1 /* Copyright (C) 1995-1999, 2000, 2001, 2002, 2003, 2004, 2006, 2008,
2 * 2009, 2010, 2011 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
34 #include "libguile/_scm.h"
35 #include "libguile/chars.h"
36 #include "libguile/continuations.h"
37 #include "libguile/smob.h"
38 #include "libguile/control.h"
39 #include "libguile/eval.h"
40 #include "libguile/macros.h"
41 #include "libguile/procprop.h"
42 #include "libguile/read.h"
43 #include "libguile/weaks.h"
44 #include "libguile/programs.h"
45 #include "libguile/alist.h"
46 #include "libguile/struct.h"
47 #include "libguile/ports.h"
48 #include "libguile/root.h"
49 #include "libguile/strings.h"
50 #include "libguile/strports.h"
51 #include "libguile/vectors.h"
52 #include "libguile/numbers.h"
53 #include "libguile/vm.h"
55 #include "libguile/validate.h"
56 #include "libguile/print.h"
58 #include "libguile/private-options.h"
62 /* Character printers. */
64 static size_t display_string (const void *, int, size_t, SCM
,
65 scm_t_string_failed_conversion_handler
);
67 static int display_character (scm_t_wchar
, SCM
,
68 scm_t_string_failed_conversion_handler
);
70 static void write_character (scm_t_wchar
, SCM
, int);
72 static void write_character_escaped (scm_t_wchar
, int, SCM
);
76 /* {Names of immediate symbols}
78 * This table must agree with the declarations in scm.h: {Immediate Symbols}.
81 /* This table must agree with the list of flags in tags.h. */
82 static const char *iflagnames
[] =
85 "#nil", /* Elisp nil value. Should print from elisp as symbol `nil'. */
86 "#<XXX UNUSED LISP FALSE -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
89 "#<XXX UNUSED BOOLEAN 0 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
90 "#<XXX UNUSED BOOLEAN 1 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
91 "#<XXX UNUSED BOOLEAN 2 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
96 /* Unbound slot marker for GOOPS. For internal use in GOOPS only. */
100 SCM_SYMBOL (sym_reader
, "reader");
102 scm_t_option scm_print_opts
[] = {
103 { SCM_OPTION_SCM
, "highlight-prefix", (scm_t_bits
)SCM_BOOL_F
,
104 "The string to print before highlighted values." },
105 { SCM_OPTION_SCM
, "highlight-suffix", (scm_t_bits
)SCM_BOOL_F
,
106 "The string to print after highlighted values." },
107 { SCM_OPTION_SCM
, "quote-keywordish-symbols", (scm_t_bits
)SCM_BOOL_F
,
108 "How to print symbols that have a colon as their first or last character. "
109 "The value '#f' does not quote the colons; '#t' quotes them; "
110 "'reader' quotes them when the reader option 'keywords' is not '#f'."
115 SCM_DEFINE (scm_print_options
, "print-options-interface", 0, 1, 0,
117 "Option interface for the print options. Instead of using\n"
118 "this procedure directly, use the procedures\n"
119 "@code{print-enable}, @code{print-disable}, @code{print-set!}\n"
120 "and @code{print-options}.")
121 #define FUNC_NAME s_scm_print_options
123 SCM ans
= scm_options (setting
,
131 /* {Printing of Scheme Objects}
134 /* Detection of circular references.
136 * Due to other constraints in the implementation, this code has bad
137 * time complexity (O (depth * N)), The printer code can be
138 * rewritten to be O(N).
140 #define PUSH_REF(pstate, obj) \
143 PSTATE_STACK_SET (pstate, pstate->top, obj); \
145 if (pstate->top == pstate->ceiling) \
146 grow_ref_stack (pstate); \
149 #define ENTER_NESTED_DATA(pstate, obj, label) \
152 register unsigned long i; \
153 for (i = 0; i < pstate->top; ++i) \
154 if (scm_is_eq (PSTATE_STACK_REF (pstate, i), (obj))) \
156 if (pstate->fancyp) \
158 if (pstate->top - pstate->list_offset >= pstate->level) \
160 scm_putc ('#', port); \
164 PUSH_REF(pstate, obj); \
167 #define EXIT_NESTED_DATA(pstate) \
171 PSTATE_STACK_SET (pstate, pstate->top, SCM_UNDEFINED); \
175 SCM scm_print_state_vtable
= SCM_BOOL_F
;
176 static SCM print_state_pool
= SCM_EOL
;
177 scm_i_pthread_mutex_t print_state_mutex
= SCM_I_PTHREAD_MUTEX_INITIALIZER
;
179 #ifdef GUILE_DEBUG /* Used for debugging purposes */
181 SCM_DEFINE (scm_current_pstate
, "current-pstate", 0, 0, 0,
183 "Return the current-pstate -- the car of the\n"
184 "@code{print_state_pool}. @code{current-pstate} is only\n"
185 "included in @code{--enable-guile-debug} builds.")
