Optimize `peek-char'.
[bpt/guile.git] / libguile / ports.c
1 /* Copyright (C) 1995,1996,1997,1998,1999,2000,2001, 2003, 2004, 2006, 2007, 2008, 2009, 2010 Free Software Foundation, Inc.
2 *
3 * This library is free software; you can redistribute it and/or
4 * modify it under the terms of the GNU Lesser General Public License
5 * as published by the Free Software Foundation; either version 3 of
6 * the License, or (at your option) any later version.
7 *
8 * This library is distributed in the hope that it will be useful, but
9 * WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * Lesser General Public License for more details.
12 *
13 * You should have received a copy of the GNU Lesser General Public
14 * License along with this library; if not, write to the Free Software
15 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
16 * 02110-1301 USA
17 */
18
19
20 \f
21 /* Headers. */
22
23 #define _LARGEFILE64_SOURCE /* ask for stat64 etc */
24
25 #ifdef HAVE_CONFIG_H
26 # include <config.h>
27 #endif
28
29 #include <stdio.h>
30 #include <errno.h>
31 #include <fcntl.h> /* for chsize on mingw */
32 #include <assert.h>
33 #include <uniconv.h>
34 #include <unistr.h>
35 #include <striconveh.h>
36
37 #include <assert.h>
38
39 #include "libguile/_scm.h"
40 #include "libguile/async.h"
41 #include "libguile/eval.h"
42 #include "libguile/fports.h" /* direct access for seek and truncate */
43 #include "libguile/goops.h"
44 #include "libguile/smob.h"
45 #include "libguile/chars.h"
46 #include "libguile/dynwind.h"
47
48 #include "libguile/keywords.h"
49 #include "libguile/hashtab.h"
50 #include "libguile/root.h"
51 #include "libguile/strings.h"
52 #include "libguile/mallocs.h"
53 #include "libguile/validate.h"
54 #include "libguile/ports.h"
55 #include "libguile/vectors.h"
56 #include "libguile/weaks.h"
57 #include "libguile/fluids.h"
58 #include "libguile/eq.h"
59
60 #ifdef HAVE_STRING_H
61 #include <string.h>
62 #endif
63
64 #ifdef HAVE_IO_H
65 #include <io.h>
66 #endif
67
68 #ifdef HAVE_UNISTD_H
69 #include <unistd.h>
70 #endif
71
72 #ifdef HAVE_SYS_IOCTL_H
73 #include <sys/ioctl.h>
74 #endif
75
76 /* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize
77 already, but have this code here in case that wasn't so in past versions,
78 or perhaps to help other minimal DOS environments.
79
80 gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which
81 might be possibilities if we've got other systems without ftruncate. */
82
83 #if defined HAVE_CHSIZE && ! defined HAVE_FTRUNCATE
84 #define ftruncate(fd, size) chsize (fd, size)
85 #undef HAVE_FTRUNCATE
86 #define HAVE_FTRUNCATE 1
87 #endif
88
89 \f
90 /* The port kind table --- a dynamically resized array of port types. */
91
92
93 /* scm_ptobs scm_numptob
94 * implement a dynamically resized array of ptob records.
95 * Indexes into this table are used when generating type
96 * tags for smobjects (if you know a tag you can get an index and conversely).
97 */
98 scm_t_ptob_descriptor *scm_ptobs = NULL;
99 long scm_numptob = 0;
100
101 /* GC marker for a port with stream of SCM type. */
102 SCM
103 scm_markstream (SCM ptr)
104 {
105 int openp;
106 openp = SCM_CELL_WORD_0 (ptr) & SCM_OPN;
107 if (openp)
108 return SCM_PACK (SCM_STREAM (ptr));
109 else
110 return SCM_BOOL_F;
111 }
112
113 /*
114 * We choose to use an interface similar to the smob interface with
115 * fill_input and write as standard fields, passed to the port
116 * type constructor, and optional fields set by setters.
117 */
118
119 static void
120 flush_port_default (SCM port SCM_UNUSED)
121 {
122 }
123
124 static void
125 end_input_default (SCM port SCM_UNUSED, int offset SCM_UNUSED)
126 {
127 }
128
129 scm_t_bits
130 scm_make_port_type (char *name,
131 int (*fill_input) (SCM port),
132 void (*write) (SCM port, const void *data, size_t size))
133 {
134 char *tmp;
135 if (SCM_I_MAX_PORT_TYPE_COUNT - 1 <= scm_numptob)
136 goto ptoberr;
137 SCM_CRITICAL_SECTION_START;
138 tmp = (char *) scm_gc_realloc ((char *) scm_ptobs,
139 scm_numptob * sizeof (scm_t_ptob_descriptor),
140 (1 + scm_numptob)
141 * sizeof (scm_t_ptob_descriptor),
142 "port-type");
143 if (tmp)
144 {
145 scm_ptobs = (scm_t_ptob_descriptor *) tmp;
146
147 scm_ptobs[scm_numptob].name = name;
148 scm_ptobs[scm_numptob].mark = 0;
149 scm_ptobs[scm_numptob].free = NULL;
150 scm_ptobs[scm_numptob].print = scm_port_print;
151 scm_ptobs[scm_numptob].equalp = 0;
152 scm_ptobs[scm_numptob].close = 0;
153
154 scm_ptobs[scm_numptob].write = write;
155 scm_ptobs[scm_numptob].flush = flush_port_default;
156
157 scm_ptobs[scm_numptob].end_input = end_input_default;
158 scm_ptobs[scm_numptob].fill_input = fill_input;
159 scm_ptobs[scm_numptob].input_waiting = 0;
160
161 scm_ptobs[scm_numptob].seek = 0;
162 scm_ptobs[scm_numptob].truncate = 0;
163
164 scm_numptob++;
165 }
166 SCM_CRITICAL_SECTION_END;
167 if (!tmp)
168 {
169 ptoberr:
170 scm_memory_error ("scm_make_port_type");
171 }
172 /* Make a class object if Goops is present */
173 if (SCM_UNPACK (scm_port_class[0]) != 0)
174 scm_make_port_classes (scm_numptob - 1, SCM_PTOBNAME (scm_numptob - 1));
175 return scm_tc7_port + (scm_numptob - 1) * 256;
176 }
177
178 void
179 scm_set_port_mark (scm_t_bits tc, SCM (*mark) (SCM))
180 {
181 scm_ptobs[SCM_TC2PTOBNUM (tc)].mark = mark;
182 }
183
184 void
185 scm_set_port_free (scm_t_bits tc, size_t (*free) (SCM))
186 {
187 scm_ptobs[SCM_TC2PTOBNUM (tc)].free = free;
188 }
189
190 void
191 scm_set_port_print (scm_t_bits tc, int (*print) (SCM exp, SCM port,
192 scm_print_state *pstate))
193 {
194 scm_ptobs[SCM_TC2PTOBNUM (tc)].print = print;
195 }
196
197 void
198 scm_set_port_equalp (scm_t_bits tc, SCM (*equalp) (SCM, SCM))
199 {
200 scm_ptobs[SCM_TC2PTOBNUM (tc)].equalp = equalp;
201 }
202
203 void
204 scm_set_port_flush (scm_t_bits tc, void (*flush) (SCM port))
205 {
206 scm_ptobs[SCM_TC2PTOBNUM (tc)].flush = flush;
207 }
208
209 void
210 scm_set_port_end_input (scm_t_bits tc, void (*end_input) (SCM port, int offset))
211 {
212 scm_ptobs[SCM_TC2PTOBNUM (tc)].end_input = end_input;
213 }
214
215 void
216 scm_set_port_close (scm_t_bits tc, int (*close) (SCM))
217 {
218 scm_ptobs[SCM_TC2PTOBNUM (tc)].close = close;
219 }
220
221 void
222 scm_set_port_seek (scm_t_bits tc,
223 scm_t_off (*seek) (SCM, scm_t_off, int))
224 {
225 scm_ptobs[SCM_TC2PTOBNUM (tc)].seek = seek;
226 }
227
228 void
229 scm_set_port_truncate (scm_t_bits tc, void (*truncate) (SCM, scm_t_off))
230 {
231 scm_ptobs[SCM_TC2PTOBNUM (tc)].truncate = truncate;
232 }
233
234 void
235 scm_set_port_input_waiting (scm_t_bits tc, int (*input_waiting) (SCM))
236 {
237 scm_ptobs[SCM_TC2PTOBNUM (tc)].input_waiting = input_waiting;
238 }
239
240 \f
241
242 SCM_DEFINE (scm_char_ready_p, "char-ready?", 0, 1, 0,
243 (SCM port),
244 "Return @code{#t} if a character is ready on input @var{port}\n"
245 "and return @code{#f} otherwise. If @code{char-ready?} returns\n"
246 "@code{#t} then the next @code{read-char} operation on\n"
247 "@var{port} is guaranteed not to hang. If @var{port} is a file\n"
248 "port at end of file then @code{char-ready?} returns @code{#t}.\n"
249 "\n"
250 "@code{char-ready?} exists to make it possible for a\n"
251 "program to accept characters from interactive ports without\n"
252 "getting stuck waiting for input. Any input editors associated\n"
253 "with such ports must make sure that characters whose existence\n"
254 "has been asserted by @code{char-ready?} cannot be rubbed out.\n"
255 "If @code{char-ready?} were to return @code{#f} at end of file,\n"
256 "a port at end of file would be indistinguishable from an\n"
257 "interactive port that has no ready characters.")
258 #define FUNC_NAME s_scm_char_ready_p
259 {
260 scm_t_port *pt;
261
262 if (SCM_UNBNDP (port))
263 port = scm_current_input_port ();
264 else
265 SCM_VALIDATE_OPINPORT (1, port);
266
267 pt = SCM_PTAB_ENTRY (port);
268
269 /* if the current read buffer is filled, or the
270 last pushed-back char has been read and the saved buffer is
271 filled, result is true. */
272 if (pt->read_pos < pt->read_end
273 || (pt->read_buf == pt->putback_buf
274 && pt->saved_read_pos < pt->saved_read_end))
275 return SCM_BOOL_T;
276 else
277 {
278 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
279
280 if (ptob->input_waiting)
281 return scm_from_bool(ptob->input_waiting (port));
282 else
283 return SCM_BOOL_T;
284 }
285 }
286 #undef FUNC_NAME
287
288 /* move up to read_len chars from port's putback and/or read buffers
289 into memory starting at dest. returns the number of chars moved. */
290 size_t scm_take_from_input_buffers (SCM port, char *dest, size_t read_len)
291 {
292 scm_t_port *pt = SCM_PTAB_ENTRY (port);
293 size_t chars_read = 0;
294 size_t from_buf = min (pt->read_end - pt->read_pos, read_len);
295
296 if (from_buf > 0)
297 {
298 memcpy (dest, pt->read_pos, from_buf);
299 pt->read_pos += from_buf;
300 chars_read += from_buf;
301 read_len -= from_buf;
302 dest += from_buf;
303 }
304
305 /* if putback was active, try the real input buffer too. */
306 if (pt->read_buf == pt->putback_buf)
307 {
308 from_buf = min (pt->saved_read_end - pt->saved_read_pos, read_len);
309 if (from_buf > 0)
310 {
311 memcpy (dest, pt->saved_read_pos, from_buf);
312 pt->saved_read_pos += from_buf;
313 chars_read += from_buf;
314 }
315 }
316 return chars_read;
317 }
318
319 /* Clear a port's read buffers, returning the contents. */
320 SCM_DEFINE (scm_drain_input, "drain-input", 1, 0, 0,
321 (SCM port),
322 "This procedure clears a port's input buffers, similar\n"
323 "to the way that force-output clears the output buffer. The\n"
324 "contents of the buffers are returned as a single string, e.g.,\n"
325 "\n"
326 "@lisp\n"
327 "(define p (open-input-file ...))\n"
328 "(drain-input p) => empty string, nothing buffered yet.\n"
329 "(unread-char (read-char p) p)\n"
330 "(drain-input p) => initial chars from p, up to the buffer size.\n"
331 "@end lisp\n\n"
332 "Draining the buffers may be useful for cleanly finishing\n"
333 "buffered I/O so that the file descriptor can be used directly\n"
334 "for further input.")
