Fix broken interaction between readline and Unicode
[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 "libguile/_scm.h"
38 #include "libguile/async.h"
39 #include "libguile/eval.h"
40 #include "libguile/fports.h" /* direct access for seek and truncate */
41 #include "libguile/objects.h"
42 #include "libguile/goops.h"
43 #include "libguile/smob.h"
44 #include "libguile/chars.h"
45 #include "libguile/dynwind.h"
46
47 #include "libguile/keywords.h"
48 #include "libguile/hashtab.h"
49 #include "libguile/root.h"
50 #include "libguile/strings.h"
51 #include "libguile/mallocs.h"
52 #include "libguile/validate.h"
53 #include "libguile/ports.h"
54 #include "libguile/vectors.h"
55 #include "libguile/weaks.h"
56 #include "libguile/fluids.h"
57 #include "libguile/eq.h"
58
59 #ifdef HAVE_STRING_H
60 #include <string.h>
61 #endif
62
63 #ifdef HAVE_MALLOC_H
64 #include <malloc.h>
65 #endif
66
67 #ifdef HAVE_IO_H
68 #include <io.h>
69 #endif
70
71 #ifdef HAVE_UNISTD_H
72 #include <unistd.h>
73 #endif
74
75 #ifdef HAVE_SYS_IOCTL_H
76 #include <sys/ioctl.h>
77 #endif
78
79 /* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize
80 already, but have this code here in case that wasn't so in past versions,
81 or perhaps to help other minimal DOS environments.
82
83 gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which
84 might be possibilities if we've got other systems without ftruncate. */
85
86 #if HAVE_CHSIZE && ! HAVE_FTRUNCATE
87 #define ftruncate(fd, size) chsize (fd, size)
88 #undef HAVE_FTRUNCATE
89 #define HAVE_FTRUNCATE 1
90 #endif
91
92 \f
93 /* The port kind table --- a dynamically resized array of port types. */
94
95
96 /* scm_ptobs scm_numptob
97 * implement a dynamically resized array of ptob records.
98 * Indexes into this table are used when generating type
99 * tags for smobjects (if you know a tag you can get an index and conversely).
100 */
101 scm_t_ptob_descriptor *scm_ptobs;
102 long scm_numptob;
103
104 /* GC marker for a port with stream of SCM type. */
105 SCM
106 scm_markstream (SCM ptr)
107 {
108 int openp;
109 openp = SCM_CELL_WORD_0 (ptr) & SCM_OPN;
110 if (openp)
111 return SCM_PACK (SCM_STREAM (ptr));
112 else
113 return SCM_BOOL_F;
114 }
115
116 /*
117 * We choose to use an interface similar to the smob interface with
118 * fill_input and write as standard fields, passed to the port
119 * type constructor, and optional fields set by setters.
120 */
121
122 static void
123 flush_port_default (SCM port SCM_UNUSED)
124 {
125 }
126
127 static void
128 end_input_default (SCM port SCM_UNUSED, int offset SCM_UNUSED)
129 {
130 }
131
132 static size_t
133 scm_port_free0 (SCM port)
134 {
135 return 0;
136 }
137
138 scm_t_bits
139 scm_make_port_type (char *name,
140 int (*fill_input) (SCM port),
141 void (*write) (SCM port, const void *data, size_t size))
142 {
143 char *tmp;
144 if (SCM_I_MAX_PORT_TYPE_COUNT - 1 <= scm_numptob)
145 goto ptoberr;
146 SCM_CRITICAL_SECTION_START;
147 SCM_SYSCALL (tmp = (char *) realloc ((char *) scm_ptobs,
148 (1 + scm_numptob)
149 * sizeof (scm_t_ptob_descriptor)));
150 if (tmp)
151 {
152 scm_ptobs = (scm_t_ptob_descriptor *) tmp;
153
154 scm_ptobs[scm_numptob].name = name;
155 scm_ptobs[scm_numptob].mark = 0;
156 scm_ptobs[scm_numptob].free = scm_port_free0;
157 scm_ptobs[scm_numptob].print = scm_port_print;
158 scm_ptobs[scm_numptob].equalp = 0;
159 scm_ptobs[scm_numptob].close = 0;
160
161 scm_ptobs[scm_numptob].write = write;
162 scm_ptobs[scm_numptob].flush = flush_port_default;
163
164 scm_ptobs[scm_numptob].end_input = end_input_default;
165 scm_ptobs[scm_numptob].fill_input = fill_input;
166 scm_ptobs[scm_numptob].input_waiting = 0;
167
168 scm_ptobs[scm_numptob].seek = 0;
169 scm_ptobs[scm_numptob].truncate = 0;
170
171 scm_numptob++;
172 }
173 SCM_CRITICAL_SECTION_END;
174 if (!tmp)
175 {
176 ptoberr:
177 scm_memory_error ("scm_make_port_type");
178 }
179 /* Make a class object if Goops is present */
180 if (SCM_UNPACK (scm_port_class[0]) != 0)
181 scm_make_port_classes (scm_numptob - 1, SCM_PTOBNAME (scm_numptob - 1));
182 return scm_tc7_port + (scm_numptob - 1) * 256;
183 }
184
185 void
186 scm_set_port_mark (scm_t_bits tc, SCM (*mark) (SCM))
187 {
188 scm_ptobs[SCM_TC2PTOBNUM (tc)].mark = mark;
189 }
190
191 void
192 scm_set_port_free (scm_t_bits tc, size_t (*free) (SCM))
193 {
194 scm_ptobs[SCM_TC2PTOBNUM (tc)].free = free;
195 }
196
197 void
198 scm_set_port_print (scm_t_bits tc, int (*print) (SCM exp, SCM port,
199 scm_print_state *pstate))
200 {
201 scm_ptobs[SCM_TC2PTOBNUM (tc)].print = print;
202 }
203
204 void
205 scm_set_port_equalp (scm_t_bits tc, SCM (*equalp) (SCM, SCM))
206 {
207 scm_ptobs[SCM_TC2PTOBNUM (tc)].equalp = equalp;
208 }
209
210 void
211 scm_set_port_flush (scm_t_bits tc, void (*flush) (SCM port))
212 {
213 scm_ptobs[SCM_TC2PTOBNUM (tc)].flush = flush;
214 }
215
216 void
217 scm_set_port_end_input (scm_t_bits tc, void (*end_input) (SCM port, int offset))
218 {
219 scm_ptobs[SCM_TC2PTOBNUM (tc)].end_input = end_input;
220 }
221
222 void
223 scm_set_port_close (scm_t_bits tc, int (*close) (SCM))
224 {
225 scm_ptobs[SCM_TC2PTOBNUM (tc)].close = close;
226 }
227
228 void
229 scm_set_port_seek (scm_t_bits tc,
230 scm_t_off (*seek) (SCM, scm_t_off, int))
231 {
232 scm_ptobs[SCM_TC2PTOBNUM (tc)].seek = seek;
233 }
234
235 void
236 scm_set_port_truncate (scm_t_bits tc, void (*truncate) (SCM, scm_t_off))
237 {
238 scm_ptobs[SCM_TC2PTOBNUM (tc)].truncate = truncate;
239 }
240
241 void
242 scm_set_port_input_waiting (scm_t_bits tc, int (*input_waiting) (SCM))
243 {
244 scm_ptobs[SCM_TC2PTOBNUM (tc)].input_waiting = input_waiting;
245 }
246
247 \f
248
249 SCM_DEFINE (scm_char_ready_p, "char-ready?", 0, 1, 0,
250 (SCM port),
251 "Return @code{#t} if a character is ready on input @var{port}\n"
252 "and return @code{#f} otherwise. If @code{char-ready?} returns\n"
253 "@code{#t} then the next @code{read-char} operation on\n"
254 "@var{port} is guaranteed not to hang. If @var{port} is a file\n"
255 "port at end of file then @code{char-ready?} returns @code{#t}.\n"
256 "\n"
257 "@code{char-ready?} exists to make it possible for a\n"
258 "program to accept characters from interactive ports without\n"
259 "getting stuck waiting for input. Any input editors associated\n"
260 "with such ports must make sure that characters whose existence\n"
261 "has been asserted by @code{char-ready?} cannot be rubbed out.\n"
262 "If @code{char-ready?} were to return @code{#f} at end of file,\n"
263 "a port at end of file would be indistinguishable from an\n"
264 "interactive port that has no ready characters.")
265 #define FUNC_NAME s_scm_char_ready_p
266 {
267 scm_t_port *pt;
268
269 if (SCM_UNBNDP (port))
270 port = scm_current_input_port ();
271 else
272 SCM_VALIDATE_OPINPORT (1, port);
273
274 pt = SCM_PTAB_ENTRY (port);
275
276 /* if the current read buffer is filled, or the
277 last pushed-back char has been read and the saved buffer is
278 filled, result is true. */
279 if (pt->read_pos < pt->read_end
280 || (pt->read_buf == pt->putback_buf
281 && pt->saved_read_pos < pt->saved_read_end))
282 return SCM_BOOL_T;
283 else
284 {
285 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
286
287 if (ptob->input_waiting)
288 return scm_from_bool(ptob->input_waiting (port));
289 else
290 return SCM_BOOL_T;
291 }
292 }
293 #undef FUNC_NAME
294
295 /* move up to read_len chars from port's putback and/or read buffers
296 into memory starting at dest. returns the number of chars moved. */
297 size_t scm_take_from_input_buffers (SCM port, char *dest, size_t read_len)
298 {
299 scm_t_port *pt = SCM_PTAB_ENTRY (port);
300 size_t chars_read = 0;
301 size_t from_buf = min (pt->read_end - pt->read_pos, read_len);
302
303 if (from_buf > 0)
304 {
305 memcpy (dest, pt->read_pos, from_buf);
306 pt->read_pos += from_buf;
307 chars_read += from_buf;
308 read_len -= from_buf;
309 dest += from_buf;
310 }
311
312 /* if putback was active, try the real input buffer too. */
313 if (pt->read_buf == pt->putback_buf)
314 {
315 from_buf = min (pt->saved_read_end - pt->saved_read_pos, read_len);
316 if (from_buf > 0)
317 {
318 memcpy (dest, pt->saved_read_pos, from_buf);
319 pt->saved_read_pos += from_buf;
320 chars_read += from_buf;
321 }
322 }
323 return chars_read;
324 }
325
326 /* Clear a port's read buffers, returning the contents. */
327 SCM_DEFINE (scm_drain_input, "drain-input", 1, 0, 0,
328 (SCM port),
329 "This procedure clears a port's input buffers, similar\n"
330 "to the way that force-output clears the output buffer. The\n"
331 "contents of the buffers are returned as a single string, e.g.,\n"
332 "\n"
333 "@lisp\n"
334 "(define p (open-input-file ...))\n"
335 "(drain-input p) => empty string, nothing buffered yet.\n"
336 "(unread-char (read-char p) p)\n"
337 "(drain-input p) => initial chars from p, up to the buffer size.\n"
338 "@end lisp\n\n"
339 "Draining the buffers may be useful for cleanly finishing\n"
340 "buffered I/O so that the file descriptor can be used directly\n"
341 "for further input.")
