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_BITS
,
104 "The string to print before highlighted values." },
105 { SCM_OPTION_SCM
, "highlight-suffix", (scm_t_bits
)SCM_BOOL_F_BITS
,
106 "The string to print after highlighted values." },
107 { SCM_OPTION_SCM
, "quote-keywordish-symbols", (scm_t_bits
)SCM_BOOL_F_BITS
,
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'." },
111 { SCM_OPTION_BOOLEAN
, "escape-newlines", 1,
112 "Render newlines as \\n when printing using `write'." },
116 SCM_DEFINE (scm_print_options
, "print-options-interface", 0, 1, 0,
118 "Option interface for the print options. Instead of using\n"
119 "this procedure directly, use the procedures\n"
120 "@code{print-enable}, @code{print-disable}, @code{print-set!}\n"
121 "and @code{print-options}.")
122 #define FUNC_NAME s_scm_print_options
124 SCM ans
= scm_options (setting
,
132 /* {Printing of Scheme Objects}
135 /* Detection of circular references.
137 * Due to other constraints in the implementation, this code has bad
138 * time complexity (O (depth * N)), The printer code can be
139 * rewritten to be O(N).
141 #define PUSH_REF(pstate, obj) \
144 PSTATE_STACK_SET (pstate, pstate->top, obj); \
146 if (pstate->top == pstate->ceiling) \
147 grow_ref_stack (pstate); \
150 #define ENTER_NESTED_DATA(pstate, obj, label) \
153 register unsigned long i; \
154 for (i = 0; i < pstate->top; ++i) \
155 if (scm_is_eq (PSTATE_STACK_REF (pstate, i), (obj))) \
157 if (pstate->fancyp) \
159 if (pstate->top - pstate->list_offset >= pstate->level) \
161 scm_putc ('#', port); \
165 PUSH_REF(pstate, obj); \
168 #define EXIT_NESTED_DATA(pstate) \
172 PSTATE_STACK_SET (pstate, pstate->top, SCM_UNDEFINED); \
176 SCM scm_print_state_vtable
= SCM_BOOL_F
;
177 static SCM print_state_pool
= SCM_EOL
;
178 scm_i_pthread_mutex_t print_state_mutex
= SCM_I_PTHREAD_MUTEX_INITIALIZER
;
180 #ifdef GUILE_DEBUG /* Used for debugging purposes */
182 SCM_DEFINE (scm_current_pstate
, "current-pstate", 0, 0, 0,
184 "Return the current-pstate -- the car of the\n"
185 "@code{print_state_pool}. @code{current-pstate} is only\n"
186 "included in @code{--enable-guile-debug} builds.")
187 #define FUNC_NAME s_scm_current_pstate
189 if (!scm_is_null (print_state_pool
))
190 return SCM_CAR (print_state_pool
);
198 #define PSTATE_SIZE 50L
201 make_print_state (void)
204 = scm_make_struct (scm_print_state_vtable
, SCM_INUM0
, SCM_EOL
);
205 scm_print_state
*pstate
= SCM_PRINT_STATE (print_state
);
206 pstate
->ref_vect
= scm_c_make_vector (PSTATE_SIZE
, SCM_UNDEFINED
);
207 pstate
->ceiling
= SCM_SIMPLE_VECTOR_LENGTH (pstate
->ref_vect
);
208 pstate
->highlight_objects
= SCM_EOL
;
213 scm_make_print_state ()
215 SCM answer
= SCM_BOOL_F
;
217 /* First try to allocate a print state from the pool */
218 scm_i_pthread_mutex_lock (&print_state_mutex
);
219 if (!scm_is_null (print_state_pool
))
221 answer
= SCM_CAR (print_state_pool
);
222 print_state_pool
= SCM_CDR (print_state_pool
);
224 scm_i_pthread_mutex_unlock (&print_state_mutex
);
226 return scm_is_false (answer
) ? make_print_state () : answer
;
230 scm_free_print_state (SCM print_state
)
233 scm_print_state
*pstate
= SCM_PRINT_STATE (print_state
);
234 /* Cleanup before returning print state to pool.
