fix a couple leaks in ports.c. thanks valgrind!
[bpt/guile.git] / libguile / ports.c
1 /* Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004,
2 * 2006, 2007, 2008, 2009, 2010, 2011 Free Software Foundation, Inc.
3 *
4 * This library is free software; you can redistribute it and/or
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.
8 *
9 * This library is distributed in the hope that it will be useful, but
10 * WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
13 *
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
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
17 * 02110-1301 USA
18 */
19
20
21 \f
22 /* Headers. */
23
24 #define _LARGEFILE64_SOURCE /* ask for stat64 etc */
25
26 #ifdef HAVE_CONFIG_H
27 # include <config.h>
28 #endif
29
30 #include <stdio.h>
31 #include <errno.h>
32 #include <fcntl.h> /* for chsize on mingw */
33 #include <assert.h>
34 #include <iconv.h>
35 #include <uniconv.h>
36 #include <unistr.h>
37 #include <striconveh.h>
38
39 #include <assert.h>
40
41 #include "libguile/_scm.h"
42 #include "libguile/async.h"
43 #include "libguile/deprecation.h"
44 #include "libguile/eval.h"
45 #include "libguile/fports.h" /* direct access for seek and truncate */
46 #include "libguile/goops.h"
47 #include "libguile/smob.h"
48 #include "libguile/chars.h"
49 #include "libguile/dynwind.h"
50
51 #include "libguile/keywords.h"
52 #include "libguile/hashtab.h"
53 #include "libguile/root.h"
54 #include "libguile/strings.h"
55 #include "libguile/mallocs.h"
56 #include "libguile/validate.h"
57 #include "libguile/ports.h"
58 #include "libguile/vectors.h"
59 #include "libguile/weaks.h"
60 #include "libguile/fluids.h"
61 #include "libguile/eq.h"
62
63 #ifdef HAVE_STRING_H
64 #include <string.h>
65 #endif
66
67 #ifdef HAVE_IO_H
68 #include <io.h>
69 #endif
70
71 #ifdef HAVE_UNISTD_H
72 #include <unistd.h>
73 #endif
74
75 #ifdef HAVE_SYS_IOCTL_H
76 #include <sys/ioctl.h>
77 #endif
78
79 /* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize
80 already, but have this code here in case that wasn't so in past versions,
81 or perhaps to help other minimal DOS environments.
82
83 gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which
84 might be possibilities if we've got other systems without ftruncate. */
85
86 #if defined HAVE_CHSIZE && ! defined HAVE_FTRUNCATE
87 #define ftruncate(fd, size) chsize (fd, size)
88 #undef HAVE_FTRUNCATE
89 #define HAVE_FTRUNCATE 1
90 #endif
91
92 \f
93 /* The port kind table --- a dynamically resized array of port types. */
94
95
96 /* scm_ptobs scm_numptob
97 * implement a dynamically resized array of ptob records.
98 * Indexes into this table are used when generating type
99 * tags for smobjects (if you know a tag you can get an index and conversely).
100 */
101 scm_t_ptob_descriptor *scm_ptobs = NULL;
102 long scm_numptob = 0;
103
104 /* GC marker for a port with stream of SCM type. */
105 SCM
106 scm_markstream (SCM ptr)
107 {
108 int openp;
109 openp = SCM_CELL_WORD_0 (ptr) & SCM_OPN;
110 if (openp)
111 return SCM_PACK (SCM_STREAM (ptr));
112 else
113 return SCM_BOOL_F;
114 }
115
116 /*
117 * We choose to use an interface similar to the smob interface with
118 * fill_input and write as standard fields, passed to the port
119 * type constructor, and optional fields set by setters.
120 */
121
122 static void
123 flush_port_default (SCM port SCM_UNUSED)
124 {
125 }
126
127 static void
128 end_input_default (SCM port SCM_UNUSED, int offset SCM_UNUSED)
129 {
130 }
131
132 scm_t_bits
133 scm_make_port_type (char *name,
134 int (*fill_input) (SCM port),
135 void (*write) (SCM port, const void *data, size_t size))
136 {
137 char *tmp;
138 if (SCM_I_MAX_PORT_TYPE_COUNT - 1 <= scm_numptob)
139 goto ptoberr;
140 SCM_CRITICAL_SECTION_START;
141 tmp = (char *) scm_gc_realloc ((char *) scm_ptobs,
142 scm_numptob * sizeof (scm_t_ptob_descriptor),
143 (1 + scm_numptob)
144 * sizeof (scm_t_ptob_descriptor),
145 "port-type");
146 if (tmp)
147 {
148 scm_ptobs = (scm_t_ptob_descriptor *) tmp;
149
150 scm_ptobs[scm_numptob].name = name;
151 scm_ptobs[scm_numptob].mark = 0;
152 scm_ptobs[scm_numptob].free = NULL;
153 scm_ptobs[scm_numptob].print = scm_port_print;
154 scm_ptobs[scm_numptob].equalp = 0;
155 scm_ptobs[scm_numptob].close = 0;
156
157 scm_ptobs[scm_numptob].write = write;
158 scm_ptobs[scm_numptob].flush = flush_port_default;
159
160 scm_ptobs[scm_numptob].end_input = end_input_default;
161 scm_ptobs[scm_numptob].fill_input = fill_input;
162 scm_ptobs[scm_numptob].input_waiting = 0;
163
164 scm_ptobs[scm_numptob].seek = 0;
165 scm_ptobs[scm_numptob].truncate = 0;
166
167 scm_numptob++;
168 }
169 SCM_CRITICAL_SECTION_END;
170 if (!tmp)
171 {
172 ptoberr:
173 scm_memory_error ("scm_make_port_type");
174 }
175 /* Make a class object if Goops is present */
176 if (SCM_UNPACK (scm_port_class[0]) != 0)
177 scm_make_port_classes (scm_numptob - 1, SCM_PTOBNAME (scm_numptob - 1));
178 return scm_tc7_port + (scm_numptob - 1) * 256;
179 }
180
181 void
182 scm_set_port_mark (scm_t_bits tc, SCM (*mark) (SCM))
183 {
184 scm_ptobs[SCM_TC2PTOBNUM (tc)].mark = mark;
185 }
186
187 void
188 scm_set_port_free (scm_t_bits tc, size_t (*free) (SCM))
189 {
190 scm_ptobs[SCM_TC2PTOBNUM (tc)].free = free;
191 }
192
193 void
194 scm_set_port_print (scm_t_bits tc, int (*print) (SCM exp, SCM port,
195 scm_print_state *pstate))
196 {
197 scm_ptobs[SCM_TC2PTOBNUM (tc)].print = print;
198 }
199
200 void
201 scm_set_port_equalp (scm_t_bits tc, SCM (*equalp) (SCM, SCM))
202 {
203 scm_ptobs[SCM_TC2PTOBNUM (tc)].equalp = equalp;
204 }
205
206 void
207 scm_set_port_flush (scm_t_bits tc, void (*flush) (SCM port))
208 {
209 scm_ptobs[SCM_TC2PTOBNUM (tc)].flush = flush;
210 }
211
212 void
213 scm_set_port_end_input (scm_t_bits tc, void (*end_input) (SCM port, int offset))
214 {
215 scm_ptobs[SCM_TC2PTOBNUM (tc)].end_input = end_input;
216 }
217
218 void
219 scm_set_port_close (scm_t_bits tc, int (*close) (SCM))
220 {
221 scm_ptobs[SCM_TC2PTOBNUM (tc)].close = close;
222 }
223
224 void
225 scm_set_port_seek (scm_t_bits tc,
226 scm_t_off (*seek) (SCM, scm_t_off, int))
227 {
228 scm_ptobs[SCM_TC2PTOBNUM (tc)].seek = seek;
229 }
230
231 void
232 scm_set_port_truncate (scm_t_bits tc, void (*truncate) (SCM, scm_t_off))
233 {
234 scm_ptobs[SCM_TC2PTOBNUM (tc)].truncate = truncate;
235 }
236
237 void
238 scm_set_port_input_waiting (scm_t_bits tc, int (*input_waiting) (SCM))
239 {
240 scm_ptobs[SCM_TC2PTOBNUM (tc)].input_waiting = input_waiting;
241 }
242
243 \f
244
245 SCM_DEFINE (scm_char_ready_p, "char-ready?", 0, 1, 0,
246 (SCM port),
247 "Return @code{#t} if a character is ready on input @var{port}\n"
248 "and return @code{#f} otherwise. If @code{char-ready?} returns\n"
249 "@code{#t} then the next @code{read-char} operation on\n"
250 "@var{port} is guaranteed not to hang. If @var{port} is a file\n"
251 "port at end of file then @code{char-ready?} returns @code{#t}.\n"
252 "\n"
253 "@code{char-ready?} exists to make it possible for a\n"
254 "program to accept characters from interactive ports without\n"
255 "getting stuck waiting for input. Any input editors associated\n"
256 "with such ports must make sure that characters whose existence\n"
257 "has been asserted by @code{char-ready?} cannot be rubbed out.\n"
258 "If @code{char-ready?} were to return @code{#f} at end of file,\n"
259 "a port at end of file would be indistinguishable from an\n"
260 "interactive port that has no ready characters.")
261 #define FUNC_NAME s_scm_char_ready_p
262 {
263 scm_t_port *pt;
264
265 if (SCM_UNBNDP (port))
266 port = scm_current_input_port ();
267 /* It's possible to close the current input port, so validate even in
268 this case. */
269 SCM_VALIDATE_OPINPORT (1, port);
270
271 pt = SCM_PTAB_ENTRY (port);
272
273 /* if the current read buffer is filled, or the
274 last pushed-back char has been read and the saved buffer is
275 filled, result is true. */
276 if (pt->read_pos < pt->read_end
277 || (pt->read_buf == pt->putback_buf
278 && pt->saved_read_pos < pt->saved_read_end))
279 return SCM_BOOL_T;
280 else
281 {
282 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
283
284 if (ptob->input_waiting)
285 return scm_from_bool(ptob->input_waiting (port));
286 else
287 return SCM_BOOL_T;
288 }
289 }
290 #undef FUNC_NAME
291
292 /* move up to read_len chars from port's putback and/or read buffers
293 into memory starting at dest. returns the number of chars moved. */
294 size_t scm_take_from_input_buffers (SCM port, char *dest, size_t read_len)
295 {
296 scm_t_port *pt = SCM_PTAB_ENTRY (port);
297 size_t chars_read = 0;
298 size_t from_buf = min (pt->read_end - pt->read_pos, read_len);
299
300 if (from_buf > 0)
301 {
302 memcpy (dest, pt->read_pos, from_buf);
303 pt->read_pos += from_buf;
304 chars_read += from_buf;
305 read_len -= from_buf;
306 dest += from_buf;
307 }
308
309 /* if putback was active, try the real input buffer too. */
310 if (pt->read_buf == pt->putback_buf)
311 {
312 from_buf = min (pt->saved_read_end - pt->saved_read_pos, read_len);
313 if (from_buf > 0)
314 {
315 memcpy (dest, pt->saved_read_pos, from_buf);
316 pt->saved_read_pos += from_buf;
317 chars_read += from_buf;
318 }
319 }
320 return chars_read;
321 }
322
323 /* Clear a port's read buffers, returning the contents. */
324 SCM_DEFINE (scm_drain_input, "drain-input", 1, 0, 0,
325 (SCM port),
326 "This procedure clears a port's input buffers, similar\n"
327 "to the way that force-output clears the output buffer. The\n"
328 "contents of the buffers are returned as a single string, e.g.,\n"
329 "\n"
330 "@lisp\n"
331 "(define p (open-input-file ...))\n"
332 "(drain-input p) => empty string, nothing buffered yet.\n"
333 "(unread-char (read-char p) p)\n"
334 "(drain-input p) => initial chars from p, up to the buffer size.\n"
335 "@end lisp\n\n"
336 "Draining the buffers may be useful for cleanly finishing\n"
337 "buffered I/O so that the file descriptor can be used directly\n"
338 "for further input.")