342 #define FUNC_NAME s_scm_drain_input
343 {
344 SCM result;
345 char *data;
346 scm_t_port *pt;
347 long count;
348
349 SCM_VALIDATE_OPINPORT (1, port);
350 pt = SCM_PTAB_ENTRY (port);
351
352 count = pt->read_end - pt->read_pos;
353 if (pt->read_buf == pt->putback_buf)
354 count += pt->saved_read_end - pt->saved_read_pos;
355
356 result = scm_i_make_string (count, &data);
357 scm_take_from_input_buffers (port, data, count);
358 return result;
359 }
360 #undef FUNC_NAME
361
362 \f
363 /* Standard ports --- current input, output, error, and more(!). */
364
365 static SCM cur_inport_fluid = 0;
366 static SCM cur_outport_fluid = 0;
367 static SCM cur_errport_fluid = 0;
368 static SCM cur_loadport_fluid = 0;
369
370 SCM_DEFINE (scm_current_input_port, "current-input-port", 0, 0, 0,
371 (),
372 "Return the current input port. This is the default port used\n"
373 "by many input procedures. Initially, @code{current-input-port}\n"
374 "returns the @dfn{standard input} in Unix and C terminology.")
375 #define FUNC_NAME s_scm_current_input_port
376 {
377 if (cur_inport_fluid)
378 return scm_fluid_ref (cur_inport_fluid);
379 else
380 return SCM_BOOL_F;
381 }
382 #undef FUNC_NAME
383
384 SCM_DEFINE (scm_current_output_port, "current-output-port", 0, 0, 0,
385 (),
386 "Return the current output port. This is the default port used\n"
387 "by many output procedures. Initially,\n"
388 "@code{current-output-port} returns the @dfn{standard output} in\n"
389 "Unix and C terminology.")
390 #define FUNC_NAME s_scm_current_output_port
391 {
392 if (cur_outport_fluid)
393 return scm_fluid_ref (cur_outport_fluid);
394 else
395 return SCM_BOOL_F;
396 }
397 #undef FUNC_NAME
398
399 SCM_DEFINE (scm_current_error_port, "current-error-port", 0, 0, 0,
400 (),
401 "Return the port to which errors and warnings should be sent (the\n"
402 "@dfn{standard error} in Unix and C terminology).")
403 #define FUNC_NAME s_scm_current_error_port
404 {
405 if (cur_errport_fluid)
406 return scm_fluid_ref (cur_errport_fluid);
407 else
408 return SCM_BOOL_F;
409 }
410 #undef FUNC_NAME
411
412 SCM_DEFINE (scm_current_load_port, "current-load-port", 0, 0, 0,
413 (),
414 "Return the current-load-port.\n"
415 "The load port is used internally by @code{primitive-load}.")
416 #define FUNC_NAME s_scm_current_load_port
417 {
418 return scm_fluid_ref (cur_loadport_fluid);
419 }
420 #undef FUNC_NAME
421
422 SCM_DEFINE (scm_set_current_input_port, "set-current-input-port", 1, 0, 0,
423 (SCM port),
424 "@deffnx {Scheme Procedure} set-current-output-port port\n"
425 "@deffnx {Scheme Procedure} set-current-error-port port\n"
426 "Change the ports returned by @code{current-input-port},\n"
427 "@code{current-output-port} and @code{current-error-port}, respectively,\n"
428 "so that they use the supplied @var{port} for input or output.")
429 #define FUNC_NAME s_scm_set_current_input_port
430 {
431 SCM oinp = scm_fluid_ref (cur_inport_fluid);
432 SCM_VALIDATE_OPINPORT (1, port);
433 scm_fluid_set_x (cur_inport_fluid, port);
434 return oinp;
435 }
436 #undef FUNC_NAME
437
438
439 SCM_DEFINE (scm_set_current_output_port, "set-current-output-port", 1, 0, 0,
440 (SCM port),
441 "Set the current default output port to @var{port}.")
442 #define FUNC_NAME s_scm_set_current_output_port
443 {
444 SCM ooutp = scm_fluid_ref (cur_outport_fluid);
445 port = SCM_COERCE_OUTPORT (port);
446 SCM_VALIDATE_OPOUTPORT (1, port);
447 scm_fluid_set_x (cur_outport_fluid, port);
448 return ooutp;
449 }
450 #undef FUNC_NAME
451
452
453 SCM_DEFINE (scm_set_current_error_port, "set-current-error-port", 1, 0, 0,
454 (SCM port),
455 "Set the current default error port to @var{port}.")
456 #define FUNC_NAME s_scm_set_current_error_port
457 {
458 SCM oerrp = scm_fluid_ref (cur_errport_fluid);
459 port = SCM_COERCE_OUTPORT (port);
460 SCM_VALIDATE_OPOUTPORT (1, port);
461 scm_fluid_set_x (cur_errport_fluid, port);
462 return oerrp;
463 }
464 #undef FUNC_NAME
465
466 void
467 scm_dynwind_current_input_port (SCM port)
468 #define FUNC_NAME NULL
469 {
470 SCM_VALIDATE_OPINPORT (1, port);
471 scm_dynwind_fluid (cur_inport_fluid, port);
472 }
473 #undef FUNC_NAME
474
475 void
476 scm_dynwind_current_output_port (SCM port)
477 #define FUNC_NAME NULL
478 {
479 port = SCM_COERCE_OUTPORT (port);
480 SCM_VALIDATE_OPOUTPORT (1, port);
481 scm_dynwind_fluid (cur_outport_fluid, port);
482 }
483 #undef FUNC_NAME
484
485 void
486 scm_dynwind_current_error_port (SCM port)
487 #define FUNC_NAME NULL
488 {
489 port = SCM_COERCE_OUTPORT (port);
490 SCM_VALIDATE_OPOUTPORT (1, port);
491 scm_dynwind_fluid (cur_errport_fluid, port);
492 }
493 #undef FUNC_NAME
494
495 void
496 scm_i_dynwind_current_load_port (SCM port)
497 {
498 scm_dynwind_fluid (cur_loadport_fluid, port);
499 }
500
501 \f
502 /* The port table --- an array of pointers to ports. */
503
504 /*
505 We need a global registry of ports to flush them all at exit, and to
506 get all the ports matching a file descriptor.
507 */
508 SCM scm_i_port_weak_hash;
509
510 scm_i_pthread_mutex_t scm_i_port_table_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER;
511
512 /* This function is not and should not be thread safe. */
513
514 SCM
515 scm_new_port_table_entry (scm_t_bits tag)
516 #define FUNC_NAME "scm_new_port_table_entry"
517 {
518 /*
519 We initialize the cell to empty, this is in case scm_gc_calloc
520 triggers GC ; we don't want the GC to scan a half-finished Z.
521 */
522
523 SCM z = scm_cons (SCM_EOL, SCM_EOL);
524 scm_t_port *entry = (scm_t_port *) scm_gc_calloc (sizeof (scm_t_port), "port");
525 const char *enc;
526
527 entry->file_name = SCM_BOOL_F;
528 entry->rw_active = SCM_PORT_NEITHER;
529 entry->port = z;
530 /* Initialize this port with the thread's current default
531 encoding. */
532 if ((enc = scm_i_get_port_encoding (SCM_BOOL_F)) == NULL)
533 entry->encoding = NULL;
534 else
535 entry->encoding = strdup (enc);
536 entry->ilseq_handler = scm_i_get_conversion_strategy (SCM_BOOL_F);
537
538 SCM_SET_CELL_TYPE (z, tag);
539 SCM_SETPTAB_ENTRY (z, entry);
540
541 scm_hashq_set_x (scm_i_port_weak_hash, z, SCM_BOOL_F);
542
543 return z;
544 }
545 #undef FUNC_NAME
546
547 #if SCM_ENABLE_DEPRECATED==1
548 SCM_API scm_t_port *
549 scm_add_to_port_table (SCM port)
550 {
551 SCM z = scm_new_port_table_entry (scm_tc7_port);
552 scm_t_port * pt = SCM_PTAB_ENTRY(z);
553
554 pt->port = port;
555 SCM_SETCAR (z, SCM_EOL);
556 SCM_SETCDR (z, SCM_EOL);
557 SCM_SETPTAB_ENTRY (port, pt);
558 return pt;
559 }
560 #endif
561
562
563 /* Remove a port from the table and destroy it. */
564
565 /* This function is not and should not be thread safe. */
566 void
567 scm_i_remove_port (SCM port)
568 #define FUNC_NAME "scm_remove_port"
569 {
570 scm_t_port *p = SCM_PTAB_ENTRY (port);
571 if (p->putback_buf)
572 scm_gc_free (p->putback_buf, p->putback_buf_size, "putback buffer");
573 if (p->encoding)
574 {
575 free (p->encoding);
576 p->encoding = NULL;
577 }
578 scm_gc_free (p, sizeof (scm_t_port), "port");
579
580 SCM_SETPTAB_ENTRY (port, 0);
581 scm_hashq_remove_x (scm_i_port_weak_hash, port);
582 }
583 #undef FUNC_NAME
584
585
586 /* Functions for debugging. */
587 #ifdef GUILE_DEBUG
588 SCM_DEFINE (scm_pt_size, "pt-size", 0, 0, 0,
589 (),
590 "Return the number of ports in the port table. @code{pt-size}\n"
591 "is only included in @code{--enable-guile-debug} builds.")
592 #define FUNC_NAME s_scm_pt_size
593 {
594 return scm_from_int (SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash));
595 }
596 #undef FUNC_NAME
597 #endif
598
599 void
600 scm_port_non_buffer (scm_t_port *pt)
601 {
602 pt->read_pos = pt->read_buf = pt->read_end = &pt->shortbuf;
603 pt->write_buf = pt->write_pos = &pt->shortbuf;
604 pt->read_buf_size = pt->write_buf_size = 1;
605 pt->write_end = pt->write_buf + pt->write_buf_size;
606 }
607
608 \f
609 /* Revealed counts --- an oddity inherited from SCSH. */
610
611 /* Find a port in the table and return its revealed count.
