1 /* Copyright (C) 1995-1999, 2000, 2001, 2002, 2003, 2004, 2006, 2008,
2 * 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
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/ports-internal.h"
49 #include "libguile/root.h"
50 #include "libguile/strings.h"
51 #include "libguile/strports.h"
52 #include "libguile/vectors.h"
53 #include "libguile/numbers.h"
54 #include "libguile/vm.h"
56 #include "libguile/validate.h"
57 #include "libguile/print.h"
59 #include "libguile/private-options.h"
63 /* Character printers. */
65 #define PORT_CONVERSION_HANDLER(port) \
66 SCM_PTAB_ENTRY (port)->ilseq_handler
68 static size_t display_string (const void *, int, size_t, SCM
,
69 scm_t_string_failed_conversion_handler
);
71 static int display_character (scm_t_wchar
, SCM
,
72 scm_t_string_failed_conversion_handler
);
74 static void write_character (scm_t_wchar
, SCM
, int);
76 static void write_character_escaped (scm_t_wchar
, int, SCM
);
80 /* {Names of immediate symbols}
82 * This table must agree with the declarations in scm.h: {Immediate Symbols}.
85 /* This table must agree with the list of flags in tags.h. */
86 static const char *iflagnames
[] =
89 "#nil", /* Elisp nil value. Should print from elisp as symbol `nil'. */
90 "#<XXX UNUSED LISP FALSE -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
93 "#<XXX UNUSED BOOLEAN 0 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
94 "#<XXX UNUSED BOOLEAN 1 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
95 "#<XXX UNUSED BOOLEAN 2 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
100 /* Unbound slot marker for GOOPS. For internal use in GOOPS only. */
104 SCM_SYMBOL (sym_reader
, "reader");
106 scm_t_option scm_print_opts
[] = {
107 { SCM_OPTION_SCM
, "highlight-prefix", (scm_t_bits
)SCM_BOOL_F_BITS
,
108 "The string to print before highlighted values." },
109 { SCM_OPTION_SCM
, "highlight-suffix", (scm_t_bits
)SCM_BOOL_F_BITS
,
110 "The string to print after highlighted values." },
111 { SCM_OPTION_SCM
, "quote-keywordish-symbols", (scm_t_bits
)SCM_BOOL_F_BITS
,
112 "How to print symbols that have a colon as their first or last character. "
113 "The value '#f' does not quote the colons; '#t' quotes them; "
114 "'reader' quotes them when the reader option 'keywords' is not '#f'." },
115 { SCM_OPTION_BOOLEAN
, "escape-newlines", 1,
116 "Render newlines as \\n when printing using `write'." },
120 SCM_DEFINE (scm_print_options
, "print-options-interface", 0, 1, 0,
122 "Option interface for the print options. Instead of using\n"
123 "this procedure directly, use the procedures\n"
124 "@code{print-enable}, @code{print-disable}, @code{print-set!}\n"
125 "and @code{print-options}.")
126 #define FUNC_NAME s_scm_print_options
128 SCM ans
= scm_options (setting
,
136 /* {Printing of Scheme Objects}
139 /* Detection of circular references.
141 * Due to other constraints in the implementation, this code has bad
142 * time complexity (O (depth * N)), The printer code can be
143 * rewritten to be O(N).
145 #define PUSH_REF(pstate, obj) \
148 PSTATE_STACK_SET (pstate, pstate->top, obj); \
150 if (pstate->top == pstate->ceiling) \
151 grow_ref_stack (pstate); \
154 #define ENTER_NESTED_DATA(pstate, obj, label) \
157 register unsigned long i; \
158 for (i = 0; i < pstate->top; ++i) \
159 if (scm_is_eq (PSTATE_STACK_REF (pstate, i), (obj))) \
161 if (pstate->fancyp) \
163 if (pstate->top - pstate->list_offset >= pstate->level) \
165 scm_putc ('#', port); \
169 PUSH_REF(pstate, obj); \
172 #define EXIT_NESTED_DATA(pstate) \
176 PSTATE_STACK_SET (pstate, pstate->top, SCM_UNDEFINED); \
180 SCM scm_print_state_vtable
= SCM_BOOL_F
;
181 static SCM print_state_pool
= SCM_EOL
;
182 scm_i_pthread_mutex_t print_state_mutex
= SCM_I_PTHREAD_MUTEX_INITIALIZER
;
184 #ifdef GUILE_DEBUG /* Used for debugging purposes */
186 SCM_DEFINE (scm_current_pstate
, "current-pstate", 0, 0, 0,
188 "Return the current-pstate -- the car of the\n"
189 "@code{print_state_pool}. @code{current-pstate} is only\n"
190 "included in @code{--enable-guile-debug} builds.")
191 #define FUNC_NAME s_scm_current_pstate
193 if (!scm_is_null (print_state_pool
))
194 return SCM_CAR (print_state_pool
);
202 #define PSTATE_SIZE 50L
205 make_print_state (void)
208 = scm_make_struct (scm_print_state_vtable
, SCM_INUM0
, SCM_EOL
);
209 scm_print_state
*pstate
= SCM_PRINT_STATE (print_state
);
210 pstate
->ref_vect
= scm_c_make_vector (PSTATE_SIZE
, SCM_UNDEFINED
);
211 pstate
->ceiling
= SCM_SIMPLE_VECTOR_LENGTH (pstate
->ref_vect
);
212 pstate
->highlight_objects
= SCM_EOL
;
217 scm_make_print_state ()
219 SCM answer
= SCM_BOOL_F
;
221 /* First try to allocate a print state from the pool */
222 scm_i_pthread_mutex_lock (&print_state_mutex
);
223 if (!scm_is_null (print_state_pool
))
225 answer
= SCM_CAR (print_state_pool
);
226 print_state_pool
= SCM_CDR (print_state_pool
);
228 scm_i_pthread_mutex_unlock (&print_state_mutex
);
230 return scm_is_false (answer
) ? make_print_state () : answer
;
234 scm_free_print_state (SCM print_state
)
237 scm_print_state
*pstate
= SCM_PRINT_STATE (print_state
);
238 /* Cleanup before returning print state to pool.