186 #define FUNC_NAME s_scm_current_pstate
188 if (!scm_is_null (print_state_pool
))
189 return SCM_CAR (print_state_pool
);
197 #define PSTATE_SIZE 50L
200 make_print_state (void)
203 = scm_make_struct (scm_print_state_vtable
, SCM_INUM0
, SCM_EOL
);
204 scm_print_state
*pstate
= SCM_PRINT_STATE (print_state
);
205 pstate
->ref_vect
= scm_c_make_vector (PSTATE_SIZE
, SCM_UNDEFINED
);
206 pstate
->ceiling
= SCM_SIMPLE_VECTOR_LENGTH (pstate
->ref_vect
);
207 pstate
->highlight_objects
= SCM_EOL
;
212 scm_make_print_state ()
214 SCM answer
= SCM_BOOL_F
;
216 /* First try to allocate a print state from the pool */
217 scm_i_pthread_mutex_lock (&print_state_mutex
);
218 if (!scm_is_null (print_state_pool
))
220 answer
= SCM_CAR (print_state_pool
);
221 print_state_pool
= SCM_CDR (print_state_pool
);
223 scm_i_pthread_mutex_unlock (&print_state_mutex
);
225 return scm_is_false (answer
) ? make_print_state () : answer
;
229 scm_free_print_state (SCM print_state
)
232 scm_print_state
*pstate
= SCM_PRINT_STATE (print_state
);
233 /* Cleanup before returning print state to pool.
234 * It is better to do it here. Doing it in scm_prin1
235 * would cost more since that function is called much more
239 pstate
->revealed
= 0;
240 pstate
->highlight_objects
= SCM_EOL
;
241 scm_i_pthread_mutex_lock (&print_state_mutex
);
242 handle
= scm_cons (print_state
, print_state_pool
);
243 print_state_pool
= handle
;
244 scm_i_pthread_mutex_unlock (&print_state_mutex
);
248 scm_i_port_with_print_state (SCM port
, SCM print_state
)
250 if (SCM_UNBNDP (print_state
))
252 if (SCM_PORT_WITH_PS_P (port
))
255 print_state
= scm_make_print_state ();
256 /* port does not need to be coerced since it doesn't have ps */
259 port
= SCM_COERCE_OUTPORT (port
);
260 SCM_RETURN_NEWSMOB (scm_tc16_port_with_ps
,
261 SCM_UNPACK (scm_cons (port
, print_state
)));
265 grow_ref_stack (scm_print_state
*pstate
)
267 SCM old_vect
= pstate
->ref_vect
;
268 size_t old_size
= SCM_SIMPLE_VECTOR_LENGTH (old_vect
);
269 size_t new_size
= 2 * pstate
->ceiling
;
270 SCM new_vect
= scm_c_make_vector (new_size
, SCM_UNDEFINED
);
273 for (i
= 0; i
!= old_size
; ++i
)
274 SCM_SIMPLE_VECTOR_SET (new_vect
, i
, SCM_SIMPLE_VECTOR_REF (old_vect
, i
));
276 pstate
->ref_vect
= new_vect
;
277 pstate
->ceiling
= new_size
;
280 #define PSTATE_STACK_REF(p,i) SCM_SIMPLE_VECTOR_REF((p)->ref_vect, (i))
281 #define PSTATE_STACK_SET(p,i,v) SCM_SIMPLE_VECTOR_SET((p)->ref_vect, (i), (v))
284 print_circref (SCM port
, scm_print_state
*pstate
, SCM ref
)
287 long self
= pstate
->top
- 1;
289 if (scm_is_pair (PSTATE_STACK_REF (pstate
, i
)))
293 if (!scm_is_pair (PSTATE_STACK_REF (pstate
, i
-1))
294 || !scm_is_eq (SCM_CDR (PSTATE_STACK_REF (pstate
, i
-1)),
295 SCM_CDR (PSTATE_STACK_REF (pstate
, i
))))
301 for (i
= pstate
->top
- 1; 1; --i
)
302 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), ref
))
304 scm_putc ('#', port
);
305 scm_intprint (i
- self
, 10, port
);
306 scm_putc ('#', port
);
309 /* Print the name of a symbol. */
312 quote_keywordish_symbol (SCM symbol
)
316 if (scm_i_symbol_ref (symbol
, 0) != ':'
317 && scm_i_symbol_ref (symbol
, scm_i_symbol_length (symbol
) - 1) != ':')
320 option
= SCM_PRINT_KEYWORD_STYLE
;
321 if (scm_is_false (option
))
323 if (scm_is_eq (option
, sym_reader
))
324 return scm_is_true (SCM_PACK (SCM_KEYWORD_STYLE
));
329 scm_i_print_symbol_name (SCM str
, SCM port
)
331 /* This points to the first character that has not yet been written to the
334 /* This points to the character we're currently looking at. */
336 /* If the name contains weird characters, we'll escape them with
337 * backslashes and set this flag; it indicates that we should surround the
338 * name with "#{" and "}#". */
340 /* Backslashes are not sufficient to make a name weird, but if a name is
341 * weird because of other characters, backslahes need to be escaped too.
342 * The first time we see a backslash, we set maybe_weird, and mw_pos points
343 * to the backslash. Then if the name turns out to be weird, we re-process
344 * everything starting from mw_pos.
