "is omitted. The current output buffer contents are passed to the\n" | |
2545 | "underlying port implementation (e.g., in the case of fports, the\n" | |
2546 | "data will be written to the file and the output buffer will be cleared.)\n" | |
2547 | "It has no effect on an unbuffered port.\n\n" | |
2548 | "The return value is unspecified.") | |
2549 | #define FUNC_NAME s_scm_force_output | |
d043d8c2 | 2550 | { |
19b8d12b AW |
2551 | if (SCM_UNBNDP (port)) |
2552 | port = scm_current_output_port (); | |
2553 | else | |
2554 | { | |
2555 | port = SCM_COERCE_OUTPORT (port); | |
2556 | SCM_VALIDATE_OPOUTPORT (1, port); | |
2557 | } | |
4251ae2e | 2558 | scm_flush_unlocked (port); |
564478fd | 2559 | return SCM_UNSPECIFIED; |
d043d8c2 | 2560 | } |
1bbd0b84 | 2561 | #undef FUNC_NAME |
d043d8c2 | 2562 | |
19b8d12b | 2563 | void |
4251ae2e | 2564 | scm_flush_unlocked (SCM port) |
0f2d19dd | 2565 | { |
19b8d12b | 2566 | SCM_PORT_DESCRIPTOR (port)->flush (port); |
0f2d19dd JB |
2567 | } |
2568 | ||
4251ae2e AW |
2569 | void |
2570 | scm_flush (SCM port) | |
2571 | { | |
92c0ebac AW |
2572 | scm_i_pthread_mutex_t *lock; |
2573 | scm_c_lock_port (port, &lock); | |
4251ae2e | 2574 | scm_flush_unlocked (port); |
92c0ebac AW |
2575 | if (lock) |
2576 | scm_i_pthread_mutex_unlock (lock); | |
2577 | ||
4251ae2e AW |
2578 | } |
2579 | ||
f6f4feb0 MW |
2580 | int |
2581 | scm_fill_input_unlocked (SCM port) | |
337edc59 | 2582 | { |
f6f4feb0 MW |
2583 | return scm_i_fill_input_unlocked (port); |
2584 | } | |
337edc59 | 2585 | |
d14af9f2 | 2586 | |
19b8d12b | 2587 | \f |
d6a6989e | 2588 | |
19b8d12b | 2589 | /* Output. */ |
889975e5 | 2590 | |
0607ebbf AW |
2591 | void |
2592 | scm_putc (char c, SCM port) | |
2593 | { | |
92c0ebac AW |
2594 | scm_i_pthread_mutex_t *lock; |
2595 | scm_c_lock_port (port, &lock); | |
0607ebbf | 2596 | scm_putc_unlocked (c, port); |
92c0ebac AW |
2597 | if (lock) |
2598 | scm_i_pthread_mutex_unlock (lock); | |
2599 | ||
0607ebbf AW |
2600 | } |
2601 | ||
2602 | void | |
2603 | scm_puts (const char *s, SCM port) | |
2604 | { | |
92c0ebac AW |
2605 | scm_i_pthread_mutex_t *lock; |
2606 | scm_c_lock_port (port, &lock); | |
0607ebbf | 2607 | scm_puts_unlocked (s, port); |
92c0ebac AW |
2608 | if (lock) |
2609 | scm_i_pthread_mutex_unlock (lock); | |
2610 | ||
0607ebbf AW |
2611 | } |
2612 | ||
19b8d12b AW |
2613 | /* scm_c_write |
2614 | * | |
2615 | * Used by an application to write arbitrary number of bytes to an SCM | |
2616 | * port. Similar semantics as libc write. However, unlike libc | |
2617 | * write, scm_c_write writes the requested number of bytes and has no | |
2618 | * return value. | |
2619 | * | |
2620 | * Warning: Doesn't update port line and column counts! | |
2621 | */ | |
9d9c66ba | 2622 | void |
f209aeee | 2623 | scm_c_write_unlocked (SCM port, const void *ptr, size_t size) |
19b8d12b | 2624 | #define FUNC_NAME "scm_c_write" |
9d9c66ba | 2625 | { |
19b8d12b AW |
2626 | scm_t_port *pt; |
2627 | scm_t_ptob_descriptor *ptob; | |
9d9c66ba | 2628 | |
19b8d12b | 2629 | SCM_VALIDATE_OPOUTPORT (1, port); |
9d9c66ba | 2630 | |
19b8d12b AW |
2631 | pt = SCM_PTAB_ENTRY (port); |
2632 | ptob = SCM_PORT_DESCRIPTOR (port); | |
9d9c66ba | 2633 | |
19b8d12b | 2634 | if (pt->rw_active == SCM_PORT_READ) |
4251ae2e | 2635 | scm_end_input_unlocked (port); |
19b8d12b AW |
2636 | |