239 * It is better to do it here. Doing it in scm_prin1
240 * would cost more since that function is called much more
244 pstate
->revealed
= 0;
245 pstate
->highlight_objects
= SCM_EOL
;
246 scm_i_pthread_mutex_lock (&print_state_mutex
);
247 handle
= scm_cons (print_state
, print_state_pool
);
248 print_state_pool
= handle
;
249 scm_i_pthread_mutex_unlock (&print_state_mutex
);
253 scm_i_port_with_print_state (SCM port
, SCM print_state
)
255 if (SCM_UNBNDP (print_state
))
257 if (SCM_PORT_WITH_PS_P (port
))
260 print_state
= scm_make_print_state ();
261 /* port does not need to be coerced since it doesn't have ps */
264 port
= SCM_COERCE_OUTPORT (port
);
265 SCM_RETURN_NEWSMOB (scm_tc16_port_with_ps
,
266 SCM_UNPACK (scm_cons (port
, print_state
)));
270 grow_ref_stack (scm_print_state
*pstate
)
272 SCM old_vect
= pstate
->ref_vect
;
273 size_t old_size
= SCM_SIMPLE_VECTOR_LENGTH (old_vect
);
274 size_t new_size
= 2 * pstate
->ceiling
;
275 SCM new_vect
= scm_c_make_vector (new_size
, SCM_UNDEFINED
);
278 for (i
= 0; i
!= old_size
; ++i
)
279 SCM_SIMPLE_VECTOR_SET (new_vect
, i
, SCM_SIMPLE_VECTOR_REF (old_vect
, i
));
281 pstate
->ref_vect
= new_vect
;
282 pstate
->ceiling
= new_size
;
285 #define PSTATE_STACK_REF(p,i) SCM_SIMPLE_VECTOR_REF((p)->ref_vect, (i))
286 #define PSTATE_STACK_SET(p,i,v) SCM_SIMPLE_VECTOR_SET((p)->ref_vect, (i), (v))
289 print_circref (SCM port
, scm_print_state
*pstate
, SCM ref
)
292 long self
= pstate
->top
- 1;
294 if (scm_is_pair (PSTATE_STACK_REF (pstate
, i
)))
298 if (!scm_is_pair (PSTATE_STACK_REF (pstate
, i
-1))
299 || !scm_is_eq (SCM_CDR (PSTATE_STACK_REF (pstate
, i
-1)),
300 SCM_CDR (PSTATE_STACK_REF (pstate
, i
))))
306 for (i
= pstate
->top
- 1; 1; --i
)
307 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), ref
))
309 scm_putc ('#', port
);
310 scm_intprint (i
- self
, 10, port
);
311 scm_putc ('#', port
);
314 /* Print the name of a symbol. */
317 quote_keywordish_symbols (void)
319 SCM 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
));
328 #define INITIAL_IDENTIFIER_MASK \
329 (UC_CATEGORY_MASK_Lu | UC_CATEGORY_MASK_Ll | UC_CATEGORY_MASK_Lt \
330 | UC_CATEGORY_MASK_Lm | UC_CATEGORY_MASK_Lo | UC_CATEGORY_MASK_Mn \
331 | UC_CATEGORY_MASK_Nl | UC_CATEGORY_MASK_No | UC_CATEGORY_MASK_Pd \
332 | UC_CATEGORY_MASK_Pc | UC_CATEGORY_MASK_Po | UC_CATEGORY_MASK_Sc \
333 | UC_CATEGORY_MASK_Sm | UC_CATEGORY_MASK_Sk | UC_CATEGORY_MASK_So \
334 | UC_CATEGORY_MASK_Co)
336 #define SUBSEQUENT_IDENTIFIER_MASK \
337 (INITIAL_IDENTIFIER_MASK \
338 | UC_CATEGORY_MASK_Nd | UC_CATEGORY_MASK_Mc | UC_CATEGORY_MASK_Me)
341 symbol_has_extended_read_syntax (SCM sym
)
343 size_t pos
, len
= scm_i_symbol_length (sym
);
346 /* The empty symbol. */
350 c
= scm_i_symbol_ref (sym
, 0);
352 /* Single dot; conflicts with dotted-pair notation. */
353 if (len
== 1 && c
== '.')
