1 /* Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004, 2006,
2 * 2007, 2008, 2009, 2010, 2011, 2012, 2013,
3 * 2014 Free Software Foundation, Inc.
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public License
7 * as published by the Free Software Foundation; either version 3 of
8 * the License, or (at your option) any later version.
10 * This library is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * Lesser General Public License for more details.
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
25 #define _LARGEFILE64_SOURCE /* ask for stat64 etc */
33 #include <fcntl.h> /* for chsize on mingw */
38 #include <striconveh.h>
39 #include <c-strcase.h>
43 #include "libguile/_scm.h"
44 #include "libguile/async.h"
45 #include "libguile/deprecation.h"
46 #include "libguile/eval.h"
47 #include "libguile/fports.h" /* direct access for seek and truncate */
48 #include "libguile/goops.h"
49 #include "libguile/smob.h"
50 #include "libguile/chars.h"
51 #include "libguile/dynwind.h"
53 #include "libguile/keywords.h"
54 #include "libguile/hashtab.h"
55 #include "libguile/root.h"
56 #include "libguile/strings.h"
57 #include "libguile/mallocs.h"
58 #include "libguile/validate.h"
59 #include "libguile/ports.h"
60 #include "libguile/ports-internal.h"
61 #include "libguile/vectors.h"
62 #include "libguile/weaks.h"
63 #include "libguile/fluids.h"
64 #include "libguile/eq.h"
65 #include "libguile/alist.h"
77 #ifdef HAVE_SYS_IOCTL_H
78 #include <sys/ioctl.h>
81 /* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize
82 already, but have this code here in case that wasn't so in past versions,
83 or perhaps to help other minimal DOS environments.
85 gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which
86 might be possibilities if we've got other systems without ftruncate. */
88 #if defined HAVE_CHSIZE && ! defined HAVE_FTRUNCATE
89 #define ftruncate(fd, size) chsize (fd, size)
91 #define HAVE_FTRUNCATE 1
95 /* The port kind table --- a dynamically resized array of port types. */
98 /* scm_ptobs scm_numptob
99 * implement a dynamically resized array of ptob records.
100 * Indexes into this table are used when generating type
101 * tags for smobjects (if you know a tag you can get an index and conversely).
103 scm_t_ptob_descriptor
*scm_ptobs
= NULL
;
104 long scm_numptob
= 0;
106 /* GC marker for a port with stream of SCM type. */
108 scm_markstream (SCM ptr
)
111 openp
= SCM_CELL_WORD_0 (ptr
) & SCM_OPN
;
113 return SCM_PACK (SCM_STREAM (ptr
));
119 * We choose to use an interface similar to the smob interface with
120 * fill_input and write as standard fields, passed to the port
121 * type constructor, and optional fields set by setters.
125 flush_port_default (SCM port SCM_UNUSED
)
130 end_input_default (SCM port SCM_UNUSED
, int offset SCM_UNUSED
)
135 scm_make_port_type (char *name
,
136 int (*fill_input
) (SCM port
),
137 void (*write
) (SCM port
, const void *data
, size_t size
))
140 if (SCM_I_MAX_PORT_TYPE_COUNT
- 1 <= scm_numptob
)
142 SCM_CRITICAL_SECTION_START
;
143 tmp
= (char *) scm_gc_realloc ((char *) scm_ptobs
,
144 scm_numptob
* sizeof (scm_t_ptob_descriptor
),
146 * sizeof (scm_t_ptob_descriptor
),
150 scm_ptobs
= (scm_t_ptob_descriptor
*) tmp
;
152 scm_ptobs
[scm_numptob
].name
= name
;
153 scm_ptobs
[scm_numptob
].mark
= 0;
154 scm_ptobs
[scm_numptob
].free
= NULL
;
155 scm_ptobs
[scm_numptob
].print
= scm_port_print
;
156 scm_ptobs
[scm_numptob
].equalp
= 0;
157 scm_ptobs
[scm_numptob
].close
= 0;
159 scm_ptobs
[scm_numptob
].write
= write
;
160 scm_ptobs
[scm_numptob
].flush
= flush_port_default
;
162 scm_ptobs
[scm_numptob
].end_input
= end_input_default
;
163 scm_ptobs
[scm_numptob
].fill_input
= fill_input
;
164 scm_ptobs
[scm_numptob
].input_waiting
= 0;
166 scm_ptobs
[scm_numptob
].seek
= 0;
167 scm_ptobs
[scm_numptob
].truncate
= 0;
171 SCM_CRITICAL_SECTION_END
;
175 scm_memory_error ("scm_make_port_type");
177 /* Make a class object if Goops is present */
178 if (SCM_UNPACK (scm_port_class
[0]) != 0)
179 scm_make_port_classes (scm_numptob
- 1, SCM_PTOBNAME (scm_numptob
- 1));
180 return scm_tc7_port
+ (scm_numptob
- 1) * 256;
184 scm_set_port_mark (scm_t_bits tc
, SCM (*mark
) (SCM
))
186 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].mark
= mark
;
190 scm_set_port_free (scm_t_bits tc
, size_t (*free
) (SCM
))
192 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].free
= free
;
196 scm_set_port_print (scm_t_bits tc
, int (*print
) (SCM exp
, SCM port
,
197 scm_print_state
*pstate
))
199 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].print
= print
;
203 scm_set_port_equalp (scm_t_bits tc
, SCM (*equalp
) (SCM
, SCM
))
205 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].equalp
= equalp
;
209 scm_set_port_flush (scm_t_bits tc
, void (*flush
) (SCM port
))
211 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].flush
= flush
;
215 scm_set_port_end_input (scm_t_bits tc
, void (*end_input
) (SCM port
, int offset
))
217 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].end_input
= end_input
;
221 scm_set_port_close (scm_t_bits tc
, int (*close
) (SCM
))
223 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].close
= close
;
227 scm_set_port_seek (scm_t_bits tc
,
228 scm_t_off (*seek
) (SCM
, scm_t_off
, int))
230 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].seek
= seek
;
234 scm_set_port_truncate (scm_t_bits tc
, void (*truncate
) (SCM
, scm_t_off
))
236 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].truncate
= truncate
;
240 scm_set_port_input_waiting (scm_t_bits tc
, int (*input_waiting
) (SCM
))
242 scm_ptobs
[SCM_TC2PTOBNUM (tc
)].input_waiting
= input_waiting
;
246 scm_i_set_pending_eof (SCM port
)
248 SCM_PORT_GET_INTERNAL (port
)->pending_eof
= 1;
252 scm_i_clear_pending_eof (SCM port
)
254 SCM_PORT_GET_INTERNAL (port
)->pending_eof
= 0;
257 SCM_DEFINE (scm_i_port_property
, "%port-property", 2, 0, 0,
259 "Return the property of @var{port} associated with @var{key}.")
260 #define FUNC_NAME s_scm_i_port_property
262 SCM_VALIDATE_OPPORT (1, port
);
263 return scm_assq_ref (SCM_PORT_GET_INTERNAL (port
)->alist
, key
);
267 SCM_DEFINE (scm_i_set_port_property_x
, "%set-port-property!", 3, 0, 0,
268 (SCM port
, SCM key
, SCM value
),
269 "Set the property of @var{port} associated with @var{key} to @var{value}.")
270 #define FUNC_NAME s_scm_i_set_port_property_x
272 scm_t_port_internal
*pti
;
274 SCM_VALIDATE_OPPORT (1, port
);
275 pti
= SCM_PORT_GET_INTERNAL (port
);
276 pti
->alist
= scm_assq_set_x (pti
->alist
, key
, value
);
277 return SCM_UNSPECIFIED
;
283 SCM_DEFINE (scm_char_ready_p
, "char-ready?", 0, 1, 0,
285 "Return @code{#t} if a character is ready on input @var{port}\n"
286 "and return @code{#f} otherwise. If @code{char-ready?} returns\n"
287 "@code{#t} then the next @code{read-char} operation on\n"
288 "@var{port} is guaranteed not to hang. If @var{port} is a file\n"
289 "port at end of file then @code{char-ready?} returns @code{#t}.\n"
291 "@code{char-ready?} exists to make it possible for a\n"
292 "program to accept characters from interactive ports without\n"
293 "getting stuck waiting for input. Any input editors associated\n"
294 "with such ports must make sure that characters whose existence\n"
295 "has been asserted by @code{char-ready?} cannot be rubbed out.\n"
296 "If @code{char-ready?} were to return @code{#f} at end of file,\n"
297 "a port at end of file would be indistinguishable from an\n"
298 "interactive port that has no ready characters.")
299 #define FUNC_NAME s_scm_char_ready_p
303 if (SCM_UNBNDP (port
))
304 port
= scm_current_input_port ();
305 /* It's possible to close the current input port, so validate even in
307 SCM_VALIDATE_OPINPORT (1, port
);
309 pt
= SCM_PTAB_ENTRY (port
);
311 /* if the current read buffer is filled, or the
312 last pushed-back char has been read and the saved buffer is
313 filled, result is true. */
314 if (pt
->read_pos
< pt
->read_end
315 || (pt
->read_buf
== pt
->putback_buf
316 && pt
->saved_read_pos
< pt
->saved_read_end
))
320 scm_t_ptob_descriptor
*ptob
= &scm_ptobs
[SCM_PTOBNUM (port
)];
322 if (ptob
->input_waiting
)
323 return scm_from_bool(ptob
->input_waiting (port
));
330 /* Move up to READ_LEN bytes from PORT's putback and/or read buffers
331 into memory starting at DEST. Return the number of bytes moved.
332 PORT's line/column numbers are left unchanged. */
334 scm_take_from_input_buffers (SCM port
, char *dest
, size_t read_len
)
336 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
337 size_t bytes_read
= 0;
338 size_t from_buf
= min (pt
->read_end
- pt
->read_pos
, read_len
);
342 memcpy (dest
, pt
->read_pos
, from_buf
);
343 pt
->read_pos
+= from_buf
;
344 bytes_read
+= from_buf
;
345 read_len
-= from_buf
;
349 /* if putback was active, try the real input buffer too. */
350 if (pt
->read_buf
== pt
->putback_buf
)
352 from_buf
= min (pt
->saved_read_end
- pt
->saved_read_pos
, read_len
);
355 memcpy (dest
, pt
->saved_read_pos
, from_buf
);
356 pt
->saved_read_pos
+= from_buf
;
357 bytes_read
+= from_buf
;
364 /* Clear a port's read buffers, returning the contents. */
365 SCM_DEFINE (scm_drain_input
, "drain-input", 1, 0, 0,
367 "This procedure clears a port's input buffers, similar\n"
368 "to the way that force-output clears the output buffer. The\n"
369 "contents of the buffers are returned as a single string, e.g.,\n"
372 "(define p (open-input-file ...))\n"
373 "(drain-input p) => empty string, nothing buffered yet.\n"
374 "(unread-char (read-char p) p)\n"
375 "(drain-input p) => initial chars from p, up to the buffer size.\n"
377 "Draining the buffers may be useful for cleanly finishing\n"
378 "buffered I/O so that the file descriptor can be used directly\n"
379 "for further input.")
