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'."
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_symbols (void)
314 SCM option
= SCM_PRINT_KEYWORD_STYLE
;
316 if (scm_is_false (option
))
318 if (scm_is_eq (option
, sym_reader
))
319 return scm_is_true (SCM_PACK (SCM_KEYWORD_STYLE
));
323 #define INITIAL_IDENTIFIER_MASK \
324 (UC_CATEGORY_MASK_Lu | UC_CATEGORY_MASK_Ll | UC_CATEGORY_MASK_Lt \
325 | UC_CATEGORY_MASK_Lm | UC_CATEGORY_MASK_Lo | UC_CATEGORY_MASK_Mn \
326 | UC_CATEGORY_MASK_Nl | UC_CATEGORY_MASK_No | UC_CATEGORY_MASK_Pd \
327 | UC_CATEGORY_MASK_Pc | UC_CATEGORY_MASK_Po | UC_CATEGORY_MASK_Sc \
328 | UC_CATEGORY_MASK_Sm | UC_CATEGORY_MASK_Sk | UC_CATEGORY_MASK_So \
329 | UC_CATEGORY_MASK_Co)
331 #define SUBSEQUENT_IDENTIFIER_MASK \
332 (INITIAL_IDENTIFIER_MASK \
333 | UC_CATEGORY_MASK_Nd | UC_CATEGORY_MASK_Mc | UC_CATEGORY_MASK_Me)
336 symbol_has_extended_read_syntax (SCM sym
)
338 size_t pos
, len
= scm_i_symbol_length (sym
);
341 /* The empty symbol. */
345 c
= scm_i_symbol_ref (sym
, 0);
347 /* Single dot; conflicts with dotted-pair notation. */
348 if (len
== 1 && c
== '.')
351 /* Other initial-character constraints. */
352 if (c
== '\'' || c
== '`' || c
== ',' || c
== '"' || c
== ';' || c
== '#')
355 /* Keywords can be identified by trailing colons too. */
356 if (c
== ':' || scm_i_symbol_ref (sym
, len
- 1) == ':')
357 return quote_keywordish_symbols ();
359 /* Number-ish symbols. */
360 if (scm_is_true (scm_i_string_to_number (scm_symbol_to_string (sym
), 10)))
363 /* Other disallowed first characters. */
364 if (!uc_is_general_category_withtable (c
, INITIAL_IDENTIFIER_MASK
))
367 /* Otherwise, any character that's in the identifier category mask is
368 fine to pass through as-is, provided it's not one of the ASCII
369 delimiters like `;'. */
370 for (pos
= 1; pos
< len
; pos
++)
372 c
= scm_i_symbol_ref (sym
, pos
);
373 if (!uc_is_general_category_withtable (c
, SUBSEQUENT_IDENTIFIER_MASK
))
375 else if (c
== '"' || c
== ';' || c
== '#')
383 print_normal_symbol (SCM sym
, SCM port
)
385 scm_display (scm_symbol_to_string (sym
), port
);
389 print_extended_symbol (SCM sym
, SCM port
)
392 scm_t_string_failed_conversion_handler strategy
;
394 len
= scm_i_symbol_length (sym
);
395 strategy
= scm_i_get_conversion_strategy (port
);
397 scm_lfwrite ("#{", 2, port
);
399 for (pos
= 0; pos
< len
; pos
++)
401 scm_t_wchar c
= scm_i_symbol_ref (sym
, pos
);
403 if (uc_is_general_category_withtable (c
,
404 SUBSEQUENT_IDENTIFIER_MASK
405 | UC_CATEGORY_MASK_Zs
))
407 if (!display_character (c
, port
, strategy
))
408 scm_encoding_error ("print_extended_symbol", errno
,
409 "cannot convert to output locale",
410 port
, SCM_MAKE_CHAR (c
));
414 display_string ("\\x", 1, 2, port
, iconveh_question_mark
);
415 scm_intprint (c
, 16, port
);
416 display_character (';', port
, iconveh_question_mark
);
420 scm_lfwrite ("}#", 2, port
);
423 /* FIXME: allow R6RS hex escapes instead of #{...}#. */
425 scm_i_print_symbol_name (SCM sym
, SCM port
)
427 if (symbol_has_extended_read_syntax (sym
))
428 print_extended_symbol (sym
, port
);
430 print_normal_symbol (sym
, port
);
434 scm_print_symbol_name (const char *str
, size_t len
, SCM port
)
436 SCM symbol
= scm_from_locale_symboln (str
, len
);
437 scm_i_print_symbol_name (symbol
, port
);
440 /* Print generally. Handles both write and display according to PSTATE.
