Merge branch 'bdw-gc-static-alloc'
[bpt/guile.git] / libguile / ports.c
1 /* Copyright (C) 1995,1996,1997,1998,1999,2000,2001, 2003, 2004, 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
2 *
3 * This library is free software; you can redistribute it and/or
4 * modify it under the terms of the GNU Lesser General Public License
5 * as published by the Free Software Foundation; either version 3 of
6 * the License, or (at your option) any later version.
7 *
8 * This library is distributed in the hope that it will be useful, but
9 * WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * Lesser General Public License for more details.
12 *
13 * You should have received a copy of the GNU Lesser General Public
14 * License along with this library; if not, write to the Free Software
15 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
16 * 02110-1301 USA
17 */
18
19
20 \f
21 /* Headers. */
22
23 #define _LARGEFILE64_SOURCE /* ask for stat64 etc */
24
25 #ifdef HAVE_CONFIG_H
26 # include <config.h>
27 #endif
28
29 #include <stdio.h>
30 #include <errno.h>
31 #include <fcntl.h> /* for chsize on mingw */
32 #include <assert.h>
33 #include <uniconv.h>
34 #include <unistr.h>
35 #include <striconveh.h>
36
37 #include <assert.h>
38
39 #include "libguile/_scm.h"
40 #include "libguile/async.h"
41 #include "libguile/eval.h"
42 #include "libguile/fports.h" /* direct access for seek and truncate */
43 #include "libguile/objects.h"
44 #include "libguile/goops.h"
45 #include "libguile/smob.h"
46 #include "libguile/chars.h"
47 #include "libguile/dynwind.h"
48
49 #include "libguile/keywords.h"
50 #include "libguile/hashtab.h"
51 #include "libguile/root.h"
52 #include "libguile/strings.h"
53 #include "libguile/mallocs.h"
54 #include "libguile/validate.h"
55 #include "libguile/ports.h"
56 #include "libguile/vectors.h"
57 #include "libguile/weaks.h"
58 #include "libguile/fluids.h"
59 #include "libguile/eq.h"
60
61 #ifdef HAVE_STRING_H
62 #include <string.h>
63 #endif
64
65 #ifdef HAVE_IO_H
66 #include <io.h>
67 #endif
68
69 #ifdef HAVE_UNISTD_H
70 #include <unistd.h>
71 #endif
72
73 #ifdef HAVE_SYS_IOCTL_H
74 #include <sys/ioctl.h>
75 #endif
76
77 /* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize
78 already, but have this code here in case that wasn't so in past versions,
79 or perhaps to help other minimal DOS environments.
80
81 gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which
82 might be possibilities if we've got other systems without ftruncate. */
83
84 #if HAVE_CHSIZE && ! HAVE_FTRUNCATE
85 #define ftruncate(fd, size) chsize (fd, size)
86 #undef HAVE_FTRUNCATE
87 #define HAVE_FTRUNCATE 1
88 #endif
89
90 \f
91 /* The port kind table --- a dynamically resized array of port types. */
92
93
94 /* scm_ptobs scm_numptob
95 * implement a dynamically resized array of ptob records.
96 * Indexes into this table are used when generating type
97 * tags for smobjects (if you know a tag you can get an index and conversely).
98 */
99 scm_t_ptob_descriptor *scm_ptobs;
100 long scm_numptob;
101
102 /* GC marker for a port with stream of SCM type. */
103 SCM
104 scm_markstream (SCM ptr)
105 {
106 int openp;
107 openp = SCM_CELL_WORD_0 (ptr) & SCM_OPN;
108 if (openp)
109 return SCM_PACK (SCM_STREAM (ptr));
110 else
111 return SCM_BOOL_F;
112 }
113
114 /*
115 * We choose to use an interface similar to the smob interface with
116 * fill_input and write as standard fields, passed to the port
117 * type constructor, and optional fields set by setters.
118 */
119
120 static void
121 flush_port_default (SCM port SCM_UNUSED)
122 {
123 }
124
125 static void
126 end_input_default (SCM port SCM_UNUSED, int offset SCM_UNUSED)
127 {
128 }
129
130 scm_t_bits
131 scm_make_port_type (char *name,
132 int (*fill_input) (SCM port),
133 void (*write) (SCM port, const void *data, size_t size))
134 {
135 char *tmp;
136 if (SCM_I_MAX_PORT_TYPE_COUNT - 1 <= scm_numptob)
137 goto ptoberr;
138 SCM_CRITICAL_SECTION_START;
139 tmp = (char *) scm_gc_realloc ((char *) scm_ptobs,
140 scm_numptob * sizeof (scm_t_ptob_descriptor),
141 (1 + scm_numptob)
142 * sizeof (scm_t_ptob_descriptor),
143 "port-type");
144 if (tmp)
145 {
146 scm_ptobs = (scm_t_ptob_descriptor *) tmp;
147
148 scm_ptobs[scm_numptob].name = name;
149 scm_ptobs[scm_numptob].mark = 0;
150 scm_ptobs[scm_numptob].free = NULL;
151 scm_ptobs[scm_numptob].print = scm_port_print;
152 scm_ptobs[scm_numptob].equalp = 0;
153 scm_ptobs[scm_numptob].close = 0;
154
155 scm_ptobs[scm_numptob].write = write;
156 scm_ptobs[scm_numptob].flush = flush_port_default;
157
158 scm_ptobs[scm_numptob].end_input = end_input_default;
159 scm_ptobs[scm_numptob].fill_input = fill_input;
160 scm_ptobs[scm_numptob].input_waiting = 0;
161
162 scm_ptobs[scm_numptob].seek = 0;
163 scm_ptobs[scm_numptob].truncate = 0;
164
165 scm_numptob++;
166 }
167 SCM_CRITICAL_SECTION_END;
168 if (!tmp)
169 {
170 ptoberr:
171 scm_memory_error ("scm_make_port_type");
172 }
173 /* Make a class object if Goops is present */
174 if (SCM_UNPACK (scm_port_class[0]) != 0)
175 scm_make_port_classes (scm_numptob - 1, SCM_PTOBNAME (scm_numptob - 1));
176 return scm_tc7_port + (scm_numptob - 1) * 256;
177 }
178
179 void
180 scm_set_port_mark (scm_t_bits tc, SCM (*mark) (SCM))
181 {
182 scm_ptobs[SCM_TC2PTOBNUM (tc)].mark = mark;
183 }
184
185 void
186 scm_set_port_free (scm_t_bits tc, size_t (*free) (SCM))
187 {
188 scm_ptobs[SCM_TC2PTOBNUM (tc)].free = free;
189 }
190
191 void
192 scm_set_port_print (scm_t_bits tc, int (*print) (SCM exp, SCM port,
193 scm_print_state *pstate))
194 {
195 scm_ptobs[SCM_TC2PTOBNUM (tc)].print = print;
196 }
197
198 void
199 scm_set_port_equalp (scm_t_bits tc, SCM (*equalp) (SCM, SCM))
200 {
201 scm_ptobs[SCM_TC2PTOBNUM (tc)].equalp = equalp;
202 }
203
204 void
205 scm_set_port_flush (scm_t_bits tc, void (*flush) (SCM port))
206 {
207 scm_ptobs[SCM_TC2PTOBNUM (tc)].flush = flush;
208 }
209
210 void
211 scm_set_port_end_input (scm_t_bits tc, void (*end_input) (SCM port, int offset))
212 {
213 scm_ptobs[SCM_TC2PTOBNUM (tc)].end_input = end_input;
214 }
215
216 void
217 scm_set_port_close (scm_t_bits tc, int (*close) (SCM))
218 {
219 scm_ptobs[SCM_TC2PTOBNUM (tc)].close = close;
220 }
221
222 void
223 scm_set_port_seek (scm_t_bits tc,
224 scm_t_off (*seek) (SCM, scm_t_off, int))
225 {
226 scm_ptobs[SCM_TC2PTOBNUM (tc)].seek = seek;
227 }
228
229 void
230 scm_set_port_truncate (scm_t_bits tc, void (*truncate) (SCM, scm_t_off))
231 {
232 scm_ptobs[SCM_TC2PTOBNUM (tc)].truncate = truncate;
233 }
234
235 void
236 scm_set_port_input_waiting (scm_t_bits tc, int (*input_waiting) (SCM))
237 {
238 scm_ptobs[SCM_TC2PTOBNUM (tc)].input_waiting = input_waiting;
239 }
240
241 \f
242
243 SCM_DEFINE (scm_char_ready_p, "char-ready?", 0, 1, 0,
244 (SCM port),
245 "Return @code{#t} if a character is ready on input @var{port}\n"
246 "and return @code{#f} otherwise. If @code{char-ready?} returns\n"
247 "@code{#t} then the next @code{read-char} operation on\n"
248 "@var{port} is guaranteed not to hang. If @var{port} is a file\n"
249 "port at end of file then @code{char-ready?} returns @code{#t}.\n"
250 "\n"
251 "@code{char-ready?} exists to make it possible for a\n"
252 "program to accept characters from interactive ports without\n"
253 "getting stuck waiting for input. Any input editors associated\n"
254 "with such ports must make sure that characters whose existence\n"
255 "has been asserted by @code{char-ready?} cannot be rubbed out.\n"
256 "If @code{char-ready?} were to return @code{#f} at end of file,\n"
257 "a port at end of file would be indistinguishable from an\n"
258 "interactive port that has no ready characters.")