235 * It is better to do it here. Doing it in scm_prin1
236 * would cost more since that function is called much more
240 pstate
->revealed
= 0;
241 pstate
->highlight_objects
= SCM_EOL
;
242 scm_i_pthread_mutex_lock (&print_state_mutex
);
243 handle
= scm_cons (print_state
, print_state_pool
);
244 print_state_pool
= handle
;
245 scm_i_pthread_mutex_unlock (&print_state_mutex
);
249 scm_i_port_with_print_state (SCM port
, SCM print_state
)
251 if (SCM_UNBNDP (print_state
))
253 if (SCM_PORT_WITH_PS_P (port
))
256 print_state
= scm_make_print_state ();
257 /* port does not need to be coerced since it doesn't have ps */
260 port
= SCM_COERCE_OUTPORT (port
);
261 SCM_RETURN_NEWSMOB (scm_tc16_port_with_ps
,
262 SCM_UNPACK (scm_cons (port
, print_state
)));
266 grow_ref_stack (scm_print_state
*pstate
)
268 SCM old_vect
= pstate
->ref_vect
;
269 size_t old_size
= SCM_SIMPLE_VECTOR_LENGTH (old_vect
);
270 size_t new_size
= 2 * pstate
->ceiling
;
271 SCM new_vect
= scm_c_make_vector (new_size
, SCM_UNDEFINED
);
274 for (i
= 0; i
!= old_size
; ++i
)
275 SCM_SIMPLE_VECTOR_SET (new_vect
, i
, SCM_SIMPLE_VECTOR_REF (old_vect
, i
));
277 pstate
->ref_vect
= new_vect
;
278 pstate
->ceiling
= new_size
;
281 #define PSTATE_STACK_REF(p,i) SCM_SIMPLE_VECTOR_REF((p)->ref_vect, (i))
282 #define PSTATE_STACK_SET(p,i,v) SCM_SIMPLE_VECTOR_SET((p)->ref_vect, (i), (v))
285 print_circref (SCM port
, scm_print_state
*pstate
, SCM ref
)
288 long self
= pstate
->top
- 1;
290 if (scm_is_pair (PSTATE_STACK_REF (pstate
, i
)))
294 if (!scm_is_pair (PSTATE_STACK_REF (pstate
, i
-1))
295 || !scm_is_eq (SCM_CDR (PSTATE_STACK_REF (pstate
, i
-1)),
296 SCM_CDR (PSTATE_STACK_REF (pstate
, i
))))
302 for (i
= pstate
->top
- 1; 1; --i
)
303 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), ref
))
305 scm_putc ('#', port
);
306 scm_intprint (i
- self
, 10, port
);
307 scm_putc ('#', port
);
310 /* Print the name of a symbol. */
313 quote_keywordish_symbols (void)
315 SCM option
= SCM_PRINT_KEYWORD_STYLE
;
317 if (scm_is_false (option
))
319 if (scm_is_eq (option
, sym_reader
))
320 return scm_is_true (SCM_PACK (SCM_KEYWORD_STYLE
));
324 #define INITIAL_IDENTIFIER_MASK \
325 (UC_CATEGORY_MASK_Lu | UC_CATEGORY_MASK_Ll | UC_CATEGORY_MASK_Lt \
326 | UC_CATEGORY_MASK_Lm | UC_CATEGORY_MASK_Lo | UC_CATEGORY_MASK_Mn \
327 | UC_CATEGORY_MASK_Nl | UC_CATEGORY_MASK_No | UC_CATEGORY_MASK_Pd \
328 | UC_CATEGORY_MASK_Pc | UC_CATEGORY_MASK_Po | UC_CATEGORY_MASK_Sc \
329 | UC_CATEGORY_MASK_Sm | UC_CATEGORY_MASK_Sk | UC_CATEGORY_MASK_So \
330 | UC_CATEGORY_MASK_Co)
332 #define SUBSEQUENT_IDENTIFIER_MASK \
333 (INITIAL_IDENTIFIER_MASK \
334 | UC_CATEGORY_MASK_Nd | UC_CATEGORY_MASK_Mc | UC_CATEGORY_MASK_Me)
337 symbol_has_extended_read_syntax (SCM sym
)
339 size_t pos
, len
= scm_i_symbol_length (sym
);
342 /* The empty symbol. */
346 c
= scm_i_symbol_ref (sym
, 0);
348 /* Single dot; conflicts with dotted-pair notation. */
349 if (len
== 1 && c
== '.')
352 /* Other initial-character constraints. */
353 if (c
== '\'' || c
== '`' || c
== ',' || c
== '"' || c
== ';' || c
== '#')
356 /* Keywords can be identified by trailing colons too. */
357 if (c
== ':' || scm_i_symbol_ref (sym
, len
- 1) == ':')
358 return quote_keywordish_symbols ();
360 /* Number-ish symbols. */
361 if (scm_is_true (scm_i_string_to_number (scm_symbol_to_string (sym
), 10)))
364 /* Other disallowed first characters. */
365 if (!uc_is_general_category_withtable (c
, INITIAL_IDENTIFIER_MASK
))
368 /* Otherwise, any character that's in the identifier category mask is
369 fine to pass through as-is, provided it's not one of the ASCII
370 delimiters like `;'. */
371 for (pos
= 1; pos
< len
; pos
++)
373 c
= scm_i_symbol_ref (sym
, pos
);
374 if (!uc_is_general_category_withtable (c
, SUBSEQUENT_IDENTIFIER_MASK
))
376 else if (c
== '"' || c
== ';' || c
== '#')
384 print_normal_symbol (SCM sym
, SCM port
)
386 scm_display (scm_symbol_to_string (sym
), port
);
390 print_extended_symbol (SCM sym
, SCM port
)
393 scm_t_string_failed_conversion_handler strategy
;
395 len
= scm_i_symbol_length (sym
);
396 strategy
= scm_i_get_conversion_strategy (port
);
398 scm_lfwrite ("#{", 2, port
);
400 for (pos
= 0; pos
< len
; pos
++)
402 scm_t_wchar c
= scm_i_symbol_ref (sym
, pos
);
404 if (uc_is_general_category_withtable (c
,
405 SUBSEQUENT_IDENTIFIER_MASK
406 | UC_CATEGORY_MASK_Zs
))
408 if (!display_character (c
, port
, strategy
))
409 scm_encoding_error ("print_extended_symbol", errno
,
410 "cannot convert to output locale",
411 port
, SCM_MAKE_CHAR (c
));
415 display_string ("\\x", 1, 2, port
, iconveh_question_mark
);
416 scm_intprint (c
, 16, port
);
417 display_character (';', port
, iconveh_question_mark
);
421 scm_lfwrite ("}#", 2, port
);
424 /* FIXME: allow R6RS hex escapes instead of #{...}#. */
426 scm_i_print_symbol_name (SCM sym
, SCM port
)
428 if (symbol_has_extended_read_syntax (sym
))
429 print_extended_symbol (sym
, port
);
431 print_normal_symbol (sym
, port
);
435 scm_print_symbol_name (const char *str
, size_t len
, SCM port
)
437 SCM symbol
= scm_from_locale_symboln (str
, len
);
438 scm_i_print_symbol_name (symbol
, port
);
441 /* Print generally. Handles both write and display according to PSTATE.