339 #define FUNC_NAME s_scm_drain_input
340 {
341 SCM result;
342 char *data;
343 scm_t_port *pt;
344 long count;
345
346 SCM_VALIDATE_OPINPORT (1, port);
347 pt = SCM_PTAB_ENTRY (port);
348
349 count = pt->read_end - pt->read_pos;
350 if (pt->read_buf == pt->putback_buf)
351 count += pt->saved_read_end - pt->saved_read_pos;
352
353 if (count)
354 {
355 result = scm_i_make_string (count, &data);
356 scm_take_from_input_buffers (port, data, count);
357 }
358 else
359 result = scm_nullstr;
360
361 return result;
362 }
363 #undef FUNC_NAME
364
365 \f
366 /* Standard ports --- current input, output, error, and more(!). */
367
368 static SCM cur_inport_fluid = 0;
369 static SCM cur_outport_fluid = 0;
370 static SCM cur_errport_fluid = 0;
371 static SCM cur_loadport_fluid = 0;
372
373 SCM_DEFINE (scm_current_input_port, "current-input-port", 0, 0, 0,
374 (),
375 "Return the current input port. This is the default port used\n"
376 "by many input procedures. Initially, @code{current-input-port}\n"
377 "returns the @dfn{standard input} in Unix and C terminology.")
378 #define FUNC_NAME s_scm_current_input_port
379 {
380 if (cur_inport_fluid)
381 return scm_fluid_ref (cur_inport_fluid);
382 else
383 return SCM_BOOL_F;
384 }
385 #undef FUNC_NAME
386
387 SCM_DEFINE (scm_current_output_port, "current-output-port", 0, 0, 0,
388 (),
389 "Return the current output port. This is the default port used\n"
390 "by many output procedures. Initially,\n"
391 "@code{current-output-port} returns the @dfn{standard output} in\n"
392 "Unix and C terminology.")
393 #define FUNC_NAME s_scm_current_output_port
394 {
395 if (cur_outport_fluid)
396 return scm_fluid_ref (cur_outport_fluid);
397 else
398 return SCM_BOOL_F;
399 }
400 #undef FUNC_NAME
401
402 SCM_DEFINE (scm_current_error_port, "current-error-port", 0, 0, 0,
403 (),
404 "Return the port to which errors and warnings should be sent (the\n"
405 "@dfn{standard error} in Unix and C terminology).")
406 #define FUNC_NAME s_scm_current_error_port
407 {
408 if (cur_errport_fluid)
409 return scm_fluid_ref (cur_errport_fluid);
410 else
411 return SCM_BOOL_F;
412 }
413 #undef FUNC_NAME
414
415 SCM_DEFINE (scm_current_load_port, "current-load-port", 0, 0, 0,
416 (),
417 "Return the current-load-port.\n"
418 "The load port is used internally by @code{primitive-load}.")
419 #define FUNC_NAME s_scm_current_load_port
420 {
421 return scm_fluid_ref (cur_loadport_fluid);
422 }
423 #undef FUNC_NAME
424
425 SCM_DEFINE (scm_set_current_input_port, "set-current-input-port", 1, 0, 0,
426 (SCM port),
427 "@deffnx {Scheme Procedure} set-current-output-port port\n"
428 "@deffnx {Scheme Procedure} set-current-error-port port\n"
429 "Change the ports returned by @code{current-input-port},\n"
430 "@code{current-output-port} and @code{current-error-port}, respectively,\n"
431 "so that they use the supplied @var{port} for input or output.")
432 #define FUNC_NAME s_scm_set_current_input_port
433 {
434 SCM oinp = scm_fluid_ref (cur_inport_fluid);
435 SCM_VALIDATE_OPINPORT (1, port);
436 scm_fluid_set_x (cur_inport_fluid, port);
437 return oinp;
438 }
439 #undef FUNC_NAME
440
441
442 SCM_DEFINE (scm_set_current_output_port, "set-current-output-port", 1, 0, 0,
443 (SCM port),
444 "Set the current default output port to @var{port}.")
445 #define FUNC_NAME s_scm_set_current_output_port
446 {
447 SCM ooutp = scm_fluid_ref (cur_outport_fluid);
448 port = SCM_COERCE_OUTPORT (port);
449 SCM_VALIDATE_OPOUTPORT (1, port);
450 scm_fluid_set_x (cur_outport_fluid, port);
451 return ooutp;
452 }
453 #undef FUNC_NAME
454
455
456 SCM_DEFINE (scm_set_current_error_port, "set-current-error-port", 1, 0, 0,
457 (SCM port),
458 "Set the current default error port to @var{port}.")
459 #define FUNC_NAME s_scm_set_current_error_port
460 {
461 SCM oerrp = scm_fluid_ref (cur_errport_fluid);
462 port = SCM_COERCE_OUTPORT (port);
463 SCM_VALIDATE_OPOUTPORT (1, port);
464 scm_fluid_set_x (cur_errport_fluid, port);
465 return oerrp;
466 }
467 #undef FUNC_NAME
468
469 void
470 scm_dynwind_current_input_port (SCM port)
471 #define FUNC_NAME NULL
472 {
473 SCM_VALIDATE_OPINPORT (1, port);
474 scm_dynwind_fluid (cur_inport_fluid, port);
475 }
476 #undef FUNC_NAME
477
478 void
479 scm_dynwind_current_output_port (SCM port)
480 #define FUNC_NAME NULL
481 {
482 port = SCM_COERCE_OUTPORT (port);
483 SCM_VALIDATE_OPOUTPORT (1, port);
484 scm_dynwind_fluid (cur_outport_fluid, port);
485 }
486 #undef FUNC_NAME
487
488 void
489 scm_dynwind_current_error_port (SCM port)
490 #define FUNC_NAME NULL
491 {
492 port = SCM_COERCE_OUTPORT (port);
493 SCM_VALIDATE_OPOUTPORT (1, port);
494 scm_dynwind_fluid (cur_errport_fluid, port);
495 }
496 #undef FUNC_NAME
497
498 void
499 scm_i_dynwind_current_load_port (SCM port)
500 {
501 scm_dynwind_fluid (cur_loadport_fluid, port);
502 }
503
504 \f
505 /* The port table --- an array of pointers to ports. */
506
507 /*
508 We need a global registry of ports to flush them all at exit, and to
509 get all the ports matching a file descriptor.
510 */
511 SCM scm_i_port_weak_hash;
512
513 scm_i_pthread_mutex_t scm_i_port_table_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER;
514
515 \f
516 /* Port finalization. */
517
518
519 static void finalize_port (GC_PTR, GC_PTR);
520
521 /* Register a finalizer for PORT. */
522 static SCM_C_INLINE_KEYWORD void
523 register_finalizer_for_port (SCM port)
524 {
525 long port_type;
526 GC_finalization_proc prev_finalizer;
527 GC_PTR prev_finalization_data;
528
529 port_type = SCM_TC2PTOBNUM (SCM_CELL_TYPE (port));
530
531 /* Register a finalizer for PORT so that its iconv CDs get freed and
532 optionally its type's `free' function gets called. */
533 GC_REGISTER_FINALIZER_NO_ORDER (SCM2PTR (port), finalize_port, 0,
534 &prev_finalizer,
535 &prev_finalization_data);
536 }
537
538 /* Finalize the object (a port) pointed to by PTR. */
539 static void
540 finalize_port (GC_PTR ptr, GC_PTR data)
541 {
542 long port_type;
543 SCM port = PTR2SCM (ptr);
544
545 if (!SCM_PORTP (port))
546 abort ();
547
548 if (SCM_OPENP (port))
549 {
550 if (SCM_REVEALED (port) > 0)
551 /* Keep "revealed" ports alive and re-register a finalizer. */
552 register_finalizer_for_port (port);
553 else
554 {
555 scm_t_port *entry;
556
557 port_type = SCM_TC2PTOBNUM (SCM_CELL_TYPE (port));
558 if (port_type >= scm_numptob)
559 abort ();
560
561 if (scm_ptobs[port_type].free)
562 /* Yes, I really do mean `.free' rather than `.close'. `.close'
563 is for explicit `close-port' by user. */
564 scm_ptobs[port_type].free (port);
565
566 entry = SCM_PTAB_ENTRY (port);
567
568 if (entry->input_cd != (iconv_t) -1)
569 iconv_close (entry->input_cd);
570 if (entry->output_cd != (iconv_t) -1)
571 iconv_close (entry->output_cd);
572
573 SCM_SETSTREAM (port, 0);
574 SCM_CLR_PORT_OPEN_FLAG (port);
575
576 scm_gc_ports_collected++;
577 }
578 }
579 }
580
581
582
583 \f
584
585 /* This function is not and should not be thread safe. */
586 SCM
587 scm_new_port_table_entry (scm_t_bits tag)
588 #define FUNC_NAME "scm_new_port_table_entry"
589 {
590 /*
591 We initialize the cell to empty, this is in case scm_gc_calloc
592 triggers GC ; we don't want the GC to scan a half-finished Z.