259 #define FUNC_NAME s_scm_char_ready_p
260 {
261 scm_t_port *pt;
262
263 if (SCM_UNBNDP (port))
264 port = scm_current_input_port ();
265 else
266 SCM_VALIDATE_OPINPORT (1, port);
267
268 pt = SCM_PTAB_ENTRY (port);
269
270 /* if the current read buffer is filled, or the
271 last pushed-back char has been read and the saved buffer is
272 filled, result is true. */
273 if (pt->read_pos < pt->read_end
274 || (pt->read_buf == pt->putback_buf
275 && pt->saved_read_pos < pt->saved_read_end))
276 return SCM_BOOL_T;
277 else
278 {
279 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
280
281 if (ptob->input_waiting)
282 return scm_from_bool(ptob->input_waiting (port));
283 else
284 return SCM_BOOL_T;
285 }
286 }
287 #undef FUNC_NAME
288
289 /* move up to read_len chars from port's putback and/or read buffers
290 into memory starting at dest. returns the number of chars moved. */
291 size_t scm_take_from_input_buffers (SCM port, char *dest, size_t read_len)
292 {
293 scm_t_port *pt = SCM_PTAB_ENTRY (port);
294 size_t chars_read = 0;
295 size_t from_buf = min (pt->read_end - pt->read_pos, read_len);
296
297 if (from_buf > 0)
298 {
299 memcpy (dest, pt->read_pos, from_buf);
300 pt->read_pos += from_buf;
301 chars_read += from_buf;
302 read_len -= from_buf;
303 dest += from_buf;
304 }
305
306 /* if putback was active, try the real input buffer too. */
307 if (pt->read_buf == pt->putback_buf)
308 {
309 from_buf = min (pt->saved_read_end - pt->saved_read_pos, read_len);
310 if (from_buf > 0)
311 {
312 memcpy (dest, pt->saved_read_pos, from_buf);
313 pt->saved_read_pos += from_buf;
314 chars_read += from_buf;
315 }
316 }
317 return chars_read;
318 }
319
320 /* Clear a port's read buffers, returning the contents. */
321 SCM_DEFINE (scm_drain_input, "drain-input", 1, 0, 0,
322 (SCM port),
323 "This procedure clears a port's input buffers, similar\n"
324 "to the way that force-output clears the output buffer. The\n"
325 "contents of the buffers are returned as a single string, e.g.,\n"
326 "\n"
327 "@lisp\n"
328 "(define p (open-input-file ...))\n"
329 "(drain-input p) => empty string, nothing buffered yet.\n"
330 "(unread-char (read-char p) p)\n"
331 "(drain-input p) => initial chars from p, up to the buffer size.\n"
332 "@end lisp\n\n"
333 "Draining the buffers may be useful for cleanly finishing\n"
334 "buffered I/O so that the file descriptor can be used directly\n"
335 "for further input.")
336 #define FUNC_NAME s_scm_drain_input
337 {
338 SCM result;
339 char *data;
340 scm_t_port *pt;
341 long count;
342
343 SCM_VALIDATE_OPINPORT (1, port);
344 pt = SCM_PTAB_ENTRY (port);
345
346 count = pt->read_end - pt->read_pos;
347 if (pt->read_buf == pt->putback_buf)
348 count += pt->saved_read_end - pt->saved_read_pos;
349
350 result = scm_i_make_string (count, &data);
351 scm_take_from_input_buffers (port, data, count);
352 return result;
353 }
354 #undef FUNC_NAME
355
356 \f
357 /* Standard ports --- current input, output, error, and more(!). */
358
359 static SCM cur_inport_fluid = 0;
360 static SCM cur_outport_fluid = 0;
361 static SCM cur_errport_fluid = 0;
362 static SCM cur_loadport_fluid = 0;
363
364 SCM_DEFINE (scm_current_input_port, "current-input-port", 0, 0, 0,
365 (),
366 "Return the current input port. This is the default port used\n"
367 "by many input procedures. Initially, @code{current-input-port}\n"
368 "returns the @dfn{standard input} in Unix and C terminology.")
369 #define FUNC_NAME s_scm_current_input_port
370 {
371 if (cur_inport_fluid)
372 return scm_fluid_ref (cur_inport_fluid);
373 else
374 return SCM_BOOL_F;
375 }
376 #undef FUNC_NAME
377
378 SCM_DEFINE (scm_current_output_port, "current-output-port", 0, 0, 0,
379 (),
380 "Return the current output port. This is the default port used\n"
381 "by many output procedures. Initially,\n"
382 "@code{current-output-port} returns the @dfn{standard output} in\n"
383 "Unix and C terminology.")
384 #define FUNC_NAME s_scm_current_output_port
385 {
386 if (cur_outport_fluid)
387 return scm_fluid_ref (cur_outport_fluid);
388 else
389 return SCM_BOOL_F;
390 }
391 #undef FUNC_NAME
392
393 SCM_DEFINE (scm_current_error_port, "current-error-port", 0, 0, 0,
394 (),
395 "Return the port to which errors and warnings should be sent (the\n"
396 "@dfn{standard error} in Unix and C terminology).")
397 #define FUNC_NAME s_scm_current_error_port
398 {
399 if (cur_errport_fluid)
400 return scm_fluid_ref (cur_errport_fluid);
401 else
402 return SCM_BOOL_F;
403 }
404 #undef FUNC_NAME
405
406 SCM_DEFINE (scm_current_load_port, "current-load-port", 0, 0, 0,
407 (),
408 "Return the current-load-port.\n"
409 "The load port is used internally by @code{primitive-load}.")
410 #define FUNC_NAME s_scm_current_load_port
411 {
412 return scm_fluid_ref (cur_loadport_fluid);
413 }
414 #undef FUNC_NAME
415
416 SCM_DEFINE (scm_set_current_input_port, "set-current-input-port", 1, 0, 0,
417 (SCM port),
418 "@deffnx {Scheme Procedure} set-current-output-port port\n"
419 "@deffnx {Scheme Procedure} set-current-error-port port\n"
420 "Change the ports returned by @code{current-input-port},\n"
421 "@code{current-output-port} and @code{current-error-port}, respectively,\n"
422 "so that they use the supplied @var{port} for input or output.")
423 #define FUNC_NAME s_scm_set_current_input_port
424 {
425 SCM oinp = scm_fluid_ref (cur_inport_fluid);
426 SCM_VALIDATE_OPINPORT (1, port);
427 scm_fluid_set_x (cur_inport_fluid, port);
428 return oinp;
429 }
430 #undef FUNC_NAME
431
432
433 SCM_DEFINE (scm_set_current_output_port, "set-current-output-port", 1, 0, 0,
434 (SCM port),
435 "Set the current default output port to @var{port}.")
436 #define FUNC_NAME s_scm_set_current_output_port
437 {
438 SCM ooutp = scm_fluid_ref (cur_outport_fluid);
439 port = SCM_COERCE_OUTPORT (port);
440 SCM_VALIDATE_OPOUTPORT (1, port);
441 scm_fluid_set_x (cur_outport_fluid, port);
442 return ooutp;
443 }
444 #undef FUNC_NAME
445
446
447 SCM_DEFINE (scm_set_current_error_port, "set-current-error-port", 1, 0, 0,
448 (SCM port),
449 "Set the current default error port to @var{port}.")
450 #define FUNC_NAME s_scm_set_current_error_port
451 {
452 SCM oerrp = scm_fluid_ref (cur_errport_fluid);
453 port = SCM_COERCE_OUTPORT (port);
454 SCM_VALIDATE_OPOUTPORT (1, port);
455 scm_fluid_set_x (cur_errport_fluid, port);
456 return oerrp;
457 }
458 #undef FUNC_NAME
459
460 void
461 scm_dynwind_current_input_port (SCM port)
462 #define FUNC_NAME NULL
463 {
464 SCM_VALIDATE_OPINPORT (1, port);
465 scm_dynwind_fluid (cur_inport_fluid, port);
466 }
467 #undef FUNC_NAME
468
469 void
470 scm_dynwind_current_output_port (SCM port)
471 #define FUNC_NAME NULL
472 {
473 port = SCM_COERCE_OUTPORT (port);
474 SCM_VALIDATE_OPOUTPORT (1, port);
475 scm_dynwind_fluid (cur_outport_fluid, port);
476 }
477 #undef FUNC_NAME
478
479 void
480 scm_dynwind_current_error_port (SCM port)
481 #define FUNC_NAME NULL
482 {
483 port = SCM_COERCE_OUTPORT (port);
484 SCM_VALIDATE_OPOUTPORT (1, port);
485 scm_dynwind_fluid (cur_errport_fluid, port);
486 }
487 #undef FUNC_NAME
488
489 void
490 scm_i_dynwind_current_load_port (SCM port)
491 {
492 scm_dynwind_fluid (cur_loadport_fluid, port);
493 }
494
495 \f
496 /* The port table --- an array of pointers to ports. */
497
498 /*
499 We need a global registry of ports to flush them all at exit, and to
500 get all the ports matching a file descriptor.
501 */
502 SCM scm_i_port_weak_hash;
503
504 scm_i_pthread_mutex_t scm_i_port_table_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER;
505
506 \f
507 /* Port finalization. */
508
509
510 static void finalize_port (GC_PTR, GC_PTR);
511
512 /* Register a finalizer for PORT, if needed by its port type. */
513 static SCM_C_INLINE_KEYWORD void
514 register_finalizer_for_port (SCM port)
515 {
516 long port_type;
517
518 port_type = SCM_TC2PTOBNUM (SCM_CELL_TYPE (port));
519 if (scm_ptobs[port_type].free)
520 {
521 GC_finalization_proc prev_finalizer;
522 GC_PTR prev_finalization_data;
523
524 GC_REGISTER_FINALIZER_NO_ORDER (SCM2PTR (port), finalize_port, 0,
525 &prev_finalizer,
526 &prev_finalization_data);
527 }
528 }
529
530 /* Finalize the object (a port) pointed to by PTR. */
531 static void
532 finalize_port (GC_PTR ptr, GC_PTR data)
533 {
534 long port_type;
535 SCM port = PTR2SCM (ptr);
536
537 if (!SCM_PORTP (port))
538 abort ();
539
540 if (SCM_OPENP (port))
541 {
542 if (SCM_REVEALED (port) > 0)
543 /* Keep "revealed" ports alive and re-register a finalizer. */
544 register_finalizer_for_port (port);
545 else
546 {
547 port_type = SCM_TC2PTOBNUM (SCM_CELL_TYPE (port));
548 if (port_type >= scm_numptob)
549 abort ();
550
551 if (scm_ptobs[port_type].free)
552 /* Yes, I really do mean `.free' rather than `.close'. `.close'
553 is for explicit `close-port' by user. */
554 scm_ptobs[port_type].free (port);
555
556 SCM_SETSTREAM (port, 0);
557 SCM_CLR_PORT_OPEN_FLAG (port);
558
559 scm_gc_ports_collected++;
560 }
561 }
562 }
563
564
565
566 \f
567
568 /* This function is not and should not be thread safe. */
569 SCM
570 scm_new_port_table_entry (scm_t_bits tag)
571 #define FUNC_NAME "scm_new_port_table_entry"
572 {
573 /*
574 We initialize the cell to empty, this is in case scm_gc_calloc
575 triggers GC ; we don't want the GC to scan a half-finished Z.
