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'." },
110 { SCM_OPTION_BOOLEAN
, "escape-newlines", 1,
111 "Render newlines as \\n when printing using `write'." },
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_unlocked ('#', 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_unlocked ('#', port
);
305 scm_intprint (i
- self
, 10, port
);
306 scm_putc_unlocked ('#', 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_unlocked ("#{", 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_unlocked ("}#", 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_utf8_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_unlocked ("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_unlocked (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_unlocked ("#<uninterned-symbol ", port
);
606 scm_i_print_symbol_name (exp
, port
);
607 scm_putc_unlocked (' ', port
);
608 scm_uintprint (SCM_UNPACK (exp
), 16, port
);
609 scm_putc_unlocked ('>', 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
);
624 case scm_tc7_weak_set
:
625 scm_i_weak_set_print (exp
, port
, pstate
);
627 case scm_tc7_weak_table
:
628 scm_i_weak_table_print (exp
, port
, pstate
);
631 scm_i_fluid_print (exp
, port
, pstate
);
633 case scm_tc7_dynamic_state
:
634 scm_i_dynamic_state_print (exp
, port
, pstate
);
637 scm_i_frame_print (exp
, port
, pstate
);
639 case scm_tc7_objcode
:
640 scm_i_objcode_print (exp
, port
, pstate
);
643 scm_i_vm_print (exp
, port
, pstate
);
645 case scm_tc7_vm_cont
:
646 scm_i_vm_cont_print (exp
, port
, pstate
);
649 scm_i_prompt_print (exp
, port
, pstate
);
651 case scm_tc7_with_fluids
:
652 scm_i_with_fluids_print (exp
, port
, pstate
);
655 ENTER_NESTED_DATA (pstate
, exp
, circref
);
656 scm_puts_unlocked ("#w(", port
);
657 goto common_vector_printer
;
659 case scm_tc7_bytevector
:
660 scm_i_print_bytevector (exp
, port
, pstate
);
663 ENTER_NESTED_DATA (pstate
, exp
, circref
);
664 scm_puts_unlocked ("#(", port
);
665 common_vector_printer
:
668 long last
= SCM_SIMPLE_VECTOR_LENGTH (exp
) - 1;
671 && SCM_SIMPLE_VECTOR_LENGTH (exp
) > pstate
->length
)
673 last
= pstate
->length
- 1;
676 for (i
= 0; i
< last
; ++i
)
678 scm_iprin1 (scm_c_vector_ref (exp
, i
), port
, pstate
);
679 scm_putc_unlocked (' ', port
);
684 scm_iprin1 (scm_c_vector_ref (exp
, i
), port
, pstate
);
687 scm_puts_unlocked (" ...", port
);
688 scm_putc_unlocked (')', port
);
690 EXIT_NESTED_DATA (pstate
);
694 scm_t_ptob_descriptor
*ptob
= SCM_PORT_DESCRIPTOR (exp
);
695 if (ptob
->print
&& ptob
->print (exp
, port
, pstate
))
700 ENTER_NESTED_DATA (pstate
, exp
, circref
);
701 SCM_SMOB_DESCRIPTOR (exp
).print (exp
, port
, pstate
);
702 EXIT_NESTED_DATA (pstate
);
705 /* case scm_tcs_closures: */
707 scm_ipruk ("type", exp
, port
);
712 /* Print states are necessary for circular reference safe printing.
713 * They are also expensive to allocate. Therefore print states are
714 * kept in a pool so that they can be reused.