380 #define FUNC_NAME s_scm_drain_input
387 SCM_VALIDATE_OPINPORT (1, port
);
388 pt
= SCM_PTAB_ENTRY (port
);
390 count
= pt
->read_end
- pt
->read_pos
;
391 if (pt
->read_buf
== pt
->putback_buf
)
392 count
+= pt
->saved_read_end
- pt
->saved_read_pos
;
396 result
= scm_i_make_string (count
, &data
, 0);
397 scm_take_from_input_buffers (port
, data
, count
);
400 result
= scm_nullstr
;
407 /* Standard ports --- current input, output, error, and more(!). */
409 static SCM cur_inport_fluid
= SCM_BOOL_F
;
410 static SCM cur_outport_fluid
= SCM_BOOL_F
;
411 static SCM cur_errport_fluid
= SCM_BOOL_F
;
412 static SCM cur_loadport_fluid
= SCM_BOOL_F
;
414 SCM_DEFINE (scm_current_input_port
, "current-input-port", 0, 0, 0,
416 "Return the current input port. This is the default port used\n"
417 "by many input procedures. Initially, @code{current-input-port}\n"
418 "returns the @dfn{standard input} in Unix and C terminology.")
419 #define FUNC_NAME s_scm_current_input_port
421 if (scm_is_true (cur_inport_fluid
))
422 return scm_fluid_ref (cur_inport_fluid
);
428 SCM_DEFINE (scm_current_output_port
, "current-output-port", 0, 0, 0,
430 "Return the current output port. This is the default port used\n"
431 "by many output procedures. Initially,\n"
432 "@code{current-output-port} returns the @dfn{standard output} in\n"
433 "Unix and C terminology.")
434 #define FUNC_NAME s_scm_current_output_port
436 if (scm_is_true (cur_outport_fluid
))
437 return scm_fluid_ref (cur_outport_fluid
);
443 SCM_DEFINE (scm_current_error_port
, "current-error-port", 0, 0, 0,
445 "Return the port to which errors and warnings should be sent (the\n"
446 "@dfn{standard error} in Unix and C terminology).")
447 #define FUNC_NAME s_scm_current_error_port
449 if (scm_is_true (cur_errport_fluid
))
450 return scm_fluid_ref (cur_errport_fluid
);
456 static SCM current_warning_port_var
;
457 static scm_i_pthread_once_t current_warning_port_once
= SCM_I_PTHREAD_ONCE_INIT
;
460 init_current_warning_port_var (void)
462 current_warning_port_var
463 = scm_c_private_variable ("guile", "current-warning-port");
467 scm_current_warning_port (void)
469 scm_i_pthread_once (¤t_warning_port_once
,
470 init_current_warning_port_var
);
471 return scm_call_0 (scm_variable_ref (current_warning_port_var
));
474 SCM_DEFINE (scm_current_load_port
, "current-load-port", 0, 0, 0,
476 "Return the current-load-port.\n"
477 "The load port is used internally by @code{primitive-load}.")
478 #define FUNC_NAME s_scm_current_load_port
480 return scm_fluid_ref (cur_loadport_fluid
);
484 SCM_DEFINE (scm_set_current_input_port
, "set-current-input-port", 1, 0, 0,
486 "@deffnx {Scheme Procedure} set-current-output-port port\n"
487 "@deffnx {Scheme Procedure} set-current-error-port port\n"
488 "Change the ports returned by @code{current-input-port},\n"
489 "@code{current-output-port} and @code{current-error-port}, respectively,\n"
490 "so that they use the supplied @var{port} for input or output.")
491 #define FUNC_NAME s_scm_set_current_input_port
493 SCM oinp
= scm_fluid_ref (cur_inport_fluid
);
494 SCM_VALIDATE_OPINPORT (1, port
);
495 scm_fluid_set_x (cur_inport_fluid
, port
);
501 SCM_DEFINE (scm_set_current_output_port
, "set-current-output-port", 1, 0, 0,
503 "Set the current default output port to @var{port}.")
504 #define FUNC_NAME s_scm_set_current_output_port
506 SCM ooutp
= scm_fluid_ref (cur_outport_fluid
);
507 port
= SCM_COERCE_OUTPORT (port
);
508 SCM_VALIDATE_OPOUTPORT (1, port
);
509 scm_fluid_set_x (cur_outport_fluid
, port
);
515 SCM_DEFINE (scm_set_current_error_port
, "set-current-error-port", 1, 0, 0,
517 "Set the current default error port to @var{port}.")
518 #define FUNC_NAME s_scm_set_current_error_port
520 SCM oerrp
= scm_fluid_ref (cur_errport_fluid
);
521 port
= SCM_COERCE_OUTPORT (port
);
522 SCM_VALIDATE_OPOUTPORT (1, port
);
523 scm_fluid_set_x (cur_errport_fluid
, port
);
530 scm_set_current_warning_port (SCM port
)
532 scm_i_pthread_once (¤t_warning_port_once
,
533 init_current_warning_port_var
);
534 return scm_call_1 (scm_variable_ref (current_warning_port_var
), port
);
539 scm_dynwind_current_input_port (SCM port
)
540 #define FUNC_NAME NULL
542 SCM_VALIDATE_OPINPORT (1, port
);
543 scm_dynwind_fluid (cur_inport_fluid
, port
);
548 scm_dynwind_current_output_port (SCM port
)
549 #define FUNC_NAME NULL
551 port
= SCM_COERCE_OUTPORT (port
);
552 SCM_VALIDATE_OPOUTPORT (1, port
);
553 scm_dynwind_fluid (cur_outport_fluid
, port
);
558 scm_dynwind_current_error_port (SCM port
)
559 #define FUNC_NAME NULL
561 port
= SCM_COERCE_OUTPORT (port
);
562 SCM_VALIDATE_OPOUTPORT (1, port
);
563 scm_dynwind_fluid (cur_errport_fluid
, port
);
568 scm_i_dynwind_current_load_port (SCM port
)
570 scm_dynwind_fluid (cur_loadport_fluid
, port
);
574 /* The port table --- an array of pointers to ports. */
577 We need a global registry of ports to flush them all at exit, and to
578 get all the ports matching a file descriptor.
580 SCM scm_i_port_weak_hash
;
582 scm_i_pthread_mutex_t scm_i_port_table_mutex
= SCM_I_PTHREAD_MUTEX_INITIALIZER
;
585 /* Port finalization. */
588 static void finalize_port (void *, void *);
590 /* Register a finalizer for PORT. */
591 static SCM_C_INLINE_KEYWORD
void
592 register_finalizer_for_port (SCM port
)
594 /* Register a finalizer for PORT so that its
595 type's `free' function gets called. */
596 scm_i_set_finalizer (SCM2PTR (port
), finalize_port
, NULL
);
599 /* Finalize the object (a port) pointed to by PTR. */
601 finalize_port (void *ptr
, void *data
)
604 SCM port
= PTR2SCM (ptr
);
606 if (!SCM_PORTP (port
))
609 if (SCM_OPENP (port
))
611 if (SCM_REVEALED (port
) > 0)
612 /* Keep "revealed" ports alive and re-register a finalizer. */
613 register_finalizer_for_port (port
);
616 port_type
= SCM_TC2PTOBNUM (SCM_CELL_TYPE (port
));
617 if (port_type
>= scm_numptob
)
620 if (scm_ptobs
[port_type
].free
)
621 /* Yes, I really do mean `.free' rather than `.close'. `.close'
622 is for explicit `close-port' by user. */
623 scm_ptobs
[port_type
].free (port
);
625 SCM_SETSTREAM (port
, 0);
626 SCM_CLR_PORT_OPEN_FLAG (port
);
628 scm_gc_ports_collected
++;
637 /* This function is not and should not be thread safe. */
639 scm_new_port_table_entry (scm_t_bits tag
)
640 #define FUNC_NAME "scm_new_port_table_entry"
643 We initialize the cell to empty, this is in case scm_gc_calloc
644 triggers GC ; we don't want the GC to scan a half-finished Z.
647 SCM z
= scm_cons (SCM_EOL
, SCM_EOL
);
648 scm_t_port
*entry
= scm_gc_typed_calloc (scm_t_port
);
649 scm_t_port_internal
*pti
= scm_gc_typed_calloc (scm_t_port_internal
);
650 const char *encoding
;
652 entry
->file_name
= SCM_BOOL_F
;
653 entry
->rw_active
= SCM_PORT_NEITHER
;
656 /* Initialize this port with the thread's current default
658 encoding
= scm_i_default_port_encoding ();
659 entry
->ilseq_handler
= scm_i_default_port_conversion_handler ();
660 entry
->encoding
= encoding
? scm_gc_strdup (encoding
, "port") : NULL
;
661 if (encoding
&& c_strcasecmp (encoding
, "UTF-8") == 0)
662 pti
->encoding_mode
= SCM_PORT_ENCODING_MODE_UTF8
;
664 pti
->encoding_mode
= SCM_PORT_ENCODING_MODE_ICONV
;
665 pti
->iconv_descriptors
= NULL
;
667 pti
->at_stream_start_for_bom_read
= 1;
668 pti
->at_stream_start_for_bom_write
= 1;
670 /* XXX These fields are not what they seem. They have been
671 repurposed, but cannot safely be renamed in 2.0 without breaking
672 ABI compatibility. This will be cleaned up in 2.2. */
673 entry
->input_cd
= pti
; /* XXX pointer to the internal port structure */
674 entry
->output_cd
= NULL
; /* XXX unused */
676 pti
->pending_eof
= 0;
677 pti
->alist
= SCM_EOL
;
679 /* Until Guile 2.0.9 included, 'setvbuf' would only work on file
680 ports. Now all port types can be supported, but it's not clear
681 that port types out in wild accept having someone else fiddle with
682 their buffer. Thus, conservatively turn it off by default. */
685 SCM_SET_CELL_TYPE (z
, tag
);
686 SCM_SETPTAB_ENTRY (z
, entry
);
688 scm_hashq_set_x (scm_i_port_weak_hash
, z
, SCM_BOOL_F
);
690 /* For each new port, register a finalizer so that it port type's free
691 function can be invoked eventually. */
692 register_finalizer_for_port (z
);
698 #if SCM_ENABLE_DEPRECATED==1
700 scm_add_to_port_table (SCM port
)
705 scm_c_issue_deprecation_warning ("scm_add_to_port_table is deprecated.");
707 scm_i_pthread_mutex_lock (&scm_i_port_table_mutex
);
708 z
= scm_new_port_table_entry (scm_tc7_port
);
709 pt
= SCM_PTAB_ENTRY(z
);
711 SCM_SETCAR (z
, SCM_EOL
);
712 SCM_SETCDR (z
, SCM_EOL
);
713 SCM_SETPTAB_ENTRY (port
, pt
);
714 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex
);
721 /* Remove a port from the table and destroy it. */
723 static void close_iconv_descriptors (scm_t_iconv_descriptors
*id
);
726 scm_i_remove_port (SCM port
)
727 #define FUNC_NAME "scm_remove_port"
730 scm_t_port_internal
*pti
;
732 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex
);
734 p
= SCM_PTAB_ENTRY (port
);
735 pti
= SCM_PORT_GET_INTERNAL (port
);
736 scm_port_non_buffer (p
);
737 p
->putback_buf
= NULL
;
738 p
->putback_buf_size
= 0;
740 if (pti
->iconv_descriptors
)
742 close_iconv_descriptors (pti
->iconv_descriptors
);
743 pti
->iconv_descriptors
= NULL
;
746 SCM_SETPTAB_ENTRY (port
, 0);
748 scm_hashq_remove_x (scm_i_port_weak_hash
, port
);
750 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex
);
755 /* Functions for debugging. */
757 SCM_DEFINE (scm_pt_size
, "pt-size", 0, 0, 0,
759 "Return the number of ports in the port table. @code{pt-size}\n"
760 "is only included in @code{--enable-guile-debug} builds.")