576 */
577
578 SCM z = scm_cons (SCM_EOL, SCM_EOL);
579 scm_t_port *entry = (scm_t_port *) scm_gc_calloc (sizeof (scm_t_port), "port");
580 const char *enc;
581
582 entry->file_name = SCM_BOOL_F;
583 entry->rw_active = SCM_PORT_NEITHER;
584 entry->port = z;
585 /* Initialize this port with the thread's current default
586 encoding. */
587 if ((enc = scm_i_get_port_encoding (SCM_BOOL_F)) == NULL)
588 entry->encoding = NULL;
589 else
590 entry->encoding = scm_gc_strdup (enc, "port");
591 entry->ilseq_handler = scm_i_get_conversion_strategy (SCM_BOOL_F);
592
593 SCM_SET_CELL_TYPE (z, tag);
594 SCM_SETPTAB_ENTRY (z, entry);
595
596 scm_hashq_set_x (scm_i_port_weak_hash, z, SCM_BOOL_F);
597
598 /* For each new port, register a finalizer so that it port type's free
599 function can be invoked eventually. */
600 register_finalizer_for_port (z);
601
602 return z;
603 }
604 #undef FUNC_NAME
605
606 #if SCM_ENABLE_DEPRECATED==1
607 SCM_API scm_t_port *
608 scm_add_to_port_table (SCM port)
609 {
610 SCM z = scm_new_port_table_entry (scm_tc7_port);
611 scm_t_port * pt = SCM_PTAB_ENTRY(z);
612
613 pt->port = port;
614 SCM_SETCAR (z, SCM_EOL);
615 SCM_SETCDR (z, SCM_EOL);
616 SCM_SETPTAB_ENTRY (port, pt);
617 return pt;
618 }
619 #endif
620
621
622 /* Remove a port from the table and destroy it. */
623
624 /* This function is not and should not be thread safe. */
625 void
626 scm_i_remove_port (SCM port)
627 #define FUNC_NAME "scm_remove_port"
628 {
629 scm_t_port *p = SCM_PTAB_ENTRY (port);
630
631 scm_port_non_buffer (p);
632
633 p->putback_buf = NULL;
634 p->putback_buf_size = 0;
635
636 SCM_SETPTAB_ENTRY (port, 0);
637 scm_hashq_remove_x (scm_i_port_weak_hash, port);
638 }
639 #undef FUNC_NAME
640
641
642 /* Functions for debugging. */
643 #ifdef GUILE_DEBUG
644 SCM_DEFINE (scm_pt_size, "pt-size", 0, 0, 0,
645 (),
646 "Return the number of ports in the port table. @code{pt-size}\n"
647 "is only included in @code{--enable-guile-debug} builds.")
648 #define FUNC_NAME s_scm_pt_size
649 {
650 return scm_from_int (SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash));
651 }
652 #undef FUNC_NAME
653 #endif
654
655 void
656 scm_port_non_buffer (scm_t_port *pt)
657 {
658 pt->read_pos = pt->read_buf = pt->read_end = &pt->shortbuf;
659 pt->write_buf = pt->write_pos = &pt->shortbuf;
660 pt->read_buf_size = pt->write_buf_size = 1;
661 pt->write_end = pt->write_buf + pt->write_buf_size;
662 }
663
664 \f
665 /* Revealed counts --- an oddity inherited from SCSH. */
666
667 /* Find a port in the table and return its revealed count.
668 Also used by the garbage collector.
669 */
670
671 int
672 scm_revealed_count (SCM port)
673 {
674 return SCM_REVEALED(port);
675 }
676
677
678
679 /* Return the revealed count for a port. */
680
681 SCM_DEFINE (scm_port_revealed, "port-revealed", 1, 0, 0,
682 (SCM port),
683 "Return the revealed count for @var{port}.")
684 #define FUNC_NAME s_scm_port_revealed
685 {
686 port = SCM_COERCE_OUTPORT (port);
687 SCM_VALIDATE_OPENPORT (1, port);
688 return scm_from_int (scm_revealed_count (port));
689 }
690 #undef FUNC_NAME
691
692 /* Set the revealed count for a port. */
693 SCM_DEFINE (scm_set_port_revealed_x, "set-port-revealed!", 2, 0, 0,
694 (SCM port, SCM rcount),
695 "Sets the revealed count for a port to a given value.\n"
696 "The return value is unspecified.")
697 #define FUNC_NAME s_scm_set_port_revealed_x
698 {
699 port = SCM_COERCE_OUTPORT (port);
700 SCM_VALIDATE_OPENPORT (1, port);
701 SCM_REVEALED (port) = scm_to_int (rcount);
702 return SCM_UNSPECIFIED;
703 }
704 #undef FUNC_NAME
705
706
707 \f
708 /* Retrieving a port's mode. */
709
710 /* Return the flags that characterize a port based on the mode
711 * string used to open a file for that port.
712 *
713 * See PORT FLAGS in scm.h
714 */
715
716 static long
717 scm_i_mode_bits_n (SCM modes)
718 {
719 return (SCM_OPN
720 | (scm_i_string_contains_char (modes, 'r')
721 || scm_i_string_contains_char (modes, '+') ? SCM_RDNG : 0)
722 | (scm_i_string_contains_char (modes, 'w')
723 || scm_i_string_contains_char (modes, 'a')
724 || scm_i_string_contains_char (modes, '+') ? SCM_WRTNG : 0)
725 | (scm_i_string_contains_char (modes, '0') ? SCM_BUF0 : 0)
726 | (scm_i_string_contains_char (modes, 'l') ? SCM_BUFLINE : 0));
727 }
728
729 long
730 scm_mode_bits (char *modes)
731 {
732 return scm_i_mode_bits (scm_from_locale_string (modes));
733 }
734
735 long
736 scm_i_mode_bits (SCM modes)
737 {
738 long bits;
739
740 if (!scm_is_string (modes))
741 scm_wrong_type_arg_msg (NULL, 0, modes, "string");
742
743 bits = scm_i_mode_bits_n (modes);
744 scm_remember_upto_here_1 (modes);
745 return bits;
746 }
747
748 /* Return the mode flags from an open port.
749 * Some modes such as "append" are only used when opening
750 * a file and are not returned here. */
751
752 SCM_DEFINE (scm_port_mode, "port-mode", 1, 0, 0,
753 (SCM port),
754 "Return the port modes associated with the open port @var{port}.\n"
755 "These will not necessarily be identical to the modes used when\n"
756 "the port was opened, since modes such as \"append\" which are\n"
757 "used only during port creation are not retained.")
758 #define FUNC_NAME s_scm_port_mode
759 {
760 char modes[4];
761 modes[0] = '\0';
762
763 port = SCM_COERCE_OUTPORT (port);
764 SCM_VALIDATE_OPPORT (1, port);
765 if (SCM_CELL_WORD_0 (port) & SCM_RDNG) {
766 if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
767 strcpy (modes, "r+");
768 else
769 strcpy (modes, "r");
770 }
771 else if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
772 strcpy (modes, "w");
773 if (SCM_CELL_WORD_0 (port) & SCM_BUF0)
774 strcat (modes, "0");
775 return scm_from_locale_string (modes);
776 }
777 #undef FUNC_NAME
778
779
780 \f
781 /* Closing ports. */
782
783 /* scm_close_port
784 * Call the close operation on a port object.
785 * see also scm_close.
786 */
787 SCM_DEFINE (scm_close_port, "close-port", 1, 0, 0,
788 (SCM port),
789 "Close the specified port object. Return @code{#t} if it\n"
790 "successfully closes a port or @code{#f} if it was already\n"
791 "closed. An exception may be raised if an error occurs, for\n"
792 "example when flushing buffered output. See also @ref{Ports and\n"
793 "File Descriptors, close}, for a procedure which can close file\n"
794 "descriptors.")
795 #define FUNC_NAME s_scm_close_port
796 {
797 size_t i;
798 int rv;
799
800 port = SCM_COERCE_OUTPORT (port);
801
802 SCM_VALIDATE_PORT (1, port);
803 if (SCM_CLOSEDP (port))
804 return SCM_BOOL_F;
805 i = SCM_PTOBNUM (port);
806 if (scm_ptobs[i].close)
807 rv = (scm_ptobs[i].close) (port);
808 else
809 rv = 0;
810 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
811 scm_i_remove_port (port);
812 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
813 SCM_CLR_PORT_OPEN_FLAG (port);
814 return scm_from_bool (rv >= 0);
815 }
816 #undef FUNC_NAME
817
818 SCM_DEFINE (scm_close_input_port, "close-input-port", 1, 0, 0,
819 (SCM port),
820 "Close the specified input port object. The routine has no effect if\n"
821 "the file has already been closed. An exception may be raised if an\n"
822 "error occurs. The value returned is unspecified.\n\n"
823 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
824 "which can close file descriptors.")
825 #define FUNC_NAME s_scm_close_input_port
826 {
827 SCM_VALIDATE_INPUT_PORT (1, port);
828 scm_close_port (port);
829 return SCM_UNSPECIFIED;
830 }
831 #undef FUNC_NAME
832
833 SCM_DEFINE (scm_close_output_port, "close-output-port", 1, 0, 0,
834 (SCM port),
835 "Close the specified output port object. The routine has no effect if\n"
836 "the file has already been closed. An exception may be raised if an\n"
837 "error occurs. The value returned is unspecified.\n\n"
838 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
839 "which can close file descriptors.")
840 #define FUNC_NAME s_scm_close_output_port
841 {
842 port = SCM_COERCE_OUTPORT (port);
843 SCM_VALIDATE_OUTPUT_PORT (1, port);
844 scm_close_port (port);
845 return SCM_UNSPECIFIED;
846 }
847 #undef FUNC_NAME
848
849 static SCM
850 scm_i_collect_keys_in_vector (void *closure, SCM key, SCM value, SCM result)
851 {
852 int *i = (int*) closure;
853 scm_c_vector_set_x (result, *i, key);
854 (*i)++;
855
856 return result;
857 }
858
859 void
860 scm_c_port_for_each (void (*proc)(void *data, SCM p), void *data)
861 {
862 int i = 0;
863 size_t n;
864 SCM ports;
865
866 /* Even without pre-emptive multithreading, running arbitrary code
867 while scanning the port table is unsafe because the port table
868 can change arbitrarily (from a GC, for example). So we first
869 collect the ports into a vector. -mvo */
870
871 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
872 n = SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash);
873 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
874 ports = scm_c_make_vector (n, SCM_BOOL_F);
875
876 scm_i_pthread_mutex_lock (&scm_i_port_table_mutex);
877 ports = scm_internal_hash_fold (scm_i_collect_keys_in_vector, &i,
878 ports, scm_i_port_weak_hash);
879 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
880
881 for (i = 0; i < n; i++) {
882 SCM p = SCM_SIMPLE_VECTOR_REF (ports, i);
883 if (SCM_PORTP (p))
884 proc (data, p);
885 }
886
887 scm_remember_upto_here_1 (ports);
888 }
889
890 SCM_DEFINE (scm_port_for_each, "port-for-each", 1, 0, 0,
891 (SCM proc),
892 "Apply @var{proc} to each port in the Guile port table\n"
893 "in turn. The return value is unspecified. More specifically,\n"
894 "@var{proc} is applied exactly once to every port that exists\n"
895 "in the system at the time @var{port-for-each} is invoked.\n"
896 "Changes to the port table while @var{port-for-each} is running\n"
897 "have no effect as far as @var{port-for-each} is concerned.")