442 SCM_GPROC(s_write
, "write", 1, 1, 0, scm_write
, g_write
);
443 SCM_GPROC(s_display
, "display", 1, 1, 0, scm_display
, g_display
);
445 static void iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
);
448 /* Print a character as an octal or hex escape. */
449 #define PRINT_CHAR_ESCAPE(i, port) \
452 if (!SCM_R6RS_ESCAPES_P) \
453 scm_intprint (i, 8, port); \
456 scm_puts ("x", port); \
457 scm_intprint (i, 16, port); \
464 scm_iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
467 && scm_is_true (scm_memq (exp
, pstate
->highlight_objects
)))
469 scm_display (SCM_PRINT_HIGHLIGHT_PREFIX
, port
);
470 iprin1 (exp
, port
, pstate
);
471 scm_display (SCM_PRINT_HIGHLIGHT_SUFFIX
, port
);
474 iprin1 (exp
, port
, pstate
);
478 iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
480 switch (SCM_ITAG3 (exp
))
484 /* These tc3 tags should never occur in an immediate value. They are
485 * only used in cell types of non-immediates, i. e. the value returned
486 * by SCM_CELL_TYPE (exp) can use these tags.
488 scm_ipruk ("immediate", exp
, port
);
492 scm_intprint (SCM_I_INUM (exp
), 10, port
);
497 if (SCM_WRITINGP (pstate
))
498 write_character (SCM_CHAR (exp
), port
, 0);
501 if (!display_character (SCM_CHAR (exp
), port
,
502 scm_i_get_conversion_strategy (port
)))
503 scm_encoding_error (__func__
, errno
,
504 "cannot convert to output locale",
508 else if (SCM_IFLAGP (exp
)
509 && ((size_t) SCM_IFLAGNUM (exp
) < (sizeof iflagnames
/ sizeof (char *))))
511 scm_puts (iflagnames
[SCM_IFLAGNUM (exp
)], port
);
515 /* unknown immediate value */
516 scm_ipruk ("immediate", exp
, port
);
520 switch (SCM_TYP7 (exp
))
524 ENTER_NESTED_DATA (pstate
, exp
, circref
);
525 if (SCM_OBJ_CLASS_FLAGS (exp
) & SCM_CLASSF_GOOPS
)
527 SCM pwps
, print
= pstate
->writingp
? g_write
: g_display
;
528 if (SCM_UNPACK (print
) == 0)
530 pwps
= scm_i_port_with_print_state (port
, pstate
->handle
);
531 pstate
->revealed
= 1;
532 scm_call_2 (print
, exp
, pwps
);
537 scm_print_struct (exp
, port
, pstate
);
539 EXIT_NESTED_DATA (pstate
);
542 case scm_tcs_cons_imcar
:
543 case scm_tcs_cons_nimcar
:
544 ENTER_NESTED_DATA (pstate
, exp
, circref
);
545 scm_iprlist ("(", exp
, ')', port
, pstate
);
546 EXIT_NESTED_DATA (pstate
);
549 print_circref (port
, pstate
, exp
);
552 switch SCM_TYP16 (exp
) {
554 scm_bigprint (exp
, port
, pstate
);
557 scm_print_real (exp
, port
, pstate
);
559 case scm_tc16_complex
:
560 scm_print_complex (exp
, port
, pstate
);
562 case scm_tc16_fraction
:
563 scm_i_print_fraction (exp
, port
, pstate
);
568 if (SCM_WRITINGP (pstate
))
572 display_character ('"', port
, iconveh_question_mark
);
573 len
= scm_i_string_length (exp
);
574 for (i
= 0; i
< len
; ++i
)
575 write_character (scm_i_string_ref (exp
, i
), port
, 1);
577 display_character ('"', port
, iconveh_question_mark
);
578 scm_remember_upto_here_1 (exp
);
584 len
= scm_i_string_length (exp
);
585 printed
= display_string (scm_i_string_data (exp
