Merge commit '5d971db802eaa8038db17e1aa5b4c69452739744'
[bpt/guile.git] / libguile / print.c
1 /* Copyright (C) 1995-1999, 2000, 2001, 2002, 2003, 2004, 2006, 2008,
2 * 2009, 2010, 2011, 2012, 2013, 2014, 2015 Free Software Foundation, Inc.
3 *
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.
8 *
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.
13 *
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
17 * 02110-1301 USA
18 */
19
20
21 \f
22 #ifdef HAVE_CONFIG_H
23 # include <config.h>
24 #endif
25
26 #include <errno.h>
27 #include <iconv.h>
28 #include <stdio.h>
29 #include <assert.h>
30
31 #include <uniconv.h>
32 #include <unictype.h>
33
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/ports-internal.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"
54
55 #include "libguile/validate.h"
56 #include "libguile/print.h"
57
58 #include "libguile/private-options.h"
59
60 \f
61
62 /* Character printers. */
63
64 #define PORT_CONVERSION_HANDLER(port) \
65 SCM_PTAB_ENTRY (port)->ilseq_handler
66
67 static size_t display_string (const void *, int, size_t, SCM,
68 scm_t_string_failed_conversion_handler);
69
70 static int display_character (scm_t_wchar, SCM,
71 scm_t_string_failed_conversion_handler);
72
73 static void write_character (scm_t_wchar, SCM, int);
74
75 static void write_character_escaped (scm_t_wchar, int, SCM);
76
77 \f
78
79 /* {Names of immediate symbols}
80 *
81 * This table must agree with the declarations in scm.h: {Immediate Symbols}.
82 */
83
84 /* This table must agree with the list of flags in tags.h. */
85 static const char *iflagnames[] =
86 {
87 "#f",
88 "#nil", /* Elisp nil value. Should print from elisp as symbol `nil'. */
89 "#<XXX UNUSED LISP FALSE -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
90 "()",
91 "#t",
92 "#<XXX UNUSED BOOLEAN 0 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
93 "#<XXX UNUSED BOOLEAN 1 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
94 "#<XXX UNUSED BOOLEAN 2 -- DO NOT USE -- SHOULD NEVER BE SEEN XXX>",
95 "#<unspecified>",
96 "#<undefined>",
97 "#<eof>",
98
99 /* Unbound slot marker for GOOPS. For internal use in GOOPS only. */
100 "#<unbound>",
101 };
102
103 SCM_SYMBOL (sym_reader, "reader");
104
105 scm_t_option scm_print_opts[] = {
106 { SCM_OPTION_SCM, "highlight-prefix", (scm_t_bits)SCM_BOOL_F_BITS,
107 "The string to print before highlighted values." },
108 { SCM_OPTION_SCM, "highlight-suffix", (scm_t_bits)SCM_BOOL_F_BITS,
109 "The string to print after highlighted values." },
110 { SCM_OPTION_SCM, "quote-keywordish-symbols", (scm_t_bits)SCM_BOOL_F_BITS,
111 "How to print symbols that have a colon as their first or last character. "
112 "The value '#f' does not quote the colons; '#t' quotes them; "
113 "'reader' quotes them when the reader option 'keywords' is not '#f'." },
114 { SCM_OPTION_BOOLEAN, "escape-newlines", 1,
115 "Render newlines as \\n when printing using `write'." },
116 { SCM_OPTION_BOOLEAN, "r7rs-symbols", 0,
117 "Escape symbols using R7RS |...| symbol notation." },
118 { 0 },
119 };
120
121 SCM_DEFINE (scm_print_options, "print-options-interface", 0, 1, 0,
122 (SCM setting),
123 "Option interface for the print options. Instead of using\n"
124 "this procedure directly, use the procedures\n"
125 "@code{print-enable}, @code{print-disable}, @code{print-set!}\n"
126 "and @code{print-options}.")
127 #define FUNC_NAME s_scm_print_options
128 {
129 SCM ans = scm_options (setting,
130 scm_print_opts,
131 FUNC_NAME);
132 return ans;
133 }
134 #undef FUNC_NAME
135
136 \f
137 /* {Printing of Scheme Objects}
138 */
139
140 /* Detection of circular references.
141 *
142 * Due to other constraints in the implementation, this code has bad
143 * time complexity (O (depth * N)), The printer code can be
144 * rewritten to be O(N).
145 */
146 #define PUSH_REF(pstate, obj) \
147 do \
148 { \
149 PSTATE_STACK_SET (pstate, pstate->top, obj); \
150 pstate->top++; \
151 if (pstate->top == pstate->ceiling) \
152 grow_ref_stack (pstate); \
153 } while(0)
154
155 #define ENTER_NESTED_DATA(pstate, obj, label) \
156 do \
157 { \
158 register unsigned long i; \
159 for (i = 0; i < pstate->top; ++i) \
160 if (scm_is_eq (PSTATE_STACK_REF (pstate, i), (obj))) \
161 goto label; \
162 if (pstate->fancyp) \
163 { \
164 if (pstate->top - pstate->list_offset >= pstate->level) \
165 { \
166 scm_putc_unlocked ('#', port); \
167 return; \
168 } \
169 } \
170 PUSH_REF(pstate, obj); \
171 } while(0)
172
173 #define EXIT_NESTED_DATA(pstate) \
174 do \
175 { \
176 --pstate->top; \
177 PSTATE_STACK_SET (pstate, pstate->top, SCM_UNDEFINED); \
178 } \
179 while (0)
180
181 SCM scm_print_state_vtable = SCM_BOOL_F;
182 static SCM print_state_pool = SCM_EOL;
183 scm_i_pthread_mutex_t print_state_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER;
184
185 #ifdef GUILE_DEBUG /* Used for debugging purposes */
186
187 SCM_DEFINE (scm_current_pstate, "current-pstate", 0, 0, 0,
188 (),
189 "Return the current-pstate -- the car of the\n"
190 "@code{print_state_pool}. @code{current-pstate} is only\n"
191 "included in @code{--enable-guile-debug} builds.")
192 #define FUNC_NAME s_scm_current_pstate
193 {
194 if (!scm_is_null (print_state_pool))
195 return SCM_CAR (print_state_pool);
196 else
197 return SCM_BOOL_F;
198 }
199 #undef FUNC_NAME
200
201 #endif
202
203 #define PSTATE_SIZE 50L
204
205 static SCM
206 make_print_state (void)
207 {
208 SCM print_state
209 = scm_make_struct (scm_print_state_vtable, SCM_INUM0, SCM_EOL);
210 scm_print_state *pstate = SCM_PRINT_STATE (print_state);
211 pstate->ref_vect = scm_c_make_vector (PSTATE_SIZE, SCM_UNDEFINED);
212 pstate->ceiling = SCM_SIMPLE_VECTOR_LENGTH (pstate->ref_vect);
213 pstate->highlight_objects = SCM_EOL;
214 return print_state;
215 }
216
217 SCM
218 scm_make_print_state ()
219 {
220 SCM answer = SCM_BOOL_F;
221
222 /* First try to allocate a print state from the pool */
223 scm_i_pthread_mutex_lock (&print_state_mutex);
224 if (!scm_is_null (print_state_pool))
225 {
226 answer = SCM_CAR (print_state_pool);
227 print_state_pool = SCM_CDR (print_state_pool);
228 }
229 scm_i_pthread_mutex_unlock (&print_state_mutex);
230
231 return scm_is_false (answer) ? make_print_state () : answer;
232 }
233
234 void
235 scm_free_print_state (SCM print_state)
236 {
237 SCM handle;
238 scm_print_state *pstate = SCM_PRINT_STATE (print_state);
239 /* Cleanup before returning print state to pool.
240 * It is better to do it here. Doing it in scm_prin1
241 * would cost more since that function is called much more
242 * often.