898 #define FUNC_NAME s_scm_port_for_each
899 {
900 SCM_VALIDATE_PROC (1, proc);
901
902 scm_c_port_for_each ((void (*)(void*,SCM))scm_call_1, proc);
903 return SCM_UNSPECIFIED;
904 }
905 #undef FUNC_NAME
906
907
908 \f
909 /* Utter miscellany. Gosh, we should clean this up some time. */
910
911 SCM_DEFINE (scm_input_port_p, "input-port?", 1, 0, 0,
912 (SCM x),
913 "Return @code{#t} if @var{x} is an input port, otherwise return\n"
914 "@code{#f}. Any object satisfying this predicate also satisfies\n"
915 "@code{port?}.")
916 #define FUNC_NAME s_scm_input_port_p
917 {
918 return scm_from_bool (SCM_INPUT_PORT_P (x));
919 }
920 #undef FUNC_NAME
921
922 SCM_DEFINE (scm_output_port_p, "output-port?", 1, 0, 0,
923 (SCM x),
924 "Return @code{#t} if @var{x} is an output port, otherwise return\n"
925 "@code{#f}. Any object satisfying this predicate also satisfies\n"
926 "@code{port?}.")
927 #define FUNC_NAME s_scm_output_port_p
928 {
929 x = SCM_COERCE_OUTPORT (x);
930 return scm_from_bool (SCM_OUTPUT_PORT_P (x));
931 }
932 #undef FUNC_NAME
933
934 SCM_DEFINE (scm_port_p, "port?", 1, 0, 0,
935 (SCM x),
936 "Return a boolean indicating whether @var{x} is a port.\n"
937 "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n"
938 "@var{x}))}.")
939 #define FUNC_NAME s_scm_port_p
940 {
941 return scm_from_bool (SCM_PORTP (x));
942 }
943 #undef FUNC_NAME
944
945 SCM_DEFINE (scm_port_closed_p, "port-closed?", 1, 0, 0,
946 (SCM port),
947 "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n"
948 "open.")
949 #define FUNC_NAME s_scm_port_closed_p
950 {
951 SCM_VALIDATE_PORT (1, port);
952 return scm_from_bool (!SCM_OPPORTP (port));
953 }
954 #undef FUNC_NAME
955
956 SCM_DEFINE (scm_eof_object_p, "eof-object?", 1, 0, 0,
957 (SCM x),
958 "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n"
959 "return @code{#f}.")
960 #define FUNC_NAME s_scm_eof_object_p
961 {
962 return scm_from_bool(SCM_EOF_OBJECT_P (x));
963 }
964 #undef FUNC_NAME
965
966 SCM_DEFINE (scm_force_output, "force-output", 0, 1, 0,
967 (SCM port),
968 "Flush the specified output port, or the current output port if @var{port}\n"
969 "is omitted. The current output buffer contents are passed to the\n"
970 "underlying port implementation (e.g., in the case of fports, the\n"
971 "data will be written to the file and the output buffer will be cleared.)\n"
972 "It has no effect on an unbuffered port.\n\n"
973 "The return value is unspecified.")
974 #define FUNC_NAME s_scm_force_output
975 {
976 if (SCM_UNBNDP (port))
977 port = scm_current_output_port ();
978 else
979 {
980 port = SCM_COERCE_OUTPORT (port);
981 SCM_VALIDATE_OPOUTPORT (1, port);
982 }
983 scm_flush (port);
984 return SCM_UNSPECIFIED;
985 }
986 #undef FUNC_NAME
987
988
989 static void
990 flush_output_port (void *closure, SCM port)
991 {
992 if (SCM_OPOUTPORTP (port))
993 scm_flush (port);
994 }
995
996 SCM_DEFINE (scm_flush_all_ports, "flush-all-ports", 0, 0, 0,
997 (),
998 "Equivalent to calling @code{force-output} on\n"
999 "all open output ports. The return value is unspecified.")
1000 #define FUNC_NAME s_scm_flush_all_ports
1001 {
1002 scm_c_port_for_each (&flush_output_port, NULL);
1003 return SCM_UNSPECIFIED;
1004 }
1005 #undef FUNC_NAME
1006
1007 SCM_DEFINE (scm_read_char, "read-char", 0, 1, 0,
1008 (SCM port),
1009 "Return the next character available from @var{port}, updating\n"
1010 "@var{port} to point to the following character. If no more\n"
1011 "characters are available, the end-of-file object is returned.")
1012 #define FUNC_NAME s_scm_read_char
1013 {
1014 scm_t_wchar c;
1015 if (SCM_UNBNDP (port))
1016 port = scm_current_input_port ();
1017 SCM_VALIDATE_OPINPORT (1, port);
1018 c = scm_getc (port);
1019 if (EOF == c)
1020 return SCM_EOF_VAL;
1021 return SCM_MAKE_CHAR (c);
1022 }
1023 #undef FUNC_NAME
1024
1025 #define SCM_MBCHAR_BUF_SIZE (4)
1026
1027 /* Get one codepoint from a file, using the port's encoding. */
1028 scm_t_wchar
1029 scm_getc (SCM port)
1030 {
1031 int c;
1032 unsigned int bufcount = 0;
1033 char buf[SCM_MBCHAR_BUF_SIZE];
1034 scm_t_wchar codepoint = 0;
1035 scm_t_uint32 *u32;
1036 size_t u32len;
1037 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1038
1039 c = scm_get_byte_or_eof (port);
1040 if (c == EOF)
1041 return (scm_t_wchar) EOF;
1042
1043 buf[0] = c;
1044 bufcount++;
1045
1046 if (pt->encoding == NULL)
1047 {
1048 /* The encoding is Latin-1: bytes are characters. */
1049 codepoint = (unsigned char) buf[0];
1050 goto success;
1051 }
1052
1053 for (;;)
1054 {
1055 u32 = u32_conv_from_encoding (pt->encoding,
1056 (enum iconv_ilseq_handler) pt->ilseq_handler,
1057 buf, bufcount, NULL, NULL, &u32len);
1058 if (u32 == NULL || u32len == 0)
1059 {
1060 if (errno == ENOMEM)
1061 scm_memory_error ("Input decoding");
1062
1063 /* Otherwise errno is EILSEQ or EINVAL, so perhaps more
1064 bytes are needed. Keep looping. */
1065 }
1066 else
1067 {
1068 /* Complete codepoint found. */
1069 codepoint = u32[0];
1070 free (u32);
1071 goto success;
1072 }
1073
1074 if (bufcount == SCM_MBCHAR_BUF_SIZE)
1075 {
1076 /* We've read several bytes and didn't find a good
1077 codepoint. Give up. */
1078 goto failure;
1079 }
1080
1081 c = scm_get_byte_or_eof (port);
1082
1083 if (c == EOF)
1084 {
1085 /* EOF before a complete character was read. Push it all
1086 back and return EOF. */
1087 while (bufcount > 0)
1088 {
1089 /* FIXME: this will probably cause errors in the port column. */
1090 scm_unget_byte (buf[bufcount-1], port);
1091 bufcount --;
1092 }
1093 return EOF;
1094 }
1095
1096 if (c == '\n')
1097 {
1098 /* It is always invalid to have EOL in the middle of a
1099 multibyte character. */
1100 scm_unget_byte ('\n', port);
1101 goto failure;
1102 }
1103
1104 buf[bufcount++] = c;
1105 }
1106
1107 success:
1108 switch (codepoint)
1109 {
1110 case '\a':
1111 break;
1112 case '\b':
1113 SCM_DECCOL (port);
1114 break;
1115 case '\n':
1116 SCM_INCLINE (port);
1117 break;
1118 case '\r':
1119 SCM_ZEROCOL (port);
1120 break;
1121 case '\t':
1122 SCM_TABCOL (port);
1123 break;
1124 default:
1125 SCM_INCCOL (port);
1126 break;
1127 }
1128
1129 return codepoint;
1130
1131 failure:
1132 {
1133 char *err_buf;
1134 SCM err_str = scm_i_make_string (bufcount, &err_buf);
1135 memcpy (err_buf, buf, bufcount);
1136
1137 if (errno == EILSEQ)
1138 scm_misc_error (NULL, "input encoding error for ~s: ~s",
1139 scm_list_2 (scm_from_locale_string (scm_i_get_port_encoding (port)),
1140 err_str));
1141 else
1142 scm_misc_error (NULL, "input encoding error (invalid) for ~s: ~s\n",
1143 scm_list_2 (scm_from_locale_string (scm_i_get_port_encoding (port)),
1144 err_str));
1145 }
1146
1147 /* Never gets here. */
1148 return 0;
1149 }
1150
1151
1152 /* this should only be called when the read buffer is empty. it
1153 tries to refill the read buffer. it returns the first char from
1154 the port, which is either EOF or *(pt->read_pos). */
1155 int
1156 scm_fill_input (SCM port)
1157 {
1158 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1159
1160 assert (pt->read_pos == pt->read_end);
1161
1162 if (pt->read_buf == pt->putback_buf)
1163 {
1164 /* finished reading put-back chars. */
1165 pt->read_buf = pt->saved_read_buf;
1166 pt->read_pos = pt->saved_read_pos;
1167 pt->read_end = pt->saved_read_end;
1168 pt->read_buf_size = pt->saved_read_buf_size;
1169 if (pt->read_pos < pt->read_end)
1170 return *(pt->read_pos);
1171 }
1172 return scm_ptobs[SCM_PTOBNUM (port)].fill_input (port);
1173 }
1174
1175
1176 /* scm_lfwrite
1177 *
1178 * This function differs from scm_c_write; it updates port line and
1179 * column. */
1180
1181 static void
1182 update_port_lf (scm_t_wchar c, SCM port)
1183 {
1184 if (c == '\a')
1185 ; /* Do nothing. */
1186 else if (c == '\b')
1187 SCM_DECCOL (port);
1188 else if (c == '\n')
1189 SCM_INCLINE (port);
1190 else if (c == '\r')
1191 SCM_ZEROCOL (port);
1192 else if (c == '\t')
1193 SCM_TABCOL (port);
1194 else
1195 SCM_INCCOL (port);
1196 }
1197
1198 void
1199 scm_lfwrite (const char *ptr, size_t size, SCM port)
1200 {
1201 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1202 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1203
1204 if (pt->rw_active == SCM_PORT_READ)
1205 scm_end_input (port);
1206
1207 ptob->write (port, ptr, size);
1208
1209 for (; size; ptr++, size--)
1210 update_port_lf ((scm_t_wchar) (unsigned char) *ptr, port);
1211
1212 if (pt->rw_random)
1213 pt->rw_active = SCM_PORT_WRITE;
1214 }
1215
1216 /* Write a scheme string STR to PORT from START inclusive to END
1217 exclusive. */
1218 void
1219 scm_lfwrite_substr (SCM str, size_t start, size_t end, SCM port)
1220 {
1221 size_t i, size = scm_i_string_length (str);
1222 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1223 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1224 scm_t_wchar p;
1225 char *buf;
1226 size_t len;
1227
1228 if (pt->rw_active == SCM_PORT_READ)
1229 scm_end_input (port);
1230
1231 if (end == (size_t) (-1))
1232 end = size;
1233 size = end - start;
1234
1235 /* Note that making a substring will likely take the
1236 stringbuf_write_mutex. So, one shouldn't use scm_lfwrite_substr
1237 if the stringbuf write mutex may still be held elsewhere. */
1238 buf = scm_to_stringn (scm_c_substring (str, start, end), &len,
1239 pt->encoding, pt->ilseq_handler);
1240 ptob->write (port, buf, len);
1241 free (buf);
1242
1243 for (i = 0; i < size; i++)
1244 {
1245 p = scm_i_string_ref (str, i + start);
1246 update_port_lf (p, port);
1247 }
1248
1249 if (pt->rw_random)
1250 pt->rw_active = SCM_PORT_WRITE;
1251 }
1252
1253 /* Write a scheme string STR to PORT. */
1254 void
1255 scm_lfwrite_str (SCM str, SCM port)
1256 {
1257 size_t i, size = scm_i_string_length (str);
1258 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1259 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1260 scm_t_wchar p;
1261 char *buf;
1262 size_t len;
1263
1264 if (pt->rw_active == SCM_PORT_READ)
1265 scm_end_input (port);
1266
1267 buf = scm_to_stringn (str, &len,
1268 pt->encoding, pt->ilseq_handler);
1269 ptob->write (port, buf, len);
1270 free (buf);
1271
1272 for (i = 0; i < size; i++)
1273 {
1274 p = scm_i_string_ref (str, i);
1275 update_port_lf (p, port);
1276 }
1277
1278 if (pt->rw_random)
1279 pt->rw_active = SCM_PORT_WRITE;
1280 }
1281
1282 /* scm_c_read
1283 *
1284 * Used by an application to read arbitrary number of bytes from an
1285 * SCM port. Same semantics as libc read, except that scm_c_read only
1286 * returns less than SIZE bytes if at end-of-file.
