1 /* Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2003, 2004,
2 * 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013 Free Software Foundation, Inc.
4 * This library is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public License
6 * as published by the Free Software Foundation; either version 3 of
7 * the License, or (at your option) any later version.
9 * This library is distributed in the hope that it will be useful, but
10 * WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
14 * You should have received a copy of the GNU Lesser General Public
15 * License along with this library; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
24 #define _LARGEFILE64_SOURCE /* ask for stat64 etc */
32 #include <fcntl.h> /* for chsize on mingw */
37 #include <striconveh.h>
41 #include "libguile/_scm.h"
42 #include "libguile/async.h"
43 #include "libguile/deprecation.h"
44 #include "libguile/eval.h"
45 #include "libguile/fports.h" /* direct access for seek and truncate */
46 #include "libguile/goops.h"
47 #include "libguile/smob.h"
48 #include "libguile/chars.h"
49 #include "libguile/dynwind.h"
51 #include "libguile/keywords.h"
52 #include "libguile/hashtab.h"
53 #include "libguile/root.h"
54 #include "libguile/strings.h"
55 #include "libguile/mallocs.h"
56 #include "libguile/validate.h"
57 #include "libguile/ports.h"
58 #include "libguile/vectors.h"
59 #include "libguile/weak-set.h"
60 #include "libguile/fluids.h"
61 #include "libguile/eq.h"
75 #ifdef HAVE_SYS_IOCTL_H
76 #include <sys/ioctl.h>
79 /* Mingw (version 3.4.5, circa 2006) has ftruncate as an alias for chsize
80 already, but have this code here in case that wasn't so in past versions,
81 or perhaps to help other minimal DOS environments.
83 gnulib ftruncate.c has code using fcntl F_CHSIZE and F_FREESP, which
84 might be possibilities if we've got other systems without ftruncate. */
86 #if defined HAVE_CHSIZE && ! defined HAVE_FTRUNCATE
87 #define ftruncate(fd, size) chsize (fd, size)
89 #define HAVE_FTRUNCATE 1
93 /* Port encodings are case-insensitive ASCII strings. */
95 ascii_toupper (char c
)
97 return (c
< 'a' || c
> 'z') ? c
: ('A' + (c
- 'a'));
100 /* It is only necessary to use this function on encodings that come from
101 the user and have not been canonicalized yet. Encodings that are set
102 on ports or in the default encoding fluid are in upper-case, and can
103 be compared with strcmp. */
105 encoding_matches (const char *enc
, const char *upper
)
111 if (ascii_toupper (*enc
++) != *upper
++)
118 canonicalize_encoding (const char *enc
)
126 ret
= scm_gc_strdup (enc
, "port");
128 for (i
= 0; ret
[i
]; i
++)
131 /* Restrict to ASCII. */
132 scm_misc_error (NULL
, "invalid character encoding ~s",
133 scm_list_1 (scm_from_latin1_string (enc
)));
135 ret
[i
] = ascii_toupper (ret
[i
]);
143 /* The port kind table --- a dynamically resized array of port types. */
146 /* scm_ptobs scm_numptob
147 * implement a dynamically resized array of ptob records.
148 * Indexes into this table are used when generating type
149 * tags for smobjects (if you know a tag you can get an index and conversely).
151 static scm_t_ptob_descriptor
**scm_ptobs
= NULL
;
152 static long scm_numptob
= 0; /* Number of port types. */
153 static long scm_ptobs_size
= 0; /* Number of slots in the port type
155 static scm_i_pthread_mutex_t scm_ptobs_lock
= SCM_I_PTHREAD_MUTEX_INITIALIZER
;
158 scm_c_num_port_types (void)
162 scm_i_pthread_mutex_lock (&scm_ptobs_lock
);
164 scm_i_pthread_mutex_unlock (&scm_ptobs_lock
);
169 scm_t_ptob_descriptor
*
170 scm_c_port_type_ref (long ptobnum
)
172 scm_t_ptob_descriptor
*ret
= NULL
;
174 scm_i_pthread_mutex_lock (&scm_ptobs_lock
);
176 if (0 <= ptobnum
&& ptobnum
< scm_numptob
)
177 ret
= scm_ptobs
[ptobnum
];
179 scm_i_pthread_mutex_unlock (&scm_ptobs_lock
);
182 scm_out_of_range ("scm_c_port_type_ref", scm_from_long (ptobnum
));
188 scm_c_port_type_add_x (scm_t_ptob_descriptor
*desc
)
192 scm_i_pthread_mutex_lock (&scm_ptobs_lock
);
194 if (scm_numptob
+ 1 < SCM_I_MAX_PORT_TYPE_COUNT
)
196 if (scm_numptob
== scm_ptobs_size
)
198 unsigned long old_size
= scm_ptobs_size
;
199 scm_t_ptob_descriptor
**old_ptobs
= scm_ptobs
;
201 /* Currently there are only 9 predefined port types, so one
202 resize will cover it. */
203 scm_ptobs_size
= old_size
+ 10;
205 if (scm_ptobs_size
>= SCM_I_MAX_PORT_TYPE_COUNT
)
206 scm_ptobs_size
= SCM_I_MAX_PORT_TYPE_COUNT
;
208 scm_ptobs
= scm_gc_malloc (sizeof (*scm_ptobs
) * scm_ptobs_size
,
211 memcpy (scm_ptobs
, old_ptobs
, sizeof (*scm_ptobs
) * scm_numptob
);
215 scm_ptobs
[ret
] = desc
;
218 scm_i_pthread_mutex_unlock (&scm_ptobs_lock
);
221 scm_out_of_range ("scm_c_port_type_add_x", scm_from_long (scm_numptob
));
227 * We choose to use an interface similar to the smob interface with
228 * fill_input and write as standard fields, passed to the port
229 * type constructor, and optional fields set by setters.
233 flush_port_default (SCM port SCM_UNUSED
)
238 end_input_default (SCM port SCM_UNUSED
, int offset SCM_UNUSED
)
243 scm_make_port_type (char *name
,
244 int (*fill_input
) (SCM port
),
245 void (*write
) (SCM port
, const void *data
, size_t size
))
247 scm_t_ptob_descriptor
*desc
;
250 desc
= scm_gc_malloc_pointerless (sizeof (*desc
), "port-type");
251 memset (desc
, 0, sizeof (*desc
));
254 desc
->print
= scm_port_print
;
256 desc
->flush
= flush_port_default
;
257 desc
->end_input
= end_input_default
;
258 desc
->fill_input
= fill_input
;
260 ptobnum
= scm_c_port_type_add_x (desc
);
262 /* Make a class object if GOOPS is present. */
263 if (SCM_UNPACK (scm_port_class
[0]) != 0)
264 scm_make_port_classes (ptobnum
, name
);
266 return scm_tc7_port
+ ptobnum
* 256;
270 scm_set_port_mark (scm_t_bits tc
, SCM (*mark
) (SCM
))
272 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->mark
= mark
;
276 scm_set_port_free (scm_t_bits tc
, size_t (*free
) (SCM
))
278 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->free
= free
;
282 scm_set_port_print (scm_t_bits tc
, int (*print
) (SCM exp
, SCM port
,
283 scm_print_state
*pstate
))
285 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->print
= print
;
289 scm_set_port_equalp (scm_t_bits tc
, SCM (*equalp
) (SCM
, SCM
))
291 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->equalp
= equalp
;
295 scm_set_port_close (scm_t_bits tc
, int (*close
) (SCM
))
297 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->close
= close
;
301 scm_set_port_flush (scm_t_bits tc
, void (*flush
) (SCM port
))
303 scm_t_ptob_descriptor
*ptob
= scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
));
305 ptob
->flags
|= SCM_PORT_TYPE_HAS_FLUSH
;
309 scm_set_port_end_input (scm_t_bits tc
, void (*end_input
) (SCM port
, int offset
))
311 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->end_input
= end_input
;
315 scm_set_port_seek (scm_t_bits tc
, scm_t_off (*seek
) (SCM
, scm_t_off
, int))
317 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->seek
= seek
;
321 scm_set_port_truncate (scm_t_bits tc
, void (*truncate
) (SCM
, scm_t_off
))
323 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->truncate
= truncate
;
327 scm_set_port_input_waiting (scm_t_bits tc
, int (*input_waiting
) (SCM
))
329 scm_c_port_type_ref (SCM_TC2PTOBNUM (tc
))->input_waiting
= input_waiting
;
334 /* Standard ports --- current input, output, error, and more(!). */
336 static SCM cur_inport_fluid
= SCM_BOOL_F
;
337 static SCM cur_outport_fluid
= SCM_BOOL_F
;
338 static SCM cur_errport_fluid
= SCM_BOOL_F
;
339 static SCM cur_loadport_fluid
= SCM_BOOL_F
;
341 SCM_DEFINE (scm_current_input_port
, "current-input-port", 0, 0, 0,
343 "Return the current input port. This is the default port used\n"
344 "by many input procedures. Initially, @code{current-input-port}\n"
345 "returns the @dfn{standard input} in Unix and C terminology.")
346 #define FUNC_NAME s_scm_current_input_port
348 if (scm_is_true (cur_inport_fluid
))
349 return scm_fluid_ref (cur_inport_fluid
);
355 SCM_DEFINE (scm_current_output_port
, "current-output-port", 0, 0, 0,
357 "Return the current output port. This is the default port used\n"
358 "by many output procedures. Initially,\n"
359 "@code{current-output-port} returns the @dfn{standard output} in\n"
360 "Unix and C terminology.")
361 #define FUNC_NAME s_scm_current_output_port
363 if (scm_is_true (cur_outport_fluid
))
364 return scm_fluid_ref (cur_outport_fluid
);
370 SCM_DEFINE (scm_current_error_port
, "current-error-port", 0, 0, 0,
372 "Return the port to which errors and warnings should be sent (the\n"
373 "@dfn{standard error} in Unix and C terminology).")
374 #define FUNC_NAME s_scm_current_error_port
376 if (scm_is_true (cur_errport_fluid
))
377 return scm_fluid_ref (cur_errport_fluid
);
384 scm_current_warning_port (void)
386 static SCM cwp_var
= SCM_BOOL_F
;
388 if (scm_is_false (cwp_var
))
389 cwp_var
= scm_c_private_lookup ("guile", "current-warning-port");
391 return scm_call_0 (scm_variable_ref (cwp_var
));
394 SCM_DEFINE (scm_current_load_port
, "current-load-port", 0, 0, 0,
396 "Return the current-load-port.\n"
397 "The load port is used internally by @code{primitive-load}.")
398 #define FUNC_NAME s_scm_current_load_port
400 return scm_fluid_ref (cur_loadport_fluid
);
404 SCM_DEFINE (scm_set_current_input_port
, "set-current-input-port", 1, 0, 0,
406 "@deffnx {Scheme Procedure} set-current-output-port port\n"
407 "@deffnx {Scheme Procedure} set-current-error-port port\n"
408 "Change the ports returned by @code{current-input-port},\n"
409 "@code{current-output-port} and @code{current-error-port}, respectively,\n"
410 "so that they use the supplied @var{port} for input or output.")
