port i/o optimizations for iso-8859-1
[bpt/guile.git] / libguile / print.c
CommitLineData
f4bc4e59 1/* Copyright (C) 1995-1999, 2000, 2001, 2002, 2003, 2004, 2006, 2008,
88c0a1d5 2 * 2009, 2010, 2011, 2012 Free Software Foundation, Inc.
f4bc4e59 3 *
73be1d9e 4 * This library is free software; you can redistribute it and/or
53befeb7
NJ
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.
0f2d19dd 8 *
53befeb7
NJ
9 * This library is distributed in the hope that it will be useful, but
10 * WITHOUT ANY WARRANTY; without even the implied warranty of
73be1d9e
MV
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * Lesser General Public License for more details.
0f2d19dd 13 *
73be1d9e
MV
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
53befeb7
NJ
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
17 * 02110-1301 USA
73be1d9e 18 */
1bbd0b84 19
1bbd0b84 20
0f2d19dd 21\f
dbb605f5
LC
22#ifdef HAVE_CONFIG_H
23# include <config.h>
24#endif
0f2d19dd 25
e6e2e95a 26#include <errno.h>
f4bc4e59
LC
27#include <iconv.h>
28#include <stdio.h>
29#include <assert.h>
30
eca29b02 31#include <uniconv.h>
904a78f1 32#include <unictype.h>
e6e2e95a 33
a0599745
MD
34#include "libguile/_scm.h"
35#include "libguile/chars.h"
a002f1a2 36#include "libguile/continuations.h"
a0599745 37#include "libguile/smob.h"
bbb2ecd1 38#include "libguile/control.h"
a0599745
MD
39#include "libguile/eval.h"
40#include "libguile/macros.h"
41#include "libguile/procprop.h"
42#include "libguile/read.h"
2fb924f6 43#include "libguile/programs.h"
a0599745
MD
44#include "libguile/alist.h"
45#include "libguile/struct.h"
a0599745
MD
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"
327967ef 51#include "libguile/numbers.h"
6f3b0cc2 52#include "libguile/vm.h"
a0599745
MD
53
54#include "libguile/validate.h"
55#include "libguile/print.h"
22fc179a
HWN
56
57#include "libguile/private-options.h"
58
0f2d19dd
JB
59\f
60
07f49ac7
LC
61/* Character printers. */
62
f4bc4e59
LC
63static size_t display_string (const void *, int, size_t, SCM,
64 scm_t_string_failed_conversion_handler);
65
07f49ac7
LC
66static int display_character (scm_t_wchar, SCM,
67 scm_t_string_failed_conversion_handler);
f4bc4e59 68
07f49ac7
LC
69static void write_character (scm_t_wchar, SCM, int);
70
f4bc4e59
LC
71static void write_character_escaped (scm_t_wchar, int, SCM);
72
07f49ac7
LC
73\f
74
0f2d19dd
JB
75/* {Names of immediate symbols}
76 *
77 * This table must agree with the declarations in scm.h: {Immediate Symbols}.
78 */
79
e17d318f
DH
80/* This table must agree with the list of flags in tags.h. */
81static const char *iflagnames[] =
82{
83 "#f",
45f4cbdf
MW
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>",
86 "()",
e17d318f 87 "#t",
f60c2c4e
MW
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>",
45f4cbdf 91 "#<unspecified>",
e17d318f
DH
92 "#<undefined>",
93 "#<eof>",
e17d318f
DH
94
95 /* Unbound slot marker for GOOPS. For internal use in GOOPS only. */
96 "#<unbound>",
e17d318f
DH
97};
98
475fa9a5
MV
99SCM_SYMBOL (sym_reader, "reader");
100
92c2555f 101scm_t_option scm_print_opts[] = {
210c0325 102 { SCM_OPTION_SCM, "highlight-prefix", (scm_t_bits)SCM_BOOL_F_BITS,
81ae25da 103 "The string to print before highlighted values." },
210c0325 104 { SCM_OPTION_SCM, "highlight-suffix", (scm_t_bits)SCM_BOOL_F_BITS,
475fa9a5 105 "The string to print after highlighted values." },
210c0325 106 { SCM_OPTION_SCM, "quote-keywordish-symbols", (scm_t_bits)SCM_BOOL_F_BITS,
475fa9a5
MV
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; "
8500b186
AW
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'." },
62560650 112 { 0 },
e6e4c9af
MD
113};
114
a1ec6916 115SCM_DEFINE (scm_print_options, "print-options-interface", 0, 1, 0,
1bbd0b84 116 (SCM setting),
71331188 117 "Option interface for the print options. Instead of using\n"
1dd05fd8
MG
118 "this procedure directly, use the procedures\n"
119 "@code{print-enable}, @code{print-disable}, @code{print-set!}\n"
120 "and @code{print-options}.")
1bbd0b84 121#define FUNC_NAME s_scm_print_options
e6e4c9af 122{
a51ea417 123 SCM ans = scm_options (setting,
b7ff98dd 124 scm_print_opts,
1bbd0b84 125 FUNC_NAME);
e6e4c9af
MD
126 return ans;
127}
1bbd0b84 128#undef FUNC_NAME
e6e4c9af 129
0f2d19dd
JB
130\f
131/* {Printing of Scheme Objects}
132 */
133
a51ea417 134/* Detection of circular references.
c62fbfe1
MD
135 *
136 * Due to other constraints in the implementation, this code has bad
5d46ebe3
MD
137 * time complexity (O (depth * N)), The printer code can be
138 * rewritten to be O(N).
a51ea417 139 */
dbb5de29
NJ
140#define PUSH_REF(pstate, obj) \
141do \
142{ \
143 PSTATE_STACK_SET (pstate, pstate->top, obj); \
144 pstate->top++; \
145 if (pstate->top == pstate->ceiling) \
146 grow_ref_stack (pstate); \
1bbd0b84 147} while(0)
a51ea417 148
dbb5de29
NJ
149#define ENTER_NESTED_DATA(pstate, obj, label) \
150do \
151{ \
152 register unsigned long i; \
153 for (i = 0; i < pstate->top; ++i) \
154 if (scm_is_eq (PSTATE_STACK_REF (pstate, i), (obj))) \
155 goto label; \
156 if (pstate->fancyp) \
157 { \
158 if (pstate->top - pstate->list_offset >= pstate->level) \
159 { \
0607ebbf 160 scm_putc_unlocked ('#', port); \
dbb5de29
NJ
161 return; \
162 } \
163 } \
164 PUSH_REF(pstate, obj); \
1bbd0b84 165} while(0)
a51ea417 166
dbb5de29
NJ
167#define EXIT_NESTED_DATA(pstate) \
168do \
169{ \
170 --pstate->top; \
171 PSTATE_STACK_SET (pstate, pstate->top, SCM_UNDEFINED); \
172} \
173while (0)
c62fbfe1 174
d5cf5324
DH
175SCM scm_print_state_vtable = SCM_BOOL_F;
176static SCM print_state_pool = SCM_EOL;
9de87eea 177scm_i_pthread_mutex_t print_state_mutex = SCM_I_PTHREAD_MUTEX_INITIALIZER;
c4f37e80 178
f843a84c 179#ifdef GUILE_DEBUG /* Used for debugging purposes */
1cc91f1b 180
3b3b36dd 181SCM_DEFINE (scm_current_pstate, "current-pstate", 0, 0, 0,
1bbd0b84 182 (),
d5cf5324 183 "Return the current-pstate -- the car of the\n"
5352393c
MG
184 "@code{print_state_pool}. @code{current-pstate} is only\n"
185 "included in @code{--enable-guile-debug} builds.")
1bbd0b84 186#define FUNC_NAME s_scm_current_pstate
c62fbfe1 187{
d2e53ed6 188 if (!scm_is_null (print_state_pool))
d5cf5324 189 return SCM_CAR (print_state_pool);
a0adfbf0 190 else
0a284a4e 191 return SCM_BOOL_F;
c62fbfe1 192}
1bbd0b84
GB
193#undef FUNC_NAME
194
c62fbfe1
MD
195#endif
196
197#define PSTATE_SIZE 50L
198
698c0295 199static SCM
1bbd0b84 200make_print_state (void)
698c0295 201{
d5cf5324
DH
202 SCM print_state
203 = scm_make_struct (scm_print_state_vtable, SCM_INUM0, SCM_EOL);
bf685b6d 204 scm_print_state *pstate = SCM_PRINT_STATE (print_state);
00ffa0e7 205 pstate->ref_vect = scm_c_make_vector (PSTATE_SIZE, SCM_UNDEFINED);
4057a3e0 206 pstate->ceiling = SCM_SIMPLE_VECTOR_LENGTH (pstate->ref_vect);
d232520a 207 pstate->highlight_objects = SCM_EOL;
698c0295
MD
208 return print_state;
209}
1cc91f1b 210
c62fbfe1
MD
211SCM
212scm_make_print_state ()
c62fbfe1 213{
230d095f 214 SCM answer = SCM_BOOL_F;
698c0295
MD
215
216 /* First try to allocate a print state from the pool */
9de87eea 217 scm_i_pthread_mutex_lock (&print_state_mutex);
d2e53ed6 218 if (!scm_is_null (print_state_pool))
698c0295 219 {
d5cf5324
DH
220 answer = SCM_CAR (print_state_pool);
221 print_state_pool = SCM_CDR (print_state_pool);
698c0295 222 }
9de87eea 223 scm_i_pthread_mutex_unlock (&print_state_mutex);
698c0295 224
7888309b 225 return scm_is_false (answer) ? make_print_state () : answer;
c62fbfe1 226}
a51ea417 227
698c0295 228void
6e8d25a6 229scm_free_print_state (SCM print_state)
698c0295
MD
230{
231 SCM handle;
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
236 * often.
