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