345 * We could instead make backslashes always weird. This is not necessary
346 * to ensure that the output is (read)-able, but it would make this code
347 * simpler and faster. */
350 size_t len
= scm_i_symbol_length (str
);
351 scm_t_wchar str0
= scm_i_symbol_ref (str
, 0);
353 if (len
== 0 || str0
== '\'' || str0
== '`' || str0
== ','
354 || quote_keywordish_symbol (str
)
355 || (str0
== '.' && len
== 1)
356 || scm_is_true (scm_i_string_to_number (scm_symbol_to_string (str
), 10)))
358 scm_lfwrite ("#{", 2, port
);
362 for (end
= pos
; end
< len
; ++end
)
363 switch (scm_i_symbol_ref (str
, end
))
365 #ifdef BRACKETS_AS_PARENS
374 case SCM_WHITE_SPACES
:
375 case SCM_LINE_INCREMENTORS
:
384 scm_lfwrite ("#{", 2, port
);
388 scm_lfwrite_substr (scm_symbol_to_string (str
), pos
, end
, port
);
392 buf
[1] = (char) (unsigned char) scm_i_symbol_ref (str
, end
);
393 scm_lfwrite (buf
, 2, port
);
410 scm_lfwrite_substr (scm_symbol_to_string (str
), pos
, end
, port
);
412 scm_lfwrite ("}#", 2, port
);
416 scm_print_symbol_name (const char *str
, size_t len
, SCM port
)
418 SCM symbol
= scm_from_locale_symboln (str
, len
);
419 scm_i_print_symbol_name (symbol
, port
);
422 /* Print generally. Handles both write and display according to PSTATE.
424 SCM_GPROC(s_write
, "write", 1, 1, 0, scm_write
, g_write
);
425 SCM_GPROC(s_display
, "display", 1, 1, 0, scm_display
, g_display
);
427 static void iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
);
430 /* Print a character as an octal or hex escape. */
431 #define PRINT_CHAR_ESCAPE(i, port) \
434 if (!SCM_R6RS_ESCAPES_P) \
435 scm_intprint (i, 8, port); \
438 scm_puts ("x", port); \
439 scm_intprint (i, 16, port); \
446 scm_iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
449 && scm_is_true (scm_memq (exp
, pstate
->highlight_objects
)))
451 scm_display (SCM_PRINT_HIGHLIGHT_PREFIX
, port
);
452 iprin1 (exp
, port
, pstate
);
453 scm_display (SCM_PRINT_HIGHLIGHT_SUFFIX
, port
);
456 iprin1 (exp
, port
, pstate
);
460 iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
462 switch (SCM_ITAG3 (exp
))
466 /* These tc3 tags should never occur in an immediate value. They are
467 * only used in cell types of non-immediates, i. e. the value returned
468 * by SCM_CELL_TYPE (exp) can use these tags.
470 scm_ipruk ("immediate", exp
, port
);
474 scm_intprint (SCM_I_INUM (exp
), 10, port
);
479 if (SCM_WRITINGP (pstate
))
480 write_character (SCM_CHAR (exp
), port
, 0);
483 if (!display_character (SCM_CHAR (exp
), port
,
484 scm_i_get_conversion_strategy (port
)))
485 scm_encoding_error (__func__
, errno
,
486 "cannot convert to output locale",
487 "UTF-32", scm_i_get_port_encoding (port
),
488 scm_string (scm_list_1 (exp
)));
491 else if (SCM_IFLAGP (exp
)
492 && ((size_t) SCM_IFLAGNUM (exp
) < (sizeof iflagnames
/ sizeof (char *))))
494 scm_puts (iflagnames
[SCM_IFLAGNUM (exp
)], port
);
498 /* unknown immediate value */
499 scm_ipruk ("immediate", exp
, port
);
503 switch (SCM_TYP7 (exp
))
507 ENTER_NESTED_DATA (pstate
, exp
, circref
);
508 if (SCM_OBJ_CLASS_FLAGS (exp
) & SCM_CLASSF_GOOPS
)
510 SCM pwps
, print
= pstate
->writingp
? g_write
: g_display
;
513 pwps
= scm_i_port_with_print_state (port
, pstate
->handle
);
514 pstate
->revealed
= 1;
515 scm_call_generic_2 (print
, exp
, pwps
);
520 scm_print_struct (exp
, port
, pstate
);
522 EXIT_NESTED_DATA (pstate
);
525 case scm_tcs_cons_imcar
:
526 case scm_tcs_cons_nimcar
:
527 ENTER_NESTED_DATA (pstate
, exp
, circref
);
528 scm_iprlist ("(", exp
, ')', port
, pstate
);
529 EXIT_NESTED_DATA (pstate
);
532 print_circref (port
, pstate
, exp
);
535 switch SCM_TYP16 (exp
) {
537 scm_bigprint (exp
, port
, pstate
);
540 scm_print_real (exp
, port
, pstate
);
542 case scm_tc16_complex
:
543 scm_print_complex (exp
, port
, pstate
);
545 case scm_tc16_fraction
:
546 scm_i_print_fraction (exp
, port
, pstate
);
551 if (SCM_WRITINGP (pstate
))
555 display_character ('"', port
, iconveh_question_mark
);
556 len
= scm_i_string_length (exp
);
557 for (i
= 0; i
< len
; ++i
)
558 write_character (scm_i_string_ref (exp
