add a test for IA64 Debian GNU/Linux Etch.
[ruby-svn.git] / sprintf.c
blob9253820f752894e61565b5628b7a2a0c83efd5cc
1 /**********************************************************************
3 sprintf.c -
5 $Author$
6 created at: Fri Oct 15 10:39:26 JST 1993
8 Copyright (C) 1993-2007 Yukihiro Matsumoto
9 Copyright (C) 2000 Network Applied Communication Laboratory, Inc.
10 Copyright (C) 2000 Information-technology Promotion Agency, Japan
12 **********************************************************************/
14 #include "ruby/ruby.h"
15 #include "ruby/re.h"
16 #include "ruby/encoding.h"
17 #include <math.h>
18 #include <stdarg.h>
20 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
21 #define BITSPERDIG (SIZEOF_BDIGITS*CHAR_BIT)
22 #define EXTENDSIGN(n, l) (((~0 << (n)) >> (((n)*(l)) % BITSPERDIG)) & ~(~0 << (n)))
24 static void fmt_setup(char*,int,int,int,int);
26 static char*
27 remove_sign_bits(char *str, int base)
29 char *s, *t;
31 s = t = str;
33 if (base == 16) {
34 while (*t == 'f') {
35 t++;
38 else if (base == 8) {
39 *t |= EXTENDSIGN(3, strlen(t));
40 while (*t == '7') {
41 t++;
44 else if (base == 2) {
45 while (*t == '1') {
46 t++;
50 return t;
53 static char
54 sign_bits(int base, const char *p)
56 char c = '.';
58 switch (base) {
59 case 16:
60 if (*p == 'X') c = 'F';
61 else c = 'f';
62 break;
63 case 8:
64 c = '7'; break;
65 case 2:
66 c = '1'; break;
68 return c;
71 #define FNONE 0
72 #define FSHARP 1
73 #define FMINUS 2
74 #define FPLUS 4
75 #define FZERO 8
76 #define FSPACE 16
77 #define FWIDTH 32
78 #define FPREC 64
79 #define FPREC0 128
81 #define CHECK(l) do {\
82 while (blen + (l) >= bsiz) {\
83 bsiz*=2;\
85 rb_str_resize(result, bsiz);\
86 buf = RSTRING_PTR(result);\
87 } while (0)
89 #define PUSH(s, l) do { \
90 CHECK(l);\
91 memcpy(&buf[blen], s, l);\
92 blen += (l);\
93 } while (0)
95 #define FILL(c, l) do { \
96 CHECK(l);\
97 memset(&buf[blen], c, l);\
98 blen += (l);\
99 } while (0)
101 #define GETARG() (nextvalue != Qundef ? nextvalue : \
102 posarg < 0 ? \
103 (rb_raise(rb_eArgError, "unnumbered(%d) mixed with numbered", nextarg), 0) : \
104 (posarg = nextarg++, GETNTHARG(posarg)))
106 #define GETPOSARG(n) (posarg > 0 ? \
107 (rb_raise(rb_eArgError, "numbered(%d) after unnumbered(%d)", n, posarg), 0) : \
108 ((n < 1) ? (rb_raise(rb_eArgError, "invalid index - %d$", n), 0) : \
109 (posarg = -1, GETNTHARG(n))))
111 #define GETNTHARG(nth) \
112 ((nth >= argc) ? (rb_raise(rb_eArgError, "too few arguments"), 0) : argv[nth])
114 #define GETNUM(n, val) \
115 for (; p < end && rb_enc_isdigit(*p, enc); p++) { \
116 int next_n = 10 * n + (*p - '0'); \
117 if (next_n / 10 != n) {\
118 rb_raise(rb_eArgError, #val " too big"); \
120 n = next_n; \
122 if (p >= end) { \
123 rb_raise(rb_eArgError, "malformed format string - %%*[0-9]"); \
126 #define GETASTER(val) do { \
127 t = p++; \
128 n = 0; \
129 GETNUM(n, val); \
130 if (*p == '$') { \
131 tmp = GETPOSARG(n); \
133 else { \
134 tmp = GETARG(); \
135 p = t; \
137 val = NUM2INT(tmp); \
138 } while (0)
142 * call-seq:
143 * format(format_string [, arguments...] ) => string
144 * sprintf(format_string [, arguments...] ) => string
146 * Returns the string resulting from applying <i>format_string</i> to
147 * any additional arguments. Within the format string, any characters
148 * other than format sequences are copied to the result.
150 * The syntax of a format sequence is follows.