),
586 scm_i_is_narrow_string (exp
),
588 scm_i_get_conversion_strategy (port
));
589 if (SCM_UNLIKELY (printed
< len
))
590 scm_encoding_error (__func__
, errno
,
591 "cannot convert to output locale",
592 port
, scm_c_string_ref (exp
, printed
));
595 scm_remember_upto_here_1 (exp
);
598 if (scm_i_symbol_is_interned (exp
))
600 scm_i_print_symbol_name (exp
, port
);
601 scm_remember_upto_here_1 (exp
);
605 scm_puts ("#<uninterned-symbol ", port
);
606 scm_i_print_symbol_name (exp
, port
);
607 scm_putc (' ', port
);
608 scm_uintprint (SCM_UNPACK (exp
), 16, port
);
609 scm_putc ('>', port
);
612 case scm_tc7_variable
:
613 scm_i_variable_print (exp
, port
, pstate
);
615 case scm_tc7_program
:
616 scm_i_program_print (exp
, port
, pstate
);
618 case scm_tc7_pointer
:
619 scm_i_pointer_print (exp
, port
, pstate
);
621 case scm_tc7_hashtable
:
622 scm_i_hashtable_print (exp
, port
, pstate
);
625 scm_i_fluid_print (exp
, port
, pstate
);
627 case scm_tc7_dynamic_state
:
628 scm_i_dynamic_state_print (exp
, port
, pstate
);
631 scm_i_frame_print (exp
, port
, pstate
);
633 case scm_tc7_objcode
:
634 scm_i_objcode_print (exp
, port
, pstate
);
637 scm_i_vm_print (exp
, port
, pstate
);
639 case scm_tc7_vm_cont
:
640 scm_i_vm_cont_print (exp
, port
, pstate
);
643 scm_i_prompt_print (exp
, port
, pstate
);
645 case scm_tc7_with_fluids
:
646 scm_i_with_fluids_print (exp
, port
, pstate
);
649 ENTER_NESTED_DATA (pstate
, exp
, circref
);
650 if (SCM_IS_WHVEC (exp
))
651 scm_puts ("#wh(", port
);
653 scm_puts ("#w(", port
);
654 goto common_vector_printer
;
656 case scm_tc7_bytevector
:
657 scm_i_print_bytevector (exp
, port
, pstate
);
660 ENTER_NESTED_DATA (pstate
, exp
, circref
);
661 scm_puts ("#(", port
);
662 common_vector_printer
:
665 long last
= SCM_SIMPLE_VECTOR_LENGTH (exp
) - 1;
668 && SCM_SIMPLE_VECTOR_LENGTH (exp
) > pstate
->length
)
670 last
= pstate
->length
- 1;
673 if (SCM_I_WVECTP (exp
))
675 /* Elements of weak vectors may not be accessed via the
676 `SIMPLE_VECTOR_REF ()' macro. */
677 for (i
= 0; i
< last
; ++i
)
679 scm_iprin1 (scm_c_vector_ref (exp
, i
),
681 scm_putc (' ', port
);
686 for (i
= 0; i
< last
; ++i
)
688 scm_iprin1 (SCM_SIMPLE_VECTOR_REF (exp
, i
), port
, pstate
);
689 scm_putc (' ', port
);
696 scm_iprin1 (scm_c_vector_ref (exp
, i
), port
, pstate
);
699 scm_puts (" ...", port
);
700 scm_putc (')', port
);
702 EXIT_NESTED_DATA (pstate
);
706 register long i
= SCM_PTOBNUM (exp
);
708 && scm_ptobs
[i
].print
709 && (scm_ptobs
[i
].print
) (exp
, port
, pstate
))
714 ENTER_NESTED_DATA (pstate
, exp
, circref
);
715 SCM_SMOB_DESCRIPTOR (exp
).print (exp
, port
, pstate
);
716 EXIT_NESTED_DATA (pstate
);
719 /* case scm_tcs_closures: */
721 scm_ipruk ("type", exp
, port
);
726 /* Print states are necessary for circular reference safe printing.
727 * They are also expensive to allocate. Therefore print states are
728 * kept in a pool so that they can be reused.