243 */
244 pstate->fancyp = 0;
245 pstate->revealed = 0;
246 pstate->highlight_objects = SCM_EOL;
247 scm_i_pthread_mutex_lock (&print_state_mutex);
248 handle = scm_cons (print_state, print_state_pool);
249 print_state_pool = handle;
250 scm_i_pthread_mutex_unlock (&print_state_mutex);
251 }
252
253 SCM
254 scm_i_port_with_print_state (SCM port, SCM print_state)
255 {
256 if (SCM_UNBNDP (print_state))
257 {
258 if (SCM_PORT_WITH_PS_P (port))
259 return port;
260 else
261 print_state = scm_make_print_state ();
262 /* port does not need to be coerced since it doesn't have ps */
263 }
264 else
265 port = SCM_COERCE_OUTPORT (port);
266 SCM_RETURN_NEWSMOB (scm_tc16_port_with_ps,
267 SCM_UNPACK (scm_cons (port, print_state)));
268 }
269
270 static void
271 grow_ref_stack (scm_print_state *pstate)
272 {
273 SCM old_vect = pstate->ref_vect;
274 size_t old_size = SCM_SIMPLE_VECTOR_LENGTH (old_vect);
275 size_t new_size = 2 * pstate->ceiling;
276 SCM new_vect = scm_c_make_vector (new_size, SCM_UNDEFINED);
277 unsigned long int i;
278
279 for (i = 0; i != old_size; ++i)
280 SCM_SIMPLE_VECTOR_SET (new_vect, i, SCM_SIMPLE_VECTOR_REF (old_vect, i));
281
282 pstate->ref_vect = new_vect;
283 pstate->ceiling = new_size;
284 }
285
286 #define PSTATE_STACK_REF(p,i) SCM_SIMPLE_VECTOR_REF((p)->ref_vect, (i))
287 #define PSTATE_STACK_SET(p,i,v) SCM_SIMPLE_VECTOR_SET((p)->ref_vect, (i), (v))
288
289 static void
290 print_circref (SCM port, scm_print_state *pstate, SCM ref)
291 {
292 register long i;
293 long self = pstate->top - 1;
294 i = pstate->top - 1;
295 if (scm_is_pair (PSTATE_STACK_REF (pstate, i)))
296 {
297 while (i > 0)
298 {
299 if (!scm_is_pair (PSTATE_STACK_REF (pstate, i-1))
300 || !scm_is_eq (SCM_CDR (PSTATE_STACK_REF (pstate, i-1)),
301 SCM_CDR (PSTATE_STACK_REF (pstate, i))))
302 break;
303 --i;
304 }
305 self = i;
306 }
307 for (i = pstate->top - 1; 1; --i)
308 if (scm_is_eq (PSTATE_STACK_REF(pstate, i), ref))
309 break;
310 scm_putc_unlocked ('#', port);
311 scm_intprint (i - self, 10, port);
312 scm_putc_unlocked ('#', port);
313 }
314
315 /* Print the name of a symbol. */
316
317 static int
318 quote_keywordish_symbols (void)
319 {
320 SCM option = SCM_PRINT_KEYWORD_STYLE;
321
322 if (scm_is_false (option))
323 return 0;
324 if (scm_is_eq (option, sym_reader))
325 return scm_is_true (SCM_PACK (SCM_KEYWORD_STYLE));
326 return 1;
327 }
328
329 #define INITIAL_IDENTIFIER_MASK \
330 (UC_CATEGORY_MASK_Lu | UC_CATEGORY_MASK_Ll | UC_CATEGORY_MASK_Lt \
331 | UC_CATEGORY_MASK_Lm | UC_CATEGORY_MASK_Lo | UC_CATEGORY_MASK_Mn \
332 | UC_CATEGORY_MASK_Nl | UC_CATEGORY_MASK_No | UC_CATEGORY_MASK_Pd \
333 | UC_CATEGORY_MASK_Pc | UC_CATEGORY_MASK_Po | UC_CATEGORY_MASK_Sc \
334 | UC_CATEGORY_MASK_Sm | UC_CATEGORY_MASK_Sk | UC_CATEGORY_MASK_So \
335 | UC_CATEGORY_MASK_Co)
336
337 #define SUBSEQUENT_IDENTIFIER_MASK \
338 (INITIAL_IDENTIFIER_MASK \
339 | UC_CATEGORY_MASK_Nd | UC_CATEGORY_MASK_Mc | UC_CATEGORY_MASK_Me)
340
341 /* FIXME: Cache this information on the symbol, somehow. */
342 static int
343 symbol_has_extended_read_syntax (SCM sym)
344 {
345 size_t pos, len = scm_i_symbol_length (sym);
346 scm_t_wchar c;
347
348 /* The empty symbol. */
349 if (len == 0)
350 return 1;
351
352 c = scm_i_symbol_ref (sym, 0);
353
354 switch (c)
355 {
356 case '\'':
357 case '`':
358 case ',':
359 case '"':
360 case ';':
361 case '#':
362 /* Some initial-character constraints. */
363 return 1;
364
365 case '|':
366 case '\\':
367 /* R7RS allows neither '|' nor '\' in bare symbols. */
368 if (SCM_PRINT_R7RS_SYMBOLS_P)
369 return 1;
370 break;
371
372 case ':':
373 /* Symbols that look like keywords. */
374 return quote_keywordish_symbols ();
375
376 case '.':
377 /* Single dot conflicts with dotted-pair notation. */
378 if (len == 1)
379 return 1;
380 /* Fall through to check numbers. */
381 case '+':
382 case '-':
383 case '0':
384 case '1':
385 case '2':
386 case '3':
387 case '4':
388 case '5':
389 case '6':
390 case '7':
391 case '8':
392 case '9':
393 /* Number-ish symbols. Numbers with radixes already caught be #
394 above. */
395 if (scm_is_true (scm_i_string_to_number (scm_symbol_to_string (sym), 10)))
396 return 1;
397 break;
398
399 default:
400 break;
401 }
402
403 /* Other disallowed first characters. */
404 if (!uc_is_general_category_withtable (c, INITIAL_IDENTIFIER_MASK))
405 return 1;
406
407 /* Keywords can be identified by trailing colons too. */
408 if (scm_i_symbol_ref (sym, len - 1) == ':')
409 return quote_keywordish_symbols ();
410
411 /* Otherwise, any character that's in the identifier category mask is
412 fine to pass through as-is, provided it's not one of the ASCII
413 delimiters like `;'. */
414 for (pos = 1; pos < len; pos++)
415 {
416 c = scm_i_symbol_ref (sym, pos);
417 if (!uc_is_general_category_withtable (c, SUBSEQUENT_IDENTIFIER_MASK))
418 return 1;
419 else if (c == '"' || c == ';' || c == '#')
420 return 1;
421 else if ((c == '|' || c == '\\') && SCM_PRINT_R7RS_SYMBOLS_P)
422 /* R7RS allows neither '|' nor '\' in bare symbols. */
423 return 1;
424 }
425
426 return 0;
427 }
428
429 static void
430 print_normal_symbol (SCM sym, SCM port)
431 {
432 size_t len;