237 */
238 pstate->fancyp = 0;
bb35f315 239 pstate->revealed = 0;
d232520a 240 pstate->highlight_objects = SCM_EOL;
9de87eea 241 scm_i_pthread_mutex_lock (&print_state_mutex);
16d4699b 242 handle = scm_cons (print_state, print_state_pool);
d5cf5324 243 print_state_pool = handle;
9de87eea 244 scm_i_pthread_mutex_unlock (&print_state_mutex);
dfd03fb9
MD
245}
246
247SCM
248scm_i_port_with_print_state (SCM port, SCM print_state)
249{
250 if (SCM_UNBNDP (print_state))
251 {
252 if (SCM_PORT_WITH_PS_P (port))
253 return port;
254 else
255 print_state = scm_make_print_state ();
256 /* port does not need to be coerced since it doesn't have ps */
257 }
258 else
259 port = SCM_COERCE_OUTPORT (port);
260 SCM_RETURN_NEWSMOB (scm_tc16_port_with_ps,
261 SCM_UNPACK (scm_cons (port, print_state)));
698c0295 262}
1cc91f1b 263
a51ea417 264static void
1bbd0b84 265grow_ref_stack (scm_print_state *pstate)
a51ea417 266{
4057a3e0
MV
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;
00ffa0e7 270 SCM new_vect = scm_c_make_vector (new_size, SCM_UNDEFINED);
b17004b8
DH
271 unsigned long int i;
272
273 for (i = 0; i != old_size; ++i)
4057a3e0 274 SCM_SIMPLE_VECTOR_SET (new_vect, i, SCM_SIMPLE_VECTOR_REF (old_vect, i));
b17004b8
DH
275
276 pstate->ref_vect = new_vect;
bf685b6d 277 pstate->ceiling = new_size;
a51ea417
MD
278}
279
509759dd
MV
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))
1cc91f1b 282
a51ea417 283static void
34d19ef6 284print_circref (SCM port, scm_print_state *pstate, SCM ref)
a51ea417 285{
c014a02e
ML
286 register long i;
287 long self = pstate->top - 1;
c62fbfe1 288 i = pstate->top - 1;
509759dd 289 if (scm_is_pair (PSTATE_STACK_REF (pstate, i)))
c62fbfe1
MD
290 {
291 while (i > 0)
292 {
509759dd
MV
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))))
c62fbfe1
MD
296 break;
297 --i;
298 }
299 self = i;
300 }
301 for (i = pstate->top - 1; 1; --i)
509759dd 302 if (scm_is_eq (PSTATE_STACK_REF(pstate, i), ref))
c62fbfe1 303 break;
0607ebbf 304 scm_putc_unlocked ('#', port);
c62fbfe1 305 scm_intprint (i - self, 10, port);
0607ebbf 306 scm_putc_unlocked ('#', port);
a51ea417
MD
307}
308
6662998f
MV
309/* Print the name of a symbol. */
310
475fa9a5 311static int
15671c6e 312quote_keywordish_symbols (void)
475fa9a5 313{
15671c6e 314 SCM option = SCM_PRINT_KEYWORD_STYLE;
475fa9a5 315
475fa9a5
MV
316 if (scm_is_false (option))
317 return 0;
318 if (scm_is_eq (option, sym_reader))
319 return scm_is_true (SCM_PACK (SCM_KEYWORD_STYLE));
320 return 1;
321}
322
2e9fc9fc
AW
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)
330
331#define SUBSEQUENT_IDENTIFIER_MASK \
332 (INITIAL_IDENTIFIER_MASK \
333 | UC_CATEGORY_MASK_Nd | UC_CATEGORY_MASK_Mc | UC_CATEGORY_MASK_Me)
334
15671c6e
AW
335static int
336symbol_has_extended_read_syntax (SCM sym)
6662998f 337{
15671c6e
AW
338 size_t pos, len = scm_i_symbol_length (sym);
339 scm_t_wchar c;
340
341 /* The empty symbol. */
342 if (len == 0)
343 return 1;
344
345 c = scm_i_symbol_ref (sym, 0);
346
347 /* Single dot; conflicts with dotted-pair notation. */
348 if (len == 1 && c == '.')
349 return 1;
350
351 /* Other initial-character constraints. */
2e9fc9fc 352 if (c == '\'' || c == '`' || c == ',' || c == '"' || c == ';' || c == '#')
15671c6e
AW
353 return 1;
354
355 /* Keywords can be identified by trailing colons too. */
356 if (c == ':' || scm_i_symbol_ref (sym, len - 1) == ':')
357 return quote_keywordish_symbols ();
358
359 /* Number-ish symbols. */
360 if (scm_is_true (scm_i_string_to_number (scm_symbol_to_string (sym), 10)))
361 return 1;
362
2e9fc9fc
AW
363 /* Other disallowed first characters. */
364 if (!uc_is_general_category_withtable (c, INITIAL_IDENTIFIER_MASK))
365 return 1;
366
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++)
6662998f 371 {
2e9fc9fc
AW
372 c = scm_i_symbol_ref (sym, pos);
373 if (!uc_is_general_category_withtable (c, SUBSEQUENT_IDENTIFIER_MASK))
374 return 1;
375 else if (c == '"' || c == ';' || c == '#')
376 return 1;
6662998f 377 }
c6b49e89 378
15671c6e
AW
379 return 0;
380}
381
382static void
383print_normal_symbol (SCM sym, SCM port)
384{
385 scm_display (scm_symbol_to_string (sym), port);
386}
387
15671c6e
AW
388static void
389print_extended_symbol (SCM sym, SCM port)
390{
391 size_t pos, len;
392 scm_t_string_failed_conversion_handler strategy;
393
394 len = scm_i_symbol_length (sym);
395 strategy = scm_i_get_conversion_strategy (port);
396
f209aeee 397 scm_lfwrite_unlocked ("#{", 2, port);
15671c6e
AW
398
399 for (pos = 0; pos < len; pos++)
400 {
401 scm_t_wchar c = scm_i_symbol_ref (sym, pos);
402
2e9fc9fc
AW
403 if (uc_is_general_category_withtable (c,
404 SUBSEQUENT_IDENTIFIER_MASK
405 | UC_CATEGORY_MASK_Zs))
15671c6e 406 {
15671c6e
AW
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));
2e9fc9fc
AW
411 }
412 else
413 {
414 display_string ("\\x", 1, 2, port, iconveh_question_mark);
415 scm_intprint (c, 16, port);
416 display_character (';', port, iconveh_question_mark);
15671c6e
AW
417 }
418 }
419
f209aeee 420 scm_lfwrite_unlocked ("}#", 2, port);
15671c6e
AW
421}
422
423/* FIXME: allow R6RS hex escapes instead of #{...}#. */
424void
425scm_i_print_symbol_name (SCM sym, SCM port)
426{
427 if (symbol_has_extended_read_syntax (sym))
428 print_extended_symbol (sym, port);
429 else
430 print_normal_symbol (sym, port);
6662998f
MV
431}
432
e23106d5
MG
433void
434scm_print_symbol_name (const char *str, size_t len, SCM port)
435{
25d50a05 436 SCM symbol = scm_from_utf8_symboln (str, len);
ab33782c 437 scm_i_print_symbol_name (symbol, port);
e23106d5
MG
438}
439
c62fbfe1 440/* Print generally. Handles both write and display according to PSTATE.