356 /* Other initial-character constraints. */
357 if (c
== '\'' || c
== '`' || c
== ',' || c
== '"' || c
== ';' || c
== '#')
360 /* Keywords can be identified by trailing colons too. */
361 if (c
== ':' || scm_i_symbol_ref (sym
, len
- 1) == ':')
362 return quote_keywordish_symbols ();
364 /* Number-ish symbols. */
365 if (scm_is_true (scm_i_string_to_number (scm_symbol_to_string (sym
), 10)))
368 /* Other disallowed first characters. */
369 if (!uc_is_general_category_withtable (c
, INITIAL_IDENTIFIER_MASK
))
372 /* Otherwise, any character that's in the identifier category mask is
373 fine to pass through as-is, provided it's not one of the ASCII
374 delimiters like `;'. */
375 for (pos
= 1; pos
< len
; pos
++)
377 c
= scm_i_symbol_ref (sym
, pos
);
378 if (!uc_is_general_category_withtable (c
, SUBSEQUENT_IDENTIFIER_MASK
))
380 else if (c
== '"' || c
== ';' || c
== '#')
388 print_normal_symbol (SCM sym
, SCM port
)
390 scm_display (scm_symbol_to_string (sym
), port
);
394 print_extended_symbol (SCM sym
, SCM port
)
397 scm_t_string_failed_conversion_handler strategy
;
399 len
= scm_i_symbol_length (sym
);
400 strategy
= PORT_CONVERSION_HANDLER (port
);
402 scm_lfwrite ("#{", 2, port
);
404 for (pos
= 0; pos
< len
; pos
++)
406 scm_t_wchar c
= scm_i_symbol_ref (sym
, pos
);
408 if (uc_is_general_category_withtable (c
,
409 SUBSEQUENT_IDENTIFIER_MASK
410 | UC_CATEGORY_MASK_Zs
))
412 if (!display_character (c
, port
, strategy
))
413 scm_encoding_error ("print_extended_symbol", errno
,
414 "cannot convert to output locale",
415 port
, SCM_MAKE_CHAR (c
));
419 display_string ("\\x", 1, 2, port
, iconveh_question_mark
);
420 scm_intprint (c
, 16, port
);
421 display_character (';', port
, iconveh_question_mark
);
425 scm_lfwrite ("}#", 2, port
);
428 /* FIXME: allow R6RS hex escapes instead of #{...}#. */
430 scm_i_print_symbol_name (SCM sym
, SCM port
)
432 if (symbol_has_extended_read_syntax (sym
))
433 print_extended_symbol (sym
, port
);
435 print_normal_symbol (sym
, port
);
439 scm_print_symbol_name (const char *str
, size_t len
, SCM port
)
441 SCM symbol
= scm_from_locale_symboln (str
, len
);
442 scm_i_print_symbol_name (symbol
, port
);
445 /* Print generally. Handles both write and display according to PSTATE.
447 SCM_GPROC(s_write
, "write", 1, 1, 0, scm_write
, g_write
);
448 SCM_GPROC(s_display
, "display", 1, 1, 0, scm_display
, g_display
);
450 static void iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
);
453 /* Print a character as an octal or hex escape. */
454 #define PRINT_CHAR_ESCAPE(i, port) \
457 if (!SCM_R6RS_ESCAPES_P) \
458 scm_intprint (i, 8, port); \
461 scm_puts ("x", port); \
462 scm_intprint (i, 16, port); \
469 scm_iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
472 && scm_is_true (scm_memq (exp
, pstate
->highlight_objects
)))
474 scm_display (SCM_PRINT_HIGHLIGHT_PREFIX
, port
);
475 iprin1 (exp
, port
, pstate
);
476 scm_display (SCM_PRINT_HIGHLIGHT_SUFFIX
, port
);
479 iprin1 (exp
, port
, pstate
);
483 iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
485 switch (SCM_ITAG3 (exp
))
489 /* These tc3 tags should never occur in an immediate value. They are
490 * only used in cell types of non-immediates, i. e. the value returned
491 * by SCM_CELL_TYPE (exp) can use these tags.
493 scm_ipruk ("immediate", exp
, port
);
497 scm_intprint (SCM_I_INUM (exp
), 10, port
);
502 if (SCM_WRITINGP (pstate
))
503 write_character (SCM_CHAR (exp
), port
, 0);
506 if (!display_character (SCM_CHAR (exp
), port
,
507 PORT_CONVERSION_HANDLER (port
)))
508 scm_encoding_error (__func__
, errno
,
509 "cannot convert to output locale",
513 else if (SCM_IFLAGP (exp
)
514 && ((size_t) SCM_IFLAGNUM (exp
) < (sizeof iflagnames
/ sizeof (char *))))
516 scm_puts (iflagnames
[SCM_IFLAGNUM (exp
)], port
);
520 /* unknown immediate value */
521 scm_ipruk ("immediate", exp
, port
);
525 switch (SCM_TYP7 (exp
))
529 ENTER_NESTED_DATA (pstate
, exp
, circref
);
530 if (SCM_OBJ_CLASS_FLAGS (exp
) & SCM_CLASSF_GOOPS
)
532 SCM pwps
, print
= pstate
->writingp
? g_write
: g_display
;
533 if (SCM_UNPACK (print
) == 0)
535 pwps
= scm_i_port_with_print_state (port
, pstate
->handle
);
536 pstate
->revealed
= 1;
537 scm_call_generic_2 (print
, exp
, pwps
);
542 scm_print_struct (exp
, port
, pstate
);
544 EXIT_NESTED_DATA (pstate
);
547 case scm_tcs_cons_imcar
:
548 case scm_tcs_cons_nimcar
:
549 ENTER_NESTED_DATA (pstate
, exp
, circref
);
550 scm_iprlist ("(", exp
, ')', port
, pstate
);
551 EXIT_NESTED_DATA (pstate
);
554 print_circref (port
, pstate
, exp
);
557 switch SCM_TYP16 (exp
) {
559 scm_bigprint (exp
, port
, pstate
);
562 scm_print_real (exp
, port
, pstate
);
564 case scm_tc16_complex
:
565 scm_print_complex (exp
, port
, pstate
);
567 case scm_tc16_fraction
:
568 scm_i_print_fraction (exp
, port
, pstate
);
573 if (SCM_WRITINGP (pstate