717 /* The PORT argument can also be a print-state/port pair, which will
718 * then be used instead of allocating a new print state. This is
719 * useful for continuing a chain of print calls from Scheme. */
722 scm_prin1 (SCM exp
, SCM port
, int writingp
)
724 SCM handle
= SCM_BOOL_F
; /* Will GC protect the handle whilst unlinked */
726 scm_print_state
*pstate
;
729 /* If PORT is a print-state/port pair, use that. Else create a new
732 if (SCM_PORT_WITH_PS_P (port
))
734 pstate_scm
= SCM_PORT_WITH_PS_PS (port
);
735 port
= SCM_PORT_WITH_PS_PORT (port
);
739 /* First try to allocate a print state from the pool */
740 scm_i_pthread_mutex_lock (&print_state_mutex
);
741 if (!scm_is_null (print_state_pool
))
743 handle
= print_state_pool
;
744 print_state_pool
= SCM_CDR (print_state_pool
);
746 scm_i_pthread_mutex_unlock (&print_state_mutex
);
747 if (scm_is_false (handle
))
748 handle
= scm_list_1 (make_print_state ());
749 pstate_scm
= SCM_CAR (handle
);
752 pstate
= SCM_PRINT_STATE (pstate_scm
);
753 old_writingp
= pstate
->writingp
;
754 pstate
->writingp
= writingp
;
755 scm_iprin1 (exp
, port
, pstate
);
756 pstate
->writingp
= old_writingp
;
758 /* Return print state to pool if it has been created above and
759 hasn't escaped to Scheme. */
761 if (scm_is_true (handle
) && !pstate
->revealed
)
763 scm_i_pthread_mutex_lock (&print_state_mutex
);
764 SCM_SETCDR (handle
, print_state_pool
);
765 print_state_pool
= handle
;
766 scm_i_pthread_mutex_unlock (&print_state_mutex
);
770 /* Convert codepoint CH to UTF-8 and store the result in UTF8. Return
771 the number of bytes of the UTF-8-encoded string. */
773 codepoint_to_utf8 (scm_t_wchar ch
, scm_t_uint8 utf8
[4])
776 scm_t_uint32 codepoint
;
778 codepoint
= (scm_t_uint32
) ch
;
780 if (codepoint
<= 0x7f)
783 utf8
[0] = (scm_t_uint8
) codepoint
;
785 else if (codepoint
<= 0x7ffUL
)
788 utf8
[0] = 0xc0 | (codepoint
>> 6);
789 utf8
[1] = 0x80 | (codepoint
& 0x3f);
791 else if (codepoint
<= 0xffffUL
)
794 utf8
[0] = 0xe0 | (codepoint
>> 12);
795 utf8
[1] = 0x80 | ((codepoint
>> 6) & 0x3f);
796 utf8
[2] = 0x80 | (codepoint
& 0x3f);
801 utf8
[0] = 0xf0 | (codepoint
>> 18);
802 utf8
[1] = 0x80 | ((codepoint
>> 12) & 0x3f);
803 utf8
[2] = 0x80 | ((codepoint
>> 6) & 0x3f);
804 utf8
[3] = 0x80 | (codepoint
& 0x3f);
810 #define STR_REF(s, x) \
812 ? (scm_t_wchar) ((unsigned char *) (s))[x] \
813 : ((scm_t_wchar *) (s))[x])
815 /* Write STR to PORT as UTF-8. STR is a LEN-codepoint string; it is
816 narrow if NARROW_P is true, wide otherwise. Return LEN. */
818 display_string_as_utf8 (const void *str
, int narrow_p
, size_t len
,
823 while (len
> printed
)
826 char *input
, utf8_buf
[256];
828 /* Convert STR to UTF-8. */
829 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
830 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
833 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
834 (scm_t_uint8
*) input
);
835 input
= utf8_buf
+ utf8_len
;
838 /* INPUT was successfully converted, entirely; print the
840 scm_lfwrite_unlocked (utf8_buf
, utf8_len
, port
);
841 printed
+= i
- printed
;
844 assert (printed
== len
);
849 /* Convert STR through PORT's output conversion descriptor and write the
850 output to PORT. Return the number of codepoints written. */
852 display_string_using_iconv (const void *str
, int narrow_p
, size_t len
,
854 scm_t_string_failed_conversion_handler strategy
)
859 pt
= SCM_PTAB_ENTRY (port
);
863 while (len
> printed
)
865 size_t done
, utf8_len
, input_left
, output_left
, i
;
866 size_t codepoints_read