1287 *
1288 * Warning: Doesn't update port line and column counts! */
1289
1290 /* This structure, and the following swap_buffer function, are used
1291 for temporarily swapping a port's own read buffer, and the buffer
1292 that the caller of scm_c_read provides. */
1293 struct port_and_swap_buffer
1294 {
1295 scm_t_port *pt;
1296 unsigned char *buffer;
1297 size_t size;
1298 };
1299
1300 static void
1301 swap_buffer (void *data)
1302 {
1303 struct port_and_swap_buffer *psb = (struct port_and_swap_buffer *) data;
1304 unsigned char *old_buf = psb->pt->read_buf;
1305 size_t old_size = psb->pt->read_buf_size;
1306
1307 /* Make the port use (buffer, size) from the struct. */
1308 psb->pt->read_pos = psb->pt->read_buf = psb->pt->read_end = psb->buffer;
1309 psb->pt->read_buf_size = psb->size;
1310
1311 /* Save the port's old (buffer, size) in the struct. */
1312 psb->buffer = old_buf;
1313 psb->size = old_size;
1314 }
1315
1316 size_t
1317 scm_c_read (SCM port, void *buffer, size_t size)
1318 #define FUNC_NAME "scm_c_read"
1319 {
1320 scm_t_port *pt;
1321 size_t n_read = 0, n_available;
1322 struct port_and_swap_buffer psb;
1323
1324 SCM_VALIDATE_OPINPORT (1, port);
1325
1326 pt = SCM_PTAB_ENTRY (port);
1327 if (pt->rw_active == SCM_PORT_WRITE)
1328 scm_ptobs[SCM_PTOBNUM (port)].flush (port);
1329
1330 if (pt->rw_random)
1331 pt->rw_active = SCM_PORT_READ;
1332
1333 /* Take bytes first from the port's read buffer. */
1334 if (pt->read_pos < pt->read_end)
1335 {
1336 n_available = min (size, pt->read_end - pt->read_pos);
1337 memcpy (buffer, pt->read_pos, n_available);
1338 buffer = (char *) buffer + n_available;
1339 pt->read_pos += n_available;
1340 n_read += n_available;
1341 size -= n_available;
1342 }
1343
1344 /* Avoid the scm_dynwind_* costs if we now have enough data. */
1345 if (size == 0)
1346 return n_read;
1347
1348 /* Now we will call scm_fill_input repeatedly until we have read the
1349 requested number of bytes. (Note that a single scm_fill_input
1350 call does not guarantee to fill the whole of the port's read
1351 buffer.) */
1352 if (pt->read_buf_size <= 1 && pt->encoding == NULL)
1353 {
1354 /* The port that we are reading from is unbuffered - i.e. does
1355 not have its own persistent buffer - but we have a buffer,
1356 provided by our caller, that is the right size for the data
1357 that is wanted. For the following scm_fill_input calls,
1358 therefore, we use the buffer in hand as the port's read
1359 buffer.
1360
1361 We need to make sure that the port's normal (1 byte) buffer
1362 is reinstated in case one of the scm_fill_input () calls
1363 throws an exception; we use the scm_dynwind_* API to achieve
1364 that.
1365
1366 A consequence of this optimization is that the fill_input
1367 functions can't unget characters. That'll push data to the
1368 pushback buffer instead of this psb buffer. */
1369 #if SCM_DEBUG == 1
1370 unsigned char *pback = pt->putback_buf;
1371 #endif
1372 psb.pt = pt;
1373 psb.buffer = buffer;
1374 psb.size = size;
1375 scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE);
1376 scm_dynwind_rewind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY);
1377 scm_dynwind_unwind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY);
1378
1379 /* Call scm_fill_input until we have all the bytes that we need,
1380 or we hit EOF. */
1381 while (pt->read_buf_size && (scm_fill_input (port) != EOF))
1382 {
1383 pt->read_buf_size -= (pt->read_end - pt->read_pos);
1384 pt->read_pos = pt->read_buf = pt->read_end;
1385 }
1386 #if SCM_DEBUG == 1
1387 if (pback != pt->putback_buf
1388 || pt->read_buf - (unsigned char *) buffer < 0)
1389 scm_misc_error (FUNC_NAME,
1390 "scm_c_read must not call a fill function that pushes "
1391 "back characters onto an unbuffered port", SCM_EOL);
1392 #endif
1393 n_read += pt->read_buf - (unsigned char *) buffer;
1394
1395 /* Reinstate the port's normal buffer. */
1396 scm_dynwind_end ();
1397 }
1398 else
1399 {
1400 /* The port has its own buffer. It is important that we use it,
1401 even if it happens to be smaller than our caller's buffer, so
1402 that a custom port implementation's entry points (in
1403 particular, fill_input) can rely on the buffer always being
1404 the same as they first set up. */
1405 while (size && (scm_fill_input (port) != EOF))
1406 {
1407 n_available = min (size, pt->read_end - pt->read_pos);
1408 memcpy (buffer, pt->read_pos, n_available);
1409 buffer = (char *) buffer + n_available;
1410 pt->read_pos += n_available;
1411 n_read += n_available;
1412 size -= n_available;
1413 }
1414 }
1415
1416 return n_read;
1417 }
1418 #undef FUNC_NAME
1419
1420 /* scm_c_write
1421 *
1422 * Used by an application to write arbitrary number of bytes to an SCM
1423 * port. Similar semantics as libc write. However, unlike libc
1424 * write, scm_c_write writes the requested number of bytes and has no
1425 * return value.
1426 *
1427 * Warning: Doesn't update port line and column counts!
1428 */
1429
1430 void
1431 scm_c_write (SCM port, const void *ptr, size_t size)
1432 #define FUNC_NAME "scm_c_write"
1433 {
1434 scm_t_port *pt;
1435 scm_t_ptob_descriptor *ptob;
1436
1437 SCM_VALIDATE_OPOUTPORT (1, port);
1438
1439 pt = SCM_PTAB_ENTRY (port);
1440 ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1441
1442 if (pt->rw_active == SCM_PORT_READ)
1443 scm_end_input (port);
1444
1445 ptob->write (port, ptr, size);
1446
1447 if (pt->rw_random)
1448 pt->rw_active = SCM_PORT_WRITE;
1449 }
1450 #undef FUNC_NAME
1451
1452 void
1453 scm_flush (SCM port)
1454 {
1455 long i = SCM_PTOBNUM (port);
1456 assert (i >= 0);
1457 (scm_ptobs[i].flush) (port);
1458 }
1459
1460 void
1461 scm_end_input (SCM port)
1462 {
1463 long offset;
1464 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1465
1466 if (pt->read_buf == pt->putback_buf)
1467 {
1468 offset = pt->read_end - pt->read_pos;
1469 pt->read_buf = pt->saved_read_buf;
1470 pt->read_pos = pt->saved_read_pos;
1471 pt->read_end = pt->saved_read_end;
1472 pt->read_buf_size = pt->saved_read_buf_size;
1473 }
1474 else
1475 offset = 0;
1476
1477 scm_ptobs[SCM_PTOBNUM (port)].end_input (port, offset);
1478 }
1479
1480 \f
1481
1482
1483 void
1484 scm_unget_byte (int c, SCM port)
1485 #define FUNC_NAME "scm_unget_byte"
1486 {
1487 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1488
1489 if (pt->read_buf == pt->putback_buf)
1490 /* already using the put-back buffer. */
1491 {
1492 /* enlarge putback_buf if necessary. */
1493 if (pt->read_end == pt->read_buf + pt->read_buf_size
1494 && pt->read_buf == pt->read_pos)
1495 {
1496 size_t new_size = pt->read_buf_size * 2;
1497 unsigned char *tmp = (unsigned char *)
1498 scm_gc_realloc (pt->putback_buf, pt->read_buf_size, new_size,
1499 "putback buffer");
1500
1501 pt->read_pos = pt->read_buf = pt->putback_buf = tmp;
1502 pt->read_end = pt->read_buf + pt->read_buf_size;
1503 pt->read_buf_size = pt->putback_buf_size = new_size;
1504 }
1505
1506 /* shift any existing bytes to buffer + 1. */
1507 if (pt->read_pos == pt->read_end)
1508 pt->read_end = pt->read_buf + 1;
1509 else if (pt->read_pos != pt->read_buf + 1)
1510 {
1511 int count = pt->read_end - pt->read_pos;
1512
1513 memmove (pt->read_buf + 1, pt->read_pos, count);
1514 pt->read_end = pt->read_buf + 1 + count;
1515 }
1516
1517 pt->read_pos = pt->read_buf;
1518 }
1519 else
1520 /* switch to the put-back buffer. */
1521 {
1522 if (pt->putback_buf == NULL)
1523 {
1524 pt->putback_buf
1525 = (unsigned char *) scm_gc_malloc_pointerless
1526 (SCM_INITIAL_PUTBACK_BUF_SIZE, "putback buffer");
1527 pt->putback_buf_size = SCM_INITIAL_PUTBACK_BUF_SIZE;
1528 }
1529
1530 pt->saved_read_buf = pt->read_buf;
1531 pt->saved_read_pos = pt->read_pos;
1532 pt->saved_read_end = pt->read_end;
1533 pt->saved_read_buf_size = pt->read_buf_size;
1534
1535 pt->read_pos = pt->read_buf = pt->putback_buf;
1536 pt->read_end = pt->read_buf + 1;
1537 pt->read_buf_size = pt->putback_buf_size;
1538 }
1539
1540 *pt->read_buf = c;
1541
1542 if (pt->rw_random)
1543 pt->rw_active = SCM_PORT_READ;
1544 }
1545 #undef FUNC_NAME
1546
1547 void
1548 scm_ungetc (scm_t_wchar c, SCM port)
1549 #define FUNC_NAME "scm_ungetc"
1550 {
1551 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1552 scm_t_wchar *wbuf;
1553 SCM str = scm_i_make_wide_string (1, &wbuf);
1554 char *buf;
1555 size_t len;
1556 int i;
1557
1558 wbuf[0] = c;
1559 buf = scm_to_stringn (str, &len, pt->encoding, pt->ilseq_handler);
1560
1561 for (i = len - 1; i >= 0; i--)
1562 scm_unget_byte (buf[i], port);
1563
1564 if (c == '\n')
1565 {
1566 /* What should col be in this case?