761 #define FUNC_NAME s_scm_pt_size
763 return scm_from_int (SCM_HASHTABLE_N_ITEMS (scm_i_port_weak_hash
));
769 scm_port_non_buffer (scm_t_port
*pt
)
771 pt
->read_pos
= pt
->read_buf
= pt
->read_end
= &pt
->shortbuf
;
772 pt
->write_buf
= pt
->write_pos
= &pt
->shortbuf
;
773 pt
->read_buf_size
= pt
->write_buf_size
= 1;
774 pt
->write_end
= pt
->write_buf
+ pt
->write_buf_size
;
778 /* Revealed counts --- an oddity inherited from SCSH. */
780 /* Find a port in the table and return its revealed count.
781 Also used by the garbage collector.
785 scm_revealed_count (SCM port
)
787 return SCM_REVEALED(port
);
792 /* Return the revealed count for a port. */
794 SCM_DEFINE (scm_port_revealed
, "port-revealed", 1, 0, 0,
796 "Return the revealed count for @var{port}.")
797 #define FUNC_NAME s_scm_port_revealed
799 port
= SCM_COERCE_OUTPORT (port
);
800 SCM_VALIDATE_OPENPORT (1, port
);
801 return scm_from_int (scm_revealed_count (port
));
805 /* Set the revealed count for a port. */
806 SCM_DEFINE (scm_set_port_revealed_x
, "set-port-revealed!", 2, 0, 0,
807 (SCM port
, SCM rcount
),
808 "Sets the revealed count for a port to a given value.\n"
809 "The return value is unspecified.")
810 #define FUNC_NAME s_scm_set_port_revealed_x
812 port
= SCM_COERCE_OUTPORT (port
);
813 SCM_VALIDATE_OPENPORT (1, port
);
814 SCM_REVEALED (port
) = scm_to_int (rcount
);
815 return SCM_UNSPECIFIED
;
821 /* Retrieving a port's mode. */
823 /* Return the flags that characterize a port based on the mode
824 * string used to open a file for that port.
826 * See PORT FLAGS in scm.h
830 scm_i_mode_bits_n (SCM modes
)
833 | (scm_i_string_contains_char (modes
, 'r')
834 || scm_i_string_contains_char (modes
, '+') ? SCM_RDNG
: 0)
835 | (scm_i_string_contains_char (modes
, 'w')
836 || scm_i_string_contains_char (modes
, 'a')
837 || scm_i_string_contains_char (modes
, '+') ? SCM_WRTNG
: 0)
838 | (scm_i_string_contains_char (modes
, '0') ? SCM_BUF0
: 0)
839 | (scm_i_string_contains_char (modes
, 'l') ? SCM_BUFLINE
: 0));
843 scm_mode_bits (char *modes
)
845 return scm_i_mode_bits (scm_from_locale_string (modes
));
849 scm_i_mode_bits (SCM modes
)
853 if (!scm_is_string (modes
))
854 scm_wrong_type_arg_msg (NULL
, 0, modes
, "string");
856 bits
= scm_i_mode_bits_n (modes
);
857 scm_remember_upto_here_1 (modes
);
861 /* Return the mode flags from an open port.
862 * Some modes such as "append" are only used when opening
863 * a file and are not returned here. */
865 SCM_DEFINE (scm_port_mode
, "port-mode", 1, 0, 0,
867 "Return the port modes associated with the open port @var{port}.\n"
868 "These will not necessarily be identical to the modes used when\n"
869 "the port was opened, since modes such as \"append\" which are\n"
870 "used only during port creation are not retained.")
871 #define FUNC_NAME s_scm_port_mode
876 port
= SCM_COERCE_OUTPORT (port
);
877 SCM_VALIDATE_OPPORT (1, port
);
878 if (SCM_CELL_WORD_0 (port
) & SCM_RDNG
) {
879 if (SCM_CELL_WORD_0 (port
) & SCM_WRTNG
)
880 strcpy (modes
, "r+");
884 else if (SCM_CELL_WORD_0 (port
) & SCM_WRTNG
)
886 if (SCM_CELL_WORD_0 (port
) & SCM_BUF0
)
888 return scm_from_locale_string (modes
);
897 * Call the close operation on a port object.
898 * see also scm_close.
900 SCM_DEFINE (scm_close_port
, "close-port", 1, 0, 0,
902 "Close the specified port object. Return @code{#t} if it\n"
903 "successfully closes a port or @code{#f} if it was already\n"
904 "closed. An exception may be raised if an error occurs, for\n"
905 "example when flushing buffered output. See also @ref{Ports and\n"
906 "File Descriptors, close}, for a procedure which can close file\n"
908 #define FUNC_NAME s_scm_close_port
913 port
= SCM_COERCE_OUTPORT (port
);
915 SCM_VALIDATE_PORT (1, port
);
916 if (SCM_CLOSEDP (port
))
918 i
= SCM_PTOBNUM (port
);
919 if (scm_ptobs
[i
].close
)
920 rv
= (scm_ptobs
[i
].close
) (port
);
923 scm_i_remove_port (port
);
924 SCM_CLR_PORT_OPEN_FLAG (port
);
925 return scm_from_bool (rv
>= 0);
929 SCM_DEFINE (scm_close_input_port
, "close-input-port", 1, 0, 0,
931 "Close the specified input port object. The routine has no effect if\n"
932 "the file has already been closed. An exception may be raised if an\n"
933 "error occurs. The value returned is unspecified.\n\n"
934 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
935 "which can close file descriptors.")
936 #define FUNC_NAME s_scm_close_input_port
938 SCM_VALIDATE_INPUT_PORT (1, port
);
939 scm_close_port (port
);
940 return SCM_UNSPECIFIED
;
944 SCM_DEFINE (scm_close_output_port
, "close-output-port", 1, 0, 0,
946 "Close the specified output port object. The routine has no effect if\n"
947 "the file has already been closed. An exception may be raised if an\n"
948 "error occurs. The value returned is unspecified.\n\n"
949 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
950 "which can close file descriptors.")
951 #define FUNC_NAME s_scm_close_output_port
953 port
= SCM_COERCE_OUTPORT (port
);
954 SCM_VALIDATE_OUTPUT_PORT (1, port
);
955 scm_close_port (port
);
956 return SCM_UNSPECIFIED
;
961 collect_keys (void *unused
, SCM key
, SCM value
, SCM result
)
963 return scm_cons (key
, result
);
967 scm_c_port_for_each (void (*proc
)(void *data
, SCM p
), void *data
)
971 /* Copy out the port table as a list so that we get strong references
972 to all the values. */
973 scm_i_pthread_mutex_lock (&scm_i_port_table_mutex
);
974 ports
= scm_internal_hash_fold (collect_keys
, NULL
,
975 SCM_EOL
, scm_i_port_weak_hash
);
976 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex
);
978 for (; scm_is_pair (ports
); ports
= scm_cdr (ports
))
980 SCM p
= scm_car (ports
);
986 SCM_DEFINE (scm_port_for_each
, "port-for-each", 1, 0, 0,
988 "Apply @var{proc} to each port in the Guile port table\n"
989 "in turn. The return value is unspecified. More specifically,\n"
990 "@var{proc} is applied exactly once to every port that exists\n"
991 "in the system at the time @code{port-for-each} is invoked.\n"
992 "Changes to the port table while @code{port-for-each} is running\n"
993 "have no effect as far as @code{port-for-each} is concerned.")
994 #define FUNC_NAME s_scm_port_for_each
998 SCM_VALIDATE_PROC (1, proc
);
1000 /* Copy out the port table as a list so that we get strong references
1001 to all the values. */
1002 scm_i_pthread_mutex_lock (&scm_i_port_table_mutex
);
1003 ports
= scm_internal_hash_fold (collect_keys
, NULL
,
1004 SCM_EOL
, scm_i_port_weak_hash
);
1005 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex
);
1007 for (; scm_is_pair (ports
); ports
= scm_cdr (ports
))
1008 if (SCM_PORTP (SCM_CAR (ports
)))
1009 scm_call_1 (proc
, SCM_CAR (ports
));
1011 return SCM_UNSPECIFIED
;
1017 /* Utter miscellany. Gosh, we should clean this up some time. */
1019 SCM_DEFINE (scm_input_port_p
, "input-port?", 1, 0, 0,
1021 "Return @code{#t} if @var{x} is an input port, otherwise return\n"
1022 "@code{#f}. Any object satisfying this predicate also satisfies\n"
1024 #define FUNC_NAME s_scm_input_port_p
1026 return scm_from_bool (SCM_INPUT_PORT_P (x
));
1030 SCM_DEFINE (scm_output_port_p
, "output-port?", 1, 0, 0,
1032 "Return @code{#t} if @var{x} is an output port, otherwise return\n"
1033 "@code{#f}. Any object satisfying this predicate also satisfies\n"
1035 #define FUNC_NAME s_scm_output_port_p
1037 x
= SCM_COERCE_OUTPORT (x
);
1038 return scm_from_bool (SCM_OUTPUT_PORT_P (x
));
1042 SCM_DEFINE (scm_port_p
, "port?", 1, 0, 0,
1044 "Return a boolean indicating whether @var{x} is a port.\n"
1045 "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n"
1047 #define FUNC_NAME s_scm_port_p
1049 return scm_from_bool (SCM_PORTP (x
));
1053 SCM_DEFINE (scm_port_closed_p
, "port-closed?", 1, 0, 0,
1055 "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n"
1057 #define FUNC_NAME s_scm_port_closed_p
1059 SCM_VALIDATE_PORT (1, port
);
1060 return scm_from_bool (!SCM_OPPORTP (port
));
1064 SCM_DEFINE (scm_eof_object_p
, "eof-object?", 1, 0, 0,
1066 "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n"
1067 "return @code{#f}.")
1068 #define FUNC_NAME s_scm_eof_object_p
1070 return scm_from_bool(SCM_EOF_OBJECT_P (x
));
1074 SCM_DEFINE (scm_force_output
, "force-output", 0, 1, 0,
1076 "Flush the specified output port, or the current output port if @var{port}\n"
1077 "is omitted. The current output buffer contents are passed to the\n"
1078 "underlying port implementation (e.g., in the case of fports, the\n"
1079 "data will be written to the file and the output buffer will be cleared.)\n"
1080 "It has no effect on an unbuffered port.\n\n"
1081 "The return value is unspecified.")
1082 #define FUNC_NAME s_scm_force_output
1084 if (SCM_UNBNDP (port
))
1085 port
= scm_current_output_port ();
1088 port
= SCM_COERCE_OUTPORT (port
);
1089 SCM_VALIDATE_OPOUTPORT (1, port
);
1092 return SCM_UNSPECIFIED
;
1098 flush_output_port (void *closure
, SCM port
)
1100 if (SCM_OPOUTPORTP (port
))
1104 SCM_DEFINE (scm_flush_all_ports
, "flush-all-ports", 0, 0, 0,
1106 "Equivalent to calling @code{force-output} on\n"
1107 "all open output ports. The return value is unspecified.")
1108 #define FUNC_NAME s_scm_flush_all_ports
1110 scm_c_port_for_each (&flush_output_port
, NULL
);
1111 return SCM_UNSPECIFIED
;
1115 SCM_DEFINE (scm_read_char
, "read-char", 0, 1, 0,
1117 "Return the next character available from @var{port}, updating\n"
1118 "@var{port} to point to the following character. If no more\n"
1119 "characters are available, the end-of-file object is returned.\n"
1121 "When @var{port}'s data cannot be decoded according to its\n"
1122 "character encoding, a @code{decoding-error} is raised and\n"
1123 "@var{port} points past the erroneous byte sequence.\n")
1124 #define FUNC_NAME s_scm_read_char
1127 if (SCM_UNBNDP (port
))
1128 port
= scm_current_input_port ();
1129 SCM_VALIDATE_OPINPORT (1, port
);
1130 c
= scm_getc (port
);
1133 return SCM_MAKE_CHAR (c
);
1137 /* Update the line and column number of PORT after consumption of C. */
1139 update_port_lf (scm_t_wchar c
, SCM port
)
1164 #define SCM_MBCHAR_BUF_SIZE (4)
1166 /* Convert the SIZE-byte UTF-8 sequence in UTF8_BUF to a codepoint.