, output_len
;
867 char *input
, *output
;
868 char utf8_buf
[256], encoded_output
[256];
871 /* Convert STR to UTF-8. */
872 for (i
= printed
, utf8_len
= 0, input
= utf8_buf
;
873 i
< len
&& utf8_len
+ 4 < sizeof (utf8_buf
);
876 offsets
[utf8_len
] = i
;
877 utf8_len
+= codepoint_to_utf8 (STR_REF (str
, i
),
878 (scm_t_uint8
*) input
);
879 input
= utf8_buf
+ utf8_len
;
883 input_left
= utf8_len
;
885 output
= encoded_output
;
886 output_left
= sizeof (encoded_output
);
888 done
= iconv (pt
->output_cd
, &input
, &input_left
,
889 &output
, &output_left
);
891 output_len
= sizeof (encoded_output
) - output_left
;
893 if (SCM_UNLIKELY (done
== (size_t) -1))
895 int errno_save
= errno
;
897 /* Reset the `iconv' state. */
898 iconv (pt
->output_cd
, NULL
, NULL
, NULL
, NULL
);
900 /* Print the OUTPUT_LEN bytes successfully converted. */
901 scm_lfwrite_unlocked (encoded_output
, output_len
, port
);
903 /* See how many input codepoints these OUTPUT_LEN bytes
905 codepoints_read
= offsets
[input
- utf8_buf
] - printed
;
906 printed
+= codepoints_read
;
908 if (errno_save
== EILSEQ
&&
909 strategy
!= SCM_FAILED_CONVERSION_ERROR
)
911 /* Conversion failed somewhere in INPUT and we want to
912 escape or substitute the offending input character. */
914 if (strategy
== SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE
)
918 /* Find CH, the offending codepoint, and escape it. */
919 ch
= STR_REF (str
, offsets
[input
- utf8_buf
]);
920 write_character_escaped (ch
, 1, port
);
923 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
924 display_string ("?", 1, 1, port
, strategy
);
929 /* Something bad happened that we can't handle: bail out. */
934 /* INPUT was successfully converted, entirely; print the
936 scm_lfwrite_unlocked (encoded_output
, output_len
, port
);
937 codepoints_read
= i
- printed
;
938 printed
+= codepoints_read
;
947 /* Display the LEN codepoints in STR to PORT according to STRATEGY;
948 return the number of codepoints successfully displayed. If NARROW_P,
949 then STR is interpreted as a sequence of `char', denoting a Latin-1
950 string; otherwise it's interpreted as a sequence of
953 display_string (const void *str
, int narrow_p
,
954 size_t len
, SCM port
,
955 scm_t_string_failed_conversion_handler strategy
)
960 pt
= SCM_PTAB_ENTRY (port
);
962 if (pt
->output_cd
== (iconv_t
) -1)
963 /* Initialize the conversion descriptors, if needed. */
964 scm_i_set_port_encoding_x (port
, pt
->encoding
);
966 /* FIXME: In 2.1, add a flag to determine whether a port is UTF-8. */
967 if (pt
->output_cd
== (iconv_t
) -1)
968 return display_string_as_utf8 (str
, narrow_p
, len
, port
);
970 return display_string_using_iconv (str
, narrow_p
, len
,
974 /* Attempt to display CH to PORT according to STRATEGY. Return non-zero
975 if CH was successfully displayed, zero otherwise (e.g., if it was not
976 representable in PORT's encoding.) */
978 display_character (scm_t_wchar ch
, SCM port
,
979 scm_t_string_failed_conversion_handler strategy
)
981 return display_string (&ch
, 0, 1, port
, strategy
) == 1;
984 /* Attempt to pretty-print CH, a combining character, to PORT. Return
985 zero upon failure, non-zero otherwise. The idea is to print CH above
986 a dotted circle to make it more visible. */
988 write_combining_character (scm_t_wchar ch
, SCM port
)
992 str
[0] = SCM_CODEPOINT_DOTTED_CIRCLE
;
995 return display_string (str
, 0, 2, port
, iconveh_error
) == 2;
998 /* Write CH to PORT in its escaped form, using the string escape syntax