152 * %[flags][width][.precision]type
154 * A format
155 * sequence consists of a percent sign, followed by optional flags,
156 * width, and precision indicators, then terminated with a field type
157 * character. The field type controls how the corresponding
158 * <code>sprintf</code> argument is to be interpreted, while the flags
159 * modify that interpretation.
161 * The field type characters are:
163 * Field | Integer Format
164 * ------+--------------------------------------------------------------
165 * b | Convert argument as a binary number.
166 * | Negative numbers will be displayed as a two's complement
167 * | prefixed with `..1'.
168 * B | Equivalent to `b', but uses an uppercase 0B for prefix
169 * | in the alternative format by #.
170 * d | Convert argument as a decimal number.
171 * i | Identical to `d'.
172 * o | Convert argument as an octal number.
173 * | Negative numbers will be displayed as a two's complement
174 * | prefixed with `..7'.
175 * u | Identical to `d'.
176 * x | Convert argument as a hexadecimal number.
177 * | Negative numbers will be displayed as a two's complement
178 * | prefixed with `..f' (representing an infinite string of
179 * | leading 'ff's).
180 * X | Equivalent to `x', but uses uppercase letters.
182 * Field | Float Format
183 * ------+--------------------------------------------------------------
184 * e | Convert floating point argument into exponential notation
185 * | with one digit before the decimal point as [-]d.dddddde[+-]dd.
186 * | The precision specifies the number of digits after the decimal
187 * | point (defaulting to six).
188 * E | Equivalent to `e', but uses an uppercase E to indicate
189 * | the exponent.
190 * f | Convert floating point argument as [-]ddd.dddddd,
191 * | where the precision specifies the number of digits after
192 * | the decimal point.
193 * g | Convert a floating point number using exponential form
194 * | if the exponent is less than -4 or greater than or
195 * | equal to the precision, or in dd.dddd form otherwise.
196 * | The precision specifies the number of significant digits.
197 * G | Equivalent to `g', but use an uppercase `E' in exponent form.
199 * Field | Other Format
200 * ------+--------------------------------------------------------------
201 * c | Argument is the numeric code for a single character or
202 * | a single character string itself.
203 * p | The valuing of argument.inspect.
204 * s | Argument is a string to be substituted. If the format
205 * | sequence contains a precision, at most that many characters
206 * | will be copied.
208 * The flags modifies the behavior of the formats.
209 * The flag characters are:
211 * Flag | Applies to | Meaning
212 * ---------+---------------+-----------------------------------------
213 * space | bBdiouxX | Leave a space at the start of
214 * | eEfgG | non-negative numbers.
215 * | (numeric fmt) | For `o', `x', `X', `b' and `B', use
216 * | | a minus sign with absolute value for
217 * | | negative values.
218 * ---------+---------------+-----------------------------------------
219 * (digit)$ | all | Specifies the absolute argument number
220 * | | for this field. Absolute and relative
221 * | | argument numbers cannot be mixed in a
222 * | | sprintf string.
223 * ---------+---------------+-----------------------------------------
224 * # | bBoxX | Use an alternative format.
225 * | eEfgG | For the conversions `o', increase the precision
226 * | | until the first digit will be `0' if
227 * | | it is not formatted as complements.
228 * | | For the conversions `x', `X', `b' and `B'
229 * | | on non-zero, prefix the result with ``0x'',
230 * | | ``0X'', ``0b'' and ``0B'', respectively.
231 * | | For `e', `E', `f', `g', and 'G',
232 * | | force a decimal point to be added,
233 * | | even if no digits follow.
234 * | | For `g' and 'G', do not remove trailing zeros.
235 * ---------+---------------+-----------------------------------------
236 * + | bBdiouxX | Add a leading plus sign to non-negative
237 * | eEfgG | numbers.
238 * | (numeric fmt) | For `o', `x', `X', `b' and `B', use
239 * | | a minus sign with absolute value for
240 * | | negative values.
241 * ---------+---------------+-----------------------------------------
242 * - | all | Left-justify the result of this conversion.
243 * ---------+---------------+-----------------------------------------
244 * 0 (zero) | bBdiouxX | Pad with zeros, not spaces.
245 * | eEfgG | For `o', `x', `X', `b' and `B', radix-1
246 * | (numeric fmt) | is used for negative numbers formatted as
247 * | | complements.
248 * ---------+---------------+-----------------------------------------
249 * * | all | Use the next argument as the field width.