731 /* The PORT argument can also be a print-state/port pair, which will
732 * then be used instead of allocating a new print state. This is
733 * useful for continuing a chain of print calls from Scheme. */
736 scm_prin1 (SCM exp
, SCM port
, int writingp
)
738 SCM handle
= SCM_BOOL_F
; /* Will GC protect the handle whilst unlinked */
740 scm_print_state
*pstate
;
743 /* If PORT is a print-state/port pair, use that. Else create a new
746 if (SCM_PORT_WITH_PS_P (port
))
748 pstate_scm
= SCM_PORT_WITH_PS_PS (port
);
749 port
= SCM_PORT_WITH_PS_PORT (port
);
753 /* First try to allocate a print state from the pool */
754 scm_i_pthread_mutex_lock (&print_state_mutex
);
755 if (!scm_is_null (print_state_pool
))
757 handle
= print_state_pool
;
758 print_state_pool
= SCM_CDR (print_state_pool
);
760 scm_i_pthread_mutex_unlock (&print_state_mutex
);
761 if (scm_is_false (handle
))
762 handle
= scm_list_1 (make_print_state ());
763 pstate_scm
= SCM_CAR (handle
);
766 pstate
= SCM_PRINT_STATE (pstate_scm
);
767 old_writingp
= pstate
->writingp
;
768 pstate
->writingp
= writingp
;
769 scm_iprin1 (exp
, port
, pstate
);
770 pstate
->writingp
= old_writingp
;
772 /* Return print state to pool if it has been created above and
773 hasn't escaped to Scheme. */
775 if (scm_is_true (handle
) && !pstate
->revealed
)
777 scm_i_pthread_mutex_lock (&print_state_mutex
);
778 SCM_SETCDR (handle
, print_state_pool
);
779 print_state_pool
= handle
;
780 scm_i_pthread_mutex_unlock (&print_state_mutex
);
784 /* Convert codepoint CH to UTF-8 and store the result in UTF8. Return
785 the number of bytes of the UTF-8-encoded string. */
787 codepoint_to_utf8 (scm_t_wchar ch
, scm_t_uint8 utf8
[4])
790 scm_t_uint32 codepoint
;
792 codepoint
= (scm_t_uint32
) ch
;
794 if (codepoint
<= 0x7f)
797 utf8
[0] = (scm_t_uint8
) codepoint
;
799 else if (codepoint
<= 0x7ffUL
)
802 utf8
[0] = 0xc0 | (codepoint
>> 6);
803 utf8
[1] = 0x80 | (codepoint
& 0x3f);
805 else if (codepoint
<= 0xffffUL
)
808 utf8
[0] = 0xe0 | (codepoint
>> 12);
809 utf8
[1] = 0x80 | ((codepoint
>> 6) & 0x3f);
810 utf8
[2] = 0x80 | (codepoint
& 0x3f);
815 utf8
[0] = 0xf0 | (codepoint
>> 18);
816 utf8
[1] = 0x80 | ((codepoint
>> 12) & 0x3f);
817 utf8
[2] = 0x80 | ((codepoint
>> 6) & 0x3f);
818 utf8
[3] = 0x80 | (codepoint
& 0x3f);
824 #define STR_REF(s, x) \
826 ? (scm_t_wchar) ((unsigned char *) (s))[x] \
827 : ((scm_t_wchar *) (s))[x])
829 /* Write STR to PORT as UTF-8. STR is a LEN-codepoint string; it is
830 narrow if NARROW_P is true, wide otherwise. Return LEN. */
832 display_string_as_utf8 (const void *str
, int narrow_p
, size_t len
,
837 while (len
> printed
)
840 char *input
, utf8_buf
[256];
842 /* Convert STR to UTF-8. */
843 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
844 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
847 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
848 (scm_t_uint8
*) input
);
849 input
= utf8_buf
+ utf8_len
;
852 /* INPUT was successfully converted, entirely; print the
854 scm_lfwrite (utf8_buf
, utf8_len
, port
);
855 printed
+= i
- printed
;
858 assert (printed
== len
);
863 /* Convert STR through PORT's output conversion descriptor and write the
864 output to PORT. Return the number of codepoints written. */
866 display_string_using_iconv (const void *str
, int narrow_p
, size_t len
,
868 scm_t_string_failed_conversion_handler strategy
)
873 pt
= SCM_PTAB_ENTRY (port
);
877 while (len
> printed
)
879 size_t done
, utf8_len
, input_left
, output_left
, i
;
880 size_t codepoints_read
, output_len
;
881 char *input
, *output
;
882 char utf8_buf
[256], encoded_output
[256];
885 /* Convert STR to UTF-8. */
886 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
887 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
890 offsets
[utf8_len
] = i
;
891 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
892 (scm_t_uint8
*) input
);
893 input
= utf8_buf
+ utf8_len
;
897 input_left
= utf8_len
;
899 output
= encoded_output
;
900 output_left
= sizeof (encoded_output
);
902 done
= iconv (pt
->output_cd
, &input
, &input_left
,
903 &output