1567 * We'll leave it at -1.
1568 */
1569 SCM_LINUM (port) -= 1;
1570 }
1571 else
1572 SCM_COL(port) -= 1;
1573 }
1574 #undef FUNC_NAME
1575
1576
1577 void
1578 scm_ungets (const char *s, int n, SCM port)
1579 {
1580 /* This is simple minded and inefficient, but unreading strings is
1581 * probably not a common operation, and remember that line and
1582 * column numbers have to be handled...
1583 *
1584 * Please feel free to write an optimized version!
1585 */
1586 while (n--)
1587 scm_ungetc (s[n], port);
1588 }
1589
1590
1591 SCM_DEFINE (scm_peek_char, "peek-char", 0, 1, 0,
1592 (SCM port),
1593 "Return the next character available from @var{port},\n"
1594 "@emph{without} updating @var{port} to point to the following\n"
1595 "character. If no more characters are available, the\n"
1596 "end-of-file object is returned.\n"
1597 "\n"
1598 "The value returned by\n"
1599 "a call to @code{peek-char} is the same as the value that would\n"
1600 "have been returned by a call to @code{read-char} on the same\n"
1601 "port. The only difference is that the very next call to\n"
1602 "@code{read-char} or @code{peek-char} on that @var{port} will\n"
1603 "return the value returned by the preceding call to\n"
1604 "@code{peek-char}. In particular, a call to @code{peek-char} on\n"
1605 "an interactive port will hang waiting for input whenever a call\n"
1606 "to @code{read-char} would have hung.")
1607 #define FUNC_NAME s_scm_peek_char
1608 {
1609 scm_t_wchar c, column;
1610 if (SCM_UNBNDP (port))
1611 port = scm_current_input_port ();
1612 else
1613 SCM_VALIDATE_OPINPORT (1, port);
1614 column = SCM_COL(port);
1615 c = scm_getc (port);
1616 if (EOF == c)
1617 return SCM_EOF_VAL;
1618 scm_ungetc (c, port);
1619 SCM_COL(port) = column;
1620 return SCM_MAKE_CHAR (c);
1621 }
1622 #undef FUNC_NAME
1623
1624 SCM_DEFINE (scm_unread_char, "unread-char", 1, 1, 0,
1625 (SCM cobj, SCM port),
1626 "Place @var{char} in @var{port} so that it will be read by the\n"
1627 "next read operation. If called multiple times, the unread characters\n"
1628 "will be read again in last-in first-out order. If @var{port} is\n"
1629 "not supplied, the current input port is used.")
1630 #define FUNC_NAME s_scm_unread_char
1631 {
1632 int c;
1633
1634 SCM_VALIDATE_CHAR (1, cobj);
1635 if (SCM_UNBNDP (port))
1636 port = scm_current_input_port ();
1637 else
1638 SCM_VALIDATE_OPINPORT (2, port);
1639
1640 c = SCM_CHAR (cobj);
1641
1642 scm_ungetc (c, port);
1643 return cobj;
1644 }
1645 #undef FUNC_NAME
1646
1647 SCM_DEFINE (scm_unread_string, "unread-string", 2, 0, 0,
1648 (SCM str, SCM port),
1649 "Place the string @var{str} in @var{port} so that its characters will be\n"
1650 "read in subsequent read operations. If called multiple times, the\n"
1651 "unread characters will be read again in last-in first-out order. If\n"
1652 "@var{port} is not supplied, the current-input-port is used.")
1653 #define FUNC_NAME s_scm_unread_string
1654 {
1655 int n;
1656 SCM_VALIDATE_STRING (1, str);
1657 if (SCM_UNBNDP (port))
1658 port = scm_current_input_port ();
1659 else
1660 SCM_VALIDATE_OPINPORT (2, port);
1661
1662 n = scm_i_string_length (str);
1663
1664 while (n--)
1665 scm_ungetc (scm_i_string_ref (str, n), port);
1666
1667 return str;
1668 }
1669 #undef FUNC_NAME
1670
1671 SCM_DEFINE (scm_seek, "seek", 3, 0, 0,
1672 (SCM fd_port, SCM offset, SCM whence),
1673 "Sets the current position of @var{fd/port} to the integer\n"
1674 "@var{offset}, which is interpreted according to the value of\n"
1675 "@var{whence}.\n"
1676 "\n"
1677 "One of the following variables should be supplied for\n"
1678 "@var{whence}:\n"
1679 "@defvar SEEK_SET\n"
1680 "Seek from the beginning of the file.\n"
1681 "@end defvar\n"
1682 "@defvar SEEK_CUR\n"
1683 "Seek from the current position.\n"
1684 "@end defvar\n"
1685 "@defvar SEEK_END\n"
1686 "Seek from the end of the file.\n"
1687 "@end defvar\n"
1688 "If @var{fd/port} is a file descriptor, the underlying system\n"
1689 "call is @code{lseek}. @var{port} may be a string port.\n"
1690 "\n"
1691 "The value returned is the new position in the file. This means\n"
1692 "that the current position of a port can be obtained using:\n"
1693 "@lisp\n"
1694 "(seek port 0 SEEK_CUR)\n"
1695 "@end lisp")
1696 #define FUNC_NAME s_scm_seek
1697 {
1698 int how;
1699
1700 fd_port = SCM_COERCE_OUTPORT (fd_port);
1701
1702 how = scm_to_int (whence);
1703 if (how != SEEK_SET && how != SEEK_CUR && how != SEEK_END)
1704 SCM_OUT_OF_RANGE (3, whence);
1705
1706 if (SCM_OPPORTP (fd_port))
1707 {
1708 scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (fd_port);
1709 off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
1710 off_t_or_off64_t rv;
1711
1712 if (!ptob->seek)
1713 SCM_MISC_ERROR ("port is not seekable",
1714 scm_cons (fd_port, SCM_EOL));
1715 else
1716 rv = ptob->seek (fd_port, off, how);
1717 return scm_from_off_t_or_off64_t (rv);
1718 }
1719 else /* file descriptor?. */
1720 {
1721 off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
1722 off_t_or_off64_t rv;
1723 rv = lseek_or_lseek64 (scm_to_int (fd_port), off, how);
1724 if (rv == -1)
1725 SCM_SYSERROR;
1726 return scm_from_off_t_or_off64_t (rv);
1727 }
1728 }
1729 #undef FUNC_NAME
1730
1731 #ifndef O_BINARY
1732 #define O_BINARY 0
1733 #endif
1734
1735 /* Mingw has ftruncate(), perhaps implemented above using chsize, but
1736 doesn't have the filename version truncate(), hence this code. */
1737 #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE
1738 static int
1739 truncate (const char *file, off_t length)
1740 {
1741 int ret, fdes;
1742
1743 fdes = open (file, O_BINARY | O_WRONLY);
1744 if (fdes == -1)
1745 return -1;
1746
1747 ret = ftruncate (fdes, length);
1748 if (ret == -1)
1749 {
1750 int save_errno = errno;
1751 close (fdes);
1752 errno = save_errno;
1753 return -1;
1754 }
1755
1756 return close (fdes);
1757 }
1758 #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */
1759
1760 SCM_DEFINE (scm_truncate_file, "truncate-file", 1, 1, 0,
1761 (SCM object, SCM length),
1762 "Truncate @var{file} to @var{length} bytes. @var{file} can be a\n"
1763 "filename string, a port object, or an integer file descriptor.\n"
1764 "The return value is unspecified.\n"
1765 "\n"
1766 "For a port or file descriptor @var{length} can be omitted, in\n"
1767 "which case the file is truncated at the current position (per\n"
1768 "@code{ftell} above).\n"
1769 "\n"
1770 "On most systems a file can be extended by giving a length\n"
1771 "greater than the current size, but this is not mandatory in the\n"
1772 "POSIX standard.")
1773 #define FUNC_NAME s_scm_truncate_file
1774 {
1775 int rv;
1776
1777 /* "object" can be a port, fdes or filename.
1778
1779 Negative "length" makes no sense, but it's left to truncate() or
1780 ftruncate() to give back an error for that (normally EINVAL).
1781 */
1782
1783 if (SCM_UNBNDP (length))
1784 {
1785 /* must supply length if object is a filename. */
1786 if (scm_is_string (object))
1787 SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL);
1788
1789 length = scm_seek (object, SCM_INUM0, scm_from_int (SEEK_CUR));
1790 }
1791
1792 object = SCM_COERCE_OUTPORT (object);
1793 if (scm_is_integer (object))
1794 {
1795 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1796 SCM_SYSCALL (rv = ftruncate_or_ftruncate64 (scm_to_int (object),
1797 c_length));
1798 }
1799 else if (SCM_OPOUTPORTP (object))
1800 {
1801 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1802 scm_t_port *pt = SCM_PTAB_ENTRY (object);
1803 scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (object);
1804
1805 if (!ptob->truncate)
1806 SCM_MISC_ERROR ("port is not truncatable", SCM_EOL);
1807 if (pt->rw_active == SCM_PORT_READ)
1808 scm_end_input (object);
1809 else if (pt->rw_active == SCM_PORT_WRITE)
1810 ptob->flush (object);
1811
1812 ptob->truncate (object, c_length);
1813 rv = 0;
1814 }
1815 else
1816 {
1817 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1818 char *str = scm_to_locale_string (object);
1819 int eno;
1820 SCM_SYSCALL (rv = truncate_or_truncate64 (str, c_length));
1821 eno = errno;
1822 free (str);
1823 errno = eno;
1824 }
1825 if (rv == -1)
1826 SCM_SYSERROR;
1827 return SCM_UNSPECIFIED;
1828 }
1829 #undef FUNC_NAME
1830
1831 SCM_DEFINE (scm_port_line, "port-line", 1, 0, 0,
1832 (SCM port),
1833 "Return the current line number for @var{port}.\n"
1834 "\n"
1835 "The first line of a file is 0. But you might want to add 1\n"
1836 "when printing line numbers, since starting from 1 is\n"
1837 "traditional in error messages, and likely to be more natural to\n"
1838 "non-programmers.")