))
577 display_character ('"', port
, iconveh_question_mark
);
578 len
= scm_i_string_length (exp
);
579 for (i
= 0; i
< len
; ++i
)
580 write_character (scm_i_string_ref (exp
, i
), port
, 1);
582 display_character ('"', port
, iconveh_question_mark
);
583 scm_remember_upto_here_1 (exp
);
589 len
= scm_i_string_length (exp
);
590 printed
= display_string (scm_i_string_data (exp
),
591 scm_i_is_narrow_string (exp
),
593 PORT_CONVERSION_HANDLER (port
));
594 if (SCM_UNLIKELY (printed
< len
))
595 scm_encoding_error (__func__
, errno
,
596 "cannot convert to output locale",
597 port
, scm_c_string_ref (exp
, printed
));
600 scm_remember_upto_here_1 (exp
);
603 if (scm_i_symbol_is_interned (exp
))
605 scm_i_print_symbol_name (exp
, port
);
606 scm_remember_upto_here_1 (exp
);
610 scm_puts ("#<uninterned-symbol ", port
);
611 scm_i_print_symbol_name (exp
, port
);
612 scm_putc (' ', port
);
613 scm_uintprint (SCM_UNPACK (exp
), 16, port
);
614 scm_putc ('>', port
);
617 case scm_tc7_variable
:
618 scm_i_variable_print (exp
, port
, pstate
);
620 case scm_tc7_program
:
621 scm_i_program_print (exp
, port
, pstate
);
623 case scm_tc7_pointer
:
624 scm_i_pointer_print (exp
, port
, pstate
);
626 case scm_tc7_hashtable
:
627 scm_i_hashtable_print (exp
, port
, pstate
);
630 scm_i_fluid_print (exp
, port
, pstate
);
632 case scm_tc7_dynamic_state
:
633 scm_i_dynamic_state_print (exp
, port
, pstate
);
636 scm_i_frame_print (exp
, port
, pstate
);
638 case scm_tc7_objcode
:
639 scm_i_objcode_print (exp
, port
, pstate
);
642 scm_i_vm_print (exp
, port
, pstate
);
644 case scm_tc7_vm_cont
:
645 scm_i_vm_cont_print (exp
, port
, pstate
);
648 scm_i_prompt_print (exp
, port
, pstate
);
650 case scm_tc7_with_fluids
:
651 scm_i_with_fluids_print (exp
, port
, pstate
);
654 ENTER_NESTED_DATA (pstate
, exp
, circref
);
655 scm_i_print_array (exp
, port
, pstate
);
656 EXIT_NESTED_DATA (pstate
);
658 case scm_tc7_bytevector
:
659 scm_i_print_bytevector (exp
, port
, pstate
);
661 case scm_tc7_bitvector
:
662 scm_i_print_bitvector (exp
, port
, pstate
);
665 ENTER_NESTED_DATA (pstate
, exp
, circref
);
666 if (SCM_IS_WHVEC (exp
))
667 scm_puts ("#wh(", port
);
669 scm_puts ("#w(", port
);
670 goto common_vector_printer
;
672 ENTER_NESTED_DATA (pstate
, exp
, circref
);
673 scm_puts ("#(", port
);
674 common_vector_printer
:
677 long last
= SCM_SIMPLE_VECTOR_LENGTH (exp
) - 1;
680 && SCM_SIMPLE_VECTOR_LENGTH (exp
) > pstate
->length
)
682 last
= pstate
->length
- 1;
685 if (SCM_I_WVECTP (exp
))
687 /* Elements of weak vectors may not be accessed via the
688 `SIMPLE_VECTOR_REF ()' macro. */
689 for (i
= 0; i
< last
; ++i
)
691 scm_iprin1 (scm_c_vector_ref (exp
, i
),
693 scm_putc (' ', port
);
698 for (i
= 0; i
< last
; ++i
)
700 scm_iprin1 (SCM_SIMPLE_VECTOR_REF (exp
, i
), port
, pstate
);
701 scm_putc (' ', port
);
708 scm_iprin1 (scm_c_vector_ref (exp
, i
), port
, pstate
);
711 scm_puts (" ...", port
);
712 scm_putc (')', port
);
714 EXIT_NESTED_DATA (pstate
);
718 register long i
= SCM_PTOBNUM (exp
);
720 && scm_ptobs
[i
].print
721 && (scm_ptobs
[i
].print
) (exp
, port
, pstate
))
726 ENTER_NESTED_DATA (pstate
, exp
, circref
);
727 SCM_SMOB_DESCRIPTOR (exp
).print (exp
, port
, pstate
);
728 EXIT_NESTED_DATA (pstate
);
731 /* case scm_tcs_closures: */
733 scm_ipruk ("type", exp
, port
);
738 /* Print states are necessary for circular reference safe printing.
739 * They are also expensive to allocate. Therefore print states are
740 * kept in a pool so that they can be reused.
743 /* The PORT argument can also be a print-state/port pair, which will
744 * then be used instead of allocating a new print state. This is
745 * useful for continuing a chain of print calls from Scheme. */
748 scm_prin1 (SCM exp
, SCM port
, int writingp
)
750 SCM handle
= SCM_BOOL_F
; /* Will GC protect the handle whilst unlinked */
752 scm_print_state
*pstate
;
755 /* If PORT is a print-state/port pair, use that. Else create a new
758 if (SCM_PORT_WITH_PS_P (port
))
760 pstate_scm
= SCM_PORT_WITH_PS_PS (port
);
761 port
= SCM_PORT_WITH_PS_PORT (port
);
765 /* First try to allocate a print state from the pool */
766 scm_i_pthread_mutex_lock (&print_state_mutex
);
767 if (!scm_is_null (print_state_pool
))
769 handle
= print_state_pool
;
770 print_state_pool
= SCM_CDR (print_state_pool
);
772 scm_i_pthread_mutex_unlock (&print_state_mutex
);
773 if (scm_is_false (handle
))
774 handle
= scm_list_1 (make_print_state ());
775 pstate_scm
= SCM_CAR (handle
);
778 pstate
= SCM_PRINT_STATE (pstate_scm
);
779 old_writingp
= pstate
->writingp
;
780 pstate
->writingp
= writingp
;
781 scm_iprin1 (exp
, port
, pstate
);
782 pstate
->writingp
= old_writingp
;
784 /* Return print state to pool if it has been created above and
785 hasn't escaped to Scheme. */
787 if (scm_is_true (handle
) && !pstate
->revealed