1167 UTF8_BUF is assumed to contain a valid UTF-8 sequence. */
1169 utf8_to_codepoint (const scm_t_uint8
*utf8_buf
, size_t size
)
1171 scm_t_wchar codepoint
;
1173 if (utf8_buf
[0] <= 0x7f)
1176 codepoint
= utf8_buf
[0];
1178 else if ((utf8_buf
[0] & 0xe0) == 0xc0)
1181 codepoint
= ((scm_t_wchar
) utf8_buf
[0] & 0x1f) << 6UL
1182 | (utf8_buf
[1] & 0x3f);
1184 else if ((utf8_buf
[0] & 0xf0) == 0xe0)
1187 codepoint
= ((scm_t_wchar
) utf8_buf
[0] & 0x0f) << 12UL
1188 | ((scm_t_wchar
) utf8_buf
[1] & 0x3f) << 6UL
1189 | (utf8_buf
[2] & 0x3f);
1194 codepoint
= ((scm_t_wchar
) utf8_buf
[0] & 0x07) << 18UL
1195 | ((scm_t_wchar
) utf8_buf
[1] & 0x3f) << 12UL
1196 | ((scm_t_wchar
) utf8_buf
[2] & 0x3f) << 6UL
1197 | (utf8_buf
[3] & 0x3f);
1203 /* Read a UTF-8 sequence from PORT. On success, return 0 and set
1204 *CODEPOINT to the codepoint that was read, fill BUF with its UTF-8
1205 representation, and set *LEN to the length in bytes. Return
1206 `EILSEQ' on error. */
1208 get_utf8_codepoint (SCM port
, scm_t_wchar
*codepoint
,
1209 scm_t_uint8 buf
[SCM_MBCHAR_BUF_SIZE
], size_t *len
)
1211 #define ASSERT_NOT_EOF(b) \
1212 if (SCM_UNLIKELY ((b) == EOF)) \
1214 #define CONSUME_PEEKED_BYTE() \
1221 pt
= SCM_PTAB_ENTRY (port
);
1223 byte
= scm_get_byte_or_eof (port
);
1230 buf
[0] = (scm_t_uint8
) byte
;
1235 *codepoint
= buf
[0];
1236 else if (buf
[0] >= 0xc2 && buf
[0] <= 0xdf)
1239 byte
= scm_peek_byte_or_eof (port
);
1240 ASSERT_NOT_EOF (byte
);
1242 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80))
1245 CONSUME_PEEKED_BYTE ();
1246 buf
[1] = (scm_t_uint8
) byte
;
1249 *codepoint
= ((scm_t_wchar
) buf
[0] & 0x1f) << 6UL
1252 else if ((buf
[0] & 0xf0) == 0xe0)
1255 byte
= scm_peek_byte_or_eof (port
);
1256 ASSERT_NOT_EOF (byte
);
1258 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80
1259 || (buf
[0] == 0xe0 && byte
< 0xa0)
1260 || (buf
[0] == 0xed && byte
> 0x9f)))
1263 CONSUME_PEEKED_BYTE ();
1264 buf
[1] = (scm_t_uint8
) byte
;
1267 byte
= scm_peek_byte_or_eof (port
);
1268 ASSERT_NOT_EOF (byte
);
1270 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80))
1273 CONSUME_PEEKED_BYTE ();
1274 buf
[2] = (scm_t_uint8
) byte
;
1277 *codepoint
= ((scm_t_wchar
) buf
[0] & 0x0f) << 12UL
1278 | ((scm_t_wchar
) buf
[1] & 0x3f) << 6UL
1281 else if (buf
[0] >= 0xf0 && buf
[0] <= 0xf4)
1284 byte
= scm_peek_byte_or_eof (port
);
1285 ASSERT_NOT_EOF (byte
);
1287 if (SCM_UNLIKELY (((byte
& 0xc0) != 0x80)
1288 || (buf
[0] == 0xf0 && byte
< 0x90)
1289 || (buf
[0] == 0xf4 && byte
> 0x8f)))
1292 CONSUME_PEEKED_BYTE ();
1293 buf
[1] = (scm_t_uint8
) byte
;
1296 byte
= scm_peek_byte_or_eof (port
);
1297 ASSERT_NOT_EOF (byte
);
1299 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80))
1302 CONSUME_PEEKED_BYTE ();
1303 buf
[2] = (scm_t_uint8
) byte
;
1306 byte
= scm_peek_byte_or_eof (port
);
1307 ASSERT_NOT_EOF (byte
);
1309 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80))
1312 CONSUME_PEEKED_BYTE ();
1313 buf
[3] = (scm_t_uint8
) byte
;
1316 *codepoint
= ((scm_t_wchar
) buf
[0] & 0x07) << 18UL
1317 | ((scm_t_wchar
) buf
[1] & 0x3f) << 12UL
1318 | ((scm_t_wchar
) buf
[2] & 0x3f) << 6UL
1327 /* Here we could choose the consume the faulty byte when it's not a
1328 valid starting byte, but it's not a requirement. What Section 3.9
1329 of Unicode 6.0.0 mandates, though, is to not consume a byte that
1330 would otherwise be a valid starting byte. */
1334 #undef CONSUME_PEEKED_BYTE
1335 #undef ASSERT_NOT_EOF
1338 /* Likewise, read a byte sequence from PORT, passing it through its
1339 input conversion descriptor. */
1341 get_iconv_codepoint (SCM port
, scm_t_wchar
*codepoint
,
1342 char buf
[SCM_MBCHAR_BUF_SIZE
], size_t *len
)
1344 scm_t_iconv_descriptors
*id
;
1345 scm_t_uint8 utf8_buf
[SCM_MBCHAR_BUF_SIZE
];
1346 size_t input_size
= 0;
1348 id
= scm_i_port_iconv_descriptors (port
, SCM_PORT_READ
);
1353 char *input
, *output
;
1354 size_t input_left
, output_left
, done
;
1356 byte_read
= scm_get_byte_or_eof (port
);
1357 if (SCM_UNLIKELY (byte_read
== EOF
))
1359 if (SCM_LIKELY (input_size
== 0))
1361 *codepoint
= (scm_t_wchar
) EOF
;
1367 /* EOF found in the middle of a multibyte character. */
1368 scm_i_set_pending_eof (port
);
1373 buf
[input_size
++] = byte_read
;
1376 input_left
= input_size
;
1377 output
= (char *) utf8_buf
;
1378 output_left
= sizeof (utf8_buf
);
1380 done
= iconv (id
->input_cd
, &input
, &input_left
, &output
, &output_left
);
1382 if (done
== (size_t) -1)
1385 if (SCM_LIKELY (err
== EINVAL
))
1386 /* The input byte sequence did not form a complete
1387 character. Read another byte and try again. */
1394 size_t output_size
= sizeof (utf8_buf
) - output_left
;
1395 if (SCM_LIKELY (output_size
> 0))
1397 /* iconv generated output. Convert the UTF8_BUF sequence
1398 to a Unicode code point. */
1399 *codepoint
= utf8_to_codepoint (utf8_buf
, output_size
);
1405 /* iconv consumed some bytes without producing any output.
1406 Most likely this means that a Unicode byte-order mark
1407 (BOM) was consumed, which should not be included in the
1408 returned buf. Shift any remaining bytes to the beginning
1409 of buf, and continue the loop. */
1410 memmove (buf
, input
, input_left
);
1411 input_size
= input_left
;
1418 /* Read a codepoint from PORT and return it in *CODEPOINT. Fill BUF
1419 with the byte representation of the codepoint in PORT's encoding, and
1420 set *LEN to the length in bytes of that representation. Return 0 on
1421 success and an errno value on error. */
1423 get_codepoint (SCM port
, scm_t_wchar
*codepoint
,
1424 char buf
[SCM_MBCHAR_BUF_SIZE
], size_t *len
)
1427 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1428 scm_t_port_internal
*pti
= SCM_PORT_GET_INTERNAL (port
);
1430 if (pti
->encoding_mode
== SCM_PORT_ENCODING_MODE_UTF8
)
1431 err
= get_utf8_codepoint (port
, codepoint
, (scm_t_uint8
*) buf
, len
);
1433 err
= get_iconv_codepoint (port
, codepoint
, buf
, len
);
1435 if (SCM_LIKELY (err
== 0))
1437 if (SCM_UNLIKELY (pti
->at_stream_start_for_bom_read
))
1439 /* Record that we're no longer at stream start. */
1440 pti
->at_stream_start_for_bom_read
= 0;
1442 pti
->at_stream_start_for_bom_write
= 0;
1444 /* If we just read a BOM in an encoding that recognizes them,
1445 then silently consume it and read another code point. */
1447 (*codepoint
== SCM_UNICODE_BOM
1448 && (pti
->encoding_mode
== SCM_PORT_ENCODING_MODE_UTF8
1449 || c_strcasecmp (pt
->encoding
, "UTF-16") == 0
1450 || c_strcasecmp (pt
->encoding
, "UTF-32") == 0)))
1451 return get_codepoint (port
, codepoint
, buf
, len
);
1453 update_port_lf (*codepoint
, port
);
1455 else if (pt
->ilseq_handler
== SCM_ICONVEH_QUESTION_MARK
)
1459 update_port_lf (*codepoint
, port
);
1465 /* Read a codepoint from PORT and return it. */
1468 #define FUNC_NAME "scm_getc"
1472 scm_t_wchar codepoint
;
1473 char buf
[SCM_MBCHAR_BUF_SIZE
];
1475 err
= get_codepoint (port
, &codepoint
, buf
, &len
);
1476 if (SCM_UNLIKELY (err
!= 0))
1477 /* At this point PORT should point past the invalid encoding, as per
1478 R6RS-lib Section 8.2.4. */
1479 scm_decoding_error (FUNC_NAME
, err
, "input decoding error", port
);
1485 /* this should only be called when the read buffer is empty. it
1486 tries to refill the read buffer. it returns the first char from
1487 the port, which is either EOF or *(pt->read_pos). */
1489 scm_i_fill_input (SCM port
)
1491 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1492 scm_t_port_internal
*pti
= SCM_PORT_GET_INTERNAL (port
);
1494 assert (pt
->read_pos
== pt
->read_end
);
1496 if (pti
->pending_eof
)
1498 pti
->pending_eof
= 0;
1502 if (pt
->read_buf
== pt
->putback_buf
)
1504 /* finished reading put-back chars. */
1505 pt
->read_buf
= pt
->saved_read_buf
;
1506 pt
->read_pos
= pt
->saved_read_pos
;
1507 pt
->read_end
= pt
->saved_read_end
;
1508 pt
->read_buf_size
= pt
->saved_read_buf_size
;
1509 if (pt
->read_pos
< pt
->read_end
)
1510 return *(pt
->read_pos
);
1512 return scm_ptobs
[SCM_PTOBNUM (port
)].fill_input (port
);
1516 scm_fill_input (SCM port
)
1518 return scm_i_fill_input (port
);
1521 /* Slow-path fallback for 'scm_get_byte_or_eof' in inline.h */
1523 scm_slow_get_byte_or_eof (SCM port
)
1525 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1527 if (pt
->rw_active
== SCM_PORT_WRITE
)
1531 pt
->rw_active
= SCM_PORT_READ
;
1533 if (pt
->read_pos
>= pt
->read_end
)
1535 if (SCM_UNLIKELY (scm_i_fill_input (port
) == EOF
))
1539 return *pt
->read_pos
++;
1542 /* Slow-path fallback for 'scm_peek_byte_or_eof' in inline.h */
1544 scm_slow_peek_byte_or_eof (SCM port
)
1546 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1548 if (pt
->rw_active
== SCM_PORT_WRITE
)
1552 pt
->rw_active
= SCM_PORT_READ
;
1554 if (pt
->read_pos
>= pt
->read_end
)
1556 if (SCM_UNLIKELY (scm_i_fill_input (port
) == EOF
))
1558 scm_i_set_pending_eof (port
);
1563 return *pt
->read_pos
;
1569 * This function differs from scm_c_write; it updates port line and
1573 scm_lfwrite (const char *ptr
, size_t size
, SCM port
)
1575 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1576 scm_t_ptob_descriptor
*ptob
= &scm_ptobs
[SCM_PTOBNUM (port
)];
1578 if (pt
->rw_active
== SCM_PORT_READ
)
1579 scm_end_input (port
);
1581 ptob
->write (port
, ptr
, size
);
1583 for (; size
; ptr
++, size
--)
1584 update_port_lf ((scm_t_wchar
) (unsigned char) *ptr
, port
);
1587 pt
->rw_active
= SCM_PORT_WRITE
;
1590 /* Write STR to PORT from START inclusive to END exclusive. */
1592 scm_lfwrite_substr (SCM str
, size_t start
, size_t end
, SCM port
)
1594 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1596 if (pt
->rw_active
== SCM_PORT_READ
)
1597 scm_end_input (port
);
1599 if (end
== (size_t) -1)
1600 end
= scm_i_string_length (str
);
1602 scm_i_display_substring (str
, start
, end
, port
);
1605 pt
->rw_active
= SCM_PORT_WRITE
;
1610 * Used by an application to read arbitrary number of bytes from an
1611 * SCM port. Same semantics as libc read, except that scm_c_read only
1612 * returns less than SIZE bytes if at end-of-file.