999 if STRING_ESCAPES_P is non-zero. */
1001 write_character_escaped (scm_t_wchar ch
, int string_escapes_p
, SCM port
)
1003 if (string_escapes_p
)
1005 /* Represent CH using the in-string escape syntax. */
1007 static const char hex
[] = "0123456789abcdef";
1008 static const char escapes
[7] = "abtnvfr";
1011 if (ch
>= 0x07 && ch
<= 0x0D && ch
!= 0x0A)
1013 /* Use special escapes for some C0 controls. */
1015 buf
[1] = escapes
[ch
- 0x07];
1016 scm_lfwrite_unlocked (buf
, 2, port
);
1018 else if (!SCM_R6RS_ESCAPES_P
)
1024 buf
[2] = hex
[ch
/ 16];
1025 buf
[3] = hex
[ch
% 16];
1026 scm_lfwrite_unlocked (buf
, 4, port
);
1028 else if (ch
<= 0xFFFF)
1032 buf
[2] = hex
[(ch
& 0xF000) >> 12];
1033 buf
[3] = hex
[(ch
& 0xF00) >> 8];
1034 buf
[4] = hex
[(ch
& 0xF0) >> 4];
1035 buf
[5] = hex
[(ch
& 0xF)];
1036 scm_lfwrite_unlocked (buf
, 6, port
);
1038 else if (ch
> 0xFFFF)
1042 buf
[2] = hex
[(ch
& 0xF00000) >> 20];
1043 buf
[3] = hex
[(ch
& 0xF0000) >> 16];
1044 buf
[4] = hex
[(ch
& 0xF000) >> 12];
1045 buf
[5] = hex
[(ch
& 0xF00) >> 8];
1046 buf
[6] = hex
[(ch
& 0xF0) >> 4];
1047 buf
[7] = hex
[(ch
& 0xF)];
1048 scm_lfwrite_unlocked (buf
, 8, port
);
1053 /* Print an R6RS variable-length hex escape: "\xNNNN;". */
1054 scm_t_wchar ch2
= ch
;
1064 buf
[i
] = hex
[ch2
& 0xF];
1071 scm_lfwrite_unlocked (buf
+ i
, 9 - i
, port
);
1076 /* Represent CH using the character escape syntax. */
1079 name
= scm_i_charname (SCM_MAKE_CHAR (ch
));
1081 scm_puts_unlocked (name
, port
);
1083 PRINT_CHAR_ESCAPE (ch
, port
);
1087 /* Write CH to PORT, escaping it if it's non-graphic or not
1088 representable in PORT's encoding. If STRING_ESCAPES_P is true and CH
1089 needs to be escaped, it is escaped using the in-string escape syntax;
1090 otherwise the character escape syntax is used. */
1092 write_character (scm_t_wchar ch
, SCM port
, int string_escapes_p
)
1095 scm_t_string_failed_conversion_handler strategy
;
1097 strategy
= scm_i_get_conversion_strategy (port
);
1099 if (string_escapes_p
)
1101 /* Check if CH deserves special treatment. */
1102 if (ch
== '"' || ch
== '\\')
1104 display_character ('\\', port
, iconveh_question_mark
);
1105 display_character (ch
, port
, strategy
);
1108 else if (ch
== '\n' && SCM_PRINT_ESCAPE_NEWLINES_P
)
1110 display_character ('\\', port
, iconveh_question_mark
);
1111 display_character ('n', port
, strategy
);
1114 else if (ch
== ' ' || ch
== '\n')
1116 display_character (ch
, port
, strategy
);
1122 display_string ("#\\", 1, 2, port
, iconveh_question_mark
);
1124 if (uc_combining_class (ch
) != UC_CCC_NR
)
1125 /* Character is a combining character, so attempt to
1127 printed
= write_combining_character (ch
, port
);
1131 && uc_is_general_category_withtable (ch
,
1132 UC_CATEGORY_MASK_L
|
1133 UC_CATEGORY_MASK_M
|
1134 UC_CATEGORY_MASK_N
|
1135 UC_CATEGORY_MASK_P
|
1136 UC_CATEGORY_MASK_S
))
1137 /* CH is graphic; attempt to display it. */
1138 printed
= display_character (ch
, port
, iconveh_error
);
1141 /* CH isn't graphic or cannot be represented in PORT's encoding. */
1142 write_character_escaped (ch
, string_escapes_p
, port
);
1145 /* Print an integer.
1149 scm_intprint (scm_t_intmax n
, int radix
, SCM port
)
1151 char num_buf
[SCM_INTBUFLEN
];
1152 scm_lfwrite_unlocked (num_buf
, scm_iint2str (n
, radix
, num_buf
), port
);
1156 scm_uintprint (scm_t_uintmax n
, int radix
, SCM port
)
1158 char num_buf
[SCM_INTBUFLEN
];
1159 scm_lfwrite_unlocked (num_buf
, scm_iuint2str (n
, radix
, num_buf
), port
);
1162 /* Print an object of unrecognized type.