335 #define FUNC_NAME s_scm_drain_input
336 {
337 SCM result;
338 char *data;
339 scm_t_port *pt;
340 long count;
341
342 SCM_VALIDATE_OPINPORT (1, port);
343 pt = SCM_PTAB_ENTRY (port);
344
345 count = pt->read_end - pt->read_pos;
346 if (pt->read_buf == pt->putback_buf)
347 count += pt->saved_read_end - pt->saved_read_pos;
348
349 result = scm_i_make_string (count, &data);
350 scm_take_from_input_buffers (port, data, count);
351 return result;
352 }
353 #undef FUNC_NAME
354
355 \f
356 /* Standard ports --- current input, output, error, and more(!). */
357
358 static SCM cur_inport_fluid = 0;
359 static SCM cur_outport_fluid = 0;
360 static SCM cur_errport_fluid = 0;
361 static SCM cur_loadport_fluid = 0;
362
363 SCM_DEFINE (scm_current_input_port, "current-input-port", 0, 0, 0,
364 (),
365 "Return the current input port. This is the default port used\n"
366 "by many input procedures. Initially, @code{current-input-port}\n"
367 "returns the @dfn{standard input} in Unix and C terminology.")
368 #define FUNC_NAME s_scm_current_input_port
369 {
370 if (cur_inport_fluid)
371 return scm_fluid_ref (cur_inport_fluid);
372 else
373 return SCM_BOOL_F;
374 }
375 #undef FUNC_NAME
376
377 SCM_DEFINE (scm_current_output_port, "current-output-port", 0, 0, 0,
378 (),
379 "Return the current output port. This is the default port used\n"
380 "by many output procedures. Initially,\n"
381 "@code{current-output-port} returns the @dfn{standard output} in\n"
382 "Unix and C terminology.")
383 #define FUNC_NAME s_scm_current_output_port
384 {
385 if (cur_outport_fluid)
386 return scm_fluid_ref (cur_outport_fluid);
387 else
388 return SCM_BOOL_F;
389 }
390 #undef FUNC_NAME
391
392 SCM_DEFINE (scm_current_error_port, "current-error-port", 0, 0, 0,
393 (),
394 "Return the port to which errors and warnings should be sent (the\n"
395 "@dfn{standard error} in Unix and C terminology).")
396 #define FUNC_NAME s_scm_current_error_port
397 {
398 if (cur_errport_fluid)
399 return scm_fluid_ref (cur_errport_fluid);
400 else
401 return SCM_BOOL_F;
402 }
403 #undef FUNC_NAME
404
405 SCM_DEFINE (scm_current_load_port, "current-load-port", 0, 0, 0,
406 (),
407 "Return the current-load-port.\n"
408 "The load port is used internally by @code{primitive-load}.")
409 #define FUNC_NAME s_scm_current_load_port
410 {
411 return scm_fluid_ref (cur_loadport_fluid);
412 }
413 #undef FUNC_NAME
414
415 SCM_DEFINE (scm_set_current_input_port, "set-current-input-port", 1, 0, 0,
416 (SCM port),
417 "@deffnx {Scheme Procedure} set-current-output-port port\n"
418 "@deffnx {Scheme Procedure} set-current-error-port port\n"
419 "Change the ports returned by @code{current-input-port},\n"
420 "@code{current-output-port} and @code{current-error-port}, respectively,\n"
421 "so that they use the supplied @var{port} for input or output.")
422 #define FUNC_NAME s_scm_set_current_input_port
423 {
424 SCM oinp = scm_fluid_ref (cur_inport_fluid);
425 SCM_VALIDATE_OPINPORT (1, port);
426 scm_fluid_set_x (cur_inport_fluid, port);
427 return oinp;
428 }
429 #undef FUNC_NAME
430
431
432 SCM_DEFINE (scm_set_current_output_port, "set-current-output-port", 1, 0, 0,
433 (SCM port),
434 "Set the current default output port to @var{port}.")
435 #define FUNC_NAME s_scm_set_current_output_port
436 {
437 SCM ooutp = scm_fluid_ref (cur_outport_fluid);
438 port = SCM_COERCE_OUTPORT (port);
439 SCM_VALIDATE_OPOUTPORT (1, port);
440 scm_fluid_set_x (cur_outport_fluid, port);
441 return ooutp;
442 }
443 #undef FUNC_NAME
444
445
446 SCM_DEFINE (scm_set_current_error_port, "set-current-error-port", 1, 0, 0,
447 (SCM port),
448 "Set the current default error port to @var{port}.")
449 #define FUNC_NAME s_scm_set_current_error_port
450 {
451 SCM oerrp = scm_fluid_ref (cur_errport_fluid);
452 port = SCM_COERCE_OUTPORT (port);
453 SCM_VALIDATE_OPOUTPORT (1, port);
454 scm_fluid_set_x (cur_errport_fluid, port);
455 return oerrp;
456 }
457 #undef FUNC_NAME
458
459 void
460 scm_dynwind_current_input_port (SCM port)
461 #define FUNC_NAME NULL
462 {
463 SCM_VALIDATE_OPINPORT (1, port);
464 scm_dynwind_fluid (cur_inport_fluid, port);
465 }
466 #undef FUNC_NAME
467
468 void
469 scm_dynwind_current_output_port (SCM port)
470 #define FUNC_NAME NULL
471 {
472 port = SCM_COERCE_OUTPORT (port);
473 SCM_VALIDATE_OPOUTPORT (1, port);
474 scm_dynwind_fluid (cur_outport_fluid, port);
475 }
476 #undef FUNC_NAME
477
478 void
479 scm_dynwind_current_error_port (SCM port)
480 #define FUNC_NAME NULL
481 {
482 port = SCM_COERCE_OUTPORT (port);
483 SCM_VALIDATE_OPOUTPORT (1, port);
484 scm_dynwind_fluid (cur_errport_fluid, port);
485 }
486 #undef FUNC_NAME
487
488 void
489 scm_i_dynwind_current_load_port (SCM port)
490 {
491 scm_dynwind_fluid (cur_loadport_fluid, port);
492 }
493
494 \f
495 /* The port table --- an array of pointers to ports. */
496
497 /*
498 We need a global registry of ports to flush them all at exit, and to
499 get all the ports matching a file descriptor.
500 */
501 SCM scm_i_port_weak_hash;
502
503 scm_i_pthread_mutex_t scm_i_port_table_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER;
504
505 \f
506 /* Port finalization. */
507
508
509 static void finalize_port (GC_PTR, GC_PTR);
510
511 /* Register a finalizer for PORT, if needed by its port type. */
512 static SCM_C_INLINE_KEYWORD void
513 register_finalizer_for_port (SCM port)
514 {
515 long port_type;
516
517 port_type = SCM_TC2PTOBNUM (SCM_CELL_TYPE (port));
518 if (scm_ptobs[port_type].free)
519 {
520 GC_finalization_proc prev_finalizer;
521 GC_PTR prev_finalization_data;
522
523 GC_REGISTER_FINALIZER_NO_ORDER (SCM2PTR (port), finalize_port, 0,
524 &prev_finalizer,
525 &prev_finalization_data);
526 }
527 }
528
529 /* Finalize the object (a port) pointed to by PTR. */
530 static void
531 finalize_port (GC_PTR ptr, GC_PTR data)
532 {
533 long port_type;
534 SCM port = PTR2SCM (ptr);
535
536 if (!SCM_PORTP (port))
537 abort ();
538
539 if (SCM_OPENP (port))
540 {
541 if (SCM_REVEALED (port) > 0)
542 /* Keep "revealed" ports alive and re-register a finalizer. */
543 register_finalizer_for_port (port);
544 else
545 {
546 port_type = SCM_TC2PTOBNUM (SCM_CELL_TYPE (port));
547 if (port_type >= scm_numptob)
548 abort ();
549
550 if (scm_ptobs[port_type].free)
551 /* Yes, I really do mean `.free' rather than `.close'. `.close'
552 is for explicit `close-port' by user. */
553 scm_ptobs[port_type].free (port);
554
555 SCM_SETSTREAM (port, 0);
556 SCM_CLR_PORT_OPEN_FLAG (port);
557
558 scm_gc_ports_collected++;
559 }
560 }
561 }
562
563
564
565 \f
566
567 /* This function is not and should not be thread safe. */
568 SCM
569 scm_new_port_table_entry (scm_t_bits tag)
570 #define FUNC_NAME "scm_new_port_table_entry"
571 {
572 /*
573 We initialize the cell to empty, this is in case scm_gc_calloc
574 triggers GC ; we don't want the GC to scan a half-finished Z.
575 */
576
577 SCM z = scm_cons (SCM_EOL, SCM_EOL);
578 scm_t_port *entry = (scm_t_port *) scm_gc_calloc (sizeof (scm_t_port), "port");
579 const char *enc;
580
581 entry->file_name = SCM_BOOL_F;
582 entry->rw_active = SCM_PORT_NEITHER;
583 entry->port = z;
584 /* Initialize this port with the thread's current default
585 encoding. */
586 if ((enc = scm_i_get_port_encoding (SCM_BOOL_F)) == NULL)
587 entry->encoding = NULL;
588 else
589 entry->encoding = scm_gc_strdup (enc, "port");
590 entry->ilseq_handler = scm_i_get_conversion_strategy (SCM_BOOL_F);
591
592 SCM_SET_CELL_TYPE (z, tag);
593 SCM_SETPTAB_ENTRY (z, entry);
594
595 scm_hashq_set_x (scm_i_port_weak_hash, z, SCM_BOOL_F);
596
597 /* For each new port, register a finalizer so that it port type's free
598 function can be invoked eventually. */
599 register_finalizer_for_port (z);
600
601 return z;
602 }
603 #undef FUNC_NAME
604
605 #if SCM_ENABLE_DEPRECATED==1
606 SCM_API scm_t_port *
607 scm_add_to_port_table (SCM port)
608 {
609 SCM z = scm_new_port_table_entry (scm_tc7_port);
610 scm_t_port * pt = SCM_PTAB_ENTRY(z);
611
612 pt->port = port;
613 SCM_SETCAR (z, SCM_EOL);
614 SCM_SETCDR (z, SCM_EOL);
615 SCM_SETPTAB_ENTRY (port, pt);
616 return pt;
617 }
618 #endif
619
620
621 /* Remove a port from the table and destroy it. */
622
623 /* This function is not and should not be thread safe. */
624 void
625 scm_i_remove_port (SCM port)
626 #define FUNC_NAME "scm_remove_port"
627 {
628 scm_t_port *p = SCM_PTAB_ENTRY (port);
629
630 scm_port_non_buffer (p);
631
632 p->putback_buf = NULL;
633 p->putback_buf_size = 0;
634
635 SCM_SETPTAB_ENTRY (port, 0);
636 scm_hashq_remove_x (scm_i_port_weak_hash, port);
637 }
638 #undef FUNC_NAME
639
640
641 /* Functions for debugging. */
642 #ifdef GUILE_DEBUG
643 SCM_DEFINE (scm_pt_size, "pt-size", 0, 0, 0,
644 (),
645 "Return the number of ports in the port table. @code{pt-size}\n"
646 "is only included in @code{--enable-guile-debug} builds.")