)
789 scm_i_pthread_mutex_lock (&print_state_mutex
);
790 SCM_SETCDR (handle
, print_state_pool
);
791 print_state_pool
= handle
;
792 scm_i_pthread_mutex_unlock (&print_state_mutex
);
796 /* Convert codepoint CH to UTF-8 and store the result in UTF8. Return
797 the number of bytes of the UTF-8-encoded string. */
799 codepoint_to_utf8 (scm_t_wchar ch
, scm_t_uint8 utf8
[4])
802 scm_t_uint32 codepoint
;
804 codepoint
= (scm_t_uint32
) ch
;
806 if (codepoint
<= 0x7f)
809 utf8
[0] = (scm_t_uint8
) codepoint
;
811 else if (codepoint
<= 0x7ffUL
)
814 utf8
[0] = 0xc0 | (codepoint
>> 6);
815 utf8
[1] = 0x80 | (codepoint
& 0x3f);
817 else if (codepoint
<= 0xffffUL
)
820 utf8
[0] = 0xe0 | (codepoint
>> 12);
821 utf8
[1] = 0x80 | ((codepoint
>> 6) & 0x3f);
822 utf8
[2] = 0x80 | (codepoint
& 0x3f);
827 utf8
[0] = 0xf0 | (codepoint
>> 18);
828 utf8
[1] = 0x80 | ((codepoint
>> 12) & 0x3f);
829 utf8
[2] = 0x80 | ((codepoint
>> 6) & 0x3f);
830 utf8
[3] = 0x80 | (codepoint
& 0x3f);
836 #define STR_REF(s, x) \
838 ? (scm_t_wchar) ((unsigned char *) (s))[x] \
839 : ((scm_t_wchar *) (s))[x])
841 /* Write STR to PORT as UTF-8. STR is a LEN-codepoint string; it is
842 narrow if NARROW_P is true, wide otherwise. Return LEN. */
844 display_string_as_utf8 (const void *str
, int narrow_p
, size_t len
,
849 while (len
> printed
)
852 char *input
, utf8_buf
[256];
854 /* Convert STR to UTF-8. */
855 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
856 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
859 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
860 (scm_t_uint8
*) input
);
861 input
= utf8_buf
+ utf8_len
;
864 /* INPUT was successfully converted, entirely; print the
866 scm_lfwrite (utf8_buf
, utf8_len
, port
);
867 printed
+= i
- printed
;
870 assert (printed
== len
);
875 /* Convert STR through PORT's output conversion descriptor and write the
876 output to PORT. Return the number of codepoints written. */
878 display_string_using_iconv (const void *str
, int narrow_p
, size_t len
,
880 scm_t_string_failed_conversion_handler strategy
)
883 scm_t_iconv_descriptors
*id
;
885 id
= scm_i_port_iconv_descriptors (port
);
889 while (len
> printed
)
891 size_t done
, utf8_len
, input_left
, output_left
, i
;
892 size_t codepoints_read
, output_len
;
893 char *input
, *output
;
894 char utf8_buf
[256], encoded_output
[256];
897 /* Convert STR to UTF-8. */
898 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
899 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
902 offsets
[utf8_len
] = i
;
903 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
904 (scm_t_uint8
*) input
);
905 input
= utf8_buf
+ utf8_len
;
909 input_left
= utf8_len
;
911 output
= encoded_output
;
912 output_left
= sizeof (encoded_output
);
914 done
= iconv (id
->output_cd
, &input
, &input_left
,
915 &output
, &output_left
);
917 output_len
= sizeof (encoded_output
) - output_left
;
919 if (SCM_UNLIKELY (done
== (size_t) -1))
921 int errno_save
= errno
;
923 /* Reset the `iconv' state. */
924 iconv (id
->output_cd
, NULL
, NULL
, NULL
, NULL
);
926 /* Print the OUTPUT_LEN bytes successfully converted. */
927 scm_lfwrite (encoded_output
, output_len
, port
);
929 /* See how many input codepoints these OUTPUT_LEN bytes
931 codepoints_read
= offsets
[input
- utf8_buf
] - printed
;
932 printed
+= codepoints_read
;
934 if (errno_save
== EILSEQ
&&
935 strategy
!= SCM_FAILED_CONVERSION_ERROR
)
937 /* Conversion failed somewhere in INPUT and we want to
938 escape or substitute the offending input character. */
940 if (strategy
== SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
)
944 /* Find CH, the offending codepoint, and escape it. */
945 ch
= STR_REF (str
, offsets
[input
- utf8_buf
]);
946 write_character_escaped (ch
, 1, port
);
949 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
950 display_string ("?", 1, 1, port
, strategy
);
955 /* Something bad happened that we can't handle: bail out. */
960 /* INPUT was successfully converted, entirely; print the
962 scm_lfwrite (encoded_output
, output_len
, port
);
963 codepoints_read
= i
- printed
;
964 printed
+= codepoints_read
;
973 /* Display the LEN codepoints in STR to PORT according to STRATEGY;
974 return the number of codepoints successfully displayed. If NARROW_P,
975 then STR is interpreted as a sequence of `char', denoting a Latin-1
976 string; otherwise it's interpreted as a sequence of
979 display_string (const void *str
, int narrow_p
,
980 size_t len
, SCM port
,
981 scm_t_string_failed_conversion_handler strategy
)
984 scm_t_port_internal
*pti
;
986 pti
= SCM_PORT_GET_INTERNAL (port
);
988 if (pti
->encoding_mode
== SCM_PORT_ENCODING_MODE_UTF8
)
989 return display_string_as_utf8 (str
, narrow_p
, len
, port
);
991 return display_string_using_iconv (str
, narrow_p
, len
,
995 /* Attempt to display CH to PORT according to STRATEGY. Return non-zero