593 */
594
595 SCM z = scm_cons (SCM_EOL, SCM_EOL);
596 scm_t_port *entry = (scm_t_port *) scm_gc_calloc (sizeof (scm_t_port), "port");
597 const char *enc;
598
599 entry->file_name = SCM_BOOL_F;
600 entry->rw_active = SCM_PORT_NEITHER;
601 entry->port = z;
602
603 /* Initialize this port with the thread's current default
604 encoding. */
605 enc = scm_i_default_port_encoding ();
606 entry->encoding = enc ? scm_gc_strdup (enc, "port") : NULL;
607
608 /* The conversion descriptors will be opened lazily. */
609 entry->input_cd = (iconv_t) -1;
610 entry->output_cd = (iconv_t) -1;
611
612 entry->ilseq_handler = scm_i_get_conversion_strategy (SCM_BOOL_F);
613
614 SCM_SET_CELL_TYPE (z, tag);
615 SCM_SETPTAB_ENTRY (z, entry);
616
617 scm_hashq_set_x (scm_i_port_weak_hash, z, SCM_BOOL_F);
618
619 /* For each new port, register a finalizer so that it port type's free
620 function can be invoked eventually. */
621 register_finalizer_for_port (z);
622
623 return z;
624 }
625 #undef FUNC_NAME
626
627 #if SCM_ENABLE_DEPRECATED==1
628 scm_t_port *
629 scm_add_to_port_table (SCM port)
630 {
631 SCM z;
632 scm_t_port * pt;
633
634 scm_c_issue_deprecation_warning ("scm_add_to_port_table is deprecated.");
635
636 scm_i_pthread_mutex_lock (&scm_i_port_table_mutex);
637 z = scm_new_port_table_entry (scm_tc7_port);
638 pt = SCM_PTAB_ENTRY(z);
639 pt->port = port;
640 SCM_SETCAR (z, SCM_EOL);
641 SCM_SETCDR (z, SCM_EOL);
642 SCM_SETPTAB_ENTRY (port, pt);
643 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
644
645 return pt;
646 }
647 #endif
648
649
650 /* Remove a port from the table and destroy it. */
651
652 static void
653 scm_i_remove_port (SCM port)
654 #define FUNC_NAME "scm_remove_port"
655 {
656 scm_t_port *p;
657
658 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
659
660 p = SCM_PTAB_ENTRY (port);
661 scm_port_non_buffer (p);
662 p->putback_buf = NULL;
663 p->putback_buf_size = 0;
664
665 if (p->input_cd != (iconv_t) -1)
666 {
667 iconv_close (p->input_cd);
668 p->input_cd = (iconv_t) -1;
669 }
670
671 if (p->output_cd != (iconv_t) -1)
672 {
673 iconv_close (p->output_cd);
674 p->output_cd = (iconv_t) -1;
675 }
676
677 SCM_SETPTAB_ENTRY (port, 0);
678
679 scm_hashq_remove_x (scm_i_port_weak_hash, port);
680
681 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
682 }
683 #undef FUNC_NAME
684
685
686 /* Functions for debugging. */
687 #ifdef GUILE_DEBUG
688 SCM_DEFINE (scm_pt_size, "pt-size", 0, 0, 0,
689 (),
690 "Return the number of ports in the port table. @code{pt-size}\n"
691 "is only included in @code{--enable-guile-debug} builds.")
692 #define FUNC_NAME s_scm_pt_size
693 {
694 return scm_from_int (SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash));
695 }
696 #undef FUNC_NAME
697 #endif
698
699 void
700 scm_port_non_buffer (scm_t_port *pt)
701 {
702 pt->read_pos = pt->read_buf = pt->read_end = &pt->shortbuf;
703 pt->write_buf = pt->write_pos = &pt->shortbuf;
704 pt->read_buf_size = pt->write_buf_size = 1;
705 pt->write_end = pt->write_buf + pt->write_buf_size;
706 }
707
708 \f
709 /* Revealed counts --- an oddity inherited from SCSH. */
710
711 /* Find a port in the table and return its revealed count.
712 Also used by the garbage collector.
713 */
714
715 int
716 scm_revealed_count (SCM port)
717 {
718 return SCM_REVEALED(port);
719 }
720
721
722
723 /* Return the revealed count for a port. */
724
725 SCM_DEFINE (scm_port_revealed, "port-revealed", 1, 0, 0,
726 (SCM port),
727 "Return the revealed count for @var{port}.")
728 #define FUNC_NAME s_scm_port_revealed
729 {
730 port = SCM_COERCE_OUTPORT (port);
731 SCM_VALIDATE_OPENPORT (1, port);
732 return scm_from_int (scm_revealed_count (port));
733 }
734 #undef FUNC_NAME
735
736 /* Set the revealed count for a port. */
737 SCM_DEFINE (scm_set_port_revealed_x, "set-port-revealed!", 2, 0, 0,
738 (SCM port, SCM rcount),
739 "Sets the revealed count for a port to a given value.\n"
740 "The return value is unspecified.")
741 #define FUNC_NAME s_scm_set_port_revealed_x
742 {
743 port = SCM_COERCE_OUTPORT (port);
744 SCM_VALIDATE_OPENPORT (1, port);
745 SCM_REVEALED (port) = scm_to_int (rcount);
746 return SCM_UNSPECIFIED;
747 }
748 #undef FUNC_NAME
749
750
751 \f
752 /* Retrieving a port's mode. */
753
754 /* Return the flags that characterize a port based on the mode
755 * string used to open a file for that port.
756 *
757 * See PORT FLAGS in scm.h
758 */
759
760 static long
761 scm_i_mode_bits_n (SCM modes)
762 {
763 return (SCM_OPN
764 | (scm_i_string_contains_char (modes, 'r')
765 || scm_i_string_contains_char (modes, '+') ? SCM_RDNG : 0)
766 | (scm_i_string_contains_char (modes, 'w')
767 || scm_i_string_contains_char (modes, 'a')
768 || scm_i_string_contains_char (modes, '+') ? SCM_WRTNG : 0)
769 | (scm_i_string_contains_char (modes, '0') ? SCM_BUF0 : 0)
770 | (scm_i_string_contains_char (modes, 'l') ? SCM_BUFLINE : 0));
771 }
772
773 long
774 scm_mode_bits (char *modes)
775 {
776 return scm_i_mode_bits (scm_from_locale_string (modes));
777 }
778
779 long
780 scm_i_mode_bits (SCM modes)
781 {
782 long bits;
783
784 if (!scm_is_string (modes))
785 scm_wrong_type_arg_msg (NULL, 0, modes, "string");
786
787 bits = scm_i_mode_bits_n (modes);
788 scm_remember_upto_here_1 (modes);
789 return bits;
790 }
791
792 /* Return the mode flags from an open port.
793 * Some modes such as "append" are only used when opening
794 * a file and are not returned here. */
795
796 SCM_DEFINE (scm_port_mode, "port-mode", 1, 0, 0,
797 (SCM port),
798 "Return the port modes associated with the open port @var{port}.\n"
799 "These will not necessarily be identical to the modes used when\n"
800 "the port was opened, since modes such as \"append\" which are\n"
801 "used only during port creation are not retained.")
802 #define FUNC_NAME s_scm_port_mode
803 {
804 char modes[4];
805 modes[0] = '\0';
806
807 port = SCM_COERCE_OUTPORT (port);
808 SCM_VALIDATE_OPPORT (1, port);
809 if (SCM_CELL_WORD_0 (port) & SCM_RDNG) {
810 if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
811 strcpy (modes, "r+");
812 else
813 strcpy (modes, "r");
814 }
815 else if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
816 strcpy (modes, "w");
817 if (SCM_CELL_WORD_0 (port) & SCM_BUF0)
818 strcat (modes, "0");
819 return scm_from_locale_string (modes);
820 }
821 #undef FUNC_NAME
822
823
824 \f
825 /* Closing ports. */
826
827 /* scm_close_port
828 * Call the close operation on a port object.
829 * see also scm_close.
830 */
831 SCM_DEFINE (scm_close_port, "close-port", 1, 0, 0,
832 (SCM port),
833 "Close the specified port object. Return @code{#t} if it\n"
834 "successfully closes a port or @code{#f} if it was already\n"
835 "closed. An exception may be raised if an error occurs, for\n"
836 "example when flushing buffered output. See also @ref{Ports and\n"
837 "File Descriptors, close}, for a procedure which can close file\n"
838 "descriptors.")
839 #define FUNC_NAME s_scm_close_port
840 {
841 size_t i;
842 int rv;
843
844 port = SCM_COERCE_OUTPORT (port);
845
846 SCM_VALIDATE_PORT (1, port);
847 if (SCM_CLOSEDP (port))
848 return SCM_BOOL_F;
849 i = SCM_PTOBNUM (port);
850 if (scm_ptobs[i].close)
851 rv = (scm_ptobs[i].close) (port);
852 else
853 rv = 0;
854 scm_i_remove_port (port);
855 SCM_CLR_PORT_OPEN_FLAG (port);
856 return scm_from_bool (rv >= 0);
857 }
858 #undef FUNC_NAME
859
860 SCM_DEFINE (scm_close_input_port, "close-input-port", 1, 0, 0,
861 (SCM port),
862 "Close the specified input port object. The routine has no effect if\n"
863 "the file has already been closed. An exception may be raised if an\n"
864 "error occurs. The value returned is unspecified.\n\n"
865 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
866 "which can close file descriptors.")
867 #define FUNC_NAME s_scm_close_input_port
868 {
869 SCM_VALIDATE_INPUT_PORT (1, port);
870 scm_close_port (port);
871 return SCM_UNSPECIFIED;
872 }
873 #undef FUNC_NAME
874
875 SCM_DEFINE (scm_close_output_port, "close-output-port", 1, 0, 0,
876 (SCM port),
877 "Close the specified output port object. The routine has no effect if\n"
878 "the file has already been closed. An exception may be raised if an\n"
879 "error occurs. The value returned is unspecified.\n\n"
880 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
881 "which can close file descriptors.")
882 #define FUNC_NAME s_scm_close_output_port
883 {
884 port = SCM_COERCE_OUTPORT (port);
885 SCM_VALIDATE_OUTPUT_PORT (1, port);
886 scm_close_port (port);
887 return SCM_UNSPECIFIED;
888 }
889 #undef FUNC_NAME
890
891 static SCM
892 collect_keys (void *unused, SCM key, SCM value, SCM result)
893 {
894 return scm_cons (key, result);
895 }
896
897 void
898 scm_c_port_for_each (void (*proc)(void *data, SCM p), void *data)
899 {
900 SCM ports;
901
902 /* Copy out the port table as a list so that we get strong references
903 to all the values. */
904 scm_i_pthread_mutex_lock (&scm_i_port_table_mutex);
905 ports = scm_internal_hash_fold (collect_keys, NULL,
906 SCM_EOL, scm_i_port_weak_hash);
907 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
908
909 for (; scm_is_pair (ports); ports = scm_cdr (ports))
910 {
911 SCM p = scm_car (ports);
912 if (SCM_PORTP (p))
913 proc (data, p);
914 }
915 }
916
917 SCM_DEFINE (scm_port_for_each, "port-for-each", 1, 0, 0,
918 (SCM proc),
919 "Apply @var{proc} to each port in the Guile port table\n"
920 "in turn. The return value is unspecified. More specifically,\n"
921 "@var{proc} is applied exactly once to every port that exists\n"
922 "in the system at the time @var{port-for-each} is invoked.\n"
923 "Changes to the port table while @var{port-for-each} is running\n"
924 "have no effect as far as @var{port-for-each} is concerned.")
925 #define FUNC_NAME s_scm_port_for_each
926 {
927 SCM_VALIDATE_PROC (1, proc);
928
929 scm_c_port_for_each ((void (*)(void*,SCM))scm_call_1, proc);
930 return SCM_UNSPECIFIED;
931 }
932 #undef FUNC_NAME
933
934
935 \f
936 /* Utter miscellany. Gosh, we should clean this up some time. */
937
938 SCM_DEFINE (scm_input_port_p, "input-port?", 1, 0, 0,
939 (SCM x),
940 "Return @code{#t} if @var{x} is an input port, otherwise return\n"
941 "@code{#f}. Any object satisfying this predicate also satisfies\n"
942 "@code{port?}.")