647 #define FUNC_NAME s_scm_pt_size
648 {
649 return scm_from_int (SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash));
650 }
651 #undef FUNC_NAME
652 #endif
653
654 void
655 scm_port_non_buffer (scm_t_port *pt)
656 {
657 pt->read_pos = pt->read_buf = pt->read_end = &pt->shortbuf;
658 pt->write_buf = pt->write_pos = &pt->shortbuf;
659 pt->read_buf_size = pt->write_buf_size = 1;
660 pt->write_end = pt->write_buf + pt->write_buf_size;
661 }
662
663 \f
664 /* Revealed counts --- an oddity inherited from SCSH. */
665
666 /* Find a port in the table and return its revealed count.
667 Also used by the garbage collector.
668 */
669
670 int
671 scm_revealed_count (SCM port)
672 {
673 return SCM_REVEALED(port);
674 }
675
676
677
678 /* Return the revealed count for a port. */
679
680 SCM_DEFINE (scm_port_revealed, "port-revealed", 1, 0, 0,
681 (SCM port),
682 "Return the revealed count for @var{port}.")
683 #define FUNC_NAME s_scm_port_revealed
684 {
685 port = SCM_COERCE_OUTPORT (port);
686 SCM_VALIDATE_OPENPORT (1, port);
687 return scm_from_int (scm_revealed_count (port));
688 }
689 #undef FUNC_NAME
690
691 /* Set the revealed count for a port. */
692 SCM_DEFINE (scm_set_port_revealed_x, "set-port-revealed!", 2, 0, 0,
693 (SCM port, SCM rcount),
694 "Sets the revealed count for a port to a given value.\n"
695 "The return value is unspecified.")
696 #define FUNC_NAME s_scm_set_port_revealed_x
697 {
698 port = SCM_COERCE_OUTPORT (port);
699 SCM_VALIDATE_OPENPORT (1, port);
700 SCM_REVEALED (port) = scm_to_int (rcount);
701 return SCM_UNSPECIFIED;
702 }
703 #undef FUNC_NAME
704
705
706 \f
707 /* Retrieving a port's mode. */
708
709 /* Return the flags that characterize a port based on the mode
710 * string used to open a file for that port.
711 *
712 * See PORT FLAGS in scm.h
713 */
714
715 static long
716 scm_i_mode_bits_n (SCM modes)
717 {
718 return (SCM_OPN
719 | (scm_i_string_contains_char (modes, 'r')
720 || scm_i_string_contains_char (modes, '+') ? SCM_RDNG : 0)
721 | (scm_i_string_contains_char (modes, 'w')
722 || scm_i_string_contains_char (modes, 'a')
723 || scm_i_string_contains_char (modes, '+') ? SCM_WRTNG : 0)
724 | (scm_i_string_contains_char (modes, '0') ? SCM_BUF0 : 0)
725 | (scm_i_string_contains_char (modes, 'l') ? SCM_BUFLINE : 0));
726 }
727
728 long
729 scm_mode_bits (char *modes)
730 {
731 return scm_i_mode_bits (scm_from_locale_string (modes));
732 }
733
734 long
735 scm_i_mode_bits (SCM modes)
736 {
737 long bits;
738
739 if (!scm_is_string (modes))
740 scm_wrong_type_arg_msg (NULL, 0, modes, "string");
741
742 bits = scm_i_mode_bits_n (modes);
743 scm_remember_upto_here_1 (modes);
744 return bits;
745 }
746
747 /* Return the mode flags from an open port.
748 * Some modes such as "append" are only used when opening
749 * a file and are not returned here. */
750
751 SCM_DEFINE (scm_port_mode, "port-mode", 1, 0, 0,
752 (SCM port),
753 "Return the port modes associated with the open port @var{port}.\n"
754 "These will not necessarily be identical to the modes used when\n"
755 "the port was opened, since modes such as \"append\" which are\n"
756 "used only during port creation are not retained.")
757 #define FUNC_NAME s_scm_port_mode
758 {
759 char modes[4];
760 modes[0] = '\0';
761
762 port = SCM_COERCE_OUTPORT (port);
763 SCM_VALIDATE_OPPORT (1, port);
764 if (SCM_CELL_WORD_0 (port) & SCM_RDNG) {
765 if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
766 strcpy (modes, "r+");
767 else
768 strcpy (modes, "r");
769 }
770 else if (SCM_CELL_WORD_0 (port) & SCM_WRTNG)
771 strcpy (modes, "w");
772 if (SCM_CELL_WORD_0 (port) & SCM_BUF0)
773 strcat (modes, "0");
774 return scm_from_locale_string (modes);
775 }
776 #undef FUNC_NAME
777
778
779 \f
780 /* Closing ports. */
781
782 /* scm_close_port
783 * Call the close operation on a port object.
784 * see also scm_close.
785 */
786 SCM_DEFINE (scm_close_port, "close-port", 1, 0, 0,
787 (SCM port),
788 "Close the specified port object. Return @code{#t} if it\n"
789 "successfully closes a port or @code{#f} if it was already\n"
790 "closed. An exception may be raised if an error occurs, for\n"
791 "example when flushing buffered output. See also @ref{Ports and\n"
792 "File Descriptors, close}, for a procedure which can close file\n"
793 "descriptors.")
794 #define FUNC_NAME s_scm_close_port
795 {
796 size_t i;
797 int rv;
798
799 port = SCM_COERCE_OUTPORT (port);
800
801 SCM_VALIDATE_PORT (1, port);
802 if (SCM_CLOSEDP (port))
803 return SCM_BOOL_F;
804 i = SCM_PTOBNUM (port);
805 if (scm_ptobs[i].close)
806 rv = (scm_ptobs[i].close) (port);
807 else
808 rv = 0;
809 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
810 scm_i_remove_port (port);
811 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
812 SCM_CLR_PORT_OPEN_FLAG (port);
813 return scm_from_bool (rv >= 0);
814 }
815 #undef FUNC_NAME
816
817 SCM_DEFINE (scm_close_input_port, "close-input-port", 1, 0, 0,
818 (SCM port),
819 "Close the specified input port object. The routine has no effect if\n"
820 "the file has already been closed. An exception may be raised if an\n"
821 "error occurs. The value returned is unspecified.\n\n"
822 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
823 "which can close file descriptors.")
824 #define FUNC_NAME s_scm_close_input_port
825 {
826 SCM_VALIDATE_INPUT_PORT (1, port);
827 scm_close_port (port);
828 return SCM_UNSPECIFIED;
829 }
830 #undef FUNC_NAME
831
832 SCM_DEFINE (scm_close_output_port, "close-output-port", 1, 0, 0,
833 (SCM port),
834 "Close the specified output port object. The routine has no effect if\n"
835 "the file has already been closed. An exception may be raised if an\n"
836 "error occurs. The value returned is unspecified.\n\n"
837 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
838 "which can close file descriptors.")
839 #define FUNC_NAME s_scm_close_output_port
840 {
841 port = SCM_COERCE_OUTPORT (port);
842 SCM_VALIDATE_OUTPUT_PORT (1, port);
843 scm_close_port (port);
844 return SCM_UNSPECIFIED;
845 }
846 #undef FUNC_NAME
847
848 static SCM
849 scm_i_collect_keys_in_vector (void *closure, SCM key, SCM value, SCM result)
850 {
851 int *i = (int*) closure;
852 scm_c_vector_set_x (result, *i, key);
853 (*i)++;
854
855 return result;
856 }
857
858 void
859 scm_c_port_for_each (void (*proc)(void *data, SCM p), void *data)
860 {
861 int i = 0;
862 size_t n;
863 SCM ports;
864
865 /* Even without pre-emptive multithreading, running arbitrary code
866 while scanning the port table is unsafe because the port table
867 can change arbitrarily (from a GC, for example). So we first
868 collect the ports into a vector. -mvo */
869
870 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
871 n = SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash);
872 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
873 ports = scm_c_make_vector (n, SCM_BOOL_F);
874
875 scm_i_pthread_mutex_lock (&scm_i_port_table_mutex);
876 ports = scm_internal_hash_fold (scm_i_collect_keys_in_vector, &i,
877 ports, scm_i_port_weak_hash);
878 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
879
880 for (i = 0; i < n; i++) {
881 SCM p = SCM_SIMPLE_VECTOR_REF (ports, i);
882 if (SCM_PORTP (p))
883 proc (data, p);
884 }
885
886 scm_remember_upto_here_1 (ports);
887 }
888
889 SCM_DEFINE (scm_port_for_each, "port-for-each", 1, 0, 0,
890 (SCM proc),
891 "Apply @var{proc} to each port in the Guile port table\n"
892 "in turn. The return value is unspecified. More specifically,\n"
893 "@var{proc} is applied exactly once to every port that exists\n"
894 "in the system at the time @var{port-for-each} is invoked.\n"
895 "Changes to the port table while @var{port-for-each} is running\n"
896 "have no effect as far as @var{port-for-each} is concerned.")
897 #define FUNC_NAME s_scm_port_for_each
898 {
899 SCM_VALIDATE_PROC (1, proc);
900
901 scm_c_port_for_each ((void (*)(void*,SCM))scm_call_1, proc);
902 return SCM_UNSPECIFIED;
903 }
904 #undef FUNC_NAME
905
906
907 \f
908 /* Utter miscellany. Gosh, we should clean this up some time. */
909
910 SCM_DEFINE (scm_input_port_p, "input-port?", 1, 0, 0,
911 (SCM x),
912 "Return @code{#t} if @var{x} is an input port, otherwise return\n"
913 "@code{#f}. Any object satisfying this predicate also satisfies\n"
914 "@code{port?}.")