0f2d19dd 441 */
8b840115
MD
442SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write);
443SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display);
1cc91f1b 444
d232520a
MV
445static void iprin1 (SCM exp, SCM port, scm_print_state *pstate);
446
dea901d6
MG
447
448/* Print a character as an octal or hex escape. */
449#define PRINT_CHAR_ESCAPE(i, port) \
450 do \
451 { \
452 if (!SCM_R6RS_ESCAPES_P) \
453 scm_intprint (i, 8, port); \
454 else \
455 { \
0607ebbf 456 scm_puts_unlocked ("x", port); \
dea901d6
MG
457 scm_intprint (i, 16, port); \
458 } \
459 } \
460 while (0)
461
462
0f2d19dd 463void
1bbd0b84 464scm_iprin1 (SCM exp, SCM port, scm_print_state *pstate)
d232520a
MV
465{
466 if (pstate->fancyp
467 && scm_is_true (scm_memq (exp, pstate->highlight_objects)))
468 {
81ae25da 469 scm_display (SCM_PRINT_HIGHLIGHT_PREFIX, port);
d232520a 470 iprin1 (exp, port, pstate);
81ae25da 471 scm_display (SCM_PRINT_HIGHLIGHT_SUFFIX, port);
d232520a
MV
472 }
473 else
474 iprin1 (exp, port, pstate);
475}
476
477static void
478iprin1 (SCM exp, SCM port, scm_print_state *pstate)
0f2d19dd 479{
54778cd3 480 switch (SCM_ITAG3 (exp))
0f2d19dd 481 {
e34f941a
DH
482 case scm_tc3_tc7_1:
483 case scm_tc3_tc7_2:
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.
487 */
488 scm_ipruk ("immediate", exp, port);
489 break;
490 case scm_tc3_int_1:
491 case scm_tc3_int_2:
e11e83f3 492 scm_intprint (SCM_I_INUM (exp), 10, port);
0f2d19dd 493 break;
e34f941a 494 case scm_tc3_imm24:
7866a09b 495 if (SCM_CHARP (exp))
0f2d19dd 496 {
b7f3516f 497 if (SCM_WRITINGP (pstate))
07f49ac7
LC
498 write_character (SCM_CHAR (exp), port, 0);
499 else
b7f3516f 500 {
07f49ac7
LC
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",
6851d3be 505 port, exp);
b7f3516f 506 }
0f2d19dd 507 }
a51ea417 508 else if (SCM_IFLAGP (exp)
e17d318f
DH
509 && ((size_t) SCM_IFLAGNUM (exp) < (sizeof iflagnames / sizeof (char *))))
510 {
0607ebbf 511 scm_puts_unlocked (iflagnames [SCM_IFLAGNUM (exp)], port);
e17d318f 512 }
0f2d19dd 513 else
e34f941a
DH
514 {
515 /* unknown immediate value */
516 scm_ipruk ("immediate", exp, port);
517 }
0f2d19dd 518 break;
e34f941a 519 case scm_tc3_cons:
0f2d19dd
JB
520 switch (SCM_TYP7 (exp))
521 {
904a077d
MV
522 case scm_tcs_struct:
523 {
524 ENTER_NESTED_DATA (pstate, exp, circref);
525 if (SCM_OBJ_CLASS_FLAGS (exp) & SCM_CLASSF_GOOPS)
526 {
527 SCM pwps, print = pstate->writingp ? g_write : g_display;
b2b33168 528 if (SCM_UNPACK (print) == 0)
904a077d 529 goto print_struct;
dfd03fb9 530 pwps = scm_i_port_with_print_state (port, pstate->handle);
7663c008 531 pstate->revealed = 1;
fa075d40 532 scm_call_2 (print, exp, pwps);
904a077d
MV
533 }
534 else
535 {
536 print_struct:
537 scm_print_struct (exp, port, pstate);
538 }
539 EXIT_NESTED_DATA (pstate);
540 }
541 break;
0f2d19dd
JB
542 case scm_tcs_cons_imcar:
543 case scm_tcs_cons_nimcar:
c62fbfe1
MD
544 ENTER_NESTED_DATA (pstate, exp, circref);
545 scm_iprlist ("(", exp, ')', port, pstate);
546 EXIT_NESTED_DATA (pstate);
a51ea417
MD
547 break;
548 circref:
c62fbfe1 549 print_circref (port, pstate, exp);
0f2d19dd 550 break;
534c55a9
DH
551 case scm_tc7_number:
552 switch SCM_TYP16 (exp) {
553 case scm_tc16_big:
554 scm_bigprint (exp, port, pstate);
555 break;
556 case scm_tc16_real:
557 scm_print_real (exp, port, pstate);
558 break;
559 case scm_tc16_complex:
560 scm_print_complex (exp, port, pstate);
561 break;
f92e85f7
MV
562 case scm_tc16_fraction:
563 scm_i_print_fraction (exp, port, pstate);
564 break;
534c55a9
DH
565 }
566 break;
9c44cd45
MG
567 case scm_tc7_string:
568 if (SCM_WRITINGP (pstate))
569 {
07f49ac7 570 size_t len, i;
9c44cd45 571
f4bc4e59 572 display_character ('"', port, iconveh_question_mark);
9c44cd45
MG
573 len = scm_i_string_length (exp);
574 for (i = 0; i < len; ++i)
07f49ac7
LC
575 write_character (scm_i_string_ref (exp, i), port, 1);
576
f4bc4e59 577 display_character ('"', port, iconveh_question_mark);
9c44cd45
MG
578 scm_remember_upto_here_1 (exp);
579 }
580 else
f4bc4e59
LC
581 {
582 size_t len, printed;
583
584 len = scm_i_string_length (exp);
585 printed = display_string (scm_i_string_data (exp),
586 scm_i_is_narrow_string (exp),
587 len, port,
588 scm_i_get_conversion_strategy (port));
589 if (SCM_UNLIKELY (printed < len))
f4bc4e59
LC
590 scm_encoding_error (__func__, errno,
591 "cannot convert to output locale",
6851d3be 592 port, scm_c_string_ref (exp, printed));
f4bc4e59
LC
593 }
594
9c44cd45
MG
595 scm_remember_upto_here_1 (exp);
596 break;
28b06554 597 case scm_tc7_symbol:
cc95e00a 598 if (scm_i_symbol_is_interned (exp))
9ff28a13 599 {
e23106d5 600 scm_i_print_symbol_name (exp, port);
9ff28a13
MV
601 scm_remember_upto_here_1 (exp);
602 }
603 else
604 {
0607ebbf 605 scm_puts_unlocked ("#<uninterned-symbol ", port);
e23106d5 606 scm_i_print_symbol_name (exp, port);
0607ebbf 607 scm_putc_unlocked (' ', port);
0345e278 608 scm_uintprint (SCM_UNPACK (exp), 16, port);
0607ebbf 609 scm_putc_unlocked ('>', port);
9ff28a13 610 }
6662998f 611 break;
e5aca4b5
MV
612 case scm_tc7_variable:
613 scm_i_variable_print (exp, port, pstate);
614 break;
2fb924f6
AW
615 case scm_tc7_program:
616 scm_i_program_print (exp, port, pstate);
617 break;
5b46a8c2
LC
618 case scm_tc7_pointer:
619 scm_i_pointer_print (exp, port, pstate);
e2c2a699 620 break;
c99de5aa
AW
621 case scm_tc7_hashtable:
622 scm_i_hashtable_print (exp, port, pstate);
623 break;
26b26354
AW
624 case scm_tc7_weak_set:
625 scm_i_weak_set_print (exp, port, pstate);
626 break;
7005c60f
AW
627 case scm_tc7_weak_table:
628 scm_i_weak_table_print (exp, port, pstate);
629 break;
9ea31741
AW
630 case scm_tc7_fluid:
631 scm_i_fluid_print (exp, port, pstate);
632 break;
45cf2428
AW
633 case scm_tc7_dynamic_state:
634 scm_i_dynamic_state_print (exp, port, pstate);
635 break;
6f3b0cc2
AW
636 case scm_tc7_frame:
637 scm_i_frame_print (exp, port, pstate);
638 break;
639 case scm_tc7_objcode:
640 scm_i_objcode_print (exp, port, pstate);
641 break;
642 case scm_tc7_vm:
643 scm_i_vm_print (exp, port, pstate);
644 break;
645 case scm_tc7_vm_cont:
646 scm_i_vm_cont_print (exp, port, pstate);
647 break;
bbb2ecd1
AW
648 case scm_tc7_prompt:
649 scm_i_prompt_print (exp, port, pstate);
650 break;
651 case scm_tc7_with_fluids:
652 scm_i_with_fluids_print (exp, port, pstate);
653 break;
b2637c98 654 case scm_tc7_array:
c62fbfe1 655 ENTER_NESTED_DATA (pstate, exp, circref);
b2637c98 656 scm_i_print_array (exp, port, pstate);
88c0a1d5 657 EXIT_NESTED_DATA (pstate);
b2637c98 658 break;
807e5a66
LC
659 case scm_tc7_bytevector:
660 scm_i_print_bytevector (exp, port, pstate);
661 break;
ff1feca9
AW
662 case scm_tc7_bitvector:
663 scm_i_print_bitvector (exp, port, pstate);
664 break;
0f2d19dd 665 case scm_tc7_wvect:
c62fbfe1 666 ENTER_NESTED_DATA (pstate, exp, circref);
91ee7515 667 scm_puts_unlocked ("#w(", port);
0f2d19dd 668 goto common_vector_printer;
0f2d19dd 669 case scm_tc7_vector:
c62fbfe1 670 ENTER_NESTED_DATA (pstate, exp, circref);
0607ebbf 671 scm_puts_unlocked ("#(", port);
0f2d19dd 672 common_vector_printer:
9fbaf27c 673 {
c014a02e 674 register long i;
4057a3e0 675 long last = SCM_SIMPLE_VECTOR_LENGTH (exp) - 1;
9fbaf27c 676 int cutp = 0;
4057a3e0
MV
677 if (pstate->fancyp
678 && SCM_SIMPLE_VECTOR_LENGTH (exp) > pstate->length)
9fbaf27c
MD
679 {
680 last = pstate->length - 1;
681 cutp = 1;
682 }
a141db86
AW
683 for (i = 0; i < last; ++i)
684 {
685 scm_iprin1 (scm_c_vector_ref (exp, i), port, pstate);
0607ebbf 686 scm_putc_unlocked (' ', port);
a141db86 687 }
9fbaf27c
MD
688 if (i == last)
689 {
690 /* CHECK_INTS; */
c367c4b4 691 scm_iprin1 (scm_c_vector_ref (exp, i), port, pstate);
9fbaf27c
MD
692 }
693 if (cutp)
0607ebbf
AW
694 scm_puts_unlocked (" ...", port);
695 scm_putc_unlocked (')', port);
9fbaf27c 696 }
c62fbfe1 697 EXIT_NESTED_DATA (pstate);
0f2d19dd 698 break;
0f2d19dd 699 case scm_tc7_port:
5ca6dc39 700 {
62bd5d66
AW
701 scm_t_ptob_descriptor *ptob = SCM_PORT_DESCRIPTOR (exp);
702 if (ptob->print && ptob->print (exp, port, pstate))
a51ea417 703 break;
5ca6dc39
JB
704 goto punk;
705 }
706 case scm_tc7_smob:
7a7f7c53
DH
707 ENTER_NESTED_DATA (pstate, exp, circref);
708 SCM_SMOB_DESCRIPTOR (exp).print (exp, port, pstate);
709 EXIT_NESTED_DATA (pstate);
710 break;
0f2d19dd 711 default:
314b8716 712 /* case scm_tcs_closures: */
a51ea417
MD
713 punk:
714 scm_ipruk ("type", exp, port);
0f2d19dd
JB
715 }
716 }
717}
718
c62fbfe1
MD
719/* Print states are necessary for circular reference safe printing.
720 * They are also expensive to allocate. Therefore print states are
721 * kept in a pool so that they can be reused.
722 */
1cc91f1b 723
bb35f315
MV
724/* The PORT argument can also be a print-state/port pair, which will
725 * then be used instead of allocating a new print state. This is
726 * useful for continuing a chain of print calls from Scheme. */
727
a51ea417 728void
1bbd0b84 729scm_prin1 (SCM exp, SCM port, int writingp)
a51ea417 730{
c4f37e80
MV
731 SCM handle = SCM_BOOL_F; /* Will GC protect the handle whilst unlinked */
732 SCM pstate_scm;
c62fbfe1 733 scm_print_state *pstate;
15635be5 734 int old_writingp;
c62fbfe1 735
bb35f315
MV
736 /* If PORT is a print-state/port pair, use that. Else create a new
737 print-state. */
c4f37e80 738
0c95b57d 739 if (SCM_PORT_WITH_PS_P (port))
bb35f315 740 {
52235e71
MD
741 pstate_scm = SCM_PORT_WITH_PS_PS (port);
742 port = SCM_PORT_WITH_PS_PORT (port);
bb35f315
MV
743 }
744 else
c62fbfe1 745 {
c4f37e80 746 /* First try to allocate a print state from the pool */
9de87eea 747 scm_i_pthread_mutex_lock (&print_state_mutex);
d2e53ed6 748 if (!scm_is_null (print_state_pool))
c4f37e80 749 {
d5cf5324
DH
750 handle = print_state_pool;
751 print_state_pool = SCM_CDR (print_state_pool);
c4f37e80 752 }
9de87eea 753 scm_i_pthread_mutex_unlock (&print_state_mutex);
7888309b 754 if (scm_is_false (handle))
d5cf5324 755 handle = scm_list_1 (make_print_state ());
c4f37e80 756 pstate_scm = SCM_CAR (handle);
c62fbfe1 757 }
c62fbfe1 758
c4f37e80 759 pstate = SCM_PRINT_STATE (pstate_scm);
15635be5 760 old_writingp = pstate->writingp;
c62fbfe1
MD
761 pstate->writingp = writingp;
762 scm_iprin1 (exp, port, pstate);
15635be5 763 pstate->writingp = old_writingp;
c62fbfe1 764
bb35f315
MV
765 /* Return print state to pool if it has been created above and
766 hasn't escaped to Scheme. */
767
7888309b 768 if (scm_is_true (handle) && !pstate->revealed)
c4f37e80 769 {
9de87eea 770 scm_i_pthread_mutex_lock (&print_state_mutex);
d5cf5324
DH
771 SCM_SETCDR (handle, print_state_pool);
772 print_state_pool = handle;
9de87eea 773 scm_i_pthread_mutex_unlock (&print_state_mutex);
c4f37e80 774 }
a51ea417
MD
775}
776
f4bc4e59
LC
777/* Convert codepoint CH to UTF-8 and store the result in UTF8. Return
778 the number of bytes of the UTF-8-encoded string. */
779static size_t
780codepoint_to_utf8 (scm_t_wchar ch, scm_t_uint8 utf8[4])
9c44cd45 781{
f4bc4e59
LC
782 size_t len;
783 scm_t_uint32 codepoint;
784
785 codepoint = (scm_t_uint32) ch;
07f49ac7 786
f4bc4e59 787 if (codepoint <= 0x7f)
07f49ac7 788 {
f4bc4e59
LC
789 len = 1;
790 utf8[0] = (scm_t_uint8) codepoint;
791 }
792 else if (codepoint <= 0x7ffUL)
793 {
794 len = 2;
795 utf8[0] = 0xc0 | (codepoint >> 6);
796 utf8[1] = 0x80 | (codepoint & 0x3f);
797 }
798 else if (codepoint <= 0xffffUL)
799 {
800 len = 3;
801 utf8[0] = 0xe0 | (codepoint >> 12);
802 utf8[1] = 0x80 | ((codepoint >> 6) & 0x3f);
803 utf8[2] = 0x80 | (codepoint & 0x3f);
07f49ac7
LC
804 }
805 else
806 {
f4bc4e59
LC
807 len = 4;
808 utf8[0] = 0xf0 | (codepoint >> 18);
809 utf8[1] = 0x80 | ((codepoint >> 12) & 0x3f);
810 utf8[2] = 0x80 | ((codepoint >> 6) & 0x3f);
811 utf8[3] = 0x80 | (codepoint & 0x3f);
812 }
813
814 return len;
815}
816
f4bc4e59
LC
817#define STR_REF(s, x) \
818 (narrow_p \
819 ? (scm_t_wchar) ((unsigned char *) (s))[x] \
820 : ((scm_t_wchar *) (s))[x])
821
7b292a9d
LC
822/* Write STR to PORT as UTF-8. STR is a LEN-codepoint string; it is
823 narrow if NARROW_P is true, wide otherwise. Return LEN. */
824static size_t
825display_string_as_utf8 (const void *str, int narrow_p, size_t len,
826 SCM port)
827{
828 size_t printed = 0;
829
830 while (len > printed)
831 {
832 size_t utf8_len, i;
833 char *input, utf8_buf[256];
834
835 /* Convert STR to UTF-8. */
836 for (i = printed, utf8_len = 0, input = utf8_buf;