612 Also used by the garbage collector.
613 */
614
615 int
616 scm_revealed_count (SCM port)
617 {
618 return SCM_REVEALED(port);
619 }
620
621
622
623 /* Return the revealed count for a port. */
624
625 SCM_DEFINE (scm_port_revealed, "port-revealed", 1, 0, 0,
626 (SCM port),
627 "Return the revealed count for @var{port}.")
628 #define FUNC_NAME s_scm_port_revealed
629 {
630 port = SCM_COERCE_OUTPORT (port);
631 SCM_VALIDATE_OPENPORT (1, port);
632 return scm_from_int (scm_revealed_count (port));
633 }
634 #undef FUNC_NAME
635
636 /* Set the revealed count for a port. */
637 SCM_DEFINE (scm_set_port_revealed_x, "set-port-revealed!", 2, 0, 0,
638 (SCM port, SCM rcount),
639 "Sets the revealed count for a port to a given value.\n"
640 "The return value is unspecified.")
641 #define FUNC_NAME s_scm_set_port_revealed_x
642 {
643 port = SCM_COERCE_OUTPORT (port);
644 SCM_VALIDATE_OPENPORT (1, port);
645 SCM_REVEALED (port) = scm_to_int (rcount);
646 return SCM_UNSPECIFIED;
647 }
648 #undef FUNC_NAME
649
650
651 \f
652 /* Retrieving a port's mode. */
653
654 /* Return the flags that characterize a port based on the mode
655 * string used to open a file for that port.
656 *
657 * See PORT FLAGS in scm.h
658 */
659
660 static long
661 scm_i_mode_bits_n (SCM modes)
662 {
663 return (SCM_OPN
664 | (scm_i_string_contains_char (modes, 'r')
665 || scm_i_string_contains_char (modes, '+') ? SCM_RDNG : 0)
666 | (scm_i_string_contains_char (modes, 'w')
667 || scm_i_string_contains_char (modes, 'a')
668 || scm_i_string_contains_char (modes, '+') ? SCM_WRTNG : 0)
669 | (scm_i_string_contains_char (modes, '0') ? SCM_BUF0 : 0)
670 | (scm_i_string_contains_char (modes, 'l') ? SCM_BUFLINE : 0));
671 }
672
673 long
674 scm_mode_bits (char *modes)
675 {
676 return scm_i_mode_bits (scm_from_locale_string (modes));
677 }
678
679 long
680 scm_i_mode_bits (SCM modes)
681 {
682 long bits;
683
684 if (!scm_is_string (modes))
685 scm_wrong_type_arg_msg (NULL, 0, modes, "string");
686
687 bits = scm_i_mode_bits_n (modes);
688 scm_remember_upto_here_1 (modes);
689 return bits;
690 }
691
692 /* Return the mode flags from an open port.
693 * Some modes such as "append" are only used when opening
694 * a file and are not returned here. */
695
696 SCM_DEFINE (scm_port_mode, "port-mode", 1, 0, 0,
697 (SCM port),
698 "Return the port modes associated with the open port @var{port}.\n"
699 "These will not necessarily be identical to the modes used when\n"
700 "the port was opened, since modes such as \"append\" which are\n"
701 "used only during port creation are not retained.")
702 #define FUNC_NAME s_scm_port_mode
703 {
704 char modes[4];
705 modes[0] = '\0';
706
707 port = SCM_COERCE_OUTPORT (port);
708 SCM_VALIDATE_OPPORT (1, port);
709 if (SCM_CELL_WORD_0 (port) & SCM_RDNG) {
710 if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
711 strcpy (modes, "r+");
712 else
713 strcpy (modes, "r");
714 }
715 else if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
716 strcpy (modes, "w");
717 if (SCM_CELL_WORD_0 (port) & SCM_BUF0)
718 strcat (modes, "0");
719 return scm_from_locale_string (modes);
720 }
721 #undef FUNC_NAME
722
723
724 \f
725 /* Closing ports. */
726
727 /* scm_close_port
728 * Call the close operation on a port object.
729 * see also scm_close.
730 */
731 SCM_DEFINE (scm_close_port, "close-port", 1, 0, 0,
732 (SCM port),
733 "Close the specified port object. Return @code{#t} if it\n"
734 "successfully closes a port or @code{#f} if it was already\n"
735 "closed. An exception may be raised if an error occurs, for\n"
736 "example when flushing buffered output. See also @ref{Ports and\n"
737 "File Descriptors, close}, for a procedure which can close file\n"
738 "descriptors.")
739 #define FUNC_NAME s_scm_close_port
740 {
741 size_t i;
742 int rv;
743
744 port = SCM_COERCE_OUTPORT (port);
745
746 SCM_VALIDATE_PORT (1, port);
747 if (SCM_CLOSEDP (port))
748 return SCM_BOOL_F;
749 i = SCM_PTOBNUM (port);
750 if (scm_ptobs[i].close)
751 rv = (scm_ptobs[i].close) (port);
752 else
753 rv = 0;
754 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
755 scm_i_remove_port (port);
756 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
757 SCM_CLR_PORT_OPEN_FLAG (port);
758 return scm_from_bool (rv >= 0);
759 }
760 #undef FUNC_NAME
761
762 SCM_DEFINE (scm_close_input_port, "close-input-port", 1, 0, 0,
763 (SCM port),
764 "Close the specified input port object. The routine has no effect if\n"
765 "the file has already been closed. An exception may be raised if an\n"
766 "error occurs. The value returned is unspecified.\n\n"
767 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
768 "which can close file descriptors.")
769 #define FUNC_NAME s_scm_close_input_port
770 {
771 SCM_VALIDATE_INPUT_PORT (1, port);
772 scm_close_port (port);
773 return SCM_UNSPECIFIED;
774 }
775 #undef FUNC_NAME
776
777 SCM_DEFINE (scm_close_output_port, "close-output-port", 1, 0, 0,
778 (SCM port),
779 "Close the specified output port object. The routine has no effect if\n"
780 "the file has already been closed. An exception may be raised if an\n"
781 "error occurs. The value returned is unspecified.\n\n"
782 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
783 "which can close file descriptors.")
784 #define FUNC_NAME s_scm_close_output_port
785 {
786 port = SCM_COERCE_OUTPORT (port);
787 SCM_VALIDATE_OUTPUT_PORT (1, port);
788 scm_close_port (port);
789 return SCM_UNSPECIFIED;
790 }
791 #undef FUNC_NAME
792
793 static SCM
794 scm_i_collect_keys_in_vector (void *closure, SCM key, SCM value, SCM result)
795 {
796 int *i = (int*) closure;
797 scm_c_vector_set_x (result, *i, key);
798 (*i)++;
799
800 return result;
801 }
802
803 void
804 scm_c_port_for_each (void (*proc)(void *data, SCM p), void *data)
805 {
806 int i = 0;
807 size_t n;
808 SCM ports;
809
810 /* Even without pre-emptive multithreading, running arbitrary code
811 while scanning the port table is unsafe because the port table
812 can change arbitrarily (from a GC, for example). So we first
813 collect the ports into a vector. -mvo */
814
815 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
816 n = SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash);
817 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
818 ports = scm_c_make_vector (n, SCM_BOOL_F);
819
820 scm_i_pthread_mutex_lock (&scm_i_port_table_mutex);
821 ports = scm_internal_hash_fold (scm_i_collect_keys_in_vector, &i,
822 ports, scm_i_port_weak_hash);
823 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
824
825 for (i = 0; i < n; i++) {
826 SCM p = SCM_SIMPLE_VECTOR_REF (ports, i);
827 if (SCM_PORTP (p))
828 proc (data, p);
829 }
830
831 scm_remember_upto_here_1 (ports);
832 }
833
834 SCM_DEFINE (scm_port_for_each, "port-for-each", 1, 0, 0,
835 (SCM proc),
836 "Apply @var{proc} to each port in the Guile port table\n"
837 "in turn. The return value is unspecified. More specifically,\n"
838 "@var{proc} is applied exactly once to every port that exists\n"
839 "in the system at the time @var{port-for-each} is invoked.\n"
840 "Changes to the port table while @var{port-for-each} is running\n"
841 "have no effect as far as @var{port-for-each} is concerned.")
842 #define FUNC_NAME s_scm_port_for_each
843 {
844 SCM_VALIDATE_PROC (1, proc);
845
846 scm_c_port_for_each ((void (*)(void*,SCM))scm_call_1, proc);
847 return SCM_UNSPECIFIED;
848 }
849 #undef FUNC_NAME
850
851
852 \f
853 /* Utter miscellany. Gosh, we should clean this up some time. */
854
855 SCM_DEFINE (scm_input_port_p, "input-port?", 1, 0, 0,
856 (SCM x),
857 "Return @code{#t} if @var{x} is an input port, otherwise return\n"
858 "@code{#f}. Any object satisfying this predicate also satisfies\n"
859 "@code{port?}.")
860 #define FUNC_NAME s_scm_input_port_p
861 {
862 return scm_from_bool (SCM_INPUT_PORT_P (x));
863 }
864 #undef FUNC_NAME
865
866 SCM_DEFINE (scm_output_port_p, "output-port?", 1, 0, 0,
867 (SCM x),
868 "Return @code{#t} if @var{x} is an output port, otherwise return\n"
869 "@code{#f}. Any object satisfying this predicate also satisfies\n"
870 "@code{port?}.")
871 #define FUNC_NAME s_scm_output_port_p
872 {
873 x = SCM_COERCE_OUTPORT (x);
874 return scm_from_bool (SCM_OUTPUT_PORT_P (x));
875 }
876 #undef FUNC_NAME
877
878 SCM_DEFINE (scm_port_p, "port?", 1, 0, 0,
879 (SCM x),
880 "Return a boolean indicating whether @var{x} is a port.\n"
881 "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n"
882 "@var{x}))}.")
883 #define FUNC_NAME s_scm_port_p
884 {
885 return scm_from_bool (SCM_PORTP (x));
886 }
887 #undef FUNC_NAME
888
889 SCM_DEFINE (scm_port_closed_p, "port-closed?", 1, 0, 0,
890 (SCM port),
891 "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n"
892 "open.")