433 scm_t_string_failed_conversion_handler strategy;
434
435 len = scm_i_symbol_length (sym);
436 strategy = SCM_PTAB_ENTRY (port)->ilseq_handler;
437
438 if (scm_i_is_narrow_symbol (sym))
439 display_string (scm_i_symbol_chars (sym), 1, len, port, strategy);
440 else
441 display_string (scm_i_symbol_wide_chars (sym), 0, len, port, strategy);
442 }
443
444 static void
445 print_extended_symbol (SCM sym, SCM port)
446 {
447 size_t pos, len;
448 scm_t_string_failed_conversion_handler strategy;
449
450 len = scm_i_symbol_length (sym);
451 strategy = PORT_CONVERSION_HANDLER (port);
452
453 scm_lfwrite_unlocked ("#{", 2, port);
454
455 for (pos = 0; pos < len; pos++)
456 {
457 scm_t_wchar c = scm_i_symbol_ref (sym, pos);
458
459 if (uc_is_general_category_withtable (c,
460 SUBSEQUENT_IDENTIFIER_MASK
461 | UC_CATEGORY_MASK_Zs))
462 {
463 if (!display_character (c, port, strategy)
464 || (c == '\\' && !display_character (c, port, strategy)))
465 scm_encoding_error ("print_extended_symbol", errno,
466 "cannot convert to output locale",
467 port, SCM_MAKE_CHAR (c));
468 }
469 else
470 {
471 scm_lfwrite_unlocked ("\\x", 2, port);
472 scm_intprint (c, 16, port);
473 scm_putc_unlocked (';', port);
474 }
475 }
476
477 scm_lfwrite_unlocked ("}#", 2, port);
478 }
479
480 static void
481 print_r7rs_extended_symbol (SCM sym, SCM port)
482 {
483 size_t pos, len;
484 scm_t_string_failed_conversion_handler strategy;
485
486 len = scm_i_symbol_length (sym);
487 strategy = PORT_CONVERSION_HANDLER (port);
488
489 scm_putc_unlocked ('|', port);
490
491 for (pos = 0; pos < len; pos++)
492 {
493 scm_t_wchar c = scm_i_symbol_ref (sym, pos);
494
495 switch (c)
496 {
497 case '\a': scm_lfwrite_unlocked ("\\a", 2, port); break;
498 case '\b': scm_lfwrite_unlocked ("\\b", 2, port); break;
499 case '\t': scm_lfwrite_unlocked ("\\t", 2, port); break;
500 case '\n': scm_lfwrite_unlocked ("\\n", 2, port); break;
501 case '\r': scm_lfwrite_unlocked ("\\r", 2, port); break;
502 case '|': scm_lfwrite_unlocked ("\\|", 2, port); break;
503 case '\\': scm_lfwrite_unlocked ("\\x5c;", 5, port); break;
504 default:
505 if (uc_is_general_category_withtable (c,
506 UC_CATEGORY_MASK_L
507 | UC_CATEGORY_MASK_M
508 | UC_CATEGORY_MASK_N
509 | UC_CATEGORY_MASK_P
510 | UC_CATEGORY_MASK_S)
511 || (c == ' '))
512 {
513 if (!display_character (c, port, strategy))
514 scm_encoding_error ("print_r7rs_extended_symbol", errno,
515 "cannot convert to output locale",
516 port, SCM_MAKE_CHAR (c));
517 }
518 else
519 {
520 scm_lfwrite_unlocked ("\\x", 2, port);
521 scm_intprint (c, 16, port);
522 scm_putc_unlocked (';', port);
523 }
524 break;
525 }
526 }
527
528 scm_putc_unlocked ('|', port);
529 }
530
531 /* FIXME: allow R6RS hex escapes instead of #{...}# or |...|. */
532 static void
533 print_symbol (SCM sym, SCM port)
534 {
535 if (!symbol_has_extended_read_syntax (sym))
536 print_normal_symbol (sym, port);
537 else if (SCM_PRINT_R7RS_SYMBOLS_P)
538 print_r7rs_extended_symbol (sym, port);
539 else
540 print_extended_symbol (sym, port);
541 }
542
543 void
544 scm_print_symbol_name (const char *str, size_t len, SCM port)
545 {
546 SCM symbol = scm_from_utf8_symboln (str, len);
547 print_symbol (symbol, port);
548 }
549
550 /* Print generally. Handles both write and display according to PSTATE.
551 */
552 SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write);
553 SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display);
554
555 static void iprin1 (SCM exp, SCM port, scm_print_state *pstate);
556
557
558 /* Print a character as an octal or hex escape. */
559 #define PRINT_CHAR_ESCAPE(i, port) \
560 do \
561 { \
562 if (!SCM_R6RS_ESCAPES_P) \
563 scm_intprint (i, 8, port); \
564 else \
565 { \
566 scm_puts_unlocked ("x", port); \
567 scm_intprint (i, 16, port); \
568 } \
569 } \
570 while (0)
571
572
573 void
574 scm_iprin1 (SCM exp, SCM port, scm_print_state *pstate)
575 {
576 if (pstate->fancyp
577 && scm_is_true (scm_memq (exp, pstate->highlight_objects)))
578 {
579 scm_display (SCM_PRINT_HIGHLIGHT_PREFIX, port);
580 iprin1 (exp, port, pstate);
581 scm_display (SCM_PRINT_HIGHLIGHT_SUFFIX, port);
582 }
583 else
584 iprin1 (exp, port, pstate);
585 }
586
587 static void
588 print_vector_or_weak_vector (SCM v, size_t len, SCM (*ref) (SCM, size_t),
589 SCM port, scm_print_state *pstate)
590 {
591 long i;
592 long last = len - 1;
593 int cutp = 0;
594 if (pstate->fancyp && len > pstate->length)
595 {
596 last = pstate->length - 1;
597 cutp = 1;
598 }
599 for (i = 0; i < last; ++i)
600 {
601 scm_iprin1 (ref (v, i), port, pstate);
602 scm_putc_unlocked (' ', port);
603 }
604 if (i == last)
605 {
606 /* CHECK_INTS; */
607 scm_iprin1 (ref (v, i), port, pstate);
608 }
609 if (cutp)
610 scm_puts_unlocked (" ...", port);
611 scm_putc_unlocked (')', port);
612 }
613
614 static void
615 iprin1 (SCM exp, SCM port, scm_print_state *pstate)
616 {
617 switch (SCM_ITAG3 (exp))
618 {
619 case scm_tc3_tc7_1:
620 case scm_tc3_tc7_2:
621 /* These tc3 tags should never occur in an immediate value. They are
622 * only used in cell types of non-immediates, i. e. the value returned
623 * by SCM_CELL_TYPE (exp) can use these tags.
