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/programs.h"
44 #include "libguile/alist.h"
45 #include "libguile/struct.h"
46 #include "libguile/ports.h"
47 #include "libguile/root.h"
48 #include "libguile/strings.h"
49 #include "libguile/strports.h"
50 #include "libguile/vectors.h"
51 #include "libguile/numbers.h"
52 #include "libguile/vm.h"
54 #include "libguile/validate.h"
55 #include "libguile/print.h"
57 #include "libguile/private-options.h"
61 /* Character printers. */
63 static size_t display_string (const void *, int, size_t, SCM
,
64 scm_t_string_failed_conversion_handler
);
66 static int display_character (scm_t_wchar
, SCM
,
67 scm_t_string_failed_conversion_handler
);
69 static void write_character (scm_t_wchar
, SCM
, int);
71 static void write_character_escaped (scm_t_wchar
, int, SCM
);
75 /* {Names of immediate symbols}
77 * This table must agree with the declarations in scm.h: {Immediate Symbols}.
80 /* This table must agree with the list of flags in tags.h. */
81 static const char *iflagnames
[] =
84 "#nil", /* Elisp nil value. Should print from elisp as symbol `nil'. */
85 "#<XXX UNUSED LISP FALSE -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
88 "#<XXX UNUSED BOOLEAN 0 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
89 "#<XXX UNUSED BOOLEAN 1 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
90 "#<XXX UNUSED BOOLEAN 2 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
95 /* Unbound slot marker for GOOPS. For internal use in GOOPS only. */
99 SCM_SYMBOL (sym_reader
, "reader");
101 scm_t_option scm_print_opts
[] = {
102 { SCM_OPTION_SCM
, "highlight-prefix", (scm_t_bits
)SCM_BOOL_F_BITS
,
103 "The string to print before highlighted values." },
104 { SCM_OPTION_SCM
, "highlight-suffix", (scm_t_bits
)SCM_BOOL_F_BITS
,
105 "The string to print after highlighted values." },
106 { SCM_OPTION_SCM
, "quote-keywordish-symbols", (scm_t_bits
)SCM_BOOL_F_BITS
,
107 "How to print symbols that have a colon as their first or last character. "
108 "The value '#f' does not quote the colons; '#t' quotes them; "
109 "'reader' quotes them when the reader option 'keywords' is not '#f'."
114 SCM_DEFINE (scm_print_options
, "print-options-interface", 0, 1, 0,
116 "Option interface for the print options. Instead of using\n"
117 "this procedure directly, use the procedures\n"
118 "@code{print-enable}, @code{print-disable}, @code{print-set!}\n"
119 "and @code{print-options}.")
120 #define FUNC_NAME s_scm_print_options
122 SCM ans
= scm_options (setting
,
130 /* {Printing of Scheme Objects}
133 /* Detection of circular references.
135 * Due to other constraints in the implementation, this code has bad
136 * time complexity (O (depth * N)), The printer code can be
137 * rewritten to be O(N).
139 #define PUSH_REF(pstate, obj) \
142 PSTATE_STACK_SET (pstate, pstate->top, obj); \
144 if (pstate->top == pstate->ceiling) \
145 grow_ref_stack (pstate); \
148 #define ENTER_NESTED_DATA(pstate, obj, label) \
151 register unsigned long i; \
152 for (i = 0; i < pstate->top; ++i) \
153 if (scm_is_eq (PSTATE_STACK_REF (pstate, i), (obj))) \
155 if (pstate->fancyp) \
157 if (pstate->top - pstate->list_offset >= pstate->level) \
159 scm_putc ('#', port); \
163 PUSH_REF(pstate, obj); \
166 #define EXIT_NESTED_DATA(pstate) \
170 PSTATE_STACK_SET (pstate, pstate->top, SCM_UNDEFINED); \
174 SCM scm_print_state_vtable
= SCM_BOOL_F
;
175 static SCM print_state_pool
= SCM_EOL
;
176 scm_i_pthread_mutex_t print_state_mutex
= SCM_I_PTHREAD_MUTEX_INITIALIZER
;
178 #ifdef GUILE_DEBUG /* Used for debugging purposes */
180 SCM_DEFINE (scm_current_pstate
, "current-pstate", 0, 0, 0,
182 "Return the current-pstate -- the car of the\n"
183 "@code{print_state_pool}. @code{current-pstate} is only\n"
184 "included in @code{--enable-guile-debug} builds.")
185 #define FUNC_NAME s_scm_current_pstate
187 if (!scm_is_null (print_state_pool
))
188 return SCM_CAR (print_state_pool
);
196 #define PSTATE_SIZE 50L
199 make_print_state (void)
202 = scm_make_struct (scm_print_state_vtable
, SCM_INUM0
, SCM_EOL
);
203 scm_print_state
*pstate
= SCM_PRINT_STATE (print_state
);
204 pstate
->ref_vect
= scm_c_make_vector (PSTATE_SIZE
, SCM_UNDEFINED
);
205 pstate
->ceiling
= SCM_SIMPLE_VECTOR_LENGTH (pstate
->ref_vect
);
206 pstate
->highlight_objects
= SCM_EOL
;
211 scm_make_print_state ()
213 SCM answer
= SCM_BOOL_F
;
215 /* First try to allocate a print state from the pool */
216 scm_i_pthread_mutex_lock (&print_state_mutex
);
217 if (!scm_is_null (print_state_pool
))
219 answer
= SCM_CAR (print_state_pool
);
220 print_state_pool
= SCM_CDR (print_state_pool
);
222 scm_i_pthread_mutex_unlock (&print_state_mutex
);
224 return scm_is_false (answer
) ? make_print_state () : answer
;
228 scm_free_print_state (SCM print_state
)
231 scm_print_state
*pstate
= SCM_PRINT_STATE (print_state
);
232 /* Cleanup before returning print state to pool.