443 SCM_GPROC(s_write
, "write", 1, 1, 0, scm_write
, g_write
);
444 SCM_GPROC(s_display
, "display", 1, 1, 0, scm_display
, g_display
);
446 static void iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
);
449 /* Print a character as an octal or hex escape. */
450 #define PRINT_CHAR_ESCAPE(i, port) \
453 if (!SCM_R6RS_ESCAPES_P) \
454 scm_intprint (i, 8, port); \
457 scm_puts ("x", port); \
458 scm_intprint (i, 16, port); \
465 scm_iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
468 && scm_is_true (scm_memq (exp
, pstate
->highlight_objects
)))
470 scm_display (SCM_PRINT_HIGHLIGHT_PREFIX
, port
);
471 iprin1 (exp
, port
, pstate
);
472 scm_display (SCM_PRINT_HIGHLIGHT_SUFFIX
, port
);
475 iprin1 (exp
, port
, pstate
);
479 iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
481 switch (SCM_ITAG3 (exp
))
485 /* These tc3 tags should never occur in an immediate value. They are
486 * only used in cell types of non-immediates, i. e. the value returned
487 * by SCM_CELL_TYPE (exp) can use these tags.
489 scm_ipruk ("immediate", exp
, port
);
493 scm_intprint (SCM_I_INUM (exp
), 10, port
);
498 if (SCM_WRITINGP (pstate
))
499 write_character (SCM_CHAR (exp
), port
, 0);
502 if (!display_character (SCM_CHAR (exp
), port
,
503 scm_i_get_conversion_strategy (port
)))
504 scm_encoding_error (__func__
, errno
,
505 "cannot convert to output locale",
509 else if (SCM_IFLAGP (exp
)
510 && ((size_t) SCM_IFLAGNUM (exp
) < (sizeof iflagnames
/ sizeof (char *))))
512 scm_puts (iflagnames
[SCM_IFLAGNUM (exp
)], port
);
516 /* unknown immediate value */
517 scm_ipruk ("immediate", exp
, port
);
521 switch (SCM_TYP7 (exp
))
525 ENTER_NESTED_DATA (pstate
, exp
, circref
);
526 if (SCM_OBJ_CLASS_FLAGS (exp
) & SCM_CLASSF_GOOPS
)
528 SCM pwps
, print
= pstate
->writingp
? g_write
: g_display
;
529 if (SCM_UNPACK (print
) == 0)
531 pwps
= scm_i_port_with_print_state (port
, pstate
->handle
);
532 pstate
->revealed
= 1;
533 scm_call_generic_2 (print
, exp
, pwps
);
538 scm_print_struct (exp
, port
, pstate
);
540 EXIT_NESTED_DATA (pstate
);
543 case scm_tcs_cons_imcar
:
544 case scm_tcs_cons_nimcar
:
545 ENTER_NESTED_DATA (pstate
, exp
, circref
);
546 scm_iprlist ("(", exp
, ')', port
, pstate
);
547 EXIT_NESTED_DATA (pstate
);
550 print_circref (port
, pstate
, exp
);
553 switch SCM_TYP16 (exp
) {
555 scm_bigprint (exp
, port
, pstate
);
558 scm_print_real (exp
, port
, pstate
);
560 case scm_tc16_complex
:
561 scm_print_complex (exp
, port
, pstate
);
563 case scm_tc16_fraction
:
564 scm_i_print_fraction (exp
, port
, pstate
);
569 if (SCM_WRITINGP (pstate
))
573 display_character ('"', port
, iconveh_question_mark
);
574 len
= scm_i_string_length (exp
);
575 for (i
= 0; i
< len
; ++i
)
576 write_character (scm_i_string_ref (exp
, i
), port
, 1);
578 display_character ('"', port
, iconveh_question_mark
);
579 scm_remember_upto_here_1 (exp
);
585 len
= scm_i_string_length (exp
);
586 printed
= display_string (scm_i_string_data (exp
),
587 scm_i_is_narrow_string (exp
),
589 scm_i_get_conversion_strategy (port
));
590 if (SCM_UNLIKELY (printed
< len
))
591 scm_encoding_error (__func__
, errno
,
592 "cannot convert to output locale",
593 port
, scm_c_string_ref (exp
, printed
));
596 scm_remember_upto_here_1 (exp
);
599 if (scm_i_symbol_is_interned (exp
))
601 scm_i_print_symbol_name (exp
, port
);
602 scm_remember_upto_here_1 (exp
);
606 scm_puts ("#<uninterned-symbol ", port
);
607 scm_i_print_symbol_name (exp
, port
);
608 scm_putc (' ', port
);
609 scm_uintprint (SCM_UNPACK (exp
), 16, port
);
610 scm_putc ('>', port
);
613 case scm_tc7_variable
:
614 scm_i_variable_print (exp
, port
, pstate
);
616 case scm_tc7_program
:
617 scm_i_program_print (exp
, port
, pstate
);
619 case scm_tc7_pointer
:
620 scm_i_pointer_print (exp
, port
, pstate
);
622 case scm_tc7_hashtable
:
623 scm_i_hashtable_print (exp
, port
, pstate
);
626 scm_i_fluid_print (exp
, port
, pstate
);
628 case scm_tc7_dynamic_state
:
629 scm_i_dynamic_state_print (exp
, port
, pstate
);
632 scm_i_frame_print (exp
, port
, pstate
);
634 case scm_tc7_objcode
:
635 scm_i_objcode_print (exp
, port
, pstate