624 */
625 scm_ipruk ("immediate", exp, port);
626 break;
627 case scm_tc3_int_1:
628 case scm_tc3_int_2:
629 scm_intprint (SCM_I_INUM (exp), 10, port);
630 break;
631 case scm_tc3_imm24:
632 if (SCM_CHARP (exp))
633 {
634 if (SCM_WRITINGP (pstate))
635 write_character (SCM_CHAR (exp), port, 0);
636 else
637 {
638 if (!display_character (SCM_CHAR (exp), port,
639 PORT_CONVERSION_HANDLER (port)))
640 scm_encoding_error (__func__, errno,
641 "cannot convert to output locale",
642 port, exp);
643 }
644 }
645 else if (SCM_IFLAGP (exp)
646 && ((size_t) SCM_IFLAGNUM (exp) < (sizeof iflagnames / sizeof (char *))))
647 {
648 scm_puts_unlocked (iflagnames [SCM_IFLAGNUM (exp)], port);
649 }
650 else
651 {
652 /* unknown immediate value */
653 scm_ipruk ("immediate", exp, port);
654 }
655 break;
656 case scm_tc3_cons:
657 switch (SCM_TYP7 (exp))
658 {
659 case scm_tcs_struct:
660 {
661 ENTER_NESTED_DATA (pstate, exp, circref);
662 if (SCM_OBJ_CLASS_FLAGS (exp) & SCM_CLASSF_GOOPS)
663 {
664 SCM pwps, print = pstate->writingp ? g_write : g_display;
665 if (SCM_UNPACK (print) == 0)
666 goto print_struct;
667 pwps = scm_i_port_with_print_state (port, pstate->handle);
668 pstate->revealed = 1;
669 scm_call_2 (print, exp, pwps);
670 }
671 else
672 {
673 print_struct:
674 scm_print_struct (exp, port, pstate);
675 }
676 EXIT_NESTED_DATA (pstate);
677 }
678 break;
679 case scm_tcs_cons_imcar:
680 case scm_tcs_cons_nimcar:
681 ENTER_NESTED_DATA (pstate, exp, circref);
682 scm_iprlist ("(", exp, ')', port, pstate);
683 EXIT_NESTED_DATA (pstate);
684 break;
685 circref:
686 print_circref (port, pstate, exp);
687 break;
688 case scm_tc7_number:
689 switch SCM_TYP16 (exp) {
690 case scm_tc16_big:
691 scm_bigprint (exp, port, pstate);
692 break;
693 case scm_tc16_real:
694 scm_print_real (exp, port, pstate);
695 break;
696 case scm_tc16_complex:
697 scm_print_complex (exp, port, pstate);
698 break;
699 case scm_tc16_fraction:
700 scm_i_print_fraction (exp, port, pstate);
701 break;
702 }
703 break;
704 case scm_tc7_stringbuf:
705 scm_i_print_stringbuf (exp, port, pstate);
706 break;
707 case scm_tc7_string:
708 if (SCM_WRITINGP (pstate))
709 {
710 size_t len, i;
711
712 display_character ('"', port, iconveh_question_mark);
713 len = scm_i_string_length (exp);
714 for (i = 0; i < len; ++i)
715 write_character (scm_i_string_ref (exp, i), port, 1);
716
717 display_character ('"', port, iconveh_question_mark);
718 scm_remember_upto_here_1 (exp);
719 }
720 else
721 {
722 size_t len, printed;
723
724 len = scm_i_string_length (exp);
725 printed = display_string (scm_i_string_data (exp),
726 scm_i_is_narrow_string (exp),
727 len, port,
728 PORT_CONVERSION_HANDLER (port));
729 if (SCM_UNLIKELY (printed < len))
730 scm_encoding_error (__func__, errno,
731 "cannot convert to output locale",
732 port, scm_c_string_ref (exp, printed));
733 }
734
735 scm_remember_upto_here_1 (exp);
736 break;
737 case scm_tc7_symbol:
738 if (scm_i_symbol_is_interned (exp))
739 {
740 print_symbol (exp, port);
741 scm_remember_upto_here_1 (exp);
742 }
743 else
744 {
745 scm_puts_unlocked ("#<uninterned-symbol ", port);
746 print_symbol (exp, port);
747 scm_putc_unlocked (' ', port);
748 scm_uintprint (SCM_UNPACK (exp), 16, port);
749 scm_putc_unlocked ('>', port);
750 }
751 break;
752 case scm_tc7_variable:
753 scm_i_variable_print (exp, port, pstate);
754 break;
755 case scm_tc7_program:
756 scm_i_program_print (exp, port, pstate);
757 break;
758 case scm_tc7_pointer:
759 scm_i_pointer_print (exp, port, pstate);
760 break;
761 case scm_tc7_hashtable:
762 scm_i_hashtable_print (exp, port, pstate);
763 break;
764 case scm_tc7_weak_set:
765 scm_i_weak_set_print (exp, port, pstate);
766 break;
767 case scm_tc7_weak_table:
768 scm_i_weak_table_print (exp, port, pstate);
769 break;
770 case scm_tc7_fluid:
771 scm_i_fluid_print (exp, port, pstate);
772 break;
773 case scm_tc7_dynamic_state:
774 scm_i_dynamic_state_print (exp, port, pstate);
775 break;
776 case scm_tc7_frame:
777 scm_i_frame_print (exp, port, pstate);
778 break;
779 case scm_tc7_keyword:
780 scm_puts_unlocked ("#:", port);
781 scm_iprin1 (scm_keyword_to_symbol (exp), port, pstate);
782 break;
783 case scm_tc7_vm_cont:
784 scm_i_vm_cont_print (exp, port, pstate);
785 break;
786 case scm_tc7_array:
787 ENTER_NESTED_DATA (pstate, exp, circref);
788 scm_i_print_array (exp, port, pstate);
789 EXIT_NESTED_DATA (pstate);
790 break;
791 case scm_tc7_bytevector:
792 scm_i_print_bytevector (exp, port, pstate);
793 break;
794 case scm_tc7_bitvector:
795 scm_i_print_bitvector (exp, port, pstate);
796 break;
797 case scm_tc7_wvect:
798 ENTER_NESTED_DATA (pstate, exp, circref);
799 scm_puts_unlocked ("#w(", port);
800 print_vector_or_weak_vector (exp, scm_c_weak_vector_length (exp),
801 scm_c_weak_vector_ref, port, pstate);
802 EXIT_NESTED_DATA (pstate);
803 break;
804 case scm_tc7_vector:
805 ENTER_NESTED_DATA (pstate, exp, circref);
806 scm_puts_unlocked ("#(", port);
807 print_vector_or_weak_vector (exp, SCM_SIMPLE_VECTOR_LENGTH (exp),
808 scm_c_vector_ref, port, pstate);
809 EXIT_NESTED_DATA (pstate);
810 break;
811 case scm_tc7_port:
812 {
813 scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (exp);
814 if (ptob->print && ptob->print (exp, port, pstate))
815 break;
816 goto punk;
817 }
818 case scm_tc7_smob:
819 ENTER_NESTED_DATA (pstate, exp, circref);
820 SCM_SMOB_DESCRIPTOR (exp).print (exp, port, pstate);
821 EXIT_NESTED_DATA (pstate);
822 break;
823 default:
824 /* case scm_tcs_closures: */
825 punk:
826 scm_ipruk ("type", exp, port);
827 }
828 }
829 }
830
831 /* Print states are necessary for circular reference safe printing.
832 * They are also expensive to allocate. Therefore print states are
833 * kept in a pool so that they can be reused.
834 */
835
836 /* The PORT argument can also be a print-state/port pair, which will
837 * then be used instead of allocating a new print state. This is
838 * useful for continuing a chain of print calls from Scheme. */
839
840 void
841 scm_prin1 (SCM exp, SCM port, int writingp)
842 {
843 SCM handle = SCM_BOOL_F; /* Will GC protect the handle whilst unlinked */
844 SCM pstate_scm;
845 scm_print_state *pstate;
846 int old_writingp;
847
848 /* If PORT is a print-state/port pair, use that. Else create a new
849 print-state. */
850
851 if (SCM_PORT_WITH_PS_P (port))
852 {
853 pstate_scm = SCM_PORT_WITH_PS_PS (port);
854 port = SCM_PORT_WITH_PS_PORT (port);
855 }
856 else
857 {
858 /* First try to allocate a print state from the pool */
859 scm_i_pthread_mutex_lock (&print_state_mutex);
860 if (!scm_is_null (print_state_pool))
861 {
862 handle = print_state_pool;
863 print_state_pool = SCM_CDR (print_state_pool);
864 }
865 scm_i_pthread_mutex_unlock (&print_state_mutex);
866 if (scm_is_false (handle))
867 handle = scm_list_1 (make_print_state ());