2637 | ptob->write (port, ptr, size); | |
2638 | ||
2639 | if (pt->rw_random) | |
2640 | pt->rw_active = SCM_PORT_WRITE; | |
889975e5 | 2641 | } |
19b8d12b | 2642 | #undef FUNC_NAME |
889975e5 | 2643 | |
f209aeee AW |
2644 | void |
2645 | scm_c_write (SCM port, const void *ptr, size_t size) | |
2646 | { | |
92c0ebac AW |
2647 | scm_i_pthread_mutex_t *lock; |
2648 | scm_c_lock_port (port, &lock); | |
f209aeee | 2649 | scm_c_write_unlocked (port, ptr, size); |
92c0ebac AW |
2650 | if (lock) |
2651 | scm_i_pthread_mutex_unlock (lock); | |
2652 | ||
f209aeee AW |
2653 | } |
2654 | ||
19b8d12b AW |
2655 | /* scm_lfwrite |
2656 | * | |
2657 | * This function differs from scm_c_write; it updates port line and | |
2658 | * column. */ | |
889975e5 | 2659 | void |
f209aeee | 2660 | scm_lfwrite_unlocked (const char *ptr, size_t size, SCM port) |
889975e5 | 2661 | { |
19b8d12b AW |
2662 | scm_t_port *pt = SCM_PTAB_ENTRY (port); |
2663 | scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (port); | |
f4bc4e59 | 2664 | |
19b8d12b | 2665 | if (pt->rw_active == SCM_PORT_READ) |
4251ae2e | 2666 | scm_end_input_unlocked (port); |
f4bc4e59 | 2667 | |
19b8d12b | 2668 | ptob->write (port, ptr, size); |
f4bc4e59 | 2669 | |
19b8d12b AW |
2670 | for (; size; ptr++, size--) |
2671 | update_port_lf ((scm_t_wchar) (unsigned char) *ptr, port); | |
d9544bf0 | 2672 | |
19b8d12b AW |
2673 | if (pt->rw_random) |
2674 | pt->rw_active = SCM_PORT_WRITE; | |
2675 | } | |
f4bc4e59 | 2676 | |
f209aeee AW |
2677 | void |
2678 | scm_lfwrite (const char *ptr, size_t size, SCM port) | |
2679 | { | |
92c0ebac AW |
2680 | scm_i_pthread_mutex_t *lock; |
2681 | scm_c_lock_port (port, &lock); | |
f209aeee | 2682 | scm_lfwrite_unlocked (ptr, size, port); |
92c0ebac AW |
2683 | if (lock) |
2684 | scm_i_pthread_mutex_unlock (lock); | |
2685 | ||
f209aeee AW |
2686 | } |
2687 | ||
19b8d12b AW |
2688 | /* Write STR to PORT from START inclusive to END exclusive. */ |
2689 | void | |
2690 | scm_lfwrite_substr (SCM str, size_t start, size_t end, SCM port) | |
2691 | { | |
2692 | scm_t_port *pt = SCM_PTAB_ENTRY (port); | |
f4bc4e59 | 2693 | |
19b8d12b | 2694 | if (pt->rw_active == SCM_PORT_READ) |
4251ae2e | 2695 | scm_end_input_unlocked (port); |
f4bc4e59 | 2696 | |
19b8d12b AW |
2697 | if (end == (size_t) -1) |
2698 | end = scm_i_string_length (str); | |
f4bc4e59 | 2699 | |
fa980bcc | 2700 | scm_i_display_substring (str, start, end, port); |
f4bc4e59 | 2701 | |
19b8d12b AW |
2702 | if (pt->rw_random) |
2703 | pt->rw_active = SCM_PORT_WRITE; | |
889975e5 MG |
2704 | } |
2705 | ||
19b8d12b AW |
2706 | |
2707 | \f | |
2708 | ||
2709 | /* Querying and setting positions, and character availability. */ | |
2710 | ||
2711 | SCM_DEFINE (scm_char_ready_p, "char-ready?", 0, 1, 0, | |
889975e5 | 2712 | (SCM port), |
19b8d12b AW |
2713 | "Return @code{#t} if a character is ready on input @var{port}\n" |
2714 | "and return @code{#f} otherwise. If @code{char-ready?} returns\n" | |
2715 | "@code{#t} then the next @code{read-char} operation on\n" | |
2716 | "@var{port} is guaranteed not to hang. If @var{port} is a file\n" | |
2717 | "port at end of file then @code{char-ready?} returns @code{#t}.\n" | |
2718 | "\n" | |
2719 | "@code{char-ready?} exists to make it possible for a\n" | |
2720 | "program to accept characters from interactive ports without\n" | |