, &output_left
);
905 output_len
= sizeof (encoded_output
) - output_left
;
907 if (SCM_UNLIKELY (done
== (size_t) -1))
909 int errno_save
= errno
;
911 /* Reset the `iconv' state. */
912 iconv (pt
->output_cd
, NULL
, NULL
, NULL
, NULL
);
914 /* Print the OUTPUT_LEN bytes successfully converted. */
915 scm_lfwrite (encoded_output
, output_len
, port
);
917 /* See how many input codepoints these OUTPUT_LEN bytes
919 codepoints_read
= offsets
[input
- utf8_buf
] - printed
;
920 printed
+= codepoints_read
;
922 if (errno_save
== EILSEQ
&&
923 strategy
!= SCM_FAILED_CONVERSION_ERROR
)
925 /* Conversion failed somewhere in INPUT and we want to
926 escape or substitute the offending input character. */
928 if (strategy
== SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
)
932 /* Find CH, the offending codepoint, and escape it. */
933 ch
= STR_REF (str
, offsets
[input
- utf8_buf
]);
934 write_character_escaped (ch
, 1, port
);
937 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
938 display_string ("?", 1, 1, port
, strategy
);
943 /* Something bad happened that we can't handle: bail out. */
948 /* INPUT was successfully converted, entirely; print the
950 scm_lfwrite (encoded_output
, output_len
, port
);
951 codepoints_read
= i
- printed
;
952 printed
+= codepoints_read
;
961 /* Display the LEN codepoints in STR to PORT according to STRATEGY;
962 return the number of codepoints successfully displayed. If NARROW_P,
963 then STR is interpreted as a sequence of `char', denoting a Latin-1
964 string; otherwise it's interpreted as a sequence of
967 display_string (const void *str
, int narrow_p
,
968 size_t len
, SCM port
,
969 scm_t_string_failed_conversion_handler strategy
)
974 pt
= SCM_PTAB_ENTRY (port
);
976 if (pt
->output_cd
== (iconv_t
) -1)
977 /* Initialize the conversion descriptors, if needed. */
978 scm_i_set_port_encoding_x (port
, pt
->encoding
);
980 /* FIXME: In 2.1, add a flag to determine whether a port is UTF-8. */
981 if (pt
->output_cd
== (iconv_t
) -1)
982 return display_string_as_utf8 (str
, narrow_p
, len
, port
);
984 return display_string_using_iconv (str
, narrow_p
, len
,
988 /* Attempt to display CH to PORT according to STRATEGY. Return non-zero
989 if CH was successfully displayed, zero otherwise (e.g., if it was not
990 representable in PORT's encoding.) */
992 display_character (scm_t_wchar ch
, SCM port
,
993 scm_t_string_failed_conversion_handler strategy
)
995 return display_string (&ch
, 0, 1, port
, strategy
) == 1;
998 /* Attempt to pretty-print CH, a combining character, to PORT. Return
999 zero upon failure, non-zero otherwise. The idea is to print CH above
1000 a dotted circle to make it more visible. */
1002 write_combining_character (scm_t_wchar ch
, SCM port
)
1006 str
[0] = SCM_CODEPOINT_DOTTED_CIRCLE
;
1009 return display_string (str
, 0, 2, port
, iconveh_error
) == 2;
1012 /* Write CH to PORT in its escaped form, using the string escape syntax
1013 if STRING_ESCAPES_P is non-zero. */
1015 write_character_escaped (scm_t_wchar ch
, int string_escapes_p
, SCM port
)
1017 if (string_escapes_p
)
1019 /* Represent CH using the in-string escape syntax. */
1021 static const char hex
[] = "0123456789abcdef";
1022 static const char escapes
[7] = "abtnvfr";
1025 if (ch
>= 0x07 && ch
<= 0x0D && ch
!= 0x0A)
1027 /* Use special escapes for some C0 controls. */
1029 buf
[1] = escapes
[ch
- 0x07];
1030 scm_lfwrite (buf
, 2, port
);
1032 else if (!SCM_R6RS_ESCAPES_P
)
1038 buf
[2] = hex
[ch
/ 16];
1039 buf
[3] = hex
[ch
% 16];
1040 scm_lfwrite (buf
, 4, port
);
1042 else if (ch
<= 0xFFFF)
1046 buf
[2] = hex
[(ch
& 0xF000) >> 12];
1047 buf
[3] = hex
[(ch
& 0xF00) >> 8];
1048 buf
[4] = hex
[(ch
& 0xF0) >> 4];
1049 buf
[5] = hex
[(ch
& 0xF)];
1050 scm_lfwrite (buf
, 6, port
);
1052 else if (ch
> 0xFFFF)
1056 buf
[2] = hex
[(ch
& 0xF00000) >> 20];
1057 buf
[3] = hex
[(ch
& 0xF0000) >> 16];
1058 buf
[4] = hex
[(ch
& 0xF000) >> 12];
1059 buf
[5] = hex
[(ch
& 0xF00) >> 8];
1060 buf
[6] = hex
[(ch
& 0xF0) >> 4];
1061 buf
[7] = hex
[(ch
& 0xF)];
1062 scm_lfwrite (buf
, 8, port
);