943 #define FUNC_NAME s_scm_input_port_p
944 {
945 return scm_from_bool (SCM_INPUT_PORT_P (x));
946 }
947 #undef FUNC_NAME
948
949 SCM_DEFINE (scm_output_port_p, "output-port?", 1, 0, 0,
950 (SCM x),
951 "Return @code{#t} if @var{x} is an output port, otherwise return\n"
952 "@code{#f}. Any object satisfying this predicate also satisfies\n"
953 "@code{port?}.")
954 #define FUNC_NAME s_scm_output_port_p
955 {
956 x = SCM_COERCE_OUTPORT (x);
957 return scm_from_bool (SCM_OUTPUT_PORT_P (x));
958 }
959 #undef FUNC_NAME
960
961 SCM_DEFINE (scm_port_p, "port?", 1, 0, 0,
962 (SCM x),
963 "Return a boolean indicating whether @var{x} is a port.\n"
964 "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n"
965 "@var{x}))}.")
966 #define FUNC_NAME s_scm_port_p
967 {
968 return scm_from_bool (SCM_PORTP (x));
969 }
970 #undef FUNC_NAME
971
972 SCM_DEFINE (scm_port_closed_p, "port-closed?", 1, 0, 0,
973 (SCM port),
974 "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n"
975 "open.")
976 #define FUNC_NAME s_scm_port_closed_p
977 {
978 SCM_VALIDATE_PORT (1, port);
979 return scm_from_bool (!SCM_OPPORTP (port));
980 }
981 #undef FUNC_NAME
982
983 SCM_DEFINE (scm_eof_object_p, "eof-object?", 1, 0, 0,
984 (SCM x),
985 "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n"
986 "return @code{#f}.")
987 #define FUNC_NAME s_scm_eof_object_p
988 {
989 return scm_from_bool(SCM_EOF_OBJECT_P (x));
990 }
991 #undef FUNC_NAME
992
993 SCM_DEFINE (scm_force_output, "force-output", 0, 1, 0,
994 (SCM port),
995 "Flush the specified output port, or the current output port if @var{port}\n"
996 "is omitted. The current output buffer contents are passed to the\n"
997 "underlying port implementation (e.g., in the case of fports, the\n"
998 "data will be written to the file and the output buffer will be cleared.)\n"
999 "It has no effect on an unbuffered port.\n\n"
1000 "The return value is unspecified.")
1001 #define FUNC_NAME s_scm_force_output
1002 {
1003 if (SCM_UNBNDP (port))
1004 port = scm_current_output_port ();
1005 else
1006 {
1007 port = SCM_COERCE_OUTPORT (port);
1008 SCM_VALIDATE_OPOUTPORT (1, port);
1009 }
1010 scm_flush (port);
1011 return SCM_UNSPECIFIED;
1012 }
1013 #undef FUNC_NAME
1014
1015
1016 static void
1017 flush_output_port (void *closure, SCM port)
1018 {
1019 if (SCM_OPOUTPORTP (port))
1020 scm_flush (port);
1021 }
1022
1023 SCM_DEFINE (scm_flush_all_ports, "flush-all-ports", 0, 0, 0,
1024 (),
1025 "Equivalent to calling @code{force-output} on\n"
1026 "all open output ports. The return value is unspecified.")
1027 #define FUNC_NAME s_scm_flush_all_ports
1028 {
1029 scm_c_port_for_each (&flush_output_port, NULL);
1030 return SCM_UNSPECIFIED;
1031 }
1032 #undef FUNC_NAME
1033
1034 SCM_DEFINE (scm_read_char, "read-char", 0, 1, 0,
1035 (SCM port),
1036 "Return the next character available from @var{port}, updating\n"
1037 "@var{port} to point to the following character. If no more\n"
1038 "characters are available, the end-of-file object is returned.\n"
1039 "\n"
1040 "When @var{port}'s data cannot be decoded according to its\n"
1041 "character encoding, a @code{decoding-error} is raised and\n"
1042 "@var{port} points past the erroneous byte sequence.\n")
1043 #define FUNC_NAME s_scm_read_char
1044 {
1045 scm_t_wchar c;
1046 if (SCM_UNBNDP (port))
1047 port = scm_current_input_port ();
1048 SCM_VALIDATE_OPINPORT (1, port);
1049 c = scm_getc (port);
1050 if (EOF == c)
1051 return SCM_EOF_VAL;
1052 return SCM_MAKE_CHAR (c);
1053 }
1054 #undef FUNC_NAME
1055
1056 /* Update the line and column number of PORT after consumption of C. */
1057 static inline void
1058 update_port_lf (scm_t_wchar c, SCM port)
1059 {
1060 switch (c)
1061 {
1062 case '\a':
1063 break;
1064 case '\b':
1065 SCM_DECCOL (port);
1066 break;
1067 case '\n':
1068 SCM_INCLINE (port);
1069 break;
1070 case '\r':
1071 SCM_ZEROCOL (port);
1072 break;
1073 case '\t':
1074 SCM_TABCOL (port);
1075 break;
1076 default:
1077 SCM_INCCOL (port);
1078 break;
1079 }
1080 }
1081
1082 #define SCM_MBCHAR_BUF_SIZE (4)
1083
1084 /* Convert the SIZE-byte UTF-8 sequence in UTF8_BUF to a codepoint.
1085 UTF8_BUF is assumed to contain a valid UTF-8 sequence. */
1086 static scm_t_wchar
1087 utf8_to_codepoint (const scm_t_uint8 *utf8_buf, size_t size)
1088 {
1089 scm_t_wchar codepoint;
1090
1091 if (utf8_buf[0] <= 0x7f)
1092 {
1093 assert (size == 1);
1094 codepoint = utf8_buf[0];
1095 }
1096 else if ((utf8_buf[0] & 0xe0) == 0xc0)
1097 {
1098 assert (size == 2);
1099 codepoint = ((scm_t_wchar) utf8_buf[0] & 0x1f) << 6UL
1100 | (utf8_buf[1] & 0x3f);
1101 }
1102 else if ((utf8_buf[0] & 0xf0) == 0xe0)
1103 {
1104 assert (size == 3);
1105 codepoint = ((scm_t_wchar) utf8_buf[0] & 0x0f) << 12UL
1106 | ((scm_t_wchar) utf8_buf[1] & 0x3f) << 6UL
1107 | (utf8_buf[2] & 0x3f);
1108 }
1109 else
1110 {
1111 assert (size == 4);
1112 codepoint = ((scm_t_wchar) utf8_buf[0] & 0x07) << 18UL
1113 | ((scm_t_wchar) utf8_buf[1] & 0x3f) << 12UL
1114 | ((scm_t_wchar) utf8_buf[2] & 0x3f) << 6UL
1115 | (utf8_buf[3] & 0x3f);
1116 }
1117
1118 return codepoint;
1119 }
1120
1121 /* Read a codepoint from PORT and return it in *CODEPOINT. Fill BUF
1122 with the byte representation of the codepoint in PORT's encoding, and
1123 set *LEN to the length in bytes of that representation. Return 0 on
1124 success and an errno value on error. */
1125 static int
1126 get_codepoint (SCM port, scm_t_wchar *codepoint,
1127 char buf[SCM_MBCHAR_BUF_SIZE], size_t *len)
1128 {
1129 int err, byte_read;
1130 size_t bytes_consumed, output_size;
1131 char *output;
1132 scm_t_uint8 utf8_buf[SCM_MBCHAR_BUF_SIZE];
1133 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1134
1135 if (SCM_UNLIKELY (pt->input_cd == (iconv_t) -1))
1136 /* Initialize the conversion descriptors. */
1137 scm_i_set_port_encoding_x (port, pt->encoding);
1138
1139 for (output_size = 0, output = (char *) utf8_buf,
1140 bytes_consumed = 0, err = 0;
1141 err == 0 && output_size == 0
1142 && (bytes_consumed == 0 || byte_read != EOF);
1143 bytes_consumed++)
1144 {
1145 char *input;
1146 size_t input_left, output_left, done;
1147
1148 byte_read = scm_get_byte_or_eof (port);
1149 if (byte_read == EOF)
1150 {
1151 if (bytes_consumed == 0)
1152 {
1153 *codepoint = (scm_t_wchar) EOF;
1154 *len = 0;
1155 return 0;
1156 }
1157 else
1158 continue;
1159 }
1160
1161 buf[bytes_consumed] = byte_read;
1162
1163 input = buf;
1164 input_left = bytes_consumed + 1;
1165 output_left = sizeof (utf8_buf);
1166
1167 done = iconv (pt->input_cd, &input, &input_left,
1168 &output, &output_left);
1169 if (done == (size_t) -1)
1170 {
1171 err = errno;
1172 if (err == EINVAL)
1173 /* Missing input: keep trying. */
1174 err = 0;
1175 }
1176 else
1177 output_size = sizeof (utf8_buf) - output_left;
1178 }
1179
1180 if (SCM_UNLIKELY (err != 0))
1181 {
1182 /* Reset the `iconv' state. */
1183 iconv (pt->input_cd, NULL, NULL, NULL, NULL);
1184
1185 if (pt->ilseq_handler == SCM_ICONVEH_QUESTION_MARK)
1186 {
1187 *codepoint = '?';
1188 err = 0;
1189 }
1190
1191 /* Fail when the strategy is SCM_ICONVEH_ERROR or
1192 SCM_ICONVEH_ESCAPE_SEQUENCE (the latter doesn't make sense for
1193 input encoding errors.) */
1194 }
1195 else
1196 /* Convert the UTF8_BUF sequence to a Unicode code point. */
1197 *codepoint = utf8_to_codepoint (utf8_buf, output_size);
1198
1199 if (SCM_LIKELY (err == 0))
1200 update_port_lf (*codepoint, port);
1201
1202 *len = bytes_consumed;
1203
1204 return err;
1205 }
1206
1207 /* Read a codepoint from PORT and return it. */
1208 scm_t_wchar
1209 scm_getc (SCM port)
1210 #define FUNC_NAME "scm_getc"
1211 {
1212 int err;
1213 size_t len;
1214 scm_t_wchar codepoint;
1215 char buf[SCM_MBCHAR_BUF_SIZE];
1216
1217 err = get_codepoint (port, &codepoint, buf, &len);
1218 if (SCM_UNLIKELY (err != 0))
1219 /* At this point PORT should point past the invalid encoding, as per
1220 R6RS-lib Section 8.2.4. */
1221 scm_decoding_error (FUNC_NAME, err, "input decoding error", port);
1222
1223 return codepoint;
1224 }
1225 #undef FUNC_NAME
1226
1227 /* this should only be called when the read buffer is empty. it
1228 tries to refill the read buffer. it returns the first char from
1229 the port, which is either EOF or *(pt->read_pos). */
1230 int
1231 scm_fill_input (SCM port)
1232 {
1233 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1234
1235 assert (pt->read_pos == pt->read_end);
1236
1237 if (pt->read_buf == pt->putback_buf)
1238 {
1239 /* finished reading put-back chars. */
1240 pt->read_buf = pt->saved_read_buf;
1241 pt->read_pos = pt->saved_read_pos;
1242 pt->read_end = pt->saved_read_end;
1243 pt->read_buf_size = pt->saved_read_buf_size;
1244 if (pt->read_pos < pt->read_end)
1245 return *(pt->read_pos);
1246 }
1247 return scm_ptobs[SCM_PTOBNUM (port)].fill_input (port);
1248 }
1249
1250
1251 /* scm_lfwrite
1252 *
1253 * This function differs from scm_c_write; it updates port line and
1254 * column. */
1255
1256 void
1257 scm_lfwrite (const char *ptr, size_t size, SCM port)
1258 {
1259 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1260 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1261
1262 if (pt->rw_active == SCM_PORT_READ)
1263 scm_end_input (port);
1264
1265 ptob->write (port, ptr, size);
1266
1267 for (; size; ptr++, size--)
1268 update_port_lf ((scm_t_wchar) (unsigned char) *ptr, port);
1269
1270 if (pt->rw_random)
1271 pt->rw_active = SCM_PORT_WRITE;
1272 }
1273
1274 /* Write STR to PORT from START inclusive to END exclusive. */
1275 void
1276 scm_lfwrite_substr (SCM str, size_t start, size_t end, SCM port)
1277 {
1278 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1279
1280 if (pt->rw_active == SCM_PORT_READ)
1281 scm_end_input (port);
1282
1283 if (end == (size_t) -1)
1284 end = scm_i_string_length (str);
1285
1286 scm_display (scm_c_substring (str, start, end), port);
1287
1288 if (pt->rw_random)
1289 pt->rw_active = SCM_PORT_WRITE;
1290 }
1291
1292 /* scm_c_read
1293 *
1294 * Used by an application to read arbitrary number of bytes from an
1295 * SCM port. Same semantics as libc read, except that scm_c_read only
1296 * returns less than SIZE bytes if at end-of-file.