2721 | "getting stuck waiting for input. Any input editors associated\n" | |
2722 | "with such ports must make sure that characters whose existence\n" | |
2723 | "has been asserted by @code{char-ready?} cannot be rubbed out.\n" | |
2724 | "If @code{char-ready?} were to return @code{#f} at end of file,\n" | |
2725 | "a port at end of file would be indistinguishable from an\n" | |
2726 | "interactive port that has no ready characters.") | |
2727 | #define FUNC_NAME s_scm_char_ready_p | |
889975e5 MG |
2728 | { |
2729 | scm_t_port *pt; | |
889975e5 | 2730 | |
19b8d12b AW |
2731 | if (SCM_UNBNDP (port)) |
2732 | port = scm_current_input_port (); | |
2733 | /* It's possible to close the current input port, so validate even in | |
2734 | this case. */ | |
2735 | SCM_VALIDATE_OPINPORT (1, port); | |
889975e5 MG |
2736 | |
2737 | pt = SCM_PTAB_ENTRY (port); | |
19b8d12b AW |
2738 | |
2739 | /* if the current read buffer is filled, or the | |
2740 | last pushed-back char has been read and the saved buffer is | |
2741 | filled, result is true. */ | |
2742 | if (pt->read_pos < pt->read_end | |
2743 | || (pt->read_buf == pt->putback_buf | |
2744 | && pt->saved_read_pos < pt->saved_read_end)) | |
2745 | return SCM_BOOL_T; | |
889975e5 | 2746 | else |
19b8d12b AW |
2747 | { |
2748 | scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (port); | |
2749 | ||
2750 | if (ptob->input_waiting) | |
2751 | return scm_from_bool(ptob->input_waiting (port)); | |
2752 | else | |
2753 | return SCM_BOOL_T; | |
2754 | } | |
889975e5 MG |
2755 | } |
2756 | #undef FUNC_NAME | |
d6a6989e | 2757 | |
19b8d12b AW |
2758 | SCM_DEFINE (scm_seek, "seek", 3, 0, 0, |
2759 | (SCM fd_port, SCM offset, SCM whence), | |
0858753e | 2760 | "Sets the current position of @var{fd_port} to the integer\n" |
19b8d12b AW |
2761 | "@var{offset}, which is interpreted according to the value of\n" |
2762 | "@var{whence}.\n" | |
2763 | "\n" | |
2764 | "One of the following variables should be supplied for\n" | |
2765 | "@var{whence}:\n" | |
2766 | "@defvar SEEK_SET\n" | |
2767 | "Seek from the beginning of the file.\n" | |
2768 | "@end defvar\n" | |
2769 | "@defvar SEEK_CUR\n" | |
2770 | "Seek from the current position.\n" | |
2771 | "@end defvar\n" | |
2772 | "@defvar SEEK_END\n" | |
2773 | "Seek from the end of the file.\n" | |
2774 | "@end defvar\n" | |
0858753e | 2775 | "If @var{fd_port} is a file descriptor, the underlying system\n" |
19b8d12b AW |
2776 | "call is @code{lseek}. @var{port} may be a string port.\n" |
2777 | "\n" | |
2778 | "The value returned is the new position in the file. This means\n" | |
2779 | "that the current position of a port can be obtained using:\n" | |
2780 | "@lisp\n" | |
2781 | "(seek port 0 SEEK_CUR)\n" | |
2782 | "@end lisp") | |
2783 | #define FUNC_NAME s_scm_seek | |
889975e5 | 2784 | { |
19b8d12b | 2785 | int how; |
889975e5 | 2786 | |
19b8d12b | 2787 | fd_port = SCM_COERCE_OUTPORT (fd_port); |
889975e5 | 2788 | |
19b8d12b AW |
2789 | how = scm_to_int (whence); |
2790 | if (how != SEEK_SET && how != SEEK_CUR && how != SEEK_END) | |
2791 | SCM_OUT_OF_RANGE (3, whence); | |
da288f50 | 2792 | |
19b8d12b AW |
2793 | if (SCM_OPPORTP (fd_port)) |
2794 | { | |
f6f4feb0 | 2795 | scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (fd_port); |
19b8d12b AW |
2796 | scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (fd_port); |