1067 /* Print an R6RS variable-length hex escape: "\xNNNN;". */
1068 scm_t_wchar ch2
= ch
;
1078 buf
[i
] = hex
[ch2
& 0xF];
1085 scm_lfwrite (buf
+ i
, 9 - i
, port
);
1090 /* Represent CH using the character escape syntax. */
1093 name
= scm_i_charname (SCM_MAKE_CHAR (ch
));
1095 scm_puts (name
, port
);
1097 PRINT_CHAR_ESCAPE (ch
, port
);
1101 /* Write CH to PORT, escaping it if it's non-graphic or not
1102 representable in PORT's encoding. If STRING_ESCAPES_P is true and CH
1103 needs to be escaped, it is escaped using the in-string escape syntax;
1104 otherwise the character escape syntax is used. */
1106 write_character (scm_t_wchar ch
, SCM port
, int string_escapes_p
)
1109 scm_t_string_failed_conversion_handler strategy
;
1111 strategy
= scm_i_get_conversion_strategy (port
);
1113 if (string_escapes_p
)
1115 /* Check if CH deserves special treatment. */
1116 if (ch
== '"' || ch
== '\\')
1118 display_character ('\\', port
, iconveh_question_mark
);
1119 display_character (ch
, port
, strategy
);
1122 else if (ch
== ' ' || ch
== '\n')
1124 display_character (ch
, port
, strategy
);
1130 display_string ("#\\", 1, 2, port
, iconveh_question_mark
);
1132 if (uc_combining_class (ch
) != UC_CCC_NR
)
1133 /* Character is a combining character, so attempt to
1135 printed
= write_combining_character (ch
, port
);
1139 && uc_is_general_category_withtable (ch
,
1140 UC_CATEGORY_MASK_L
|
1141 UC_CATEGORY_MASK_M
|
1142 UC_CATEGORY_MASK_N
|
1143 UC_CATEGORY_MASK_P
|
1144 UC_CATEGORY_MASK_S
))
1145 /* CH is graphic; attempt to display it. */
1146 printed
= display_character (ch
, port
, iconveh_error
);
1149 /* CH isn't graphic or cannot be represented in PORT's encoding. */
1150 write_character_escaped (ch
, string_escapes_p
, port
);
1153 /* Print an integer.
1157 scm_intprint (scm_t_intmax n
, int radix
, SCM port
)
1159 char num_buf
[SCM_INTBUFLEN
];
1160 scm_lfwrite (num_buf
, scm_iint2str (n
, radix
, num_buf
), port
);
1164 scm_uintprint (scm_t_uintmax n
, int radix
, SCM port
)
1166 char num_buf
[SCM_INTBUFLEN
];
1167 scm_lfwrite (num_buf
, scm_iuint2str (n
, radix
, num_buf
), port
);
1170 /* Print an object of unrecognized type.
1174 scm_ipruk (char *hdr
, SCM ptr
, SCM port
)
1176 scm_puts ("#<unknown-", port
);
1177 scm_puts (hdr
, port
);
1178 if (1) /* (scm_in_heap_p (ptr)) */ /* FIXME */
1180 scm_puts (" (0x", port
);
1181 scm_uintprint (SCM_CELL_WORD_0 (ptr
), 16, port
);
1182 scm_puts (" . 0x", port
);
1183 scm_uintprint (SCM_CELL_WORD_1 (ptr
), 16, port
);
1184 scm_puts (") @", port
);
1186 scm_puts (" 0x", port
);
1187 scm_uintprint (SCM_UNPACK (ptr
), 16, port
);
1188 scm_putc ('>', port
);
1195 scm_iprlist (char *hdr
, SCM exp
, int tlr
, SCM port
, scm_print_state
*pstate
)
1197 register SCM hare
, tortoise
;
1198 long floor
= pstate
->top
- 2;
1199 scm_puts (hdr
, port
);
1202 goto fancy_printing
;
1204 /* Run a hare and tortoise so that total time complexity will be
1205 O(depth * N) instead of O(N^2). */
1206 hare
= SCM_CDR (exp
);
1208 while (scm_is_pair (hare
))
1210 if (scm_is_eq (hare
, tortoise
))
1211 goto fancy_printing
;
1212 hare
= SCM_CDR (hare
);
1213 if (!scm_is_pair (hare
))
1215 hare
= SCM_CDR (hare
);
1216 tortoise
= SCM_CDR (tortoise
);
1219 /* No cdr cycles intrinsic to this list */
1220 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1221 for (exp
= SCM_CDR (exp
); scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1225 for (i
= floor
; i
>= 0; --i
)
1226 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1228 PUSH_REF (pstate
, exp
);
1229 scm_putc (' ', port
);
1231 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1233 if (!SCM_NULL_OR_NIL_P (exp
))
1235 scm_puts (" . ", port
);
1236 scm_iprin1 (exp
, port
, pstate
);
1240 scm_putc (tlr
, port
);
1241 pstate
->top
= floor
+ 2;
1246 long n
= pstate
->length
;
1248 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1249 exp
= SCM_CDR (exp
); --n
;
1250 for (; scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1252 register unsigned long i
;
1254 for (i
= 0; i
< pstate
->top
; ++i
)
1255 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1261 scm_puts (" ...", port