411 #define FUNC_NAME s_scm_set_current_input_port
413 SCM oinp
= scm_fluid_ref (cur_inport_fluid
);
414 SCM_VALIDATE_OPINPORT (1, port
);
415 scm_fluid_set_x (cur_inport_fluid
, port
);
421 SCM_DEFINE (scm_set_current_output_port
, "set-current-output-port", 1, 0, 0,
423 "Set the current default output port to @var{port}.")
424 #define FUNC_NAME s_scm_set_current_output_port
426 SCM ooutp
= scm_fluid_ref (cur_outport_fluid
);
427 port
= SCM_COERCE_OUTPORT (port
);
428 SCM_VALIDATE_OPOUTPORT (1, port
);
429 scm_fluid_set_x (cur_outport_fluid
, port
);
435 SCM_DEFINE (scm_set_current_error_port
, "set-current-error-port", 1, 0, 0,
437 "Set the current default error port to @var{port}.")
438 #define FUNC_NAME s_scm_set_current_error_port
440 SCM oerrp
= scm_fluid_ref (cur_errport_fluid
);
441 port
= SCM_COERCE_OUTPORT (port
);
442 SCM_VALIDATE_OPOUTPORT (1, port
);
443 scm_fluid_set_x (cur_errport_fluid
, port
);
450 scm_set_current_warning_port (SCM port
)
452 static SCM cwp_var
= SCM_BOOL_F
;
454 if (scm_is_false (cwp_var
))
455 cwp_var
= scm_c_private_lookup ("guile", "current-warning-port");
457 return scm_call_1 (scm_variable_ref (cwp_var
), port
);
462 scm_dynwind_current_input_port (SCM port
)
463 #define FUNC_NAME NULL
465 SCM_VALIDATE_OPINPORT (1, port
);
466 scm_dynwind_fluid (cur_inport_fluid
, port
);
471 scm_dynwind_current_output_port (SCM port
)
472 #define FUNC_NAME NULL
474 port
= SCM_COERCE_OUTPORT (port
);
475 SCM_VALIDATE_OPOUTPORT (1, port
);
476 scm_dynwind_fluid (cur_outport_fluid
, port
);
481 scm_dynwind_current_error_port (SCM port
)
482 #define FUNC_NAME NULL
484 port
= SCM_COERCE_OUTPORT (port
);
485 SCM_VALIDATE_OPOUTPORT (1, port
);
486 scm_dynwind_fluid (cur_errport_fluid
, port
);
491 scm_i_dynwind_current_load_port (SCM port
)
493 scm_dynwind_fluid (cur_loadport_fluid
, port
);
499 /* Retrieving a port's mode. */
501 /* Return the flags that characterize a port based on the mode
502 * string used to open a file for that port.
504 * See PORT FLAGS in scm.h
508 scm_i_mode_bits_n (SCM modes
)
511 | (scm_i_string_contains_char (modes
, 'r')
512 || scm_i_string_contains_char (modes
, '+') ? SCM_RDNG
: 0)
513 | (scm_i_string_contains_char (modes
, 'w')
514 || scm_i_string_contains_char (modes
, 'a')
515 || scm_i_string_contains_char (modes
, '+') ? SCM_WRTNG
: 0)
516 | (scm_i_string_contains_char (modes
, '0') ? SCM_BUF0
: 0)
517 | (scm_i_string_contains_char (modes
, 'l') ? SCM_BUFLINE
: 0));
521 scm_mode_bits (char *modes
)
523 /* Valid characters are rw+a0l. So, use latin1. */
524 return scm_i_mode_bits (scm_from_latin1_string (modes
));
528 scm_i_mode_bits (SCM modes
)
532 if (!scm_is_string (modes
))
533 scm_wrong_type_arg_msg (NULL
, 0, modes
, "string");
535 bits
= scm_i_mode_bits_n (modes
);
536 scm_remember_upto_here_1 (modes
);
540 /* Return the mode flags from an open port.
541 * Some modes such as "append" are only used when opening
542 * a file and are not returned here. */
544 SCM_DEFINE (scm_port_mode
, "port-mode", 1, 0, 0,
546 "Return the port modes associated with the open port @var{port}.\n"
547 "These will not necessarily be identical to the modes used when\n"
548 "the port was opened, since modes such as \"append\" which are\n"
549 "used only during port creation are not retained.")
550 #define FUNC_NAME s_scm_port_mode
555 port
= SCM_COERCE_OUTPORT (port
);
556 SCM_VALIDATE_OPPORT (1, port
);
557 if (SCM_CELL_WORD_0 (port
) & SCM_RDNG
) {
558 if (SCM_CELL_WORD_0 (port
) & SCM_WRTNG
)
559 strcpy (modes
, "r+");
563 else if (SCM_CELL_WORD_0 (port
) & SCM_WRTNG
)
565 if (SCM_CELL_WORD_0 (port
) & SCM_BUF0
)
568 return scm_from_latin1_string (modes
);
575 /* The port table --- a weak set of all 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. */
579 SCM scm_i_port_weak_set
;
584 /* Port finalization. */
588 scm_t_ptob_descriptor
*ptob
;
593 do_free (void *body_data
)
595 struct do_free_data
*data
= body_data
;
597 /* `close' is for explicit `close-port' by user. `free' is for this
598 purpose: ports collected by the GC. */
599 data
->ptob
->free (data
->port
);
604 /* Finalize the object (a port) pointed to by PTR. */
606 finalize_port (void *ptr
, void *data
)
608 SCM port
= SCM_PACK_POINTER (ptr
);
610 if (!SCM_PORTP (port
))
613 if (SCM_OPENP (port
))
615 struct do_free_data data
;
617 SCM_CLR_PORT_OPEN_FLAG (port
);
619 data
.ptob
= SCM_PORT_DESCRIPTOR (port
);
622 scm_internal_catch (SCM_BOOL_T
, do_free
, &data
,
623 scm_handle_by_message_noexit
, NULL
);
625 scm_gc_ports_collected
++;
633 scm_c_make_port_with_encoding (scm_t_bits tag
, unsigned long mode_bits
,
634 const char *encoding
,
635 scm_t_string_failed_conversion_handler handler
,
640 scm_t_ptob_descriptor
*ptob
;
642 entry
= (scm_t_port
*) scm_gc_calloc (sizeof (scm_t_port
), "port");
643 ptob
= scm_c_port_type_ref (SCM_TC2PTOBNUM (tag
));
645 ret
= scm_words (tag
| mode_bits
, 3);
646 SCM_SET_CELL_WORD_1 (ret
, (scm_t_bits
) entry
);
647 SCM_SET_CELL_WORD_2 (ret
, (scm_t_bits
) ptob
);
649 entry
->lock
= scm_gc_malloc_pointerless (sizeof (*entry
->lock
), "port lock");
650 scm_i_pthread_mutex_init (entry
->lock
, scm_i_pthread_mutexattr_recursive
);
652 entry
->file_name
= SCM_BOOL_F
;
653 entry
->rw_active
= SCM_PORT_NEITHER
;
655 entry
->stream
= stream
;
657 if (encoding_matches (encoding
, "UTF-8"))
659 entry
->encoding_mode
= SCM_PORT_ENCODING_MODE_UTF8
;
660 entry
->encoding
= "UTF-8";
662 else if (encoding_matches (encoding
, "ISO-8859-1"))
664 entry
->encoding_mode
= SCM_PORT_ENCODING_MODE_LATIN1
;
665 entry
->encoding
= "ISO-8859-1";
669 entry
->encoding_mode
= SCM_PORT_ENCODING_MODE_ICONV
;
670 entry
->encoding
= canonicalize_encoding (encoding
);
673 entry
->ilseq_handler
= handler
;
674 entry
->iconv_descriptors
= NULL
;
676 entry
->alist
= SCM_EOL
;
678 if (SCM_PORT_DESCRIPTOR (ret
)->free
)
679 scm_i_set_finalizer (SCM2PTR (ret
), finalize_port
, NULL
);
681 if (SCM_PORT_DESCRIPTOR (ret
)->flags
& SCM_PORT_TYPE_HAS_FLUSH
)
682 scm_weak_set_add_x (scm_i_port_weak_set
, ret
);
688 scm_c_make_port (scm_t_bits tag
, unsigned long mode_bits
, scm_t_bits stream
)
690 return scm_c_make_port_with_encoding (tag
, mode_bits
,
691 scm_i_default_port_encoding (),
692 scm_i_default_port_conversion_handler (),
697 scm_new_port_table_entry (scm_t_bits tag
)
699 return scm_c_make_port (tag
, 0, 0);
706 SCM_DEFINE (scm_port_p
, "port?", 1, 0, 0,
708 "Return a boolean indicating whether @var{x} is a port.\n"
709 "Equivalent to @code{(or (input-port? @var{x}) (output-port?\n"
711 #define FUNC_NAME s_scm_port_p
713 return scm_from_bool (SCM_PORTP (x
));
717 SCM_DEFINE (scm_input_port_p
, "input-port?", 1, 0, 0,
719 "Return @code{#t} if @var{x} is an input port, otherwise return\n"
720 "@code{#f}. Any object satisfying this predicate also satisfies\n"
722 #define FUNC_NAME s_scm_input_port_p
724 return scm_from_bool (SCM_INPUT_PORT_P (x
));
728 SCM_DEFINE (scm_output_port_p
, "output-port?", 1, 0, 0,
730 "Return @code{#t} if @var{x} is an output port, otherwise return\n"
731 "@code{#f}. Any object satisfying this predicate also satisfies\n"
733 #define FUNC_NAME s_scm_output_port_p
735 x
= SCM_COERCE_OUTPORT (x
);
736 return scm_from_bool (SCM_OUTPUT_PORT_P (x
));
740 SCM_DEFINE (scm_port_closed_p
, "port-closed?", 1, 0, 0,
742 "Return @code{#t} if @var{port} is closed or @code{#f} if it is\n"
744 #define FUNC_NAME s_scm_port_closed_p
746 SCM_VALIDATE_PORT (1, port
);
747 return scm_from_bool (!SCM_OPPORTP (port
));
751 SCM_DEFINE (scm_eof_object_p
, "eof-object?", 1, 0, 0,
753 "Return @code{#t} if @var{x} is an end-of-file object; otherwise\n"
755 #define FUNC_NAME s_scm_eof_object_p
757 return scm_from_bool (SCM_EOF_OBJECT_P (x
));
766 static void close_iconv_descriptors (scm_t_iconv_descriptors
*id
);
769 * Call the close operation on a port object.
770 * see also scm_close.