, i
), port
, 1);
560 display_character ('"', port
, iconveh_question_mark
);
561 scm_remember_upto_here_1 (exp
);
567 len
= scm_i_string_length (exp
);
568 printed
= display_string (scm_i_string_data (exp
),
569 scm_i_is_narrow_string (exp
),
571 scm_i_get_conversion_strategy (port
));
572 if (SCM_UNLIKELY (printed
< len
))
573 /* FIXME: Provide the error location. */
574 scm_encoding_error (__func__
, errno
,
575 "cannot convert to output locale",
576 "UTF-32", scm_i_get_port_encoding (port
),
580 scm_remember_upto_here_1 (exp
);
583 if (scm_i_symbol_is_interned (exp
))
585 scm_i_print_symbol_name (exp
, port
);
586 scm_remember_upto_here_1 (exp
);
590 scm_puts ("#<uninterned-symbol ", port
);
591 scm_i_print_symbol_name (exp
, port
);
592 scm_putc (' ', port
);
593 scm_uintprint (SCM_UNPACK (exp
), 16, port
);
594 scm_putc ('>', port
);
597 case scm_tc7_variable
:
598 scm_i_variable_print (exp
, port
, pstate
);
600 case scm_tc7_program
:
601 scm_i_program_print (exp
, port
, pstate
);
603 case scm_tc7_pointer
:
604 scm_i_pointer_print (exp
, port
, pstate
);
606 case scm_tc7_hashtable
:
607 scm_i_hashtable_print (exp
, port
, pstate
);
610 scm_i_fluid_print (exp
, port
, pstate
);
612 case scm_tc7_dynamic_state
:
613 scm_i_dynamic_state_print (exp
, port
, pstate
);
616 scm_i_frame_print (exp
, port
, pstate
);
618 case scm_tc7_objcode
:
619 scm_i_objcode_print (exp
, port
, pstate
);
622 scm_i_vm_print (exp
, port
, pstate
);
624 case scm_tc7_vm_cont
:
625 scm_i_vm_cont_print (exp
, port
, pstate
);
628 scm_i_prompt_print (exp
, port
, pstate
);
630 case scm_tc7_with_fluids
:
631 scm_i_with_fluids_print (exp
, port
, pstate
);
634 ENTER_NESTED_DATA (pstate
, exp
, circref
);
635 if (SCM_IS_WHVEC (exp
))
636 scm_puts ("#wh(", port
);
638 scm_puts ("#w(", port
);
639 goto common_vector_printer
;
641 case scm_tc7_bytevector
:
642 scm_i_print_bytevector (exp
, port
, pstate
);
645 ENTER_NESTED_DATA (pstate
, exp
, circref
);
646 scm_puts ("#(", port
);
647 common_vector_printer
:
650 long last
= SCM_SIMPLE_VECTOR_LENGTH (exp
) - 1;
653 && SCM_SIMPLE_VECTOR_LENGTH (exp
) > pstate
->length
)
655 last
= pstate
->length
- 1;
658 if (SCM_I_WVECTP (exp
))
660 /* Elements of weak vectors may not be accessed via the
661 `SIMPLE_VECTOR_REF ()' macro. */
662 for (i
= 0; i
< last
; ++i
)
664 scm_iprin1 (scm_c_vector_ref (exp
, i
),
666 scm_putc (' ', port
);
671 for (i
= 0; i
< last
; ++i
)
673 scm_iprin1 (SCM_SIMPLE_VECTOR_REF (exp
, i
), port
, pstate
);
674 scm_putc (' ', port
);
681 scm_iprin1 (scm_c_vector_ref (exp
, i
), port
, pstate
);
684 scm_puts (" ...", port
);
685 scm_putc (')', port
);
687 EXIT_NESTED_DATA (pstate
);
691 register long i
= SCM_PTOBNUM (exp
);
693 && scm_ptobs
[i
].print
694 && (scm_ptobs
[i
].print
) (exp
, port
, pstate
))
699 ENTER_NESTED_DATA (pstate
, exp
, circref
);
700 SCM_SMOB_DESCRIPTOR (exp
).print (exp
, port
, pstate
);
701 EXIT_NESTED_DATA (pstate
);
704 /* case scm_tcs_closures: */
706 scm_ipruk ("type", exp
, port
);
711 /* Print states are necessary for circular reference safe printing.
712 * They are also expensive to allocate. Therefore print states are
713 * kept in a pool so that they can be reused.
716 /* The PORT argument can also be a print-state/port pair, which will
717 * then be used instead of allocating a new print state. This is
718 * useful for continuing a chain of print calls from Scheme. */
721 scm_prin1 (SCM exp
, SCM port
, int writingp
)
723 SCM handle
= SCM_BOOL_F
; /* Will GC protect the handle whilst unlinked */
725 scm_print_state
*pstate
;
728 /* If PORT is a print-state/port pair, use that. Else create a new
731 if (SCM_PORT_WITH_PS_P (port
))
733 pstate_scm
= SCM_PORT_WITH_PS_PS (port
);
734 port
= SCM_PORT_WITH_PS_PORT (port
);
738 /* First try to allocate a print state from the pool */
739 scm_i_pthread_mutex_lock (&print_state_mutex
);
740 if (!scm_is_null (print_state_pool
))
742 handle
= print_state_pool
;
743 print_state_pool
= SCM_CDR (print_state_pool
);
745 scm_i_pthread_mutex_unlock (&print_state_mutex
);
746 if (scm_is_false (handle