996 if CH was successfully displayed, zero otherwise (e.g., if it was not
997 representable in PORT's encoding.) */
999 display_character (scm_t_wchar ch
, SCM port
,
1000 scm_t_string_failed_conversion_handler strategy
)
1002 return display_string (&ch
, 0, 1, port
, strategy
) == 1;
1005 /* Attempt to pretty-print CH, a combining character, to PORT. Return
1006 zero upon failure, non-zero otherwise. The idea is to print CH above
1007 a dotted circle to make it more visible. */
1009 write_combining_character (scm_t_wchar ch
, SCM port
)
1013 str
[0] = SCM_CODEPOINT_DOTTED_CIRCLE
;
1016 return display_string (str
, 0, 2, port
, iconveh_error
) == 2;
1019 /* Write CH to PORT in its escaped form, using the string escape syntax
1020 if STRING_ESCAPES_P is non-zero. */
1022 write_character_escaped (scm_t_wchar ch
, int string_escapes_p
, SCM port
)
1024 if (string_escapes_p
)
1026 /* Represent CH using the in-string escape syntax. */
1028 static const char hex
[] = "0123456789abcdef";
1029 static const char escapes
[7] = "abtnvfr";
1032 if (ch
>= 0x07 && ch
<= 0x0D && ch
!= 0x0A)
1034 /* Use special escapes for some C0 controls. */
1036 buf
[1] = escapes
[ch
- 0x07];
1037 scm_lfwrite (buf
, 2, port
);
1039 else if (!SCM_R6RS_ESCAPES_P
)
1045 buf
[2] = hex
[ch
/ 16];
1046 buf
[3] = hex
[ch
% 16];
1047 scm_lfwrite (buf
, 4, port
);
1049 else if (ch
<= 0xFFFF)
1053 buf
[2] = hex
[(ch
& 0xF000) >> 12];
1054 buf
[3] = hex
[(ch
& 0xF00) >> 8];
1055 buf
[4] = hex
[(ch
& 0xF0) >> 4];
1056 buf
[5] = hex
[(ch
& 0xF)];
1057 scm_lfwrite (buf
, 6, port
);
1059 else if (ch
> 0xFFFF)
1063 buf
[2] = hex
[(ch
& 0xF00000) >> 20];
1064 buf
[3] = hex
[(ch
& 0xF0000) >> 16];
1065 buf
[4] = hex
[(ch
& 0xF000) >> 12];
1066 buf
[5] = hex
[(ch
& 0xF00) >> 8];
1067 buf
[6] = hex
[(ch
& 0xF0) >> 4];
1068 buf
[7] = hex
[(ch
& 0xF)];
1069 scm_lfwrite (buf
, 8, port
);
1074 /* Print an R6RS variable-length hex escape: "\xNNNN;". */
1075 scm_t_wchar ch2
= ch
;
1085 buf
[i
] = hex
[ch2
& 0xF];
1092 scm_lfwrite (buf
+ i
, 9 - i
, port
);
1097 /* Represent CH using the character escape syntax. */
1100 name
= scm_i_charname (SCM_MAKE_CHAR (ch
));
1102 scm_puts (name
, port
);
1104 PRINT_CHAR_ESCAPE (ch
, port
);
1108 /* Write CH to PORT, escaping it if it's non-graphic or not
1109 representable in PORT's encoding. If STRING_ESCAPES_P is true and CH
1110 needs to be escaped, it is escaped using the in-string escape syntax;
1111 otherwise the character escape syntax is used. */
1113 write_character (scm_t_wchar ch
, SCM port
, int string_escapes_p
)
1116 scm_t_string_failed_conversion_handler strategy
;
1118 strategy
= PORT_CONVERSION_HANDLER (port
);
1120 if (string_escapes_p
)
1122 /* Check if CH deserves special treatment. */
1123 if (ch
== '"' || ch
== '\\')
1125 display_character ('\\', port
, iconveh_question_mark
);
1126 display_character (ch
, port
, strategy
);
1129 else if (ch
== '\n' && SCM_PRINT_ESCAPE_NEWLINES_P
)
1131 display_character ('\\', port
, iconveh_question_mark
);
1132 display_character ('n', port
, strategy
);
1135 else if (ch
== ' ' || ch
== '\n')
1137 display_character (ch
, port
, strategy
);
1143 display_string ("#\\", 1, 2, port
, iconveh_question_mark
);
1145 if (uc_combining_class (ch
) != UC_CCC_NR
)
1146 /* Character is a combining character, so attempt to
1148 printed
= write_combining_character (ch
, port
);
1152 && uc_is_general_category_withtable (ch
,
1153 UC_CATEGORY_MASK_L
|
1154 UC_CATEGORY_MASK_M
|
1155 UC_CATEGORY_MASK_N
|
1156 UC_CATEGORY_MASK_P
|
1157 UC_CATEGORY_MASK_S
))
1158 /* CH is graphic; attempt to display it. */
1159 printed
= display_character (ch
, port
, iconveh_error
);
1162 /* CH isn't graphic or cannot be represented in PORT's encoding. */
1163 write_character_escaped (ch
, string_escapes_p
, port
);
1166 /* Display STR to PORT from START inclusive to END exclusive. */
1168 scm_i_display_substring (SCM str
, size_t start
, size_t end
, SCM port
)
1172 size_t len
, printed
;
1174 buf
= scm_i_string_data (str
);
1176 narrow_p
= scm_i_is_narrow_string (str
);
1177 buf
+= start
* (narrow_p
? sizeof (char) : sizeof (scm_t_wchar
));
1179 printed
= display_string (buf
, narrow_p
, end
- start
, port
,
1180 PORT_CONVERSION_HANDLER (port
));
1182 if (SCM_UNLIKELY (printed
< len
))
1183 scm_encoding_error (__func__
, errno
,
1184 "cannot convert to output locale",
1185 port
, scm_c_string_ref (str
, printed
+ start
));
1189 /* Print an integer.
1193 scm_intprint (scm_t_intmax n
, int radix
, SCM port
)
1195 char num_buf
[SCM_INTBUFLEN
];
1196 scm_lfwrite (num_buf
, scm_iint2str (n
, radix
, num_buf
), port
);
1200 scm_uintprint (scm_t_uintmax n
, int radix
, SCM port
)
1202 char num_buf
[SCM_INTBUFLEN
];
1203 scm_lfwrite (num_buf
, scm_iuint2str (n
, radix
, num_buf
), port
);
1206 /* Print an object of unrecognized type.