2797 | off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset); | |
2798 | off_t_or_off64_t rv; | |
2799 | ||
2800 | if (!ptob->seek) | |
2801 | SCM_MISC_ERROR ("port is not seekable", | |
2802 | scm_cons (fd_port, SCM_EOL)); | |
2803 | else | |
f6f4feb0 | 2804 | rv = ptob->seek (fd_port, off, how); |
889975e5 | 2805 | |
f6f4feb0 MW |
2806 | /* Set stream-start flags according to new position. */ |
2807 | pti->at_stream_start_for_bom_read = (rv == 0); | |
2808 | pti->at_stream_start_for_bom_write = (rv == 0); | |
b22e94db | 2809 | |
f6f4feb0 | 2810 | scm_i_clear_pending_eof (fd_port); |
889975e5 | 2811 | |
19b8d12b AW |
2812 | return scm_from_off_t_or_off64_t (rv); |
2813 | } | |
2814 | else /* file descriptor?. */ | |
2815 | { | |
2816 | off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset); | |
2817 | off_t_or_off64_t rv; | |
2818 | rv = lseek_or_lseek64 (scm_to_int (fd_port), off, how); | |
2819 | if (rv == -1) | |
2820 | SCM_SYSERROR; | |
2821 | return scm_from_off_t_or_off64_t (rv); | |
2822 | } | |
889975e5 MG |
2823 | } |
2824 | #undef FUNC_NAME | |
2825 | ||
19b8d12b AW |
2826 | #ifndef O_BINARY |
2827 | #define O_BINARY 0 | |
2828 | #endif | |
889975e5 | 2829 | |
19b8d12b AW |
2830 | /* Mingw has ftruncate(), perhaps implemented above using chsize, but |
2831 | doesn't have the filename version truncate(), hence this code. */ | |
2832 | #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE | |
2833 | static int | |
2834 | truncate (const char *file, off_t length) | |
889975e5 | 2835 | { |
19b8d12b AW |
2836 | int ret, fdes; |
2837 | ||
2838 | fdes = open (file, O_BINARY | O_WRONLY); | |
2839 | if (fdes == -1) | |
2840 | return -1; | |
2841 | ||
2842 | ret = ftruncate (fdes, length); | |
2843 | if (ret == -1) | |
889975e5 | 2844 | { |
19b8d12b AW |
2845 | int save_errno = errno; |
2846 | close (fdes); | |
2847 | errno = save_errno; | |
2848 | return -1; | |
889975e5 | 2849 | } |
19b8d12b AW |
2850 | |
2851 | return close (fdes); | |
889975e5 | 2852 | } |
19b8d12b | 2853 | #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */ |
889975e5 | 2854 | |
19b8d12b AW |
2855 | SCM_DEFINE (scm_truncate_file, "truncate-file", 1, 1, 0, |
2856 | (SCM object, SCM length), | |
0858753e AW |
2857 | "Truncate file @var{object} to @var{length} bytes. @var{object}\n" |
2858 | "can be a filename string, a port object, or an integer file\n" | |
2859 | "descriptor.\n" | |
19b8d12b AW |
2860 | "The return value is unspecified.\n" |
2861 | "\n" | |
2862 | "For a port or file descriptor @var{length} can be omitted, in\n" | |
2863 | "which case the file is truncated at the current position (per\n" | |
2864 | "@code{ftell} above).\n" | |
2865 | "\n" | |
2866 | "On most systems a file can be extended by giving a length\n" | |
2867 | "greater than the current size, but this is not mandatory in the\n" | |
2868 | "POSIX standard.") | |
2869 | #define FUNC_NAME s_scm_truncate_file | |
889975e5 | 2870 | { |
19b8d12b AW |
2871 | int rv; |
2872 | ||
2873 | /* "object" can be a port, fdes or filename. | |
2874 | ||
2875 | Negative "length" makes no sense, but it's left to truncate() or | |
2876 | ftruncate() to give back an error for that (normally EINVAL). | |
2877 | */ | |
2878 | ||
2879 | if (SCM_UNBNDP (length)) | |
889975e5 | 2880 | { |
19b8d12b AW |
2881 | /* must supply length if object is a filename. */ |
2882 | if (scm_is_string (object)) | |