772 SCM_DEFINE (scm_close_port
, "close-port", 1, 0, 0,
774 "Close the specified port object. Return @code{#t} if it\n"
775 "successfully closes a port or @code{#f} if it was already\n"
776 "closed. An exception may be raised if an error occurs, for\n"
777 "example when flushing buffered output. See also @ref{Ports and\n"
778 "File Descriptors, close}, for a procedure which can close file\n"
780 #define FUNC_NAME s_scm_close_port
785 port
= SCM_COERCE_OUTPORT (port
);
787 SCM_VALIDATE_PORT (1, port
);
788 if (SCM_CLOSEDP (port
))
791 p
= SCM_PTAB_ENTRY (port
);
792 SCM_CLR_PORT_OPEN_FLAG (port
);
794 if (SCM_PORT_DESCRIPTOR (port
)->flags
& SCM_PORT_TYPE_HAS_FLUSH
)
795 scm_weak_set_remove_x (scm_i_port_weak_set
, port
);
797 if (SCM_PORT_DESCRIPTOR (port
)->close
)
798 /* Note! This may throw an exception. Anything after this point
799 should be resilient to non-local exits. */
800 rv
= SCM_PORT_DESCRIPTOR (port
)->close (port
);
804 if (p
->iconv_descriptors
)
806 /* If we don't get here, the iconv_descriptors finalizer will
808 close_iconv_descriptors (p
->iconv_descriptors
);
809 p
->iconv_descriptors
= NULL
;
812 return scm_from_bool (rv
>= 0);
816 SCM_DEFINE (scm_close_input_port
, "close-input-port", 1, 0, 0,
818 "Close the specified input port object. The routine has no effect if\n"
819 "the file has already been closed. An exception may be raised if an\n"
820 "error occurs. The value returned is unspecified.\n\n"
821 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
822 "which can close file descriptors.")
823 #define FUNC_NAME s_scm_close_input_port
825 SCM_VALIDATE_INPUT_PORT (1, port
);
826 scm_close_port (port
);
827 return SCM_UNSPECIFIED
;
831 SCM_DEFINE (scm_close_output_port
, "close-output-port", 1, 0, 0,
833 "Close the specified output port object. The routine has no effect if\n"
834 "the file has already been closed. An exception may be raised if an\n"
835 "error occurs. The value returned is unspecified.\n\n"
836 "See also @ref{Ports and File Descriptors, close}, for a procedure\n"
837 "which can close file descriptors.")
838 #define FUNC_NAME s_scm_close_output_port
840 port
= SCM_COERCE_OUTPORT (port
);
841 SCM_VALIDATE_OUTPUT_PORT (1, port
);
842 scm_close_port (port
);
843 return SCM_UNSPECIFIED
;
850 /* Encoding characters to byte streams, and decoding byte streams to
853 /* A fluid specifying the default encoding for newly created ports. If it is
854 a string, that is the encoding. If it is #f, it is in the "native"
855 (Latin-1) encoding. */
856 SCM_VARIABLE (default_port_encoding_var
, "%default-port-encoding");
858 static int scm_port_encoding_init
= 0;
860 /* Use ENCODING as the default encoding for future ports. */
862 scm_i_set_default_port_encoding (const char *encoding
)
864 if (!scm_port_encoding_init
865 || !scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var
)))
866 scm_misc_error (NULL
, "tried to set port encoding fluid before it is initialized",
869 if (encoding_matches (encoding
, "ISO-8859-1"))
870 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var
), SCM_BOOL_F
);
872 scm_fluid_set_x (SCM_VARIABLE_REF (default_port_encoding_var
),
873 scm_from_latin1_string (canonicalize_encoding (encoding
)));
876 /* Return the name of the default encoding for newly created ports. */
878 scm_i_default_port_encoding (void)
880 if (!scm_port_encoding_init
)
882 else if (!scm_is_fluid (SCM_VARIABLE_REF (default_port_encoding_var
)))
888 encoding
= scm_fluid_ref (SCM_VARIABLE_REF (default_port_encoding_var
));
889 if (!scm_is_string (encoding
))
892 return scm_i_string_chars (encoding
);
896 /* A fluid specifying the default conversion handler for newly created
897 ports. Its value should be one of the symbols below. */
898 SCM_VARIABLE (default_conversion_strategy_var
,
899 "%default-port-conversion-strategy");
901 /* Whether the above fluid is initialized. */
902 static int scm_conversion_strategy_init
= 0;
904 /* The possible conversion strategies. */
905 SCM_SYMBOL (sym_error
, "error");
906 SCM_SYMBOL (sym_substitute
, "substitute");
907 SCM_SYMBOL (sym_escape
, "escape");
909 /* Return the default failed encoding conversion policy for new created
911 scm_t_string_failed_conversion_handler
912 scm_i_default_port_conversion_handler (void)
914 scm_t_string_failed_conversion_handler handler
;
916 if (!scm_conversion_strategy_init
917 || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var
)))
918 handler
= SCM_FAILED_CONVERSION_QUESTION_MARK
;
923 fluid
= SCM_VARIABLE_REF (default_conversion_strategy_var
);
924 value
= scm_fluid_ref (fluid
);
926 if (scm_is_eq (sym_substitute
, value
))
927 handler
= SCM_FAILED_CONVERSION_QUESTION_MARK
;
928 else if (scm_is_eq (sym_escape
, value
))
929 handler
= SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
;
931 /* Default to 'error also when the fluid's value is not one of
932 the valid symbols. */
933 handler
= SCM_FAILED_CONVERSION_ERROR
;
939 /* Use HANDLER as the default conversion strategy for future ports. */
941 scm_i_set_default_port_conversion_handler (scm_t_string_failed_conversion_handler
946 if (!scm_conversion_strategy_init
947 || !scm_is_fluid (SCM_VARIABLE_REF (default_conversion_strategy_var
)))
948 scm_misc_error (NULL
, "tried to set conversion strategy fluid before it is initialized",
953 case SCM_FAILED_CONVERSION_ERROR
:
954 strategy
= sym_error
;
957 case SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
:
958 strategy
= sym_escape
;
961 case SCM_FAILED_CONVERSION_QUESTION_MARK
:
962 strategy
= sym_substitute
;
969 scm_fluid_set_x (SCM_VARIABLE_REF (default_conversion_strategy_var
),
974 finalize_iconv_descriptors (void *ptr
, void *data
)
976 close_iconv_descriptors (ptr
);
979 static scm_t_iconv_descriptors
*
980 open_iconv_descriptors (const char *encoding
, int reading
, int writing
)
982 scm_t_iconv_descriptors
*id
;
983 iconv_t input_cd
, output_cd
;
986 input_cd
= (iconv_t
) -1;
987 output_cd
= (iconv_t
) -1;
989 for (i
= 0; encoding
[i
]; i
++)
990 if (encoding
[i
] > 127)
991 goto invalid_encoding
;
995 /* Open an input iconv conversion descriptor, from ENCODING
996 to UTF-8. We choose UTF-8, not UTF-32, because iconv
997 implementations can typically convert from anything to
998 UTF-8, but not to UTF-32 (see
999 <http://lists.gnu.org/archive/html/bug-libunistring/2010-09/msg00007.html>). */
1001 /* Assume opening an iconv descriptor causes about 16 KB of
1003 scm_gc_register_allocation (16 * 1024);
1005 input_cd
= iconv_open ("UTF-8", encoding
);
1006 if (input_cd
== (iconv_t
) -1)
1007 goto invalid_encoding
;
1012 /* Assume opening an iconv descriptor causes about 16 KB of
1014 scm_gc_register_allocation (16 * 1024);
1016 output_cd
= iconv_open (encoding
, "UTF-8");
1017 if (output_cd
== (iconv_t
) -1)
1019 if (input_cd
!= (iconv_t
) -1)
1020 iconv_close (input_cd
);
1021 goto invalid_encoding
;
1025 id
= scm_gc_malloc_pointerless (sizeof (*id
), "iconv descriptors");
1026 id
->input_cd
= input_cd
;
1027 id
->output_cd
= output_cd
;
1029 /* Register a finalizer to close the descriptors. */
1030 scm_i_set_finalizer (id
, finalize_iconv_descriptors
, NULL
);
1037 err
= scm_from_latin1_string (encoding
);
1038 scm_misc_error ("open_iconv_descriptors",
1039 "invalid or unknown character encoding ~s",
1045 close_iconv_descriptors (scm_t_iconv_descriptors
*id
)
1047 if (id
->input_cd
!= (iconv_t
) -1)
1048 iconv_close (id
->input_cd
);
1049 if (id
->output_cd
!= (iconv_t
) -1)
1050 iconv_close (id
->output_cd
);
1051 id
->input_cd
= (void *) -1;
1052 id
->output_cd
= (void *) -1;
1055 scm_t_iconv_descriptors
*
1056 scm_i_port_iconv_descriptors (SCM port
)
1060 pt
= SCM_PTAB_ENTRY (port
);
1062 assert (pt
->encoding_mode
== SCM_PORT_ENCODING_MODE_ICONV
);
1064 if (!pt
->iconv_descriptors
)
1067 pt
->encoding
= "ISO-8859-1";
1068 pt
->iconv_descriptors
=
1069 open_iconv_descriptors (pt
->encoding
,
1070 SCM_INPUT_PORT_P (port
),
1071 SCM_OUTPUT_PORT_P (port
));
1074 return pt
->iconv_descriptors
;
1077 /* The name of the encoding is itself encoded in ASCII. */
1079 scm_i_set_port_encoding_x (SCM port
, const char *encoding
)
1082 scm_t_iconv_descriptors
*prev
;
1084 /* Set the character encoding for this port. */
1085 pt
= SCM_PTAB_ENTRY (port
);
1086 prev
= pt
->iconv_descriptors
;
1088 if (encoding_matches (encoding
, "UTF-8"))
1090 pt
->encoding
= "UTF-8";
1091 pt
->encoding_mode
= SCM_PORT_ENCODING_MODE_UTF8
;
1092 pt
->iconv_descriptors
= NULL
;
1094 else if (encoding_matches (encoding
, "ISO-8859-1"))
1096 pt
->encoding
= "ISO-8859-1";
1097 pt
->encoding_mode
= SCM_PORT_ENCODING_MODE_LATIN1
;
1098 pt
->iconv_descriptors
= NULL
;
1102 /* Open descriptors before mutating the port. */
1103 char *gc_encoding
= canonicalize_encoding (encoding
);
1104 pt
->iconv_descriptors
=
1105 open_iconv_descriptors (gc_encoding
,
1106 SCM_INPUT_PORT_P (port
),
1107 SCM_OUTPUT_PORT_P (port
));
1108 pt
->encoding
= gc_encoding
;
1109 pt
->encoding_mode
= SCM_PORT_ENCODING_MODE_ICONV
;
1113 close_iconv_descriptors (prev
);
1116 SCM_DEFINE (scm_port_encoding
, "port-encoding", 1, 0, 0,
1118 "Returns, as a string, the character encoding that @var{port}\n"
1119 "uses to interpret its input and output.\n")
1120 #define FUNC_NAME s_scm_port_encoding
1122 SCM_VALIDATE_PORT (1, port
);
1124 return scm_from_latin1_string (SCM_PTAB_ENTRY (port
)->encoding
);
1128 SCM_DEFINE (scm_set_port_encoding_x
, "set-port-encoding!", 2, 0, 0,