1297 *
1298 * Warning: Doesn't update port line and column counts! */
1299
1300 /* This structure, and the following swap_buffer function, are used
1301 for temporarily swapping a port's own read buffer, and the buffer
1302 that the caller of scm_c_read provides. */
1303 struct port_and_swap_buffer
1304 {
1305 scm_t_port *pt;
1306 unsigned char *buffer;
1307 size_t size;
1308 };
1309
1310 static void
1311 swap_buffer (void *data)
1312 {
1313 struct port_and_swap_buffer *psb = (struct port_and_swap_buffer *) data;
1314 unsigned char *old_buf = psb->pt->read_buf;
1315 size_t old_size = psb->pt->read_buf_size;
1316
1317 /* Make the port use (buffer, size) from the struct. */
1318 psb->pt->read_pos = psb->pt->read_buf = psb->pt->read_end = psb->buffer;
1319 psb->pt->read_buf_size = psb->size;
1320
1321 /* Save the port's old (buffer, size) in the struct. */
1322 psb->buffer = old_buf;
1323 psb->size = old_size;
1324 }
1325
1326 size_t
1327 scm_c_read (SCM port, void *buffer, size_t size)
1328 #define FUNC_NAME "scm_c_read"
1329 {
1330 scm_t_port *pt;
1331 size_t n_read = 0, n_available;
1332 struct port_and_swap_buffer psb;
1333
1334 SCM_VALIDATE_OPINPORT (1, port);
1335
1336 pt = SCM_PTAB_ENTRY (port);
1337 if (pt->rw_active == SCM_PORT_WRITE)
1338 scm_ptobs[SCM_PTOBNUM (port)].flush (port);
1339
1340 if (pt->rw_random)
1341 pt->rw_active = SCM_PORT_READ;
1342
1343 /* Take bytes first from the port's read buffer. */
1344 if (pt->read_pos < pt->read_end)
1345 {
1346 n_available = min (size, pt->read_end - pt->read_pos);
1347 memcpy (buffer, pt->read_pos, n_available);
1348 buffer = (char *) buffer + n_available;
1349 pt->read_pos += n_available;
1350 n_read += n_available;
1351 size -= n_available;
1352 }
1353
1354 /* Avoid the scm_dynwind_* costs if we now have enough data. */
1355 if (size == 0)
1356 return n_read;
1357
1358 /* Now we will call scm_fill_input repeatedly until we have read the
1359 requested number of bytes. (Note that a single scm_fill_input
1360 call does not guarantee to fill the whole of the port's read
1361 buffer.) */
1362 if (pt->read_buf_size <= 1 && pt->encoding == NULL)
1363 {
1364 /* The port that we are reading from is unbuffered - i.e. does
1365 not have its own persistent buffer - but we have a buffer,
1366 provided by our caller, that is the right size for the data
1367 that is wanted. For the following scm_fill_input calls,
1368 therefore, we use the buffer in hand as the port's read
1369 buffer.
1370
1371 We need to make sure that the port's normal (1 byte) buffer
1372 is reinstated in case one of the scm_fill_input () calls
1373 throws an exception; we use the scm_dynwind_* API to achieve
1374 that.
1375
1376 A consequence of this optimization is that the fill_input
1377 functions can't unget characters. That'll push data to the
1378 pushback buffer instead of this psb buffer. */
1379 #if SCM_DEBUG == 1
1380 unsigned char *pback = pt->putback_buf;
1381 #endif
1382 psb.pt = pt;
1383 psb.buffer = buffer;
1384 psb.size = size;
1385 scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE);
1386 scm_dynwind_rewind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY);
1387 scm_dynwind_unwind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY);
1388
1389 /* Call scm_fill_input until we have all the bytes that we need,
1390 or we hit EOF. */
1391 while (pt->read_buf_size && (scm_fill_input (port) != EOF))
1392 {
1393 pt->read_buf_size -= (pt->read_end - pt->read_pos);
1394 pt->read_pos = pt->read_buf = pt->read_end;
1395 }
1396 #if SCM_DEBUG == 1
1397 if (pback != pt->putback_buf
1398 || pt->read_buf - (unsigned char *) buffer < 0)
1399 scm_misc_error (FUNC_NAME,
1400 "scm_c_read must not call a fill function that pushes "
1401 "back characters onto an unbuffered port", SCM_EOL);
1402 #endif
1403 n_read += pt->read_buf - (unsigned char *) buffer;
1404
1405 /* Reinstate the port's normal buffer. */
1406 scm_dynwind_end ();
1407 }
1408 else
1409 {
1410 /* The port has its own buffer. It is important that we use it,
1411 even if it happens to be smaller than our caller's buffer, so
1412 that a custom port implementation's entry points (in
1413 particular, fill_input) can rely on the buffer always being
1414 the same as they first set up. */
1415 while (size && (scm_fill_input (port) != EOF))
1416 {
1417 n_available = min (size, pt->read_end - pt->read_pos);
1418 memcpy (buffer, pt->read_pos, n_available);
1419 buffer = (char *) buffer + n_available;
1420 pt->read_pos += n_available;
1421 n_read += n_available;
1422 size -= n_available;
1423 }
1424 }
1425
1426 return n_read;
1427 }
1428 #undef FUNC_NAME
1429
1430 /* scm_c_write
1431 *
1432 * Used by an application to write arbitrary number of bytes to an SCM
1433 * port. Similar semantics as libc write. However, unlike libc
1434 * write, scm_c_write writes the requested number of bytes and has no
1435 * return value.
1436 *
1437 * Warning: Doesn't update port line and column counts!
1438 */
1439
1440 void
1441 scm_c_write (SCM port, const void *ptr, size_t size)
1442 #define FUNC_NAME "scm_c_write"
1443 {
1444 scm_t_port *pt;
1445 scm_t_ptob_descriptor *ptob;
1446
1447 SCM_VALIDATE_OPOUTPORT (1, port);
1448
1449 pt = SCM_PTAB_ENTRY (port);
1450 ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1451
1452 if (pt->rw_active == SCM_PORT_READ)
1453 scm_end_input (port);
1454
1455 ptob->write (port, ptr, size);
1456
1457 if (pt->rw_random)
1458 pt->rw_active = SCM_PORT_WRITE;
1459 }
1460 #undef FUNC_NAME
1461
1462 void
1463 scm_flush (SCM port)
1464 {
1465 long i = SCM_PTOBNUM (port);
1466 assert (i >= 0);
1467 (scm_ptobs[i].flush) (port);
1468 }
1469
1470 void
1471 scm_end_input (SCM port)
1472 {
1473 long offset;
1474 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1475
1476 if (pt->read_buf == pt->putback_buf)
1477 {
1478 offset = pt->read_end - pt->read_pos;
1479 pt->read_buf = pt->saved_read_buf;
1480 pt->read_pos = pt->saved_read_pos;
1481 pt->read_end = pt->saved_read_end;
1482 pt->read_buf_size = pt->saved_read_buf_size;
1483 }
1484 else
1485 offset = 0;
1486
1487 scm_ptobs[SCM_PTOBNUM (port)].end_input (port, offset);
1488 }
1489
1490 \f
1491
1492
1493 void
1494 scm_unget_byte (int c, SCM port)
1495 #define FUNC_NAME "scm_unget_byte"
1496 {
1497 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1498
1499 if (pt->read_buf == pt->putback_buf)
1500 /* already using the put-back buffer. */
1501 {
1502 /* enlarge putback_buf if necessary. */
1503 if (pt->read_end == pt->read_buf + pt->read_buf_size
1504 && pt->read_buf == pt->read_pos)
1505 {
1506 size_t new_size = pt->read_buf_size * 2;
1507 unsigned char *tmp = (unsigned char *)
1508 scm_gc_realloc (pt->putback_buf, pt->read_buf_size, new_size,
1509 "putback buffer");
1510
1511 pt->read_pos = pt->read_buf = pt->putback_buf = tmp;
1512 pt->read_end = pt->read_buf + pt->read_buf_size;
1513 pt->read_buf_size = pt->putback_buf_size = new_size;
1514 }
1515
1516 /* shift any existing bytes to buffer + 1. */
1517 if (pt->read_pos == pt->read_end)
1518 pt->read_end = pt->read_buf + 1;
1519 else if (pt->read_pos != pt->read_buf + 1)
1520 {
1521 int count = pt->read_end - pt->read_pos;
1522
1523 memmove (pt->read_buf + 1, pt->read_pos, count);
1524 pt->read_end = pt->read_buf + 1 + count;
1525 }
1526
1527 pt->read_pos = pt->read_buf;
1528 }
1529 else
1530 /* switch to the put-back buffer. */
1531 {
1532 if (pt->putback_buf == NULL)
1533 {
1534 pt->putback_buf
1535 = (unsigned char *) scm_gc_malloc_pointerless
1536 (SCM_INITIAL_PUTBACK_BUF_SIZE, "putback buffer");
1537 pt->putback_buf_size = SCM_INITIAL_PUTBACK_BUF_SIZE;
1538 }
1539
1540 pt->saved_read_buf = pt->read_buf;
1541 pt->saved_read_pos = pt->read_pos;
1542 pt->saved_read_end = pt->read_end;
1543 pt->saved_read_buf_size = pt->read_buf_size;
1544
1545 pt->read_pos = pt->read_buf = pt->putback_buf;
1546 pt->read_end = pt->read_buf + 1;
1547 pt->read_buf_size = pt->putback_buf_size;
1548 }
1549
1550 *pt->read_buf = c;
1551
1552 if (pt->rw_random)
1553 pt->rw_active = SCM_PORT_READ;
1554 }
1555 #undef FUNC_NAME
1556
1557 void
1558 scm_ungetc (scm_t_wchar c, SCM port)
1559 #define FUNC_NAME "scm_ungetc"
1560 {
1561 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1562 char *result;
1563 char result_buf[10];
1564 const char *encoding;
1565 size_t len;
1566 int i;
1567
1568 if (pt->encoding != NULL)
1569 encoding = pt->encoding;
1570 else
1571 encoding = "ISO-8859-1";
1572
1573 len = sizeof (result_buf);
1574 result = u32_conv_to_encoding (encoding,
1575 (enum iconv_ilseq_handler) pt->ilseq_handler,
1576 (uint32_t *) &c, 1, NULL,
1577 result_buf, &len);
1578
1579 if (SCM_UNLIKELY (result == NULL || len == 0))
1580 scm_encoding_error (FUNC_NAME, errno,
1581 "conversion to port encoding failed",
1582 SCM_BOOL_F, SCM_MAKE_CHAR (c));
1583
1584 for (i = len - 1; i >= 0; i--)
1585 scm_unget_byte (result[i], port);
1586
1587 if (SCM_UNLIKELY (result != result_buf))
1588 free (result);
1589
1590 if (c == '\n')
1591 {
1592 /* What should col be in this case?