250 * | | If negative, left-justify the result. If the
251 * | | asterisk is followed by a number and a dollar
252 * | | sign, use the indicated argument as the width.
254 * Examples of flags:
256 * # `+' and space flag specifies the sign of non-negative numbers.
257 * sprintf("%d", 123) #=> "123"
258 * sprintf("%+d", 123) #=> "+123"
259 * sprintf("% d", 123) #=> " 123"
261 * # `#' flag for `o' increases number of digits to show `0'.
262 * # `+' and space flag changes format of negative numbers.
263 * sprintf("%o", 123) #=> "173"
264 * sprintf("%#o", 123) #=> "0173"
265 * sprintf("%+o", -123) #=> "-173"
266 * sprintf("%o", -123) #=> "..7605"
267 * sprintf("%#o", -123) #=> "..7605"
269 * # `#' flag for `x' add a prefix `0x' for non-zero numbers.
270 * # `+' and space flag disables complements for negative numbers.
271 * sprintf("%x", 123) #=> "7b"
272 * sprintf("%#x", 123) #=> "0x7b"
273 * sprintf("%+x", -123) #=> "-7b"
274 * sprintf("%x", -123) #=> "..f85"
275 * sprintf("%#x", -123) #=> "0x..f85"
276 * sprintf("%#x", 0) #=> "0"
278 * # `#' for `X' uses the prefix `0X'.
279 * sprintf("%X", 123) #=> "7B"
280 * sprintf("%#X", 123) #=> "0X7B"
282 * # `#' flag for `b' add a prefix `0b' for non-zero numbers.
283 * # `+' and space flag disables complements for negative numbers.
284 * sprintf("%b", 123) #=> "1111011"
285 * sprintf("%#b", 123) #=> "0b1111011"
286 * sprintf("%+b", -123) #=> "-1111011"
287 * sprintf("%b", -123) #=> "..10000101"
288 * sprintf("%#b", -123) #=> "0b..10000101"
289 * sprintf("%#b", 0) #=> "0"
291 * # `#' for `B' uses the prefix `0B'.
292 * sprintf("%B", 123) #=> "1111011"
293 * sprintf("%#B", 123) #=> "0B1111011"
295 * # `#' for `e' forces to show the decimal point.
296 * sprintf("%.0e", 1) #=> "1e+00"
297 * sprintf("%#.0e", 1) #=> "1.e+00"
299 * # `#' for `f' forces to show the decimal point.
300 * sprintf("%.0f", 1234) #=> "1234"
301 * sprintf("%#.0f", 1234) #=> "1234."
303 * # `#' for `g' forces to show the decimal point.
304 * # It also disables stripping lowest zeros.
305 * sprintf("%g", 123.4) #=> "123.4"
306 * sprintf("%#g", 123.4) #=> "123.400"
307 * sprintf("%g", 123456) #=> "123456"
308 * sprintf("%#g", 123456) #=> "123456."
310 * The field width is an optional integer, followed optionally by a
311 * period and a precision. The width specifies the minimum number of
312 * characters that will be written to the result for this field.
314 * Examples of width:
316 * # padding is done by spaces, width=20
317 * # 0 or radix-1. <------------------>
318 * sprintf("%20d", 123) #=> " 123"
319 * sprintf("%+20d", 123) #=> " +123"
320 * sprintf("%020d", 123) #=> "00000000000000000123"
321 * sprintf("%+020d", 123) #=> "+0000000000000000123"
322 * sprintf("% 020d", 123) #=> " 0000000000000000123"
323 * sprintf("%-20d", 123) #=> "123 "
324 * sprintf("%-+20d", 123) #=> "+123 "
325 * sprintf("%- 20d", 123) #=> " 123 "
326 * sprintf("%020x", -123) #=> "..ffffffffffffffff85"
328 * For
329 * numeric fields, the precision controls the number of decimal places
330 * displayed. For string fields, the precision determines the maximum
331 * number of characters to be copied from the string. (Thus, the format
332 * sequence <code>%10.10s</code> will always contribute exactly ten
333 * characters to the result.)