233 * It is better to do it here. Doing it in scm_prin1
234 * would cost more since that function is called much more
238 pstate
->revealed
= 0;
239 pstate
->highlight_objects
= SCM_EOL
;
240 scm_i_pthread_mutex_lock (&print_state_mutex
);
241 handle
= scm_cons (print_state
, print_state_pool
);
242 print_state_pool
= handle
;
243 scm_i_pthread_mutex_unlock (&print_state_mutex
);
247 scm_i_port_with_print_state (SCM port
, SCM print_state
)
249 if (SCM_UNBNDP (print_state
))
251 if (SCM_PORT_WITH_PS_P (port
))
254 print_state
= scm_make_print_state ();
255 /* port does not need to be coerced since it doesn't have ps */
258 port
= SCM_COERCE_OUTPORT (port
);
259 SCM_RETURN_NEWSMOB (scm_tc16_port_with_ps
,
260 SCM_UNPACK (scm_cons (port
, print_state
)));
264 grow_ref_stack (scm_print_state
*pstate
)
266 SCM old_vect
= pstate
->ref_vect
;
267 size_t old_size
= SCM_SIMPLE_VECTOR_LENGTH (old_vect
);
268 size_t new_size
= 2 * pstate
->ceiling
;
269 SCM new_vect
= scm_c_make_vector (new_size
, SCM_UNDEFINED
);
272 for (i
= 0; i
!= old_size
; ++i
)
273 SCM_SIMPLE_VECTOR_SET (new_vect
, i
, SCM_SIMPLE_VECTOR_REF (old_vect
, i
));
275 pstate
->ref_vect
= new_vect
;
276 pstate
->ceiling
= new_size
;
279 #define PSTATE_STACK_REF(p,i) SCM_SIMPLE_VECTOR_REF((p)->ref_vect, (i))
280 #define PSTATE_STACK_SET(p,i,v) SCM_SIMPLE_VECTOR_SET((p)->ref_vect, (i), (v))
283 print_circref (SCM port
, scm_print_state
*pstate
, SCM ref
)
286 long self
= pstate
->top
- 1;
288 if (scm_is_pair (PSTATE_STACK_REF (pstate
, i
)))
292 if (!scm_is_pair (PSTATE_STACK_REF (pstate
, i
-1))
293 || !scm_is_eq (SCM_CDR (PSTATE_STACK_REF (pstate
, i
-1)),
294 SCM_CDR (PSTATE_STACK_REF (pstate
, i
))))
300 for (i
= pstate
->top
- 1; 1; --i
)
301 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), ref
))
303 scm_putc ('#', port
);
304 scm_intprint (i
- self
, 10, port
);
305 scm_putc ('#', port
);
308 /* Print the name of a symbol. */
311 quote_keywordish_symbols (void)
313 SCM option
= SCM_PRINT_KEYWORD_STYLE
;
315 if (scm_is_false (option
))
317 if (scm_is_eq (option
, sym_reader
))
318 return scm_is_true (SCM_PACK (SCM_KEYWORD_STYLE
));
322 #define INITIAL_IDENTIFIER_MASK \
323 (UC_CATEGORY_MASK_Lu | UC_CATEGORY_MASK_Ll | UC_CATEGORY_MASK_Lt \
324 | UC_CATEGORY_MASK_Lm | UC_CATEGORY_MASK_Lo | UC_CATEGORY_MASK_Mn \
325 | UC_CATEGORY_MASK_Nl | UC_CATEGORY_MASK_No | UC_CATEGORY_MASK_Pd \
326 | UC_CATEGORY_MASK_Pc | UC_CATEGORY_MASK_Po | UC_CATEGORY_MASK_Sc \
327 | UC_CATEGORY_MASK_Sm | UC_CATEGORY_MASK_Sk | UC_CATEGORY_MASK_So \
328 | UC_CATEGORY_MASK_Co)
330 #define SUBSEQUENT_IDENTIFIER_MASK \
331 (INITIAL_IDENTIFIER_MASK \
332 | UC_CATEGORY_MASK_Nd | UC_CATEGORY_MASK_Mc | UC_CATEGORY_MASK_Me)
335 symbol_has_extended_read_syntax (SCM sym
)
337 size_t pos
, len
= scm_i_symbol_length (sym
);
340 /* The empty symbol. */
344 c
= scm_i_symbol_ref (sym
, 0);
346 /* Single dot; conflicts with dotted-pair notation. */
347 if (len
== 1 && c
== '.')