893 #define FUNC_NAME s_scm_port_closed_p
894 {
895 SCM_VALIDATE_PORT (1, port);
896 return scm_from_bool (!SCM_OPPORTP (port));
897 }
898 #undef FUNC_NAME
899
900 SCM_DEFINE (scm_eof_object_p, "eof-object?", 1, 0, 0,
901 (SCM x),
902 "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n"
903 "return @code{#f}.")
904 #define FUNC_NAME s_scm_eof_object_p
905 {
906 return scm_from_bool(SCM_EOF_OBJECT_P (x));
907 }
908 #undef FUNC_NAME
909
910 SCM_DEFINE (scm_force_output, "force-output", 0, 1, 0,
911 (SCM port),
912 "Flush the specified output port, or the current output port if @var{port}\n"
913 "is omitted. The current output buffer contents are passed to the\n"
914 "underlying port implementation (e.g., in the case of fports, the\n"
915 "data will be written to the file and the output buffer will be cleared.)\n"
916 "It has no effect on an unbuffered port.\n\n"
917 "The return value is unspecified.")
918 #define FUNC_NAME s_scm_force_output
919 {
920 if (SCM_UNBNDP (port))
921 port = scm_current_output_port ();
922 else
923 {
924 port = SCM_COERCE_OUTPORT (port);
925 SCM_VALIDATE_OPOUTPORT (1, port);
926 }
927 scm_flush (port);
928 return SCM_UNSPECIFIED;
929 }
930 #undef FUNC_NAME
931
932
933 static void
934 flush_output_port (void *closure, SCM port)
935 {
936 if (SCM_OPOUTPORTP (port))
937 scm_flush (port);
938 }
939
940 SCM_DEFINE (scm_flush_all_ports, "flush-all-ports", 0, 0, 0,
941 (),
942 "Equivalent to calling @code{force-output} on\n"
943 "all open output ports. The return value is unspecified.")
944 #define FUNC_NAME s_scm_flush_all_ports
945 {
946 scm_c_port_for_each (&flush_output_port, NULL);
947 return SCM_UNSPECIFIED;
948 }
949 #undef FUNC_NAME
950
951 SCM_DEFINE (scm_read_char, "read-char", 0, 1, 0,
952 (SCM port),
953 "Return the next character available from @var{port}, updating\n"
954 "@var{port} to point to the following character. If no more\n"
955 "characters are available, the end-of-file object is returned.")
956 #define FUNC_NAME s_scm_read_char
957 {
958 scm_t_wchar c;
959 if (SCM_UNBNDP (port))
960 port = scm_current_input_port ();
961 SCM_VALIDATE_OPINPORT (1, port);
962 c = scm_getc (port);
963 if (EOF == c)
964 return SCM_EOF_VAL;
965 return SCM_MAKE_CHAR (c);
966 }
967 #undef FUNC_NAME
968
969 #define SCM_MBCHAR_BUF_SIZE (4)
970
971 /* Get one codepoint from a file, using the port's encoding. */
972 scm_t_wchar
973 scm_getc (SCM port)
974 {
975 int c;
976 unsigned int bufcount = 0;
977 char buf[SCM_MBCHAR_BUF_SIZE];
978 scm_t_wchar codepoint = 0;
979 scm_t_uint32 *u32;
980 size_t u32len;
981 scm_t_port *pt = SCM_PTAB_ENTRY (port);
982
983 c = scm_get_byte_or_eof (port);
984 if (c == EOF)
985 return (scm_t_wchar) EOF;
986
987 buf[0] = c;
988 bufcount++;
989
990 if (pt->encoding == NULL)
991 {
992 /* The encoding is Latin-1: bytes are characters. */
993 codepoint = (unsigned char) buf[0];
994 goto success;
995 }
996
997 for (;;)
998 {
999 u32 = u32_conv_from_encoding (pt->encoding,
1000 (enum iconv_ilseq_handler) pt->ilseq_handler,
1001 buf, bufcount, NULL, NULL, &u32len);
1002 if (u32 == NULL || u32len == 0)
1003 {
1004 if (errno == ENOMEM)
1005 scm_memory_error ("Input decoding");
1006
1007 /* Otherwise errno is EILSEQ or EINVAL, so perhaps more
1008 bytes are needed. Keep looping. */
1009 }
1010 else
1011 {
1012 /* Complete codepoint found. */
1013 codepoint = u32[0];
1014 free (u32);
1015 goto success;
1016 }
1017
1018 if (bufcount == SCM_MBCHAR_BUF_SIZE)
1019 {
1020 /* We've read several bytes and didn't find a good
1021 codepoint. Give up. */
1022 goto failure;
1023 }
1024
1025 c = scm_get_byte_or_eof (port);
1026
1027 if (c == EOF)
1028 {
1029 /* EOF before a complete character was read. Push it all
1030 back and return EOF. */
1031 while (bufcount > 0)
1032 {
1033 /* FIXME: this will probably cause errors in the port column. */
1034 scm_unget_byte (buf[bufcount-1], port);
1035 bufcount --;
1036 }
1037 return EOF;
1038 }
1039
1040 if (c == '\n')
1041 {
1042 /* It is always invalid to have EOL in the middle of a
1043 multibyte character. */
1044 scm_unget_byte ('\n', port);
1045 goto failure;
1046 }
1047
1048 buf[bufcount++] = c;
1049 }
1050
1051 success:
1052 switch (codepoint)
1053 {
1054 case '\a':
1055 break;
1056 case '\b':
1057 SCM_DECCOL (port);
1058 break;
1059 case '\n':
1060 SCM_INCLINE (port);
1061 break;
1062 case '\r':
1063 SCM_ZEROCOL (port);
1064 break;
1065 case '\t':
1066 SCM_TABCOL (port);
1067 break;
1068 default:
1069 SCM_INCCOL (port);
1070 break;
1071 }
1072
1073 return codepoint;
1074
1075 failure:
1076 {
1077 char *err_buf;
1078 SCM err_str = scm_i_make_string (bufcount, &err_buf);
1079 memcpy (err_buf, buf, bufcount);
1080
1081 if (errno == EILSEQ)
1082 scm_misc_error (NULL, "input encoding error for ~s: ~s",
1083 scm_list_2 (scm_from_locale_string (scm_i_get_port_encoding (port)),
1084 err_str));
1085 else
1086 scm_misc_error (NULL, "input encoding error (invalid) for ~s: ~s\n",
1087 scm_list_2 (scm_from_locale_string (scm_i_get_port_encoding (port)),
1088 err_str));
1089 }
1090
1091 /* Never gets here. */
1092 return 0;
1093 }
1094
1095
1096 /* this should only be called when the read buffer is empty. it
1097 tries to refill the read buffer. it returns the first char from
1098 the port, which is either EOF or *(pt->read_pos). */
1099 int
1100 scm_fill_input (SCM port)
1101 {
1102 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1103
1104 assert (pt->read_pos == pt->read_end);
1105
1106 if (pt->read_buf == pt->putback_buf)
1107 {
1108 /* finished reading put-back chars. */
1109 pt->read_buf = pt->saved_read_buf;
1110 pt->read_pos = pt->saved_read_pos;
1111 pt->read_end = pt->saved_read_end;
1112 pt->read_buf_size = pt->saved_read_buf_size;
1113 if (pt->read_pos < pt->read_end)
1114 return *(pt->read_pos);
1115 }
1116 return scm_ptobs[SCM_PTOBNUM (port)].fill_input (port);
1117 }
1118
1119
1120 /* scm_lfwrite
1121 *
1122 * This function differs from scm_c_write; it updates port line and
1123 * column. */
1124
1125 static void
1126 update_port_lf (scm_t_wchar c, SCM port)
1127 {
1128 if (c == '\a')
1129 ; /* Do nothing. */
1130 else if (c == '\b')
1131 SCM_DECCOL (port);
1132 else if (c == '\n')
1133 SCM_INCLINE (port);
1134 else if (c == '\r')
1135 SCM_ZEROCOL (port);
1136 else if (c == '\t')
1137 SCM_TABCOL (port);
1138 else
1139 SCM_INCCOL (port);
1140 }
1141
1142 void
1143 scm_lfwrite (const char *ptr, size_t size, SCM port)
1144 {
1145 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1146 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1147
1148 if (pt->rw_active == SCM_PORT_READ)
1149 scm_end_input (port);
1150
1151 ptob->write (port, ptr, size);
1152
1153 for (; size; ptr++, size--)
1154 update_port_lf ((scm_t_wchar) (unsigned char) *ptr, port);
1155
1156 if (pt->rw_random)
1157 pt->rw_active = SCM_PORT_WRITE;
1158 }
1159
1160 /* Write a scheme string STR to PORT from START inclusive to END
1161 exclusive. */
1162 void
1163 scm_lfwrite_substr (SCM str, size_t start, size_t end, SCM port)
1164 {
1165 size_t i, size = scm_i_string_length (str);
1166 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1167 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1168 scm_t_wchar p;
1169 char *buf;
1170 size_t len;
1171
1172 if (pt->rw_active == SCM_PORT_READ)
1173 scm_end_input (port);
1174
1175 if (end == (size_t) (-1))
1176 end = size;
1177 size = end - start;
1178
1179 /* Note that making a substring will likely take the
1180 stringbuf_write_mutex. So, one shouldn't use scm_lfwrite_substr
1181 if the stringbuf write mutex may still be held elsewhere. */
1182 buf = scm_to_stringn (scm_c_substring (str, start, end), &len,
1183 pt->encoding, pt->ilseq_handler);
1184 ptob->write (port, buf, len);
1185 free (buf);
1186
1187 for (i = 0; i < size; i++)
1188 {
1189 p = scm_i_string_ref (str, i + start);
1190 update_port_lf (p, port);
1191 }
1192
1193 if (pt->rw_random)
1194 pt->rw_active = SCM_PORT_WRITE;
1195 }
1196
1197 /* Write a scheme string STR to PORT. */
1198 void
1199 scm_lfwrite_str (SCM str, SCM port)
1200 {
1201 size_t i, size = scm_i_string_length (str);
1202 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1203 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1204 scm_t_wchar p;
1205 char *buf;
1206 size_t len;
1207
1208 if (pt->rw_active == SCM_PORT_READ)
1209 scm_end_input (port);
1210
1211 buf = scm_to_stringn (str, &len,
1212 pt->encoding, pt->ilseq_handler);
1213 ptob->write (port, buf, len);
1214 free (buf);
1215
1216 for (i = 0; i < size; i++)
1217 {
1218 p = scm_i_string_ref (str, i);
1219 update_port_lf (p, port);
1220 }
1221
1222 if (pt->rw_random)
1223 pt->rw_active = SCM_PORT_WRITE;
1224 }
1225
1226 /* scm_c_read
1227 *
1228 * Used by an application to read arbitrary number of bytes from an
1229 * SCM port. Same semantics as libc read, except that scm_c_read only
1230 * returns less than SIZE bytes if at end-of-file.