);
638 scm_i_vm_print (exp
, port
, pstate
);
640 case scm_tc7_vm_cont
:
641 scm_i_vm_cont_print (exp
, port
, pstate
);
644 scm_i_prompt_print (exp
, port
, pstate
);
646 case scm_tc7_with_fluids
:
647 scm_i_with_fluids_print (exp
, port
, pstate
);
650 ENTER_NESTED_DATA (pstate
, exp
, circref
);
651 scm_i_print_array (exp
, port
, pstate
);
653 case scm_tc7_bytevector
:
654 scm_i_print_bytevector (exp
, port
, pstate
);
657 ENTER_NESTED_DATA (pstate
, exp
, circref
);
658 if (SCM_IS_WHVEC (exp
))
659 scm_puts ("#wh(", port
);
661 scm_puts ("#w(", port
);
662 goto common_vector_printer
;
664 ENTER_NESTED_DATA (pstate
, exp
, circref
);
665 scm_puts ("#(", port
);
666 common_vector_printer
:
669 long last
= SCM_SIMPLE_VECTOR_LENGTH (exp
) - 1;
672 && SCM_SIMPLE_VECTOR_LENGTH (exp
) > pstate
->length
)
674 last
= pstate
->length
- 1;
677 if (SCM_I_WVECTP (exp
))
679 /* Elements of weak vectors may not be accessed via the
680 `SIMPLE_VECTOR_REF ()' macro. */
681 for (i
= 0; i
< last
; ++i
)
683 scm_iprin1 (scm_c_vector_ref (exp
, i
),
685 scm_putc (' ', port
);
690 for (i
= 0; i
< last
; ++i
)
692 scm_iprin1 (SCM_SIMPLE_VECTOR_REF (exp
, i
), port
, pstate
);
693 scm_putc (' ', port
);
700 scm_iprin1 (scm_c_vector_ref (exp
, i
), port
, pstate
);
703 scm_puts (" ...", port
);
704 scm_putc (')', port
);
706 EXIT_NESTED_DATA (pstate
);
710 register long i
= SCM_PTOBNUM (exp
);
712 && scm_ptobs
[i
].print
713 && (scm_ptobs
[i
].print
) (exp
, port
, pstate
))
718 ENTER_NESTED_DATA (pstate
, exp
, circref
);
719 SCM_SMOB_DESCRIPTOR (exp
).print (exp
, port
, pstate
);
720 EXIT_NESTED_DATA (pstate
);
723 /* case scm_tcs_closures: */
725 scm_ipruk ("type", exp
, port
);
730 /* Print states are necessary for circular reference safe printing.
731 * They are also expensive to allocate. Therefore print states are
732 * kept in a pool so that they can be reused.
735 /* The PORT argument can also be a print-state/port pair, which will
736 * then be used instead of allocating a new print state. This is
737 * useful for continuing a chain of print calls from Scheme. */
740 scm_prin1 (SCM exp
, SCM port
, int writingp
)
742 SCM handle
= SCM_BOOL_F
; /* Will GC protect the handle whilst unlinked */
744 scm_print_state
*pstate
;
747 /* If PORT is a print-state/port pair, use that. Else create a new
750 if (SCM_PORT_WITH_PS_P (port
))
752 pstate_scm
= SCM_PORT_WITH_PS_PS (port
);
753 port
= SCM_PORT_WITH_PS_PORT (port
);
757 /* First try to allocate a print state from the pool */
758 scm_i_pthread_mutex_lock (&print_state_mutex
);
759 if (!scm_is_null (print_state_pool
))
761 handle
= print_state_pool
;
762 print_state_pool
= SCM_CDR (print_state_pool
);
764 scm_i_pthread_mutex_unlock (&print_state_mutex
);
765 if (scm_is_false (handle
))
766 handle
= scm_list_1 (make_print_state ());
767 pstate_scm
= SCM_CAR (handle
);
770 pstate
= SCM_PRINT_STATE (pstate_scm
);
771 old_writingp
= pstate
->writingp
;
772 pstate
->writingp
= writingp
;
773 scm_iprin1 (exp
, port
, pstate
);
774 pstate
->writingp
= old_writingp
;
776 /* Return print state to pool if it has been created above and
777 hasn't escaped to Scheme. */
779 if (scm_is_true (handle
) && !pstate
->revealed
)
781 scm_i_pthread_mutex_lock (&print_state_mutex
);
782 SCM_SETCDR (handle
, print_state_pool
);
783 print_state_pool
= handle
;
784 scm_i_pthread_mutex_unlock (&print_state_mutex
);
788 /* Convert codepoint CH to UTF-8 and store the result in UTF8. Return
789 the number of bytes of the UTF-8-encoded string. */
791 codepoint_to_utf8 (scm_t_wchar ch
, scm_t_uint8 utf8
[4])
794 scm_t_uint32 codepoint
;
796 codepoint
= (scm_t_uint32
) ch
;
798 if (codepoint
<= 0x7f)
801 utf8
[0] = (scm_t_uint8
) codepoint
;
803 else if (codepoint
<= 0x7ffUL
)
806 utf8
[0] = 0xc0 | (codepoint
>> 6);
807 utf8
[1] = 0x80 | (codepoint
& 0x3f);
809 else if (codepoint
<= 0xffffUL
)
812 utf8
[0] = 0xe0 | (codepoint
>> 12);
813 utf8
[1] = 0x80 | ((codepoint
>> 6) & 0x3f);
814 utf8
[2] = 0x80 | (codepoint
& 0x3f);
819 utf8
[0] = 0xf0 | (codepoint
>> 18);
820 utf8
[1] = 0x80 | ((codepoint
>> 12) & 0x3f);
821 utf8
[2] = 0x80 | ((codepoint