1129 (SCM port
, SCM enc
),
1130 "Sets the character encoding that will be used to interpret all\n"
1131 "port I/O. New ports are created with the encoding\n"
1132 "appropriate for the current locale if @code{setlocale} has \n"
1133 "been called or ISO-8859-1 otherwise\n"
1134 "and this procedure can be used to modify that encoding.\n")
1135 #define FUNC_NAME s_scm_set_port_encoding_x
1139 SCM_VALIDATE_PORT (1, port
);
1140 SCM_VALIDATE_STRING (2, enc
);
1142 enc_str
= scm_to_latin1_string (enc
);
1143 scm_i_set_port_encoding_x (port
, enc_str
);
1146 return SCM_UNSPECIFIED
;
1150 SCM_DEFINE (scm_port_conversion_strategy
, "port-conversion-strategy",
1151 1, 0, 0, (SCM port
),
1152 "Returns the behavior of the port when handling a character that\n"
1153 "is not representable in the port's current encoding.\n"
1154 "It returns the symbol @code{error} if unrepresentable characters\n"
1155 "should cause exceptions, @code{substitute} if the port should\n"
1156 "try to replace unrepresentable characters with question marks or\n"
1157 "approximate characters, or @code{escape} if unrepresentable\n"
1158 "characters should be converted to string escapes.\n"
1160 "If @var{port} is @code{#f}, then the current default behavior\n"
1161 "will be returned. New ports will have this default behavior\n"
1162 "when they are created.\n")
1163 #define FUNC_NAME s_scm_port_conversion_strategy
1165 scm_t_string_failed_conversion_handler h
;
1167 SCM_VALIDATE_OPPORT (1, port
);
1169 if (scm_is_false (port
))
1170 h
= scm_i_default_port_conversion_handler ();
1175 SCM_VALIDATE_OPPORT (1, port
);
1176 pt
= SCM_PTAB_ENTRY (port
);
1178 h
= pt
->ilseq_handler
;
1181 if (h
== SCM_FAILED_CONVERSION_ERROR
)
1182 return scm_from_latin1_symbol ("error");
1183 else if (h
== SCM_FAILED_CONVERSION_QUESTION_MARK
)
1184 return scm_from_latin1_symbol ("substitute");
1185 else if (h
== SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
)
1186 return scm_from_latin1_symbol ("escape");
1190 /* Never gets here. */
1191 return SCM_UNDEFINED
;
1195 SCM_DEFINE (scm_set_port_conversion_strategy_x
, "set-port-conversion-strategy!",
1197 (SCM port
, SCM sym
),
1198 "Sets the behavior of the interpreter when outputting a character\n"
1199 "that is not representable in the port's current encoding.\n"
1200 "@var{sym} can be either @code{'error}, @code{'substitute}, or\n"
1201 "@code{'escape}. If it is @code{'error}, an error will be thrown\n"
1202 "when an unconvertible character is encountered. If it is\n"
1203 "@code{'substitute}, then unconvertible characters will \n"
1204 "be replaced with approximate characters, or with question marks\n"
1205 "if no approximately correct character is available.\n"
1206 "If it is @code{'escape},\n"
1207 "it will appear as a hex escape when output.\n"
1209 "If @var{port} is an open port, the conversion error behavior\n"
1210 "is set for that port. If it is @code{#f}, it is set as the\n"
1211 "default behavior for any future ports that get created in\n"
1213 #define FUNC_NAME s_scm_set_port_conversion_strategy_x
1215 scm_t_string_failed_conversion_handler handler
;
1217 if (scm_is_eq (sym
, sym_error
))
1218 handler
= SCM_FAILED_CONVERSION_ERROR
;
1219 else if (scm_is_eq (sym
, sym_substitute
))
1220 handler
= SCM_FAILED_CONVERSION_QUESTION_MARK
;
1221 else if (scm_is_eq (sym
, sym_escape
))
1222 handler
= SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
;
1224 SCM_MISC_ERROR ("unknown conversion strategy ~s", scm_list_1 (sym
));
1226 if (scm_is_false (port
))
1227 scm_i_set_default_port_conversion_handler (handler
);
1230 SCM_VALIDATE_OPPORT (1, port
);
1231 SCM_PTAB_ENTRY (port
)->ilseq_handler
= handler
;
1234 return SCM_UNSPECIFIED
;
1241 /* The port lock. */
1244 lock_port (void *mutex
)
1246 scm_i_pthread_mutex_lock (mutex
);
1250 unlock_port (void *mutex
)
1252 scm_i_pthread_mutex_unlock (mutex
);
1256 scm_dynwind_lock_port (SCM port
)
1257 #define FUNC_NAME "dynwind-lock-port"
1259 scm_i_pthread_mutex_t
*lock
;
1260 SCM_VALIDATE_OPPORT (SCM_ARG1
, port
);
1261 scm_c_lock_port (port
, &lock
);
1264 scm_dynwind_unwind_handler (unlock_port
, lock
, SCM_F_WIND_EXPLICITLY
);
1265 scm_dynwind_rewind_handler (lock_port
, lock
, 0);
1276 scm_get_byte_or_eof (SCM port
)
1278 scm_i_pthread_mutex_t
*lock
;
1281 scm_c_lock_port (port
, &lock
);
1282 ret
= scm_get_byte_or_eof_unlocked (port
);
1284 scm_i_pthread_mutex_unlock (lock
);
1290 scm_peek_byte_or_eof (SCM port
)
1292 scm_i_pthread_mutex_t
*lock
;
1295 scm_c_lock_port (port
, &lock
);
1296 ret
= scm_peek_byte_or_eof_unlocked (port
);
1298 scm_i_pthread_mutex_unlock (lock
);
1305 * Used by an application to read arbitrary number of bytes from an
1306 * SCM port. Same semantics as libc read, except that scm_c_read only
1307 * returns less than SIZE bytes if at end-of-file.
1309 * Warning: Doesn't update port line and column counts! */
1311 /* This structure, and the following swap_buffer function, are used
1312 for temporarily swapping a port's own read buffer, and the buffer
1313 that the caller of scm_c_read provides. */
1314 struct port_and_swap_buffer
1317 unsigned char *buffer
;
1322 swap_buffer (void *data
)
1324 struct port_and_swap_buffer
*psb
= (struct port_and_swap_buffer
*) data
;
1325 unsigned char *old_buf
= psb
->pt
->read_buf
;
1326 size_t old_size
= psb
->pt
->read_buf_size
;
1328 /* Make the port use (buffer, size) from the struct. */
1329 psb
->pt
->read_pos
= psb
->pt
->read_buf
= psb
->pt
->read_end
= psb
->buffer
;
1330 psb
->pt
->read_buf_size
= psb
->size
;
1332 /* Save the port's old (buffer, size) in the struct. */
1333 psb
->buffer
= old_buf
;
1334 psb
->size
= old_size
;
1338 scm_c_read_unlocked (SCM port
, void *buffer
, size_t size
)
1339 #define FUNC_NAME "scm_c_read"
1342 size_t n_read
= 0, n_available
;
1343 struct port_and_swap_buffer psb
;
1345 SCM_VALIDATE_OPINPORT (1, port
);
1347 pt
= SCM_PTAB_ENTRY (port
);
1348 if (pt
->rw_active
== SCM_PORT_WRITE
)
1349 SCM_PORT_DESCRIPTOR (port
)->flush (port
);
1352 pt
->rw_active
= SCM_PORT_READ
;
1354 /* Take bytes first from the port's read buffer. */
1355 if (pt
->read_pos
< pt
->read_end
)
1357 n_available
= min (size
, pt
->read_end
- pt
->read_pos
);
1358 memcpy (buffer
, pt
->read_pos
, n_available
);
1359 buffer
= (char *) buffer
+ n_available
;
1360 pt
->read_pos
+= n_available
;
1361 n_read
+= n_available
;
1362 size
-= n_available
;
1365 /* Avoid the scm_dynwind_* costs if we now have enough data. */
1369 /* Now we will call scm_fill_input repeatedly until we have read the
1370 requested number of bytes. (Note that a single scm_fill_input
1371 call does not guarantee to fill the whole of the port's read
1373 if (pt
->read_buf_size
<= 1
1374 && pt
->encoding_mode
== SCM_PORT_ENCODING_MODE_LATIN1
)
1376 /* The port that we are reading from is unbuffered - i.e. does
1377 not have its own persistent buffer - but we have a buffer,
1378 provided by our caller, that is the right size for the data
1379 that is wanted. For the following scm_fill_input calls,
1380 therefore, we use the buffer in hand as the port's read
1383 We need to make sure that the port's normal (1 byte) buffer
1384 is reinstated in case one of the scm_fill_input () calls
1385 throws an exception; we use the scm_dynwind_* API to achieve
1388 A consequence of this optimization is that the fill_input
1389 functions can't unget characters. That'll push data to the
1390 pushback buffer instead of this psb buffer. */
1392 unsigned char *pback
= pt
->putback_buf
;
1395 psb
.buffer
= buffer
;
1397 scm_dynwind_begin (SCM_F_DYNWIND_REWINDABLE
);
1398 scm_dynwind_rewind_handler (swap_buffer
, &psb
, SCM_F_WIND_EXPLICITLY
);
1399 scm_dynwind_unwind_handler (swap_buffer
, &psb
, SCM_F_WIND_EXPLICITLY
);
1401 /* Call scm_fill_input until we have all the bytes that we need,
1403 while (pt
->read_buf_size
&& (scm_fill_input_unlocked (port
) != EOF
))
1405 pt
->read_buf_size
-= (pt
->read_end
- pt
->read_pos
);
1406 pt
->read_pos
= pt
->read_buf
= pt
->read_end
;
1409 if (pback
!= pt
->putback_buf
1410 || pt
->read_buf
- (unsigned char *) buffer
< 0)
1411 scm_misc_error (FUNC_NAME
,
1412 "scm_c_read must not call a fill function that pushes "
1413 "back characters onto an unbuffered port", SCM_EOL
);
1415 n_read
+= pt
->read_buf
- (unsigned char *) buffer
;
1417 /* Reinstate the port's normal buffer. */
1422 /* The port has its own buffer. It is important that we use it,
1423 even if it happens to be smaller than our caller's buffer, so
1424 that a custom port implementation's entry points (in
1425 particular, fill_input) can rely on the buffer always being
1426 the same as they first set up. */
1427 while (size
&& (scm_fill_input_unlocked (port
) != EOF
))
1429 n_available
= min (size
, pt
->read_end
- pt
->read_pos
);
1430 memcpy (buffer
, pt
->read_pos
, n_available
);
1431 buffer
= (char *) buffer
+ n_available
;
1432 pt
->read_pos
+= n_available
;
1433 n_read
+= n_available
;
1434 size
-= n_available
;
1443 scm_c_read (SCM port
, void *buffer
, size_t size
)
1445 scm_i_pthread_mutex_t
*lock
;
1448 scm_c_lock_port (port
, &lock
);
1449 ret
= scm_c_read_unlocked (port
, buffer
, size
);
1451 scm_i_pthread_mutex_unlock (lock
);
1457 /* Update the line and column number of PORT after consumption of C. */
1459 update_port_lf (scm_t_wchar c
, SCM port
)
1484 #define SCM_MBCHAR_BUF_SIZE (4)
1486 /* Convert the SIZE-byte UTF-8 sequence in UTF8_BUF to a codepoint.