1231 *
1232 * Warning: Doesn't update port line and column counts! */
1233
1234 /* This structure, and the following swap_buffer function, are used
1235 for temporarily swapping a port's own read buffer, and the buffer
1236 that the caller of scm_c_read provides. */
1237 struct port_and_swap_buffer
1238 {
1239 scm_t_port *pt;
1240 unsigned char *buffer;
1241 size_t size;
1242 };
1243
1244 static void
1245 swap_buffer (void *data)
1246 {
1247 struct port_and_swap_buffer *psb = (struct port_and_swap_buffer *) data;
1248 unsigned char *old_buf = psb->pt->read_buf;
1249 size_t old_size = psb->pt->read_buf_size;
1250
1251 /* Make the port use (buffer, size) from the struct. */
1252 psb->pt->read_pos = psb->pt->read_buf = psb->pt->read_end = psb->buffer;
1253 psb->pt->read_buf_size = psb->size;
1254
1255 /* Save the port's old (buffer, size) in the struct. */
1256 psb->buffer = old_buf;
1257 psb->size = old_size;
1258 }
1259
1260 size_t
1261 scm_c_read (SCM port, void *buffer, size_t size)
1262 #define FUNC_NAME "scm_c_read"
1263 {
1264 scm_t_port *pt;
1265 size_t n_read = 0, n_available;
1266 struct port_and_swap_buffer psb;
1267
1268 SCM_VALIDATE_OPINPORT (1, port);
1269
1270 pt = SCM_PTAB_ENTRY (port);
1271 if (pt->rw_active == SCM_PORT_WRITE)
1272 scm_ptobs[SCM_PTOBNUM (port)].flush (port);
1273
1274 if (pt->rw_random)
1275 pt->rw_active = SCM_PORT_READ;
1276
1277 /* Take bytes first from the port's read buffer. */
1278 if (pt->read_pos < pt->read_end)
1279 {
1280 n_available = min (size, pt->read_end - pt->read_pos);
1281 memcpy (buffer, pt->read_pos, n_available);
1282 buffer = (char *) buffer + n_available;
1283 pt->read_pos += n_available;
1284 n_read += n_available;
1285 size -= n_available;
1286 }
1287
1288 /* Avoid the scm_dynwind_* costs if we now have enough data. */
1289 if (size == 0)
1290 return n_read;
1291
1292 /* Now we will call scm_fill_input repeatedly until we have read the
1293 requested number of bytes. (Note that a single scm_fill_input
1294 call does not guarantee to fill the whole of the port's read
1295 buffer.) */
1296 if (pt->read_buf_size <= 1 && pt->encoding == NULL)
1297 {
1298 /* The port that we are reading from is unbuffered - i.e. does
1299 not have its own persistent buffer - but we have a buffer,
1300 provided by our caller, that is the right size for the data
1301 that is wanted. For the following scm_fill_input calls,
1302 therefore, we use the buffer in hand as the port's read
1303 buffer.
1304
1305 We need to make sure that the port's normal (1 byte) buffer
1306 is reinstated in case one of the scm_fill_input () calls
1307 throws an exception; we use the scm_dynwind_* API to achieve
1308 that.
1309
1310 A consequence of this optimization is that the fill_input
1311 functions can't unget characters. That'll push data to the
1312 pushback buffer instead of this psb buffer. */
1313 #if SCM_DEBUG == 1
1314 unsigned char *pback = pt->putback_buf;
1315 #endif
1316 psb.pt = pt;
1317 psb.buffer = buffer;
1318 psb.size = size;
1319 scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE);
1320 scm_dynwind_rewind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY);
1321 scm_dynwind_unwind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY);
1322
1323 /* Call scm_fill_input until we have all the bytes that we need,
1324 or we hit EOF. */
1325 while (pt->read_buf_size && (scm_fill_input (port) != EOF))
1326 {
1327 pt->read_buf_size -= (pt->read_end - pt->read_pos);
1328 pt->read_pos = pt->read_buf = pt->read_end;
1329 }
1330 #if SCM_DEBUG == 1
1331 if (pback != pt->putback_buf
1332 || pt->read_buf - (unsigned char *) buffer < 0)
1333 scm_misc_error (FUNC_NAME,
1334 "scm_c_read must not call a fill function that pushes "
1335 "back characters onto an unbuffered port", SCM_EOL);
1336 #endif
1337 n_read += pt->read_buf - (unsigned char *) buffer;
1338
1339 /* Reinstate the port's normal buffer. */
1340 scm_dynwind_end ();
1341 }
1342 else
1343 {
1344 /* The port has its own buffer. It is important that we use it,
1345 even if it happens to be smaller than our caller's buffer, so
1346 that a custom port implementation's entry points (in
1347 particular, fill_input) can rely on the buffer always being
1348 the same as they first set up. */
1349 while (size && (scm_fill_input (port) != EOF))
1350 {
1351 n_available = min (size, pt->read_end - pt->read_pos);
1352 memcpy (buffer, pt->read_pos, n_available);
1353 buffer = (char *) buffer + n_available;
1354 pt->read_pos += n_available;
1355 n_read += n_available;
1356 size -= n_available;
1357 }
1358 }
1359
1360 return n_read;
1361 }
1362 #undef FUNC_NAME
1363
1364 /* scm_c_write
1365 *
1366 * Used by an application to write arbitrary number of bytes to an SCM
1367 * port. Similar semantics as libc write. However, unlike libc
1368 * write, scm_c_write writes the requested number of bytes and has no
1369 * return value.
1370 *
1371 * Warning: Doesn't update port line and column counts!
1372 */
1373
1374 void
1375 scm_c_write (SCM port, const void *ptr, size_t size)
1376 #define FUNC_NAME "scm_c_write"
1377 {
1378 scm_t_port *pt;
1379 scm_t_ptob_descriptor *ptob;
1380
1381 SCM_VALIDATE_OPOUTPORT (1, port);
1382
1383 pt = SCM_PTAB_ENTRY (port);
1384 ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1385
1386 if (pt->rw_active == SCM_PORT_READ)
1387 scm_end_input (port);
1388
1389 ptob->write (port, ptr, size);
1390
1391 if (pt->rw_random)
1392 pt->rw_active = SCM_PORT_WRITE;
1393 }
1394 #undef FUNC_NAME
1395
1396 void
1397 scm_flush (SCM port)
1398 {
1399 long i = SCM_PTOBNUM (port);
1400 (scm_ptobs[i].flush) (port);
1401 }
1402
1403 void
1404 scm_end_input (SCM port)
1405 {
1406 long offset;
1407 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1408
1409 if (pt->read_buf == pt->putback_buf)
1410 {
1411 offset = pt->read_end - pt->read_pos;
1412 pt->read_buf = pt->saved_read_buf;
1413 pt->read_pos = pt->saved_read_pos;
1414 pt->read_end = pt->saved_read_end;
1415 pt->read_buf_size = pt->saved_read_buf_size;
1416 }
1417 else
1418 offset = 0;
1419
1420 scm_ptobs[SCM_PTOBNUM (port)].end_input (port, offset);
1421 }
1422
1423 \f
1424
1425
1426 void
1427 scm_unget_byte (int c, SCM port)
1428 #define FUNC_NAME "scm_unget_byte"
1429 {
1430 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1431
1432 if (pt->read_buf == pt->putback_buf)
1433 /* already using the put-back buffer. */
1434 {
1435 /* enlarge putback_buf if necessary. */
1436 if (pt->read_end == pt->read_buf + pt->read_buf_size
1437 && pt->read_buf == pt->read_pos)
1438 {
1439 size_t new_size = pt->read_buf_size * 2;
1440 unsigned char *tmp = (unsigned char *)
1441 scm_gc_realloc (pt->putback_buf, pt->read_buf_size, new_size,
1442 "putback buffer");
1443
1444 pt->read_pos = pt->read_buf = pt->putback_buf = tmp;
1445 pt->read_end = pt->read_buf + pt->read_buf_size;
1446 pt->read_buf_size = pt->putback_buf_size = new_size;
1447 }
1448
1449 /* shift any existing bytes to buffer + 1. */
1450 if (pt->read_pos == pt->read_end)
1451 pt->read_end = pt->read_buf + 1;
1452 else if (pt->read_pos != pt->read_buf + 1)
1453 {
1454 int count = pt->read_end - pt->read_pos;
1455
1456 memmove (pt->read_buf + 1, pt->read_pos, count);
1457 pt->read_end = pt->read_buf + 1 + count;
1458 }
1459
1460 pt->read_pos = pt->read_buf;
1461 }
1462 else
1463 /* switch to the put-back buffer. */
1464 {
1465 if (pt->putback_buf == NULL)
1466 {
1467 pt->putback_buf
1468 = (unsigned char *) scm_gc_malloc (SCM_INITIAL_PUTBACK_BUF_SIZE,
1469 "putback buffer");
1470 pt->putback_buf_size = SCM_INITIAL_PUTBACK_BUF_SIZE;
1471 }
1472
1473 pt->saved_read_buf = pt->read_buf;
1474 pt->saved_read_pos = pt->read_pos;
1475 pt->saved_read_end = pt->read_end;
1476 pt->saved_read_buf_size = pt->read_buf_size;
1477
1478 pt->read_pos = pt->read_buf = pt->putback_buf;
1479 pt->read_end = pt->read_buf + 1;
1480 pt->read_buf_size = pt->putback_buf_size;
1481 }
1482
1483 *pt->read_buf = c;
1484
1485 if (pt->rw_random)
1486 pt->rw_active = SCM_PORT_READ;
1487 }
1488 #undef FUNC_NAME
1489
1490 void
1491 scm_ungetc (scm_t_wchar c, SCM port)
1492 #define FUNC_NAME "scm_ungetc"
1493 {
1494 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1495 scm_t_wchar *wbuf;
1496 SCM str = scm_i_make_wide_string (1, &wbuf);
1497 char *buf;
1498 size_t len;
1499 int i;
1500
1501 wbuf[0] = c;
1502 buf = scm_to_stringn (str, &len, pt->encoding, pt->ilseq_handler);
1503
1504 for (i = len - 1; i >= 0; i--)
1505 scm_unget_byte (buf[i], port);
1506
1507 if (c == '\n')
1508 {
1509 /* What should col be in this case?