868 pstate_scm = SCM_CAR (handle);
869 }
870
871 pstate = SCM_PRINT_STATE (pstate_scm);
872 old_writingp = pstate->writingp;
873 pstate->writingp = writingp;
874 scm_iprin1 (exp, port, pstate);
875 pstate->writingp = old_writingp;
876
877 /* Return print state to pool if it has been created above and
878 hasn't escaped to Scheme. */
879
880 if (scm_is_true (handle) && !pstate->revealed)
881 {
882 scm_i_pthread_mutex_lock (&print_state_mutex);
883 SCM_SETCDR (handle, print_state_pool);
884 print_state_pool = handle;
885 scm_i_pthread_mutex_unlock (&print_state_mutex);
886 }
887 }
888
889 /* Convert codepoint CH to UTF-8 and store the result in UTF8. Return
890 the number of bytes of the UTF-8-encoded string. */
891 static size_t
892 codepoint_to_utf8 (scm_t_wchar ch, scm_t_uint8 utf8[4])
893 {
894 size_t len;
895 scm_t_uint32 codepoint;
896
897 codepoint = (scm_t_uint32) ch;
898
899 if (codepoint <= 0x7f)
900 {
901 len = 1;
902 utf8[0] = (scm_t_uint8) codepoint;
903 }
904 else if (codepoint <= 0x7ffUL)
905 {
906 len = 2;
907 utf8[0] = 0xc0 | (codepoint >> 6);
908 utf8[1] = 0x80 | (codepoint & 0x3f);
909 }
910 else if (codepoint <= 0xffffUL)
911 {
912 len = 3;
913 utf8[0] = 0xe0 | (codepoint >> 12);
914 utf8[1] = 0x80 | ((codepoint >> 6) & 0x3f);
915 utf8[2] = 0x80 | (codepoint & 0x3f);
916 }
917 else
918 {
919 len = 4;
920 utf8[0] = 0xf0 | (codepoint >> 18);
921 utf8[1] = 0x80 | ((codepoint >> 12) & 0x3f);
922 utf8[2] = 0x80 | ((codepoint >> 6) & 0x3f);
923 utf8[3] = 0x80 | (codepoint & 0x3f);
924 }
925
926 return len;
927 }
928
929 #define STR_REF(s, x) \
930 (narrow_p \
931 ? (scm_t_wchar) ((unsigned char *) (s))[x] \
932 : ((scm_t_wchar *) (s))[x])
933
934 /* Write STR to PORT as UTF-8. STR is a LEN-codepoint string; it is
935 narrow if NARROW_P is true, wide otherwise. Return LEN. */
936 static size_t
937 display_string_as_utf8 (const void *str, int narrow_p, size_t len,
938 SCM port)
939 {
940 size_t printed = 0;
941
942 while (len > printed)
943 {
944 size_t utf8_len, i;
945 char *input, utf8_buf[256];
946
947 /* Convert STR to UTF-8. */
948 for (i = printed, utf8_len = 0, input = utf8_buf;
949 i < len && utf8_len + 4 < sizeof (utf8_buf);
950 i++)
951 {
952 utf8_len += codepoint_to_utf8 (STR_REF (str, i),
953 (scm_t_uint8 *) input);
954 input = utf8_buf + utf8_len;
955 }
956
957 /* INPUT was successfully converted, entirely; print the
958 result. */
959 scm_lfwrite_unlocked (utf8_buf, utf8_len, port);
960 printed += i - printed;
961 }
962
963 assert (printed == len);
964
965 return len;
966 }
967
968 /* Write STR to PORT as ISO-8859-1. STR is a LEN-codepoint string; it
969 is narrow if NARROW_P is true, wide otherwise. Return LEN. */
970 static size_t
971 display_string_as_latin1 (const void *str, int narrow_p, size_t len,
972 SCM port,
973 scm_t_string_failed_conversion_handler strategy)
974 {
975 size_t printed = 0;
976
977 if (narrow_p)
978 {
979 scm_lfwrite_unlocked (str, len, port);
980 return len;
981 }
982
983 while (printed < len)
984 {
985 char buf[256];
986 size_t i;
987
988 for (i = 0; i < sizeof(buf) && printed < len; i++, printed++)
989 {
990 scm_t_wchar c = STR_REF (str, printed);
991
992 if (c < 256)
993 buf[i] = c;
994 else
995 break;
996 }
997
998 scm_lfwrite_unlocked (buf, i, port);
999
1000 if (i < sizeof(buf) && printed < len)
1001 {
1002 if (strategy == SCM_FAILED_CONVERSION_ERROR)
1003 break;
1004 else if (strategy == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
1005 write_character_escaped (STR_REF (str, printed), 1, port);
1006 else
1007 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
1008 display_string ("?", 1, 1, port, strategy);
1009 printed++;
1010 }
1011 }
1012
1013 return printed;
1014 }
1015
1016 /* Convert STR through PORT's output conversion descriptor and write the
1017 output to PORT. Return the number of codepoints written. */
1018 static size_t
1019 display_string_using_iconv (const void *str, int narrow_p, size_t len,
1020 SCM port,
1021 scm_t_string_failed_conversion_handler strategy)
1022 {
1023 size_t printed;
1024 scm_t_iconv_descriptors *id;
1025 scm_t_port_internal *pti = SCM_PORT_GET_INTERNAL (port);
1026
1027 id = scm_i_port_iconv_descriptors (port, SCM_PORT_WRITE);
1028
1029 if (SCM_UNLIKELY (pti->at_stream_start_for_bom_write && len > 0))
1030 {
1031 scm_t_port *pt = SCM_PTAB_ENTRY (port);
1032
1033 /* Record that we're no longer at stream start. */
1034 pti->at_stream_start_for_bom_write = 0;
1035 if (pt->rw_random)
1036 pti->at_stream_start_for_bom_read = 0;
1037
1038 /* Write a BOM if appropriate. */
1039 if (SCM_UNLIKELY (strcmp(pt->encoding, "UTF-16") == 0
1040 || strcmp(pt->encoding, "UTF-32") == 0))
1041 display_character (SCM_UNICODE_BOM, port, iconveh_error);
1042 }
1043
1044 printed = 0;
1045
1046 while (len > printed)
1047 {
1048 size_t done, utf8_len, input_left, output_left, i;
1049 size_t codepoints_read, output_len;
1050 char *input, *output;
1051 char utf8_buf[256], encoded_output[256];
1052 size_t offsets[256];
1053
1054 /* Convert STR to UTF-8. */
1055 for (i = printed, utf8_len = 0, input = utf8_buf;
1056 i < len && utf8_len + 4 < sizeof (utf8_buf);
1057 i++)
1058 {
1059 offsets[utf8_len] = i;
1060 utf8_len += codepoint_to_utf8 (STR_REF (str, i),
1061 (scm_t_uint8 *) input);
1062 input = utf8_buf + utf8_len;
1063 }
1064
1065 input = utf8_buf;
1066 input_left = utf8_len;
1067
1068 output = encoded_output;
1069 output_left = sizeof (encoded_output);
1070
1071 done = iconv (id->output_cd, &input, &input_left,
1072 &output, &output_left);
1073
1074 output_len = sizeof (encoded_output) - output_left;
1075
1076 if (SCM_UNLIKELY (done == (size_t) -1))
1077 {
1078 int errno_save = errno;
1079
1080 /* Reset the `iconv' state. */
1081 iconv (id->output_cd, NULL, NULL, NULL, NULL);
1082
1083 /* Print the OUTPUT_LEN bytes successfully converted. */
1084 scm_lfwrite_unlocked (encoded_output, output_len, port);
1085
1086 /* See how many input codepoints these OUTPUT_LEN bytes
1087 corresponds to. */
1088 codepoints_read = offsets[input - utf8_buf] - printed;
1089 printed += codepoints_read;
1090
1091 if (errno_save == EILSEQ &&
1092 strategy != SCM_FAILED_CONVERSION_ERROR)
1093 {
1094 /* Conversion failed somewhere in INPUT and we want to
1095 escape or substitute the offending input character. */
1096
1097 if (strategy == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
1098 {
1099 scm_t_wchar ch;
1100
1101 /* Find CH, the offending codepoint, and escape it. */
1102 ch = STR_REF (str, offsets[input - utf8_buf]);
1103 write_character_escaped (ch, 1, port);
1104 }
1105 else
1106 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
1107 display_string ("?", 1, 1, port, strategy);
1108
1109 printed++;
1110 }
1111 else
1112 /* Something bad happened that we can't handle: bail out. */
1113 break;
1114 }