350 /* Other initial-character constraints. */
351 if (c
== '\'' || c
== '`' || c
== ',' || c
== '"' || c
== ';' || c
== '#')
354 /* Keywords can be identified by trailing colons too. */
355 if (c
== ':' || scm_i_symbol_ref (sym
, len
- 1) == ':')
356 return quote_keywordish_symbols ();
358 /* Number-ish symbols. */
359 if (scm_is_true (scm_i_string_to_number (scm_symbol_to_string (sym
), 10)))
362 /* Other disallowed first characters. */
363 if (!uc_is_general_category_withtable (c
, INITIAL_IDENTIFIER_MASK
))
366 /* Otherwise, any character that's in the identifier category mask is
367 fine to pass through as-is, provided it's not one of the ASCII
368 delimiters like `;'. */
369 for (pos
= 1; pos
< len
; pos
++)
371 c
= scm_i_symbol_ref (sym
, pos
);
372 if (!uc_is_general_category_withtable (c
, SUBSEQUENT_IDENTIFIER_MASK
))
374 else if (c
== '"' || c
== ';' || c
== '#')
382 print_normal_symbol (SCM sym
, SCM port
)
384 scm_display (scm_symbol_to_string (sym
), port
);
388 print_extended_symbol (SCM sym
, SCM port
)
391 scm_t_string_failed_conversion_handler strategy
;
393 len
= scm_i_symbol_length (sym
);
394 strategy
= scm_i_get_conversion_strategy (port
);
396 scm_lfwrite ("#{", 2, port
);
398 for (pos
= 0; pos
< len
; pos
++)
400 scm_t_wchar c
= scm_i_symbol_ref (sym
, pos
);
402 if (uc_is_general_category_withtable (c
,
403 SUBSEQUENT_IDENTIFIER_MASK
404 | UC_CATEGORY_MASK_Zs
))
406 if (!display_character (c
, port
, strategy
))
407 scm_encoding_error ("print_extended_symbol", errno
,
408 "cannot convert to output locale",
409 port
, SCM_MAKE_CHAR (c
));
413 display_string ("\\x", 1, 2, port
, iconveh_question_mark
);
414 scm_intprint (c
, 16, port
);
415 display_character (';', port
, iconveh_question_mark
);
419 scm_lfwrite ("}#", 2, port
);
422 /* FIXME: allow R6RS hex escapes instead of #{...}#. */
424 scm_i_print_symbol_name (SCM sym
, SCM port
)
426 if (symbol_has_extended_read_syntax (sym
))
427 print_extended_symbol (sym
, port
);
429 print_normal_symbol (sym
, port
);
433 scm_print_symbol_name (const char *str
, size_t len
, SCM port
)
435 SCM symbol
= scm_from_locale_symboln (str
, len
);
436 scm_i_print_symbol_name (symbol
, port
);
439 /* Print generally. Handles both write and display according to PSTATE.
441 SCM_GPROC(s_write
, "write", 1, 1, 0, scm_write
, g_write
);
442 SCM_GPROC(s_display
, "display", 1, 1, 0, scm_display
, g_display
);
444 static void iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
);
447 /* Print a character as an octal or hex escape. */
448 #define PRINT_CHAR_ESCAPE(i, port) \
451 if (!SCM_R6RS_ESCAPES_P) \
452 scm_intprint (i, 8, port); \
455 scm_puts ("x", port); \
456 scm_intprint (i, 16, port); \
463 scm_iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
466 && scm_is_true (scm_memq (exp
, pstate
->highlight_objects
)))
468 scm_display (SCM_PRINT_HIGHLIGHT_PREFIX
, port
);
469 iprin1 (exp
, port
, pstate
);
470 scm_display (SCM_PRINT_HIGHLIGHT_SUFFIX
, port
);
473 iprin1 (exp
, port
, pstate
);
477 iprin1 (SCM exp
, SCM port
, scm_print_state
*pstate
)
479 switch (SCM_ITAG3 (exp
))
483 /* These tc3 tags should never occur in an immediate value. They are
484 * only used in cell types of non-immediates, i. e. the value returned
485 * by SCM_CELL_TYPE (exp) can use these tags.
487 scm_ipruk ("immediate", exp
, port
);
491 scm_intprint (SCM_I_INUM (exp
), 10, port
);
496 if (SCM_WRITINGP (pstate
))
497 write_character (SCM_CHAR (exp
), port
, 0);
500 if (!display_character (SCM_CHAR (exp
), port
,
501 scm_i_get_conversion_strategy (port
)))
502 scm_encoding_error (__func__
, errno
,
503 "cannot convert to output locale",
507 else if (SCM_IFLAGP (exp
)
508 && ((size_t) SCM_IFLAGNUM (exp
) < (sizeof iflagnames
/ sizeof (char *))))
510 scm_puts (iflagnames
[SCM_IFLAGNUM (exp
)], port
);
514 /* unknown immediate value */
515 scm_ipruk ("immediate", exp
, port
);
519 switch (SCM_TYP7 (exp
))
523 ENTER_NESTED_DATA (pstate
, exp
, circref
);
524 if (SCM_OBJ_CLASS_FLAGS (exp
) & SCM_CLASSF_GOOPS
)
526 SCM pwps
, print
= pstate
->writingp
? g_write
: g_display
;
527 if (SCM_UNPACK (print
) == 0)
529 pwps
= scm_i_port_with_print_state (port
, pstate
->handle
);
530 pstate
->revealed
= 1;
531 scm_call_2 (print
, exp
, pwps
);
536 scm_print_struct (exp
, port
, pstate
);
538 EXIT_NESTED_DATA (pstate
);
541 case scm_tcs_cons_imcar
:
542 case scm_tcs_cons_nimcar
:
543 ENTER_NESTED_DATA (pstate
, exp
, circref
);
544 scm_iprlist ("(", exp