))
747 handle
= scm_list_1 (make_print_state ());
748 pstate_scm
= SCM_CAR (handle
);
751 pstate
= SCM_PRINT_STATE (pstate_scm
);
752 old_writingp
= pstate
->writingp
;
753 pstate
->writingp
= writingp
;
754 scm_iprin1 (exp
, port
, pstate
);
755 pstate
->writingp
= old_writingp
;
757 /* Return print state to pool if it has been created above and
758 hasn't escaped to Scheme. */
760 if (scm_is_true (handle
) && !pstate
->revealed
)
762 scm_i_pthread_mutex_lock (&print_state_mutex
);
763 SCM_SETCDR (handle
, print_state_pool
);
764 print_state_pool
= handle
;
765 scm_i_pthread_mutex_unlock (&print_state_mutex
);
769 /* Convert codepoint CH to UTF-8 and store the result in UTF8. Return
770 the number of bytes of the UTF-8-encoded string. */
772 codepoint_to_utf8 (scm_t_wchar ch
, scm_t_uint8 utf8
[4])
775 scm_t_uint32 codepoint
;
777 codepoint
= (scm_t_uint32
) ch
;
779 if (codepoint
<= 0x7f)
782 utf8
[0] = (scm_t_uint8
) codepoint
;
784 else if (codepoint
<= 0x7ffUL
)
787 utf8
[0] = 0xc0 | (codepoint
>> 6);
788 utf8
[1] = 0x80 | (codepoint
& 0x3f);
790 else if (codepoint
<= 0xffffUL
)
793 utf8
[0] = 0xe0 | (codepoint
>> 12);
794 utf8
[1] = 0x80 | ((codepoint
>> 6) & 0x3f);
795 utf8
[2] = 0x80 | (codepoint
& 0x3f);
800 utf8
[0] = 0xf0 | (codepoint
>> 18);
801 utf8
[1] = 0x80 | ((codepoint
>> 12) & 0x3f);
802 utf8
[2] = 0x80 | ((codepoint
>> 6) & 0x3f);
803 utf8
[3] = 0x80 | (codepoint
& 0x3f);
809 /* Display the LEN codepoints in STR to PORT according to STRATEGY;
810 return the number of codepoints successfully displayed. If NARROW_P,
811 then STR is interpreted as a sequence of `char', denoting a Latin-1
812 string; otherwise it's interpreted as a sequence of
815 display_string (const void *str
, int narrow_p
,
816 size_t len
, SCM port
,
817 scm_t_string_failed_conversion_handler strategy
)
820 #define STR_REF(s, x) \
822 ? (scm_t_wchar) ((unsigned char *) (s))[x] \
823 : ((scm_t_wchar *) (s))[x])
828 pt
= SCM_PTAB_ENTRY (port
);
830 if (SCM_UNLIKELY (pt
->output_cd
== (iconv_t
) -1))
831 /* Initialize the conversion descriptors. */
832 scm_i_set_port_encoding_x (port
, pt
->encoding
);
836 while (len
> printed
)
838 size_t done
, utf8_len
, input_left
, output_left
, i
;
839 size_t codepoints_read
, output_len
;
840 char *input
, *output
;
841 char utf8_buf
[256], encoded_output
[256];
844 /* Convert STR to UTF-8. */
845 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
846 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
849 offsets
[utf8_len
] = i
;
850 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
851 (scm_t_uint8
*) input
);
852 input
= utf8_buf
+ utf8_len
;
856 input_left
= utf8_len
;
858 output
= encoded_output
;
859 output_left
= sizeof (encoded_output
);
861 done
= iconv (pt
->output_cd
, &input
, &input_left
,
862 &output
, &output_left
);
864 output_len
= sizeof (encoded_output
) - output_left
;
866 if (SCM_UNLIKELY (done
== (size_t) -1))
868 /* Reset the `iconv' state. */
869 iconv (pt
->output_cd
, NULL
, NULL
, NULL
, NULL
);
871 if (errno
== EILSEQ
&&
872 strategy
!= SCM_FAILED_CONVERSION_ERROR
)
874 /* Conversion failed somewhere in INPUT and we want to
875 escape or substitute the offending input character. */
877 /* Print the OUTPUT_LEN bytes successfully converted. */
878 scm_lfwrite (encoded_output
, output_len
, port
);
880 /* See how many input codepoints these OUTPUT_LEN bytes
882 codepoints_read
= offsets
[input
- utf8_buf
] - printed
;
883 printed
+= codepoints_read
;
885 if (strategy
== SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
)
889 /* Find CH, the offending codepoint, and escape it. */
890 ch
= STR_REF (str
, offsets
[input
- utf8_buf
]);
891 write_character_escaped (ch
, 1, port
);
894 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
895 display_string ("?", 1, 1, port
, strategy
);
900 /* Something bad happened that we can't handle: bail out. */
905 /* INPUT was successfully converted, entirely; print the
907 scm_lfwrite (encoded_output
, output_len
, port
);
908 codepoints_read
= i
- printed
;
909 printed
+= codepoints_read