1487 UTF8_BUF is assumed to contain a valid UTF-8 sequence. */
1489 utf8_to_codepoint (const scm_t_uint8
*utf8_buf
, size_t size
)
1491 scm_t_wchar codepoint
;
1493 if (utf8_buf
[0] <= 0x7f)
1496 codepoint
= utf8_buf
[0];
1498 else if ((utf8_buf
[0] & 0xe0) == 0xc0)
1501 codepoint
= ((scm_t_wchar
) utf8_buf
[0] & 0x1f) << 6UL
1502 | (utf8_buf
[1] & 0x3f);
1504 else if ((utf8_buf
[0] & 0xf0) == 0xe0)
1507 codepoint
= ((scm_t_wchar
) utf8_buf
[0] & 0x0f) << 12UL
1508 | ((scm_t_wchar
) utf8_buf
[1] & 0x3f) << 6UL
1509 | (utf8_buf
[2] & 0x3f);
1514 codepoint
= ((scm_t_wchar
) utf8_buf
[0] & 0x07) << 18UL
1515 | ((scm_t_wchar
) utf8_buf
[1] & 0x3f) << 12UL
1516 | ((scm_t_wchar
) utf8_buf
[2] & 0x3f) << 6UL
1517 | (utf8_buf
[3] & 0x3f);
1523 /* Read a UTF-8 sequence from PORT. On success, return 0 and set
1524 *CODEPOINT to the codepoint that was read, fill BUF with its UTF-8
1525 representation, and set *LEN to the length in bytes. Return
1526 `EILSEQ' on error. */
1528 get_utf8_codepoint (SCM port
, scm_t_wchar
*codepoint
,
1529 scm_t_uint8 buf
[SCM_MBCHAR_BUF_SIZE
], size_t *len
)
1531 #define ASSERT_NOT_EOF(b) \
1532 if (SCM_UNLIKELY ((b) == EOF)) \
1534 #define CONSUME_PEEKED_BYTE() \
1541 pt
= SCM_PTAB_ENTRY (port
);
1543 byte
= scm_get_byte_or_eof_unlocked (port
);
1550 buf
[0] = (scm_t_uint8
) byte
;
1555 *codepoint
= buf
[0];
1556 else if (buf
[0] >= 0xc2 && buf
[0] <= 0xdf)
1559 byte
= scm_peek_byte_or_eof_unlocked (port
);
1560 ASSERT_NOT_EOF (byte
);
1562 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80))
1565 CONSUME_PEEKED_BYTE ();
1566 buf
[1] = (scm_t_uint8
) byte
;
1569 *codepoint
= ((scm_t_wchar
) buf
[0] & 0x1f) << 6UL
1572 else if ((buf
[0] & 0xf0) == 0xe0)
1575 byte
= scm_peek_byte_or_eof_unlocked (port
);
1576 ASSERT_NOT_EOF (byte
);
1578 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80
1579 || (buf
[0] == 0xe0 && byte
< 0xa0)
1580 || (buf
[0] == 0xed && byte
> 0x9f)))
1583 CONSUME_PEEKED_BYTE ();
1584 buf
[1] = (scm_t_uint8
) byte
;
1587 byte
= scm_peek_byte_or_eof_unlocked (port
);
1588 ASSERT_NOT_EOF (byte
);
1590 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80))
1593 CONSUME_PEEKED_BYTE ();
1594 buf
[2] = (scm_t_uint8
) byte
;
1597 *codepoint
= ((scm_t_wchar
) buf
[0] & 0x0f) << 12UL
1598 | ((scm_t_wchar
) buf
[1] & 0x3f) << 6UL
1601 else if (buf
[0] >= 0xf0 && buf
[0] <= 0xf4)
1604 byte
= scm_peek_byte_or_eof_unlocked (port
);
1605 ASSERT_NOT_EOF (byte
);
1607 if (SCM_UNLIKELY (((byte
& 0xc0) != 0x80)
1608 || (buf
[0] == 0xf0 && byte
< 0x90)
1609 || (buf
[0] == 0xf4 && byte
> 0x8f)))
1612 CONSUME_PEEKED_BYTE ();
1613 buf
[1] = (scm_t_uint8
) byte
;
1616 byte
= scm_peek_byte_or_eof_unlocked (port
);
1617 ASSERT_NOT_EOF (byte
);
1619 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80))
1622 CONSUME_PEEKED_BYTE ();
1623 buf
[2] = (scm_t_uint8
) byte
;
1626 byte
= scm_peek_byte_or_eof_unlocked (port
);
1627 ASSERT_NOT_EOF (byte
);
1629 if (SCM_UNLIKELY ((byte
& 0xc0) != 0x80))
1632 CONSUME_PEEKED_BYTE ();
1633 buf
[3] = (scm_t_uint8
) byte
;
1636 *codepoint
= ((scm_t_wchar
) buf
[0] & 0x07) << 18UL
1637 | ((scm_t_wchar
) buf
[1] & 0x3f) << 12UL
1638 | ((scm_t_wchar
) buf
[2] & 0x3f) << 6UL
1647 /* Here we could choose the consume the faulty byte when it's not a
1648 valid starting byte, but it's not a requirement. What Section 3.9
1649 of Unicode 6.0.0 mandates, though, is to not consume a byte that
1650 would otherwise be a valid starting byte. */
1654 #undef CONSUME_PEEKED_BYTE
1655 #undef ASSERT_NOT_EOF
1658 /* Read an ISO-8859-1 codepoint (a byte) from PORT. On success, return
1659 0 and set *CODEPOINT to the codepoint that was read, fill BUF with
1660 its UTF-8 representation, and set *LEN to the length in bytes.
1661 Return `EILSEQ' on error. */
1663 get_latin1_codepoint (SCM port
, scm_t_wchar
*codepoint
,
1664 char buf
[SCM_MBCHAR_BUF_SIZE
], size_t *len
)
1666 *codepoint
= scm_get_byte_or_eof_unlocked (port
);
1668 if (*codepoint
== EOF
)
1673 buf
[0] = *codepoint
;
1678 /* Likewise, read a byte sequence from PORT, passing it through its
1679 input conversion descriptor. */
1681 get_iconv_codepoint (SCM port
, scm_t_wchar
*codepoint
,
1682 char buf
[SCM_MBCHAR_BUF_SIZE
], size_t *len
)
1684 scm_t_iconv_descriptors
*id
;
1686 size_t bytes_consumed
, output_size
;
1688 scm_t_uint8 utf8_buf
[SCM_MBCHAR_BUF_SIZE
];
1690 id
= scm_i_port_iconv_descriptors (port
);
1692 for (output_size
= 0, output
= (char *) utf8_buf
,
1693 bytes_consumed
= 0, err
= 0;
1694 err
== 0 && output_size
== 0
1695 && (bytes_consumed
== 0 || byte_read
!= EOF
);
1699 size_t input_left
, output_left
, done
;
1701 byte_read
= scm_get_byte_or_eof_unlocked (port
);
1702 if (byte_read
== EOF
)
1704 if (bytes_consumed
== 0)
1706 *codepoint
= (scm_t_wchar
) EOF
;
1714 buf
[bytes_consumed
] = byte_read
;
1717 input_left
= bytes_consumed
+ 1;
1718 output_left
= sizeof (utf8_buf
);
1720 done
= iconv (id
->input_cd
, &input
, &input_left
, &output
, &output_left
);
1721 if (done
== (size_t) -1)
1725 /* Missing input: keep trying. */
1729 output_size
= sizeof (utf8_buf
) - output_left
;
1732 if (SCM_UNLIKELY (output_size
== 0))
1733 /* An unterminated sequence. */
1735 else if (SCM_LIKELY (err
== 0))
1737 /* Convert the UTF8_BUF sequence to a Unicode code point. */
1738 *codepoint
= utf8_to_codepoint (utf8_buf
, output_size
);
1739 *len
= bytes_consumed
;
1745 /* Read a codepoint from PORT and return it in *CODEPOINT. Fill BUF
1746 with the byte representation of the codepoint in PORT's encoding, and
1747 set *LEN to the length in bytes of that representation. Return 0 on
1748 success and an errno value on error. */
1749 static SCM_C_INLINE
int
1750 get_codepoint (SCM port
, scm_t_wchar
*codepoint
,
1751 char buf
[SCM_MBCHAR_BUF_SIZE
], size_t *len
)
1754 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1756 if (pt
->encoding_mode
== SCM_PORT_ENCODING_MODE_UTF8
)
1757 err
= get_utf8_codepoint (port
, codepoint
, (scm_t_uint8
*) buf
, len
);
1758 else if (pt
->encoding_mode
== SCM_PORT_ENCODING_MODE_LATIN1
)
1759 err
= get_latin1_codepoint (port
, codepoint
, buf
, len
);
1761 err
= get_iconv_codepoint (port
, codepoint
, buf
, len
);
1763 if (SCM_LIKELY (err
== 0))
1764 update_port_lf (*codepoint
, port
);
1765 else if (pt
->ilseq_handler
== SCM_ICONVEH_QUESTION_MARK
)
1769 update_port_lf (*codepoint
, port
);
1775 /* Read a codepoint from PORT and return it. */
1777 scm_getc_unlocked (SCM port
)
1778 #define FUNC_NAME "scm_getc"
1782 scm_t_wchar codepoint
;
1783 char buf
[SCM_MBCHAR_BUF_SIZE
];
1785 err
= get_codepoint (port
, &codepoint
, buf
, &len
);
1786 if (SCM_UNLIKELY (err
!= 0))
1787 /* At this point PORT should point past the invalid encoding, as per
1788 R6RS-lib Section 8.2.4. */
1789 scm_decoding_error (FUNC_NAME
, err
, "input decoding error", port
);
1798 scm_i_pthread_mutex_t
*lock
;
1801 scm_c_lock_port (port
, &lock
);
1802 ret
= scm_getc_unlocked (port
);
1804 scm_i_pthread_mutex_unlock (lock
);
1810 SCM_DEFINE (scm_read_char
, "read-char", 0, 1, 0,
1812 "Return the next character available from @var{port}, updating\n"
1813 "@var{port} to point to the following character. If no more\n"
1814 "characters are available, the end-of-file object is returned.\n"
1816 "When @var{port}'s data cannot be decoded according to its\n"
1817 "character encoding, a @code{decoding-error} is raised and\n"
1818 "@var{port} points past the erroneous byte sequence.\n")
1819 #define FUNC_NAME s_scm_read_char
1822 if (SCM_UNBNDP (port
))
1823 port
= scm_current_input_port ();
1824 SCM_VALIDATE_OPINPORT (1, port
);
1825 c
= scm_getc_unlocked (port
);
1828 return SCM_MAKE_CHAR (c
);
1838 scm_unget_byte_unlocked (int c
, SCM port
)
1839 #define FUNC_NAME "scm_unget_byte"
1841 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
1843 if (pt
->read_buf
== pt
->putback_buf
)
1844 /* already using the put-back buffer. */
1846 /* enlarge putback_buf if necessary. */
1847 if (pt
->read_end
== pt
->read_buf
+ pt
->read_buf_size
1848 && pt
->read_buf
== pt
->read_pos
)
1850 size_t new_size
= pt
->read_buf_size
* 2;
1851 unsigned char *tmp
= (unsigned char *)
1852 scm_gc_realloc (pt
->putback_buf
, pt
->read_buf_size
, new_size
,
1855 pt
->read_pos
= pt
->read_buf
= pt
->putback_buf
= tmp
;
1856 pt
->read_end
= pt
->read_buf
+ pt
->read_buf_size
;
1857 pt
->read_buf_size
= pt
->putback_buf_size
= new_size
;
1860 /* shift any existing bytes to buffer + 1. */
1861 if (pt
->read_pos
== pt
->read_end
)
1862 pt
->read_end
= pt
->read_buf
+ 1;
1863 else if (pt
->read_pos
!= pt
->read_buf
+ 1)
1865 int count
= pt
->read_end
- pt
->read_pos
;
1867 memmove (pt
->read_buf
+ 1, pt
->read_pos
, count
);
1868 pt
->read_end
= pt
->read_buf
+ 1 + count
;
1871 pt
->read_pos
= pt
->read_buf
;
1874 /* switch to the put-back buffer. */
1876 if (pt
->putback_buf
== NULL
)
1879 = (unsigned char *) scm_gc_malloc_pointerless
1880 (SCM_INITIAL_PUTBACK_BUF_SIZE
, "putback buffer");
1881 pt
->putback_buf_size
= SCM_INITIAL_PUTBACK_BUF_SIZE
;
1884 pt
->saved_read_buf
= pt
->read_buf
;
1885 pt
->saved_read_pos
= pt
->read_pos
;
1886 pt
->saved_read_end
= pt
->read_end
;
1887 pt
->saved_read_buf_size
= pt
->read_buf_size
;
1889 pt
->read_pos
= pt
->read_buf
= pt
->putback_buf
;
1890 pt
->read_end
= pt
->read_buf
+ 1;
1891 pt
->read_buf_size
= pt
->putback_buf_size
;
1897 pt
->rw_active
= SCM_PORT_READ
;
1902 scm_unget_byte (int c
, SCM port
)
1904 scm_i_pthread_mutex_t
*lock
;
1905 scm_c_lock_port (port
, &lock
);
1906 scm_unget_byte_unlocked (c
, port
);
1908 scm_i_pthread_mutex_unlock (lock
);
1913 scm_ungetc_unlocked (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];
1922 len
= sizeof (result_buf
);
1923 result
= u32_conv_to_encoding (pt
->encoding
,
1924 (enum iconv_ilseq_handler
) pt
->ilseq_handler
,
1925 (uint32_t *) &c
, 1, NULL
,
1928 if (SCM_UNLIKELY (result
== NULL
|| len
== 0))
1929 scm_encoding_error (FUNC_NAME
, errno
,
1930 "conversion to port encoding failed",
1931 SCM_BOOL_F
, SCM_MAKE_CHAR (c
));
1933 for (i
= len
- 1; i
>= 0; i
--)
1934 scm_unget_byte_unlocked (result
[i
], port
);
1936 if (SCM_UNLIKELY (result
!= result_buf
))
1941 /* What should col be in this case?