1115 else
1116 {
1117 /* INPUT was successfully converted, entirely; print the
1118 result. */
1119 scm_lfwrite_unlocked (encoded_output, output_len, port);
1120 codepoints_read = i - printed;
1121 printed += codepoints_read;
1122 }
1123 }
1124
1125 return printed;
1126 }
1127
1128 #undef STR_REF
1129
1130 /* Display the LEN codepoints in STR to PORT according to STRATEGY;
1131 return the number of codepoints successfully displayed. If NARROW_P,
1132 then STR is interpreted as a sequence of `char', denoting a Latin-1
1133 string; otherwise it's interpreted as a sequence of
1134 `scm_t_wchar'. */
1135 static size_t
1136 display_string (const void *str, int narrow_p,
1137 size_t len, SCM port,
1138 scm_t_string_failed_conversion_handler strategy)
1139 {
1140 scm_t_port_internal *pti;
1141
1142 pti = SCM_PORT_GET_INTERNAL (port);
1143
1144 if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8)
1145 return display_string_as_utf8 (str, narrow_p, len, port);
1146 else if (pti->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1)
1147 return display_string_as_latin1 (str, narrow_p, len, port, strategy);
1148 else
1149 return display_string_using_iconv (str, narrow_p, len, port, strategy);
1150 }
1151
1152 /* Attempt to display CH to PORT according to STRATEGY. Return non-zero
1153 if CH was successfully displayed, zero otherwise (e.g., if it was not
1154 representable in PORT's encoding.) */
1155 static int
1156 display_character (scm_t_wchar ch, SCM port,
1157 scm_t_string_failed_conversion_handler strategy)
1158 {
1159 return display_string (&ch, 0, 1, port, strategy) == 1;
1160 }
1161
1162 /* Attempt to pretty-print CH, a combining character, to PORT. Return
1163 zero upon failure, non-zero otherwise. The idea is to print CH above
1164 a dotted circle to make it more visible. */
1165 static int
1166 write_combining_character (scm_t_wchar ch, SCM port)
1167 {
1168 scm_t_wchar str[2];
1169
1170 str[0] = SCM_CODEPOINT_DOTTED_CIRCLE;
1171 str[1] = ch;
1172
1173 return display_string (str, 0, 2, port, iconveh_error) == 2;
1174 }
1175
1176 /* Write CH to PORT in its escaped form, using the string escape syntax
1177 if STRING_ESCAPES_P is non-zero. */
1178 static void
1179 write_character_escaped (scm_t_wchar ch, int string_escapes_p, SCM port)
1180 {
1181 if (string_escapes_p)
1182 {
1183 /* Represent CH using the in-string escape syntax. */
1184
1185 static const char hex[] = "0123456789abcdef";
1186 static const char escapes[7] = "abtnvfr";
1187 char buf[9];
1188
1189 if (ch >= 0x07 && ch <= 0x0D && ch != 0x0A)
1190 {
1191 /* Use special escapes for some C0 controls. */
1192 buf[0] = '\\';
1193 buf[1] = escapes[ch - 0x07];
1194 scm_lfwrite_unlocked (buf, 2, port);
1195 }
1196 else if (!SCM_R6RS_ESCAPES_P)
1197 {
1198 if (ch <= 0xFF)
1199 {
1200 buf[0] = '\\';
1201 buf[1] = 'x';
1202 buf[2] = hex[ch / 16];
1203 buf[3] = hex[ch % 16];
1204 scm_lfwrite_unlocked (buf, 4, port);
1205 }
1206 else if (ch <= 0xFFFF)
1207 {
1208 buf[0] = '\\';
1209 buf[1] = 'u';
1210 buf[2] = hex[(ch & 0xF000) >> 12];
1211 buf[3] = hex[(ch & 0xF00) >> 8];
1212 buf[4] = hex[(ch & 0xF0) >> 4];
1213 buf[5] = hex[(ch & 0xF)];
1214 scm_lfwrite_unlocked (buf, 6, port);
1215 }
1216 else if (ch > 0xFFFF)
1217 {
1218 buf[0] = '\\';
1219 buf[1] = 'U';
1220 buf[2] = hex[(ch & 0xF00000) >> 20];
1221 buf[3] = hex[(ch & 0xF0000) >> 16];
1222 buf[4] = hex[(ch & 0xF000) >> 12];
1223 buf[5] = hex[(ch & 0xF00) >> 8];
1224 buf[6] = hex[(ch & 0xF0) >> 4];
1225 buf[7] = hex[(ch & 0xF)];
1226 scm_lfwrite_unlocked (buf, 8, port);
1227 }
1228 }
1229 else
1230 {
1231 /* Print an R6RS variable-length hex escape: "\xNNNN;". */
1232 scm_t_wchar ch2 = ch;
1233
1234 int i = 8;
1235 buf[i] = ';';
1236 i --;
1237 if (ch == 0)
1238 buf[i--] = '0';
1239 else
1240 while (ch2 > 0)
1241 {
1242 buf[i] = hex[ch2 & 0xF];
1243 ch2 >>= 4;
1244 i --;
1245 }
1246 buf[i] = 'x';
1247 i --;
1248 buf[i] = '\\';
1249 scm_lfwrite_unlocked (buf + i, 9 - i, port);
1250 }
1251 }
1252 else
1253 {
1254 /* Represent CH using the character escape syntax. */
1255 const char *name;
1256
1257 name = scm_i_charname (SCM_MAKE_CHAR (ch));
1258 if (name != NULL)
1259 scm_puts_unlocked (name, port);
1260 else
1261 PRINT_CHAR_ESCAPE (ch, port);
1262 }
1263 }
1264
1265 /* Write CH to PORT, escaping it if it's non-graphic or not
1266 representable in PORT's encoding. If STRING_ESCAPES_P is true and CH
1267 needs to be escaped, it is escaped using the in-string escape syntax;
1268 otherwise the character escape syntax is used. */
1269 static void
1270 write_character (scm_t_wchar ch, SCM port, int string_escapes_p)
1271 {
1272 int printed = 0;
1273 scm_t_string_failed_conversion_handler strategy;
1274
1275 strategy = PORT_CONVERSION_HANDLER (port);
1276
1277 if (string_escapes_p)
1278 {
1279 /* Check if CH deserves special treatment. */
1280 if (ch == '"' || ch == '\\')
1281 {
1282 display_character ('\\', port, iconveh_question_mark);
1283 display_character (ch, port, strategy);
1284 printed = 1;
1285 }
1286 else if (ch == '\n' && SCM_PRINT_ESCAPE_NEWLINES_P)
1287 {
1288 display_character ('\\', port, iconveh_question_mark);
1289 display_character ('n', port, strategy);
1290 printed = 1;
1291 }
1292 else if (ch == ' ' || ch == '\n')
1293 {
1294 display_character (ch, port, strategy);
1295 printed = 1;
1296 }
1297 }
1298 else
1299 {
1300 display_string ("#\\", 1, 2, port, iconveh_question_mark);
1301
1302 if (uc_combining_class (ch) != UC_CCC_NR)
1303 /* Character is a combining character, so attempt to
1304 pretty-print it. */
1305 printed = write_combining_character (ch, port);
1306 }
1307
1308 if (!printed
1309 && uc_is_general_category_withtable (ch,
1310 UC_CATEGORY_MASK_L |
1311 UC_CATEGORY_MASK_M |
1312 UC_CATEGORY_MASK_N |
1313 UC_CATEGORY_MASK_P |
1314 UC_CATEGORY_MASK_S))
1315 /* CH is graphic; attempt to display it. */
1316 printed = display_character (ch, port, iconveh_error);
1317
1318 if (!printed)
1319 /* CH isn't graphic or cannot be represented in PORT's encoding. */
1320 write_character_escaped (ch, string_escapes_p, port);
1321 }
1322
1323 /* Display STR to PORT from START inclusive to END exclusive. */
1324 void
1325 scm_i_display_substring (SCM str, size_t start, size_t end, SCM port)
1326 {
1327 int narrow_p;
1328 const char *buf;
1329 size_t len, printed;
1330
1331 buf = scm_i_string_data (str);
1332 len = end - start;
1333 narrow_p = scm_i_is_narrow_string (str);
1334 buf += start * (narrow_p ? sizeof (char) : sizeof (scm_t_wchar));
1335
1336 printed = display_string (buf, narrow_p, end - start, port,
1337 PORT_CONVERSION_HANDLER (port));
1338
1339 if (SCM_UNLIKELY (printed < len))
1340 scm_encoding_error (__func__, errno,
1341 "cannot convert to output locale",
1342 port, scm_c_string_ref (str, printed + start));
1343 }
1344
1345 \f
1346 /* Print an integer.