1210 scm_ipruk (char *hdr
, SCM ptr
, SCM port
)
1212 scm_puts ("#<unknown-", port
);
1213 scm_puts (hdr
, port
);
1214 if (1) /* (scm_in_heap_p (ptr)) */ /* FIXME */
1216 scm_puts (" (0x", port
);
1217 scm_uintprint (SCM_CELL_WORD_0 (ptr
), 16, port
);
1218 scm_puts (" . 0x", port
);
1219 scm_uintprint (SCM_CELL_WORD_1 (ptr
), 16, port
);
1220 scm_puts (") @", port
);
1222 scm_puts (" 0x", port
);
1223 scm_uintprint (SCM_UNPACK (ptr
), 16, port
);
1224 scm_putc ('>', port
);
1231 scm_iprlist (char *hdr
, SCM exp
, int tlr
, SCM port
, scm_print_state
*pstate
)
1233 register SCM hare
, tortoise
;
1234 long floor
= pstate
->top
- 2;
1235 scm_puts (hdr
, port
);
1238 goto fancy_printing
;
1240 /* Run a hare and tortoise so that total time complexity will be
1241 O(depth * N) instead of O(N^2). */
1242 hare
= SCM_CDR (exp
);
1244 while (scm_is_pair (hare
))
1246 if (scm_is_eq (hare
, tortoise
))
1247 goto fancy_printing
;
1248 hare
= SCM_CDR (hare
);
1249 if (!scm_is_pair (hare
))
1251 hare
= SCM_CDR (hare
);
1252 tortoise
= SCM_CDR (tortoise
);
1255 /* No cdr cycles intrinsic to this list */
1256 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1257 for (exp
= SCM_CDR (exp
); scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1261 for (i
= floor
; i
>= 0; --i
)
1262 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1264 PUSH_REF (pstate
, exp
);
1265 scm_putc (' ', port
);
1267 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1269 if (!SCM_NULL_OR_NIL_P (exp
))
1271 scm_puts (" . ", port
);
1272 scm_iprin1 (exp
, port
, pstate
);
1276 scm_putc (tlr
, port
);
1277 pstate
->top
= floor
+ 2;
1282 long n
= pstate
->length
;
1284 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1285 exp
= SCM_CDR (exp
); --n
;
1286 for (; scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1288 register unsigned long i
;
1290 for (i
= 0; i
< pstate
->top
; ++i
)
1291 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1297 scm_puts (" ...", port
);
1303 PUSH_REF(pstate
, exp
);
1304 ++pstate
->list_offset
;
1305 scm_putc (' ', port
);
1307 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1310 if (!SCM_NULL_OR_NIL_P (exp
))
1312 scm_puts (" . ", port
);
1313 scm_iprin1 (exp
, port
, pstate
);
1316 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1320 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1323 scm_puts (" . ", port
);
1324 print_circref (port
, pstate
, exp
);
1331 scm_valid_oport_value_p (SCM val
)
1333 return (SCM_OPOUTPORTP (val
)
1334 || (SCM_PORT_WITH_PS_P (val
)
1335 && SCM_OPOUTPORTP (SCM_PORT_WITH_PS_PORT (val
))));
1338 /* SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write); */
1341 scm_write (SCM obj
, SCM port
)
1343 if (SCM_UNBNDP (port
))
1344 port
= scm_current_output_port ();
1346 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_write
);
1348 scm_prin1 (obj
, port
, 1);
1349 return SCM_UNSPECIFIED
;
1353 /* SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display); */
1356 scm_display (SCM obj
, SCM port
)
1358 if (SCM_UNBNDP (port
))
1359 port
= scm_current_output_port ();
1361 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_display
);
1363 scm_prin1 (obj
, port
, 0);
1364 return SCM_UNSPECIFIED
;
1368 SCM_DEFINE (scm_simple_format
, "simple-format", 2, 0, 1,
1369 (SCM destination
, SCM message
, SCM args
),
1370 "Write @var{message} to @var{destination}, defaulting to\n"
1371 "the current output port.\n"
1372 "@var{message} can contain @code{~A} (was @code{%s}) and\n"
1373 "@code{~S} (was @code{%S}) escapes. When printed,\n"
1374 "the escapes are replaced with corresponding members of\n"
1376 "@code{~A} formats using @code{display} and @code{~S} formats\n"
1377 "using @code{write}.\n"
1378 "If @var{destination} is @code{#t}, then use the current output\n"
1379 "port, if @var{destination} is @code{#f}, then return a string\n"
1380 "containing the formatted text. Does not add a trailing newline.")
1381 #define FUNC_NAME s_scm_simple_format
1383 SCM port
, answer
= SCM_UNSPECIFIED
;
1384 int fReturnString
= 0;
1386 size_t start
, p
, end
;
1388 if (scm_is_eq (destination
, SCM_BOOL_T
))
1390 destination
= port
= scm_current_output_port ();
1392 else if (scm_is_false (destination
))
1395 port
= scm_mkstrport (SCM_INUM0
, SCM_BOOL_F
,
1396 SCM_OPN
| SCM_WRTNG
,
1402 SCM_VALIDATE_OPORT_VALUE (1, destination
);
1403 port
= SCM_COERCE_OUTPORT (destination
);
1405 SCM_VALIDATE_STRING (2, message
);
1406 SCM_VALIDATE_REST_ARGUMENT (args
);
1410 end
= scm_i_string_length (message
);
1411 for (p
= start
; p
!= end
; ++p
)
1412 if (scm_i_string_ref (message
, p
) == '~')
1417 switch (scm_i_string_ref (message
, p
))
1426 scm_lfwrite_substr (message
, start
, p
, port
);
1430 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1435 SCM_MISC_ERROR ("FORMAT: Unsupported format option ~~~A - use (ice-9 format) instead",
1436 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1441 if (!scm_is_pair (args
))
1442 SCM_MISC_ERROR ("FORMAT: Missing argument for ~~~A",
1443 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1445 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1446 /* we pass destination here */
1447 scm_prin1 (SCM_CAR (args
), destination
, writingp
);
1448 args
= SCM_CDR (args
);
1452 scm_lfwrite_substr (message
, start
, p
, port
);
1453 if (!scm_is_eq (args
, SCM_EOL
))
1454 SCM_MISC_ERROR ("FORMAT: ~A superfluous arguments",
1455 scm_list_1 (scm_length (args
)));
1458 answer
= scm_strport_to_string (destination
);
1460 return scm_return_first (answer
, message
);
1465 SCM_DEFINE (scm_newline
, "newline", 0, 1, 0,
1467 "Send a newline to @var{port}.\n"
1468 "If @var{port} is omitted, send to the current output port.")