915 #define FUNC_NAME s_scm_input_port_p
916 {
917 return scm_from_bool (SCM_INPUT_PORT_P (x));
918 }
919 #undef FUNC_NAME
920
921 SCM_DEFINE (scm_output_port_p, "output-port?", 1, 0, 0,
922 (SCM x),
923 "Return @code{#t} if @var{x} is an output port, otherwise return\n"
924 "@code{#f}. Any object satisfying this predicate also satisfies\n"
925 "@code{port?}.")
926 #define FUNC_NAME s_scm_output_port_p
927 {
928 x = SCM_COERCE_OUTPORT (x);
929 return scm_from_bool (SCM_OUTPUT_PORT_P (x));
930 }
931 #undef FUNC_NAME
932
933 SCM_DEFINE (scm_port_p, "port?", 1, 0, 0,
934 (SCM x),
935 "Return a boolean indicating whether @var{x} is a port.\n"
936 "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n"
937 "@var{x}))}.")
938 #define FUNC_NAME s_scm_port_p
939 {
940 return scm_from_bool (SCM_PORTP (x));
941 }
942 #undef FUNC_NAME
943
944 SCM_DEFINE (scm_port_closed_p, "port-closed?", 1, 0, 0,
945 (SCM port),
946 "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n"
947 "open.")
948 #define FUNC_NAME s_scm_port_closed_p
949 {
950 SCM_VALIDATE_PORT (1, port);
951 return scm_from_bool (!SCM_OPPORTP (port));
952 }
953 #undef FUNC_NAME
954
955 SCM_DEFINE (scm_eof_object_p, "eof-object?", 1, 0, 0,
956 (SCM x),
957 "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n"
958 "return @code{#f}.")
959 #define FUNC_NAME s_scm_eof_object_p
960 {
961 return scm_from_bool(SCM_EOF_OBJECT_P (x));
962 }
963 #undef FUNC_NAME
964
965 SCM_DEFINE (scm_force_output, "force-output", 0, 1, 0,
966 (SCM port),
967 "Flush the specified output port, or the current output port if @var{port}\n"
968 "is omitted. The current output buffer contents are passed to the\n"
969 "underlying port implementation (e.g., in the case of fports, the\n"
970 "data will be written to the file and the output buffer will be cleared.)\n"
971 "It has no effect on an unbuffered port.\n\n"
972 "The return value is unspecified.")
973 #define FUNC_NAME s_scm_force_output
974 {
975 if (SCM_UNBNDP (port))
976 port = scm_current_output_port ();
977 else
978 {
979 port = SCM_COERCE_OUTPORT (port);
980 SCM_VALIDATE_OPOUTPORT (1, port);
981 }
982 scm_flush (port);
983 return SCM_UNSPECIFIED;
984 }
985 #undef FUNC_NAME
986
987
988 static void
989 flush_output_port (void *closure, SCM port)
990 {
991 if (SCM_OPOUTPORTP (port))
992 scm_flush (port);
993 }
994
995 SCM_DEFINE (scm_flush_all_ports, "flush-all-ports", 0, 0, 0,
996 (),
997 "Equivalent to calling @code{force-output} on\n"
998 "all open output ports. The return value is unspecified.")
999 #define FUNC_NAME s_scm_flush_all_ports
1000 {
1001 scm_c_port_for_each (&flush_output_port, NULL);
1002 return SCM_UNSPECIFIED;
1003 }
1004 #undef FUNC_NAME
1005
1006 SCM_DEFINE (scm_read_char, "read-char", 0, 1, 0,
1007 (SCM port),
1008 "Return the next character available from @var{port}, updating\n"
1009 "@var{port} to point to the following character. If no more\n"
1010 "characters are available, the end-of-file object is returned.")
1011 #define FUNC_NAME s_scm_read_char
1012 {
1013 scm_t_wchar c;
1014 if (SCM_UNBNDP (port))
1015 port = scm_current_input_port ();
1016 SCM_VALIDATE_OPINPORT (1, port);
1017 c = scm_getc (port);
1018 if (EOF == c)
1019 return SCM_EOF_VAL;
1020 return SCM_MAKE_CHAR (c);
1021 }
1022 #undef FUNC_NAME
1023
1024 #define SCM_MBCHAR_BUF_SIZE (4)
1025
1026 /* Read a codepoint from PORT and return it. Fill BUF with the byte
1027 representation of the codepoint in PORT's encoding, and set *LEN to
1028 the length in bytes of that representation. Raise an error on
1029 failure. */
1030 static scm_t_wchar
1031 get_codepoint (SCM port, char buf[SCM_MBCHAR_BUF_SIZE], size_t *len)
1032 {
1033 int c;
1034 size_t bufcount = 0;
1035 scm_t_uint32 result_buf;
1036 scm_t_wchar codepoint = 0;
1037 scm_t_uint32 *u32;
1038 size_t u32len;
1039 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1040
1041 c = scm_get_byte_or_eof (port);
1042 if (c == EOF)
1043 return (scm_t_wchar) EOF;
1044
1045 buf[0] = c;
1046 bufcount++;
1047
1048 if (pt->encoding == NULL)
1049 {
1050 /* The encoding is Latin-1: bytes are characters. */
1051 codepoint = (unsigned char) buf[0];
1052 goto success;
1053 }
1054
1055 for (;;)
1056 {
1057 u32len = sizeof (result_buf) / sizeof (scm_t_uint32);
1058 u32 = u32_conv_from_encoding (pt->encoding,
1059 (enum iconv_ilseq_handler) pt->ilseq_handler,
1060 buf, bufcount, NULL, &result_buf, &u32len);
1061 if (u32 == NULL || u32len == 0)
1062 {
1063 if (errno == ENOMEM)
1064 scm_memory_error ("Input decoding");
1065
1066 /* Otherwise errno is EILSEQ or EINVAL, so perhaps more
1067 bytes are needed. Keep looping. */
1068 }
1069 else
1070 {
1071 /* Complete codepoint found. */
1072 codepoint = u32[0];
1073
1074 if (SCM_UNLIKELY (u32 != &result_buf))
1075 /* libunistring up to 0.9.3 (included) would always heap-allocate
1076 the result even when a large-enough RESULT_BUF is supplied, see
1077 <http://lists.gnu.org/archive/html/bug-libunistring/2010-07/msg00003.html>. */
1078 free (u32);
1079
1080 goto success;
1081 }
1082
1083 if (bufcount == SCM_MBCHAR_BUF_SIZE)
1084 {
1085 /* We've read several bytes and didn't find a good
1086 codepoint. Give up. */
1087 goto failure;
1088 }
1089
1090 c = scm_get_byte_or_eof (port);
1091
1092 if (c == EOF)
1093 {
1094 /* EOF before a complete character was read. Push it all
1095 back and return EOF. */
1096 while (bufcount > 0)
1097 {
1098 /* FIXME: this will probably cause errors in the port column. */
1099 scm_unget_byte (buf[bufcount-1], port);
1100 bufcount --;
1101 }
1102 return EOF;
1103 }
1104
1105 if (c == '\n')
1106 {
1107 /* It is always invalid to have EOL in the middle of a
1108 multibyte character. */
1109 scm_unget_byte ('\n', port);
1110 goto failure;
1111 }
1112
1113 buf[bufcount++] = c;
1114 }
1115
1116 success:
1117 switch (codepoint)
1118 {
1119 case '\a':
1120 break;
1121 case '\b':
1122 SCM_DECCOL (port);
1123 break;
1124 case '\n':
1125 SCM_INCLINE (port);
1126 break;
1127 case '\r':
1128 SCM_ZEROCOL (port);
1129 break;
1130 case '\t':
1131 SCM_TABCOL (port);
1132 break;
1133 default:
1134 SCM_INCCOL (port);
1135 break;
1136 }
1137
1138 *len = bufcount;
1139
1140 return codepoint;
1141
1142 failure:
1143 {
1144 char *err_buf;
1145 SCM err_str = scm_i_make_string (bufcount, &err_buf);
1146 memcpy (err_buf, buf, bufcount);
1147
1148 if (errno == EILSEQ)
1149 scm_misc_error (NULL, "input encoding error for ~s: ~s",
1150 scm_list_2 (scm_from_locale_string (scm_i_get_port_encoding (port)),
1151 err_str));
1152 else
1153 scm_misc_error (NULL, "input encoding error (invalid) for ~s: ~s\n",
1154 scm_list_2 (scm_from_locale_string (scm_i_get_port_encoding (port)),
1155 err_str));
1156 }
1157
1158 /* Never gets here. */
1159 return 0;
1160 }
1161
1162 /* Read a codepoint from PORT and return it. */
1163 scm_t_wchar
1164 scm_getc (SCM port)
1165 {
1166 size_t len;
1167 char buf[SCM_MBCHAR_BUF_SIZE];
1168
1169 return get_codepoint (port, buf, &len);
1170 }
1171
1172 /* this should only be called when the read buffer is empty. it
1173 tries to refill the read buffer. it returns the first char from
1174 the port, which is either EOF or *(pt->read_pos). */
1175 int
1176 scm_fill_input (SCM port)
1177 {
1178 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1179
1180 assert (pt->read_pos == pt->read_end);
1181
1182 if (pt->read_buf == pt->putback_buf)
1183 {
1184 /* finished reading put-back chars. */
1185 pt->read_buf = pt->saved_read_buf;
1186 pt->read_pos = pt->saved_read_pos;
1187 pt->read_end = pt->saved_read_end;
1188 pt->read_buf_size = pt->saved_read_buf_size;
1189 if (pt->read_pos < pt->read_end)
1190 return *(pt->read_pos);
1191 }
1192 return scm_ptobs[SCM_PTOBNUM (port)].fill_input (port);
1193 }
1194
1195
1196 /* scm_lfwrite
1197 *
1198 * This function differs from scm_c_write; it updates port line and
1199 * column. */
1200
1201 static void
1202 update_port_lf (scm_t_wchar c, SCM port)
1203 {
1204 if (c == '\a')
1205 ; /* Do nothing. */
1206 else if (c == '\b')
1207 SCM_DECCOL (port);
1208 else if (c == '\n')
1209 SCM_INCLINE (port);
1210 else if (c == '\r')
1211 SCM_ZEROCOL (port);
1212 else if (c == '\t')
1213 SCM_TABCOL (port);
1214 else
1215 SCM_INCCOL (port);
1216 }
1217
1218 void
1219 scm_lfwrite (const char *ptr, size_t size, SCM port)
1220 {
1221 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1222 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1223
1224 if (pt->rw_active == SCM_PORT_READ)
1225 scm_end_input (port);
1226
1227 ptob->write (port, ptr, size);
1228
1229 for (; size; ptr++, size--)
1230 update_port_lf ((scm_t_wchar) (unsigned char) *ptr, port);
1231
1232 if (pt->rw_random)
1233 pt->rw_active = SCM_PORT_WRITE;
1234 }
1235
1236 /* Write a scheme string STR to PORT from START inclusive to END
1237 exclusive. */
1238 void
1239 scm_lfwrite_substr (SCM str, size_t start, size_t end, SCM port)
1240 {
1241 size_t i, size = scm_i_string_length (str);
1242 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1243 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1244 scm_t_wchar p;
1245 char *buf;
1246 size_t len;
1247
1248 if (pt->rw_active == SCM_PORT_READ)
1249 scm_end_input (port);
1250
1251 if (end == (size_t) (-1))
1252 end = size;
1253 size = end - start;
1254
1255 /* Note that making a substring will likely take the
1256 stringbuf_write_mutex. So, one shouldn't use scm_lfwrite_substr
1257 if the stringbuf write mutex may still be held elsewhere. */
1258 buf = scm_to_stringn (scm_c_substring (str, start, end), &len,
1259 pt->encoding, pt->ilseq_handler);
1260 ptob->write (port, buf, len);
1261 free (buf);
1262
1263 for (i = 0; i < size; i++)
1264 {
1265 p = scm_i_string_ref (str, i + start);
1266 update_port_lf (p, port);
1267 }
1268
1269 if (pt->rw_random)
1270 pt->rw_active = SCM_PORT_WRITE;
1271 }
1272
1273 /* Write a scheme string STR to PORT. */
1274 void
1275 scm_lfwrite_str (SCM str, SCM port)
1276 {
1277 size_t i, size = scm_i_string_length (str);
1278 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1279 scm_t_ptob_descriptor *ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1280 scm_t_wchar p;
1281 char *buf;
1282 size_t len;
1283
1284 if (pt->rw_active == SCM_PORT_READ)
1285 scm_end_input (port);
1286
1287 buf = scm_to_stringn (str, &len,
1288 pt->encoding, pt->ilseq_handler);
1289 ptob->write (port, buf, len);
1290 free (buf);
1291
1292 for (i = 0; i < size; i++)
1293 {
1294 p = scm_i_string_ref (str, i);
1295 update_port_lf (p, port);
1296 }
1297
1298 if (pt->rw_random)
1299 pt->rw_active = SCM_PORT_WRITE;
1300 }
1301
1302 /* scm_c_read
1303 *
1304 * Used by an application to read arbitrary number of bytes from an
1305 * SCM port. Same semantics as libc read, except that scm_c_read only
1306 * returns less than SIZE bytes if at end-of-file.
