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