1614 * Warning: Doesn't update port line and column counts! */
1616 /* This structure, and the following swap_buffer function, are used
1617 for temporarily swapping a port's own read buffer, and the buffer
1618 that the caller of scm_c_read provides. */
1619 struct port_and_swap_buffer
1622 unsigned char *buffer
;
1627 swap_buffer (void *data
)
1629 struct port_and_swap_buffer
*psb
= (struct port_and_swap_buffer
*) data
;
1630 unsigned char *old_buf
= psb
->pt
->read_buf
;
1631 size_t old_size
= psb
->pt
->read_buf_size
;
1633 /* Make the port use (buffer, size) from the struct. */
1634 psb
->pt
->read_pos
= psb
->pt
->read_buf
= psb
->pt
->read_end
= psb
->buffer
;
1635 psb
->pt
->read_buf_size
= psb
->size
;
1637 /* Save the port's old (buffer, size) in the struct. */
1638 psb
->buffer
= old_buf
;
1639 psb
->size
= old_size
;
1643 scm_c_read (SCM port
, void *buffer
, size_t size
)
1644 #define FUNC_NAME "scm_c_read"
1647 size_t n_read
= 0, n_available
;
1648 struct port_and_swap_buffer psb
;
1650 SCM_VALIDATE_OPINPORT (1, port
);
1652 pt
= SCM_PTAB_ENTRY (port
);
1653 if (pt
->rw_active
== SCM_PORT_WRITE
)
1654 scm_ptobs
[SCM_PTOBNUM (port
)].flush (port
);
1657 pt
->rw_active
= SCM_PORT_READ
;
1659 /* Take bytes first from the port's read buffer. */
1660 if (pt
->read_pos
< pt
->read_end
)
1662 n_available
= min (size
, pt
->read_end
- pt
->read_pos
);
1663 memcpy (buffer
, pt
->read_pos
, n_available
);
1664 buffer
= (char *) buffer
+ n_available
;
1665 pt
->read_pos
+= n_available
;
1666 n_read
+= n_available
;
1667 size
-= n_available
;
1670 /* Avoid the scm_dynwind_* costs if we now have enough data. */
1674 /* Now we will call scm_i_fill_input repeatedly until we have read the
1675 requested number of bytes. (Note that a single scm_i_fill_input
1676 call does not guarantee to fill the whole of the port's read
1678 if (pt
->read_buf_size
<= 1 &&
1679 (pt
->encoding
== NULL
1680 || c_strcasecmp (pt
->encoding
, "ISO-8859-1") == 0))
1682 /* The port that we are reading from is unbuffered - i.e. does
1683 not have its own persistent buffer - but we have a buffer,
1684 provided by our caller, that is the right size for the data
1685 that is wanted. For the following scm_i_fill_input calls,
1686 therefore, we use the buffer in hand as the port's read
1689 We need to make sure that the port's normal (1 byte) buffer
1690 is reinstated in case one of the scm_i_fill_input () calls
1691 throws an exception; we use the scm_dynwind_* API to achieve
1694 A consequence of this optimization is that the fill_input
1695 functions can't unget characters. That'll push data to the
1696 pushback buffer instead of this psb buffer. */
1698 unsigned char *pback
= pt
->putback_buf
;
1701 psb
.buffer
= buffer
;
1703 scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE
);
1704 scm_dynwind_rewind_handler (swap_buffer
, &psb
, SCM_F_WIND_EXPLICITLY
);
1705 scm_dynwind_unwind_handler (swap_buffer
, &psb
, SCM_F_WIND_EXPLICITLY
);
1707 /* Call scm_i_fill_input until we have all the bytes that we need,
1709 while (pt
->read_buf_size
&& (scm_i_fill_input (port
) != EOF
))
1711 pt
->read_buf_size
-= (pt
->read_end
- pt
->read_pos
);
1712 pt
->read_pos
= pt
->read_buf
= pt
->read_end
;
1715 if (pback
!= pt
->putback_buf
1716 || pt
->read_buf
- (unsigned char *) buffer
< 0)
1717 scm_misc_error (FUNC_NAME
,
1718 "scm_c_read must not call a fill function that pushes "
1719 "back characters onto an unbuffered port", SCM_EOL
);
1721 n_read
+= pt
->read_buf
- (unsigned char *) buffer
;
1723 /* Reinstate the port's normal buffer. */
1728 /* The port has its own buffer. It is important that we use it,
1729 even if it happens to be smaller than our caller's buffer, so
1730 that a custom port implementation's entry points (in
1731 particular, fill_input) can rely on the buffer always being
1732 the same as they first set up. */
1733 while (size
&& (scm_i_fill_input (port
) != EOF
))
1735 n_available
= min (size
, pt
->read_end
- pt
->read_pos
);
1736 memcpy (buffer
, pt
->read_pos
, n_available
);
1737 buffer
= (char *) buffer
+ n_available
;
1738 pt
->read_pos
+= n_available
;
1739 n_read
+= n_available
;
1740 size
-= n_available
;
1750 * Used by an application to write arbitrary number of bytes to an SCM
1751 * port. Similar semantics as libc write. However, unlike libc
1752 * write, scm_c_write writes the requested number of bytes and has no
1755 * Warning: Doesn't update port line and column counts!
1759 scm_c_write (SCM port
, const void *ptr
, size_t size
)
1760 #define FUNC_NAME "scm_c_write"
1763 scm_t_ptob_descriptor
*ptob
;
1765 SCM_VALIDATE_OPOUTPORT (1, port
);
1767 pt
= SCM_PTAB_ENTRY (port
);
1768 ptob
= &scm_ptobs
[SCM_PTOBNUM (port
)];
1770 if (pt
->rw_active
== SCM_PORT_READ
)
1771 scm_end_input (port
);
1773 ptob
->write (port
, ptr
, size
);
1776 pt
->rw_active
= SCM_PORT_WRITE
;
1781 scm_flush (SCM port
)
1783 long i
= SCM_PTOBNUM (port
);
1785 (scm_ptobs
[i
].flush
) (port
);
1789 scm_end_input (SCM port
)
1792 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1794 scm_i_clear_pending_eof (port
);
1795 if (pt
->read_buf
== pt
->putback_buf
)
1797 offset
= pt
->read_end
- pt
->read_pos
;
1798 pt
->read_buf
= pt
->saved_read_buf
;
1799 pt
->read_pos
= pt
->saved_read_pos
;
1800 pt
->read_end
= pt
->saved_read_end
;
1801 pt
->read_buf_size
= pt
->saved_read_buf_size
;
1806 scm_ptobs
[SCM_PTOBNUM (port
)].end_input (port
, offset
);
1813 scm_i_unget_bytes (const unsigned char *buf
, size_t len
, SCM port
)
1814 #define FUNC_NAME "scm_unget_bytes"
1816 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1817 size_t old_len
, new_len
;
1819 scm_i_clear_pending_eof (port
);
1821 if (pt
->read_buf
!= pt
->putback_buf
)
1822 /* switch to the put-back buffer. */
1824 if (pt
->putback_buf
== NULL
)
1826 pt
->putback_buf_size
= (len
> SCM_INITIAL_PUTBACK_BUF_SIZE
1827 ? len
: SCM_INITIAL_PUTBACK_BUF_SIZE
);
1829 = (unsigned char *) scm_gc_malloc_pointerless
1830 (pt
->putback_buf_size
, "putback buffer");
1833 pt
->saved_read_buf
= pt
->read_buf
;
1834 pt
->saved_read_pos
= pt
->read_pos
;
1835 pt
->saved_read_end
= pt
->read_end
;
1836 pt
->saved_read_buf_size
= pt
->read_buf_size
;
1838 /* Put read_pos at the end of the buffer, so that ungets will not
1839 have to shift the buffer contents each time. */
1840 pt
->read_buf
= pt
->putback_buf
;
1841 pt
->read_pos
= pt
->read_end
= pt
->putback_buf
+ pt
->putback_buf_size
;
1842 pt
->read_buf_size
= pt
->putback_buf_size
;
1845 old_len
= pt
->read_end
- pt
->read_pos
;
1846 new_len
= old_len
+ len
;
1848 if (new_len
> pt
->read_buf_size
)
1849 /* The putback buffer needs to be enlarged. */
1851 size_t new_buf_size
;
1852 unsigned char *new_buf
, *new_end
, *new_pos
;
1854 new_buf_size
= pt
->read_buf_size
* 2;
1855 if (new_buf_size
< new_len
)
1856 new_buf_size
= new_len
;
1858 new_buf
= (unsigned char *)
1859 scm_gc_malloc_pointerless (new_buf_size
, "putback buffer");
1861 /* Put the bytes at the end of the buffer, so that future
1862 ungets won't need to shift the buffer. */
1863 new_end
= new_buf
+ new_buf_size
;
1864 new_pos
= new_end
- old_len
;
1865 memcpy (new_pos
, pt
->read_pos
, old_len
);
1867 pt
->read_buf
= pt
->putback_buf
= new_buf
;
1868 pt
->read_pos
= new_pos
;
1869 pt
->read_end
= new_end
;
1870 pt
->read_buf_size
= pt
->putback_buf_size
= new_buf_size
;
1872 else if (pt
->read_buf
+ len
< pt
->read_pos
)
1873 /* If needed, shift the existing buffer contents up.