;
917 /* Attempt to display CH to PORT according to STRATEGY. Return non-zero
918 if CH was successfully displayed, zero otherwise (e.g., if it was not
919 representable in PORT's encoding.) */
921 display_character (scm_t_wchar ch
, SCM port
,
922 scm_t_string_failed_conversion_handler strategy
)
924 return display_string (&ch
, 0, 1, port
, strategy
) == 1;
927 /* Attempt to pretty-print CH, a combining character, to PORT. Return
928 zero upon failure, non-zero otherwise. The idea is to print CH above
929 a dotted circle to make it more visible. */
931 write_combining_character (scm_t_wchar ch
, SCM port
)
935 str
[0] = SCM_CODEPOINT_DOTTED_CIRCLE
;
938 return display_string (str
, 0, 2, port
, iconveh_error
) == 2;
941 /* Write CH to PORT in its escaped form, using the string escape syntax
942 if STRING_ESCAPES_P is non-zero. */
944 write_character_escaped (scm_t_wchar ch
, int string_escapes_p
, SCM port
)
946 if (string_escapes_p
)
948 /* Represent CH using the in-string escape syntax. */
950 static const char hex
[] = "0123456789abcdef";
951 static const char escapes
[7] = "abtnvfr";
954 if (ch
>= 0x07 && ch
<= 0x0D && ch
!= 0x0A)
956 /* Use special escapes for some C0 controls. */
958 buf
[1] = escapes
[ch
- 0x07];
959 scm_lfwrite (buf
, 2, port
);
961 else if (!SCM_R6RS_ESCAPES_P
)
967 buf
[2] = hex
[ch
/ 16];
968 buf
[3] = hex
[ch
% 16];
969 scm_lfwrite (buf
, 4, port
);
971 else if (ch
<= 0xFFFF)
975 buf
[2] = hex
[(ch
& 0xF000) >> 12];
976 buf
[3] = hex
[(ch
& 0xF00) >> 8];
977 buf
[4] = hex
[(ch
& 0xF0) >> 4];
978 buf
[5] = hex
[(ch
& 0xF)];
979 scm_lfwrite (buf
, 6, port
);
981 else if (ch
> 0xFFFF)
985 buf
[2] = hex
[(ch
& 0xF00000) >> 20];
986 buf
[3] = hex
[(ch
& 0xF0000) >> 16];
987 buf
[4] = hex
[(ch
& 0xF000) >> 12];
988 buf
[5] = hex
[(ch
& 0xF00) >> 8];
989 buf
[6] = hex
[(ch
& 0xF0) >> 4];
990 buf
[7] = hex
[(ch
& 0xF)];
991 scm_lfwrite (buf
, 8, port
);
996 /* Print an R6RS variable-length hex escape: "\xNNNN;". */
997 scm_t_wchar ch2
= ch
;
1007 buf
[i
] = hex
[ch2
& 0xF];
1014 scm_lfwrite (buf
+ i
, 9 - i
, port
);
1019 /* Represent CH using the character escape syntax. */
1022 name
= scm_i_charname (SCM_MAKE_CHAR (ch
));
1024 scm_puts (name
, port
);
1026 PRINT_CHAR_ESCAPE (ch
, port
);
1030 /* Write CH to PORT, escaping it if it's non-graphic or not
1031 representable in PORT's encoding. If STRING_ESCAPES_P is true and CH
1032 needs to be escaped, it is escaped using the in-string escape syntax;
1033 otherwise the character escape syntax is used. */
1035 write_character (scm_t_wchar ch
, SCM port
, int string_escapes_p
)
1038 scm_t_string_failed_conversion_handler strategy
;
1040 strategy
= scm_i_get_conversion_strategy (port
);
1042 if (string_escapes_p
)
1044 /* Check if CH deserves special treatment. */
1045 if (ch
== '"' || ch
== '\\')
1047 display_character ('\\', port
, iconveh_question_mark
);
1048 display_character (ch
, port
, strategy
);
1051 else if (ch
== ' ' || ch
== '\n')
1053 display_character (ch
, port
, strategy
);
1059 display_string ("#\\", 1, 2, port
, iconveh_question_mark
);
1061 if (uc_combining_class (ch
) != UC_CCC_NR
)
1062 /* Character is a combining character, so attempt to
1064 printed
= write_combining_character (ch
, port
);
1068 && uc_is_general_category_withtable (ch
,
1069 UC_CATEGORY_MASK_L
|
1070 UC_CATEGORY_MASK_M
|
1071 UC_CATEGORY_MASK_N
|
1072 UC_CATEGORY_MASK_P
|
1073 UC_CATEGORY_MASK_S
))
1074 /* CH is graphic; attempt to display it. */
1075 printed
= display_character (ch
, port
, iconveh_error
);
1078 /* CH isn't graphic or cannot be represented in PORT's encoding. */
1079 write_character_escaped (ch
, string_escapes_p
, port
);
1082 /* Print an integer.
1086 scm_intprint (scm_t_intmax n
, int radix
, SCM port
)
1088 char num_buf
[SCM_INTBUFLEN
];
1089 scm_lfwrite (num_buf
, scm_iint2str (n
, radix
, num_buf
), port
);
1093 scm_uintprint (scm_t_uintmax n
, int radix
, SCM port
)
1095 char num_buf
[SCM_INTBUFLEN
];
1096 scm_lfwrite (num_buf
, scm_iuint2str (n
, radix
, num_buf
), port
);
1099 /* Print an object of unrecognized type.