1510 * We'll leave it at -1.
1511 */
1512 SCM_LINUM (port) -= 1;
1513 }
1514 else
1515 SCM_COL(port) -= 1;
1516 }
1517 #undef FUNC_NAME
1518
1519
1520 void
1521 scm_ungets (const char *s, int n, SCM port)
1522 {
1523 /* This is simple minded and inefficient, but unreading strings is
1524 * probably not a common operation, and remember that line and
1525 * column numbers have to be handled...
1526 *
1527 * Please feel free to write an optimized version!
1528 */
1529 while (n--)
1530 scm_ungetc (s[n], port);
1531 }
1532
1533
1534 SCM_DEFINE (scm_peek_char, "peek-char", 0, 1, 0,
1535 (SCM port),
1536 "Return the next character available from @var{port},\n"
1537 "@emph{without} updating @var{port} to point to the following\n"
1538 "character. If no more characters are available, the\n"
1539 "end-of-file object is returned.\n"
1540 "\n"
1541 "The value returned by\n"
1542 "a call to @code{peek-char} is the same as the value that would\n"
1543 "have been returned by a call to @code{read-char} on the same\n"
1544 "port. The only difference is that the very next call to\n"
1545 "@code{read-char} or @code{peek-char} on that @var{port} will\n"
1546 "return the value returned by the preceding call to\n"
1547 "@code{peek-char}. In particular, a call to @code{peek-char} on\n"
1548 "an interactive port will hang waiting for input whenever a call\n"
1549 "to @code{read-char} would have hung.")
1550 #define FUNC_NAME s_scm_peek_char
1551 {
1552 scm_t_wchar c, column;
1553 if (SCM_UNBNDP (port))
1554 port = scm_current_input_port ();
1555 else
1556 SCM_VALIDATE_OPINPORT (1, port);
1557 column = SCM_COL(port);
1558 c = scm_getc (port);
1559 if (EOF == c)
1560 return SCM_EOF_VAL;
1561 scm_ungetc (c, port);
1562 SCM_COL(port) = column;
1563 return SCM_MAKE_CHAR (c);
1564 }
1565 #undef FUNC_NAME
1566
1567 SCM_DEFINE (scm_unread_char, "unread-char", 1, 1, 0,
1568 (SCM cobj, SCM port),
1569 "Place @var{char} in @var{port} so that it will be read by the\n"
1570 "next read operation. If called multiple times, the unread characters\n"
1571 "will be read again in last-in first-out order. If @var{port} is\n"
1572 "not supplied, the current input port is used.")
1573 #define FUNC_NAME s_scm_unread_char
1574 {
1575 int c;
1576
1577 SCM_VALIDATE_CHAR (1, cobj);
1578 if (SCM_UNBNDP (port))
1579 port = scm_current_input_port ();
1580 else
1581 SCM_VALIDATE_OPINPORT (2, port);
1582
1583 c = SCM_CHAR (cobj);
1584
1585 scm_ungetc (c, port);
1586 return cobj;
1587 }
1588 #undef FUNC_NAME
1589
1590 SCM_DEFINE (scm_unread_string, "unread-string", 2, 0, 0,
1591 (SCM str, SCM port),
1592 "Place the string @var{str} in @var{port} so that its characters will be\n"
1593 "read in subsequent read operations. If called multiple times, the\n"
1594 "unread characters will be read again in last-in first-out order. If\n"
1595 "@var{port} is not supplied, the current-input-port is used.")
1596 #define FUNC_NAME s_scm_unread_string
1597 {
1598 int n;
1599 SCM_VALIDATE_STRING (1, str);
1600 if (SCM_UNBNDP (port))
1601 port = scm_current_input_port ();
1602 else
1603 SCM_VALIDATE_OPINPORT (2, port);
1604
1605 n = scm_i_string_length (str);
1606
1607 while (n--)
1608 scm_ungetc (scm_i_string_ref (str, n), port);
1609
1610 return str;
1611 }
1612 #undef FUNC_NAME
1613
1614 SCM_DEFINE (scm_seek, "seek", 3, 0, 0,
1615 (SCM fd_port, SCM offset, SCM whence),
1616 "Sets the current position of @var{fd/port} to the integer\n"
1617 "@var{offset}, which is interpreted according to the value of\n"
1618 "@var{whence}.\n"
1619 "\n"
1620 "One of the following variables should be supplied for\n"
1621 "@var{whence}:\n"
1622 "@defvar SEEK_SET\n"
1623 "Seek from the beginning of the file.\n"
1624 "@end defvar\n"
1625 "@defvar SEEK_CUR\n"
1626 "Seek from the current position.\n"
1627 "@end defvar\n"
1628 "@defvar SEEK_END\n"
1629 "Seek from the end of the file.\n"
1630 "@end defvar\n"
1631 "If @var{fd/port} is a file descriptor, the underlying system\n"
1632 "call is @code{lseek}. @var{port} may be a string port.\n"
1633 "\n"
1634 "The value returned is the new position in the file. This means\n"
1635 "that the current position of a port can be obtained using:\n"
1636 "@lisp\n"
1637 "(seek port 0 SEEK_CUR)\n"
1638 "@end lisp")
1639 #define FUNC_NAME s_scm_seek
1640 {
1641 int how;
1642
1643 fd_port = SCM_COERCE_OUTPORT (fd_port);
1644
1645 how = scm_to_int (whence);
1646 if (how != SEEK_SET && how != SEEK_CUR && how != SEEK_END)
1647 SCM_OUT_OF_RANGE (3, whence);
1648
1649 if (SCM_OPPORTP (fd_port))
1650 {
1651 scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (fd_port);
1652 off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
1653 off_t_or_off64_t rv;
1654
1655 if (!ptob->seek)
1656 SCM_MISC_ERROR ("port is not seekable",
1657 scm_cons (fd_port, SCM_EOL));
1658 else
1659 rv = ptob->seek (fd_port, off, how);
1660 return scm_from_off_t_or_off64_t (rv);
1661 }
1662 else /* file descriptor?. */
1663 {
1664 off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
1665 off_t_or_off64_t rv;
1666 rv = lseek_or_lseek64 (scm_to_int (fd_port), off, how);
1667 if (rv == -1)
1668 SCM_SYSERROR;
1669 return scm_from_off_t_or_off64_t (rv);
1670 }
1671 }
1672 #undef FUNC_NAME
1673
1674 #ifndef O_BINARY
1675 #define O_BINARY 0
1676 #endif
1677
1678 /* Mingw has ftruncate(), perhaps implemented above using chsize, but
1679 doesn't have the filename version truncate(), hence this code. */
1680 #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE
1681 static int
1682 truncate (const char *file, off_t length)
1683 {
1684 int ret, fdes;
1685
1686 fdes = open (file, O_BINARY | O_WRONLY);
1687 if (fdes == -1)
1688 return -1;
1689
1690 ret = ftruncate (fdes, length);
1691 if (ret == -1)
1692 {
1693 int save_errno = errno;
1694 close (fdes);
1695 errno = save_errno;
1696 return -1;
1697 }
1698
1699 return close (fdes);
1700 }
1701 #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */
1702
1703 SCM_DEFINE (scm_truncate_file, "truncate-file", 1, 1, 0,
1704 (SCM object, SCM length),
1705 "Truncate @var{file} to @var{length} bytes. @var{file} can be a\n"
1706 "filename string, a port object, or an integer file descriptor.\n"
1707 "The return value is unspecified.\n"
1708 "\n"
1709 "For a port or file descriptor @var{length} can be omitted, in\n"
1710 "which case the file is truncated at the current position (per\n"
1711 "@code{ftell} above).\n"
1712 "\n"
1713 "On most systems a file can be extended by giving a length\n"
1714 "greater than the current size, but this is not mandatory in the\n"
1715 "POSIX standard.")
1716 #define FUNC_NAME s_scm_truncate_file
1717 {
1718 int rv;
1719
1720 /* "object" can be a port, fdes or filename.
1721
1722 Negative "length" makes no sense, but it's left to truncate() or
1723 ftruncate() to give back an error for that (normally EINVAL).
1724 */
1725
1726 if (SCM_UNBNDP (length))
1727 {
1728 /* must supply length if object is a filename. */
1729 if (scm_is_string (object))
1730 SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL);
1731
1732 length = scm_seek (object, SCM_INUM0, scm_from_int (SEEK_CUR));
1733 }
1734
1735 object = SCM_COERCE_OUTPORT (object);
1736 if (scm_is_integer (object))
1737 {
1738 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1739 SCM_SYSCALL (rv = ftruncate_or_ftruncate64 (scm_to_int (object),
1740 c_length));
1741 }
1742 else if (SCM_OPOUTPORTP (object))
1743 {
1744 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1745 scm_t_port *pt = SCM_PTAB_ENTRY (object);
1746 scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (object);
1747
1748 if (!ptob->truncate)
1749 SCM_MISC_ERROR ("port is not truncatable", SCM_EOL);
1750 if (pt->rw_active == SCM_PORT_READ)
1751 scm_end_input (object);
1752 else if (pt->rw_active == SCM_PORT_WRITE)
1753 ptob->flush (object);
1754
1755 ptob->truncate (object, c_length);
1756 rv = 0;
1757 }
1758 else
1759 {
1760 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1761 char *str = scm_to_locale_string (object);
1762 int eno;
1763 SCM_SYSCALL (rv = truncate_or_truncate64 (str, c_length));
1764 eno = errno;
1765 free (str);
1766 errno = eno;
1767 }
1768 if (rv == -1)
1769 SCM_SYSERROR;
1770 return SCM_UNSPECIFIED;
1771 }
1772 #undef FUNC_NAME
1773
1774 SCM_DEFINE (scm_port_line, "port-line", 1, 0, 0,
1775 (SCM port),
1776 "Return the current line number for @var{port}.\n"
1777 "\n"
1778 "The first line of a file is 0. But you might want to add 1\n"
1779 "when printing line numbers, since starting from 1 is\n"
1780 "traditional in error messages, and likely to be more natural to\n"
1781 "non-programmers.")