1874 This should not happen unless some external code
1875 manipulates the putback buffer pointers. */
1877 unsigned char *new_end
= pt
->read_buf
+ pt
->read_buf_size
;
1878 unsigned char *new_pos
= new_end
- old_len
;
1880 memmove (new_pos
, pt
->read_pos
, old_len
);
1881 pt
->read_pos
= new_pos
;
1882 pt
->read_end
= new_end
;
1885 /* Move read_pos back and copy the bytes there. */
1886 pt
->read_pos
-= len
;
1887 memcpy (pt
->read_buf
+ (pt
->read_pos
- pt
->read_buf
), buf
, len
);
1889 if (pt
->rw_active
== SCM_PORT_WRITE
)
1893 pt
->rw_active
= SCM_PORT_READ
;
1898 scm_unget_bytes (const unsigned char *buf
, size_t len
, SCM port
)
1900 scm_i_unget_bytes (buf
, len
, port
);
1904 scm_unget_byte (int c
, SCM port
)
1909 scm_i_unget_bytes (&byte
, 1, port
);
1913 scm_ungetc (scm_t_wchar c
, SCM port
)
1914 #define FUNC_NAME "scm_ungetc"
1916 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1918 char result_buf
[10];
1919 const char *encoding
;
1922 if (pt
->encoding
!= NULL
)
1923 encoding
= pt
->encoding
;
1925 encoding
= "ISO-8859-1";
1927 len
= sizeof (result_buf
);
1928 result
= u32_conv_to_encoding (encoding
,
1929 (enum iconv_ilseq_handler
) pt
->ilseq_handler
,
1930 (uint32_t *) &c
, 1, NULL
,
1933 if (SCM_UNLIKELY (result
== NULL
|| len
== 0))
1934 scm_encoding_error (FUNC_NAME
, errno
,
1935 "conversion to port encoding failed",
1936 SCM_BOOL_F
, SCM_MAKE_CHAR (c
));
1938 scm_i_unget_bytes ((unsigned char *) result
, len
, port
);
1940 if (SCM_UNLIKELY (result
!= result_buf
))
1945 /* What should col be in this case?
1946 * We'll leave it at -1.
1948 SCM_LINUM (port
) -= 1;
1957 scm_ungets (const char *s
, int n
, SCM port
)
1959 /* This is simple minded and inefficient, but unreading strings is
1960 * probably not a common operation, and remember that line and
1961 * column numbers have to be handled...
1963 * Please feel free to write an optimized version!
1966 scm_ungetc (s
[n
], port
);
1970 SCM_DEFINE (scm_peek_char
, "peek-char", 0, 1, 0,
1972 "Return the next character available from @var{port},\n"
1973 "@emph{without} updating @var{port} to point to the following\n"
1974 "character. If no more characters are available, the\n"
1975 "end-of-file object is returned.\n"
1977 "The value returned by\n"
1978 "a call to @code{peek-char} is the same as the value that would\n"
1979 "have been returned by a call to @code{read-char} on the same\n"
1980 "port. The only difference is that the very next call to\n"
1981 "@code{read-char} or @code{peek-char} on that @var{port} will\n"
1982 "return the value returned by the preceding call to\n"
1983 "@code{peek-char}. In particular, a call to @code{peek-char} on\n"
1984 "an interactive port will hang waiting for input whenever a call\n"
1985 "to @code{read-char} would have hung.\n"
1987 "As for @code{read-char}, a @code{decoding-error} may be raised\n"
1988 "if such a situation occurs. However, unlike with @code{read-char},\n"
1989 "@var{port} still points at the beginning of the erroneous byte\n"
1990 "sequence when the error is raised.\n")
1991 #define FUNC_NAME s_scm_peek_char
1996 char bytes
[SCM_MBCHAR_BUF_SIZE
];
2000 if (SCM_UNBNDP (port
))
2001 port
= scm_current_input_port ();
2002 SCM_VALIDATE_OPINPORT (1, port
);
2004 column
= SCM_COL (port
);
2005 line
= SCM_LINUM (port
);
2007 err
= get_codepoint (port
, &c
, bytes
, &len
);
2009 scm_i_unget_bytes ((unsigned char *) bytes
, len
, port
);
2011 SCM_COL (port
) = column
;
2012 SCM_LINUM (port
) = line
;
2014 if (SCM_UNLIKELY (err
!= 0))
2016 scm_decoding_error (FUNC_NAME
, err
, "input decoding error", port
);
2018 /* Shouldn't happen since `catch' always aborts to prompt. */
2019 result
= SCM_BOOL_F
;
2023 scm_i_set_pending_eof (port
);
2024 result
= SCM_EOF_VAL
;
2027 result
= SCM_MAKE_CHAR (c
);
2033 SCM_DEFINE (scm_unread_char
, "unread-char", 1, 1, 0,
2034 (SCM cobj
, SCM port
),
2035 "Place character @var{cobj} in @var{port} so that it will be\n"
2036 "read by the next read operation. If called multiple times, the\n"
2037 "unread characters will be read again in last-in first-out\n"
2038 "order. If @var{port} is not supplied, the current input port\n"
2040 #define FUNC_NAME s_scm_unread_char
2044 SCM_VALIDATE_CHAR (1, cobj
);
2045 if (SCM_UNBNDP (port
))
2046 port
= scm_current_input_port ();
2047 SCM_VALIDATE_OPINPORT (2, port
);
2049 c
= SCM_CHAR (cobj
);
2051 scm_ungetc (c
, port
);
2056 SCM_DEFINE (scm_unread_string
, "unread-string", 2, 0, 0,
2057 (SCM str
, SCM port
),
2058 "Place the string @var{str} in @var{port} so that its characters will be\n"
2059 "read in subsequent read operations. If called multiple times, the\n"
2060 "unread characters will be read again in last-in first-out order. If\n"
2061 "@var{port} is not supplied, the current-input-port is used.")
2062 #define FUNC_NAME s_scm_unread_string
2065 SCM_VALIDATE_STRING (1, str
);
2066 if (SCM_UNBNDP (port
))
2067 port
= scm_current_input_port ();
2068 SCM_VALIDATE_OPINPORT (2, port
);
2070 n
= scm_i_string_length (str
);
2073 scm_ungetc (scm_i_string_ref (str
, n
), port
);
2079 SCM_DEFINE (scm_seek
, "seek", 3, 0, 0,
2080 (SCM fd_port
, SCM offset
, SCM whence
),
2081 "Sets the current position of @var{fd_port} to the integer\n"
2082 "@var{offset}, which is interpreted according to the value of\n"
2085 "One of the following variables should be supplied for\n"
2087 "@defvar SEEK_SET\n"
2088 "Seek from the beginning of the file.\n"
2090 "@defvar SEEK_CUR\n"
2091 "Seek from the current position.\n"
2093 "@defvar SEEK_END\n"
2094 "Seek from the end of the file.\n"
2096 "If @var{fd_port} is a file descriptor, the underlying system\n"
2097 "call is @code{lseek}. @var{port} may be a string port.\n"
2099 "The value returned is the new position in the file. This means\n"
2100 "that the current position of a port can be obtained using:\n"
2102 "(seek port 0 SEEK_CUR)\n"
2104 #define FUNC_NAME s_scm_seek
2108 fd_port
= SCM_COERCE_OUTPORT (fd_port
);
2110 how
= scm_to_int (whence
);
2111 if (how
!= SEEK_SET
&& how
!= SEEK_CUR
&& how
!= SEEK_END
)
2112 SCM_OUT_OF_RANGE (3, whence
);
2114 if (SCM_OPPORTP (fd_port
))
2116 scm_t_port_internal
*pti
= SCM_PORT_GET_INTERNAL (fd_port
);
2117 scm_t_ptob_descriptor
*ptob
= scm_ptobs
+ SCM_PTOBNUM (fd_port
);
2118 off_t_or_off64_t off
= scm_to_off_t_or_off64_t (offset
);
2119 off_t_or_off64_t rv
;
2122 SCM_MISC_ERROR ("port is not seekable",
2123 scm_cons (fd_port
, SCM_EOL
));
2125 rv
= ptob
->seek (fd_port
, off
, how
);
2127 /* Set stream-start flags according to new position. */
2128 pti
->at_stream_start_for_bom_read
= (rv
== 0);
2129 pti
->at_stream_start_for_bom_write
= (rv
== 0);
2131 scm_i_clear_pending_eof (fd_port
);
2133 return scm_from_off_t_or_off64_t (rv
);
2135 else /* file descriptor?. */
2137 off_t_or_off64_t off
= scm_to_off_t_or_off64_t (offset
);
2138 off_t_or_off64_t rv
;
2139 rv
= lseek_or_lseek64 (scm_to_int (fd_port
), off
, how
);
2142 return scm_from_off_t_or_off64_t (rv
);
2151 /* Mingw has ftruncate(), perhaps implemented above using chsize, but
2152 doesn't have the filename version truncate(), hence this code. */
2153 #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE
2155 truncate (const char *file
, off_t length
)
2159 fdes
= open (file
, O_BINARY
| O_WRONLY
);
2163 ret
= ftruncate (fdes
, length
);
2166 int save_errno
= errno
;
2172 return close (fdes
);
2174 #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */
2176 SCM_DEFINE (scm_truncate_file
, "truncate-file", 1, 1, 0,
2177 (SCM object
, SCM length
),
2178 "Truncate file @var{object} to @var{length} bytes. @var{object}\n"
2179 "can be a filename string, a port object, or an integer file\n"
2181 "The return value is unspecified.\n"
2183 "For a port or file descriptor @var{length} can be omitted, in\n"
2184 "which case the file is truncated at the current position (per\n"
2185 "@code{ftell} above).\n"
2187 "On most systems a file can be extended by giving a length\n"
2188 "greater than the current size, but this is not mandatory in the\n"
2190 #define FUNC_NAME s_scm_truncate_file
2194 /* "object" can be a port, fdes or filename.
2196 Negative "length" makes no sense, but it's left to truncate() or
2197 ftruncate() to give back an error for that (normally EINVAL).
2200 if (SCM_UNBNDP (length
))
2202 /* must supply length if object is a filename. */
2203 if (scm_is_string (object
))
2204 SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL
);
2206 length
= scm_seek (object
, SCM_INUM0
, scm_from_int (SEEK_CUR
));
2209 object
= SCM_COERCE_OUTPORT (object
);
2210 if (scm_is_integer (object
))
2212 off_t_or_off64_t c_length
= scm_to_off_t_or_off64_t (length
);
2213 SCM_SYSCALL (rv
= ftruncate_or_ftruncate64 (scm_to_int (object
),
2216 else if (SCM_OPOUTPORTP (object
))
2218 off_t_or_off64_t c_length
= scm_to_off_t_or_off64_t (length
);
2219 scm_t_port
*pt
= SCM_PTAB_ENTRY (object
);
2220 scm_t_ptob_descriptor
*ptob
= scm_ptobs
+ SCM_PTOBNUM (object
);
2222 if (!ptob
->truncate
)
2223 SCM_MISC_ERROR ("port is not truncatable", SCM_EOL
);
2225 scm_i_clear_pending_eof (object
);
2226 if (pt
->rw_active
== SCM_PORT_READ
)
2227 scm_end_input (object
);
2228 else if (pt
->rw_active
== SCM_PORT_WRITE
)
2229 ptob
->flush (object
);
2231 ptob
->truncate (object
, c_length
);
2236 off_t_or_off64_t c_length
= scm_to_off_t_or_off64_t (length
);
2237 char *str
= scm_to_locale_string (object
);
2239 SCM_SYSCALL (rv
= truncate_or_truncate64 (str
, c_length
));
2246 return SCM_UNSPECIFIED
;
2250 SCM_DEFINE (scm_port_line
, "port-line", 1, 0, 0,
2252 "Return the current line number for @var{port}.\n"
2254 "The first line of a file is 0. But you might want to add 1\n"
2255 "when printing line numbers, since starting from 1 is\n"
2256 "traditional in error messages, and likely to be more natural to\n"
2258 #define FUNC_NAME s_scm_port_line
2260 port
= SCM_COERCE_OUTPORT (port
);
2261 SCM_VALIDATE_OPENPORT (1, port
);
2262 return scm_from_long (SCM_LINUM (port
));
2266 SCM_DEFINE (scm_set_port_line_x
, "set-port-line!", 2, 0, 0,
2267 (SCM port
, SCM line
),
2268 "Set the current line number for @var{port} to @var{line}. The\n"
2269 "first line of a file is 0.")