1593 * We'll leave it at -1.
1594 */
1595 SCM_LINUM (port) -= 1;
1596 }
1597 else
1598 SCM_COL(port) -= 1;
1599 }
1600 #undef FUNC_NAME
1601
1602
1603 void
1604 scm_ungets (const char *s, int n, SCM port)
1605 {
1606 /* This is simple minded and inefficient, but unreading strings is
1607 * probably not a common operation, and remember that line and
1608 * column numbers have to be handled...
1609 *
1610 * Please feel free to write an optimized version!
1611 */
1612 while (n--)
1613 scm_ungetc (s[n], port);
1614 }
1615
1616
1617 SCM_DEFINE (scm_peek_char, "peek-char", 0, 1, 0,
1618 (SCM port),
1619 "Return the next character available from @var{port},\n"
1620 "@emph{without} updating @var{port} to point to the following\n"
1621 "character. If no more characters are available, the\n"
1622 "end-of-file object is returned.\n"
1623 "\n"
1624 "The value returned by\n"
1625 "a call to @code{peek-char} is the same as the value that would\n"
1626 "have been returned by a call to @code{read-char} on the same\n"
1627 "port. The only difference is that the very next call to\n"
1628 "@code{read-char} or @code{peek-char} on that @var{port} will\n"
1629 "return the value returned by the preceding call to\n"
1630 "@code{peek-char}. In particular, a call to @code{peek-char} on\n"
1631 "an interactive port will hang waiting for input whenever a call\n"
1632 "to @code{read-char} would have hung.\n"
1633 "\n"
1634 "As for @code{read-char}, a @code{decoding-error} may be raised\n"
1635 "if such a situation occurs. However, unlike with @code{read-char},\n"
1636 "@var{port} still points at the beginning of the erroneous byte\n"
1637 "sequence when the error is raised.\n")
1638 #define FUNC_NAME s_scm_peek_char
1639 {
1640 int err;
1641 SCM result;
1642 scm_t_wchar c;
1643 char bytes[SCM_MBCHAR_BUF_SIZE];
1644 long column, line, i;
1645 size_t len;
1646
1647 if (SCM_UNBNDP (port))
1648 port = scm_current_input_port ();
1649 SCM_VALIDATE_OPINPORT (1, port);
1650
1651 column = SCM_COL (port);
1652 line = SCM_LINUM (port);
1653
1654 err = get_codepoint (port, &c, bytes, &len);
1655
1656 for (i = len - 1; i >= 0; i--)
1657 scm_unget_byte (bytes[i], port);
1658
1659 SCM_COL (port) = column;
1660 SCM_LINUM (port) = line;
1661
1662 if (SCM_UNLIKELY (err != 0))
1663 {
1664 scm_decoding_error (FUNC_NAME, err, "input decoding error", port);
1665
1666 /* Shouldn't happen since `catch' always aborts to prompt. */
1667 result = SCM_BOOL_F;
1668 }
1669 else if (c == EOF)
1670 result = SCM_EOF_VAL;
1671 else
1672 result = SCM_MAKE_CHAR (c);
1673
1674 return result;
1675 }
1676 #undef FUNC_NAME
1677
1678 SCM_DEFINE (scm_unread_char, "unread-char", 1, 1, 0,
1679 (SCM cobj, SCM port),
1680 "Place @var{char} in @var{port} so that it will be read by the\n"
1681 "next read operation. If called multiple times, the unread characters\n"
1682 "will be read again in last-in first-out order. If @var{port} is\n"
1683 "not supplied, the current input port is used.")
1684 #define FUNC_NAME s_scm_unread_char
1685 {
1686 int c;
1687
1688 SCM_VALIDATE_CHAR (1, cobj);
1689 if (SCM_UNBNDP (port))
1690 port = scm_current_input_port ();
1691 SCM_VALIDATE_OPINPORT (2, port);
1692
1693 c = SCM_CHAR (cobj);
1694
1695 scm_ungetc (c, port);
1696 return cobj;
1697 }
1698 #undef FUNC_NAME
1699
1700 SCM_DEFINE (scm_unread_string, "unread-string", 2, 0, 0,
1701 (SCM str, SCM port),
1702 "Place the string @var{str} in @var{port} so that its characters will be\n"
1703 "read in subsequent read operations. If called multiple times, the\n"
1704 "unread characters will be read again in last-in first-out order. If\n"
1705 "@var{port} is not supplied, the current-input-port is used.")
1706 #define FUNC_NAME s_scm_unread_string
1707 {
1708 int n;
1709 SCM_VALIDATE_STRING (1, str);
1710 if (SCM_UNBNDP (port))
1711 port = scm_current_input_port ();
1712 SCM_VALIDATE_OPINPORT (2, port);
1713
1714 n = scm_i_string_length (str);
1715
1716 while (n--)
1717 scm_ungetc (scm_i_string_ref (str, n), port);
1718
1719 return str;
1720 }
1721 #undef FUNC_NAME
1722
1723 SCM_DEFINE (scm_seek, "seek", 3, 0, 0,
1724 (SCM fd_port, SCM offset, SCM whence),
1725 "Sets the current position of @var{fd/port} to the integer\n"
1726 "@var{offset}, which is interpreted according to the value of\n"
1727 "@var{whence}.\n"
1728 "\n"
1729 "One of the following variables should be supplied for\n"
1730 "@var{whence}:\n"
1731 "@defvar SEEK_SET\n"
1732 "Seek from the beginning of the file.\n"
1733 "@end defvar\n"
1734 "@defvar SEEK_CUR\n"
1735 "Seek from the current position.\n"
1736 "@end defvar\n"
1737 "@defvar SEEK_END\n"
1738 "Seek from the end of the file.\n"
1739 "@end defvar\n"
1740 "If @var{fd/port} is a file descriptor, the underlying system\n"
1741 "call is @code{lseek}. @var{port} may be a string port.\n"
1742 "\n"
1743 "The value returned is the new position in the file. This means\n"
1744 "that the current position of a port can be obtained using:\n"
1745 "@lisp\n"
1746 "(seek port 0 SEEK_CUR)\n"
1747 "@end lisp")
1748 #define FUNC_NAME s_scm_seek
1749 {
1750 int how;
1751
1752 fd_port = SCM_COERCE_OUTPORT (fd_port);
1753
1754 how = scm_to_int (whence);
1755 if (how != SEEK_SET && how != SEEK_CUR && how != SEEK_END)
1756 SCM_OUT_OF_RANGE (3, whence);
1757
1758 if (SCM_OPPORTP (fd_port))
1759 {
1760 scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (fd_port);
1761 off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
1762 off_t_or_off64_t rv;
1763
1764 if (!ptob->seek)
1765 SCM_MISC_ERROR ("port is not seekable",
1766 scm_cons (fd_port, SCM_EOL));
1767 else
1768 rv = ptob->seek (fd_port, off, how);
1769 return scm_from_off_t_or_off64_t (rv);
1770 }
1771 else /* file descriptor?. */
1772 {
1773 off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
1774 off_t_or_off64_t rv;
1775 rv = lseek_or_lseek64 (scm_to_int (fd_port), off, how);
1776 if (rv == -1)
1777 SCM_SYSERROR;
1778 return scm_from_off_t_or_off64_t (rv);
1779 }
1780 }
1781 #undef FUNC_NAME
1782
1783 #ifndef O_BINARY
1784 #define O_BINARY 0
1785 #endif
1786
1787 /* Mingw has ftruncate(), perhaps implemented above using chsize, but
1788 doesn't have the filename version truncate(), hence this code. */
1789 #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE
1790 static int
1791 truncate (const char *file, off_t length)
1792 {
1793 int ret, fdes;
1794
1795 fdes = open (file, O_BINARY | O_WRONLY);
1796 if (fdes == -1)
1797 return -1;
1798
1799 ret = ftruncate (fdes, length);
1800 if (ret == -1)
1801 {
1802 int save_errno = errno;
1803 close (fdes);
1804 errno = save_errno;
1805 return -1;
1806 }
1807
1808 return close (fdes);
1809 }
1810 #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */
1811
1812 SCM_DEFINE (scm_truncate_file, "truncate-file", 1, 1, 0,
1813 (SCM object, SCM length),
1814 "Truncate @var{file} to @var{length} bytes. @var{file} can be a\n"
1815 "filename string, a port object, or an integer file descriptor.\n"
1816 "The return value is unspecified.\n"
1817 "\n"
1818 "For a port or file descriptor @var{length} can be omitted, in\n"
1819 "which case the file is truncated at the current position (per\n"
1820 "@code{ftell} above).\n"
1821 "\n"
1822 "On most systems a file can be extended by giving a length\n"
1823 "greater than the current size, but this is not mandatory in the\n"
1824 "POSIX standard.")
1825 #define FUNC_NAME s_scm_truncate_file
1826 {
1827 int rv;
1828
1829 /* "object" can be a port, fdes or filename.
1830
1831 Negative "length" makes no sense, but it's left to truncate() or
1832 ftruncate() to give back an error for that (normally EINVAL).