837 i < len && utf8_len + 4 < sizeof (utf8_buf);
838 i++)
839 {
840 utf8_len += codepoint_to_utf8 (STR_REF (str, i),
841 (scm_t_uint8 *) input);
842 input = utf8_buf + utf8_len;
843 }
844
845 /* INPUT was successfully converted, entirely; print the
846 result. */
f209aeee 847 scm_lfwrite_unlocked (utf8_buf, utf8_len, port);
7b292a9d
LC
848 printed += i - printed;
849 }
850
851 assert (printed == len);
852
853 return len;
854}
855
79eb47ea
AW
856/* Write STR to PORT as ISO-8859-1. STR is a LEN-codepoint string; it
857 is narrow if NARROW_P is true, wide otherwise. Return LEN. */
858static size_t
859display_string_as_latin1 (const void *str, int narrow_p, size_t len,
860 SCM port,
861 scm_t_string_failed_conversion_handler strategy)
862{
863 size_t printed = 0;
864
865 if (narrow_p)
866 {
867 scm_lfwrite_unlocked (str, len, port);
868 return len;
869 }
870
871 while (printed < len)
872 {
873 char buf[256];
874 size_t i;
875
876 for (i = 0; i < sizeof(buf) && printed < len; i++, printed++)
877 {
878 scm_t_wchar c = STR_REF (str, printed);
879
880 if (c < 256)
881 buf[i] = c;
882 else
883 break;
884 }
885
886 scm_lfwrite_unlocked (buf, i, port);
887
888 if (i < sizeof(buf) && printed < len)
889 {
890 if (strategy == SCM_FAILED_CONVERSION_ERROR)
891 break;
892 else if (strategy == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
893 write_character_escaped (STR_REF (str, printed), 1, port);
894 else
895 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
896 display_string ("?", 1, 1, port, strategy);
897 printed++;
898 }
899 }
900
901 return printed;
902}
903
7b292a9d
LC
904/* Convert STR through PORT's output conversion descriptor and write the
905 output to PORT. Return the number of codepoints written. */
906static size_t
907display_string_using_iconv (const void *str, int narrow_p, size_t len,
908 SCM port,
909 scm_t_string_failed_conversion_handler strategy)
910{
f4bc4e59 911 size_t printed;
6c98257f 912 scm_t_iconv_descriptors *id;
f4bc4e59 913
6c98257f 914 id = scm_i_port_iconv_descriptors (port);
f4bc4e59 915
f4bc4e59
LC
916 printed = 0;
917
918 while (len > printed)
919 {
920 size_t done, utf8_len, input_left, output_left, i;
921 size_t codepoints_read, output_len;
922 char *input, *output;
923 char utf8_buf[256], encoded_output[256];
924 size_t offsets[256];
925
926 /* Convert STR to UTF-8. */
927 for (i = printed, utf8_len = 0, input = utf8_buf;
928 i < len && utf8_len + 4 < sizeof (utf8_buf);
929 i++)
07f49ac7 930 {
f4bc4e59
LC
931 offsets[utf8_len] = i;
932 utf8_len += codepoint_to_utf8 (STR_REF (str, i),
933 (scm_t_uint8 *) input);
934 input = utf8_buf + utf8_len;
935 }
936
937 input = utf8_buf;
938 input_left = utf8_len;
939
940 output = encoded_output;
941 output_left = sizeof (encoded_output);
942
6c98257f 943 done = iconv (id->output_cd, &input, &input_left,
f4bc4e59 944 &output, &output_left);
07f49ac7 945
f4bc4e59
LC
946 output_len = sizeof (encoded_output) - output_left;
947
948 if (SCM_UNLIKELY (done == (size_t) -1))
949 {
b2548e23
AW
950 int errno_save = errno;
951
f4bc4e59 952 /* Reset the `iconv' state. */
6c98257f 953 iconv (id->output_cd, NULL, NULL, NULL, NULL);
f4bc4e59 954
7174bc08 955 /* Print the OUTPUT_LEN bytes successfully converted. */
f209aeee 956 scm_lfwrite_unlocked (encoded_output, output_len, port);
7174bc08
LC
957
958 /* See how many input codepoints these OUTPUT_LEN bytes
959 corresponds to. */
960 codepoints_read = offsets[input - utf8_buf] - printed;
961 printed += codepoints_read;
962
b2548e23 963 if (errno_save == EILSEQ &&
f4bc4e59 964 strategy != SCM_FAILED_CONVERSION_ERROR)
07f49ac7 965 {
f4bc4e59
LC
966 /* Conversion failed somewhere in INPUT and we want to
967 escape or substitute the offending input character. */
968
f4bc4e59 969 if (strategy == SCM_FAILED_CONVERSION_ESCAPE_SEQUENCE)
f1ee6d54 970 {
f4bc4e59
LC
971 scm_t_wchar ch;
972
973 /* Find CH, the offending codepoint, and escape it. */
974 ch = STR_REF (str, offsets[input - utf8_buf]);
975 write_character_escaped (ch, 1, port);
f1ee6d54 976 }
07f49ac7 977 else
f4bc4e59
LC
978 /* STRATEGY is `SCM_FAILED_CONVERSION_QUESTION_MARK'. */
979 display_string ("?", 1, 1, port, strategy);
9c44cd45 980
f4bc4e59
LC
981 printed++;
982 }
983 else
984 /* Something bad happened that we can't handle: bail out. */
985 break;
07f49ac7
LC
986 }
987 else
f4bc4e59
LC
988 {
989 /* INPUT was successfully converted, entirely; print the
990 result. */
f209aeee 991 scm_lfwrite_unlocked (encoded_output, output_len, port);
f4bc4e59
LC
992 codepoints_read = i - printed;
993 printed += codepoints_read;
994 }
07f49ac7
LC
995 }
996
997 return printed;
7b292a9d
LC
998}
999
f4bc4e59 1000#undef STR_REF
7b292a9d
LC
1001
1002/* Display the LEN codepoints in STR to PORT according to STRATEGY;
1003 return the number of codepoints successfully displayed. If NARROW_P,
1004 then STR is interpreted as a sequence of `char', denoting a Latin-1
1005 string; otherwise it's interpreted as a sequence of
1006 `scm_t_wchar'. */
1007static size_t
1008display_string (const void *str, int narrow_p,
1009 size_t len, SCM port,
1010 scm_t_string_failed_conversion_handler strategy)
1011
1012{
1013 scm_t_port *pt;
1014
1015 pt = SCM_PTAB_ENTRY (port);
1016
6c98257f 1017 if (pt->encoding_mode == SCM_PORT_ENCODING_MODE_UTF8)
7b292a9d 1018 return display_string_as_utf8 (str, narrow_p, len, port);
79eb47ea
AW
1019 else if (pt->encoding_mode == SCM_PORT_ENCODING_MODE_LATIN1)
1020 return display_string_as_latin1 (str, narrow_p, len, port, strategy);
7b292a9d 1021 else
79eb47ea 1022 return display_string_using_iconv (str, narrow_p, len, port, strategy);
f4bc4e59
LC
1023}
1024
1025/* Attempt to display CH to PORT according to STRATEGY. Return non-zero
1026 if CH was successfully displayed, zero otherwise (e.g., if it was not
1027 representable in PORT's encoding.) */
1028static int
1029display_character (scm_t_wchar ch, SCM port,
1030 scm_t_string_failed_conversion_handler strategy)
1031{
1032 return display_string (&ch, 0, 1, port, strategy) == 1;
07f49ac7
LC
1033}
1034
33d92fe6
LC
1035/* Attempt to pretty-print CH, a combining character, to PORT. Return
1036 zero upon failure, non-zero otherwise. The idea is to print CH above