1839 #define FUNC_NAME s_scm_port_line
1840 {
1841 port = SCM_COERCE_OUTPORT (port);
1842 SCM_VALIDATE_OPENPORT (1, port);
1843 return scm_from_long (SCM_LINUM (port));
1844 }
1845 #undef FUNC_NAME
1846
1847 SCM_DEFINE (scm_set_port_line_x, "set-port-line!", 2, 0, 0,
1848 (SCM port, SCM line),
1849 "Set the current line number for @var{port} to @var{line}. The\n"
1850 "first line of a file is 0.")
1851 #define FUNC_NAME s_scm_set_port_line_x
1852 {
1853 port = SCM_COERCE_OUTPORT (port);
1854 SCM_VALIDATE_OPENPORT (1, port);
1855 SCM_PTAB_ENTRY (port)->line_number = scm_to_long (line);
1856 return SCM_UNSPECIFIED;
1857 }
1858 #undef FUNC_NAME
1859
1860 SCM_DEFINE (scm_port_column, "port-column", 1, 0, 0,
1861 (SCM port),
1862 "Return the current column number of @var{port}.\n"
1863 "If the number is\n"
1864 "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n"
1865 "- i.e. the first character of the first line is line 0, column 0.\n"
1866 "(However, when you display a file position, for example in an error\n"
1867 "message, we recommend you add 1 to get 1-origin integers. This is\n"
1868 "because lines and column numbers traditionally start with 1, and that is\n"
1869 "what non-programmers will find most natural.)")
1870 #define FUNC_NAME s_scm_port_column
1871 {
1872 port = SCM_COERCE_OUTPORT (port);
1873 SCM_VALIDATE_OPENPORT (1, port);
1874 return scm_from_int (SCM_COL (port));
1875 }
1876 #undef FUNC_NAME
1877
1878 SCM_DEFINE (scm_set_port_column_x, "set-port-column!", 2, 0, 0,
1879 (SCM port, SCM column),
1880 "Set the current column of @var{port}. Before reading the first\n"
1881 "character on a line the column should be 0.")
1882 #define FUNC_NAME s_scm_set_port_column_x
1883 {
1884 port = SCM_COERCE_OUTPORT (port);
1885 SCM_VALIDATE_OPENPORT (1, port);
1886 SCM_PTAB_ENTRY (port)->column_number = scm_to_int (column);
1887 return SCM_UNSPECIFIED;
1888 }
1889 #undef FUNC_NAME
1890
1891 SCM_DEFINE (scm_port_filename, "port-filename", 1, 0, 0,
1892 (SCM port),
1893 "Return the filename associated with @var{port}. This function returns\n"
1894 "the strings \"standard input\", \"standard output\" and \"standard error\"\n"
1895 "when called on the current input, output and error ports respectively.")
1896 #define FUNC_NAME s_scm_port_filename
1897 {
1898 port = SCM_COERCE_OUTPORT (port);
1899 SCM_VALIDATE_OPENPORT (1, port);
1900 return SCM_FILENAME (port);
1901 }
1902 #undef FUNC_NAME
1903
1904 SCM_DEFINE (scm_set_port_filename_x, "set-port-filename!", 2, 0, 0,
1905 (SCM port, SCM filename),
1906 "Change the filename associated with @var{port}, using the current input\n"
1907 "port if none is specified. Note that this does not change the port's\n"
1908 "source of data, but only the value that is returned by\n"
1909 "@code{port-filename} and reported in diagnostic output.")
1910 #define FUNC_NAME s_scm_set_port_filename_x
1911 {
1912 port = SCM_COERCE_OUTPORT (port);
1913 SCM_VALIDATE_OPENPORT (1, port);
1914 /* We allow the user to set the filename to whatever he likes. */
1915 SCM_SET_FILENAME (port, filename);
1916 return SCM_UNSPECIFIED;
1917 }
1918 #undef FUNC_NAME
1919
1920 /* The default port encoding for this locale. New ports will have this
1921 encoding. If it is a string, that is the encoding. If it #f, it
1922 is in the native (Latin-1) encoding. */
1923 SCM_GLOBAL_VARIABLE (scm_port_encoding_var, "%port-encoding");
1924 static int scm_port_encoding_init = 0;
1925
1926 /* Return a C string representation of the current encoding. */
1927 const char *
1928 scm_i_get_port_encoding (SCM port)
1929 {
1930 SCM encoding;
1931
1932 if (scm_is_false (port))
1933 {
1934 if (!scm_port_encoding_init)
1935 return NULL;
1936 else if (!scm_is_fluid (SCM_VARIABLE_REF (scm_port_encoding_var)))
1937 return NULL;
1938 else
1939 {
1940 encoding = scm_fluid_ref (SCM_VARIABLE_REF (scm_port_encoding_var));
1941 if (!scm_is_string (encoding))
1942 return NULL;
1943 else
1944 return scm_i_string_chars (encoding);
1945 }
1946 }
1947 else
1948 {
1949 scm_t_port *pt;
1950 pt = SCM_PTAB_ENTRY (port);
1951 if (pt->encoding)
1952 return pt->encoding;
1953 else
1954 return NULL;
1955 }
1956 }
1957
1958 /* Returns ENC if it is a recognized encoding. If it isn't, it tries
1959 to find an alias of ENC that is valid. Otherwise, it returns
1960 NULL. */
1961 static const char *
1962 find_valid_encoding (const char *enc)
1963 {
1964 int isvalid = 0;
1965 const char str[] = " ";
1966 scm_t_uint32 *u32;
1967 size_t u32len;
1968
1969 u32 = u32_conv_from_encoding (enc, iconveh_error, str, 1,
1970 NULL, NULL, &u32len);
1971 isvalid = (u32 != NULL);
1972 free (u32);
1973
1974 if (isvalid)
1975 return enc;
1976
1977 return NULL;
1978 }
1979
1980 void
1981 scm_i_set_port_encoding_x (SCM port, const char *enc)
1982 {
1983 const char *valid_enc;
1984 scm_t_port *pt;
1985
1986 /* Null is shorthand for the native, Latin-1 encoding. */
1987 if (enc == NULL)
1988 valid_enc = NULL;
1989 else
1990 {
1991 valid_enc = find_valid_encoding (enc);
1992 if (valid_enc == NULL)
1993 {
1994 SCM err;
1995 err = scm_from_locale_string (enc);
1996 scm_misc_error (NULL, "invalid or unknown character encoding ~s",
1997 scm_list_1 (err));
1998 }
1999 }
2000
2001 if (scm_is_false (port))
2002 {
2003 /* Set the default encoding for future ports. */
2004 if (!scm_port_encoding_init
2005 || !scm_is_fluid (SCM_VARIABLE_REF (scm_port_encoding_var)))
2006 scm_misc_error (NULL, "tried to set port encoding fluid before it is initialized",
2007 SCM_EOL);
2008
2009 if (valid_enc == NULL
2010 || !strcmp (valid_enc, "ASCII")
2011 || !strcmp (valid_enc, "ANSI_X3.4-1968")
2012 || !strcmp (valid_enc, "ISO-8859-1"))
2013 scm_fluid_set_x (SCM_VARIABLE_REF (scm_port_encoding_var), SCM_BOOL_F);
2014 else
2015 scm_fluid_set_x (SCM_VARIABLE_REF (scm_port_encoding_var),
2016 scm_from_locale_string (valid_enc));
2017 }
2018 else
2019 {
2020 /* Set the character encoding for this port. */
2021 pt = SCM_PTAB_ENTRY (port);
2022 if (valid_enc == NULL)
2023 pt->encoding = NULL;
2024 else
2025 pt->encoding = scm_gc_strdup (valid_enc, "port");
2026 }
2027 }
2028
2029 SCM_DEFINE (scm_port_encoding, "port-encoding", 1, 0, 0,
2030 (SCM port),
2031 "Returns, as a string, the character encoding that @var{port}\n"
2032 "uses to interpret its input and output.\n")
2033 #define FUNC_NAME s_scm_port_encoding
2034 {
2035 scm_t_port *pt;
2036 const char *enc;
2037
2038 SCM_VALIDATE_PORT (1, port);
2039
2040 pt = SCM_PTAB_ENTRY (port);
2041 enc = scm_i_get_port_encoding (port);
2042 if (enc)
2043 return scm_from_locale_string (pt->encoding);
2044 else
2045 return scm_from_locale_string ("NONE");
2046 }
2047 #undef FUNC_NAME
2048
2049 SCM_DEFINE (scm_set_port_encoding_x, "set-port-encoding!", 2, 0, 0,
2050 (SCM port, SCM enc),
2051 "Sets the character encoding that will be used to interpret all\n"
2052 "port I/O. New ports are created with the encoding\n"
2053 "appropriate for the current locale if @code{setlocale} has \n"
2054 "been called or ISO-8859-1 otherwise\n"
2055 "and this procedure can be used to modify that encoding.\n")
2056
2057 #define FUNC_NAME s_scm_set_port_encoding_x
2058 {
2059 char *enc_str;
2060 const char *valid_enc_str;
2061
2062 SCM_VALIDATE_PORT (1, port);
2063 SCM_VALIDATE_STRING (2, enc);
2064
2065 enc_str = scm_to_locale_string (enc);
2066 valid_enc_str = find_valid_encoding (enc_str);
2067 if (valid_enc_str == NULL)
2068 {
2069 free (enc_str);
2070 scm_misc_error (FUNC_NAME, "invalid or unknown character encoding ~s",
2071 scm_list_1 (enc));
2072 }
2073 else
2074 {
2075 scm_i_set_port_encoding_x (port, valid_enc_str);
2076 free (enc_str);
2077 }
2078 return SCM_UNSPECIFIED;
2079 }
2080 #undef FUNC_NAME
2081
2082
2083 /* This determines how conversions handle unconvertible characters. */
2084 SCM_GLOBAL_VARIABLE (scm_conversion_strategy, "%port-conversion-strategy");
2085 static int scm_conversion_strategy_init = 0;
2086
2087 scm_t_string_failed_conversion_handler
2088 scm_i_get_conversion_strategy (SCM port)
2089 {
2090 SCM encoding;
2091
2092 if (scm_is_false (port))
2093 {
2094 if (!scm_conversion_strategy_init
2095 || !scm_is_fluid (SCM_VARIABLE_REF (scm_conversion_strategy)))
2096 return SCM_FAILED_CONVERSION_QUESTION_MARK;
2097 else
2098 {