1347 */
1348
1349 void
1350 scm_intprint (scm_t_intmax n, int radix, SCM port)
1351 {
1352 char num_buf[SCM_INTBUFLEN];
1353 scm_lfwrite_unlocked (num_buf, scm_iint2str (n, radix, num_buf), port);
1354 }
1355
1356 void
1357 scm_uintprint (scm_t_uintmax n, int radix, SCM port)
1358 {
1359 char num_buf[SCM_INTBUFLEN];
1360 scm_lfwrite_unlocked (num_buf, scm_iuint2str (n, radix, num_buf), port);
1361 }
1362
1363 /* Print an object of unrecognized type.
1364 */
1365
1366 void
1367 scm_ipruk (char *hdr, SCM ptr, SCM port)
1368 {
1369 scm_puts_unlocked ("#<unknown-", port);
1370 scm_puts_unlocked (hdr, port);
1371 if (1) /* (scm_in_heap_p (ptr)) */ /* FIXME */
1372 {
1373 scm_puts_unlocked (" (0x", port);
1374 scm_uintprint (SCM_CELL_WORD_0 (ptr), 16, port);
1375 scm_puts_unlocked (" . 0x", port);
1376 scm_uintprint (SCM_CELL_WORD_1 (ptr), 16, port);
1377 scm_puts_unlocked (") @", port);
1378 }
1379 scm_puts_unlocked (" 0x", port);
1380 scm_uintprint (SCM_UNPACK (ptr), 16, port);
1381 scm_putc_unlocked ('>', port);
1382 }
1383
1384
1385 /* Print a list.
1386 */
1387 void
1388 scm_iprlist (char *hdr, SCM exp, int tlr, SCM port, scm_print_state *pstate)
1389 {
1390 register SCM hare, tortoise;
1391 long floor = pstate->top - 2;
1392 scm_puts_unlocked (hdr, port);
1393 /* CHECK_INTS; */
1394 if (pstate->fancyp)
1395 goto fancy_printing;
1396
1397 /* Run a hare and tortoise so that total time complexity will be
1398 O(depth * N) instead of O(N^2). */
1399 hare = SCM_CDR (exp);
1400 tortoise = exp;
1401 while (scm_is_pair (hare))
1402 {
1403 if (scm_is_eq (hare, tortoise))
1404 goto fancy_printing;
1405 hare = SCM_CDR (hare);
1406 if (!scm_is_pair (hare))
1407 break;
1408 hare = SCM_CDR (hare);
1409 tortoise = SCM_CDR (tortoise);
1410 }
1411
1412 /* No cdr cycles intrinsic to this list */
1413 scm_iprin1 (SCM_CAR (exp), port, pstate);
1414 for (exp = SCM_CDR (exp); scm_is_pair (exp); exp = SCM_CDR (exp))
1415 {
1416 register long i;
1417
1418 for (i = floor; i >= 0; --i)
1419 if (scm_is_eq (PSTATE_STACK_REF(pstate, i), exp))
1420 goto circref;
1421 PUSH_REF (pstate, exp);
1422 scm_putc_unlocked (' ', port);
1423 /* CHECK_INTS; */
1424 scm_iprin1 (SCM_CAR (exp), port, pstate);
1425 }
1426 if (!SCM_NULL_OR_NIL_P (exp))
1427 {
1428 scm_puts_unlocked (" . ", port);
1429 scm_iprin1 (exp, port, pstate);
1430 }
1431
1432 end:
1433 scm_putc_unlocked (tlr, port);
1434 pstate->top = floor + 2;
1435 return;
1436
1437 fancy_printing:
1438 {
1439 long n = pstate->length;
1440
1441 scm_iprin1 (SCM_CAR (exp), port, pstate);
1442 exp = SCM_CDR (exp); --n;
1443 for (; scm_is_pair (exp); exp = SCM_CDR (exp))
1444 {
1445 register unsigned long i;
1446
1447 for (i = 0; i < pstate->top; ++i)
1448 if (scm_is_eq (PSTATE_STACK_REF(pstate, i), exp))
1449 goto fancy_circref;
1450 if (pstate->fancyp)
1451 {
1452 if (n == 0)
1453 {
1454 scm_puts_unlocked (" ...", port);
1455 goto skip_tail;
1456 }
1457 else
1458 --n;
1459 }
1460 PUSH_REF(pstate, exp);
1461 ++pstate->list_offset;
1462 scm_putc_unlocked (' ', port);
1463 /* CHECK_INTS; */
1464 scm_iprin1 (SCM_CAR (exp), port, pstate);
1465 }
1466 }
1467 if (!SCM_NULL_OR_NIL_P (exp))
1468 {
1469 scm_puts_unlocked (" . ", port);
1470 scm_iprin1 (exp, port, pstate);
1471 }
1472 skip_tail:
1473 pstate->list_offset -= pstate->top - floor - 2;
1474 goto end;
1475
1476 fancy_circref:
1477 pstate->list_offset -= pstate->top - floor - 2;
1478
1479 circref:
1480 scm_puts_unlocked (" . ", port);
1481 print_circref (port, pstate, exp);
1482 goto end;
1483 }
1484
1485 \f
1486
1487 int
1488 scm_valid_oport_value_p (SCM val)
1489 {
1490 return (SCM_OPOUTPORTP (val)
1491 || (SCM_PORT_WITH_PS_P (val)
1492 && SCM_OPOUTPORTP (SCM_PORT_WITH_PS_PORT (val))));
1493 }
1494
1495 /* SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write); */
1496
1497 SCM
1498 scm_write (SCM obj, SCM port)
1499 {
1500 if (SCM_UNBNDP (port))
1501 port = scm_current_output_port ();
1502
1503 SCM_ASSERT (scm_valid_oport_value_p (port), port, SCM_ARG2, s_write);
1504
1505 scm_dynwind_begin (0);
1506 scm_dynwind_lock_port (SCM_COERCE_OUTPORT (port));
1507 scm_prin1 (obj, port, 1);
1508 scm_dynwind_end ();
1509
1510 return SCM_UNSPECIFIED;
1511 }
1512
1513
1514 /* SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display); */
1515
1516 SCM
1517 scm_display (SCM obj, SCM port)
1518 {
1519 if (SCM_UNBNDP (port))
1520 port = scm_current_output_port ();
1521
1522 SCM_ASSERT (scm_valid_oport_value_p (port), port, SCM_ARG2, s_display);
1523
1524 scm_dynwind_begin (0);
1525 scm_dynwind_lock_port (SCM_COERCE_OUTPORT (port));
1526 scm_prin1 (obj, port, 0);
1527 scm_dynwind_end ();
1528
1529 return SCM_UNSPECIFIED;
1530 }
1531
1532
1533 SCM_DEFINE (scm_simple_format, "simple-format", 2, 0, 1,
1534 (SCM destination, SCM message, SCM args),
1535 "Write @var{message} to @var{destination}, defaulting to\n"
1536 "the current output port.\n"
1537 "@var{message} can contain @code{~A} (was @code{%s}) and\n"
1538 "@code{~S} (was @code{%S}) escapes. When printed,\n"
1539 "the escapes are replaced with corresponding members of\n"
1540 "@var{args}:\n"
1541 "@code{~A} formats using @code{display} and @code{~S} formats\n"
1542 "using @code{write}.\n"
1543 "If @var{destination} is @code{#t}, then use the current output\n"
1544 "port, if @var{destination} is @code{#f}, then return a string\n"
1545 "containing the formatted text. Does not add a trailing newline.")