2883 | SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL); | |
2884 | ||
2885 | length = scm_seek (object, SCM_INUM0, scm_from_int (SEEK_CUR)); | |
2886 | } | |
2887 | ||
2888 | object = SCM_COERCE_OUTPORT (object); | |
2889 | if (scm_is_integer (object)) | |
2890 | { | |
2891 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); | |
2892 | SCM_SYSCALL (rv = ftruncate_or_ftruncate64 (scm_to_int (object), | |
2893 | c_length)); | |
2894 | } | |
2895 | else if (SCM_OPOUTPORTP (object)) | |
2896 | { | |
2897 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); | |
2898 | scm_t_port *pt = SCM_PTAB_ENTRY (object); | |
2899 | scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (object); | |
b22e94db | 2900 | |
19b8d12b AW |
2901 | if (!ptob->truncate) |
2902 | SCM_MISC_ERROR ("port is not truncatable", SCM_EOL); | |
889975e5 | 2903 | |
f6f4feb0 | 2904 | scm_i_clear_pending_eof (object); |
19b8d12b | 2905 | if (pt->rw_active == SCM_PORT_READ) |
4251ae2e | 2906 | scm_end_input_unlocked (object); |
19b8d12b AW |
2907 | else if (pt->rw_active == SCM_PORT_WRITE) |
2908 | ptob->flush (object); | |
889975e5 | 2909 | |
19b8d12b AW |
2910 | ptob->truncate (object, c_length); |
2911 | rv = 0; | |
889975e5 MG |
2912 | } |
2913 | else | |
2914 | { | |
19b8d12b AW |
2915 | off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length); |
2916 | char *str = scm_to_locale_string (object); | |
2917 | int eno; | |
2918 | SCM_SYSCALL (rv = truncate_or_truncate64 (str, c_length)); | |
2919 | eno = errno; | |
2920 | free (str); | |
2921 | errno = eno; | |
889975e5 | 2922 | } |
19b8d12b AW |
2923 | if (rv == -1) |
2924 | SCM_SYSERROR; | |
2925 | return SCM_UNSPECIFIED; | |
889975e5 | 2926 | } |
19b8d12b | 2927 | #undef FUNC_NAME |
889975e5 | 2928 | |
19b8d12b AW |
2929 | SCM_DEFINE (scm_port_line, "port-line", 1, 0, 0, |
2930 | (SCM port), | |
2931 | "Return the current line number for @var{port}.\n" | |
889975e5 | 2932 | "\n" |
19b8d12b AW |
2933 | "The first line of a file is 0. But you might want to add 1\n" |
2934 | "when printing line numbers, since starting from 1 is\n" | |
2935 | "traditional in error messages, and likely to be more natural to\n" | |
2936 | "non-programmers.") | |
2937 | #define FUNC_NAME s_scm_port_line | |
889975e5 | 2938 | { |
19b8d12b AW |
2939 | port = SCM_COERCE_OUTPORT (port); |
2940 | SCM_VALIDATE_OPENPORT (1, port); | |
2941 | return scm_from_long (SCM_LINUM (port)); | |
889975e5 MG |
2942 | } |
2943 | #undef FUNC_NAME | |
2944 | ||
19b8d12b AW |
2945 | SCM_DEFINE (scm_set_port_line_x, "set-port-line!", 2, 0, 0, |
2946 | (SCM port, SCM line), | |
2947 | "Set the current line number for @var{port} to @var{line}. The\n" | |
2948 | "first line of a file is 0.") | |
2949 | #define FUNC_NAME s_scm_set_port_line_x | |
889975e5 | 2950 | { |
19b8d12b AW |
2951 | port = SCM_COERCE_OUTPORT (port); |
2952 | SCM_VALIDATE_OPENPORT (1, port); | |
2953 | SCM_PTAB_ENTRY (port)->line_number = scm_to_long (line); | |
2954 | return SCM_UNSPECIFIED; | |
2955 | } | |
2956 | #undef FUNC_NAME | |
889975e5 | 2957 | |
19b8d12b AW |
2958 | SCM_DEFINE (scm_port_column, "port-column", 1, 0, 0, |
2959 | (SCM port), | |
2960 | "Return the current column number of @var{port}.\n" | |
2961 | "If the number is\n" | |
2962 | "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n" | |