);
1267 PUSH_REF(pstate
, exp
);
1268 ++pstate
->list_offset
;
1269 scm_putc (' ', port
);
1271 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1274 if (!SCM_NULL_OR_NIL_P (exp
))
1276 scm_puts (" . ", port
);
1277 scm_iprin1 (exp
, port
, pstate
);
1280 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1284 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1287 scm_puts (" . ", port
);
1288 print_circref (port
, pstate
, exp
);
1295 scm_valid_oport_value_p (SCM val
)
1297 return (SCM_OPOUTPORTP (val
)
1298 || (SCM_PORT_WITH_PS_P (val
)
1299 && SCM_OPOUTPORTP (SCM_PORT_WITH_PS_PORT (val
))));
1302 /* SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write); */
1305 scm_write (SCM obj
, SCM port
)
1307 if (SCM_UNBNDP (port
))
1308 port
= scm_current_output_port ();
1310 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_write
);
1312 scm_prin1 (obj
, port
, 1);
1313 return SCM_UNSPECIFIED
;
1317 /* SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display); */
1320 scm_display (SCM obj
, SCM port
)
1322 if (SCM_UNBNDP (port
))
1323 port
= scm_current_output_port ();
1325 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_display
);
1327 scm_prin1 (obj
, port
, 0);
1328 return SCM_UNSPECIFIED
;
1332 SCM_DEFINE (scm_simple_format
, "simple-format", 2, 0, 1,
1333 (SCM destination
, SCM message
, SCM args
),
1334 "Write @var{message} to @var{destination}, defaulting to\n"
1335 "the current output port.\n"
1336 "@var{message} can contain @code{~A} (was @code{%s}) and\n"
1337 "@code{~S} (was @code{%S}) escapes. When printed,\n"
1338 "the escapes are replaced with corresponding members of\n"
1340 "@code{~A} formats using @code{display} and @code{~S} formats\n"
1341 "using @code{write}.\n"
1342 "If @var{destination} is @code{#t}, then use the current output\n"
1343 "port, if @var{destination} is @code{#f}, then return a string\n"
1344 "containing the formatted text. Does not add a trailing newline.")
1345 #define FUNC_NAME s_scm_simple_format
1347 SCM port
, answer
= SCM_UNSPECIFIED
;
1348 int fReturnString
= 0;
1350 size_t start
, p
, end
;
1352 if (scm_is_eq (destination
, SCM_BOOL_T
))
1354 destination
= port
= scm_current_output_port ();
1356 else if (scm_is_false (destination
))
1359 port
= scm_mkstrport (SCM_INUM0
, SCM_BOOL_F
,
1360 SCM_OPN
| SCM_WRTNG
,
1366 SCM_VALIDATE_OPORT_VALUE (1, destination
);
1367 port
= SCM_COERCE_OUTPORT (destination
);
1369 SCM_VALIDATE_STRING (2, message
);
1370 SCM_VALIDATE_REST_ARGUMENT (args
);
1374 end
= scm_i_string_length (message
);
1375 for (p
= start
; p
!= end
; ++p
)
1376 if (scm_i_string_ref (message
, p
) == '~')
1381 switch (scm_i_string_ref (message
, p
))
1390 scm_lfwrite_substr (message
, start
, p
, port
);
1394 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1399 SCM_MISC_ERROR ("FORMAT: Unsupported format option ~~~A - use (ice-9 format) instead",
1400 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1405 if (!scm_is_pair (args
))
1406 SCM_MISC_ERROR ("FORMAT: Missing argument for ~~~A",
1407 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1409 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1410 /* we pass destination here */
1411 scm_prin1 (SCM_CAR (args
), destination
, writingp
);
1412 args
= SCM_CDR (args
);
1416 scm_lfwrite_substr (message
, start
, p
, port
);
1417 if (!scm_is_eq (args
, SCM_EOL
))
1418 SCM_MISC_ERROR ("FORMAT: ~A superfluous arguments",
1419 scm_list_1 (scm_length (args
)));
1422 answer
= scm_strport_to_string (destination
);
1424 return scm_return_first (answer
, message
);
1429 SCM_DEFINE (scm_newline
, "newline", 0, 1, 0,
1431 "Send a newline to @var{port}.\n"
1432 "If @var{port} is omitted, send to the current output port.")
1433 #define FUNC_NAME s_scm_newline
1435 if (SCM_UNBNDP (port
))
1436 port
= scm_current_output_port ();
1438 SCM_VALIDATE_OPORT_VALUE (1, port
);
1440 scm_putc ('\n', SCM_COERCE_OUTPORT (port
));
1441 return SCM_UNSPECIFIED
;
1445 SCM_DEFINE (scm_write_char
, "write-char", 1, 1, 0,
1446 (SCM chr
, SCM port
),
1447 "Send character @var{chr} to @var{port}.")