335 * Examples of precisions:
337 * # precision for `d', 'o', 'x' and 'b' is
338 * # minimum number of digits <------>
339 * sprintf("%20.8d", 123) #=> " 00000123"
340 * sprintf("%20.8o", 123) #=> " 00000173"
341 * sprintf("%20.8x", 123) #=> " 0000007b"
342 * sprintf("%20.8b", 123) #=> " 01111011"
343 * sprintf("%20.8d", -123) #=> " -00000123"
344 * sprintf("%20.8o", -123) #=> " ..777605"
345 * sprintf("%20.8x", -123) #=> " ..ffff85"
346 * sprintf("%20.8b", -11) #=> " ..110101"
348 * # "0x" and "0b" for `#x' and `#b' is not counted for
349 * # precision but "0" for `#o' is counted. <------>
350 * sprintf("%#20.8d", 123) #=> " 00000123"
351 * sprintf("%#20.8o", 123) #=> " 00000173"
352 * sprintf("%#20.8x", 123) #=> " 0x0000007b"
353 * sprintf("%#20.8b", 123) #=> " 0b01111011"
354 * sprintf("%#20.8d", -123) #=> " -00000123"
355 * sprintf("%#20.8o", -123) #=> " ..777605"
356 * sprintf("%#20.8x", -123) #=> " 0x..ffff85"
357 * sprintf("%#20.8b", -11) #=> " 0b..110101"
359 * # precision for `e' is number of
360 * # digits after the decimal point <------>
361 * sprintf("%20.8e", 1234.56789) #=> " 1.23456789e+03"
363 * # precision for `f' is number of
364 * # digits after the decimal point <------>
365 * sprintf("%20.8f", 1234.56789) #=> " 1234.56789000"
367 * # precision for `g' is number of
368 * # significant digits <------->
369 * sprintf("%20.8g", 1234.56789) #=> " 1234.5679"
371 * # <------->
372 * sprintf("%20.8g", 123456789) #=> " 1.2345679e+08"
374 * # precision for `s' is
375 * # maximum number of characters <------>
376 * sprintf("%20.8s", "string test") #=> " string t"
378 * Examples:
380 * sprintf("%d %04x", 123, 123) #=> "123 007b"
381 * sprintf("%08b '%4s'", 123, 123) #=> "01111011 ' 123'"
382 * sprintf("%1$*2$s %2$d %1$s", "hello", 8) #=> " hello 8 hello"
383 * sprintf("%1$*2$s %2$d", "hello", -8) #=> "hello -8"
384 * sprintf("%+g:% g:%-g", 1.23, 1.23, 1.23) #=> "+1.23: 1.23:1.23"
385 * sprintf("%u", -123) #=> "-123"
388 VALUE
389 rb_f_sprintf(int argc, const VALUE *argv)
391 return rb_str_format(argc - 1, argv + 1, GETNTHARG(0));
394 VALUE
395 rb_str_format(int argc, const VALUE *argv, VALUE fmt)
397 rb_encoding *enc;
398 const char *p, *end;
399 char *buf;
400 int blen, bsiz;
401 VALUE result;
403 int width, prec, flags = FNONE;
404 int nextarg = 1;
405 int posarg = 0;
406 int tainted = 0;
407 VALUE nextvalue;
408 VALUE tmp;
409 VALUE str;
411 #define CHECK_FOR_WIDTH(f) \
412 if ((f) & FWIDTH) { \
413 rb_raise(rb_eArgError, "width given twice"); \
415 if ((f) & FPREC0) { \
416 rb_raise(rb_eArgError, "width after precision"); \
418 #define CHECK_FOR_FLAGS(f) \
419 if ((f) & FWIDTH) { \
420 rb_raise(rb_eArgError, "flag after width"); \
422 if ((f) & FPREC0) { \
423 rb_raise(rb_eArgError, "flag after precision"); \
426 ++argc;
427 --argv;
428 if (OBJ_TAINTED(fmt)) tainted = 1;
429 StringValue(fmt);
430 enc = rb_enc_get(fmt);
431 fmt = rb_str_new4(fmt);
432 p = RSTRING_PTR(fmt);
433 end = p + RSTRING_LEN(fmt);
434 blen = 0;
435 bsiz = 120;
436 result = rb_str_buf_new(bsiz);
437 rb_enc_copy(result, fmt);
438 buf = RSTRING_PTR(result);
439 memset(buf, 0, bsiz);
441 for (; p < end; p++) {
442 const char *t;
443 int n;
445 for (t = p; t < end && *t != '%'; t++) ;
446 PUSH(p, t - p);