>> 6) & 0x3f);
822 utf8
[3] = 0x80 | (codepoint
& 0x3f);
828 #define STR_REF(s, x) \
830 ? (scm_t_wchar) ((unsigned char *) (s))[x] \
831 : ((scm_t_wchar *) (s))[x])
833 /* Write STR to PORT as UTF-8. STR is a LEN-codepoint string; it is
834 narrow if NARROW_P is true, wide otherwise. Return LEN. */
836 display_string_as_utf8 (const void *str
, int narrow_p
, size_t len
,
841 while (len
> printed
)
844 char *input
, utf8_buf
[256];
846 /* Convert STR to UTF-8. */
847 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
848 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
851 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
852 (scm_t_uint8
*) input
);
853 input
= utf8_buf
+ utf8_len
;
856 /* INPUT was successfully converted, entirely; print the
858 scm_lfwrite (utf8_buf
, utf8_len
, port
);
859 printed
+= i
- printed
;
862 assert (printed
== len
);
867 /* Convert STR through PORT's output conversion descriptor and write the
868 output to PORT. Return the number of codepoints written. */
870 display_string_using_iconv (const void *str
, int narrow_p
, size_t len
,
872 scm_t_string_failed_conversion_handler strategy
)
877 pt
= SCM_PTAB_ENTRY (port
);
881 while (len
> printed
)
883 size_t done
, utf8_len
, input_left
, output_left
, i
;
884 size_t codepoints_read
, output_len
;
885 char *input
, *output
;
886 char utf8_buf
[256], encoded_output
[256];
889 /* Convert STR to UTF-8. */
890 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
891 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
894 offsets
[utf8_len
] = i
;
895 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
896 (scm_t_uint8
*) input
);
897 input
= utf8_buf
+ utf8_len
;
901 input_left
= utf8_len
;
903 output
= encoded_output
;
904 output_left
= sizeof (encoded_output
);
906 done
= iconv (pt
->output_cd
, &input
, &input_left
,
907 &output
, &output_left
);
909 output_len
= sizeof (encoded_output
) - output_left
;
911 if (SCM_UNLIKELY (done
== (size_t) -1))
913 int errno_save
= errno
;
915 /* Reset the `iconv' state. */
916 iconv (pt
->output_cd
, NULL
, NULL
, NULL
, NULL
);
918 /* Print the OUTPUT_LEN bytes successfully converted. */
919 scm_lfwrite (encoded_output
, output_len
, port
);
921 /* See how many input codepoints these OUTPUT_LEN bytes
923 codepoints_read
= offsets
[input
- utf8_buf
] - printed
;
924 printed
+= codepoints_read
;
926 if (errno_save
== EILSEQ
&&
927 strategy
!= SCM_FAILED_CONVERSION_ERROR
)
929 /* Conversion failed somewhere in INPUT and we want to
930 escape or substitute the offending input character. */
932 if (strategy
== SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
)
936 /* Find CH, the offending codepoint, and escape it. */
937 ch
= STR_REF (str
, offsets
[input
- utf8_buf
]);
938 write_character_escaped (ch
, 1, port
);
941 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
942 display_string ("?", 1, 1, port
, strategy
);
947 /* Something bad happened that we can't handle: bail out. */
952 /* INPUT was successfully converted, entirely; print the
954 scm_lfwrite (encoded_output
, output_len
, port
);
955 codepoints_read
= i
- printed
;
956 printed
+= codepoints_read
;
965 /* Display the LEN codepoints in STR to PORT according to STRATEGY;
966 return the number of codepoints successfully displayed. If NARROW_P,
967 then STR is interpreted as a sequence of `char', denoting a Latin-1
968 string; otherwise it's interpreted as a sequence of
971 display_string (const void *str
, int narrow_p
,
972 size_t len
, SCM port
,
973 scm_t_string_failed_conversion_handler strategy
)
978 pt
= SCM_PTAB_ENTRY (port
);
980 if (pt
->output_cd
== (iconv_t
) -1)
981 /* Initialize the conversion descriptors, if needed. */
982 scm_i_set_port_encoding_x (port
, pt
->encoding
);
984 /* FIXME: In 2.1, add a flag to determine whether a port is UTF-8. */
985 if (pt
->output_cd
== (iconv_t
) -1)
986 return display_string_as_utf8 (str
, narrow_p
, len
, port
);
988 return display_string_using_iconv (str
, narrow_p
, len
,
992 /* Attempt to display CH to PORT according to STRATEGY. Return non-zero
993 if CH was successfully displayed, zero otherwise (e.g., if it was not