1942 * We'll leave it at -1.
1944 SCM_LINUM (port
) -= 1;
1952 scm_ungetc (scm_t_wchar c
, SCM port
)
1954 scm_i_pthread_mutex_t
*lock
;
1955 scm_c_lock_port (port
, &lock
);
1956 scm_ungetc_unlocked (c
, port
);
1958 scm_i_pthread_mutex_unlock (lock
);
1963 scm_ungets_unlocked (const char *s
, int n
, SCM port
)
1965 /* This is simple minded and inefficient, but unreading strings is
1966 * probably not a common operation, and remember that line and
1967 * column numbers have to be handled...
1969 * Please feel free to write an optimized version!
1972 scm_ungetc_unlocked (s
[n
], port
);
1976 scm_ungets (const char *s
, int n
, SCM port
)
1978 scm_i_pthread_mutex_t
*lock
;
1979 scm_c_lock_port (port
, &lock
);
1980 scm_ungets_unlocked (s
, n
, port
);
1982 scm_i_pthread_mutex_unlock (lock
);
1986 SCM_DEFINE (scm_peek_char
, "peek-char", 0, 1, 0,
1988 "Return the next character available from @var{port},\n"
1989 "@emph{without} updating @var{port} to point to the following\n"
1990 "character. If no more characters are available, the\n"
1991 "end-of-file object is returned.\n"
1993 "The value returned by\n"
1994 "a call to @code{peek-char} is the same as the value that would\n"
1995 "have been returned by a call to @code{read-char} on the same\n"
1996 "port. The only difference is that the very next call to\n"
1997 "@code{read-char} or @code{peek-char} on that @var{port} will\n"
1998 "return the value returned by the preceding call to\n"
1999 "@code{peek-char}. In particular, a call to @code{peek-char} on\n"
2000 "an interactive port will hang waiting for input whenever a call\n"
2001 "to @code{read-char} would have hung.\n"
2003 "As for @code{read-char}, a @code{decoding-error} may be raised\n"
2004 "if such a situation occurs. However, unlike with @code{read-char},\n"
2005 "@var{port} still points at the beginning of the erroneous byte\n"
2006 "sequence when the error is raised.\n")
2007 #define FUNC_NAME s_scm_peek_char
2012 char bytes
[SCM_MBCHAR_BUF_SIZE
];
2013 long column
, line
, i
;
2016 if (SCM_UNBNDP (port
))
2017 port
= scm_current_input_port ();
2018 SCM_VALIDATE_OPINPORT (1, port
);
2020 column
= SCM_COL (port
);
2021 line
= SCM_LINUM (port
);
2023 err
= get_codepoint (port
, &c
, bytes
, &len
);
2025 for (i
= len
- 1; i
>= 0; i
--)
2026 scm_unget_byte_unlocked (bytes
[i
], port
);
2028 SCM_COL (port
) = column
;
2029 SCM_LINUM (port
) = line
;
2031 if (SCM_UNLIKELY (err
!= 0))
2033 scm_decoding_error (FUNC_NAME
, err
, "input decoding error", port
);
2035 /* Shouldn't happen since `catch' always aborts to prompt. */
2036 result
= SCM_BOOL_F
;
2039 result
= SCM_EOF_VAL
;
2041 result
= SCM_MAKE_CHAR (c
);
2047 SCM_DEFINE (scm_unread_char
, "unread-char", 1, 1, 0,
2048 (SCM cobj
, SCM port
),
2049 "Place character @var{cobj} in @var{port} so that it will be\n"
2050 "read by the next read operation. If called multiple times, the\n"
2051 "unread characters will be read again in last-in first-out\n"
2052 "order. If @var{port} is not supplied, the current input port\n"
2054 #define FUNC_NAME s_scm_unread_char
2058 SCM_VALIDATE_CHAR (1, cobj
);
2059 if (SCM_UNBNDP (port
))
2060 port
= scm_current_input_port ();
2061 SCM_VALIDATE_OPINPORT (2, port
);
2063 c
= SCM_CHAR (cobj
);
2065 scm_ungetc_unlocked (c
, port
);
2070 SCM_DEFINE (scm_unread_string
, "unread-string", 2, 0, 0,
2071 (SCM str
, SCM port
),
2072 "Place the string @var{str} in @var{port} so that its characters will be\n"
2073 "read in subsequent read operations. If called multiple times, the\n"
2074 "unread characters will be read again in last-in first-out order. If\n"
2075 "@var{port} is not supplied, the current-input-port is used.")
2076 #define FUNC_NAME s_scm_unread_string
2079 SCM_VALIDATE_STRING (1, str
);
2080 if (SCM_UNBNDP (port
))
2081 port
= scm_current_input_port ();
2082 SCM_VALIDATE_OPINPORT (2, port
);
2084 n
= scm_i_string_length (str
);
2087 scm_ungetc_unlocked (scm_i_string_ref (str
, n
), port
);
2096 /* Manipulating the buffers. */
2098 /* This routine does not take any locks, as it is usually called as part
2099 of a port implementation. */
2101 scm_port_non_buffer (scm_t_port
*pt
)
2103 pt
->read_pos
= pt
->read_buf
= pt
->read_end
= &pt
->shortbuf
;
2104 pt
->write_buf
= pt
->write_pos
= &pt
->shortbuf
;
2105 pt
->read_buf_size
= pt
->write_buf_size
= 1;
2106 pt
->write_end
= pt
->write_buf
+ pt
->write_buf_size
;
2109 /* this should only be called when the read buffer is empty. it
2110 tries to refill the read buffer. it returns the first char from
2111 the port, which is either EOF or *(pt->read_pos). */
2113 scm_fill_input_unlocked (SCM port
)
2115 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
2117 assert (pt
->read_pos
== pt
->read_end
);
2119 if (pt
->read_buf
== pt
->putback_buf
)
2121 /* finished reading put-back chars. */
2122 pt
->read_buf
= pt
->saved_read_buf
;
2123 pt
->read_pos
= pt
->saved_read_pos
;
2124 pt
->read_end
= pt
->saved_read_end
;
2125 pt
->read_buf_size
= pt
->saved_read_buf_size
;
2126 if (pt
->read_pos
< pt
->read_end
)
2127 return *(pt
->read_pos
);
2129 return SCM_PORT_DESCRIPTOR (port
)->fill_input (port
);
2133 scm_fill_input (SCM port
)
2135 scm_i_pthread_mutex_t
*lock
;
2138 scm_c_lock_port (port
, &lock
);
2139 ret
= scm_fill_input_unlocked (port
);
2141 scm_i_pthread_mutex_unlock (lock
);
2147 /* Move up to READ_LEN bytes from PORT's putback and/or read buffers
2148 into memory starting at DEST. Return the number of bytes moved.
2149 PORT's line/column numbers are left unchanged. */
2151 scm_take_from_input_buffers (SCM port
, char *dest
, size_t read_len
)
2153 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
2154 size_t bytes_read
= 0;
2155 size_t from_buf
= min (pt
->read_end
- pt
->read_pos
, read_len
);
2159 memcpy (dest
, pt
->read_pos
, from_buf
);
2160 pt
->read_pos
+= from_buf
;
2161 bytes_read
+= from_buf
;
2162 read_len
-= from_buf
;
2166 /* if putback was active, try the real input buffer too. */
2167 if (pt
->read_buf
== pt
->putback_buf
)
2169 from_buf
= min (pt
->saved_read_end
- pt
->saved_read_pos
, read_len
);
2172 memcpy (dest
, pt
->saved_read_pos
, from_buf
);
2173 pt
->saved_read_pos
+= from_buf
;
2174 bytes_read
+= from_buf
;
2181 /* Clear a port's read buffers, returning the contents. */
2182 SCM_DEFINE (scm_drain_input
, "drain-input", 1, 0, 0,
2184 "This procedure clears a port's input buffers, similar\n"
2185 "to the way that force-output clears the output buffer. The\n"
2186 "contents of the buffers are returned as a single string, e.g.,\n"
2189 "(define p (open-input-file ...))\n"
2190 "(drain-input p) => empty string, nothing buffered yet.\n"
2191 "(unread-char (read-char p) p)\n"
2192 "(drain-input p) => initial chars from p, up to the buffer size.\n"
2194 "Draining the buffers may be useful for cleanly finishing\n"
2195 "buffered I/O so that the file descriptor can be used directly\n"
2196 "for further input.")