, ')', port
, pstate
);
545 EXIT_NESTED_DATA (pstate
);
548 print_circref (port
, pstate
, exp
);
551 switch SCM_TYP16 (exp
) {
553 scm_bigprint (exp
, port
, pstate
);
556 scm_print_real (exp
, port
, pstate
);
558 case scm_tc16_complex
:
559 scm_print_complex (exp
, port
, pstate
);
561 case scm_tc16_fraction
:
562 scm_i_print_fraction (exp
, port
, pstate
);
567 if (SCM_WRITINGP (pstate
))
571 display_character ('"', port
, iconveh_question_mark
);
572 len
= scm_i_string_length (exp
);
573 for (i
= 0; i
< len
; ++i
)
574 write_character (scm_i_string_ref (exp
, i
), port
, 1);
576 display_character ('"', port
, iconveh_question_mark
);
577 scm_remember_upto_here_1 (exp
);
583 len
= scm_i_string_length (exp
);
584 printed
= display_string (scm_i_string_data (exp
),
585 scm_i_is_narrow_string (exp
),
587 scm_i_get_conversion_strategy (port
));
588 if (SCM_UNLIKELY (printed
< len
))
589 scm_encoding_error (__func__
, errno
,
590 "cannot convert to output locale",
591 port
, scm_c_string_ref (exp
, printed
));
594 scm_remember_upto_here_1 (exp
);
597 if (scm_i_symbol_is_interned (exp
))
599 scm_i_print_symbol_name (exp
, port
);
600 scm_remember_upto_here_1 (exp
);
604 scm_puts ("#<uninterned-symbol ", port
);
605 scm_i_print_symbol_name (exp
, port
);
606 scm_putc (' ', port
);
607 scm_uintprint (SCM_UNPACK (exp
), 16, port
);
608 scm_putc ('>', port
);
611 case scm_tc7_variable
:
612 scm_i_variable_print (exp
, port
, pstate
);
614 case scm_tc7_program
:
615 scm_i_program_print (exp
, port
, pstate
);
617 case scm_tc7_pointer
:
618 scm_i_pointer_print (exp
, port
, pstate
);
620 case scm_tc7_hashtable
:
621 scm_i_hashtable_print (exp
, port
, pstate
);
623 case scm_tc7_weak_set
:
624 scm_i_weak_set_print (exp
, port
, pstate
);
626 case scm_tc7_weak_table
:
627 scm_i_weak_table_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_puts ("#w(", port
);
656 goto common_vector_printer
;
658 case scm_tc7_bytevector
:
659 scm_i_print_bytevector (exp
, port
, pstate
);
662 ENTER_NESTED_DATA (pstate
, exp
, circref
);
663 scm_puts ("#(", port
);
664 common_vector_printer
:
667 long last
= SCM_SIMPLE_VECTOR_LENGTH (exp
) - 1;
670 && SCM_SIMPLE_VECTOR_LENGTH (exp
) > pstate
->length
)
672 last
= pstate
->length
- 1;
675 for (i
= 0; i
< last
; ++i
)
677 scm_iprin1 (scm_c_vector_ref (exp
, i
), port
, pstate
);
678 scm_putc (' ', port
);
683 scm_iprin1 (scm_c_vector_ref (exp
, i
), port
, pstate
);
686 scm_puts (" ...", port
);
687 scm_putc (')', port
);
689 EXIT_NESTED_DATA (pstate
);
693 register long i
= SCM_PTOBNUM (exp
);
695 && scm_ptobs
[i
].print
696 && (scm_ptobs
[i
].print
) (exp
, port
, pstate
))
701 ENTER_NESTED_DATA (pstate
, exp
, circref
);
702 SCM_SMOB_DESCRIPTOR (exp
).print (exp
, port
, pstate
);
703 EXIT_NESTED_DATA (pstate
);
706 /* case scm_tcs_closures: */
708 scm_ipruk ("type", exp
, port
);
713 /* Print states are necessary for circular reference safe printing.
714 * They are also expensive to allocate. Therefore print states are
715 * kept in a pool so that they can be reused.
718 /* The PORT argument can also be a print-state/port pair, which will
719 * then be used instead of allocating a new print state. This is
720 * useful for continuing a chain of print calls from Scheme. */
723 scm_prin1 (SCM exp
, SCM port
, int writingp
)
725 SCM handle
= SCM_BOOL_F
; /* Will GC protect the handle whilst unlinked */
727 scm_print_state
*pstate
;
730 /* If PORT is a print-state/port pair, use that. Else create a new
733 if (SCM_PORT_WITH_PS_P (port
))
735 pstate_scm
= SCM_PORT_WITH_PS_PS (port
);
736 port
= SCM_PORT_WITH_PS_PORT (port
);
740 /* First try to allocate a print state from the pool */
741 scm_i_pthread_mutex_lock (&print_state_mutex
);
742 if (!scm_is_null (print_state_pool
))
744 handle
= print_state_pool
;
745 print_state_pool
= SCM_CDR (print_state_pool
);
747 scm_i_pthread_mutex_unlock (&print_state_mutex
);
748 if (scm_is_false (handle
))
749 handle
= scm_list_1 (make_print_state ());
750 pstate_scm
= SCM_CAR (handle
);
753 pstate
= SCM_PRINT_STATE (pstate_scm
);
754 old_writingp
= pstate
->writingp
;
755 pstate
->writingp
= writingp
;
756 scm_iprin1 (exp
, port
, pstate
);
757 pstate
->writingp
= old_writingp
;
759 /* Return print state to pool if it has been created above and
760 hasn't escaped to Scheme. */
762 if (scm_is_true (handle
) && !pstate
->revealed
)
764 scm_i_pthread_mutex_lock (&print_state_mutex
);
765 SCM_SETCDR (handle
, print_state_pool