994 representable in PORT's encoding.) */
996 display_character (scm_t_wchar ch
, SCM port
,
997 scm_t_string_failed_conversion_handler strategy
)
999 return display_string (&ch
, 0, 1, port
, strategy
) == 1;
1002 /* Attempt to pretty-print CH, a combining character, to PORT. Return
1003 zero upon failure, non-zero otherwise. The idea is to print CH above
1004 a dotted circle to make it more visible. */
1006 write_combining_character (scm_t_wchar ch
, SCM port
)
1010 str
[0] = SCM_CODEPOINT_DOTTED_CIRCLE
;
1013 return display_string (str
, 0, 2, port
, iconveh_error
) == 2;
1016 /* Write CH to PORT in its escaped form, using the string escape syntax
1017 if STRING_ESCAPES_P is non-zero. */
1019 write_character_escaped (scm_t_wchar ch
, int string_escapes_p
, SCM port
)
1021 if (string_escapes_p
)
1023 /* Represent CH using the in-string escape syntax. */
1025 static const char hex
[] = "0123456789abcdef";
1026 static const char escapes
[7] = "abtnvfr";
1029 if (ch
>= 0x07 && ch
<= 0x0D && ch
!= 0x0A)
1031 /* Use special escapes for some C0 controls. */
1033 buf
[1] = escapes
[ch
- 0x07];
1034 scm_lfwrite (buf
, 2, port
);
1036 else if (!SCM_R6RS_ESCAPES_P
)
1042 buf
[2] = hex
[ch
/ 16];
1043 buf
[3] = hex
[ch
% 16];
1044 scm_lfwrite (buf
, 4, port
);
1046 else if (ch
<= 0xFFFF)
1050 buf
[2] = hex
[(ch
& 0xF000) >> 12];
1051 buf
[3] = hex
[(ch
& 0xF00) >> 8];
1052 buf
[4] = hex
[(ch
& 0xF0) >> 4];
1053 buf
[5] = hex
[(ch
& 0xF)];
1054 scm_lfwrite (buf
, 6, port
);
1056 else if (ch
> 0xFFFF)
1060 buf
[2] = hex
[(ch
& 0xF00000) >> 20];
1061 buf
[3] = hex
[(ch
& 0xF0000) >> 16];
1062 buf
[4] = hex
[(ch
& 0xF000) >> 12];
1063 buf
[5] = hex
[(ch
& 0xF00) >> 8];
1064 buf
[6] = hex
[(ch
& 0xF0) >> 4];
1065 buf
[7] = hex
[(ch
& 0xF)];
1066 scm_lfwrite (buf
, 8, port
);
1071 /* Print an R6RS variable-length hex escape: "\xNNNN;". */
1072 scm_t_wchar ch2
= ch
;
1082 buf
[i
] = hex
[ch2
& 0xF];
1089 scm_lfwrite (buf
+ i
, 9 - i
, port
);
1094 /* Represent CH using the character escape syntax. */
1097 name
= scm_i_charname (SCM_MAKE_CHAR (ch
));
1099 scm_puts (name
, port
);
1101 PRINT_CHAR_ESCAPE (ch
, port
);
1105 /* Write CH to PORT, escaping it if it's non-graphic or not
1106 representable in PORT's encoding. If STRING_ESCAPES_P is true and CH
1107 needs to be escaped, it is escaped using the in-string escape syntax;
1108 otherwise the character escape syntax is used. */
1110 write_character (scm_t_wchar ch
, SCM port
, int string_escapes_p
)
1113 scm_t_string_failed_conversion_handler strategy
;
1115 strategy
= scm_i_get_conversion_strategy (port
);
1117 if (string_escapes_p
)
1119 /* Check if CH deserves special treatment. */
1120 if (ch
== '"' || ch
== '\\')
1122 display_character ('\\', port
, iconveh_question_mark
);
1123 display_character (ch
, port
, strategy
);
1126 else if (ch
== '\n' && SCM_PRINT_ESCAPE_NEWLINES_P
)
1128 display_character ('\\', port
, iconveh_question_mark
);
1129 display_character ('n', port
, strategy
);
1132 else if (ch
== ' ' || ch
== '\n')
1134 display_character (ch
, port
, strategy
);
1140 display_string ("#\\", 1, 2, port
, iconveh_question_mark
);
1142 if (uc_combining_class (ch
) != UC_CCC_NR
)
1143 /* Character is a combining character, so attempt to
1145 printed
= write_combining_character (ch
, port
);
1149 && uc_is_general_category_withtable (ch
,
1150 UC_CATEGORY_MASK_L
|
1151 UC_CATEGORY_MASK_M
|
1152 UC_CATEGORY_MASK_N
|
1153 UC_CATEGORY_MASK_P
|
1154 UC_CATEGORY_MASK_S
))
1155 /* CH is graphic; attempt to display it. */
1156 printed
= display_character (ch
, port
, iconveh_error
);
1159 /* CH isn't graphic or cannot be represented in PORT's encoding. */
1160 write_character_escaped (ch
, string_escapes_p
, port
);
1163 /* Print an integer.
1167 scm_intprint (scm_t_intmax n
, int radix
, SCM port
)
1169 char num_buf
[SCM_INTBUFLEN
];
1170 scm_lfwrite (num_buf
, scm_iint2str (n
, radix
, num_buf
), port
);
1174 scm_uintprint (scm_t_uintmax n
, int radix
, SCM port
)
1176 char num_buf
[SCM_INTBUFLEN
];
1177 scm_lfwrite (num_buf
, scm_iuint2str (n
, radix
, num_buf
), port
);
1180 /* Print an object of unrecognized type.