1448 #define FUNC_NAME s_scm_write_char
1450 if (SCM_UNBNDP (port
))
1451 port
= scm_current_output_port ();
1453 SCM_VALIDATE_CHAR (1, chr
);
1454 SCM_VALIDATE_OPORT_VALUE (2, port
);
1456 port
= SCM_COERCE_OUTPORT (port
);
1457 if (!display_character (SCM_CHAR (chr
), port
,
1458 scm_i_get_conversion_strategy (port
)))
1459 scm_encoding_error (__func__
, errno
,
1460 "cannot convert to output locale",
1463 return SCM_UNSPECIFIED
;
1469 /* Call back to Scheme code to do the printing of special objects
1470 * (like structs). SCM_PRINTER_APPLY applies PROC to EXP and a smob
1471 * containing PORT and PSTATE. This object can be used as the port for
1472 * display/write etc to continue the current print chain. The REVEALED
1473 * field of PSTATE is set to true to indicate that the print state has
1474 * escaped to Scheme and thus has to be freed by the GC.
1477 scm_t_bits scm_tc16_port_with_ps
;
1479 /* Print exactly as the port itself would */
1482 port_with_ps_print (SCM obj
, SCM port
, scm_print_state
*pstate
)
1484 obj
= SCM_PORT_WITH_PS_PORT (obj
);
1485 return scm_ptobs
[SCM_PTOBNUM (obj
)].print (obj
, port
, pstate
);
1489 scm_printer_apply (SCM proc
, SCM exp
, SCM port
, scm_print_state
*pstate
)
1491 pstate
->revealed
= 1;
1492 return scm_call_2 (proc
, exp
,
1493 scm_i_port_with_print_state (port
, pstate
->handle
));
1496 SCM_DEFINE (scm_port_with_print_state
, "port-with-print-state", 1, 1, 0,
1497 (SCM port
, SCM pstate
),
1498 "Create a new port which behaves like @var{port}, but with an\n"
1499 "included print state @var{pstate}. @var{pstate} is optional.\n"
1500 "If @var{pstate} isn't supplied and @var{port} already has\n"
1501 "a print state, the old print state is reused.")
1502 #define FUNC_NAME s_scm_port_with_print_state
1504 SCM_VALIDATE_OPORT_VALUE (1, port
);
1505 if (!SCM_UNBNDP (pstate
))
1506 SCM_VALIDATE_PRINTSTATE (2, pstate
);
1507 return scm_i_port_with_print_state (port
, pstate
);
1511 SCM_DEFINE (scm_get_print_state
, "get-print-state", 1, 0, 0,
1513 "Return the print state of the port @var{port}. If @var{port}\n"
1514 "has no associated print state, @code{#f} is returned.")
1515 #define FUNC_NAME s_scm_get_print_state
1517 if (SCM_PORT_WITH_PS_P (port
))
1518 return SCM_PORT_WITH_PS_PS (port
);
1519 if (SCM_OUTPUT_PORT_P (port
))
1521 SCM_WRONG_TYPE_ARG (1, port
);
1530 SCM vtable
, layout
, type
;
1532 scm_init_opts (scm_print_options
, scm_print_opts
);
1534 scm_print_options (scm_list_4 (scm_from_latin1_symbol ("highlight-prefix"),
1535 scm_from_locale_string ("{"),
1536 scm_from_latin1_symbol ("highlight-suffix"),
1537 scm_from_locale_string ("}")));
1539 scm_gc_register_root (&print_state_pool
);
1540 scm_gc_register_root (&scm_print_state_vtable
);
1541 vtable
= scm_make_vtable_vtable (scm_nullstr
, SCM_INUM0
, SCM_EOL
);
1543 scm_make_struct_layout (scm_from_locale_string (SCM_PRINT_STATE_LAYOUT
));
1544 type
= scm_make_struct (vtable
, SCM_INUM0
, scm_list_1 (layout
));
1545 scm_set_struct_vtable_name_x (type
, scm_from_latin1_symbol ("print-state"));
1546 scm_print_state_vtable
= type
;
1548 /* Don't want to bind a wrapper class in GOOPS, so pass 0 as arg1. */
1549 scm_tc16_port_with_ps
= scm_make_smob_type (0, 0);
1550 scm_set_smob_print (scm_tc16_port_with_ps
, port_with_ps_print
);
1552 #include "libguile/print.x"
1554 scm_print_opts
[SCM_PRINT_KEYWORD_STYLE_I
].val
= SCM_UNPACK (sym_reader
);