1782 #define FUNC_NAME s_scm_port_line
1783 {
1784 port = SCM_COERCE_OUTPORT (port);
1785 SCM_VALIDATE_OPENPORT (1, port);
1786 return scm_from_long (SCM_LINUM (port));
1787 }
1788 #undef FUNC_NAME
1789
1790 SCM_DEFINE (scm_set_port_line_x, "set-port-line!", 2, 0, 0,
1791 (SCM port, SCM line),
1792 "Set the current line number for @var{port} to @var{line}. The\n"
1793 "first line of a file is 0.")
1794 #define FUNC_NAME s_scm_set_port_line_x
1795 {
1796 port = SCM_COERCE_OUTPORT (port);
1797 SCM_VALIDATE_OPENPORT (1, port);
1798 SCM_PTAB_ENTRY (port)->line_number = scm_to_long (line);
1799 return SCM_UNSPECIFIED;
1800 }
1801 #undef FUNC_NAME
1802
1803 SCM_DEFINE (scm_port_column, "port-column", 1, 0, 0,
1804 (SCM port),
1805 "Return the current column number of @var{port}.\n"
1806 "If the number is\n"
1807 "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n"
1808 "- i.e. the first character of the first line is line 0, column 0.\n"
1809 "(However, when you display a file position, for example in an error\n"
1810 "message, we recommend you add 1 to get 1-origin integers. This is\n"
1811 "because lines and column numbers traditionally start with 1, and that is\n"
1812 "what non-programmers will find most natural.)")
1813 #define FUNC_NAME s_scm_port_column
1814 {
1815 port = SCM_COERCE_OUTPORT (port);
1816 SCM_VALIDATE_OPENPORT (1, port);
1817 return scm_from_int (SCM_COL (port));
1818 }
1819 #undef FUNC_NAME
1820
1821 SCM_DEFINE (scm_set_port_column_x, "set-port-column!", 2, 0, 0,
1822 (SCM port, SCM column),
1823 "Set the current column of @var{port}. Before reading the first\n"
1824 "character on a line the column should be 0.")
1825 #define FUNC_NAME s_scm_set_port_column_x
1826 {
1827 port = SCM_COERCE_OUTPORT (port);
1828 SCM_VALIDATE_OPENPORT (1, port);
1829 SCM_PTAB_ENTRY (port)->column_number = scm_to_int (column);
1830 return SCM_UNSPECIFIED;
1831 }
1832 #undef FUNC_NAME
1833
1834 SCM_DEFINE (scm_port_filename, "port-filename", 1, 0, 0,
1835 (SCM port),
1836 "Return the filename associated with @var{port}. This function returns\n"
1837 "the strings \"standard input\", \"standard output\" and \"standard error\"\n"
1838 "when called on the current input, output and error ports respectively.")
1839 #define FUNC_NAME s_scm_port_filename
1840 {
1841 port = SCM_COERCE_OUTPORT (port);
1842 SCM_VALIDATE_OPENPORT (1, port);
1843 return SCM_FILENAME (port);
1844 }
1845 #undef FUNC_NAME
1846
1847 SCM_DEFINE (scm_set_port_filename_x, "set-port-filename!", 2, 0, 0,
1848 (SCM port, SCM filename),
1849 "Change the filename associated with @var{port}, using the current input\n"
1850 "port if none is specified. Note that this does not change the port's\n"
1851 "source of data, but only the value that is returned by\n"
1852 "@code{port-filename} and reported in diagnostic output.")
1853 #define FUNC_NAME s_scm_set_port_filename_x
1854 {
1855 port = SCM_COERCE_OUTPORT (port);
1856 SCM_VALIDATE_OPENPORT (1, port);
1857 /* We allow the user to set the filename to whatever he likes. */
1858 SCM_SET_FILENAME (port, filename);
1859 return SCM_UNSPECIFIED;
1860 }
1861 #undef FUNC_NAME
1862
1863 /* The default port encoding for this locale. New ports will have this
1864 encoding. If it is a string, that is the encoding. If it #f, it
1865 is in the native (Latin-1) encoding. */
1866 SCM_GLOBAL_VARIABLE (scm_port_encoding_var, "%port-encoding");
1867 static int scm_port_encoding_init = 0;
1868
1869 /* Return a C string representation of the current encoding. */
1870 const char *
1871 scm_i_get_port_encoding (SCM port)
1872 {
1873 SCM encoding;
1874
1875 if (scm_is_false (port))
1876 {
1877 if (!scm_port_encoding_init)
1878 return NULL;
1879 else if (!scm_is_fluid (SCM_VARIABLE_REF (scm_port_encoding_var)))
1880 return NULL;
1881 else
1882 {
1883 encoding = scm_fluid_ref (SCM_VARIABLE_REF (scm_port_encoding_var));
1884 if (!scm_is_string (encoding))
1885 return NULL;
1886 else
1887 return scm_i_string_chars (encoding);
1888 }
1889 }
1890 else
1891 {
1892 scm_t_port *pt;
1893 pt = SCM_PTAB_ENTRY (port);
1894 if (pt->encoding)
1895 return pt->encoding;
1896 else
1897 return NULL;
1898 }
1899 }
1900
1901 /* Returns ENC is if is a recognized encoding. If it isn't, it tries
1902 to find an alias of ENC that is valid. Otherwise, it returns
1903 NULL. */
1904 static const char *
1905 find_valid_encoding (const char *enc)
1906 {
1907 int isvalid = 0;
1908 const char str[] = " ";
1909 scm_t_uint32 *u32;
1910 size_t u32len;
1911
1912 u32 = u32_conv_from_encoding (enc, iconveh_error, str, 1,
1913 NULL, NULL, &u32len);
1914 isvalid = (u32 != NULL);
1915 free (u32);
1916
1917 if (isvalid)
1918 return enc;
1919
1920 return NULL;
1921 }
1922
1923 void
1924 scm_i_set_port_encoding_x (SCM port, const char *enc)
1925 {
1926 const char *valid_enc;
1927 scm_t_port *pt;
1928
1929 /* Null is shorthand for the native, Latin-1 encoding. */
1930 if (enc == NULL)
1931 valid_enc = NULL;
1932 else
1933 {
1934 valid_enc = find_valid_encoding (enc);
1935 if (valid_enc == NULL)
1936 {
1937 SCM err;
1938 err = scm_from_locale_string (enc);
1939 scm_misc_error (NULL, "invalid or unknown character encoding ~s",
1940 scm_list_1 (err));
1941 }
1942 }
1943
1944 if (scm_is_false (port))
1945 {
1946 /* Set the default encoding for future ports. */
1947 if (!scm_port_encoding_init
1948 || !scm_is_fluid (SCM_VARIABLE_REF (scm_port_encoding_var)))
1949 scm_misc_error (NULL, "tried to set port encoding fluid before it is initialized",
1950 SCM_EOL);
1951
1952 if (valid_enc == NULL
1953 || !strcmp (valid_enc, "ASCII")
1954 || !strcmp (valid_enc, "ANSI_X3.4-1968")
1955 || !strcmp (valid_enc, "ISO-8859-1"))
1956 scm_fluid_set_x (SCM_VARIABLE_REF (scm_port_encoding_var), SCM_BOOL_F);
1957 else
1958 scm_fluid_set_x (SCM_VARIABLE_REF (scm_port_encoding_var),
1959 scm_from_locale_string (valid_enc));
1960 }
1961 else
1962 {
1963 /* Set the character encoding for this port. */
1964 pt = SCM_PTAB_ENTRY (port);
1965 if (pt->encoding)
1966 free (pt->encoding);
1967 if (valid_enc == NULL)
1968 pt->encoding = NULL;
1969 else
1970 pt->encoding = strdup (valid_enc);
1971 }
1972 }
1973
1974 SCM_DEFINE (scm_port_encoding, "port-encoding", 1, 0, 0,
1975 (SCM port),
1976 "Returns, as a string, the character encoding that @var{port}\n"
1977 "uses to interpret its input and output.\n")
1978 #define FUNC_NAME s_scm_port_encoding
1979 {
1980 scm_t_port *pt;
1981 const char *enc;
1982
1983 SCM_VALIDATE_PORT (1, port);
1984
1985 pt = SCM_PTAB_ENTRY (port);
1986 enc = scm_i_get_port_encoding (port);
1987 if (enc)
1988 return scm_from_locale_string (pt->encoding);
1989 else
1990 return scm_from_locale_string ("NONE");
1991 }
1992 #undef FUNC_NAME
1993
1994 SCM_DEFINE (scm_set_port_encoding_x, "set-port-encoding!", 2, 0, 0,
1995 (SCM port, SCM enc),
1996 "Sets the character encoding that will be used to interpret all\n"
1997 "port I/O. New ports are created with the encoding\n"
1998 "appropriate for the current locale if @code{setlocale} has \n"
1999 "been called or ISO-8859-1 otherwise\n"
2000 "and this procedure can be used to modify that encoding.\n")
2001
2002 #define FUNC_NAME s_scm_set_port_encoding_x
2003 {
2004 char *enc_str;
2005 const char *valid_enc_str;
2006
2007 SCM_VALIDATE_PORT (1, port);
2008 SCM_VALIDATE_STRING (2, enc);
2009
2010 enc_str = scm_to_locale_string (enc);
2011 valid_enc_str = find_valid_encoding (enc_str);
2012 if (valid_enc_str == NULL)
2013 {
2014 free (enc_str);
2015 scm_misc_error (FUNC_NAME, "invalid or unknown character encoding ~s",
2016 scm_list_1 (enc));
2017 }
2018 else
2019 {
2020 scm_i_set_port_encoding_x (port, valid_enc_str);
2021 free (enc_str);
2022 }
2023 return SCM_UNSPECIFIED;
2024 }
2025 #undef FUNC_NAME
2026
2027
2028 /* This determines how conversions handle unconvertible characters. */
2029 SCM_GLOBAL_VARIABLE (scm_conversion_strategy, "%port-conversion-strategy");
2030 static int scm_conversion_strategy_init = 0;
2031
2032 scm_t_string_failed_conversion_handler
2033 scm_i_get_conversion_strategy (SCM port)
2034 {
2035 SCM encoding;
2036
2037 if (scm_is_false (port))
2038 {
2039 if (!scm_conversion_strategy_init
2040 || !scm_is_fluid (SCM_VARIABLE_REF (scm_conversion_strategy)))
2041 return SCM_FAILED_CONVERSION_QUESTION_MARK;
2042 else
2043 {