1037 a dotted circle to make it more visible. */
1038static int
1039write_combining_character (scm_t_wchar ch, SCM port)
1040{
f4bc4e59
LC
1041 scm_t_wchar str[2];
1042
1043 str[0] = SCM_CODEPOINT_DOTTED_CIRCLE;
1044 str[1] = ch;
1045
1046 return display_string (str, 0, 2, port, iconveh_error) == 2;
1047}
33d92fe6 1048
f4bc4e59
LC
1049/* Write CH to PORT in its escaped form, using the string escape syntax
1050 if STRING_ESCAPES_P is non-zero. */
1051static void
1052write_character_escaped (scm_t_wchar ch, int string_escapes_p, SCM port)
1053{
1054 if (string_escapes_p)
33d92fe6 1055 {
f4bc4e59
LC
1056 /* Represent CH using the in-string escape syntax. */
1057
1058 static const char hex[] = "0123456789abcdef";
1059 static const char escapes[7] = "abtnvfr";
1060 char buf[9];
1061
1062 if (ch >= 0x07 && ch <= 0x0D && ch != 0x0A)
33d92fe6 1063 {
f4bc4e59
LC
1064 /* Use special escapes for some C0 controls. */
1065 buf[0] = '\\';
1066 buf[1] = escapes[ch - 0x07];
f209aeee 1067 scm_lfwrite_unlocked (buf, 2, port);
f4bc4e59
LC
1068 }
1069 else if (!SCM_R6RS_ESCAPES_P)
1070 {
1071 if (ch <= 0xFF)
1072 {
1073 buf[0] = '\\';
1074 buf[1] = 'x';
1075 buf[2] = hex[ch / 16];
1076 buf[3] = hex[ch % 16];
f209aeee 1077 scm_lfwrite_unlocked (buf, 4, port);
f4bc4e59
LC
1078 }
1079 else if (ch <= 0xFFFF)
1080 {
1081 buf[0] = '\\';
1082 buf[1] = 'u';
1083 buf[2] = hex[(ch & 0xF000) >> 12];
1084 buf[3] = hex[(ch & 0xF00) >> 8];
1085 buf[4] = hex[(ch & 0xF0) >> 4];
1086 buf[5] = hex[(ch & 0xF)];
f209aeee 1087 scm_lfwrite_unlocked (buf, 6, port);
f4bc4e59
LC
1088 }
1089 else if (ch > 0xFFFF)
1090 {
1091 buf[0] = '\\';
1092 buf[1] = 'U';
1093 buf[2] = hex[(ch & 0xF00000) >> 20];
1094 buf[3] = hex[(ch & 0xF0000) >> 16];
1095 buf[4] = hex[(ch & 0xF000) >> 12];
1096 buf[5] = hex[(ch & 0xF00) >> 8];
1097 buf[6] = hex[(ch & 0xF0) >> 4];
1098 buf[7] = hex[(ch & 0xF)];
f209aeee 1099 scm_lfwrite_unlocked (buf, 8, port);
f4bc4e59 1100 }
33d92fe6
LC
1101 }
1102 else
f4bc4e59
LC
1103 {
1104 /* Print an R6RS variable-length hex escape: "\xNNNN;". */
1105 scm_t_wchar ch2 = ch;
1106
1107 int i = 8;
1108 buf[i] = ';';
1109 i --;
1110 if (ch == 0)
1111 buf[i--] = '0';
1112 else
1113 while (ch2 > 0)
1114 {
1115 buf[i] = hex[ch2 & 0xF];
1116 ch2 >>= 4;
1117 i --;
1118 }
1119 buf[i] = 'x';
1120 i --;
1121 buf[i] = '\\';
f209aeee 1122 scm_lfwrite_unlocked (buf + i, 9 - i, port);
f4bc4e59 1123 }
33d92fe6
LC
1124 }
1125 else
f4bc4e59
LC
1126 {
1127 /* Represent CH using the character escape syntax. */
1128 const char *name;
33d92fe6 1129
f4bc4e59
LC
1130 name = scm_i_charname (SCM_MAKE_CHAR (ch));
1131 if (name != NULL)
0607ebbf 1132 scm_puts_unlocked (name, port);
f4bc4e59
LC
1133 else
1134 PRINT_CHAR_ESCAPE (ch, port);
1135 }
33d92fe6
LC
1136}
1137
07f49ac7
LC
1138/* Write CH to PORT, escaping it if it's non-graphic or not
1139 representable in PORT's encoding. If STRING_ESCAPES_P is true and CH
1140 needs to be escaped, it is escaped using the in-string escape syntax;
1141 otherwise the character escape syntax is used. */
1142static void
1143write_character (scm_t_wchar ch, SCM port, int string_escapes_p)
1144{
1145 int printed = 0;
f4bc4e59
LC
1146 scm_t_string_failed_conversion_handler strategy;
1147
1148 strategy = scm_i_get_conversion_strategy (port);
07f49ac7
LC
1149
1150 if (string_escapes_p)
1151 {
1152 /* Check if CH deserves special treatment. */
1153 if (ch == '"' || ch == '\\')
1154 {
f4bc4e59
LC
1155 display_character ('\\', port, iconveh_question_mark);
1156 display_character (ch, port, strategy);
07f49ac7
LC
1157 printed = 1;
1158 }
8500b186
AW
1159 else if (ch == '\n' && SCM_PRINT_ESCAPE_NEWLINES_P)
1160 {
1161 display_character ('\\', port, iconveh_question_mark);
1162 display_character ('n', port, strategy);
1163 printed = 1;
1164 }
07f49ac7
LC
1165 else if (ch == ' ' || ch == '\n')
1166 {
f4bc4e59 1167 display_character (ch, port, strategy);
07f49ac7
LC
1168 printed = 1;
1169 }
1170 }
1171 else
33d92fe6 1172 {
f4bc4e59 1173 display_string ("#\\", 1, 2, port, iconveh_question_mark);
33d92fe6
LC
1174
1175 if (uc_combining_class (ch) != UC_CCC_NR)
1176 /* Character is a combining character, so attempt to
1177 pretty-print it. */
1178 printed = write_combining_character (ch, port);
1179 }
07f49ac7
LC
1180
1181 if (!printed
1182 && uc_is_general_category_withtable (ch,
1183 UC_CATEGORY_MASK_L |
1184 UC_CATEGORY_MASK_M |
1185 UC_CATEGORY_MASK_N |
1186 UC_CATEGORY_MASK_P |
1187 UC_CATEGORY_MASK_S))
1188 /* CH is graphic; attempt to display it. */
1189 printed = display_character (ch, port, iconveh_error);
1190
1191 if (!printed)
f4bc4e59
LC
1192 /* CH isn't graphic or cannot be represented in PORT's encoding. */
1193 write_character_escaped (ch, string_escapes_p, port);
9c44cd45 1194}
0f2d19dd
JB
1195
1196/* Print an integer.
1197 */
1cc91f1b 1198
0f2d19dd 1199void
a406c9e9 1200scm_intprint (scm_t_intmax n, int radix, SCM port)
0f2d19dd
JB
1201{
1202 char num_buf[SCM_INTBUFLEN];
f209aeee 1203 scm_lfwrite_unlocked (num_buf, scm_iint2str (n, radix, num_buf), port);
0f2d19dd
JB
1204}
1205
a406c9e9
MV
1206void
1207scm_uintprint (scm_t_uintmax n, int radix, SCM port)
1208{
1209 char num_buf[SCM_INTBUFLEN];
f209aeee 1210 scm_lfwrite_unlocked (num_buf, scm_iuint2str (n, radix, num_buf), port);
a406c9e9
MV
1211}
1212
0f2d19dd
JB
1213/* Print an object of unrecognized type.
1214 */
1cc91f1b 1215
0f2d19dd 1216void
1bbd0b84 1217scm_ipruk (char *hdr, SCM ptr, SCM port)
0f2d19dd 1218{
0607ebbf
AW
1219 scm_puts_unlocked ("#<unknown-", port);
1220 scm_puts_unlocked (hdr, port);
26224b3f 1221 if (1) /* (scm_in_heap_p (ptr)) */ /* FIXME */
0f2d19dd 1222 {
0607ebbf 1223 scm_puts_unlocked (" (0x", port);
0345e278 1224 scm_uintprint (SCM_CELL_WORD_0 (ptr), 16, port);
0607ebbf 1225 scm_puts_unlocked (" . 0x", port);
0345e278 1226 scm_uintprint (SCM_CELL_WORD_1 (ptr), 16, port);
0607ebbf 1227 scm_puts_unlocked (") @", port);
0f2d19dd 1228 }
0607ebbf 1229 scm_puts_unlocked (" 0x", port);
0345e278 1230 scm_uintprint (SCM_UNPACK (ptr), 16, port);
0607ebbf 1231 scm_putc_unlocked ('>', port);
0f2d19dd
JB
1232}
1233
1cc91f1b 1234
904a077d 1235/* Print a list.