1307 *
1308 * Warning: Doesn't update port line and column counts! */
1309
1310 /* This structure, and the following swap_buffer function, are used
1311 for temporarily swapping a port's own read buffer, and the buffer
1312 that the caller of scm_c_read provides. */
1313 struct port_and_swap_buffer
1314 {
1315 scm_t_port *pt;
1316 unsigned char *buffer;
1317 size_t size;
1318 };
1319
1320 static void
1321 swap_buffer (void *data)
1322 {
1323 struct port_and_swap_buffer *psb = (struct port_and_swap_buffer *) data;
1324 unsigned char *old_buf = psb->pt->read_buf;
1325 size_t old_size = psb->pt->read_buf_size;
1326
1327 /* Make the port use (buffer, size) from the struct. */
1328 psb->pt->read_pos = psb->pt->read_buf = psb->pt->read_end = psb->buffer;
1329 psb->pt->read_buf_size = psb->size;
1330
1331 /* Save the port's old (buffer, size) in the struct. */
1332 psb->buffer = old_buf;
1333 psb->size = old_size;
1334 }
1335
1336 size_t
1337 scm_c_read (SCM port, void *buffer, size_t size)
1338 #define FUNC_NAME "scm_c_read"
1339 {
1340 scm_t_port *pt;
1341 size_t n_read = 0, n_available;
1342 struct port_and_swap_buffer psb;
1343
1344 SCM_VALIDATE_OPINPORT (1, port);
1345
1346 pt = SCM_PTAB_ENTRY (port);
1347 if (pt->rw_active == SCM_PORT_WRITE)
1348 scm_ptobs[SCM_PTOBNUM (port)].flush (port);
1349
1350 if (pt->rw_random)
1351 pt->rw_active = SCM_PORT_READ;
1352
1353 /* Take bytes first from the port's read buffer. */
1354 if (pt->read_pos < pt->read_end)
1355 {
1356 n_available = min (size, pt->read_end - pt->read_pos);
1357 memcpy (buffer, pt->read_pos, n_available);
1358 buffer = (char *) buffer + n_available;
1359 pt->read_pos += n_available;
1360 n_read += n_available;
1361 size -= n_available;
1362 }
1363
1364 /* Avoid the scm_dynwind_* costs if we now have enough data. */
1365 if (size == 0)
1366 return n_read;
1367
1368 /* Now we will call scm_fill_input repeatedly until we have read the
1369 requested number of bytes. (Note that a single scm_fill_input
1370 call does not guarantee to fill the whole of the port's read
1371 buffer.) */
1372 if (pt->read_buf_size <= 1 && pt->encoding == NULL)
1373 {
1374 /* The port that we are reading from is unbuffered - i.e. does
1375 not have its own persistent buffer - but we have a buffer,
1376 provided by our caller, that is the right size for the data
1377 that is wanted. For the following scm_fill_input calls,
1378 therefore, we use the buffer in hand as the port's read
1379 buffer.
1380
1381 We need to make sure that the port's normal (1 byte) buffer
1382 is reinstated in case one of the scm_fill_input () calls
1383 throws an exception; we use the scm_dynwind_* API to achieve
1384 that.
1385
1386 A consequence of this optimization is that the fill_input
1387 functions can't unget characters. That'll push data to the
1388 pushback buffer instead of this psb buffer. */
1389 #if SCM_DEBUG == 1
1390 unsigned char *pback = pt->putback_buf;
1391 #endif
1392 psb.pt = pt;
1393 psb.buffer = buffer;
1394 psb.size = size;
1395 scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE);
1396 scm_dynwind_rewind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY);
1397 scm_dynwind_unwind_handler (swap_buffer, &psb, SCM_F_WIND_EXPLICITLY);
1398
1399 /* Call scm_fill_input until we have all the bytes that we need,
1400 or we hit EOF. */
1401 while (pt->read_buf_size && (scm_fill_input (port) != EOF))
1402 {
1403 pt->read_buf_size -= (pt->read_end - pt->read_pos);
1404 pt->read_pos = pt->read_buf = pt->read_end;
1405 }
1406 #if SCM_DEBUG == 1
1407 if (pback != pt->putback_buf
1408 || pt->read_buf - (unsigned char *) buffer < 0)
1409 scm_misc_error (FUNC_NAME,
1410 "scm_c_read must not call a fill function that pushes "
1411 "back characters onto an unbuffered port", SCM_EOL);
1412 #endif
1413 n_read += pt->read_buf - (unsigned char *) buffer;
1414
1415 /* Reinstate the port's normal buffer. */
1416 scm_dynwind_end ();
1417 }
1418 else
1419 {
1420 /* The port has its own buffer. It is important that we use it,
1421 even if it happens to be smaller than our caller's buffer, so
1422 that a custom port implementation's entry points (in
1423 particular, fill_input) can rely on the buffer always being
1424 the same as they first set up. */
1425 while (size && (scm_fill_input (port) != EOF))
1426 {
1427 n_available = min (size, pt->read_end - pt->read_pos);
1428 memcpy (buffer, pt->read_pos, n_available);
1429 buffer = (char *) buffer + n_available;
1430 pt->read_pos += n_available;
1431 n_read += n_available;
1432 size -= n_available;
1433 }
1434 }
1435
1436 return n_read;
1437 }
1438 #undef FUNC_NAME
1439
1440 /* scm_c_write
1441 *
1442 * Used by an application to write arbitrary number of bytes to an SCM
1443 * port. Similar semantics as libc write. However, unlike libc
1444 * write, scm_c_write writes the requested number of bytes and has no
1445 * return value.
1446 *
1447 * Warning: Doesn't update port line and column counts!
1448 */
1449
1450 void
1451 scm_c_write (SCM port, const void *ptr, size_t size)
1452 #define FUNC_NAME "scm_c_write"
1453 {
1454 scm_t_port *pt;
1455 scm_t_ptob_descriptor *ptob;
1456
1457 SCM_VALIDATE_OPOUTPORT (1, port);
1458
1459 pt = SCM_PTAB_ENTRY (port);
1460 ptob = &scm_ptobs[SCM_PTOBNUM (port)];
1461
1462 if (pt->rw_active == SCM_PORT_READ)
1463 scm_end_input (port);
1464
1465 ptob->write (port, ptr, size);
1466
1467 if (pt->rw_random)
1468 pt->rw_active = SCM_PORT_WRITE;
1469 }
1470 #undef FUNC_NAME
1471
1472 void
1473 scm_flush (SCM port)
1474 {
1475 long i = SCM_PTOBNUM (port);
1476 assert (i >= 0);
1477 (scm_ptobs[i].flush) (port);
1478 }
1479
1480 void
1481 scm_end_input (SCM port)
1482 {
1483 long offset;
1484 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1485
1486 if (pt->read_buf == pt->putback_buf)
1487 {
1488 offset = pt->read_end - pt->read_pos;
1489 pt->read_buf = pt->saved_read_buf;
1490 pt->read_pos = pt->saved_read_pos;
1491 pt->read_end = pt->saved_read_end;
1492 pt->read_buf_size = pt->saved_read_buf_size;
1493 }
1494 else
1495 offset = 0;
1496
1497 scm_ptobs[SCM_PTOBNUM (port)].end_input (port, offset);
1498 }
1499
1500 \f
1501
1502
1503 void
1504 scm_unget_byte (int c, SCM port)
1505 #define FUNC_NAME "scm_unget_byte"
1506 {
1507 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1508
1509 if (pt->read_buf == pt->putback_buf)
1510 /* already using the put-back buffer. */
1511 {
1512 /* enlarge putback_buf if necessary. */
1513 if (pt->read_end == pt->read_buf + pt->read_buf_size
1514 && pt->read_buf == pt->read_pos)
1515 {
1516 size_t new_size = pt->read_buf_size * 2;
1517 unsigned char *tmp = (unsigned char *)
1518 scm_gc_realloc (pt->putback_buf, pt->read_buf_size, new_size,
1519 "putback buffer");
1520
1521 pt->read_pos = pt->read_buf = pt->putback_buf = tmp;
1522 pt->read_end = pt->read_buf + pt->read_buf_size;
1523 pt->read_buf_size = pt->putback_buf_size = new_size;
1524 }
1525
1526 /* shift any existing bytes to buffer + 1. */
1527 if (pt->read_pos == pt->read_end)
1528 pt->read_end = pt->read_buf + 1;
1529 else if (pt->read_pos != pt->read_buf + 1)
1530 {
1531 int count = pt->read_end - pt->read_pos;
1532
1533 memmove (pt->read_buf + 1, pt->read_pos, count);
1534 pt->read_end = pt->read_buf + 1 + count;
1535 }
1536
1537 pt->read_pos = pt->read_buf;
1538 }
1539 else
1540 /* switch to the put-back buffer. */
1541 {
1542 if (pt->putback_buf == NULL)
1543 {
1544 pt->putback_buf
1545 = (unsigned char *) scm_gc_malloc_pointerless
1546 (SCM_INITIAL_PUTBACK_BUF_SIZE, "putback buffer");
1547 pt->putback_buf_size = SCM_INITIAL_PUTBACK_BUF_SIZE;
1548 }
1549
1550 pt->saved_read_buf = pt->read_buf;
1551 pt->saved_read_pos = pt->read_pos;
1552 pt->saved_read_end = pt->read_end;
1553 pt->saved_read_buf_size = pt->read_buf_size;
1554
1555 pt->read_pos = pt->read_buf = pt->putback_buf;
1556 pt->read_end = pt->read_buf + 1;
1557 pt->read_buf_size = pt->putback_buf_size;
1558 }
1559
1560 *pt->read_buf = c;
1561
1562 if (pt->rw_random)
1563 pt->rw_active = SCM_PORT_READ;
1564 }
1565 #undef FUNC_NAME
1566
1567 void
1568 scm_ungetc (scm_t_wchar c, SCM port)
1569 #define FUNC_NAME "scm_ungetc"
1570 {
1571 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1572 char *result;
1573 char result_buf[10];
1574 const char *encoding;
1575 size_t len;
1576 int i;
1577
1578 if (pt->encoding != NULL)
1579 encoding = pt->encoding;
1580 else
1581 encoding = "ISO-8859-1";
1582
1583 len = sizeof (result_buf);
1584 result = u32_conv_to_encoding (encoding,
1585 (enum iconv_ilseq_handler) pt->ilseq_handler,
1586 (uint32_t *) &c, 1, NULL,
1587 result_buf, &len);
1588
1589 if (SCM_UNLIKELY (result == NULL || len == 0))
1590 {
1591 SCM chr;
1592
1593 chr = scm_integer_to_char (scm_from_uint32 (c));
1594 scm_encoding_error (FUNC_NAME, errno,
1595 "conversion to port encoding failed",
1596 "UTF-32", encoding,
1597 scm_string (scm_list_1 (chr)));
1598 }
1599
1600 for (i = len - 1; i >= 0; i--)
1601 scm_unget_byte (result[i], port);
1602
1603 if (SCM_UNLIKELY (result != result_buf))
1604 free (result);
1605
1606 if (c == '\n')
1607 {
1608 /* What should col be in this case?