2044 encoding = scm_fluid_ref (SCM_VARIABLE_REF (scm_conversion_strategy));
2045 if (scm_is_false (encoding))
2046 return SCM_FAILED_CONVERSION_QUESTION_MARK;
2047 else
2048 return (scm_t_string_failed_conversion_handler) scm_to_int (encoding);
2049 }
2050 }
2051 else
2052 {
2053 scm_t_port *pt;
2054 pt = SCM_PTAB_ENTRY (port);
2055 return pt->ilseq_handler;
2056 }
2057
2058 }
2059
2060 void
2061 scm_i_set_conversion_strategy_x (SCM port,
2062 scm_t_string_failed_conversion_handler handler)
2063 {
2064 SCM strategy;
2065 scm_t_port *pt;
2066
2067 strategy = scm_from_int ((int) handler);
2068
2069 if (scm_is_false (port))
2070 {
2071 /* Set the default encoding for future ports. */
2072 if (!scm_conversion_strategy
2073 || !scm_is_fluid (SCM_VARIABLE_REF (scm_conversion_strategy)))
2074 scm_misc_error (NULL, "tried to set conversion strategy fluid before it is initialized",
2075 SCM_EOL);
2076 scm_fluid_set_x (SCM_VARIABLE_REF (scm_conversion_strategy), strategy);
2077 }
2078 else
2079 {
2080 /* Set the character encoding for this port. */
2081 pt = SCM_PTAB_ENTRY (port);
2082 pt->ilseq_handler = handler;
2083 }
2084 }
2085
2086 SCM_DEFINE (scm_port_conversion_strategy, "port-conversion-strategy",
2087 1, 0, 0, (SCM port),
2088 "Returns the behavior of the port when handling a character that\n"
2089 "is not representable in the port's current encoding.\n"
2090 "It returns the symbol @code{error} if unrepresentable characters\n"
2091 "should cause exceptions, @code{substitute} if the port should\n"
2092 "try to replace unrepresentable characters with question marks or\n"
2093 "approximate characters, or @code{escape} if unrepresentable\n"
2094 "characters should be converted to string escapes.\n"
2095 "\n"
2096 "If @var{port} is @code{#f}, then the current default behavior\n"
2097 "will be returned. New ports will have this default behavior\n"
2098 "when they are created.\n")
2099 #define FUNC_NAME s_scm_port_conversion_strategy
2100 {
2101 scm_t_string_failed_conversion_handler h;
2102
2103 SCM_VALIDATE_OPPORT (1, port);
2104
2105 if (!scm_is_false (port))
2106 {
2107 SCM_VALIDATE_OPPORT (1, port);
2108 }
2109
2110 h = scm_i_get_conversion_strategy (port);
2111 if (h == SCM_FAILED_CONVERSION_ERROR)
2112 return scm_from_locale_symbol ("error");
2113 else if (h == SCM_FAILED_CONVERSION_QUESTION_MARK)
2114 return scm_from_locale_symbol ("substitute");
2115 else if (h == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
2116 return scm_from_locale_symbol ("escape");
2117 else
2118 abort ();
2119
2120 /* Never gets here. */
2121 return SCM_UNDEFINED;
2122 }
2123 #undef FUNC_NAME
2124
2125 SCM_DEFINE (scm_set_port_conversion_strategy_x, "set-port-conversion-strategy!",
2126 2, 0, 0,
2127 (SCM port, SCM sym),
2128 "Sets the behavior of the interpreter when outputting a character\n"
2129 "that is not representable in the port's current encoding.\n"
2130 "@var{sym} can be either @code{'error}, @code{'substitute}, or\n"
2131 "@code{'escape}. If it is @code{'error}, an error will be thrown\n"
2132 "when an unconvertible character is encountered. If it is\n"
2133 "@code{'substitute}, then unconvertible characters will \n"
2134 "be replaced with approximate characters, or with question marks\n"
2135 "if no approximately correct character is available.\n"
2136 "If it is @code{'escape},\n"
2137 "it will appear as a hex escape when output.\n"
2138 "\n"
2139 "If @var{port} is an open port, the conversion error behavior\n"
2140 "is set for that port. If it is @code{#f}, it is set as the\n"
2141 "default behavior for any future ports that get created in\n"
2142 "this thread.\n")
2143 #define FUNC_NAME s_scm_set_port_conversion_strategy_x
2144 {
2145 SCM err;
2146 SCM qm;
2147 SCM esc;
2148
2149 if (!scm_is_false (port))
2150 {
2151 SCM_VALIDATE_OPPORT (1, port);
2152 }
2153
2154 err = scm_from_locale_symbol ("error");
2155 if (scm_is_true (scm_eqv_p (sym, err)))
2156 {
2157 scm_i_set_conversion_strategy_x (port, SCM_FAILED_CONVERSION_ERROR);
2158 return SCM_UNSPECIFIED;
2159 }
2160
2161 qm = scm_from_locale_symbol ("substitute");
2162 if (scm_is_true (scm_eqv_p (sym, qm)))
2163 {
2164 scm_i_set_conversion_strategy_x (port,
2165 SCM_FAILED_CONVERSION_QUESTION_MARK);
2166 return SCM_UNSPECIFIED;
2167 }
2168
2169 esc = scm_from_locale_symbol ("escape");
2170 if (scm_is_true (scm_eqv_p (sym, esc)))
2171 {
2172 scm_i_set_conversion_strategy_x (port,
2173 SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE);
2174 return SCM_UNSPECIFIED;
2175 }
2176
2177 SCM_MISC_ERROR ("unknown conversion behavior ~s", scm_list_1 (sym));
2178
2179 return SCM_UNSPECIFIED;
2180 }
2181 #undef FUNC_NAME
2182
2183
2184
2185 void
2186 scm_print_port_mode (SCM exp, SCM port)
2187 {
2188 scm_puts (SCM_CLOSEDP (exp)
2189 ? "closed: "
2190 : (SCM_RDNG & SCM_CELL_WORD_0 (exp)
2191 ? (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
2192 ? "input-output: "
2193 : "input: ")
2194 : (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
2195 ? "output: "
2196 : "bogus: ")),
2197 port);
2198 }
2199
2200 int
2201 scm_port_print (SCM exp, SCM port, scm_print_state *pstate SCM_UNUSED)
2202 {
2203 char *type = SCM_PTOBNAME (SCM_PTOBNUM (exp));
2204 if (!type)
2205 type = "port";
2206 scm_puts ("#<", port);
2207 scm_print_port_mode (exp, port);
2208 scm_puts (type, port);
2209 scm_putc (' ', port);
2210 scm_uintprint (SCM_CELL_WORD_1 (exp), 16, port);
2211 scm_putc ('>', port);
2212 return 1;
2213 }
2214
2215 void
2216 scm_ports_prehistory ()
2217 {
2218 scm_numptob = 0;
2219 scm_ptobs = (scm_t_ptob_descriptor *) scm_malloc (sizeof (scm_t_ptob_descriptor));
2220 }
2221
2222 \f
2223
2224 /* Void ports. */
2225
2226 scm_t_bits scm_tc16_void_port = 0;
2227
2228 static int fill_input_void_port (SCM port SCM_UNUSED)
2229 {
2230 return EOF;
2231 }
2232
2233 static void
2234 write_void_port (SCM port SCM_UNUSED,
2235 const void *data SCM_UNUSED,
2236 size_t size SCM_UNUSED)
2237 {
2238 }
2239
2240 static SCM
2241 scm_i_void_port (long mode_bits)
2242 {
2243 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
2244 {
2245 SCM answer = scm_new_port_table_entry (scm_tc16_void_port);
2246 scm_t_port * pt = SCM_PTAB_ENTRY(answer);
2247
2248 scm_port_non_buffer (pt);
2249
2250 SCM_SETSTREAM (answer, 0);
2251 SCM_SET_CELL_TYPE (answer, scm_tc16_void_port | mode_bits);
2252 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
2253 return answer;
2254 }
2255 }
2256
2257 SCM
2258 scm_void_port (char *mode_str)
2259 {
2260 return scm_i_void_port (scm_mode_bits (mode_str));
2261 }
2262
2263 SCM_DEFINE (scm_sys_make_void_port, "%make-void-port", 1, 0, 0,
2264 (SCM mode),
2265 "Create and return a new void port. A void port acts like\n"
2266 "@file{/dev/null}. The @var{mode} argument\n"
2267 "specifies the input/output modes for this port: see the\n"
2268 "documentation for @code{open-file} in @ref{File Ports}.")
2269 #define FUNC_NAME s_scm_sys_make_void_port
2270 {
2271 return scm_i_void_port (scm_i_mode_bits (mode));
2272 }
2273 #undef FUNC_NAME
2274
2275 \f
2276 /* Initialization. */
2277
2278 void
2279 scm_init_ports ()
2280 {
2281 /* lseek() symbols. */
2282 scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET));
2283 scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR));
2284 scm_c_define ("SEEK_END", scm_from_int (SEEK_END));
2285
2286 scm_tc16_void_port = scm_make_port_type ("void", fill_input_void_port,
2287 write_void_port);
2288
2289 cur_inport_fluid = scm_permanent_object (scm_make_fluid ());
2290 cur_outport_fluid = scm_permanent_object (scm_make_fluid ());
2291 cur_errport_fluid = scm_permanent_object (scm_make_fluid ());
2292 cur_loadport_fluid = scm_permanent_object (scm_make_fluid ());
2293
2294 scm_i_port_weak_hash = scm_permanent_object (scm_make_weak_key_hash_table (SCM_I_MAKINUM(31)));
2295 #include "libguile/ports.x"
2296
2297 SCM_VARIABLE_SET (scm_port_encoding_var, scm_make_fluid ());
2298 scm_fluid_set_x (SCM_VARIABLE_REF (scm_port_encoding_var), SCM_BOOL_F);
2299 scm_port_encoding_init = 1;
2300
2301 SCM_VARIABLE_SET (scm_conversion_strategy, scm_make_fluid ());
2302 scm_fluid_set_x (SCM_VARIABLE_REF (scm_conversion_strategy),
2303 scm_from_int ((int) SCM_FAILED_CONVERSION_QUESTION_MARK));
2304 scm_conversion_strategy_init = 1;
2305
2306 }
2307
2308 /*
2309 Local Variables:
2310 c-file-style: "gnu"
2311 End:
2312 */