2963 | "- i.e. the first character of the first line is line 0, column 0.\n" | |
2964 | "(However, when you display a file position, for example in an error\n" | |
2965 | "message, we recommend you add 1 to get 1-origin integers. This is\n" | |
2966 | "because lines and column numbers traditionally start with 1, and that is\n" | |
2967 | "what non-programmers will find most natural.)") | |
2968 | #define FUNC_NAME s_scm_port_column | |
2969 | { | |
2970 | port = SCM_COERCE_OUTPORT (port); | |
2971 | SCM_VALIDATE_OPENPORT (1, port); | |
2972 | return scm_from_int (SCM_COL (port)); | |
2973 | } | |
2974 | #undef FUNC_NAME | |
889975e5 | 2975 | |
19b8d12b AW |
2976 | SCM_DEFINE (scm_set_port_column_x, "set-port-column!", 2, 0, 0, |
2977 | (SCM port, SCM column), | |
2978 | "Set the current column of @var{port}. Before reading the first\n" | |
2979 | "character on a line the column should be 0.") | |
2980 | #define FUNC_NAME s_scm_set_port_column_x | |
2981 | { | |
2982 | port = SCM_COERCE_OUTPORT (port); | |
2983 | SCM_VALIDATE_OPENPORT (1, port); | |
2984 | SCM_PTAB_ENTRY (port)->column_number = scm_to_int (column); | |
2985 | return SCM_UNSPECIFIED; | |
2986 | } | |
2987 | #undef FUNC_NAME | |
889975e5 | 2988 | |
19b8d12b AW |
2989 | SCM_DEFINE (scm_port_filename, "port-filename", 1, 0, 0, |
2990 | (SCM port), | |
2991 | "Return the filename associated with @var{port}, or @code{#f}\n" | |
2992 | "if no filename is associated with the port.") | |
2993 | #define FUNC_NAME s_scm_port_filename | |
2994 | { | |
2995 | port = SCM_COERCE_OUTPORT (port); | |
2996 | SCM_VALIDATE_OPENPORT (1, port); | |
2997 | return SCM_FILENAME (port); | |
2998 | } | |
2999 | #undef FUNC_NAME | |
889975e5 | 3000 | |
19b8d12b AW |
3001 | SCM_DEFINE (scm_set_port_filename_x, "set-port-filename!", 2, 0, 0, |
3002 | (SCM port, SCM filename), | |
3003 | "Change the filename associated with @var{port}, using the current input\n" | |
3004 | "port if none is specified. Note that this does not change the port's\n" | |
3005 | "source of data, but only the value that is returned by\n" | |
3006 | "@code{port-filename} and reported in diagnostic output.") | |
3007 | #define FUNC_NAME s_scm_set_port_filename_x | |
3008 | { | |
3009 | port = SCM_COERCE_OUTPORT (port); | |
3010 | SCM_VALIDATE_OPENPORT (1, port); | |
3011 | /* We allow the user to set the filename to whatever he likes. */ | |
3012 | SCM_SET_FILENAME (port, filename); | |
889975e5 MG |
3013 | return SCM_UNSPECIFIED; |
3014 | } | |
3015 | #undef FUNC_NAME | |
3016 | ||
3017 | ||
19b8d12b AW |
3018 | \f |
3019 | ||
3020 | /* Implementation helpers for port printing functions. */ | |
889975e5 | 3021 | |
f12733c9 MD |
3022 | void |
3023 | scm_print_port_mode (SCM exp, SCM port) | |
3024 | { | |
0607ebbf | 3025 | scm_puts_unlocked (SCM_CLOSEDP (exp) |
f12733c9 | 3026 | ? "closed: " |
f9a64404 DH |
3027 | : (SCM_RDNG & SCM_CELL_WORD_0 (exp) |
3028 | ? (SCM_WRTNG & SCM_CELL_WORD_0 (exp) | |
f12733c9 MD |
3029 | ? "input-output: " |
3030 | : "input: ") | |
f9a64404 | 3031 | : (SCM_WRTNG & SCM_CELL_WORD_0 (exp) |
f12733c9 MD |
3032 | ? "output: " |
3033 | : "bogus: ")), | |
3034 | port); | |
3035 | } | |
1cc91f1b | 3036 | |
f12733c9 | 3037 | int |
e81d98ec | 3038 | scm_port_print (SCM exp, SCM port, scm_print_state *pstate SCM_UNUSED) |
0f2d19dd | 3039 | { |
f12733c9 MD |
3040 | char *type = SCM_PTOBNAME (SCM_PTOBNUM (exp)); |