2270 #define FUNC_NAME s_scm_set_port_line_x
2272 port
= SCM_COERCE_OUTPORT (port
);
2273 SCM_VALIDATE_OPENPORT (1, port
);
2274 SCM_PTAB_ENTRY (port
)->line_number
= scm_to_long (line
);
2275 return SCM_UNSPECIFIED
;
2279 SCM_DEFINE (scm_port_column
, "port-column", 1, 0, 0,
2281 "Return the current column number of @var{port}.\n"
2282 "If the number is\n"
2283 "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n"
2284 "- i.e. the first character of the first line is line 0, column 0.\n"
2285 "(However, when you display a file position, for example in an error\n"
2286 "message, we recommend you add 1 to get 1-origin integers. This is\n"
2287 "because lines and column numbers traditionally start with 1, and that is\n"
2288 "what non-programmers will find most natural.)")
2289 #define FUNC_NAME s_scm_port_column
2291 port
= SCM_COERCE_OUTPORT (port
);
2292 SCM_VALIDATE_OPENPORT (1, port
);
2293 return scm_from_int (SCM_COL (port
));
2297 SCM_DEFINE (scm_set_port_column_x
, "set-port-column!", 2, 0, 0,
2298 (SCM port
, SCM column
),
2299 "Set the current column of @var{port}. Before reading the first\n"
2300 "character on a line the column should be 0.")
2301 #define FUNC_NAME s_scm_set_port_column_x
2303 port
= SCM_COERCE_OUTPORT (port
);
2304 SCM_VALIDATE_OPENPORT (1, port
);
2305 SCM_PTAB_ENTRY (port
)->column_number
= scm_to_int (column
);
2306 return SCM_UNSPECIFIED
;
2310 SCM_DEFINE (scm_port_filename
, "port-filename", 1, 0, 0,
2312 "Return the filename associated with @var{port}, or @code{#f}\n"
2313 "if no filename is associated with the port.")
2314 #define FUNC_NAME s_scm_port_filename
2316 port
= SCM_COERCE_OUTPORT (port
);
2317 SCM_VALIDATE_OPENPORT (1, port
);
2318 return SCM_FILENAME (port
);
2322 SCM_DEFINE (scm_set_port_filename_x
, "set-port-filename!", 2, 0, 0,
2323 (SCM port
, SCM filename
),
2324 "Change the filename associated with @var{port}, using the current input\n"
2325 "port if none is specified. Note that this does not change the port's\n"
2326 "source of data, but only the value that is returned by\n"
2327 "@code{port-filename} and reported in diagnostic output.")
2328 #define FUNC_NAME s_scm_set_port_filename_x
2330 port
= SCM_COERCE_OUTPORT (port
);
2331 SCM_VALIDATE_OPENPORT (1, port
);
2332 /* We allow the user to set the filename to whatever he likes. */
2333 SCM_SET_FILENAME (port
, filename
);
2334 return SCM_UNSPECIFIED
;
2338 /* A fluid specifying the default encoding for newly created ports. If it is
2339 a string, that is the encoding. If it is #f, it is in the "native"
2340 (Latin-1) encoding. */
2341 SCM_VARIABLE (default_port_encoding_var
, "%default-port-encoding");
2343 static int scm_port_encoding_init
= 0;
2345 /* Use ENCODING as the default encoding for future ports. */
2347 scm_i_set_default_port_encoding (const char *encoding
)
2349 if (!scm_port_encoding_init
2350 || !scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var
)))
2351 scm_misc_error (NULL
, "tried to set port encoding fluid before it is initialized",
2354 if (encoding
== NULL
2355 || c_strcasecmp (encoding
, "ASCII") == 0
2356 || c_strcasecmp (encoding
, "ANSI_X3.4-1968") == 0
2357 || c_strcasecmp (encoding
, "ISO-8859-1") == 0)
2358 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var
), SCM_BOOL_F
);
2360 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var
),
2361 scm_from_locale_string (encoding
));
2364 /* Return the name of the default encoding for newly created ports; a
2365 return value of NULL means "ISO-8859-1". */
2367 scm_i_default_port_encoding (void)
2369 if (!scm_port_encoding_init
)
2371 else if (!scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var
)))
2377 encoding
= scm_fluid_ref (SCM_VARIABLE_REF (default_port_encoding_var
));
2378 if (!scm_is_string (encoding
))
2381 return scm_i_string_chars (encoding
);
2385 /* If the next LEN bytes from PORT are equal to those in BYTES, then
2386 return 1, else return 0. Leave the port position unchanged. */
2388 looking_at_bytes (SCM port
, const unsigned char *bytes
, int len
)
2390 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
2393 while (i
< len
&& scm_peek_byte_or_eof (port
) == bytes
[i
])
2398 scm_i_unget_bytes (bytes
, i
, port
);
2402 static const unsigned char scm_utf8_bom
[3] = {0xEF, 0xBB, 0xBF};
2403 static const unsigned char scm_utf16be_bom
[2] = {0xFE, 0xFF};
2404 static const unsigned char scm_utf16le_bom
[2] = {0xFF, 0xFE};
2405 static const unsigned char scm_utf32be_bom
[4] = {0x00, 0x00, 0xFE, 0xFF};
2406 static const unsigned char scm_utf32le_bom
[4] = {0xFF, 0xFE, 0x00, 0x00};
2408 /* Decide what byte order to use for a UTF-16 port. Return "UTF-16BE"
2409 or "UTF-16LE". MODE must be either SCM_PORT_READ or SCM_PORT_WRITE,
2410 and specifies which operation is about to be done. The MODE
2411 determines how we will decide the byte order. We deliberately avoid
2412 reading from the port unless the user is about to do so. If the user
2413 is about to read, then we look for a BOM, and if present, we use it
2414 to determine the byte order. Otherwise we choose big endian, as
2415 recommended by the Unicode Standard. Note that the BOM (if any) is
2416 not consumed here. */
2418 decide_utf16_encoding (SCM port
, scm_t_port_rw_active mode
)
2420 if (mode
== SCM_PORT_READ
2421 && SCM_PORT_GET_INTERNAL (port
)->at_stream_start_for_bom_read
2422 && looking_at_bytes (port
, scm_utf16le_bom
, sizeof scm_utf16le_bom
))
2428 /* Decide what byte order to use for a UTF-32 port. Return "UTF-32BE"
2429 or "UTF-32LE". See the comment above 'decide_utf16_encoding' for
2432 decide_utf32_encoding (SCM port
, scm_t_port_rw_active mode
)
2434 if (mode
== SCM_PORT_READ
2435 && SCM_PORT_GET_INTERNAL (port
)->at_stream_start_for_bom_read
2436 && looking_at_bytes (port
, scm_utf32le_bom
, sizeof scm_utf32le_bom
))
2443 finalize_iconv_descriptors (void *ptr
, void *data
)
2445 close_iconv_descriptors (ptr
);
2448 static scm_t_iconv_descriptors
*
2449 open_iconv_descriptors (const char *encoding
, int reading
, int writing
)
2451 scm_t_iconv_descriptors
*id
;
2452 iconv_t input_cd
, output_cd
;
2454 input_cd
= (iconv_t
) -1;
2455 output_cd
= (iconv_t
) -1;
2458 /* Open an input iconv conversion descriptor, from ENCODING
2459 to UTF-8. We choose UTF-8, not UTF-32, because iconv
2460 implementations can typically convert from anything to
2461 UTF-8, but not to UTF-32 (see
2462 <http://lists.gnu.org/archive/html/bug-libunistring/2010-09/msg00007.html>). */
2464 /* Assume opening an iconv descriptor causes about 16 KB of
2466 scm_gc_register_allocation (16 * 1024);
2468 input_cd
= iconv_open ("UTF-8", encoding
);
2469 if (input_cd
== (iconv_t
) -1)
2470 goto invalid_encoding
;
2475 /* Assume opening an iconv descriptor causes about 16 KB of
2477 scm_gc_register_allocation (16 * 1024);
2479 output_cd
= iconv_open (encoding
, "UTF-8");
2480 if (output_cd
== (iconv_t
) -1)
2482 if (input_cd
!= (iconv_t
) -1)
2483 iconv_close (input_cd
);
2484 goto invalid_encoding
;
2488 id
= scm_gc_malloc_pointerless (sizeof (*id
), "iconv descriptors");
2489 id
->input_cd
= input_cd
;
2490 id
->output_cd
= output_cd
;
2492 /* Register a finalizer to close the descriptors. */
2493 scm_i_set_finalizer (id
, finalize_iconv_descriptors
, NULL
);
2500 err
= scm_from_locale_string (encoding
);
2501 scm_misc_error ("open_iconv_descriptors",
2502 "invalid or unknown character encoding ~s",
2508 close_iconv_descriptors (scm_t_iconv_descriptors
*id
)
2510 if (id
->input_cd
!= (iconv_t
) -1)
2511 iconv_close (id
->input_cd
);
2512 if (id
->output_cd
!= (iconv_t
) -1)
2513 iconv_close (id
->output_cd
);
2514 id
->input_cd
= (void *) -1;
2515 id
->output_cd
= (void *) -1;
2518 /* Return the iconv_descriptors, initializing them if necessary. MODE
2519 must be either SCM_PORT_READ or SCM_PORT_WRITE, and specifies which
2520 operation is about to be done. We deliberately avoid reading from
2521 the port unless the user was about to do so. */
2522 scm_t_iconv_descriptors
*
2523 scm_i_port_iconv_descriptors (SCM port
, scm_t_port_rw_active mode
)
2525 scm_t_port_internal
*pti
= SCM_PORT_GET_INTERNAL (port
);
2527 assert (pti
->encoding_mode
== SCM_PORT_ENCODING_MODE_ICONV
);
2529 if (!pti
->iconv_descriptors
)
2531 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
2532 const char *precise_encoding
;
2535 pt
->encoding
= "ISO-8859-1";
2537 /* If the specified encoding is UTF-16 or UTF-32, then make
2538 that more precise by deciding what byte order to use. */
2539 if (c_strcasecmp (pt
->encoding
, "UTF-16") == 0)
2540 precise_encoding
= decide_utf16_encoding (port
, mode
);
2541 else if (c_strcasecmp (pt
->encoding
, "UTF-32") == 0)
2542 precise_encoding
= decide_utf32_encoding (port
, mode
);
2544 precise_encoding
= pt
->encoding
;
2546 pti
->iconv_descriptors
=
2547 open_iconv_descriptors (precise_encoding
,
2548 SCM_INPUT_PORT_P (port
),
2549 SCM_OUTPUT_PORT_P (port
));
2552 return pti
->iconv_descriptors
;
2556 scm_i_set_port_encoding_x (SCM port
, const char *encoding
)
2559 scm_t_port_internal
*pti
;
2560 scm_t_iconv_descriptors
*prev
;
2562 /* Set the character encoding for this port. */
2563 pt
= SCM_PTAB_ENTRY (port
);
2564 pti
= SCM_PORT_GET_INTERNAL (port
);
2565 prev
= pti
->iconv_descriptors
;
2567 /* In order to handle cases where the encoding changes mid-stream
2568 (e.g. within an HTTP stream, or within a file that is composed of
2569 segments with different encodings), we consider this to be "stream
2570 start" for purposes of BOM handling, regardless of our actual file
2572 pti
->at_stream_start_for_bom_read
= 1;
2573 pti
->at_stream_start_for_bom_write