447 if (t >= end) {
448 /* end of fmt string */
449 goto sprint_exit;
451 p = t + 1; /* skip `%' */
453 width = prec = -1;
454 nextvalue = Qundef;
455 retry:
456 switch (*p) {
457 default:
458 if (rb_enc_isprint(*p, enc))
459 rb_raise(rb_eArgError, "malformed format string - %%%c", *p);
460 else
461 rb_raise(rb_eArgError, "malformed format string");
462 break;
464 case ' ':
465 CHECK_FOR_FLAGS(flags);
466 flags |= FSPACE;
467 p++;
468 goto retry;
470 case '#':
471 CHECK_FOR_FLAGS(flags);
472 flags |= FSHARP;
473 p++;
474 goto retry;
476 case '+':
477 CHECK_FOR_FLAGS(flags);
478 flags |= FPLUS;
479 p++;
480 goto retry;
482 case '-':
483 CHECK_FOR_FLAGS(flags);
484 flags |= FMINUS;
485 p++;
486 goto retry;
488 case '0':
489 CHECK_FOR_FLAGS(flags);
490 flags |= FZERO;
491 p++;
492 goto retry;
494 case '1': case '2': case '3': case '4':
495 case '5': case '6': case '7': case '8': case '9':
496 n = 0;
497 GETNUM(n, width);
498 if (*p == '$') {
499 if (nextvalue != Qundef) {
500 rb_raise(rb_eArgError, "value given twice - %d$", n);
502 nextvalue = GETPOSARG(n);
503 p++;
504 goto retry;
506 CHECK_FOR_WIDTH(flags);
507 width = n;
508 flags |= FWIDTH;
509 goto retry;
511 case '*':
512 CHECK_FOR_WIDTH(flags);
513 flags |= FWIDTH;
514 GETASTER(width);
515 if (width < 0) {
516 flags |= FMINUS;
517 width = -width;
519 p++;
520 goto retry;
522 case '.':
523 if (flags & FPREC0) {
524 rb_raise(rb_eArgError, "precision given twice");
526 flags |= FPREC|FPREC0;
528 prec = 0;
529 p++;
530 if (*p == '*') {
531 GETASTER(prec);
532 if (prec < 0) { /* ignore negative precision */
533 flags &= ~FPREC;
535 p++;
536 goto retry;
539 GETNUM(prec, precision);
540 goto retry;
542 case '\n':
543 case '\0':
544 p--;
545 case '%':
546 if (flags != FNONE) {
547 rb_raise(rb_eArgError, "invalid format character - %%");
549 PUSH("%", 1);
550 break;
552 case 'c':
554 VALUE val = GETARG();
555 VALUE tmp;
556 int c, n;
558 tmp = rb_check_string_type(val);
559 if (!NIL_P(tmp)) {
560 if (rb_enc_strlen(RSTRING_PTR(tmp),RSTRING_END(tmp),enc) != 1) {
561 rb_raise(rb_eArgError, "%%c requires a character");
563 c = rb_enc_codepoint(RSTRING_PTR(tmp), RSTRING_END(tmp), enc);
565 else {
566 c = NUM2INT(val);
568 n = rb_enc_codelen(c, enc);
569 if (n == 0) {
570 rb_raise(rb_eArgError, "invalid character");
572 if (!(flags & FWIDTH)) {
573 CHECK(n);
574 rb_enc_mbcput(c, &buf[blen], enc);
575 blen += n;
577 else if ((flags & FMINUS)) {
578 CHECK(n);
579 rb_enc_mbcput(c, &buf[blen], enc);
580 blen += n;
581 FILL(' ', width-1);
583 else {
584 FILL(' ', width-1);
585 CHECK(n);
586 rb_enc_mbcput(c, &buf[blen], enc);
587 blen += n;
590 break;
592 case 's':
593 case 'p':
595 VALUE arg = GETARG();
596 long len, slen;
598 if (*p == 'p') arg = rb_inspect(arg);
599 str = rb_obj_as_string(arg);
600 if (OBJ_TAINTED(str)) tainted = 1;
601 len = RSTRING_LEN(str);
602 enc = rb_enc_check(result, str);
603 if (flags&(FPREC|FWIDTH)) {
604 slen = rb_enc_strlen(RSTRING_PTR(str),RSTRING_END(str),enc);
605 if (slen < 0) {
606 rb_raise(rb_eArgError, "invalid mbstring sequence");
608 if ((flags&FPREC) && (prec < slen)) {
609 char *p = rb_enc_nth(RSTRING_PTR(str), RSTRING_END(str),
610 prec, enc);
611 slen = prec;
612 len = p - RSTRING_PTR(str);
614 /* need to adjust multi-byte string pos */