1833 */
1834
1835 if (SCM_UNBNDP (length))
1836 {
1837 /* must supply length if object is a filename. */
1838 if (scm_is_string (object))
1839 SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL);
1840
1841 length = scm_seek (object, SCM_INUM0, scm_from_int (SEEK_CUR));
1842 }
1843
1844 object = SCM_COERCE_OUTPORT (object);
1845 if (scm_is_integer (object))
1846 {
1847 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1848 SCM_SYSCALL (rv = ftruncate_or_ftruncate64 (scm_to_int (object),
1849 c_length));
1850 }
1851 else if (SCM_OPOUTPORTP (object))
1852 {
1853 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1854 scm_t_port *pt = SCM_PTAB_ENTRY (object);
1855 scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (object);
1856
1857 if (!ptob->truncate)
1858 SCM_MISC_ERROR ("port is not truncatable", SCM_EOL);
1859 if (pt->rw_active == SCM_PORT_READ)
1860 scm_end_input (object);
1861 else if (pt->rw_active == SCM_PORT_WRITE)
1862 ptob->flush (object);
1863
1864 ptob->truncate (object, c_length);
1865 rv = 0;
1866 }
1867 else
1868 {
1869 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1870 char *str = scm_to_locale_string (object);
1871 int eno;
1872 SCM_SYSCALL (rv = truncate_or_truncate64 (str, c_length));
1873 eno = errno;
1874 free (str);
1875 errno = eno;
1876 }
1877 if (rv == -1)
1878 SCM_SYSERROR;
1879 return SCM_UNSPECIFIED;
1880 }
1881 #undef FUNC_NAME
1882
1883 SCM_DEFINE (scm_port_line, "port-line", 1, 0, 0,
1884 (SCM port),
1885 "Return the current line number for @var{port}.\n"
1886 "\n"
1887 "The first line of a file is 0. But you might want to add 1\n"
1888 "when printing line numbers, since starting from 1 is\n"
1889 "traditional in error messages, and likely to be more natural to\n"
1890 "non-programmers.")
1891 #define FUNC_NAME s_scm_port_line
1892 {
1893 port = SCM_COERCE_OUTPORT (port);
1894 SCM_VALIDATE_OPENPORT (1, port);
1895 return scm_from_long (SCM_LINUM (port));
1896 }
1897 #undef FUNC_NAME
1898
1899 SCM_DEFINE (scm_set_port_line_x, "set-port-line!", 2, 0, 0,
1900 (SCM port, SCM line),
1901 "Set the current line number for @var{port} to @var{line}. The\n"
1902 "first line of a file is 0.")
1903 #define FUNC_NAME s_scm_set_port_line_x
1904 {
1905 port = SCM_COERCE_OUTPORT (port);
1906 SCM_VALIDATE_OPENPORT (1, port);
1907 SCM_PTAB_ENTRY (port)->line_number = scm_to_long (line);
1908 return SCM_UNSPECIFIED;
1909 }
1910 #undef FUNC_NAME
1911
1912 SCM_DEFINE (scm_port_column, "port-column", 1, 0, 0,
1913 (SCM port),
1914 "Return the current column number of @var{port}.\n"
1915 "If the number is\n"
1916 "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n"
1917 "- i.e. the first character of the first line is line 0, column 0.\n"
1918 "(However, when you display a file position, for example in an error\n"
1919 "message, we recommend you add 1 to get 1-origin integers. This is\n"
1920 "because lines and column numbers traditionally start with 1, and that is\n"
1921 "what non-programmers will find most natural.)")
1922 #define FUNC_NAME s_scm_port_column
1923 {
1924 port = SCM_COERCE_OUTPORT (port);
1925 SCM_VALIDATE_OPENPORT (1, port);
1926 return scm_from_int (SCM_COL (port));
1927 }
1928 #undef FUNC_NAME
1929
1930 SCM_DEFINE (scm_set_port_column_x, "set-port-column!", 2, 0, 0,
1931 (SCM port, SCM column),
1932 "Set the current column of @var{port}. Before reading the first\n"
1933 "character on a line the column should be 0.")
1934 #define FUNC_NAME s_scm_set_port_column_x
1935 {
1936 port = SCM_COERCE_OUTPORT (port);
1937 SCM_VALIDATE_OPENPORT (1, port);
1938 SCM_PTAB_ENTRY (port)->column_number = scm_to_int (column);
1939 return SCM_UNSPECIFIED;
1940 }
1941 #undef FUNC_NAME
1942
1943 SCM_DEFINE (scm_port_filename, "port-filename", 1, 0, 0,
1944 (SCM port),
1945 "Return the filename associated with @var{port}. This function returns\n"
1946 "the strings \"standard input\", \"standard output\" and \"standard error\"\n"
1947 "when called on the current input, output and error ports respectively.")
1948 #define FUNC_NAME s_scm_port_filename
1949 {
1950 port = SCM_COERCE_OUTPORT (port);
1951 SCM_VALIDATE_OPENPORT (1, port);
1952 return SCM_FILENAME (port);
1953 }
1954 #undef FUNC_NAME
1955
1956 SCM_DEFINE (scm_set_port_filename_x, "set-port-filename!", 2, 0, 0,
1957 (SCM port, SCM filename),
1958 "Change the filename associated with @var{port}, using the current input\n"
1959 "port if none is specified. Note that this does not change the port's\n"
1960 "source of data, but only the value that is returned by\n"
1961 "@code{port-filename} and reported in diagnostic output.")
1962 #define FUNC_NAME s_scm_set_port_filename_x
1963 {
1964 port = SCM_COERCE_OUTPORT (port);
1965 SCM_VALIDATE_OPENPORT (1, port);
1966 /* We allow the user to set the filename to whatever he likes. */
1967 SCM_SET_FILENAME (port, filename);
1968 return SCM_UNSPECIFIED;
1969 }
1970 #undef FUNC_NAME
1971
1972 /* A fluid specifying the default encoding for newly created ports. If it is
1973 a string, that is the encoding. If it is #f, it is in the "native"
1974 (Latin-1) encoding. */
1975 SCM_VARIABLE (default_port_encoding_var, "%default-port-encoding");
1976
1977 static int scm_port_encoding_init = 0;
1978
1979 /* Use ENCODING as the default encoding for future ports. */
1980 void
1981 scm_i_set_default_port_encoding (const char *encoding)
1982 {
1983 if (!scm_port_encoding_init
1984 || !scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var)))
1985 scm_misc_error (NULL, "tried to set port encoding fluid before it is initialized",
1986 SCM_EOL);
1987
1988 if (encoding == NULL
1989 || !strcmp (encoding, "ASCII")
1990 || !strcmp (encoding, "ANSI_X3.4-1968")
1991 || !strcmp (encoding, "ISO-8859-1"))
1992 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), SCM_BOOL_F);
1993 else
1994 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var),
1995 scm_from_locale_string (encoding));
1996 }
1997
1998 /* Return the name of the default encoding for newly created ports; a
1999 return value of NULL means "ISO-8859-1". */
2000 const char *
2001 scm_i_default_port_encoding (void)
2002 {
2003 if (!scm_port_encoding_init)
2004 return NULL;
2005 else if (!scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var)))
2006 return NULL;
2007 else
2008 {
2009 SCM encoding;
2010
2011 encoding = scm_fluid_ref (SCM_VARIABLE_REF (default_port_encoding_var));
2012 if (!scm_is_string (encoding))
2013 return NULL;
2014 else
2015 return scm_i_string_chars (encoding);
2016 }
2017 }
2018
2019 void
2020 scm_i_set_port_encoding_x (SCM port, const char *encoding)
2021 {
2022 scm_t_port *pt;
2023 iconv_t new_input_cd, new_output_cd;
2024
2025 new_input_cd = (iconv_t) -1;
2026 new_output_cd = (iconv_t) -1;
2027
2028 /* Set the character encoding for this port. */
2029 pt = SCM_PTAB_ENTRY (port);
2030
2031 if (encoding == NULL)
2032 encoding = "ISO-8859-1";
2033
2034 pt->encoding = scm_gc_strdup (encoding, "port");
2035
2036 if (SCM_CELL_WORD_0 (port) & SCM_RDNG)
2037 {
2038 /* Open an input iconv conversion descriptor, from ENCODING
2039 to UTF-8. We choose UTF-8, not UTF-32, because iconv
2040 implementations can typically convert from anything to
2041 UTF-8, but not to UTF-32 (see
2042 <http://lists.gnu.org/archive/html/bug-libunistring/2010-09/msg00007.html>). */
2043 new_input_cd = iconv_open ("UTF-8", encoding);
2044 if (new_input_cd == (iconv_t) -1)
2045 goto invalid_encoding;
2046 }
2047
2048 if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
2049 {
2050 new_output_cd = iconv_open (encoding, "UTF-8");
2051 if (new_output_cd == (iconv_t) -1)
2052 {
2053 if (new_input_cd != (iconv_t) -1)
2054 iconv_close (new_input_cd);
2055 goto invalid_encoding;
2056 }
2057 }
2058
2059 if (pt->input_cd != (iconv_t) -1)
2060 iconv_close (pt->input_cd);
2061 if (pt->output_cd != (iconv_t) -1)
2062 iconv_close (pt->output_cd);
2063
2064 pt->input_cd = new_input_cd;
2065 pt->output_cd = new_output_cd;
2066
2067 return;
2068
2069 invalid_encoding:
2070 {
2071 SCM err;
2072 err = scm_from_locale_string (encoding);
2073 scm_misc_error ("scm_i_set_port_encoding_x",
2074 "invalid or unknown character encoding ~s",
2075 scm_list_1 (err));
2076 }
2077 }
2078
2079 SCM_DEFINE (scm_port_encoding, "port-encoding", 1, 0, 0,
2080 (SCM port),
2081 "Returns, as a string, the character encoding that @var{port}\n"
2082 "uses to interpret its input and output.\n")
2083 #define FUNC_NAME s_scm_port_encoding
2084 {
2085 scm_t_port *pt;
2086 const char *enc;
2087
2088 SCM_VALIDATE_PORT (1, port);
2089
2090 pt = SCM_PTAB_ENTRY (port);
2091 enc = pt->encoding;
2092 if (enc)
2093 return scm_from_locale_string (pt->encoding);
2094 else
2095 return SCM_BOOL_F;
2096 }
2097 #undef FUNC_NAME
2098
2099 SCM_DEFINE (scm_set_port_encoding_x, "set-port-encoding!", 2, 0, 0,
2100 (SCM port, SCM enc),
2101 "Sets the character encoding that will be used to interpret all\n"
2102 "port I/O. New ports are created with the encoding\n"
2103 "appropriate for the current locale if @code{setlocale} has \n"
2104 "been called or ISO-8859-1 otherwise\n"