1609 * We'll leave it at -1.
1610 */
1611 SCM_LINUM (port) -= 1;
1612 }
1613 else
1614 SCM_COL(port) -= 1;
1615 }
1616 #undef FUNC_NAME
1617
1618
1619 void
1620 scm_ungets (const char *s, int n, SCM port)
1621 {
1622 /* This is simple minded and inefficient, but unreading strings is
1623 * probably not a common operation, and remember that line and
1624 * column numbers have to be handled...
1625 *
1626 * Please feel free to write an optimized version!
1627 */
1628 while (n--)
1629 scm_ungetc (s[n], port);
1630 }
1631
1632
1633 SCM_DEFINE (scm_peek_char, "peek-char", 0, 1, 0,
1634 (SCM port),
1635 "Return the next character available from @var{port},\n"
1636 "@emph{without} updating @var{port} to point to the following\n"
1637 "character. If no more characters are available, the\n"
1638 "end-of-file object is returned.\n"
1639 "\n"
1640 "The value returned by\n"
1641 "a call to @code{peek-char} is the same as the value that would\n"
1642 "have been returned by a call to @code{read-char} on the same\n"
1643 "port. The only difference is that the very next call to\n"
1644 "@code{read-char} or @code{peek-char} on that @var{port} will\n"
1645 "return the value returned by the preceding call to\n"
1646 "@code{peek-char}. In particular, a call to @code{peek-char} on\n"
1647 "an interactive port will hang waiting for input whenever a call\n"
1648 "to @code{read-char} would have hung.")
1649 #define FUNC_NAME s_scm_peek_char
1650 {
1651 SCM result;
1652 scm_t_wchar c;
1653 char bytes[SCM_MBCHAR_BUF_SIZE];
1654 long column, line;
1655 size_t len;
1656
1657 if (SCM_UNBNDP (port))
1658 port = scm_current_input_port ();
1659 else
1660 SCM_VALIDATE_OPINPORT (1, port);
1661
1662 column = SCM_COL (port);
1663 line = SCM_LINUM (port);
1664
1665 c = get_codepoint (port, bytes, &len);
1666 if (c == EOF)
1667 result = SCM_EOF_VAL;
1668 else
1669 {
1670 long i;
1671
1672 result = SCM_MAKE_CHAR (c);
1673
1674 for (i = len - 1; i >= 0; i--)
1675 scm_unget_byte (bytes[i], port);
1676
1677 SCM_COL (port) = column;
1678 SCM_LINUM (port) = line;
1679 }
1680
1681 return result;
1682 }
1683 #undef FUNC_NAME
1684
1685 SCM_DEFINE (scm_unread_char, "unread-char", 1, 1, 0,
1686 (SCM cobj, SCM port),
1687 "Place @var{char} in @var{port} so that it will be read by the\n"
1688 "next read operation. If called multiple times, the unread characters\n"
1689 "will be read again in last-in first-out order. If @var{port} is\n"
1690 "not supplied, the current input port is used.")
1691 #define FUNC_NAME s_scm_unread_char
1692 {
1693 int c;
1694
1695 SCM_VALIDATE_CHAR (1, cobj);
1696 if (SCM_UNBNDP (port))
1697 port = scm_current_input_port ();
1698 else
1699 SCM_VALIDATE_OPINPORT (2, port);
1700
1701 c = SCM_CHAR (cobj);
1702
1703 scm_ungetc (c, port);
1704 return cobj;
1705 }
1706 #undef FUNC_NAME
1707
1708 SCM_DEFINE (scm_unread_string, "unread-string", 2, 0, 0,
1709 (SCM str, SCM port),
1710 "Place the string @var{str} in @var{port} so that its characters will be\n"
1711 "read in subsequent read operations. If called multiple times, the\n"
1712 "unread characters will be read again in last-in first-out order. If\n"
1713 "@var{port} is not supplied, the current-input-port is used.")
1714 #define FUNC_NAME s_scm_unread_string
1715 {
1716 int n;
1717 SCM_VALIDATE_STRING (1, str);
1718 if (SCM_UNBNDP (port))
1719 port = scm_current_input_port ();
1720 else
1721 SCM_VALIDATE_OPINPORT (2, port);
1722
1723 n = scm_i_string_length (str);
1724
1725 while (n--)
1726 scm_ungetc (scm_i_string_ref (str, n), port);
1727
1728 return str;
1729 }
1730 #undef FUNC_NAME
1731
1732 SCM_DEFINE (scm_seek, "seek", 3, 0, 0,
1733 (SCM fd_port, SCM offset, SCM whence),
1734 "Sets the current position of @var{fd/port} to the integer\n"
1735 "@var{offset}, which is interpreted according to the value of\n"
1736 "@var{whence}.\n"
1737 "\n"
1738 "One of the following variables should be supplied for\n"
1739 "@var{whence}:\n"
1740 "@defvar SEEK_SET\n"
1741 "Seek from the beginning of the file.\n"
1742 "@end defvar\n"
1743 "@defvar SEEK_CUR\n"
1744 "Seek from the current position.\n"
1745 "@end defvar\n"
1746 "@defvar SEEK_END\n"
1747 "Seek from the end of the file.\n"
1748 "@end defvar\n"
1749 "If @var{fd/port} is a file descriptor, the underlying system\n"
1750 "call is @code{lseek}. @var{port} may be a string port.\n"
1751 "\n"
1752 "The value returned is the new position in the file. This means\n"
1753 "that the current position of a port can be obtained using:\n"
1754 "@lisp\n"
1755 "(seek port 0 SEEK_CUR)\n"
1756 "@end lisp")
1757 #define FUNC_NAME s_scm_seek
1758 {
1759 int how;
1760
1761 fd_port = SCM_COERCE_OUTPORT (fd_port);
1762
1763 how = scm_to_int (whence);
1764 if (how != SEEK_SET && how != SEEK_CUR && how != SEEK_END)
1765 SCM_OUT_OF_RANGE (3, whence);
1766
1767 if (SCM_OPPORTP (fd_port))
1768 {
1769 scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (fd_port);
1770 off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
1771 off_t_or_off64_t rv;
1772
1773 if (!ptob->seek)
1774 SCM_MISC_ERROR ("port is not seekable",
1775 scm_cons (fd_port, SCM_EOL));
1776 else
1777 rv = ptob->seek (fd_port, off, how);
1778 return scm_from_off_t_or_off64_t (rv);
1779 }
1780 else /* file descriptor?. */
1781 {
1782 off_t_or_off64_t off = scm_to_off_t_or_off64_t (offset);
1783 off_t_or_off64_t rv;
1784 rv = lseek_or_lseek64 (scm_to_int (fd_port), off, how);
1785 if (rv == -1)
1786 SCM_SYSERROR;
1787 return scm_from_off_t_or_off64_t (rv);
1788 }
1789 }
1790 #undef FUNC_NAME
1791
1792 #ifndef O_BINARY
1793 #define O_BINARY 0
1794 #endif
1795
1796 /* Mingw has ftruncate(), perhaps implemented above using chsize, but
1797 doesn't have the filename version truncate(), hence this code. */
1798 #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE
1799 static int
1800 truncate (const char *file, off_t length)
1801 {
1802 int ret, fdes;
1803
1804 fdes = open (file, O_BINARY | O_WRONLY);
1805 if (fdes == -1)
1806 return -1;
1807
1808 ret = ftruncate (fdes, length);
1809 if (ret == -1)
1810 {
1811 int save_errno = errno;
1812 close (fdes);
1813 errno = save_errno;
1814 return -1;
1815 }
1816
1817 return close (fdes);
1818 }
1819 #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */
1820
1821 SCM_DEFINE (scm_truncate_file, "truncate-file", 1, 1, 0,
1822 (SCM object, SCM length),
1823 "Truncate @var{file} to @var{length} bytes. @var{file} can be a\n"
1824 "filename string, a port object, or an integer file descriptor.\n"
1825 "The return value is unspecified.\n"
1826 "\n"
1827 "For a port or file descriptor @var{length} can be omitted, in\n"
1828 "which case the file is truncated at the current position (per\n"
1829 "@code{ftell} above).\n"
1830 "\n"
1831 "On most systems a file can be extended by giving a length\n"
1832 "greater than the current size, but this is not mandatory in the\n"
1833 "POSIX standard.")
1834 #define FUNC_NAME s_scm_truncate_file
1835 {
1836 int rv;
1837
1838 /* "object" can be a port, fdes or filename.
1839
1840 Negative "length" makes no sense, but it's left to truncate() or
1841 ftruncate() to give back an error for that (normally EINVAL).