615 if ((flags&FWIDTH) && (width > slen)) {
616 width -= slen;
617 if (!(flags&FMINUS)) {
618 CHECK(width);
619 while (width--) {
620 buf[blen++] = ' ';
623 CHECK(len);
624 memcpy(&buf[blen], RSTRING_PTR(str), len);
625 blen += len;
626 if (flags&FMINUS) {
627 CHECK(width);
628 while (width--) {
629 buf[blen++] = ' ';
632 rb_enc_associate(result, enc);
633 break;
636 PUSH(RSTRING_PTR(str), len);
637 rb_enc_associate(result, enc);
639 break;
641 case 'd':
642 case 'i':
643 case 'o':
644 case 'x':
645 case 'X':
646 case 'b':
647 case 'B':
648 case 'u':
650 volatile VALUE tmp1;
651 volatile VALUE val = GETARG();
652 char fbuf[32], nbuf[64], *s;
653 const char *prefix = 0;
654 int sign = 0, dots = 0;
655 char sc = 0;
656 long v = 0;
657 int base, bignum = 0;
658 int len, pos;
660 switch (*p) {
661 case 'd':
662 case 'i':
663 case 'u':
664 sign = 1; break;
665 case 'o':
666 case 'x':
667 case 'X':
668 case 'b':
669 case 'B':
670 if (flags&(FPLUS|FSPACE)) sign = 1;
671 break;
673 if (flags & FSHARP) {
674 switch (*p) {
675 case 'o':
676 prefix = "0"; break;
677 case 'x':
678 prefix = "0x"; break;
679 case 'X':
680 prefix = "0X"; break;
681 case 'b':
682 prefix = "0b"; break;
683 case 'B':
684 prefix = "0B"; break;
688 bin_retry:
689 switch (TYPE(val)) {
690 case T_FLOAT:
691 if (FIXABLE(RFLOAT_VALUE(val))) {
692 val = LONG2FIX((long)RFLOAT_VALUE(val));
693 goto bin_retry;
695 val = rb_dbl2big(RFLOAT_VALUE(val));
696 if (FIXNUM_P(val)) goto bin_retry;
697 bignum = 1;
698 break;
699 case T_STRING:
700 val = rb_str_to_inum(val, 0, Qtrue);
701 goto bin_retry;
702 case T_BIGNUM:
703 bignum = 1;
704 break;
705 case T_FIXNUM:
706 v = FIX2LONG(val);
707 break;
708 default:
709 val = rb_Integer(val);
710 goto bin_retry;
713 switch (*p) {
714 case 'o':
715 base = 8; break;
716 case 'x':
717 case 'X':
718 base = 16; break;
719 case 'b':
720 case 'B':
721 base = 2; break;
722 case 'u':
723 case 'd':
724 case 'i':
725 default:
726 base = 10; break;
729 if (!bignum) {
730 if (base == 2) {
731 val = rb_int2big(v);
732 goto bin_retry;
734 if (sign) {
735 char c = *p;
736 if (c == 'i') c = 'd'; /* %d and %i are identical */
737 if (v < 0) {
738 v = -v;
739 sc = '-';
740 width--;
742 else if (flags & FPLUS) {
743 sc = '+';
744 width--;
746 else if (flags & FSPACE) {
747 sc = ' ';
748 width--;
750 sprintf(fbuf, "%%l%c", c);
751 sprintf(nbuf, fbuf, v);
752 s = nbuf;
754 else {
755 s = nbuf;
756 if (v < 0) {
757 dots = 1;
759 sprintf(fbuf, "%%l%c", *p == 'X' ? 'x' : *p);
760 sprintf(++s, fbuf, v);
761 if (v < 0) {
762 char d = 0;
764 s = remove_sign_bits(s, base);
765 switch (base) {
766 case 16:
767 d = 'f'; break;
768 case 8:
769 d = '7'; break;
771 if (d && *s != d) {
772 *--s = d;
777 else {
778 if (sign) {
779 tmp = rb_big2str(val, base);
780 s = RSTRING_PTR(tmp);
781 if (s[0] == '-') {
782 s++;
783 sc = '-';
784 width--;
786 else if (flags & FPLUS) {
787 sc = '+';
788 width--;
790 else if (flags & FSPACE) {
791 sc = ' ';
792 width--;
795 else {
796 if (!RBIGNUM_SIGN(val)) {
797 val = rb_big_clone(val);
798 rb_big_2comp(val);
800 tmp1 = tmp = rb_big2str0(val, base, RBIGNUM_SIGN(val));
801 s = RSTRING_PTR(tmp);
802 if (*s == '-') {
803 dots = 1;
804 if (base == 10) {
805 rb_warning("negative number for %%u specifier");