);
766 print_state_pool
= handle
;
767 scm_i_pthread_mutex_unlock (&print_state_mutex
);
771 /* Convert codepoint CH to UTF-8 and store the result in UTF8. Return
772 the number of bytes of the UTF-8-encoded string. */
774 codepoint_to_utf8 (scm_t_wchar ch
, scm_t_uint8 utf8
[4])
777 scm_t_uint32 codepoint
;
779 codepoint
= (scm_t_uint32
) ch
;
781 if (codepoint
<= 0x7f)
784 utf8
[0] = (scm_t_uint8
) codepoint
;
786 else if (codepoint
<= 0x7ffUL
)
789 utf8
[0] = 0xc0 | (codepoint
>> 6);
790 utf8
[1] = 0x80 | (codepoint
& 0x3f);
792 else if (codepoint
<= 0xffffUL
)
795 utf8
[0] = 0xe0 | (codepoint
>> 12);
796 utf8
[1] = 0x80 | ((codepoint
>> 6) & 0x3f);
797 utf8
[2] = 0x80 | (codepoint
& 0x3f);
802 utf8
[0] = 0xf0 | (codepoint
>> 18);
803 utf8
[1] = 0x80 | ((codepoint
>> 12) & 0x3f);
804 utf8
[2] = 0x80 | ((codepoint
>> 6) & 0x3f);
805 utf8
[3] = 0x80 | (codepoint
& 0x3f);
811 #define STR_REF(s, x) \
813 ? (scm_t_wchar) ((unsigned char *) (s))[x] \
814 : ((scm_t_wchar *) (s))[x])
816 /* Write STR to PORT as UTF-8. STR is a LEN-codepoint string; it is
817 narrow if NARROW_P is true, wide otherwise. Return LEN. */
819 display_string_as_utf8 (const void *str
, int narrow_p
, size_t len
,
824 while (len
> printed
)
827 char *input
, utf8_buf
[256];
829 /* Convert STR to UTF-8. */
830 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
831 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
834 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
835 (scm_t_uint8
*) input
);
836 input
= utf8_buf
+ utf8_len
;
839 /* INPUT was successfully converted, entirely; print the
841 scm_lfwrite (utf8_buf
, utf8_len
, port
);
842 printed
+= i
- printed
;
845 assert (printed
== len
);
850 /* Convert STR through PORT's output conversion descriptor and write the
851 output to PORT. Return the number of codepoints written. */
853 display_string_using_iconv (const void *str
, int narrow_p
, size_t len
,
855 scm_t_string_failed_conversion_handler strategy
)
860 pt
= SCM_PTAB_ENTRY (port
);
864 while (len
> printed
)
866 size_t done
, utf8_len
, input_left
, output_left
, i
;
867 size_t codepoints_read
, output_len
;
868 char *input
, *output
;
869 char utf8_buf
[256], encoded_output
[256];
872 /* Convert STR to UTF-8. */
873 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
874 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
877 offsets
[utf8_len
] = i
;
878 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
879 (scm_t_uint8
*) input
);
880 input
= utf8_buf
+ utf8_len
;
884 input_left
= utf8_len
;
886 output
= encoded_output
;
887 output_left
= sizeof (encoded_output
);
889 done
= iconv (pt
->output_cd
, &input
, &input_left
,
890 &output
, &output_left
);
892 output_len
= sizeof (encoded_output
) - output_left
;
894 if (SCM_UNLIKELY (done
== (size_t) -1))
896 int errno_save
= errno
;
898 /* Reset the `iconv' state. */
899 iconv (pt
->output_cd
, NULL
, NULL
, NULL
, NULL
);
901 /* Print the OUTPUT_LEN bytes successfully converted. */
902 scm_lfwrite (encoded_output
, output_len
, port
);
904 /* See how many input codepoints these OUTPUT_LEN bytes
906 codepoints_read
= offsets
[input
- utf8_buf
] - printed
;
907 printed
+= codepoints_read
;
909 if (errno_save
== EILSEQ
&&
910 strategy
!= SCM_FAILED_CONVERSION_ERROR
)
912 /* Conversion failed somewhere in INPUT and we want to
913 escape or substitute the offending input character. */
915 if (strategy
== SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
)
919 /* Find CH, the offending codepoint, and escape it. */
920 ch
= STR_REF (str
, offsets
[input
- utf8_buf
]);
921 write_character_escaped (ch
, 1, port
);
924 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
925 display_string ("?", 1, 1, port
, strategy
);
930 /* Something bad happened that we can't handle: bail out. */
935 /* INPUT was successfully converted, entirely; print the
937 scm_lfwrite (encoded_output
, output_len
, port
);
938 codepoints_read
= i
- printed
;
939 printed
+= codepoints_read
;
948 /* Display the LEN codepoints in STR to PORT according to STRATEGY;
949 return the number of codepoints successfully displayed. If NARROW_P,
950 then STR is interpreted as a sequence of `char', denoting a Latin-1
951 string; otherwise it's interpreted as a sequence of
954 display_string (const void *str
, int narrow_p
,
955 size_t len
, SCM port
,
956 scm_t_string_failed_conversion_handler strategy
)
961 pt
= SCM_PTAB_ENTRY (port
);
963 if (pt
->output_cd
== (iconv_t
) -1)