= 1;
2575 if (encoding
== NULL
)
2576 encoding
= "ISO-8859-1";
2578 /* If ENCODING is UTF-8, then no conversion descriptor is opened
2579 because we do I/O ourselves. This saves 100+ KiB for each
2581 pt
->encoding
= scm_gc_strdup (encoding
, "port");
2582 if (c_strcasecmp (encoding
, "UTF-8") == 0)
2583 pti
->encoding_mode
= SCM_PORT_ENCODING_MODE_UTF8
;
2585 pti
->encoding_mode
= SCM_PORT_ENCODING_MODE_ICONV
;
2587 pti
->iconv_descriptors
= NULL
;
2589 close_iconv_descriptors (prev
);
2592 SCM_DEFINE (scm_port_encoding
, "port-encoding", 1, 0, 0,
2594 "Returns, as a string, the character encoding that @var{port}\n"
2595 "uses to interpret its input and output.\n")
2596 #define FUNC_NAME s_scm_port_encoding
2601 SCM_VALIDATE_PORT (1, port
);
2603 pt
= SCM_PTAB_ENTRY (port
);
2606 return scm_from_locale_string (pt
->encoding
);
2612 SCM_DEFINE (scm_set_port_encoding_x
, "set-port-encoding!", 2, 0, 0,
2613 (SCM port
, SCM enc
),
2614 "Sets the character encoding that will be used to interpret all\n"
2615 "port I/O. New ports are created with the encoding\n"
2616 "appropriate for the current locale if @code{setlocale} has \n"
2617 "been called or ISO-8859-1 otherwise\n"
2618 "and this procedure can be used to modify that encoding.\n")
2619 #define FUNC_NAME s_scm_set_port_encoding_x
2623 SCM_VALIDATE_PORT (1, port
);
2624 SCM_VALIDATE_STRING (2, enc
);
2626 enc_str
= scm_to_locale_string (enc
);
2627 scm_i_set_port_encoding_x (port
, enc_str
);
2630 return SCM_UNSPECIFIED
;
2635 /* A fluid specifying the default conversion handler for newly created
2636 ports. Its value should be one of the symbols below. */
2637 SCM_VARIABLE (default_conversion_strategy_var
,
2638 "%default-port-conversion-strategy");
2640 /* Whether the above fluid is initialized. */
2641 static int scm_conversion_strategy_init
= 0;
2643 /* The possible conversion strategies. */
2644 SCM_SYMBOL (sym_error
, "error");
2645 SCM_SYMBOL (sym_substitute
, "substitute");
2646 SCM_SYMBOL (sym_escape
, "escape");
2648 /* Return the default failed encoding conversion policy for new created
2650 scm_t_string_failed_conversion_handler
2651 scm_i_default_port_conversion_handler (void)
2653 scm_t_string_failed_conversion_handler handler
;
2655 if (!scm_conversion_strategy_init
2656 || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var
)))
2657 handler
= SCM_FAILED_CONVERSION_QUESTION_MARK
;
2662 fluid
= SCM_VARIABLE_REF (default_conversion_strategy_var
);
2663 value
= scm_fluid_ref (fluid
);
2665 if (scm_is_eq (sym_substitute
, value
))
2666 handler
= SCM_FAILED_CONVERSION_QUESTION_MARK
;
2667 else if (scm_is_eq (sym_escape
, value
))
2668 handler
= SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
;
2670 /* Default to 'error also when the fluid's value is not one of
2671 the valid symbols. */
2672 handler
= SCM_FAILED_CONVERSION_ERROR
;
2678 /* Use HANDLER as the default conversion strategy for future ports. */
2680 scm_i_set_default_port_conversion_handler (scm_t_string_failed_conversion_handler
2685 if (!scm_conversion_strategy_init
2686 || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var
)))
2687 scm_misc_error (NULL
, "tried to set conversion strategy fluid before it is initialized",
2692 case SCM_FAILED_CONVERSION_ERROR
:
2693 strategy
= sym_error
;
2696 case SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
:
2697 strategy
= sym_escape
;
2700 case SCM_FAILED_CONVERSION_QUESTION_MARK
:
2701 strategy
= sym_substitute
;
2708 scm_fluid_set_x (SCM_VARIABLE_REF (default_conversion_strategy_var
),
2712 SCM_DEFINE (scm_port_conversion_strategy
, "port-conversion-strategy",
2713 1, 0, 0, (SCM port
),
2714 "Returns the behavior of the port when handling a character that\n"
2715 "is not representable in the port's current encoding.\n"
2716 "It returns the symbol @code{error} if unrepresentable characters\n"
2717 "should cause exceptions, @code{substitute} if the port should\n"
2718 "try to replace unrepresentable characters with question marks or\n"
2719 "approximate characters, or @code{escape} if unrepresentable\n"
2720 "characters should be converted to string escapes.\n"
2722 "If @var{port} is @code{#f}, then the current default behavior\n"
2723 "will be returned. New ports will have this default behavior\n"
2724 "when they are created.\n")
2725 #define FUNC_NAME s_scm_port_conversion_strategy
2727 scm_t_string_failed_conversion_handler h
;
2729 if (scm_is_false (port
))
2730 h
= scm_i_default_port_conversion_handler ();
2735 SCM_VALIDATE_OPPORT (1, port
);
2736 pt
= SCM_PTAB_ENTRY (port
);
2738 h
= pt
->ilseq_handler
;
2741 if (h
== SCM_FAILED_CONVERSION_ERROR
)
2742 return scm_from_latin1_symbol ("error");
2743 else if (h
== SCM_FAILED_CONVERSION_QUESTION_MARK
)
2744 return scm_from_latin1_symbol ("substitute");
2745 else if (h
== SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
)
2746 return scm_from_latin1_symbol ("escape");
2750 /* Never gets here. */
2751 return SCM_UNDEFINED
;
2755 SCM_DEFINE (scm_set_port_conversion_strategy_x
, "set-port-conversion-strategy!",
2757 (SCM port
, SCM sym
),
2758 "Sets the behavior of the interpreter when outputting a character\n"
2759 "that is not representable in the port's current encoding.\n"
2760 "@var{sym} can be either @code{'error}, @code{'substitute}, or\n"
2761 "@code{'escape}. If it is @code{'error}, an error will be thrown\n"
2762 "when an unconvertible character is encountered. If it is\n"
2763 "@code{'substitute}, then unconvertible characters will \n"
2764 "be replaced with approximate characters, or with question marks\n"
2765 "if no approximately correct character is available.\n"
2766 "If it is @code{'escape},\n"
2767 "it will appear as a hex escape when output.\n"
2769 "If @var{port} is an open port, the conversion error behavior\n"
2770 "is set for that port. If it is @code{#f}, it is set as the\n"
2771 "default behavior for any future ports that get created in\n"
2773 #define FUNC_NAME s_scm_set_port_conversion_strategy_x
2775 scm_t_string_failed_conversion_handler handler
;
2777 if (scm_is_eq (sym
, sym_error
))
2778 handler
= SCM_FAILED_CONVERSION_ERROR
;
2779 else if (scm_is_eq (sym
, sym_substitute
))
2780 handler
= SCM_FAILED_CONVERSION_QUESTION_MARK
;
2781 else if (scm_is_eq (sym
, sym_escape
))
2782 handler
= SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
;
2784 SCM_MISC_ERROR ("unknown conversion strategy ~s", scm_list_1 (sym
));
2786 if (scm_is_false (port
))
2787 scm_i_set_default_port_conversion_handler (handler
);
2790 SCM_VALIDATE_OPPORT (1, port
);
2791 SCM_PTAB_ENTRY (port
)->ilseq_handler
= handler
;
2794 return SCM_UNSPECIFIED
;
2801 scm_print_port_mode (SCM exp
, SCM port
)
2803 scm_puts (SCM_CLOSEDP (exp
)
2805 : (SCM_RDNG
& SCM_CELL_WORD_0 (exp
)
2806 ? (SCM_WRTNG
& SCM_CELL_WORD_0 (exp
)
2809 : (SCM_WRTNG
& SCM_CELL_WORD_0 (exp
)
2816 scm_port_print (SCM exp
, SCM port
, scm_print_state
*pstate SCM_UNUSED
)
2818 char *type
= SCM_PTOBNAME (SCM_PTOBNUM (exp
));
2821 scm_puts ("#<", port
);
2822 scm_print_port_mode (exp
, port
);
2823 scm_puts (type
, port
);
2824 scm_putc (' ', port
);
2825 scm_uintprint (SCM_CELL_WORD_1 (exp
), 16, port
);
2826 scm_putc ('>', port
);
2834 scm_t_bits scm_tc16_void_port
= 0;
2836 static int fill_input_void_port (SCM port SCM_UNUSED
)
2842 write_void_port (SCM port SCM_UNUSED
,
2843 const void *data SCM_UNUSED
,
2844 size_t size SCM_UNUSED
)
2849 scm_i_void_port (long mode_bits
)
2851 scm_i_scm_pthread_mutex_lock (&scm_i_port_table_mutex
);
2853 SCM answer
= scm_new_port_table_entry (scm_tc16_void_port
);
2854 scm_t_port
* pt
= SCM_PTAB_ENTRY(answer
);
2856 scm_port_non_buffer (pt
);
2858 SCM_SETSTREAM (answer
, 0);
2859 SCM_SET_CELL_TYPE (answer
, scm_tc16_void_port
| mode_bits
);
2860 scm_i_pthread_mutex_unlock (&scm_i_port_table_mutex
);
2866 scm_void_port (char *mode_str
)
2868 return scm_i_void_port (scm_mode_bits (mode_str
));
2871 SCM_DEFINE (scm_sys_make_void_port
, "%make-void-port", 1, 0, 0,
2873 "Create and return a new void port. A void port acts like\n"
2874 "@file{/dev/null}. The @var{mode} argument\n"
2875 "specifies the input/output modes for this port: see the\n"
2876 "documentation for @code{open-file} in @ref{File Ports}.")
2877 #define FUNC_NAME s_scm_sys_make_void_port
2879 return scm_i_void_port (scm_i_mode_bits (mode
));
2884 /* Initialization. */
2889 /* lseek() symbols. */
2890 scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET
));
2891 scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR
));
2892 scm_c_define ("SEEK_END", scm_from_int (SEEK_END
));
2894 scm_tc16_void_port
= scm_make_port_type ("void", fill_input_void_port
,
2897 cur_inport_fluid
= scm_make_fluid ();
2898 cur_outport_fluid
= scm_make_fluid ();
2899 cur_errport_fluid
= scm_make_fluid ();
2900 cur_loadport_fluid
= scm_make_fluid ();
2902 scm_i_port_weak_hash
= scm_make_weak_key_hash_table (SCM_I_MAKINUM(31));
2904 #include "libguile/ports.x"
2906 /* Use Latin-1 as the default port encoding. */
2907 SCM_VARIABLE_SET (default_port_encoding_var
,
2908 scm_make_fluid_with_default (SCM_BOOL_F
));
2909 scm_port_encoding_init
= 1;
2911 SCM_VARIABLE_SET (default_conversion_strategy_var
,
2912 scm_make_fluid_with_default (sym_substitute
));
2913 scm_conversion_strategy_init
= 1;
2915 /* These bindings are used when boot-9 turns `current-input-port' et
2916 al into parameters. They are then removed from the guile module. */
2917 scm_c_define ("%current-input-port-fluid", cur_inport_fluid
);
2918 scm_c_define ("%current-output-port-fluid", cur_outport_fluid
);
2919 scm_c_define ("%current-error-port-fluid", cur_errport_fluid
);