807 s = remove_sign_bits(++s, base);
808 switch (base) {
809 case 16:
810 if (s[0] != 'f') *--s = 'f'; break;
811 case 8:
812 if (s[0] != '7') *--s = '7'; break;
813 case 2:
814 if (s[0] != '1') *--s = '1'; break;
820 pos = -1;
821 len = strlen(s);
822 if (dots) {
823 prec -= 2;
824 width -= 2;
827 if (*p == 'X') {
828 char *pp = s;
829 int c;
830 while ((c = (int)(unsigned char)*pp) != 0) {
831 *pp = rb_enc_toupper(c, enc);
832 pp++;
835 if (prefix && !prefix[1]) { /* octal */
836 if (dots) {
837 prefix = 0;
839 else if (len == 1 && *s == '0') {
840 len = 0;
841 if (flags & FPREC) prec--;
843 else if ((flags & FPREC) && (prec > len)) {
844 prefix = 0;
847 else if (len == 1 && *s == '0') {
848 prefix = 0;
850 if (prefix) {
851 width -= strlen(prefix);
853 if ((flags & (FZERO|FMINUS|FPREC)) == FZERO) {
854 prec = width;
855 width = 0;
857 else {
858 if (prec < len) {
859 if (!prefix && prec == 0 && len == 1 && *s == '0') len = 0;
860 prec = len;
862 width -= prec;
864 if (!(flags&FMINUS)) {
865 CHECK(width);
866 while (width-- > 0) {
867 buf[blen++] = ' ';
870 if (sc) PUSH(&sc, 1);
871 if (prefix) {
872 int plen = strlen(prefix);
873 PUSH(prefix, plen);
875 CHECK(prec - len);
876 if (dots) PUSH("..", 2);
877 if (!bignum && v < 0) {
878 char c = sign_bits(base, p);
879 while (len < prec--) {
880 buf[blen++] = c;
883 else if ((flags & (FMINUS|FPREC)) != FMINUS) {
884 char c;
886 if (!sign && bignum && !RBIGNUM_SIGN(val))
887 c = sign_bits(base, p);
888 else
889 c = '0';
890 while (len < prec--) {
891 buf[blen++] = c;
894 PUSH(s, len);
895 CHECK(width);
896 while (width-- > 0) {
897 buf[blen++] = ' ';
900 break;
902 case 'f':
903 case 'g':
904 case 'G':
905 case 'e':
906 case 'E':
908 VALUE val = GETARG();
909 double fval;
910 int i, need = 6;
911 char fbuf[32];
913 fval = RFLOAT_VALUE(rb_Float(val));
914 if (isnan(fval) || isinf(fval)) {
915 const char *expr;
917 if (isnan(fval)) {
918 expr = "NaN";
920 else {
921 expr = "Inf";
923 need = strlen(expr);
924 if ((!isnan(fval) && fval < 0.0) || (flags & FPLUS))
925 need++;
926 if ((flags & FWIDTH) && need < width)
927 need = width;
929 CHECK(need);
930 sprintf(&buf[blen], "%*s", need, "");
931 if (flags & FMINUS) {
932 if (!isnan(fval) && fval < 0.0)
933 buf[blen++] = '-';
934 else if (flags & FPLUS)
935 buf[blen++] = '+';
936 else if (flags & FSPACE)
937 blen++;
938 strncpy(&buf[blen], expr, strlen(expr));
940 else {
941 if (!isnan(fval) && fval < 0.0)
942 buf[blen + need - strlen(expr) - 1] = '-';
943 else if (flags & FPLUS)
944 buf[blen + need - strlen(expr) - 1] = '+';
945 else if ((flags & FSPACE) && need > width)
946 blen++;
947 strncpy(&buf[blen + need - strlen(expr)], expr,
948 strlen(expr));
950 blen += strlen(&buf[blen]);
951 break;
954 fmt_setup(fbuf, *p, flags, width, prec);
955 need = 0;
956 if (*p != 'e' && *p != 'E') {
957 i = INT_MIN;
958 frexp(fval, &i);
959 if (i > 0)
960 need = BIT_DIGITS(i);
962 need += (flags&FPREC) ? prec : 6;
963 if ((flags&FWIDTH) && need < width)
964 need = width;
965 need += 20;
967 CHECK(need);
968 sprintf(&buf[blen], fbuf, fval);
969 blen += strlen(&buf[blen]);
971 break;
973 flags = FNONE;
976 sprint_exit:
977 /* XXX - We cannot validate the number of arguments if (digit)$ style used.