964 /* Initialize the conversion descriptors, if needed. */
965 scm_i_set_port_encoding_x (port
, pt
->encoding
);
967 /* FIXME: In 2.1, add a flag to determine whether a port is UTF-8. */
968 if (pt
->output_cd
== (iconv_t
) -1)
969 return display_string_as_utf8 (str
, narrow_p
, len
, port
);
971 return display_string_using_iconv (str
, narrow_p
, len
,
975 /* Attempt to display CH to PORT according to STRATEGY. Return non-zero
976 if CH was successfully displayed, zero otherwise (e.g., if it was not
977 representable in PORT's encoding.) */
979 display_character (scm_t_wchar ch
, SCM port
,
980 scm_t_string_failed_conversion_handler strategy
)
982 return display_string (&ch
, 0, 1, port
, strategy
) == 1;
985 /* Attempt to pretty-print CH, a combining character, to PORT. Return
986 zero upon failure, non-zero otherwise. The idea is to print CH above
987 a dotted circle to make it more visible. */
989 write_combining_character (scm_t_wchar ch
, SCM port
)
993 str
[0] = SCM_CODEPOINT_DOTTED_CIRCLE
;
996 return display_string (str
, 0, 2, port
, iconveh_error
) == 2;
999 /* Write CH to PORT in its escaped form, using the string escape syntax
1000 if STRING_ESCAPES_P is non-zero. */
1002 write_character_escaped (scm_t_wchar ch
, int string_escapes_p
, SCM port
)
1004 if (string_escapes_p
)
1006 /* Represent CH using the in-string escape syntax. */
1008 static const char hex
[] = "0123456789abcdef";
1009 static const char escapes
[7] = "abtnvfr";
1012 if (ch
>= 0x07 && ch
<= 0x0D && ch
!= 0x0A)
1014 /* Use special escapes for some C0 controls. */
1016 buf
[1] = escapes
[ch
- 0x07];
1017 scm_lfwrite (buf
, 2, port
);
1019 else if (!SCM_R6RS_ESCAPES_P
)
1025 buf
[2] = hex
[ch
/ 16];
1026 buf
[3] = hex
[ch
% 16];
1027 scm_lfwrite (buf
, 4, port
);
1029 else if (ch
<= 0xFFFF)
1033 buf
[2] = hex
[(ch
& 0xF000) >> 12];
1034 buf
[3] = hex
[(ch
& 0xF00) >> 8];
1035 buf
[4] = hex
[(ch
& 0xF0) >> 4];
1036 buf
[5] = hex
[(ch
& 0xF)];
1037 scm_lfwrite (buf
, 6, port
);
1039 else if (ch
> 0xFFFF)
1043 buf
[2] = hex
[(ch
& 0xF00000) >> 20];
1044 buf
[3] = hex
[(ch
& 0xF0000) >> 16];
1045 buf
[4] = hex
[(ch
& 0xF000) >> 12];
1046 buf
[5] = hex
[(ch
& 0xF00) >> 8];
1047 buf
[6] = hex
[(ch
& 0xF0) >> 4];
1048 buf
[7] = hex
[(ch
& 0xF)];
1049 scm_lfwrite (buf
, 8, port
);
1054 /* Print an R6RS variable-length hex escape: "\xNNNN;". */
1055 scm_t_wchar ch2
= ch
;
1065 buf
[i
] = hex
[ch2
& 0xF];
1072 scm_lfwrite (buf
+ i
, 9 - i
, port
);
1077 /* Represent CH using the character escape syntax. */
1080 name
= scm_i_charname (SCM_MAKE_CHAR (ch
));
1082 scm_puts (name
, port
);
1084 PRINT_CHAR_ESCAPE (ch
, port
);
1088 /* Write CH to PORT, escaping it if it's non-graphic or not
1089 representable in PORT's encoding. If STRING_ESCAPES_P is true and CH
1090 needs to be escaped, it is escaped using the in-string escape syntax;
1091 otherwise the character escape syntax is used. */
1093 write_character (scm_t_wchar ch
, SCM port
, int string_escapes_p
)
1096 scm_t_string_failed_conversion_handler strategy
;
1098 strategy
= scm_i_get_conversion_strategy (port
);
1100 if (string_escapes_p
)
1102 /* Check if CH deserves special treatment. */
1103 if (ch
== '"' || ch
== '\\')
1105 display_character ('\\', port
, iconveh_question_mark
);
1106 display_character (ch
, port
, strategy
);
1109 else if (ch
== ' ' || ch
== '\n')
1111 display_character (ch
, port
, strategy
);
1117 display_string ("#\\", 1, 2, port
, iconveh_question_mark
);
1119 if (uc_combining_class (ch
) != UC_CCC_NR
)
1120 /* Character is a combining character, so attempt to
1122 printed
= write_combining_character (ch
, port
);
1126 && uc_is_general_category_withtable (ch
,
1127 UC_CATEGORY_MASK_L
|
1128 UC_CATEGORY_MASK_M
|
1129 UC_CATEGORY_MASK_N
|
1130 UC_CATEGORY_MASK_P
|
1131 UC_CATEGORY_MASK_S
))
1132 /* CH is graphic; attempt to display it. */
1133 printed
= display_character (ch
, port
, iconveh_error
);
1136 /* CH isn't graphic or cannot be represented in PORT's encoding. */
1137 write_character_escaped (ch
, string_escapes_p
, port
);
1140 /* Print an integer.
1144 scm_intprint (scm_t_intmax n
, int radix
, SCM port
)
1146 char num_buf
[SCM_INTBUFLEN
];
1147 scm_lfwrite (num_buf
, scm_iint2str (n
, radix
, num_buf
), port
);
1151 scm_uintprint (scm_t_uintmax n
, int radix
, SCM port
)
1153 char num_buf
[SCM_INTBUFLEN
];
1154 scm_lfwrite (num_buf
, scm_iuint2str (n
, radix
, num_buf
), port
);
1157 /* Print an object of unrecognized type.