3041 | if (!type) | |
3042 | type = "port"; | |
0607ebbf | 3043 | scm_puts_unlocked ("#<", port); |
f12733c9 | 3044 | scm_print_port_mode (exp, port); |
0607ebbf AW |
3045 | scm_puts_unlocked (type, port); |
3046 | scm_putc_unlocked (' ', port); | |
0345e278 | 3047 | scm_uintprint (SCM_CELL_WORD_1 (exp), 16, port); |
0607ebbf | 3048 | scm_putc_unlocked ('>', port); |
f12733c9 | 3049 | return 1; |
0f2d19dd JB |
3050 | } |
3051 | ||
19b8d12b AW |
3052 | |
3053 | \f | |
3054 | ||
3055 | /* Iterating over all ports. */ | |
3056 | ||
3057 | struct for_each_data | |
3058 | { | |
3059 | void (*proc) (void *data, SCM p); | |
3060 | void *data; | |
3061 | }; | |
3062 | ||
3063 | static SCM | |
3064 | for_each_trampoline (void *data, SCM port, SCM result) | |
3065 | { | |
3066 | struct for_each_data *d = data; | |
3067 | ||
3068 | d->proc (d->data, port); | |
3069 | ||
3070 | return result; | |
3071 | } | |
3072 | ||
3073 | void | |
3074 | scm_c_port_for_each (void (*proc)(void *data, SCM p), void *data) | |
3075 | { | |
3076 | struct for_each_data d; | |
3077 | ||
3078 | d.proc = proc; | |
3079 | d.data = data; | |
3080 | ||
3081 | scm_c_weak_set_fold (for_each_trampoline, &d, SCM_EOL, | |
3082 | scm_i_port_weak_set); | |
3083 | } | |
3084 | ||
3085 | static void | |
3086 | scm_for_each_trampoline (void *data, SCM port) | |
3087 | { | |
3088 | scm_call_1 (SCM_PACK_POINTER (data), port); | |
3089 | } | |
3090 | ||
3091 | SCM_DEFINE (scm_port_for_each, "port-for-each", 1, 0, 0, | |
3092 | (SCM proc), | |
3093 | "Apply @var{proc} to each port in the Guile port table\n" | |
3094 | "in turn. The return value is unspecified. More specifically,\n" | |
3095 | "@var{proc} is applied exactly once to every port that exists\n" | |
0858753e AW |
3096 | "in the system at the time @code{port-for-each} is invoked.\n" |
3097 | "Changes to the port table while @code{port-for-each} is running\n" | |
3098 | "have no effect as far as @code{port-for-each} is concerned.") | |
19b8d12b AW |
3099 | #define FUNC_NAME s_scm_port_for_each |
3100 | { | |
3101 | SCM_VALIDATE_PROC (1, proc); | |
3102 | ||
3103 | scm_c_port_for_each (scm_for_each_trampoline, SCM_UNPACK_POINTER (proc)); | |
3104 | ||
3105 | return SCM_UNSPECIFIED; | |
3106 | } | |
3107 | #undef FUNC_NAME | |
3108 | ||
3109 | static void | |
3110 | flush_output_port (void *closure, SCM port) | |
3111 | { | |
3112 | if (SCM_OPOUTPORTP (port)) | |
4251ae2e | 3113 | scm_flush_unlocked (port); |
19b8d12b AW |
3114 | } |
3115 | ||
3116 | SCM_DEFINE (scm_flush_all_ports, "flush-all-ports", 0, 0, 0, | |
3117 | (), | |
3118 | "Equivalent to calling @code{force-output} on\n" | |
3119 | "all open output ports. The return value is unspecified.") | |
3120 | #define FUNC_NAME s_scm_flush_all_ports | |
3121 | { | |
3122 | scm_c_port_for_each (&flush_output_port, NULL); | |
3123 | return SCM_UNSPECIFIED; | |
3124 | } | |
3125 | #undef FUNC_NAME | |
3126 | ||
3127 | ||
0f2d19dd | 3128 | \f |
ee149d03 | 3129 | |
d68fee48 | 3130 | /* Void ports. */ |
0f2d19dd | 3131 | |
92c2555f | 3132 | scm_t_bits scm_tc16_void_port = 0; |
0f2d19dd | 3133 | |
e81d98ec | 3134 | static int fill_input_void_port (SCM port SCM_UNUSED) |
283a1a0e | 3135 | { |
70df8af6 | 3136 | return EOF; |
283a1a0e GH |
3137 | } |
3138 | ||
31703ab8 | 3139 | static void |
e81d98ec DH |
3140 | write_void_port (SCM port SCM_UNUSED, |