979 if (posarg >= 0 && nextarg < argc) {
980 const char *mesg = "too many arguments for format string";
981 if (RTEST(ruby_debug)) rb_raise(rb_eArgError, mesg);
982 if (RTEST(ruby_verbose)) rb_warn(mesg);
984 rb_str_resize(result, blen);
986 if (tainted) OBJ_TAINT(result);
987 return result;
990 static void
991 fmt_setup(char *buf, int c, int flags, int width, int prec)
993 *buf++ = '%';
994 if (flags & FSHARP) *buf++ = '#';
995 if (flags & FPLUS) *buf++ = '+';
996 if (flags & FMINUS) *buf++ = '-';
997 if (flags & FZERO) *buf++ = '0';
998 if (flags & FSPACE) *buf++ = ' ';
1000 if (flags & FWIDTH) {
1001 sprintf(buf, "%d", width);
1002 buf += strlen(buf);
1005 if (flags & FPREC) {
1006 sprintf(buf, ".%d", prec);
1007 buf += strlen(buf);
1010 *buf++ = c;
1011 *buf = '\0';
1014 #undef FILE
1015 #define FILE rb_printf_buffer
1016 #define __sbuf rb_printf_sbuf
1017 #define __sFILE rb_printf_sfile
1018 #undef feof
1019 #undef ferror
1020 #undef clearerr
1021 #undef fileno
1022 #if SIZEOF_LONG < SIZEOF_VOIDP
1023 # if SIZEOF_LONG_LONG == SIZEOF_VOIDP
1024 # define _HAVE_SANE_QUAD_
1025 # define _HAVE_LLP64_
1026 # define quad_t LONG_LONG
1027 # define u_quad_t unsigned LONG_LONG
1028 # endif
1029 #endif
1030 #undef vsnprintf
1031 #undef snprintf
1032 #include "missing/vsnprintf.c"
1034 static int
1035 ruby__sfvwrite(register rb_printf_buffer *fp, register struct __suio *uio)
1037 struct __siov *iov;
1038 VALUE result = (VALUE)fp->_bf._base;
1039 char *buf = (char*)fp->_p;
1040 size_t len, n;
1041 int blen = buf - RSTRING_PTR(result), bsiz = fp->_w;
1043 if (RBASIC(result)->klass) {
1044 rb_raise(rb_eRuntimeError, "rb_vsprintf reentered");
1046 if ((len = uio->uio_resid) == 0)
1047 return 0;
1048 CHECK(len);
1049 buf += blen;
1050 fp->_w = bsiz;
1051 for (iov = uio->uio_iov; len > 0; ++iov) {
1052 MEMCPY(buf, iov->iov_base, char, n = iov->iov_len);
1053 buf += n;
1054 len -= n;
1056 fp->_p = (unsigned char *)buf;
1057 return 0;
1060 VALUE
1061 rb_enc_vsprintf(rb_encoding *enc, const char *fmt, va_list ap)
1063 rb_printf_buffer f;
1064 VALUE result;
1066 f._flags = __SWR | __SSTR;
1067 f._bf._size = 0;
1068 f._w = 120;
1069 result = rb_str_buf_new(f._w);
1070 if (enc) rb_enc_associate(result, enc);
1071 f._bf._base = (unsigned char *)result;
1072 f._p = (unsigned char *)RSTRING_PTR(result);
1073 RBASIC(result)->klass = 0;
1074 f.vwrite = ruby__sfvwrite;
1075 BSD_vfprintf(&f, fmt, ap);
1076 RBASIC(result)->klass = rb_cString;
1077 rb_str_resize(result, (char *)f._p - RSTRING_PTR(result));
1079 return result;
1082 VALUE
1083 rb_enc_sprintf(rb_encoding *enc, const char *format, ...)
1085 VALUE result;
1086 va_list ap;
1088 va_start(ap, format);
1089 result = rb_enc_vsprintf(enc, format, ap);
1090 va_end(ap);
1092 return result;
1095 VALUE
1096 rb_vsprintf(const char *fmt, va_list ap)
1098 return rb_enc_vsprintf(NULL, fmt, ap);
1101 VALUE
1102 rb_sprintf(const char *format, ...)
1104 VALUE result;
1105 va_list ap;
1107 va_start(ap, format);
1108 result = rb_vsprintf(format, ap);
1109 va_end(ap);
1111 return result;