2099 encoding = scm_fluid_ref (SCM_VARIABLE_REF (scm_conversion_strategy));
2100 if (scm_is_false (encoding))
2101 return SCM_FAILED_CONVERSION_QUESTION_MARK;
2102 else
2103 return (scm_t_string_failed_conversion_handler) scm_to_int (encoding);
2104 }
2105 }
2106 else
2107 {
2108 scm_t_port *pt;
2109 pt = SCM_PTAB_ENTRY (port);
2110 return pt->ilseq_handler;
2111 }
2112
2113 }
2114
2115 void
2116 scm_i_set_conversion_strategy_x (SCM port,
2117 scm_t_string_failed_conversion_handler handler)
2118 {
2119 SCM strategy;
2120 scm_t_port *pt;
2121
2122 strategy = scm_from_int ((int) handler);
2123
2124 if (scm_is_false (port))
2125 {
2126 /* Set the default encoding for future ports. */
2127 if (!scm_conversion_strategy
2128 || !scm_is_fluid (SCM_VARIABLE_REF (scm_conversion_strategy)))
2129 scm_misc_error (NULL, "tried to set conversion strategy fluid before it is initialized",
2130 SCM_EOL);
2131 scm_fluid_set_x (SCM_VARIABLE_REF (scm_conversion_strategy), strategy);
2132 }
2133 else
2134 {
2135 /* Set the character encoding for this port. */
2136 pt = SCM_PTAB_ENTRY (port);
2137 pt->ilseq_handler = handler;
2138 }
2139 }
2140
2141 SCM_DEFINE (scm_port_conversion_strategy, "port-conversion-strategy",
2142 1, 0, 0, (SCM port),
2143 "Returns the behavior of the port when handling a character that\n"
2144 "is not representable in the port's current encoding.\n"
2145 "It returns the symbol @code{error} if unrepresentable characters\n"
2146 "should cause exceptions, @code{substitute} if the port should\n"
2147 "try to replace unrepresentable characters with question marks or\n"
2148 "approximate characters, or @code{escape} if unrepresentable\n"
2149 "characters should be converted to string escapes.\n"
2150 "\n"
2151 "If @var{port} is @code{#f}, then the current default behavior\n"
2152 "will be returned. New ports will have this default behavior\n"
2153 "when they are created.\n")
2154 #define FUNC_NAME s_scm_port_conversion_strategy
2155 {
2156 scm_t_string_failed_conversion_handler h;
2157
2158 SCM_VALIDATE_OPPORT (1, port);
2159
2160 if (!scm_is_false (port))
2161 {
2162 SCM_VALIDATE_OPPORT (1, port);
2163 }
2164
2165 h = scm_i_get_conversion_strategy (port);
2166 if (h == SCM_FAILED_CONVERSION_ERROR)
2167 return scm_from_locale_symbol ("error");
2168 else if (h == SCM_FAILED_CONVERSION_QUESTION_MARK)
2169 return scm_from_locale_symbol ("substitute");
2170 else if (h == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
2171 return scm_from_locale_symbol ("escape");
2172 else
2173 abort ();
2174
2175 /* Never gets here. */
2176 return SCM_UNDEFINED;
2177 }
2178 #undef FUNC_NAME
2179
2180 SCM_DEFINE (scm_set_port_conversion_strategy_x, "set-port-conversion-strategy!",
2181 2, 0, 0,
2182 (SCM port, SCM sym),
2183 "Sets the behavior of the interpreter when outputting a character\n"
2184 "that is not representable in the port's current encoding.\n"
2185 "@var{sym} can be either @code{'error}, @code{'substitute}, or\n"
2186 "@code{'escape}. If it is @code{'error}, an error will be thrown\n"
2187 "when an unconvertible character is encountered. If it is\n"
2188 "@code{'substitute}, then unconvertible characters will \n"
2189 "be replaced with approximate characters, or with question marks\n"
2190 "if no approximately correct character is available.\n"
2191 "If it is @code{'escape},\n"
2192 "it will appear as a hex escape when output.\n"
2193 "\n"
2194 "If @var{port} is an open port, the conversion error behavior\n"
2195 "is set for that port. If it is @code{#f}, it is set as the\n"
2196 "default behavior for any future ports that get created in\n"
2197 "this thread.\n")
2198 #define FUNC_NAME s_scm_set_port_conversion_strategy_x
2199 {
2200 SCM err;
2201 SCM qm;
2202 SCM esc;
2203
2204 if (!scm_is_false (port))
2205 {
2206 SCM_VALIDATE_OPPORT (1, port);
2207 }
2208
2209 err = scm_from_locale_symbol ("error");
2210 if (scm_is_true (scm_eqv_p (sym, err)))
2211 {
2212 scm_i_set_conversion_strategy_x (port, SCM_FAILED_CONVERSION_ERROR);
2213 return SCM_UNSPECIFIED;
2214 }
2215
2216 qm = scm_from_locale_symbol ("substitute");
2217 if (scm_is_true (scm_eqv_p (sym, qm)))
2218 {
2219 scm_i_set_conversion_strategy_x (port,
2220 SCM_FAILED_CONVERSION_QUESTION_MARK);
2221 return SCM_UNSPECIFIED;
2222 }
2223
2224 esc = scm_from_locale_symbol ("escape");
2225 if (scm_is_true (scm_eqv_p (sym, esc)))
2226 {
2227 scm_i_set_conversion_strategy_x (port,
2228 SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE);
2229 return SCM_UNSPECIFIED;
2230 }
2231
2232 SCM_MISC_ERROR ("unknown conversion behavior ~s", scm_list_1 (sym));
2233
2234 return SCM_UNSPECIFIED;
2235 }
2236 #undef FUNC_NAME
2237
2238
2239
2240 void
2241 scm_print_port_mode (SCM exp, SCM port)
2242 {
2243 scm_puts (SCM_CLOSEDP (exp)
2244 ? "closed: "
2245 : (SCM_RDNG & SCM_CELL_WORD_0 (exp)
2246 ? (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
2247 ? "input-output: "
2248 : "input: ")
2249 : (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
2250 ? "output: "
2251 : "bogus: ")),
2252 port);
2253 }
2254
2255 int
2256 scm_port_print (SCM exp, SCM port, scm_print_state *pstate SCM_UNUSED)
2257 {
2258 char *type = SCM_PTOBNAME (SCM_PTOBNUM (exp));
2259 if (!type)
2260 type = "port";
2261 scm_puts ("#<", port);
2262 scm_print_port_mode (exp, port);
2263 scm_puts (type, port);
2264 scm_putc (' ', port);
2265 scm_uintprint (SCM_CELL_WORD_1 (exp), 16, port);
2266 scm_putc ('>', port);
2267 return 1;
2268 }
2269
2270 void
2271 scm_ports_prehistory ()
2272 {
2273 scm_numptob = 0;
2274 scm_ptobs = NULL;
2275 }
2276
2277 \f
2278
2279 /* Void ports. */
2280
2281 scm_t_bits scm_tc16_void_port = 0;
2282
2283 static int fill_input_void_port (SCM port SCM_UNUSED)
2284 {
2285 return EOF;
2286 }
2287
2288 static void
2289 write_void_port (SCM port SCM_UNUSED,
2290 const void *data SCM_UNUSED,
2291 size_t size SCM_UNUSED)
2292 {
2293 }
2294
2295 static SCM
2296 scm_i_void_port (long mode_bits)
2297 {
2298 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
2299 {
2300 SCM answer = scm_new_port_table_entry (scm_tc16_void_port);
2301 scm_t_port * pt = SCM_PTAB_ENTRY(answer);
2302
2303 scm_port_non_buffer (pt);
2304
2305 SCM_SETSTREAM (answer, 0);
2306 SCM_SET_CELL_TYPE (answer, scm_tc16_void_port | mode_bits);
2307 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
2308 return answer;
2309 }
2310 }
2311
2312 SCM
2313 scm_void_port (char *mode_str)
2314 {
2315 return scm_i_void_port (scm_mode_bits (mode_str));
2316 }
2317
2318 SCM_DEFINE (scm_sys_make_void_port, "%make-void-port", 1, 0, 0,
2319 (SCM mode),
2320 "Create and return a new void port. A void port acts like\n"
2321 "@file{/dev/null}. The @var{mode} argument\n"
2322 "specifies the input/output modes for this port: see the\n"
2323 "documentation for @code{open-file} in @ref{File Ports}.")
2324 #define FUNC_NAME s_scm_sys_make_void_port
2325 {
2326 return scm_i_void_port (scm_i_mode_bits (mode));
2327 }
2328 #undef FUNC_NAME
2329
2330 \f
2331 /* Initialization. */
2332
2333 void
2334 scm_init_ports ()
2335 {
2336 /* lseek() symbols. */
2337 scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET));
2338 scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR));
2339 scm_c_define ("SEEK_END", scm_from_int (SEEK_END));
2340
2341 scm_tc16_void_port = scm_make_port_type ("void", fill_input_void_port,
2342 write_void_port);
2343
2344 cur_inport_fluid = scm_permanent_object (scm_make_fluid ());
2345 cur_outport_fluid = scm_permanent_object (scm_make_fluid ());
2346 cur_errport_fluid = scm_permanent_object (scm_make_fluid ());
2347 cur_loadport_fluid = scm_permanent_object (scm_make_fluid ());
2348
2349 scm_i_port_weak_hash = scm_permanent_object (scm_make_weak_key_hash_table (SCM_I_MAKINUM(31)));
2350 #include "libguile/ports.x"
2351
2352 SCM_VARIABLE_SET (scm_port_encoding_var, scm_make_fluid ());
2353 scm_fluid_set_x (SCM_VARIABLE_REF (scm_port_encoding_var), SCM_BOOL_F);
2354 scm_port_encoding_init = 1;
2355
2356 SCM_VARIABLE_SET (scm_conversion_strategy, scm_make_fluid ());
2357 scm_fluid_set_x (SCM_VARIABLE_REF (scm_conversion_strategy),
2358 scm_from_int ((int) SCM_FAILED_CONVERSION_QUESTION_MARK));
2359 scm_conversion_strategy_init = 1;
2360
2361 }
2362
2363 /*
2364 Local Variables:
2365 c-file-style: "gnu"
2366 End:
2367 */