2197 #define FUNC_NAME s_scm_drain_input
2204 SCM_VALIDATE_OPINPORT (1, port
);
2205 pt
= SCM_PTAB_ENTRY (port
);
2207 count
= pt
->read_end
- pt
->read_pos
;
2208 if (pt
->read_buf
== pt
->putback_buf
)
2209 count
+= pt
->saved_read_end
- pt
->saved_read_pos
;
2213 result
= scm_i_make_string (count
, &data
, 0);
2214 scm_take_from_input_buffers (port
, data
, count
);
2217 result
= scm_nullstr
;
2224 scm_end_input_unlocked (SCM port
)
2227 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
2229 if (pt
->read_buf
== pt
->putback_buf
)
2231 offset
= pt
->read_end
- pt
->read_pos
;
2232 pt
->read_buf
= pt
->saved_read_buf
;
2233 pt
->read_pos
= pt
->saved_read_pos
;
2234 pt
->read_end
= pt
->saved_read_end
;
2235 pt
->read_buf_size
= pt
->saved_read_buf_size
;
2240 SCM_PORT_DESCRIPTOR (port
)->end_input (port
, offset
);
2244 scm_end_input (SCM port
)
2246 scm_i_pthread_mutex_t
*lock
;
2247 scm_c_lock_port (port
, &lock
);
2248 scm_end_input_unlocked (port
);
2250 scm_i_pthread_mutex_unlock (lock
);
2254 SCM_DEFINE (scm_force_output
, "force-output", 0, 1, 0,
2256 "Flush the specified output port, or the current output port if @var{port}\n"
2257 "is omitted. The current output buffer contents are passed to the\n"
2258 "underlying port implementation (e.g., in the case of fports, the\n"
2259 "data will be written to the file and the output buffer will be cleared.)\n"
2260 "It has no effect on an unbuffered port.\n\n"
2261 "The return value is unspecified.")
2262 #define FUNC_NAME s_scm_force_output
2264 if (SCM_UNBNDP (port
))
2265 port
= scm_current_output_port ();
2268 port
= SCM_COERCE_OUTPORT (port
);
2269 SCM_VALIDATE_OPOUTPORT (1, port
);
2271 scm_flush_unlocked (port
);
2272 return SCM_UNSPECIFIED
;
2277 scm_flush_unlocked (SCM port
)
2279 SCM_PORT_DESCRIPTOR (port
)->flush (port
);
2283 scm_flush (SCM port
)
2285 scm_i_pthread_mutex_t
*lock
;
2286 scm_c_lock_port (port
, &lock
);
2287 scm_flush_unlocked (port
);
2289 scm_i_pthread_mutex_unlock (lock
);
2299 scm_putc (char c
, SCM port
)
2301 scm_i_pthread_mutex_t
*lock
;
2302 scm_c_lock_port (port
, &lock
);
2303 scm_putc_unlocked (c
, port
);
2305 scm_i_pthread_mutex_unlock (lock
);
2310 scm_puts (const char *s
, SCM port
)
2312 scm_i_pthread_mutex_t
*lock
;
2313 scm_c_lock_port (port
, &lock
);
2314 scm_puts_unlocked (s
, port
);
2316 scm_i_pthread_mutex_unlock (lock
);
2322 * Used by an application to write arbitrary number of bytes to an SCM
2323 * port. Similar semantics as libc write. However, unlike libc
2324 * write, scm_c_write writes the requested number of bytes and has no
2327 * Warning: Doesn't update port line and column counts!
2330 scm_c_write_unlocked (SCM port
, const void *ptr
, size_t size
)
2331 #define FUNC_NAME "scm_c_write"
2334 scm_t_ptob_descriptor
*ptob
;
2336 SCM_VALIDATE_OPOUTPORT (1, port
);
2338 pt
= SCM_PTAB_ENTRY (port
);
2339 ptob
= SCM_PORT_DESCRIPTOR (port
);
2341 if (pt
->rw_active
== SCM_PORT_READ
)
2342 scm_end_input_unlocked (port
);
2344 ptob
->write (port
, ptr
, size
);
2347 pt
->rw_active
= SCM_PORT_WRITE
;
2352 scm_c_write (SCM port
, const void *ptr
, size_t size
)
2354 scm_i_pthread_mutex_t
*lock
;
2355 scm_c_lock_port (port
, &lock
);
2356 scm_c_write_unlocked (port
, ptr
, size
);
2358 scm_i_pthread_mutex_unlock (lock
);
2364 * This function differs from scm_c_write; it updates port line and
2367 scm_lfwrite_unlocked (const char *ptr
, size_t size
, SCM port
)
2369 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
2370 scm_t_ptob_descriptor
*ptob
= SCM_PORT_DESCRIPTOR (port
);
2372 if (pt
->rw_active
== SCM_PORT_READ
)
2373 scm_end_input_unlocked (port
);
2375 ptob
->write (port
, ptr
, size
);
2377 for (; size
; ptr
++, size
--)
2378 update_port_lf ((scm_t_wchar
) (unsigned char) *ptr
, port
);
2381 pt
->rw_active
= SCM_PORT_WRITE
;
2385 scm_lfwrite (const char *ptr
, size_t size
, SCM port
)
2387 scm_i_pthread_mutex_t
*lock
;
2388 scm_c_lock_port (port
, &lock
);
2389 scm_lfwrite_unlocked (ptr
, size
, port
);
2391 scm_i_pthread_mutex_unlock (lock
);
2395 /* Write STR to PORT from START inclusive to END exclusive. */
2397 scm_lfwrite_substr (SCM str
, size_t start
, size_t end
, SCM port
)
2399 scm_t_port
*pt
= SCM_PTAB_ENTRY (port
);
2401 if (pt
->rw_active
== SCM_PORT_READ
)
2402 scm_end_input_unlocked (port
);
2404 if (end
== (size_t) -1)
2405 end
= scm_i_string_length (str
);
2407 scm_i_display_substring (str
, start
, end
, port
);
2410 pt
->rw_active
= SCM_PORT_WRITE
;
2416 /* Querying and setting positions, and character availability. */
2418 SCM_DEFINE (scm_char_ready_p
, "char-ready?", 0, 1, 0,
2420 "Return @code{#t} if a character is ready on input @var{port}\n"
2421 "and return @code{#f} otherwise. If @code{char-ready?} returns\n"
2422 "@code{#t} then the next @code{read-char} operation on\n"
2423 "@var{port} is guaranteed not to hang. If @var{port} is a file\n"
2424 "port at end of file then @code{char-ready?} returns @code{#t}.\n"
2426 "@code{char-ready?} exists to make it possible for a\n"
2427 "program to accept characters from interactive ports without\n"
2428 "getting stuck waiting for input. Any input editors associated\n"
2429 "with such ports must make sure that characters whose existence\n"
2430 "has been asserted by @code{char-ready?} cannot be rubbed out.\n"
2431 "If @code{char-ready?} were to return @code{#f} at end of file,\n"
2432 "a port at end of file would be indistinguishable from an\n"
2433 "interactive port that has no ready characters.")
2434 #define FUNC_NAME s_scm_char_ready_p
2438 if (SCM_UNBNDP (port
))
2439 port
= scm_current_input_port ();
2440 /* It's possible to close the current input port, so validate even in
2442 SCM_VALIDATE_OPINPORT (1, port
);
2444 pt
= SCM_PTAB_ENTRY (port
);
2446 /* if the current read buffer is filled, or the
2447 last pushed-back char has been read and the saved buffer is
2448 filled, result is true. */
2449 if (pt
->read_pos
< pt
->read_end
2450 || (pt
->read_buf
== pt
->putback_buf
2451 && pt
->saved_read_pos
< pt
->saved_read_end
))
2455 scm_t_ptob_descriptor
*ptob
= SCM_PORT_DESCRIPTOR (port
);
2457 if (ptob
->input_waiting
)
2458 return scm_from_bool(ptob
->input_waiting (port
));
2465 SCM_DEFINE (scm_seek
, "seek", 3, 0, 0,
2466 (SCM fd_port
, SCM offset
, SCM whence
),
2467 "Sets the current position of @var{fd_port} to the integer\n"
2468 "@var{offset}, which is interpreted according to the value of\n"
2471 "One of the following variables should be supplied for\n"
2473 "@defvar SEEK_SET\n"
2474 "Seek from the beginning of the file.\n"
2476 "@defvar SEEK_CUR\n"
2477 "Seek from the current position.\n"
2479 "@defvar SEEK_END\n"
2480 "Seek from the end of the file.\n"
2482 "If @var{fd_port} is a file descriptor, the underlying system\n"
2483 "call is @code{lseek}. @var{port} may be a string port.\n"
2485 "The value returned is the new position in the file. This means\n"
2486 "that the current position of a port can be obtained using:\n"
2488 "(seek port 0 SEEK_CUR)\n"
2490 #define FUNC_NAME s_scm_seek
2494 fd_port
= SCM_COERCE_OUTPORT (fd_port
);
2496 how
= scm_to_int (whence
);
2497 if (how
!= SEEK_SET
&& how
!= SEEK_CUR
&& how
!= SEEK_END
)
2498 SCM_OUT_OF_RANGE (3, whence
);
2500 if (SCM_OPPORTP (fd_port
))
2502 scm_t_ptob_descriptor
*ptob
= SCM_PORT_DESCRIPTOR (fd_port
);
2503 off_t_or_off64_t off
= scm_to_off_t_or_off64_t (offset
);
2504 off_t_or_off64_t rv
;
2507 SCM_MISC_ERROR ("port is not seekable",
2508 scm_cons (fd_port
, SCM_EOL
));
2510 rv
= ptob
->seek (fd_port
, off
, how
);
2511 return scm_from_off_t_or_off64_t (rv
);
2513 else /* file descriptor?. */
2515 off_t_or_off64_t off
= scm_to_off_t_or_off64_t (offset
);
2516 off_t_or_off64_t rv
;
2517 rv
= lseek_or_lseek64 (scm_to_int (fd_port
), off
, how
);
2520 return scm_from_off_t_or_off64_t (rv
);
2529 /* Mingw has ftruncate(), perhaps implemented above using chsize, but
2530 doesn't have the filename version truncate(), hence this code. */
2531 #if HAVE_FTRUNCATE && ! HAVE_TRUNCATE
2533 truncate (const char *file
, off_t length
)
2537 fdes
= open (file
, O_BINARY
| O_WRONLY
);
2541 ret
= ftruncate (fdes
, length
);
2544 int save_errno
= errno
;
2550 return close (fdes
);
2552 #endif /* HAVE_FTRUNCATE && ! HAVE_TRUNCATE */
2554 SCM_DEFINE (scm_truncate_file
, "truncate-file", 1, 1, 0,
2555 (SCM object
, SCM length
),
2556 "Truncate file @var{object} to @var{length} bytes. @var{object}\n"
2557 "can be a filename string, a port object, or an integer file\n"
2559 "The return value is unspecified.\n"
2561 "For a port or file descriptor @var{length} can be omitted, in\n"
2562 "which case the file is truncated at the current position (per\n"
2563 "@code{ftell} above).\n"
2565 "On most systems a file can be extended by giving a length\n"
2566 "greater than the current size, but this is not mandatory in the\n"
2568 #define FUNC_NAME s_scm_truncate_file
2572 /* "object" can be a port, fdes or filename.
2574 Negative "length" makes no sense, but it's left to truncate() or
2575 ftruncate() to give back an error for that (normally EINVAL).