3141 | const void *data SCM_UNUSED, | |
3142 | size_t size SCM_UNUSED) | |
31703ab8 GH |
3143 | { |
3144 | } | |
3145 | ||
d617ee18 MV |
3146 | static SCM |
3147 | scm_i_void_port (long mode_bits) | |
0f2d19dd | 3148 | { |
2721f918 AW |
3149 | SCM ret; |
3150 | ||
3151 | ret = scm_c_make_port (scm_tc16_void_port, mode_bits, 0); | |
da220f27 | 3152 | |
2721f918 | 3153 | scm_port_non_buffer (SCM_PTAB_ENTRY (ret)); |
402788a9 | 3154 | |
2721f918 | 3155 | return ret; |
0f2d19dd JB |
3156 | } |
3157 | ||
d617ee18 MV |
3158 | SCM |
3159 | scm_void_port (char *mode_str) | |
3160 | { | |
3161 | return scm_i_void_port (scm_mode_bits (mode_str)); | |
3162 | } | |
3163 | ||
a1ec6916 | 3164 | SCM_DEFINE (scm_sys_make_void_port, "%make-void-port", 1, 0, 0, |
1bbd0b84 | 3165 | (SCM mode), |
70df8af6 | 3166 | "Create and return a new void port. A void port acts like\n" |
bb2c02f2 | 3167 | "@file{/dev/null}. The @var{mode} argument\n" |
70df8af6 | 3168 | "specifies the input/output modes for this port: see the\n" |
b380b885 | 3169 | "documentation for @code{open-file} in @ref{File Ports}.") |
1bbd0b84 | 3170 | #define FUNC_NAME s_scm_sys_make_void_port |
0f2d19dd | 3171 | { |
d617ee18 | 3172 | return scm_i_void_port (scm_i_mode_bits (mode)); |
0f2d19dd | 3173 | } |
1bbd0b84 | 3174 | #undef FUNC_NAME |
0f2d19dd | 3175 | |
19b8d12b | 3176 | |
0f2d19dd | 3177 | \f |
19b8d12b | 3178 | |
89545eba | 3179 | /* Initialization. */ |
1cc91f1b | 3180 | |
0f2d19dd JB |
3181 | void |
3182 | scm_init_ports () | |
0f2d19dd | 3183 | { |
840ae05d | 3184 | /* lseek() symbols. */ |
e11e83f3 MV |
3185 | scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET)); |
3186 | scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR)); | |
3187 | scm_c_define ("SEEK_END", scm_from_int (SEEK_END)); | |
840ae05d | 3188 | |
70df8af6 GH |
3189 | scm_tc16_void_port = scm_make_port_type ("void", fill_input_void_port, |
3190 | write_void_port); | |
9de87eea | 3191 | |
f39448c5 AW |
3192 | cur_inport_fluid = scm_make_fluid (); |
3193 | cur_outport_fluid = scm_make_fluid (); | |
3194 | cur_errport_fluid = scm_make_fluid (); | |
3195 | cur_loadport_fluid = scm_make_fluid (); | |
9de87eea | 3196 | |
2721f918 | 3197 | scm_i_port_weak_set = scm_c_make_weak_set (31); |
d6a6989e | 3198 | |
a0599745 | 3199 | #include "libguile/ports.x" |
889975e5 | 3200 | |
d6a6989e | 3201 | /* Use Latin-1 as the default port encoding. */ |
c81c2ad3 AW |
3202 | SCM_VARIABLE_SET (default_port_encoding_var, |
3203 | scm_make_fluid_with_default (SCM_BOOL_F)); | |
889975e5 | 3204 | scm_port_encoding_init = 1; |
d6a6989e | 3205 | |
b22e94db LC |
3206 | SCM_VARIABLE_SET (default_conversion_strategy_var, |
3207 | scm_make_fluid_with_default (sym_substitute)); | |
889975e5 | 3208 | scm_conversion_strategy_init = 1; |
b22e94db | 3209 | |
9670f238 AW |
3210 | /* These bindings are used when boot-9 turns `current-input-port' et |
3211 | al into parameters. They are then removed from the guile module. */ | |
3212 | scm_c_define ("%current-input-port-fluid", cur_inport_fluid); | |
3213 | scm_c_define ("%current-output-port-fluid", cur_outport_fluid); | |
3214 | scm_c_define ("%current-error-port-fluid", cur_errport_fluid); | |
0f2d19dd | 3215 | } |
89e00824 ML |
3216 | |
3217 | /* | |
3218 | Local Variables: | |
3219 | c-file-style: "gnu" | |
3220 | End: | |
3221 | */ |