1184 scm_ipruk (char *hdr
, SCM ptr
, SCM port
)
1186 scm_puts ("#<unknown-", port
);
1187 scm_puts (hdr
, port
);
1188 if (1) /* (scm_in_heap_p (ptr)) */ /* FIXME */
1190 scm_puts (" (0x", port
);
1191 scm_uintprint (SCM_CELL_WORD_0 (ptr
), 16, port
);
1192 scm_puts (" . 0x", port
);
1193 scm_uintprint (SCM_CELL_WORD_1 (ptr
), 16, port
);
1194 scm_puts (") @", port
);
1196 scm_puts (" 0x", port
);
1197 scm_uintprint (SCM_UNPACK (ptr
), 16, port
);
1198 scm_putc ('>', port
);
1205 scm_iprlist (char *hdr
, SCM exp
, int tlr
, SCM port
, scm_print_state
*pstate
)
1207 register SCM hare
, tortoise
;
1208 long floor
= pstate
->top
- 2;
1209 scm_puts (hdr
, port
);
1212 goto fancy_printing
;
1214 /* Run a hare and tortoise so that total time complexity will be
1215 O(depth * N) instead of O(N^2). */
1216 hare
= SCM_CDR (exp
);
1218 while (scm_is_pair (hare
))
1220 if (scm_is_eq (hare
, tortoise
))
1221 goto fancy_printing
;
1222 hare
= SCM_CDR (hare
);
1223 if (!scm_is_pair (hare
))
1225 hare
= SCM_CDR (hare
);
1226 tortoise
= SCM_CDR (tortoise
);
1229 /* No cdr cycles intrinsic to this list */
1230 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1231 for (exp
= SCM_CDR (exp
); scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1235 for (i
= floor
; i
>= 0; --i
)
1236 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1238 PUSH_REF (pstate
, exp
);
1239 scm_putc (' ', port
);
1241 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1243 if (!SCM_NULL_OR_NIL_P (exp
))
1245 scm_puts (" . ", port
);
1246 scm_iprin1 (exp
, port
, pstate
);
1250 scm_putc (tlr
, port
);
1251 pstate
->top
= floor
+ 2;
1256 long n
= pstate
->length
;
1258 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1259 exp
= SCM_CDR (exp
); --n
;
1260 for (; scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1262 register unsigned long i
;
1264 for (i
= 0; i
< pstate
->top
; ++i
)
1265 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1271 scm_puts (" ...", port
);
1277 PUSH_REF(pstate
, exp
);
1278 ++pstate
->list_offset
;
1279 scm_putc (' ', port
);
1281 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1284 if (!SCM_NULL_OR_NIL_P (exp
))
1286 scm_puts (" . ", port
);
1287 scm_iprin1 (exp
, port
, pstate
);
1290 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1294 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1297 scm_puts (" . ", port
);
1298 print_circref (port
, pstate
, exp
);
1305 scm_valid_oport_value_p (SCM val
)
1307 return (SCM_OPOUTPORTP (val
)
1308 || (SCM_PORT_WITH_PS_P (val
)
1309 && SCM_OPOUTPORTP (SCM_PORT_WITH_PS_PORT (val
))));
1312 /* SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write); */
1315 scm_write (SCM obj
, SCM port
)
1317 if (SCM_UNBNDP (port
))
1318 port
= scm_current_output_port ();
1320 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_write
);
1322 scm_prin1 (obj
, port
, 1);
1323 return SCM_UNSPECIFIED
;
1327 /* SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display); */
1330 scm_display (SCM obj
, SCM port
)
1332 if (SCM_UNBNDP (port
))
1333 port
= scm_current_output_port ();
1335 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_display
);
1337 scm_prin1 (obj
, port
, 0);
1338 return SCM_UNSPECIFIED
;
1342 SCM_DEFINE (scm_simple_format
, "simple-format", 2, 0, 1,
1343 (SCM destination
, SCM message
, SCM args
),
1344 "Write @var{message} to @var{destination}, defaulting to\n"
1345 "the current output port.\n"
1346 "@var{message} can contain @code{~A} (was @code{%s}) and\n"
1347 "@code{~S} (was @code{%S}) escapes. When printed,\n"
1348 "the escapes are replaced with corresponding members of\n"
1350 "@code{~A} formats using @code{display} and @code{~S} formats\n"
1351 "using @code{write}.\n"
1352 "If @var{destination} is @code{#t}, then use the current output\n"
1353 "port, if @var{destination} is @code{#f}, then return a string\n"
1354 "containing the formatted text. Does not add a trailing newline.")
1355 #define FUNC_NAME s_scm_simple_format
1357 SCM port
, answer
= SCM_UNSPECIFIED
;
1358 int fReturnString
= 0;
1360 size_t start
, p
, end
;
1362 if (scm_is_eq (destination
, SCM_BOOL_T
))
1364 destination
= port
= scm_current_output_port ();
1366 else if (scm_is_false (destination
))
1369 port
= scm_mkstrport (SCM_INUM0
, SCM_BOOL_F
,
1370 SCM_OPN
| SCM_WRTNG
,
1376 SCM_VALIDATE_OPORT_VALUE (1, destination
);
1377 port
= SCM_COERCE_OUTPORT (destination
);
1379 SCM_VALIDATE_STRING (2, message
);
1380 SCM_VALIDATE_REST_ARGUMENT (args
);
1384 end
= scm_i_string_length (message
);
1385 for (p
= start
; p
!= end
; ++p
)
1386 if (scm_i_string_ref (message
, p
) == '~')
1391 switch (scm_i_string_ref (message
, p
))
1400 scm_lfwrite_substr (message
, start
, p
, port
);
1404 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1409 SCM_MISC_ERROR ("FORMAT: Unsupported format option ~~~A - use (ice-9 format) instead",
1410 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1415 if (!scm_is_pair (args
))
1416 SCM_MISC_ERROR ("FORMAT: Missing argument for ~~~A",
1417 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1419 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1420 /* we pass destination here */
1421 scm_prin1 (SCM_CAR (args
), destination
, writingp
);
1422 args
= SCM_CDR (args
);
1426 scm_lfwrite_substr (message
, start
, p
, port
);
1427 if (!scm_is_eq (args
, SCM_EOL
))
1428 SCM_MISC_ERROR ("FORMAT: ~A superfluous arguments",
1429 scm_list_1 (scm_length (args
)));
1432 answer
= scm_strport_to_string (destination
);
1434 return scm_return_first (answer
, message
);
1439 SCM_DEFINE (scm_newline
, "newline", 0, 1, 0,
1441 "Send a newline to @var{port}.\n"
1442 "If @var{port} is omitted, send to the current output port.")