1842 */
1843
1844 if (SCM_UNBNDP (length))
1845 {
1846 /* must supply length if object is a filename. */
1847 if (scm_is_string (object))
1848 SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL);
1849
1850 length = scm_seek (object, SCM_INUM0, scm_from_int (SEEK_CUR));
1851 }
1852
1853 object = SCM_COERCE_OUTPORT (object);
1854 if (scm_is_integer (object))
1855 {
1856 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1857 SCM_SYSCALL (rv = ftruncate_or_ftruncate64 (scm_to_int (object),
1858 c_length));
1859 }
1860 else if (SCM_OPOUTPORTP (object))
1861 {
1862 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1863 scm_t_port *pt = SCM_PTAB_ENTRY (object);
1864 scm_t_ptob_descriptor *ptob = scm_ptobs + SCM_PTOBNUM (object);
1865
1866 if (!ptob->truncate)
1867 SCM_MISC_ERROR ("port is not truncatable", SCM_EOL);
1868 if (pt->rw_active == SCM_PORT_READ)
1869 scm_end_input (object);
1870 else if (pt->rw_active == SCM_PORT_WRITE)
1871 ptob->flush (object);
1872
1873 ptob->truncate (object, c_length);
1874 rv = 0;
1875 }
1876 else
1877 {
1878 off_t_or_off64_t c_length = scm_to_off_t_or_off64_t (length);
1879 char *str = scm_to_locale_string (object);
1880 int eno;
1881 SCM_SYSCALL (rv = truncate_or_truncate64 (str, c_length));
1882 eno = errno;
1883 free (str);
1884 errno = eno;
1885 }
1886 if (rv == -1)
1887 SCM_SYSERROR;
1888 return SCM_UNSPECIFIED;
1889 }
1890 #undef FUNC_NAME
1891
1892 SCM_DEFINE (scm_port_line, "port-line", 1, 0, 0,
1893 (SCM port),
1894 "Return the current line number for @var{port}.\n"
1895 "\n"
1896 "The first line of a file is 0. But you might want to add 1\n"
1897 "when printing line numbers, since starting from 1 is\n"
1898 "traditional in error messages, and likely to be more natural to\n"
1899 "non-programmers.")
1900 #define FUNC_NAME s_scm_port_line
1901 {
1902 port = SCM_COERCE_OUTPORT (port);
1903 SCM_VALIDATE_OPENPORT (1, port);
1904 return scm_from_long (SCM_LINUM (port));
1905 }
1906 #undef FUNC_NAME
1907
1908 SCM_DEFINE (scm_set_port_line_x, "set-port-line!", 2, 0, 0,
1909 (SCM port, SCM line),
1910 "Set the current line number for @var{port} to @var{line}. The\n"
1911 "first line of a file is 0.")
1912 #define FUNC_NAME s_scm_set_port_line_x
1913 {
1914 port = SCM_COERCE_OUTPORT (port);
1915 SCM_VALIDATE_OPENPORT (1, port);
1916 SCM_PTAB_ENTRY (port)->line_number = scm_to_long (line);
1917 return SCM_UNSPECIFIED;
1918 }
1919 #undef FUNC_NAME
1920
1921 SCM_DEFINE (scm_port_column, "port-column", 1, 0, 0,
1922 (SCM port),
1923 "Return the current column number of @var{port}.\n"
1924 "If the number is\n"
1925 "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n"
1926 "- i.e. the first character of the first line is line 0, column 0.\n"
1927 "(However, when you display a file position, for example in an error\n"
1928 "message, we recommend you add 1 to get 1-origin integers. This is\n"
1929 "because lines and column numbers traditionally start with 1, and that is\n"
1930 "what non-programmers will find most natural.)")
1931 #define FUNC_NAME s_scm_port_column
1932 {
1933 port = SCM_COERCE_OUTPORT (port);
1934 SCM_VALIDATE_OPENPORT (1, port);
1935 return scm_from_int (SCM_COL (port));
1936 }
1937 #undef FUNC_NAME
1938
1939 SCM_DEFINE (scm_set_port_column_x, "set-port-column!", 2, 0, 0,
1940 (SCM port, SCM column),
1941 "Set the current column of @var{port}. Before reading the first\n"
1942 "character on a line the column should be 0.")
1943 #define FUNC_NAME s_scm_set_port_column_x
1944 {
1945 port = SCM_COERCE_OUTPORT (port);
1946 SCM_VALIDATE_OPENPORT (1, port);
1947 SCM_PTAB_ENTRY (port)->column_number = scm_to_int (column);
1948 return SCM_UNSPECIFIED;
1949 }
1950 #undef FUNC_NAME
1951
1952 SCM_DEFINE (scm_port_filename, "port-filename", 1, 0, 0,
1953 (SCM port),
1954 "Return the filename associated with @var{port}. This function returns\n"
1955 "the strings \"standard input\", \"standard output\" and \"standard error\"\n"
1956 "when called on the current input, output and error ports respectively.")
1957 #define FUNC_NAME s_scm_port_filename
1958 {
1959 port = SCM_COERCE_OUTPORT (port);
1960 SCM_VALIDATE_OPENPORT (1, port);
1961 return SCM_FILENAME (port);
1962 }
1963 #undef FUNC_NAME
1964
1965 SCM_DEFINE (scm_set_port_filename_x, "set-port-filename!", 2, 0, 0,
1966 (SCM port, SCM filename),
1967 "Change the filename associated with @var{port}, using the current input\n"
1968 "port if none is specified. Note that this does not change the port's\n"
1969 "source of data, but only the value that is returned by\n"
1970 "@code{port-filename} and reported in diagnostic output.")
1971 #define FUNC_NAME s_scm_set_port_filename_x
1972 {
1973 port = SCM_COERCE_OUTPORT (port);
1974 SCM_VALIDATE_OPENPORT (1, port);
1975 /* We allow the user to set the filename to whatever he likes. */
1976 SCM_SET_FILENAME (port, filename);
1977 return SCM_UNSPECIFIED;
1978 }
1979 #undef FUNC_NAME
1980
1981 /* A fluid specifying the default encoding for newly created ports. If it is
1982 a string, that is the encoding. If it is #f, it is in the "native"
1983 (Latin-1) encoding. */
1984 SCM_VARIABLE (default_port_encoding_var, "%default-port-encoding");
1985
1986 static int scm_port_encoding_init = 0;
1987
1988 /* Return a C string representation of the current encoding. */
1989 const char *
1990 scm_i_get_port_encoding (SCM port)
1991 {
1992 SCM encoding;
1993
1994 if (scm_is_false (port))
1995 {
1996 if (!scm_port_encoding_init)
1997 return NULL;
1998 else if (!scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var)))
1999 return NULL;
2000 else
2001 {
2002 encoding = scm_fluid_ref (SCM_VARIABLE_REF (default_port_encoding_var));
2003 if (!scm_is_string (encoding))
2004 return NULL;
2005 else
2006 return scm_i_string_chars (encoding);
2007 }
2008 }
2009 else
2010 {
2011 scm_t_port *pt;
2012 pt = SCM_PTAB_ENTRY (port);
2013 if (pt->encoding)
2014 return pt->encoding;
2015 else
2016 return NULL;
2017 }
2018 }
2019
2020 /* Returns ENC if it is a recognized encoding. If it isn't, it tries
2021 to find an alias of ENC that is valid. Otherwise, it returns
2022 NULL. */
2023 static const char *
2024 find_valid_encoding (const char *enc)
2025 {
2026 int isvalid = 0;
2027 const char str[] = " ";
2028 scm_t_uint32 result_buf;
2029 scm_t_uint32 *u32;
2030 size_t u32len;
2031
2032 u32len = sizeof (result_buf) / sizeof (scm_t_uint32);
2033 u32 = u32_conv_from_encoding (enc, iconveh_error, str, 1,
2034 NULL, &result_buf, &u32len);
2035 isvalid = (u32 != NULL);
2036
2037 if (SCM_UNLIKELY (u32 != &result_buf))
2038 free (u32);
2039
2040 if (isvalid)
2041 return enc;
2042
2043 return NULL;
2044 }
2045
2046 void
2047 scm_i_set_port_encoding_x (SCM port, const char *enc)
2048 {
2049 const char *valid_enc;
2050 scm_t_port *pt;
2051
2052 /* Null is shorthand for the native, Latin-1 encoding. */
2053 if (enc == NULL)
2054 valid_enc = NULL;
2055 else
2056 {
2057 valid_enc = find_valid_encoding (enc);
2058 if (valid_enc == NULL)
2059 {
2060 SCM err;
2061 err = scm_from_locale_string (enc);
2062 scm_misc_error (NULL, "invalid or unknown character encoding ~s",
2063 scm_list_1 (err));
2064 }
2065 }
2066
2067 if (scm_is_false (port))
2068 {
2069 /* Set the default encoding for future ports. */
2070 if (!scm_port_encoding_init
2071 || !scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var)))
2072 scm_misc_error (NULL, "tried to set port encoding fluid before it is initialized",
2073 SCM_EOL);
2074
2075 if (valid_enc == NULL
2076 || !strcmp (valid_enc, "ASCII")
2077 || !strcmp (valid_enc, "ANSI_X3.4-1968")
2078 || !strcmp (valid_enc, "ISO-8859-1"))
2079 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), SCM_BOOL_F);
2080 else
2081 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var),
2082 scm_from_locale_string (valid_enc));
2083 }
2084 else
2085 {
2086 /* Set the character encoding for this port. */
2087 pt = SCM_PTAB_ENTRY (port);
2088 if (valid_enc == NULL)
2089 pt->encoding = NULL;
2090 else
2091 pt->encoding = scm_gc_strdup (valid_enc, "port");
2092 }
2093 }
2094
2095 SCM_DEFINE (scm_port_encoding, "port-encoding", 1, 0, 0,
2096 (SCM port),
2097 "Returns, as a string, the character encoding that @var{port}\n"
2098 "uses to interpret its input and output.\n")
2099 #define FUNC_NAME s_scm_port_encoding
2100 {
2101 scm_t_port *pt;
2102 const char *enc;
2103
2104 SCM_VALIDATE_PORT (1, port);
2105
2106 pt = SCM_PTAB_ENTRY (port);
2107 enc = scm_i_get_port_encoding (port);
2108 if (enc)
2109 return scm_from_locale_string (pt->encoding);
2110 else
2111 return SCM_BOOL_F;
2112 }
2113 #undef FUNC_NAME
2114
2115 SCM_DEFINE (scm_set_port_encoding_x, "set-port-encoding!", 2, 0, 0,
2116 (SCM port, SCM enc),
2117 "Sets the character encoding that will be used to interpret all\n"
2118 "port I/O. New ports are created with the encoding\n"
2119 "appropriate for the current locale if @code{setlocale} has \n"
2120 "been called or ISO-8859-1 otherwise\n"
2121 "and this procedure can be used to modify that encoding.\n")
2122 #define FUNC_NAME s_scm_set_port_encoding_x
2123 {
2124 char *enc_str;
2125 const char *valid_enc_str;
2126
2127 SCM_VALIDATE_PORT (1, port);