1546 #define FUNC_NAME s_scm_simple_format
1547 {
1548 SCM port, answer = SCM_UNSPECIFIED;
1549 int fReturnString = 0;
1550 int writingp;
1551 size_t start, p, end;
1552
1553 if (scm_is_eq (destination, SCM_BOOL_T))
1554 {
1555 destination = port = scm_current_output_port ();
1556 SCM_VALIDATE_OPORT_VALUE (1, destination);
1557 }
1558 else if (scm_is_false (destination))
1559 {
1560 fReturnString = 1;
1561 port = scm_mkstrport (SCM_INUM0, SCM_BOOL_F,
1562 SCM_OPN | SCM_WRTNG,
1563 FUNC_NAME);
1564 destination = port;
1565 }
1566 else
1567 {
1568 SCM_VALIDATE_OPORT_VALUE (1, destination);
1569 port = SCM_COERCE_OUTPORT (destination);
1570 }
1571 SCM_VALIDATE_STRING (2, message);
1572 SCM_VALIDATE_REST_ARGUMENT (args);
1573
1574 p = 0;
1575 start = 0;
1576 end = scm_i_string_length (message);
1577 for (p = start; p != end; ++p)
1578 if (scm_i_string_ref (message, p) == '~')
1579 {
1580 if (++p == end)
1581 break;
1582
1583 switch (scm_i_string_ref (message, p))
1584 {
1585 case 'A': case 'a':
1586 writingp = 0;
1587 break;
1588 case 'S': case 's':
1589 writingp = 1;
1590 break;
1591 case '~':
1592 scm_lfwrite_substr (message, start, p, port);
1593 start = p + 1;
1594 continue;
1595 case '%':
1596 scm_lfwrite_substr (message, start, p - 1, port);
1597 scm_newline (port);
1598 start = p + 1;
1599 continue;
1600 default:
1601 SCM_MISC_ERROR ("FORMAT: Unsupported format option ~~~A - use (ice-9 format) instead",
1602 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message, p))));
1603
1604 }
1605
1606
1607 if (!scm_is_pair (args))
1608 SCM_MISC_ERROR ("FORMAT: Missing argument for ~~~A",
1609 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message, p))));
1610
1611 scm_lfwrite_substr (message, start, p - 1, port);
1612 /* we pass destination here */
1613 scm_prin1 (SCM_CAR (args), destination, writingp);
1614 args = SCM_CDR (args);
1615 start = p + 1;
1616 }
1617
1618 scm_lfwrite_substr (message, start, p, port);
1619 if (!scm_is_eq (args, SCM_EOL))
1620 SCM_MISC_ERROR ("FORMAT: ~A superfluous arguments",
1621 scm_list_1 (scm_length (args)));
1622
1623 if (fReturnString)
1624 answer = scm_strport_to_string (destination);
1625
1626 return scm_return_first (answer, message);
1627 }
1628 #undef FUNC_NAME
1629
1630
1631 SCM_DEFINE (scm_newline, "newline", 0, 1, 0,
1632 (SCM port),
1633 "Send a newline to @var{port}.\n"
1634 "If @var{port} is omitted, send to the current output port.")
1635 #define FUNC_NAME s_scm_newline
1636 {
1637 if (SCM_UNBNDP (port))
1638 port = scm_current_output_port ();
1639
1640 SCM_VALIDATE_OPORT_VALUE (1, port);
1641
1642 scm_putc_unlocked ('\n', SCM_COERCE_OUTPORT (port));
1643 return SCM_UNSPECIFIED;
1644 }
1645 #undef FUNC_NAME
1646
1647 SCM_DEFINE (scm_write_char, "write-char", 1, 1, 0,
1648 (SCM chr, SCM port),
1649 "Send character @var{chr} to @var{port}.")
1650 #define FUNC_NAME s_scm_write_char
1651 {
1652 if (SCM_UNBNDP (port))
1653 port = scm_current_output_port ();
1654
1655 SCM_VALIDATE_CHAR (1, chr);
1656 SCM_VALIDATE_OPORT_VALUE (2, port);
1657
1658 port = SCM_COERCE_OUTPORT (port);
1659 if (!display_character (SCM_CHAR (chr), port,
1660 PORT_CONVERSION_HANDLER (port)))
1661 scm_encoding_error (__func__, errno,
1662 "cannot convert to output locale",
1663 port, chr);
1664
1665 return SCM_UNSPECIFIED;
1666 }
1667 #undef FUNC_NAME
1668
1669 \f
1670
1671 /* Call back to Scheme code to do the printing of special objects
1672 * (like structs). SCM_PRINTER_APPLY applies PROC to EXP and a smob
1673 * containing PORT and PSTATE. This object can be used as the port for
1674 * display/write etc to continue the current print chain. The REVEALED
1675 * field of PSTATE is set to true to indicate that the print state has
1676 * escaped to Scheme and thus has to be freed by the GC.
1677 */
1678
1679 scm_t_bits scm_tc16_port_with_ps;
1680
1681 /* Print exactly as the port itself would */
1682
1683 static int
1684 port_with_ps_print (SCM obj, SCM port, scm_print_state *pstate)
1685 {
1686 obj = SCM_PORT_WITH_PS_PORT (obj);
1687 return SCM_PORT_DESCRIPTOR (obj)->print (obj, port, pstate);
1688 }
1689
1690 SCM
1691 scm_printer_apply (SCM proc, SCM exp, SCM port, scm_print_state *pstate)
1692 {
1693 pstate->revealed = 1;
1694 return scm_call_2 (proc, exp,
1695 scm_i_port_with_print_state (port, pstate->handle));
1696 }
1697
1698 SCM_DEFINE (scm_port_with_print_state, "port-with-print-state", 1, 1, 0,
1699 (SCM port, SCM pstate),
1700 "Create a new port which behaves like @var{port}, but with an\n"
1701 "included print state @var{pstate}. @var{pstate} is optional.\n"
1702 "If @var{pstate} isn't supplied and @var{port} already has\n"
1703 "a print state, the old print state is reused.")
1704 #define FUNC_NAME s_scm_port_with_print_state
1705 {
1706 SCM_VALIDATE_OPORT_VALUE (1, port);
1707 if (!SCM_UNBNDP (pstate))
1708 SCM_VALIDATE_PRINTSTATE (2, pstate);
1709 return scm_i_port_with_print_state (port, pstate);
1710 }
1711 #undef FUNC_NAME
1712
1713 SCM_DEFINE (scm_get_print_state, "get-print-state", 1, 0, 0,
1714 (SCM port),
1715 "Return the print state of the port @var{port}. If @var{port}\n"
1716 "has no associated print state, @code{#f} is returned.")
1717 #define FUNC_NAME s_scm_get_print_state
1718 {
1719 if (SCM_PORT_WITH_PS_P (port))
1720 return SCM_PORT_WITH_PS_PS (port);
1721 if (SCM_OUTPUT_PORT_P (port))
1722 return SCM_BOOL_F;
1723 SCM_WRONG_TYPE_ARG (1, port);
1724 }
1725 #undef FUNC_NAME
1726
1727 \f
1728
1729 void
1730 scm_init_print ()
1731 {
1732 SCM type;
1733
1734 type = scm_make_vtable (scm_from_locale_string (SCM_PRINT_STATE_LAYOUT),
1735 SCM_BOOL_F);
1736 scm_set_struct_vtable_name_x (type, scm_from_latin1_symbol ("print-state"));
1737 scm_print_state_vtable = type;
1738
1739 /* Don't want to bind a wrapper class in GOOPS, so pass 0 as arg1. */
1740 scm_tc16_port_with_ps = scm_make_smob_type (0, 0);
1741 scm_set_smob_print (scm_tc16_port_with_ps, port_with_ps_print);
1742
1743 #include "libguile/print.x"
1744
1745 scm_init_opts (scm_print_options, scm_print_opts);
1746 scm_print_opts[SCM_PRINT_HIGHLIGHT_PREFIX_I].val =
1747 SCM_UNPACK (scm_from_locale_string ("{"));
1748 scm_print_opts[SCM_PRINT_HIGHLIGHT_SUFFIX_I].val =
1749 SCM_UNPACK (scm_from_locale_string ("}"));
1750 scm_print_opts[SCM_PRINT_KEYWORD_STYLE_I].val = SCM_UNPACK (sym_reader);
1751 }
1752
1753 /*
1754 Local Variables:
1755 c-file-style: "gnu"
1756 End:
1757 */