1161 scm_ipruk (char *hdr
, SCM ptr
, SCM port
)
1163 scm_puts ("#<unknown-", port
);
1164 scm_puts (hdr
, port
);
1165 if (1) /* (scm_in_heap_p (ptr)) */ /* FIXME */
1167 scm_puts (" (0x", port
);
1168 scm_uintprint (SCM_CELL_WORD_0 (ptr
), 16, port
);
1169 scm_puts (" . 0x", port
);
1170 scm_uintprint (SCM_CELL_WORD_1 (ptr
), 16, port
);
1171 scm_puts (") @", port
);
1173 scm_puts (" 0x", port
);
1174 scm_uintprint (SCM_UNPACK (ptr
), 16, port
);
1175 scm_putc ('>', port
);
1182 scm_iprlist (char *hdr
, SCM exp
, int tlr
, SCM port
, scm_print_state
*pstate
)
1184 register SCM hare
, tortoise
;
1185 long floor
= pstate
->top
- 2;
1186 scm_puts (hdr
, port
);
1189 goto fancy_printing
;
1191 /* Run a hare and tortoise so that total time complexity will be
1192 O(depth * N) instead of O(N^2). */
1193 hare
= SCM_CDR (exp
);
1195 while (scm_is_pair (hare
))
1197 if (scm_is_eq (hare
, tortoise
))
1198 goto fancy_printing
;
1199 hare
= SCM_CDR (hare
);
1200 if (!scm_is_pair (hare
))
1202 hare
= SCM_CDR (hare
);
1203 tortoise
= SCM_CDR (tortoise
);
1206 /* No cdr cycles intrinsic to this list */
1207 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1208 for (exp
= SCM_CDR (exp
); scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1212 for (i
= floor
; i
>= 0; --i
)
1213 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1215 PUSH_REF (pstate
, exp
);
1216 scm_putc (' ', port
);
1218 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1220 if (!SCM_NULL_OR_NIL_P (exp
))
1222 scm_puts (" . ", port
);
1223 scm_iprin1 (exp
, port
, pstate
);
1227 scm_putc (tlr
, port
);
1228 pstate
->top
= floor
+ 2;
1233 long n
= pstate
->length
;
1235 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1236 exp
= SCM_CDR (exp
); --n
;
1237 for (; scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1239 register unsigned long i
;
1241 for (i
= 0; i
< pstate
->top
; ++i
)
1242 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1248 scm_puts (" ...", port
);
1254 PUSH_REF(pstate
, exp
);
1255 ++pstate
->list_offset
;
1256 scm_putc (' ', port
);
1258 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1261 if (!SCM_NULL_OR_NIL_P (exp
))
1263 scm_puts (" . ", port
);
1264 scm_iprin1 (exp
, port
, pstate
);
1267 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1271 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1274 scm_puts (" . ", port
);
1275 print_circref (port
, pstate
, exp
);
1282 scm_valid_oport_value_p (SCM val
)
1284 return (SCM_OPOUTPORTP (val
)
1285 || (SCM_PORT_WITH_PS_P (val
)
1286 && SCM_OPOUTPORTP (SCM_PORT_WITH_PS_PORT (val
))));
1289 /* SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write); */
1292 scm_write (SCM obj
, SCM port
)
1294 if (SCM_UNBNDP (port
))
1295 port
= scm_current_output_port ();
1297 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_write
);
1299 scm_prin1 (obj
, port
, 1);
1300 return SCM_UNSPECIFIED
;
1304 /* SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display); */
1307 scm_display (SCM obj
, SCM port
)
1309 if (SCM_UNBNDP (port
))
1310 port
= scm_current_output_port ();
1312 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_display
);
1314 scm_prin1 (obj
, port
, 0);
1315 return SCM_UNSPECIFIED
;
1319 SCM_DEFINE (scm_simple_format
, "simple-format", 2, 0, 1,
1320 (SCM destination
, SCM message
, SCM args
),
1321 "Write @var{message} to @var{destination}, defaulting to\n"
1322 "the current output port.\n"
1323 "@var{message} can contain @code{~A} (was @code{%s}) and\n"
1324 "@code{~S} (was @code{%S}) escapes. When printed,\n"
1325 "the escapes are replaced with corresponding members of\n"
1327 "@code{~A} formats using @code{display} and @code{~S} formats\n"
1328 "using @code{write}.\n"
1329 "If @var{destination} is @code{#t}, then use the current output\n"
1330 "port, if @var{destination} is @code{#f}, then return a string\n"
1331 "containing the formatted text. Does not add a trailing newline.")
1332 #define FUNC_NAME s_scm_simple_format
1334 SCM port
, answer
= SCM_UNSPECIFIED
;
1335 int fReturnString
= 0;
1337 size_t start
, p
, end
;
1339 if (scm_is_eq (destination
, SCM_BOOL_T
))
1341 destination
= port
= scm_current_output_port ();
1343 else if (scm_is_false (destination
))
1346 port
= scm_mkstrport (SCM_INUM0
, SCM_BOOL_F
,
1347 SCM_OPN
| SCM_WRTNG
,
1353 SCM_VALIDATE_OPORT_VALUE (1, destination
);
1354 port
= SCM_COERCE_OUTPORT (destination
);
1356 SCM_VALIDATE_STRING (2, message
);
1357 SCM_VALIDATE_REST_ARGUMENT (args
);
1361 end
= scm_i_string_length (message
);
1362 for (p
= start
; p
!= end
; ++p
)
1363 if (scm_i_string_ref (message
, p
) == '~')
1368 switch (scm_i_string_ref (message
, p
))
1377 scm_lfwrite_substr (message
, start
, p
, port
);
1381 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1386 SCM_MISC_ERROR ("FORMAT: Unsupported format option ~~~A - use (ice-9 format) instead",
1387 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1392 if (!scm_is_pair (args
))
1393 SCM_MISC_ERROR ("FORMAT: Missing argument for ~~~A",
1394 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message
, p
))));
1396 scm_lfwrite_substr (message
, start
, p
- 1, port
);
1397 /* we pass destination here */
1398 scm_prin1 (SCM_CAR (args
), destination
, writingp
);
1399 args
= SCM_CDR (args
);
1403 scm_lfwrite_substr (message
, start
, p
, port
);
1404 if (!scm_is_eq (args
, SCM_EOL
))
1405 SCM_MISC_ERROR ("FORMAT: ~A superfluous arguments",
1406 scm_list_1 (scm_length (args
)));
1409 answer
= scm_strport_to_string (destination
);
1411 return scm_return_first (answer
, message
);
1416 SCM_DEFINE (scm_newline
, "newline", 0, 1, 0,
1418 "Send a newline to @var{port}.\n"
1419 "If @var{port} is omitted, send to the current output port.")