1166 scm_ipruk (char *hdr
, SCM ptr
, SCM port
)
1168 scm_puts_unlocked ("#<unknown-", port
);
1169 scm_puts_unlocked (hdr
, port
);
1170 if (1) /* (scm_in_heap_p (ptr)) */ /* FIXME */
1172 scm_puts_unlocked (" (0x", port
);
1173 scm_uintprint (SCM_CELL_WORD_0 (ptr
), 16, port
);
1174 scm_puts_unlocked (" . 0x", port
);
1175 scm_uintprint (SCM_CELL_WORD_1 (ptr
), 16, port
);
1176 scm_puts_unlocked (") @", port
);
1178 scm_puts_unlocked (" 0x", port
);
1179 scm_uintprint (SCM_UNPACK (ptr
), 16, port
);
1180 scm_putc_unlocked ('>', port
);
1187 scm_iprlist (char *hdr
, SCM exp
, int tlr
, SCM port
, scm_print_state
*pstate
)
1189 register SCM hare
, tortoise
;
1190 long floor
= pstate
->top
- 2;
1191 scm_puts_unlocked (hdr
, port
);
1194 goto fancy_printing
;
1196 /* Run a hare and tortoise so that total time complexity will be
1197 O(depth * N) instead of O(N^2). */
1198 hare
= SCM_CDR (exp
);
1200 while (scm_is_pair (hare
))
1202 if (scm_is_eq (hare
, tortoise
))
1203 goto fancy_printing
;
1204 hare
= SCM_CDR (hare
);
1205 if (!scm_is_pair (hare
))
1207 hare
= SCM_CDR (hare
);
1208 tortoise
= SCM_CDR (tortoise
);
1211 /* No cdr cycles intrinsic to this list */
1212 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1213 for (exp
= SCM_CDR (exp
); scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1217 for (i
= floor
; i
>= 0; --i
)
1218 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1220 PUSH_REF (pstate
, exp
);
1221 scm_putc_unlocked (' ', port
);
1223 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1225 if (!SCM_NULL_OR_NIL_P (exp
))
1227 scm_puts_unlocked (" . ", port
);
1228 scm_iprin1 (exp
, port
, pstate
);
1232 scm_putc_unlocked (tlr
, port
);
1233 pstate
->top
= floor
+ 2;
1238 long n
= pstate
->length
;
1240 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1241 exp
= SCM_CDR (exp
); --n
;
1242 for (; scm_is_pair (exp
); exp
= SCM_CDR (exp
))
1244 register unsigned long i
;
1246 for (i
= 0; i
< pstate
->top
; ++i
)
1247 if (scm_is_eq (PSTATE_STACK_REF(pstate
, i
), exp
))
1253 scm_puts_unlocked (" ...", port
);
1259 PUSH_REF(pstate
, exp
);
1260 ++pstate
->list_offset
;
1261 scm_putc_unlocked (' ', port
);
1263 scm_iprin1 (SCM_CAR (exp
), port
, pstate
);
1266 if (!SCM_NULL_OR_NIL_P (exp
))
1268 scm_puts_unlocked (" . ", port
);
1269 scm_iprin1 (exp
, port
, pstate
);
1272 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1276 pstate
->list_offset
-= pstate
->top
- floor
- 2;
1279 scm_puts_unlocked (" . ", port
);
1280 print_circref (port
, pstate
, exp
);
1287 scm_valid_oport_value_p (SCM val
)
1289 return (SCM_OPOUTPORTP (val
)
1290 || (SCM_PORT_WITH_PS_P (val
)
1291 && SCM_OPOUTPORTP (SCM_PORT_WITH_PS_PORT (val
))));
1294 /* SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write); */
1297 scm_write (SCM obj
, SCM port
)
1299 if (SCM_UNBNDP (port
))
1300 port
= scm_current_output_port ();
1302 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_write
);
1304 scm_dynwind_begin (0);
1305 scm_dynwind_lock_port (SCM_COERCE_OUTPORT (port
));
1306 scm_prin1 (obj
, port
, 1);
1309 return SCM_UNSPECIFIED
;
1313 /* SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display); */
1316 scm_display (SCM obj
, SCM port
)
1318 if (SCM_UNBNDP (port
))
1319 port
= scm_current_output_port ();
1321 SCM_ASSERT (scm_valid_oport_value_p (port
), port
, SCM_ARG2
, s_display
);
1323 scm_dynwind_begin (0);
1324 scm_dynwind_lock_port (SCM_COERCE_OUTPORT (port
));
1325 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_unlocked ('\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_PORT_DESCRIPTOR (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
);
1532 type
= scm_make_vtable (scm_from_locale_string (SCM_PRINT_STATE_LAYOUT
),
1534 scm_set_struct_vtable_name_x (type
, scm_from_latin1_symbol ("print-state"));
1535 scm_print_state_vtable
= type
;
1537 /* Don't want to bind a wrapper class in GOOPS, so pass 0 as arg1. */
1538 scm_tc16_port_with_ps
= scm_make_smob_type (0, 0);
1539 scm_set_smob_print (scm_tc16_port_with_ps
, port_with_ps_print
);
1541 #include "libguile/print.x"
1543 scm_init_opts (scm_print_options
, scm_print_opts
);
1544 scm_print_opts
[SCM_PRINT_HIGHLIGHT_PREFIX_I
].val
=
1545 SCM_UNPACK (scm_from_locale_string ("{"));
1546 scm_print_opts
[SCM_PRINT_HIGHLIGHT_SUFFIX_I
].val
=
1547 SCM_UNPACK (scm_from_locale_string ("}"));
1548 scm_print_opts
[SCM_PRINT_KEYWORD_STYLE_I
].val
= SCM_UNPACK (sym_reader
);