22a52da1 1236 */
0f2d19dd 1237void
34d19ef6 1238scm_iprlist (char *hdr, SCM exp, int tlr, SCM port, scm_print_state *pstate)
0f2d19dd 1239{
c62fbfe1 1240 register SCM hare, tortoise;
c014a02e 1241 long floor = pstate->top - 2;
0607ebbf 1242 scm_puts_unlocked (hdr, port);
0f2d19dd 1243 /* CHECK_INTS; */
c62fbfe1
MD
1244 if (pstate->fancyp)
1245 goto fancy_printing;
1246
1247 /* Run a hare and tortoise so that total time complexity will be
1248 O(depth * N) instead of O(N^2). */
1249 hare = SCM_CDR (exp);
1250 tortoise = exp;
d2e53ed6 1251 while (scm_is_pair (hare))
c62fbfe1 1252 {
bc36d050 1253 if (scm_is_eq (hare, tortoise))
c62fbfe1
MD
1254 goto fancy_printing;
1255 hare = SCM_CDR (hare);
d2e53ed6 1256 if (!scm_is_pair (hare))
c62fbfe1
MD
1257 break;
1258 hare = SCM_CDR (hare);
1259 tortoise = SCM_CDR (tortoise);
1260 }
1261
1262 /* No cdr cycles intrinsic to this list */
1263 scm_iprin1 (SCM_CAR (exp), port, pstate);
d2e53ed6 1264 for (exp = SCM_CDR (exp); scm_is_pair (exp); exp = SCM_CDR (exp))
0f2d19dd 1265 {
c014a02e 1266 register long i;
5ca6dc39 1267
c62fbfe1 1268 for (i = floor; i >= 0; --i)
509759dd 1269 if (scm_is_eq (PSTATE_STACK_REF(pstate, i), exp))
c62fbfe1
MD
1270 goto circref;
1271 PUSH_REF (pstate, exp);
0607ebbf 1272 scm_putc_unlocked (' ', port);
0f2d19dd 1273 /* CHECK_INTS; */
c62fbfe1 1274 scm_iprin1 (SCM_CAR (exp), port, pstate);
0f2d19dd 1275 }
c96d76b8 1276 if (!SCM_NULL_OR_NIL_P (exp))
0f2d19dd 1277 {
0607ebbf 1278 scm_puts_unlocked (" . ", port);
c62fbfe1 1279 scm_iprin1 (exp, port, pstate);
0f2d19dd 1280 }
c62fbfe1 1281
a51ea417 1282end:
0607ebbf 1283 scm_putc_unlocked (tlr, port);
c62fbfe1 1284 pstate->top = floor + 2;
a51ea417 1285 return;
c62fbfe1
MD
1286
1287fancy_printing:
1288 {
c014a02e 1289 long n = pstate->length;
c62fbfe1
MD
1290
1291 scm_iprin1 (SCM_CAR (exp), port, pstate);
1292 exp = SCM_CDR (exp); --n;
d2e53ed6 1293 for (; scm_is_pair (exp); exp = SCM_CDR (exp))
c62fbfe1 1294 {
c014a02e 1295 register unsigned long i;
5ca6dc39 1296
c62fbfe1 1297 for (i = 0; i < pstate->top; ++i)
509759dd 1298 if (scm_is_eq (PSTATE_STACK_REF(pstate, i), exp))
c62fbfe1
MD
1299 goto fancy_circref;
1300 if (pstate->fancyp)
1301 {
1302 if (n == 0)
1303 {
0607ebbf 1304 scm_puts_unlocked (" ...", port);
c62fbfe1
MD
1305 goto skip_tail;
1306 }
1307 else
1308 --n;
1309 }
1310 PUSH_REF(pstate, exp);
1311 ++pstate->list_offset;
0607ebbf 1312 scm_putc_unlocked (' ', port);
c62fbfe1
MD
1313 /* CHECK_INTS; */
1314 scm_iprin1 (SCM_CAR (exp), port, pstate);
1315 }
1316 }
c96d76b8 1317 if (!SCM_NULL_OR_NIL_P (exp))
c62fbfe1 1318 {
0607ebbf 1319 scm_puts_unlocked (" . ", port);
c62fbfe1
MD
1320 scm_iprin1 (exp, port, pstate);
1321 }
1322skip_tail:
1323 pstate->list_offset -= pstate->top - floor - 2;
a51ea417 1324 goto end;
a51ea417 1325
c62fbfe1
MD
1326fancy_circref:
1327 pstate->list_offset -= pstate->top - floor - 2;
1328
1329circref:
0607ebbf 1330 scm_puts_unlocked (" . ", port);
c62fbfe1
MD
1331 print_circref (port, pstate, exp);
1332 goto end;
0f2d19dd
JB
1333}
1334
1335\f
1336
bb35f315
MV
1337int
1338scm_valid_oport_value_p (SCM val)
1339{
368cf54d
GB
1340 return (SCM_OPOUTPORTP (val)
1341 || (SCM_PORT_WITH_PS_P (val)
1342 && SCM_OPOUTPORTP (SCM_PORT_WITH_PS_PORT (val))));
bb35f315
MV
1343}
1344
8b840115 1345/* SCM_GPROC(s_write, "write", 1, 1, 0, scm_write, g_write); */
1cc91f1b 1346
0f2d19dd 1347SCM
1bbd0b84 1348scm_write (SCM obj, SCM port)
0f2d19dd
JB
1349{
1350 if (SCM_UNBNDP (port))
9de87eea 1351 port = scm_current_output_port ();
3eb7e6ee
JB
1352
1353 SCM_ASSERT (scm_valid_oport_value_p (port), port, SCM_ARG2, s_write);
bb35f315 1354
215fe3a8 1355 scm_dynwind_begin (0);
92c0ebac 1356 scm_dynwind_lock_port (SCM_COERCE_OUTPORT (port));
a51ea417 1357 scm_prin1 (obj, port, 1);
215fe3a8
AW
1358 scm_dynwind_end ();
1359
0f2d19dd
JB
1360 return SCM_UNSPECIFIED;
1361}
1362
1363
8b840115 1364/* SCM_GPROC(s_display, "display", 1, 1, 0, scm_display, g_display); */
1cc91f1b 1365
0f2d19dd 1366SCM
1bbd0b84 1367scm_display (SCM obj, SCM port)
0f2d19dd
JB
1368{
1369 if (SCM_UNBNDP (port))
9de87eea 1370 port = scm_current_output_port ();
3eb7e6ee
JB
1371
1372 SCM_ASSERT (scm_valid_oport_value_p (port), port, SCM_ARG2, s_display);
bb35f315 1373
215fe3a8 1374 scm_dynwind_begin (0);
92c0ebac 1375 scm_dynwind_lock_port (SCM_COERCE_OUTPORT (port));
a51ea417 1376 scm_prin1 (obj, port, 0);
215fe3a8
AW
1377 scm_dynwind_end ();
1378
0f2d19dd
JB
1379 return SCM_UNSPECIFIED;
1380}
1381
70d63753
GB
1382
1383SCM_DEFINE (scm_simple_format, "simple-format", 2, 0, 1,
1384 (SCM destination, SCM message, SCM args),
eca65e90
MG
1385 "Write @var{message} to @var{destination}, defaulting to\n"
1386 "the current output port.\n"
1387 "@var{message} can contain @code{~A} (was @code{%s}) and\n"
1388 "@code{~S} (was @code{%S}) escapes. When printed,\n"
1389 "the escapes are replaced with corresponding members of\n"
b7e64f8b 1390 "@var{args}:\n"
eca65e90
MG
1391 "@code{~A} formats using @code{display} and @code{~S} formats\n"
1392 "using @code{write}.\n"
1393 "If @var{destination} is @code{#t}, then use the current output\n"
1394 "port, if @var{destination} is @code{#f}, then return a string\n"
1395 "containing the formatted text. Does not add a trailing newline.")
70d63753
GB
1396#define FUNC_NAME s_scm_simple_format
1397{
dfd03fb9 1398 SCM port, answer = SCM_UNSPECIFIED;
70d63753
GB
1399 int fReturnString = 0;
1400 int writingp;
889975e5 1401 size_t start, p, end;
70d63753 1402
bc36d050 1403 if (scm_is_eq (destination, SCM_BOOL_T))
daba1a71 1404 {
9de87eea 1405 destination = port = scm_current_output_port ();
daba1a71 1406 }
7888309b 1407 else if (scm_is_false (destination))
daba1a71
MD
1408 {
1409 fReturnString = 1;
0b2c2ba3 1410 port = scm_mkstrport (SCM_INUM0, SCM_BOOL_F,
dfd03fb9
MD
1411 SCM_OPN | SCM_WRTNG,
1412 FUNC_NAME);
1413 destination = port;
daba1a71
MD
1414 }
1415 else
1416 {
1417 SCM_VALIDATE_OPORT_VALUE (1, destination);
dfd03fb9 1418 port = SCM_COERCE_OUTPORT (destination);
daba1a71
MD
1419 }
1420 SCM_VALIDATE_STRING (2, message);
af45e3b0 1421 SCM_VALIDATE_REST_ARGUMENT (args);
70d63753 1422
889975e5
MG
1423 p = 0;
1424 start = 0;
1425 end = scm_i_string_length (message);
b24b5e13 1426 for (p = start; p != end; ++p)
889975e5 1427 if (scm_i_string_ref (message, p) == '~')
70d63753 1428 {
b24b5e13 1429 if (++p == end)
6662998f
MV
1430 break;
1431
889975e5 1432 switch (scm_i_string_ref (message, p))
6662998f
MV
1433 {
1434 case 'A': case 'a':
1435 writingp = 0;
1436 break;
1437 case 'S': case 's':
1438 writingp = 1;
1439 break;
1440 case '~':
889975e5 1441 scm_lfwrite_substr (message, start, p, port);
6662998f
MV
1442 start = p + 1;
1443 continue;
1444 case '%':
889975e5 1445 scm_lfwrite_substr (message, start, p - 1, port);
dfd03fb9 1446 scm_newline (port);
6662998f
MV
1447 start = p + 1;
1448 continue;
1449 default:
1afff620 1450 SCM_MISC_ERROR ("FORMAT: Unsupported format option ~~~A - use (ice-9 format) instead",
889975e5 1451 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message, p))));
6662998f
MV
1452
1453 }
70d63753 1454
6662998f 1455
d2e53ed6 1456 if (!scm_is_pair (args))
1afff620 1457 SCM_MISC_ERROR ("FORMAT: Missing argument for ~~~A",
889975e5 1458 scm_list_1 (SCM_MAKE_CHAR (scm_i_string_ref (message, p))));
6662998f 1459
889975e5 1460 scm_lfwrite_substr (message, start, p - 1, port);
dfd03fb9 1461 /* we pass destination here */
70d63753
GB
1462 scm_prin1 (SCM_CAR (args), destination, writingp);
1463 args = SCM_CDR (args);
1464 start = p + 1;
1465 }
6662998f 1466
889975e5 1467 scm_lfwrite_substr (message, start, p, port);
bc36d050 1468 if (!scm_is_eq (args, SCM_EOL))
1afff620
KN
1469 SCM_MISC_ERROR ("FORMAT: ~A superfluous arguments",
1470 scm_list_1 (scm_length (args)));
70d63753
GB
1471
1472 if (fReturnString)
1473 answer = scm_strport_to_string (destination);
1474
daba1a71 1475 return scm_return_first (answer, message);
70d63753
GB
1476}
1477#undef FUNC_NAME
1478
1479
3b3b36dd 1480SCM_DEFINE (scm_newline, "newline", 0, 1, 0,
b450f070 1481 (SCM port),
8f85c0c6
NJ
1482 "Send a newline to @var{port}.\n"
1483 "If @var{port} is omitted, send to the current output port.")