1443 #define FUNC_NAME s_scm_newline
1445 if (SCM_UNBNDP (port
))
1446 port
= scm_current_output_port ();
1448 SCM_VALIDATE_OPORT_VALUE (1, port
);
1450 scm_putc ('\n', SCM_COERCE_OUTPORT (port
));
1451 return SCM_UNSPECIFIED
;
1455 SCM_DEFINE (scm_write_char
, "write-char", 1, 1, 0,
1456 (SCM chr
, SCM port
),
1457 "Send character @var{chr} to @var{port}.")
1458 #define FUNC_NAME s_scm_write_char
1460 if (SCM_UNBNDP (port
))
1461 port
= scm_current_output_port ();
1463 SCM_VALIDATE_CHAR (1, chr
);
1464 SCM_VALIDATE_OPORT_VALUE (2, port
);
1466 port
= SCM_COERCE_OUTPORT (port
);
1467 if (!display_character (SCM_CHAR (chr
), port
,
1468 scm_i_get_conversion_strategy (port
)))
1469 scm_encoding_error (__func__
, errno
,
1470 "cannot convert to output locale",
1473 return SCM_UNSPECIFIED
;
1479 /* Call back to Scheme code to do the printing of special objects
1480 * (like structs). SCM_PRINTER_APPLY applies PROC to EXP and a smob
1481 * containing PORT and PSTATE. This object can be used as the port for
1482 * display/write etc to continue the current print chain. The REVEALED
1483 * field of PSTATE is set to true to indicate that the print state has
1484 * escaped to Scheme and thus has to be freed by the GC.
1487 scm_t_bits scm_tc16_port_with_ps
;
1489 /* Print exactly as the port itself would */
1492 port_with_ps_print (SCM obj
, SCM port
, scm_print_state
*pstate
)
1494 obj
= SCM_PORT_WITH_PS_PORT (obj
);
1495 return scm_ptobs
[SCM_PTOBNUM (obj
)].print (obj
, port
, pstate
);
1499 scm_printer_apply (SCM proc
, SCM exp
, SCM port
, scm_print_state
*pstate
)
1501 pstate
->revealed
= 1;
1502 return scm_call_2 (proc
, exp
,
1503 scm_i_port_with_print_state (port
, pstate
->handle
));
1506 SCM_DEFINE (scm_port_with_print_state
, "port-with-print-state", 1, 1, 0,
1507 (SCM port
, SCM pstate
),
1508 "Create a new port which behaves like @var{port}, but with an\n"
1509 "included print state @var{pstate}. @var{pstate} is optional.\n"
1510 "If @var{pstate} isn't supplied and @var{port} already has\n"
1511 "a print state, the old print state is reused.")
1512 #define FUNC_NAME s_scm_port_with_print_state
1514 SCM_VALIDATE_OPORT_VALUE (1, port
);
1515 if (!SCM_UNBNDP (pstate
))
1516 SCM_VALIDATE_PRINTSTATE (2, pstate
);
1517 return scm_i_port_with_print_state (port
, pstate
);
1521 SCM_DEFINE (scm_get_print_state
, "get-print-state", 1, 0, 0,
1523 "Return the print state of the port @var{port}. If @var{port}\n"
1524 "has no associated print state, @code{#f} is returned.")
1525 #define FUNC_NAME s_scm_get_print_state
1527 if (SCM_PORT_WITH_PS_P (port
))
1528 return SCM_PORT_WITH_PS_PS (port
);
1529 if (SCM_OUTPUT_PORT_P (port
))
1531 SCM_WRONG_TYPE_ARG (1, port
);
1540 SCM vtable
, layout
, type
;
1542 scm_gc_register_root (&print_state_pool
);
1543 scm_gc_register_root (&scm_print_state_vtable
);
1544 vtable
= scm_make_vtable_vtable (scm_nullstr
, SCM_INUM0
, SCM_EOL
);
1546 scm_make_struct_layout (scm_from_locale_string (SCM_PRINT_STATE_LAYOUT
));
1547 type
= scm_make_struct (vtable
, SCM_INUM0
, scm_list_1 (layout
));
1548 scm_set_struct_vtable_name_x (type
, scm_from_latin1_symbol ("print-state"));
1549 scm_print_state_vtable
= type
;
1551 /* Don't want to bind a wrapper class in GOOPS, so pass 0 as arg1. */
1552 scm_tc16_port_with_ps
= scm_make_smob_type (0, 0);
1553 scm_set_smob_print (scm_tc16_port_with_ps
, port_with_ps_print
);
1555 #include "libguile/print.x"
1557 scm_init_opts (scm_print_options
, scm_print_opts
);
1558 scm_print_opts
[SCM_PRINT_HIGHLIGHT_PREFIX_I
].val
=
1559 SCM_UNPACK (scm_from_locale_string ("{"));
1560 scm_print_opts
[SCM_PRINT_HIGHLIGHT_SUFFIX_I
].val
=
1561 SCM_UNPACK (scm_from_locale_string ("}"));
1562 scm_print_opts
[SCM_PRINT_KEYWORD_STYLE_I
].val
= SCM_UNPACK (sym_reader
);