1103 scm_ipruk (char *hdr
, SCM ptr
, SCM port
)
1105 scm_puts ("#<unknown-", port
);
1106 scm_puts (hdr
, port
);
1107 if (1) /* (scm_in_heap_p (ptr)) */ /* FIXME */
1109 scm_puts (" (0x", port
);
1110 scm_uintprint (SCM_CELL_WORD_0 (ptr
), 16, port
);
1111 scm_puts (" . 0x", port
);
1112 scm_uintprint (SCM_CELL_WORD_1 (ptr
), 16, port
);
1113 scm_puts (") @", port
);
1115 scm_puts (" 0x", port
);
1116 scm_uintprint (SCM_UNPACK (ptr
), 16, port
);
1117 scm_putc ('>', port
);
1124 scm_iprlist (char *hdr
, SCM exp
, int tlr
, SCM port
, scm_print_state
*pstate
)
1126 register SCM hare
, tortoise
;
1127 long floor
= pstate
->top
- 2;
1128 scm_puts (hdr
, port
);
1131 goto fancy_printing
;
1133 /* Run a hare and tortoise so that total time complexity will be
1134 O(depth * N) instead of O(N^2). */
1135 hare
= SCM_CDR (exp
);
1137 while (scm_is_pair (hare
))
1139 if (scm_is_eq (hare
, tortoise
))
1140 goto fancy_printing
;
1141 hare
= SCM_CDR (hare
);
1142 if (!scm_is_pair (hare
))
1144 hare
= SCM_CDR (hare
);
1145 tortoise
= SCM_CDR (tortoise
);
1148 /* No cdr cycles intrinsic to this list */
1149 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1150 for (exp
= SCM_CDR (exp
); scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1154 for (i
= floor
; i
>= 0; --i
)
1155 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1157 PUSH_REF (pstate
, exp
);
1158 scm_putc (' ', port
);
1160 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1162 if (!SCM_NULL_OR_NIL_P (exp
))
1164 scm_puts (" . ", port
);
1165 scm_iprin1 (exp
, port
, pstate
);
1169 scm_putc (tlr
, port
);
1170 pstate
->top
= floor
+ 2;
1175 long n
= pstate
->length
;
1177 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1178 exp
= SCM_CDR (exp
); --n
;
1179 for (; scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1181 register unsigned long i
;
1183 for (i
= 0; i
< pstate
->top
; ++i
)
1184 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1190 scm_puts (" ...", port
);
1196 PUSH_REF(pstate
, exp
);
1197 ++pstate
->list_offset
;
1198 scm_putc (' ', port
);
1200 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1203 if (!SCM_NULL_OR_NIL_P (exp
))
1205 scm_puts (" . ", port
);
1206 scm_iprin1 (exp
, port
, pstate
);
1209 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1213 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1216 scm_puts (" . ", port
);
1217 print_circref (port
, pstate
, exp
);
1224 scm_valid_oport_value_p (SCM val
)
1226 return (SCM_OPOUTPORTP (val
)
1227 || (SCM_PORT_WITH_PS_P (val
)
1228 && SCM_OPOUTPORTP (SCM_PORT_WITH_PS_PORT (val
))));
1231 /* SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write); */
1234 scm_write (SCM obj
, SCM port
)
1236 if (SCM_UNBNDP (port
))
1237 port
= scm_current_output_port ();
1239 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_write
);
1241 scm_prin1 (obj
, port
, 1);
1242 return SCM_UNSPECIFIED
;
1246 /* SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display); */
1249 scm_display (SCM obj
, SCM port
)
1251 if (SCM_UNBNDP (port
))
1252 port
= scm_current_output_port ();
1254 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_display
);
1256 scm_prin1 (obj
, port
, 0);
1257 return SCM_UNSPECIFIED
;
1261 SCM_DEFINE (scm_simple_format
, "simple-format", 2, 0, 1,
1262 (SCM destination
, SCM message
, SCM args
),
1263 "Write @var{message} to @var{destination}, defaulting to\n"
1264 "the current output port.\n"
1265 "@var{message} can contain @code{~A} (was @code{%s}) and\n"
1266 "@code{~S} (was @code{%S}) escapes. When printed,\n"
1267 "the escapes are replaced with corresponding members of\n"
1269 "@code{~A} formats using @code{display} and @code{~S} formats\n"
1270 "using @code{write}.\n"
1271 "If @var{destination} is @code{#t}, then use the current output\n"
1272 "port, if @var{destination} is @code{#f}, then return a string\n"
1273 "containing the formatted text. Does not add a trailing newline.")
1274 #define FUNC_NAME s_scm_simple_format
1276 SCM port
, answer
= SCM_UNSPECIFIED
;
1277 int fReturnString
= 0;
1279 size_t start
, p
, end
;
1281 if (scm_is_eq (destination
, SCM_BOOL_T
))
1283 destination
= port
= scm_current_output_port ();
1285 else if (scm_is_false (destination
))
1288 port
= scm_mkstrport (SCM_INUM0
,
1289 scm_make_string (SCM_INUM0
, SCM_UNDEFINED
),
1290 SCM_OPN
| SCM_WRTNG
,
1296 SCM_VALIDATE_OPORT_VALUE (1, destination
);
1297 port
= SCM_COERCE_OUTPORT (destination
);
1299 SCM_VALIDATE_STRING (2, message
);
1300 SCM_VALIDATE_REST_ARGUMENT (args
);
1304 end
= scm_i_string_length (message
);
1305 for (p
= start
; p
!= end
; ++p
)
1306 if (scm_i_string_ref (message
, p
) == '~')
1311 switch (scm_i_string_ref (message
, p
))
1320 scm_lfwrite_substr (message
, start
, p
, port
);
1324 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1329 SCM_MISC_ERROR ("FORMAT: Unsupported format option ~~~A - use (ice-9 format) instead",
1330 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1335 if (!scm_is_pair (args
))
1336 SCM_MISC_ERROR ("FORMAT: Missing argument for ~~~A",
1337 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1339 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1340 /* we pass destination here */
1341 scm_prin1 (SCM_CAR (args
), destination
, writingp
);
1342 args
= SCM_CDR (args
);
1346 scm_lfwrite_substr (message
, start
, p
, port
);
1347 if (!scm_is_eq (args
, SCM_EOL
))
1348 SCM_MISC_ERROR ("FORMAT: ~A superfluous arguments",
1349 scm_list_1 (scm_length (args
)));
1352 answer
= scm_strport_to_string (destination
);
1354 return scm_return_first (answer
, message
);
1359 SCM_DEFINE (scm_newline
, "newline", 0, 1, 0,
1361 "Send a newline to @var{port}.\n"
1362 "If @var{port} is omitted, send to the current output port.")