2105 "and this procedure can be used to modify that encoding.\n")
2106 #define FUNC_NAME s_scm_set_port_encoding_x
2107 {
2108 char *enc_str;
2109
2110 SCM_VALIDATE_PORT (1, port);
2111 SCM_VALIDATE_STRING (2, enc);
2112
2113 enc_str = scm_to_locale_string (enc);
2114 scm_i_set_port_encoding_x (port, enc_str);
2115 free (enc_str);
2116
2117 return SCM_UNSPECIFIED;
2118 }
2119 #undef FUNC_NAME
2120
2121
2122 /* This determines how conversions handle unconvertible characters. */
2123 SCM_GLOBAL_VARIABLE (scm_conversion_strategy, "%port-conversion-strategy");
2124 static int scm_conversion_strategy_init = 0;
2125
2126 scm_t_string_failed_conversion_handler
2127 scm_i_get_conversion_strategy (SCM port)
2128 {
2129 SCM encoding;
2130
2131 if (scm_is_false (port))
2132 {
2133 if (!scm_conversion_strategy_init
2134 || !scm_is_fluid (SCM_VARIABLE_REF (scm_conversion_strategy)))
2135 return SCM_FAILED_CONVERSION_QUESTION_MARK;
2136 else
2137 {
2138 encoding = scm_fluid_ref (SCM_VARIABLE_REF (scm_conversion_strategy));
2139 if (scm_is_false (encoding))
2140 return SCM_FAILED_CONVERSION_QUESTION_MARK;
2141 else
2142 return (scm_t_string_failed_conversion_handler) scm_to_int (encoding);
2143 }
2144 }
2145 else
2146 {
2147 scm_t_port *pt;
2148 pt = SCM_PTAB_ENTRY (port);
2149 return pt->ilseq_handler;
2150 }
2151
2152 }
2153
2154 void
2155 scm_i_set_conversion_strategy_x (SCM port,
2156 scm_t_string_failed_conversion_handler handler)
2157 {
2158 SCM strategy;
2159 scm_t_port *pt;
2160
2161 strategy = scm_from_int ((int) handler);
2162
2163 if (scm_is_false (port))
2164 {
2165 /* Set the default encoding for future ports. */
2166 if (!scm_conversion_strategy
2167 || !scm_is_fluid (SCM_VARIABLE_REF (scm_conversion_strategy)))
2168 scm_misc_error (NULL, "tried to set conversion strategy fluid before it is initialized",
2169 SCM_EOL);
2170 scm_fluid_set_x (SCM_VARIABLE_REF (scm_conversion_strategy), strategy);
2171 }
2172 else
2173 {
2174 /* Set the character encoding for this port. */
2175 pt = SCM_PTAB_ENTRY (port);
2176 pt->ilseq_handler = handler;
2177 }
2178 }
2179
2180 SCM_DEFINE (scm_port_conversion_strategy, "port-conversion-strategy",
2181 1, 0, 0, (SCM port),
2182 "Returns the behavior of the port when handling a character that\n"
2183 "is not representable in the port's current encoding.\n"
2184 "It returns the symbol @code{error} if unrepresentable characters\n"
2185 "should cause exceptions, @code{substitute} if the port should\n"
2186 "try to replace unrepresentable characters with question marks or\n"
2187 "approximate characters, or @code{escape} if unrepresentable\n"
2188 "characters should be converted to string escapes.\n"
2189 "\n"
2190 "If @var{port} is @code{#f}, then the current default behavior\n"
2191 "will be returned. New ports will have this default behavior\n"
2192 "when they are created.\n")
2193 #define FUNC_NAME s_scm_port_conversion_strategy
2194 {
2195 scm_t_string_failed_conversion_handler h;
2196
2197 SCM_VALIDATE_OPPORT (1, port);
2198
2199 if (!scm_is_false (port))
2200 {
2201 SCM_VALIDATE_OPPORT (1, port);
2202 }
2203
2204 h = scm_i_get_conversion_strategy (port);
2205 if (h == SCM_FAILED_CONVERSION_ERROR)
2206 return scm_from_latin1_symbol ("error");
2207 else if (h == SCM_FAILED_CONVERSION_QUESTION_MARK)
2208 return scm_from_latin1_symbol ("substitute");
2209 else if (h == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
2210 return scm_from_latin1_symbol ("escape");
2211 else
2212 abort ();
2213
2214 /* Never gets here. */
2215 return SCM_UNDEFINED;
2216 }
2217 #undef FUNC_NAME
2218
2219 SCM_DEFINE (scm_set_port_conversion_strategy_x, "set-port-conversion-strategy!",
2220 2, 0, 0,
2221 (SCM port, SCM sym),
2222 "Sets the behavior of the interpreter when outputting a character\n"
2223 "that is not representable in the port's current encoding.\n"
2224 "@var{sym} can be either @code{'error}, @code{'substitute}, or\n"
2225 "@code{'escape}. If it is @code{'error}, an error will be thrown\n"
2226 "when an unconvertible character is encountered. If it is\n"
2227 "@code{'substitute}, then unconvertible characters will \n"
2228 "be replaced with approximate characters, or with question marks\n"
2229 "if no approximately correct character is available.\n"
2230 "If it is @code{'escape},\n"
2231 "it will appear as a hex escape when output.\n"
2232 "\n"
2233 "If @var{port} is an open port, the conversion error behavior\n"
2234 "is set for that port. If it is @code{#f}, it is set as the\n"
2235 "default behavior for any future ports that get created in\n"
2236 "this thread.\n")
2237 #define FUNC_NAME s_scm_set_port_conversion_strategy_x
2238 {
2239 SCM err;
2240 SCM qm;
2241 SCM esc;
2242
2243 if (!scm_is_false (port))
2244 {
2245 SCM_VALIDATE_OPPORT (1, port);
2246 }
2247
2248 err = scm_from_latin1_symbol ("error");
2249 if (scm_is_true (scm_eqv_p (sym, err)))
2250 {
2251 scm_i_set_conversion_strategy_x (port, SCM_FAILED_CONVERSION_ERROR);
2252 return SCM_UNSPECIFIED;
2253 }
2254
2255 qm = scm_from_latin1_symbol ("substitute");
2256 if (scm_is_true (scm_eqv_p (sym, qm)))
2257 {
2258 scm_i_set_conversion_strategy_x (port,
2259 SCM_FAILED_CONVERSION_QUESTION_MARK);
2260 return SCM_UNSPECIFIED;
2261 }
2262
2263 esc = scm_from_latin1_symbol ("escape");
2264 if (scm_is_true (scm_eqv_p (sym, esc)))
2265 {
2266 scm_i_set_conversion_strategy_x (port,
2267 SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE);
2268 return SCM_UNSPECIFIED;
2269 }
2270
2271 SCM_MISC_ERROR ("unknown conversion behavior ~s", scm_list_1 (sym));
2272
2273 return SCM_UNSPECIFIED;
2274 }
2275 #undef FUNC_NAME
2276
2277
2278
2279 void
2280 scm_print_port_mode (SCM exp, SCM port)
2281 {
2282 scm_puts (SCM_CLOSEDP (exp)
2283 ? "closed: "
2284 : (SCM_RDNG & SCM_CELL_WORD_0 (exp)
2285 ? (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
2286 ? "input-output: "
2287 : "input: ")
2288 : (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
2289 ? "output: "
2290 : "bogus: ")),
2291 port);
2292 }
2293
2294 int
2295 scm_port_print (SCM exp, SCM port, scm_print_state *pstate SCM_UNUSED)
2296 {
2297 char *type = SCM_PTOBNAME (SCM_PTOBNUM (exp));
2298 if (!type)
2299 type = "port";
2300 scm_puts ("#<", port);
2301 scm_print_port_mode (exp, port);
2302 scm_puts (type, port);
2303 scm_putc (' ', port);
2304 scm_uintprint (SCM_CELL_WORD_1 (exp), 16, port);
2305 scm_putc ('>', port);
2306 return 1;
2307 }
2308
2309 \f
2310
2311 /* Void ports. */
2312
2313 scm_t_bits scm_tc16_void_port = 0;
2314
2315 static int fill_input_void_port (SCM port SCM_UNUSED)
2316 {
2317 return EOF;
2318 }
2319
2320 static void
2321 write_void_port (SCM port SCM_UNUSED,
2322 const void *data SCM_UNUSED,
2323 size_t size SCM_UNUSED)
2324 {
2325 }
2326
2327 static SCM
2328 scm_i_void_port (long mode_bits)
2329 {
2330 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
2331 {
2332 SCM answer = scm_new_port_table_entry (scm_tc16_void_port);
2333 scm_t_port * pt = SCM_PTAB_ENTRY(answer);
2334
2335 scm_port_non_buffer (pt);
2336
2337 SCM_SETSTREAM (answer, 0);
2338 SCM_SET_CELL_TYPE (answer, scm_tc16_void_port | mode_bits);
2339 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
2340 return answer;
2341 }
2342 }
2343
2344 SCM
2345 scm_void_port (char *mode_str)
2346 {
2347 return scm_i_void_port (scm_mode_bits (mode_str));
2348 }
2349
2350 SCM_DEFINE (scm_sys_make_void_port, "%make-void-port", 1, 0, 0,
2351 (SCM mode),
2352 "Create and return a new void port. A void port acts like\n"
2353 "@file{/dev/null}. The @var{mode} argument\n"
2354 "specifies the input/output modes for this port: see the\n"
2355 "documentation for @code{open-file} in @ref{File Ports}.")
2356 #define FUNC_NAME s_scm_sys_make_void_port
2357 {
2358 return scm_i_void_port (scm_i_mode_bits (mode));
2359 }
2360 #undef FUNC_NAME
2361
2362 \f
2363 /* Initialization. */
2364
2365 void
2366 scm_init_ports ()
2367 {
2368 /* lseek() symbols. */
2369 scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET));
2370 scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR));
2371 scm_c_define ("SEEK_END", scm_from_int (SEEK_END));
2372
2373 scm_tc16_void_port = scm_make_port_type ("void", fill_input_void_port,
2374 write_void_port);
2375
2376 cur_inport_fluid = scm_make_fluid ();
2377 cur_outport_fluid = scm_make_fluid ();
2378 cur_errport_fluid = scm_make_fluid ();
2379 cur_loadport_fluid = scm_make_fluid ();
2380
2381 scm_i_port_weak_hash = scm_make_weak_key_hash_table (SCM_I_MAKINUM(31));
2382
2383 #include "libguile/ports.x"
2384
2385 /* Use Latin-1 as the default port encoding. */
2386 SCM_VARIABLE_SET (default_port_encoding_var, scm_make_fluid ());
2387 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), SCM_BOOL_F);
2388 scm_port_encoding_init = 1;
2389
2390 SCM_VARIABLE_SET (scm_conversion_strategy, scm_make_fluid ());
2391 scm_fluid_set_x (SCM_VARIABLE_REF (scm_conversion_strategy),
2392 scm_from_int ((int) SCM_FAILED_CONVERSION_QUESTION_MARK));
2393 scm_conversion_strategy_init = 1;
2394
2395 }
2396
2397 /*
2398 Local Variables:
2399 c-file-style: "gnu"
2400 End:
2401 */