1bbd0b84 1484#define FUNC_NAME s_scm_newline
0f2d19dd
JB
1485{
1486 if (SCM_UNBNDP (port))
9de87eea 1487 port = scm_current_output_port ();
3eb7e6ee 1488
34d19ef6 1489 SCM_VALIDATE_OPORT_VALUE (1, port);
bb35f315 1490
0607ebbf 1491 scm_putc_unlocked ('\n', SCM_COERCE_OUTPORT (port));
0f2d19dd
JB
1492 return SCM_UNSPECIFIED;
1493}
1bbd0b84 1494#undef FUNC_NAME
0f2d19dd 1495
3b3b36dd 1496SCM_DEFINE (scm_write_char, "write-char", 1, 1, 0,
b450f070 1497 (SCM chr, SCM port),
eca65e90 1498 "Send character @var{chr} to @var{port}.")
1bbd0b84 1499#define FUNC_NAME s_scm_write_char
0f2d19dd
JB
1500{
1501 if (SCM_UNBNDP (port))
9de87eea 1502 port = scm_current_output_port ();
3eb7e6ee 1503
34d19ef6
HWN
1504 SCM_VALIDATE_CHAR (1, chr);
1505 SCM_VALIDATE_OPORT_VALUE (2, port);
07f49ac7
LC
1506
1507 port = SCM_COERCE_OUTPORT (port);
1508 if (!display_character (SCM_CHAR (chr), port,
1509 scm_i_get_conversion_strategy (port)))
1510 scm_encoding_error (__func__, errno,
1511 "cannot convert to output locale",
6851d3be 1512 port, chr);
07f49ac7 1513
0f2d19dd
JB
1514 return SCM_UNSPECIFIED;
1515}
1bbd0b84 1516#undef FUNC_NAME
0f2d19dd 1517
0f2d19dd
JB
1518\f
1519
bb35f315 1520/* Call back to Scheme code to do the printing of special objects
c19bc088
MD
1521 * (like structs). SCM_PRINTER_APPLY applies PROC to EXP and a smob
1522 * containing PORT and PSTATE. This object can be used as the port for
1523 * display/write etc to continue the current print chain. The REVEALED
1524 * field of PSTATE is set to true to indicate that the print state has
1525 * escaped to Scheme and thus has to be freed by the GC.
1526 */
1527
92c2555f 1528scm_t_bits scm_tc16_port_with_ps;
c19bc088
MD
1529
1530/* Print exactly as the port itself would */
1531
1532static int
e841c3e0 1533port_with_ps_print (SCM obj, SCM port, scm_print_state *pstate)
c19bc088
MD
1534{
1535 obj = SCM_PORT_WITH_PS_PORT (obj);
62bd5d66 1536 return SCM_PORT_DESCRIPTOR (obj)->print (obj, port, pstate);
c19bc088 1537}
c4f37e80
MV
1538
1539SCM
1bbd0b84 1540scm_printer_apply (SCM proc, SCM exp, SCM port, scm_print_state *pstate)
c4f37e80 1541{
bb35f315 1542 pstate->revealed = 1;
dfd03fb9
MD
1543 return scm_call_2 (proc, exp,
1544 scm_i_port_with_print_state (port, pstate->handle));
c19bc088
MD
1545}
1546
dfd03fb9 1547SCM_DEFINE (scm_port_with_print_state, "port-with-print-state", 1, 1, 0,
1bbd0b84 1548 (SCM port, SCM pstate),
71331188 1549 "Create a new port which behaves like @var{port}, but with an\n"
dfd03fb9
MD
1550 "included print state @var{pstate}. @var{pstate} is optional.\n"
1551 "If @var{pstate} isn't supplied and @var{port} already has\n"
1552 "a print state, the old print state is reused.")
1bbd0b84 1553#define FUNC_NAME s_scm_port_with_print_state
c19bc088 1554{
34d19ef6 1555 SCM_VALIDATE_OPORT_VALUE (1, port);
dfd03fb9
MD
1556 if (!SCM_UNBNDP (pstate))
1557 SCM_VALIDATE_PRINTSTATE (2, pstate);
1558 return scm_i_port_with_print_state (port, pstate);
c19bc088 1559}
1bbd0b84 1560#undef FUNC_NAME
c19bc088 1561
a1ec6916 1562SCM_DEFINE (scm_get_print_state, "get-print-state", 1, 0, 0,
1bbd0b84 1563 (SCM port),
71331188
MG
1564 "Return the print state of the port @var{port}. If @var{port}\n"
1565 "has no associated print state, @code{#f} is returned.")
1bbd0b84 1566#define FUNC_NAME s_scm_get_print_state
c19bc088 1567{
368cf54d
GB
1568 if (SCM_PORT_WITH_PS_P (port))
1569 return SCM_PORT_WITH_PS_PS (port);
f5f2dcff 1570 if (SCM_OUTPUT_PORT_P (port))
368cf54d 1571 return SCM_BOOL_F;
276dd677 1572 SCM_WRONG_TYPE_ARG (1, port);
c4f37e80 1573}
1bbd0b84 1574#undef FUNC_NAME
bb35f315 1575
c4f37e80 1576\f
1cc91f1b 1577
0f2d19dd
JB
1578void
1579scm_init_print ()
0f2d19dd 1580{
231dd356 1581 SCM type;
d5cf5324 1582
231dd356
AW
1583 type = scm_make_vtable (scm_from_locale_string (SCM_PRINT_STATE_LAYOUT),
1584 SCM_BOOL_F);
4a655e50 1585 scm_set_struct_vtable_name_x (type, scm_from_latin1_symbol ("print-state"));
bb35f315 1586 scm_print_state_vtable = type;
c4f37e80 1587
c19bc088
MD
1588 /* Don't want to bind a wrapper class in GOOPS, so pass 0 as arg1. */
1589 scm_tc16_port_with_ps = scm_make_smob_type (0, 0);
e841c3e0 1590 scm_set_smob_print (scm_tc16_port_with_ps, port_with_ps_print);
81ae25da 1591
a0599745 1592#include "libguile/print.x"
475fa9a5 1593
8500b186
AW
1594 scm_init_opts (scm_print_options, scm_print_opts);
1595 scm_print_opts[SCM_PRINT_HIGHLIGHT_PREFIX_I].val =
1596 SCM_UNPACK (scm_from_locale_string ("{"));
1597 scm_print_opts[SCM_PRINT_HIGHLIGHT_SUFFIX_I].val =
1598 SCM_UNPACK (scm_from_locale_string ("}"));
475fa9a5 1599 scm_print_opts[SCM_PRINT_KEYWORD_STYLE_I].val = SCM_UNPACK (sym_reader);
0f2d19dd 1600}
89e00824
ML
1601
1602/*
1603 Local Variables:
1604 c-file-style: "gnu"
1605 End:
1606*/