1363 #define FUNC_NAME s_scm_newline
1365 if (SCM_UNBNDP (port
))
1366 port
= scm_current_output_port ();
1368 SCM_VALIDATE_OPORT_VALUE (1, port
);
1370 scm_putc ('\n', SCM_COERCE_OUTPORT (port
));
1371 return SCM_UNSPECIFIED
;
1375 SCM_DEFINE (scm_write_char
, "write-char", 1, 1, 0,
1376 (SCM chr
, SCM port
),
1377 "Send character @var{chr} to @var{port}.")
1378 #define FUNC_NAME s_scm_write_char
1380 if (SCM_UNBNDP (port
))
1381 port
= scm_current_output_port ();
1383 SCM_VALIDATE_CHAR (1, chr
);
1384 SCM_VALIDATE_OPORT_VALUE (2, port
);
1386 port
= SCM_COERCE_OUTPORT (port
);
1387 if (!display_character (SCM_CHAR (chr
), port
,
1388 scm_i_get_conversion_strategy (port
)))
1389 scm_encoding_error (__func__
, errno
,
1390 "cannot convert to output locale",
1391 "UTF-32", scm_i_get_port_encoding (port
),
1392 scm_string (scm_list_1 (chr
)));
1394 return SCM_UNSPECIFIED
;
1400 /* Call back to Scheme code to do the printing of special objects
1401 * (like structs). SCM_PRINTER_APPLY applies PROC to EXP and a smob
1402 * containing PORT and PSTATE. This object can be used as the port for
1403 * display/write etc to continue the current print chain. The REVEALED
1404 * field of PSTATE is set to true to indicate that the print state has
1405 * escaped to Scheme and thus has to be freed by the GC.
1408 scm_t_bits scm_tc16_port_with_ps
;
1410 /* Print exactly as the port itself would */
1413 port_with_ps_print (SCM obj
, SCM port
, scm_print_state
*pstate
)
1415 obj
= SCM_PORT_WITH_PS_PORT (obj
);
1416 return scm_ptobs
[SCM_PTOBNUM (obj
)].print (obj
, port
, pstate
);
1420 scm_printer_apply (SCM proc
, SCM exp
, SCM port
, scm_print_state
*pstate
)
1422 pstate
->revealed
= 1;
1423 return scm_call_2 (proc
, exp
,
1424 scm_i_port_with_print_state (port
, pstate
->handle
));
1427 SCM_DEFINE (scm_port_with_print_state
, "port-with-print-state", 1, 1, 0,
1428 (SCM port
, SCM pstate
),
1429 "Create a new port which behaves like @var{port}, but with an\n"
1430 "included print state @var{pstate}. @var{pstate} is optional.\n"
1431 "If @var{pstate} isn't supplied and @var{port} already has\n"
1432 "a print state, the old print state is reused.")
1433 #define FUNC_NAME s_scm_port_with_print_state
1435 SCM_VALIDATE_OPORT_VALUE (1, port
);
1436 if (!SCM_UNBNDP (pstate
))
1437 SCM_VALIDATE_PRINTSTATE (2, pstate
);
1438 return scm_i_port_with_print_state (port
, pstate
);
1442 SCM_DEFINE (scm_get_print_state
, "get-print-state", 1, 0, 0,
1444 "Return the print state of the port @var{port}. If @var{port}\n"
1445 "has no associated print state, @code{#f} is returned.")
1446 #define FUNC_NAME s_scm_get_print_state
1448 if (SCM_PORT_WITH_PS_P (port
))
1449 return SCM_PORT_WITH_PS_PS (port
);
1450 if (SCM_OUTPUT_PORT_P (port
))
1452 SCM_WRONG_TYPE_ARG (1, port
);
1461 SCM vtable
, layout
, type
;
1463 scm_init_opts (scm_print_options
, scm_print_opts
);
1465 scm_print_options (scm_list_4 (scm_from_latin1_symbol ("highlight-prefix"),
1466 scm_from_locale_string ("{"),
1467 scm_from_latin1_symbol ("highlight-suffix"),
1468 scm_from_locale_string ("}")));
1470 scm_gc_register_root (&print_state_pool
);
1471 scm_gc_register_root (&scm_print_state_vtable
);
1472 vtable
= scm_make_vtable_vtable (scm_nullstr
, SCM_INUM0
, SCM_EOL
);
1474 scm_make_struct_layout (scm_from_locale_string (SCM_PRINT_STATE_LAYOUT
));
1475 type
= scm_make_struct (vtable
, SCM_INUM0
, scm_list_1 (layout
));
1476 scm_set_struct_vtable_name_x (type
, scm_from_latin1_symbol ("print-state"));
1477 scm_print_state_vtable
= type
;
1479 /* Don't want to bind a wrapper class in GOOPS, so pass 0 as arg1. */
1480 scm_tc16_port_with_ps
= scm_make_smob_type (0, 0);
1481 scm_set_smob_print (scm_tc16_port_with_ps
, port_with_ps_print
);
1483 #include "libguile/print.x"
1485 scm_print_opts
[SCM_PRINT_KEYWORD_STYLE_I
].val
= SCM_UNPACK (sym_reader
);