1420 #define FUNC_NAME s_scm_newline
1422 if (SCM_UNBNDP (port
))
1423 port
= scm_current_output_port ();
1425 SCM_VALIDATE_OPORT_VALUE (1, port
);
1427 scm_putc ('\n', SCM_COERCE_OUTPORT (port
));
1428 return SCM_UNSPECIFIED
;
1432 SCM_DEFINE (scm_write_char
, "write-char", 1, 1, 0,
1433 (SCM chr
, SCM port
),
1434 "Send character @var{chr} to @var{port}.")
1435 #define FUNC_NAME s_scm_write_char
1437 if (SCM_UNBNDP (port
))
1438 port
= scm_current_output_port ();
1440 SCM_VALIDATE_CHAR (1, chr
);
1441 SCM_VALIDATE_OPORT_VALUE (2, port
);
1443 port
= SCM_COERCE_OUTPORT (port
);
1444 if (!display_character (SCM_CHAR (chr
), port
,
1445 scm_i_get_conversion_strategy (port
)))
1446 scm_encoding_error (__func__
, errno
,
1447 "cannot convert to output locale",
1450 return SCM_UNSPECIFIED
;
1456 /* Call back to Scheme code to do the printing of special objects
1457 * (like structs). SCM_PRINTER_APPLY applies PROC to EXP and a smob
1458 * containing PORT and PSTATE. This object can be used as the port for
1459 * display/write etc to continue the current print chain. The REVEALED
1460 * field of PSTATE is set to true to indicate that the print state has
1461 * escaped to Scheme and thus has to be freed by the GC.
1464 scm_t_bits scm_tc16_port_with_ps
;
1466 /* Print exactly as the port itself would */
1469 port_with_ps_print (SCM obj
, SCM port
, scm_print_state
*pstate
)
1471 obj
= SCM_PORT_WITH_PS_PORT (obj
);
1472 return scm_ptobs
[SCM_PTOBNUM (obj
)].print (obj
, port
, pstate
);
1476 scm_printer_apply (SCM proc
, SCM exp
, SCM port
, scm_print_state
*pstate
)
1478 pstate
->revealed
= 1;
1479 return scm_call_2 (proc
, exp
,
1480 scm_i_port_with_print_state (port
, pstate
->handle
));
1483 SCM_DEFINE (scm_port_with_print_state
, "port-with-print-state", 1, 1, 0,
1484 (SCM port
, SCM pstate
),
1485 "Create a new port which behaves like @var{port}, but with an\n"
1486 "included print state @var{pstate}. @var{pstate} is optional.\n"
1487 "If @var{pstate} isn't supplied and @var{port} already has\n"
1488 "a print state, the old print state is reused.")
1489 #define FUNC_NAME s_scm_port_with_print_state
1491 SCM_VALIDATE_OPORT_VALUE (1, port
);
1492 if (!SCM_UNBNDP (pstate
))
1493 SCM_VALIDATE_PRINTSTATE (2, pstate
);
1494 return scm_i_port_with_print_state (port
, pstate
);
1498 SCM_DEFINE (scm_get_print_state
, "get-print-state", 1, 0, 0,
1500 "Return the print state of the port @var{port}. If @var{port}\n"
1501 "has no associated print state, @code{#f} is returned.")
1502 #define FUNC_NAME s_scm_get_print_state
1504 if (SCM_PORT_WITH_PS_P (port
))
1505 return SCM_PORT_WITH_PS_PS (port
);
1506 if (SCM_OUTPUT_PORT_P (port
))
1508 SCM_WRONG_TYPE_ARG (1, port
);
1517 SCM vtable
, layout
, type
;
1519 scm_init_opts (scm_print_options
, scm_print_opts
);
1521 scm_print_options (scm_list_4 (scm_from_latin1_symbol ("highlight-prefix"),
1522 scm_from_locale_string ("{"),
1523 scm_from_latin1_symbol ("highlight-suffix"),
1524 scm_from_locale_string ("}")));
1526 scm_gc_register_root (&print_state_pool
);
1527 scm_gc_register_root (&scm_print_state_vtable
);
1528 vtable
= scm_make_vtable_vtable (scm_nullstr
, SCM_INUM0
, SCM_EOL
);
1530 scm_make_struct_layout (scm_from_locale_string (SCM_PRINT_STATE_LAYOUT
));
1531 type
= scm_make_struct (vtable
, SCM_INUM0
, scm_list_1 (layout
));
1532 scm_set_struct_vtable_name_x (type
, scm_from_latin1_symbol ("print-state"));
1533 scm_print_state_vtable
= type
;
1535 /* Don't want to bind a wrapper class in GOOPS, so pass 0 as arg1. */
1536 scm_tc16_port_with_ps
= scm_make_smob_type (0, 0);
1537 scm_set_smob_print (scm_tc16_port_with_ps
, port_with_ps_print
);
1539 #include "libguile/print.x"
1541 scm_print_opts
[SCM_PRINT_KEYWORD_STYLE_I
].val
= SCM_UNPACK (sym_reader
);