1469 #define FUNC_NAME s_scm_newline
1471 if (SCM_UNBNDP (port
))
1472 port
= scm_current_output_port ();
1474 SCM_VALIDATE_OPORT_VALUE (1, port
);
1476 scm_putc ('\n', SCM_COERCE_OUTPORT (port
));
1477 return SCM_UNSPECIFIED
;
1481 SCM_DEFINE (scm_write_char
, "write-char", 1, 1, 0,
1482 (SCM chr
, SCM port
),
1483 "Send character @var{chr} to @var{port}.")
1484 #define FUNC_NAME s_scm_write_char
1486 if (SCM_UNBNDP (port
))
1487 port
= scm_current_output_port ();
1489 SCM_VALIDATE_CHAR (1, chr
);
1490 SCM_VALIDATE_OPORT_VALUE (2, port
);
1492 port
= SCM_COERCE_OUTPORT (port
);
1493 if (!display_character (SCM_CHAR (chr
), port
,
1494 PORT_CONVERSION_HANDLER (port
)))
1495 scm_encoding_error (__func__
, errno
,
1496 "cannot convert to output locale",
1499 return SCM_UNSPECIFIED
;
1505 /* Call back to Scheme code to do the printing of special objects
1506 * (like structs). SCM_PRINTER_APPLY applies PROC to EXP and a smob
1507 * containing PORT and PSTATE. This object can be used as the port for
1508 * display/write etc to continue the current print chain. The REVEALED
1509 * field of PSTATE is set to true to indicate that the print state has
1510 * escaped to Scheme and thus has to be freed by the GC.
1513 scm_t_bits scm_tc16_port_with_ps
;
1515 /* Print exactly as the port itself would */
1518 port_with_ps_print (SCM obj
, SCM port
, scm_print_state
*pstate
)
1520 obj
= SCM_PORT_WITH_PS_PORT (obj
);
1521 return scm_ptobs
[SCM_PTOBNUM (obj
)].print (obj
, port
, pstate
);
1525 scm_printer_apply (SCM proc
, SCM exp
, SCM port
, scm_print_state
*pstate
)
1527 pstate
->revealed
= 1;
1528 return scm_call_2 (proc
, exp
,
1529 scm_i_port_with_print_state (port
, pstate
->handle
));
1532 SCM_DEFINE (scm_port_with_print_state
, "port-with-print-state", 1, 1, 0,
1533 (SCM port
, SCM pstate
),
1534 "Create a new port which behaves like @var{port}, but with an\n"
1535 "included print state @var{pstate}. @var{pstate} is optional.\n"
1536 "If @var{pstate} isn't supplied and @var{port} already has\n"
1537 "a print state, the old print state is reused.")
1538 #define FUNC_NAME s_scm_port_with_print_state
1540 SCM_VALIDATE_OPORT_VALUE (1, port
);
1541 if (!SCM_UNBNDP (pstate
))
1542 SCM_VALIDATE_PRINTSTATE (2, pstate
);
1543 return scm_i_port_with_print_state (port
, pstate
);
1547 SCM_DEFINE (scm_get_print_state
, "get-print-state", 1, 0, 0,
1549 "Return the print state of the port @var{port}. If @var{port}\n"
1550 "has no associated print state, @code{#f} is returned.")
1551 #define FUNC_NAME s_scm_get_print_state
1553 if (SCM_PORT_WITH_PS_P (port
))
1554 return SCM_PORT_WITH_PS_PS (port
);
1555 if (SCM_OUTPUT_PORT_P (port
))
1557 SCM_WRONG_TYPE_ARG (1, port
);
1568 scm_gc_register_root (&print_state_pool
);
1569 scm_gc_register_root (&scm_print_state_vtable
);
1570 type
= scm_make_vtable (scm_from_locale_string (SCM_PRINT_STATE_LAYOUT
),
1572 scm_set_struct_vtable_name_x (type
, scm_from_latin1_symbol ("print-state"));
1573 scm_print_state_vtable
= type
;
1575 /* Don't want to bind a wrapper class in GOOPS, so pass 0 as arg1. */
1576 scm_tc16_port_with_ps
= scm_make_smob_type (0, 0);
1577 scm_set_smob_print (scm_tc16_port_with_ps
, port_with_ps_print
);
1579 #include "libguile/print.x"
1581 scm_init_opts (scm_print_options
, scm_print_opts
);
1582 scm_print_opts
[SCM_PRINT_HIGHLIGHT_PREFIX_I
].val
=
1583 SCM_UNPACK (scm_from_locale_string ("{"));
1584 scm_print_opts
[SCM_PRINT_HIGHLIGHT_SUFFIX_I
].val
=
1585 SCM_UNPACK (scm_from_locale_string ("}"));
1586 scm_print_opts
[SCM_PRINT_KEYWORD_STYLE_I
].val
= SCM_UNPACK (sym_reader
);