2128 SCM_VALIDATE_STRING (2, enc);
2129
2130 enc_str = scm_to_locale_string (enc);
2131 valid_enc_str = find_valid_encoding (enc_str);
2132 if (valid_enc_str == NULL)
2133 {
2134 free (enc_str);
2135 scm_misc_error (FUNC_NAME, "invalid or unknown character encoding ~s",
2136 scm_list_1 (enc));
2137 }
2138 else
2139 {
2140 scm_i_set_port_encoding_x (port, valid_enc_str);
2141 free (enc_str);
2142 }
2143 return SCM_UNSPECIFIED;
2144 }
2145 #undef FUNC_NAME
2146
2147
2148 /* This determines how conversions handle unconvertible characters. */
2149 SCM_GLOBAL_VARIABLE (scm_conversion_strategy, "%port-conversion-strategy");
2150 static int scm_conversion_strategy_init = 0;
2151
2152 scm_t_string_failed_conversion_handler
2153 scm_i_get_conversion_strategy (SCM port)
2154 {
2155 SCM encoding;
2156
2157 if (scm_is_false (port))
2158 {
2159 if (!scm_conversion_strategy_init
2160 || !scm_is_fluid (SCM_VARIABLE_REF (scm_conversion_strategy)))
2161 return SCM_FAILED_CONVERSION_QUESTION_MARK;
2162 else
2163 {
2164 encoding = scm_fluid_ref (SCM_VARIABLE_REF (scm_conversion_strategy));
2165 if (scm_is_false (encoding))
2166 return SCM_FAILED_CONVERSION_QUESTION_MARK;
2167 else
2168 return (scm_t_string_failed_conversion_handler) scm_to_int (encoding);
2169 }
2170 }
2171 else
2172 {
2173 scm_t_port *pt;
2174 pt = SCM_PTAB_ENTRY (port);
2175 return pt->ilseq_handler;
2176 }
2177
2178 }
2179
2180 void
2181 scm_i_set_conversion_strategy_x (SCM port,
2182 scm_t_string_failed_conversion_handler handler)
2183 {
2184 SCM strategy;
2185 scm_t_port *pt;
2186
2187 strategy = scm_from_int ((int) handler);
2188
2189 if (scm_is_false (port))
2190 {
2191 /* Set the default encoding for future ports. */
2192 if (!scm_conversion_strategy
2193 || !scm_is_fluid (SCM_VARIABLE_REF (scm_conversion_strategy)))
2194 scm_misc_error (NULL, "tried to set conversion strategy fluid before it is initialized",
2195 SCM_EOL);
2196 scm_fluid_set_x (SCM_VARIABLE_REF (scm_conversion_strategy), strategy);
2197 }
2198 else
2199 {
2200 /* Set the character encoding for this port. */
2201 pt = SCM_PTAB_ENTRY (port);
2202 pt->ilseq_handler = handler;
2203 }
2204 }
2205
2206 SCM_DEFINE (scm_port_conversion_strategy, "port-conversion-strategy",
2207 1, 0, 0, (SCM port),
2208 "Returns the behavior of the port when handling a character that\n"
2209 "is not representable in the port's current encoding.\n"
2210 "It returns the symbol @code{error} if unrepresentable characters\n"
2211 "should cause exceptions, @code{substitute} if the port should\n"
2212 "try to replace unrepresentable characters with question marks or\n"
2213 "approximate characters, or @code{escape} if unrepresentable\n"
2214 "characters should be converted to string escapes.\n"
2215 "\n"
2216 "If @var{port} is @code{#f}, then the current default behavior\n"
2217 "will be returned. New ports will have this default behavior\n"
2218 "when they are created.\n")
2219 #define FUNC_NAME s_scm_port_conversion_strategy
2220 {
2221 scm_t_string_failed_conversion_handler h;
2222
2223 SCM_VALIDATE_OPPORT (1, port);
2224
2225 if (!scm_is_false (port))
2226 {
2227 SCM_VALIDATE_OPPORT (1, port);
2228 }
2229
2230 h = scm_i_get_conversion_strategy (port);
2231 if (h == SCM_FAILED_CONVERSION_ERROR)
2232 return scm_from_locale_symbol ("error");
2233 else if (h == SCM_FAILED_CONVERSION_QUESTION_MARK)
2234 return scm_from_locale_symbol ("substitute");
2235 else if (h == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
2236 return scm_from_locale_symbol ("escape");
2237 else
2238 abort ();
2239
2240 /* Never gets here. */
2241 return SCM_UNDEFINED;
2242 }
2243 #undef FUNC_NAME
2244
2245 SCM_DEFINE (scm_set_port_conversion_strategy_x, "set-port-conversion-strategy!",
2246 2, 0, 0,
2247 (SCM port, SCM sym),
2248 "Sets the behavior of the interpreter when outputting a character\n"
2249 "that is not representable in the port's current encoding.\n"
2250 "@var{sym} can be either @code{'error}, @code{'substitute}, or\n"
2251 "@code{'escape}. If it is @code{'error}, an error will be thrown\n"
2252 "when an unconvertible character is encountered. If it is\n"
2253 "@code{'substitute}, then unconvertible characters will \n"
2254 "be replaced with approximate characters, or with question marks\n"
2255 "if no approximately correct character is available.\n"
2256 "If it is @code{'escape},\n"
2257 "it will appear as a hex escape when output.\n"
2258 "\n"
2259 "If @var{port} is an open port, the conversion error behavior\n"
2260 "is set for that port. If it is @code{#f}, it is set as the\n"
2261 "default behavior for any future ports that get created in\n"
2262 "this thread.\n")
2263 #define FUNC_NAME s_scm_set_port_conversion_strategy_x
2264 {
2265 SCM err;
2266 SCM qm;
2267 SCM esc;
2268
2269 if (!scm_is_false (port))
2270 {
2271 SCM_VALIDATE_OPPORT (1, port);
2272 }
2273
2274 err = scm_from_locale_symbol ("error");
2275 if (scm_is_true (scm_eqv_p (sym, err)))
2276 {
2277 scm_i_set_conversion_strategy_x (port, SCM_FAILED_CONVERSION_ERROR);
2278 return SCM_UNSPECIFIED;
2279 }
2280
2281 qm = scm_from_locale_symbol ("substitute");
2282 if (scm_is_true (scm_eqv_p (sym, qm)))
2283 {
2284 scm_i_set_conversion_strategy_x (port,
2285 SCM_FAILED_CONVERSION_QUESTION_MARK);
2286 return SCM_UNSPECIFIED;
2287 }
2288
2289 esc = scm_from_locale_symbol ("escape");
2290 if (scm_is_true (scm_eqv_p (sym, esc)))
2291 {
2292 scm_i_set_conversion_strategy_x (port,
2293 SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE);
2294 return SCM_UNSPECIFIED;
2295 }
2296
2297 SCM_MISC_ERROR ("unknown conversion behavior ~s", scm_list_1 (sym));
2298
2299 return SCM_UNSPECIFIED;
2300 }
2301 #undef FUNC_NAME
2302
2303
2304
2305 void
2306 scm_print_port_mode (SCM exp, SCM port)
2307 {
2308 scm_puts (SCM_CLOSEDP (exp)
2309 ? "closed: "
2310 : (SCM_RDNG & SCM_CELL_WORD_0 (exp)
2311 ? (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
2312 ? "input-output: "
2313 : "input: ")
2314 : (SCM_WRTNG & SCM_CELL_WORD_0 (exp)
2315 ? "output: "
2316 : "bogus: ")),
2317 port);
2318 }
2319
2320 int
2321 scm_port_print (SCM exp, SCM port, scm_print_state *pstate SCM_UNUSED)
2322 {
2323 char *type = SCM_PTOBNAME (SCM_PTOBNUM (exp));
2324 if (!type)
2325 type = "port";
2326 scm_puts ("#<", port);
2327 scm_print_port_mode (exp, port);
2328 scm_puts (type, port);
2329 scm_putc (' ', port);
2330 scm_uintprint (SCM_CELL_WORD_1 (exp), 16, port);
2331 scm_putc ('>', port);
2332 return 1;
2333 }
2334
2335 \f
2336
2337 /* Void ports. */
2338
2339 scm_t_bits scm_tc16_void_port = 0;
2340
2341 static int fill_input_void_port (SCM port SCM_UNUSED)
2342 {
2343 return EOF;
2344 }
2345
2346 static void
2347 write_void_port (SCM port SCM_UNUSED,
2348 const void *data SCM_UNUSED,
2349 size_t size SCM_UNUSED)
2350 {
2351 }
2352
2353 static SCM
2354 scm_i_void_port (long mode_bits)
2355 {
2356 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex);
2357 {
2358 SCM answer = scm_new_port_table_entry (scm_tc16_void_port);
2359 scm_t_port * pt = SCM_PTAB_ENTRY(answer);
2360
2361 scm_port_non_buffer (pt);
2362
2363 SCM_SETSTREAM (answer, 0);
2364 SCM_SET_CELL_TYPE (answer, scm_tc16_void_port | mode_bits);
2365 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex);
2366 return answer;
2367 }
2368 }
2369
2370 SCM
2371 scm_void_port (char *mode_str)
2372 {
2373 return scm_i_void_port (scm_mode_bits (mode_str));
2374 }
2375
2376 SCM_DEFINE (scm_sys_make_void_port, "%make-void-port", 1, 0, 0,
2377 (SCM mode),
2378 "Create and return a new void port. A void port acts like\n"
2379 "@file{/dev/null}. The @var{mode} argument\n"
2380 "specifies the input/output modes for this port: see the\n"
2381 "documentation for @code{open-file} in @ref{File Ports}.")
2382 #define FUNC_NAME s_scm_sys_make_void_port
2383 {
2384 return scm_i_void_port (scm_i_mode_bits (mode));
2385 }
2386 #undef FUNC_NAME
2387
2388 \f
2389 /* Initialization. */
2390
2391 void
2392 scm_init_ports ()
2393 {
2394 /* lseek() symbols. */
2395 scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET));
2396 scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR));
2397 scm_c_define ("SEEK_END", scm_from_int (SEEK_END));
2398
2399 scm_tc16_void_port = scm_make_port_type ("void", fill_input_void_port,
2400 write_void_port);
2401
2402 cur_inport_fluid = scm_make_fluid ();
2403 cur_outport_fluid = scm_make_fluid ();
2404 cur_errport_fluid = scm_make_fluid ();
2405 cur_loadport_fluid = scm_make_fluid ();
2406
2407 scm_i_port_weak_hash = scm_make_weak_key_hash_table (SCM_I_MAKINUM(31));
2408
2409 #include "libguile/ports.x"
2410
2411 /* Use Latin-1 as the default port encoding. */
2412 SCM_VARIABLE_SET (default_port_encoding_var, scm_make_fluid ());
2413 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var), SCM_BOOL_F);
2414 scm_port_encoding_init = 1;
2415
2416 SCM_VARIABLE_SET (scm_conversion_strategy, scm_make_fluid ());
2417 scm_fluid_set_x (SCM_VARIABLE_REF (scm_conversion_strategy),
2418 scm_from_int ((int) SCM_FAILED_CONVERSION_QUESTION_MARK));
2419 scm_conversion_strategy_init = 1;
2420
2421 }
2422
2423 /*
2424 Local Variables:
2425 c-file-style: "gnu"
2426 End:
2427 */