2578 if (SCM_UNBNDP (length
))
2580 /* must supply length if object is a filename. */
2581 if (scm_is_string (object
))
2582 SCM_MISC_ERROR("must supply length if OBJECT is a filename", SCM_EOL
);
2584 length
= scm_seek (object
, SCM_INUM0
, scm_from_int (SEEK_CUR
));
2587 object
= SCM_COERCE_OUTPORT (object
);
2588 if (scm_is_integer (object
))
2590 off_t_or_off64_t c_length
= scm_to_off_t_or_off64_t (length
);
2591 SCM_SYSCALL (rv
= ftruncate_or_ftruncate64 (scm_to_int (object
),
2594 else if (SCM_OPOUTPORTP (object
))
2596 off_t_or_off64_t c_length
= scm_to_off_t_or_off64_t (length
);
2597 scm_t_port
*pt
= SCM_PTAB_ENTRY (object
);
2598 scm_t_ptob_descriptor
*ptob
= SCM_PORT_DESCRIPTOR (object
);
2600 if (!ptob
->truncate
)
2601 SCM_MISC_ERROR ("port is not truncatable", SCM_EOL
);
2602 if (pt
->rw_active
== SCM_PORT_READ
)
2603 scm_end_input_unlocked (object
);
2604 else if (pt
->rw_active
== SCM_PORT_WRITE
)
2605 ptob
->flush (object
);
2607 ptob
->truncate (object
, c_length
);
2612 off_t_or_off64_t c_length
= scm_to_off_t_or_off64_t (length
);
2613 char *str
= scm_to_locale_string (object
);
2615 SCM_SYSCALL (rv
= truncate_or_truncate64 (str
, c_length
));
2622 return SCM_UNSPECIFIED
;
2626 SCM_DEFINE (scm_port_line
, "port-line", 1, 0, 0,
2628 "Return the current line number for @var{port}.\n"
2630 "The first line of a file is 0. But you might want to add 1\n"
2631 "when printing line numbers, since starting from 1 is\n"
2632 "traditional in error messages, and likely to be more natural to\n"
2634 #define FUNC_NAME s_scm_port_line
2636 port
= SCM_COERCE_OUTPORT (port
);
2637 SCM_VALIDATE_OPENPORT (1, port
);
2638 return scm_from_long (SCM_LINUM (port
));
2642 SCM_DEFINE (scm_set_port_line_x
, "set-port-line!", 2, 0, 0,
2643 (SCM port
, SCM line
),
2644 "Set the current line number for @var{port} to @var{line}. The\n"
2645 "first line of a file is 0.")
2646 #define FUNC_NAME s_scm_set_port_line_x
2648 port
= SCM_COERCE_OUTPORT (port
);
2649 SCM_VALIDATE_OPENPORT (1, port
);
2650 SCM_PTAB_ENTRY (port
)->line_number
= scm_to_long (line
);
2651 return SCM_UNSPECIFIED
;
2655 SCM_DEFINE (scm_port_column
, "port-column", 1, 0, 0,
2657 "Return the current column number of @var{port}.\n"
2658 "If the number is\n"
2659 "unknown, the result is #f. Otherwise, the result is a 0-origin integer\n"
2660 "- i.e. the first character of the first line is line 0, column 0.\n"
2661 "(However, when you display a file position, for example in an error\n"
2662 "message, we recommend you add 1 to get 1-origin integers. This is\n"
2663 "because lines and column numbers traditionally start with 1, and that is\n"
2664 "what non-programmers will find most natural.)")
2665 #define FUNC_NAME s_scm_port_column
2667 port
= SCM_COERCE_OUTPORT (port
);
2668 SCM_VALIDATE_OPENPORT (1, port
);
2669 return scm_from_int (SCM_COL (port
));
2673 SCM_DEFINE (scm_set_port_column_x
, "set-port-column!", 2, 0, 0,
2674 (SCM port
, SCM column
),
2675 "Set the current column of @var{port}. Before reading the first\n"
2676 "character on a line the column should be 0.")
2677 #define FUNC_NAME s_scm_set_port_column_x
2679 port
= SCM_COERCE_OUTPORT (port
);
2680 SCM_VALIDATE_OPENPORT (1, port
);
2681 SCM_PTAB_ENTRY (port
)->column_number
= scm_to_int (column
);
2682 return SCM_UNSPECIFIED
;
2686 SCM_DEFINE (scm_port_filename
, "port-filename", 1, 0, 0,
2688 "Return the filename associated with @var{port}, or @code{#f}\n"
2689 "if no filename is associated with the port.")
2690 #define FUNC_NAME s_scm_port_filename
2692 port
= SCM_COERCE_OUTPORT (port
);
2693 SCM_VALIDATE_OPENPORT (1, port
);
2694 return SCM_FILENAME (port
);
2698 SCM_DEFINE (scm_set_port_filename_x
, "set-port-filename!", 2, 0, 0,
2699 (SCM port
, SCM filename
),
2700 "Change the filename associated with @var{port}, using the current input\n"
2701 "port if none is specified. Note that this does not change the port's\n"
2702 "source of data, but only the value that is returned by\n"
2703 "@code{port-filename} and reported in diagnostic output.")
2704 #define FUNC_NAME s_scm_set_port_filename_x
2706 port
= SCM_COERCE_OUTPORT (port
);
2707 SCM_VALIDATE_OPENPORT (1, port
);
2708 /* We allow the user to set the filename to whatever he likes. */
2709 SCM_SET_FILENAME (port
, filename
);
2710 return SCM_UNSPECIFIED
;
2717 /* Implementation helpers for port printing functions. */
2720 scm_print_port_mode (SCM exp
, SCM port
)
2722 scm_puts_unlocked (SCM_CLOSEDP (exp
)
2724 : (SCM_RDNG
& SCM_CELL_WORD_0 (exp
)
2725 ? (SCM_WRTNG
& SCM_CELL_WORD_0 (exp
)
2728 : (SCM_WRTNG
& SCM_CELL_WORD_0 (exp
)
2735 scm_port_print (SCM exp
, SCM port
, scm_print_state
*pstate SCM_UNUSED
)
2737 char *type
= SCM_PTOBNAME (SCM_PTOBNUM (exp
));
2740 scm_puts_unlocked ("#<", port
);
2741 scm_print_port_mode (exp
, port
);
2742 scm_puts_unlocked (type
, port
);
2743 scm_putc_unlocked (' ', port
);
2744 scm_uintprint (SCM_CELL_WORD_1 (exp
), 16, port
);
2745 scm_putc_unlocked ('>', port
);
2752 /* Iterating over all ports. */
2754 struct for_each_data
2756 void (*proc
) (void *data
, SCM p
);
2761 for_each_trampoline (void *data
, SCM port
, SCM result
)
2763 struct for_each_data
*d
= data
;
2765 d
->proc (d
->data
, port
);
2771 scm_c_port_for_each (void (*proc
)(void *data
, SCM p
), void *data
)
2773 struct for_each_data d
;
2778 scm_c_weak_set_fold (for_each_trampoline
, &d
, SCM_EOL
,
2779 scm_i_port_weak_set
);
2783 scm_for_each_trampoline (void *data
, SCM port
)
2785 scm_call_1 (SCM_PACK_POINTER (data
), port
);
2788 SCM_DEFINE (scm_port_for_each
, "port-for-each", 1, 0, 0,
2790 "Apply @var{proc} to each port in the Guile port table\n"
2791 "in turn. The return value is unspecified. More specifically,\n"
2792 "@var{proc} is applied exactly once to every port that exists\n"
2793 "in the system at the time @code{port-for-each} is invoked.\n"
2794 "Changes to the port table while @code{port-for-each} is running\n"
2795 "have no effect as far as @code{port-for-each} is concerned.")
2796 #define FUNC_NAME s_scm_port_for_each
2798 SCM_VALIDATE_PROC (1, proc
);
2800 scm_c_port_for_each (scm_for_each_trampoline
, SCM_UNPACK_POINTER (proc
));
2802 return SCM_UNSPECIFIED
;
2807 flush_output_port (void *closure
, SCM port
)
2809 if (SCM_OPOUTPORTP (port
))
2810 scm_flush_unlocked (port
);
2813 SCM_DEFINE (scm_flush_all_ports
, "flush-all-ports", 0, 0, 0,
2815 "Equivalent to calling @code{force-output} on\n"
2816 "all open output ports. The return value is unspecified.")
2817 #define FUNC_NAME s_scm_flush_all_ports
2819 scm_c_port_for_each (&flush_output_port
, NULL
);
2820 return SCM_UNSPECIFIED
;
2829 scm_t_bits scm_tc16_void_port
= 0;
2831 static int fill_input_void_port (SCM port SCM_UNUSED
)
2837 write_void_port (SCM port SCM_UNUSED
,
2838 const void *data SCM_UNUSED
,
2839 size_t size SCM_UNUSED
)
2844 scm_i_void_port (long mode_bits
)
2848 ret
= scm_c_make_port (scm_tc16_void_port
, mode_bits
, 0);
2850 scm_port_non_buffer (SCM_PTAB_ENTRY (ret
));
2856 scm_void_port (char *mode_str
)
2858 return scm_i_void_port (scm_mode_bits (mode_str
));
2861 SCM_DEFINE (scm_sys_make_void_port
, "%make-void-port", 1, 0, 0,
2863 "Create and return a new void port. A void port acts like\n"
2864 "@file{/dev/null}. The @var{mode} argument\n"
2865 "specifies the input/output modes for this port: see the\n"
2866 "documentation for @code{open-file} in @ref{File Ports}.")
2867 #define FUNC_NAME s_scm_sys_make_void_port
2869 return scm_i_void_port (scm_i_mode_bits (mode
));
2876 /* Initialization. */
2881 /* lseek() symbols. */
2882 scm_c_define ("SEEK_SET", scm_from_int (SEEK_SET
));
2883 scm_c_define ("SEEK_CUR", scm_from_int (SEEK_CUR
));
2884 scm_c_define ("SEEK_END", scm_from_int (SEEK_END
));
2886 scm_tc16_void_port
= scm_make_port_type ("void", fill_input_void_port
,
2889 cur_inport_fluid
= scm_make_fluid ();
2890 cur_outport_fluid
= scm_make_fluid ();
2891 cur_errport_fluid
= scm_make_fluid ();
2892 cur_loadport_fluid
= scm_make_fluid ();
2894 scm_i_port_weak_set
= scm_c_make_weak_set (31);
2896 #include "libguile/ports.x"
2898 /* Use Latin-1 as the default port encoding. */
2899 SCM_VARIABLE_SET (default_port_encoding_var
,
2900 scm_make_fluid_with_default (SCM_BOOL_F
));
2901 scm_port_encoding_init
= 1;
2903 SCM_VARIABLE_SET (default_conversion_strategy_var
,
2904 scm_make_fluid_with_default (sym_substitute
));
2905 scm_conversion_strategy_init
= 1;
2907 /* These bindings are used when boot-9 turns `current-input-port' et
2908 al into parameters. They are then removed from the guile module. */
2909 scm_c_define ("%current-input-port-fluid", cur_inport_fluid
);
2910 scm_c_define ("%current-output-port-fluid", cur_outport_fluid
);
2911 scm_c_define ("%current-error-port-fluid", cur_errport_fluid
);