* configure.in (rb_cv_broken_glibc_ia64_erfc): renamed from
[ruby-svn.git] / sprintf.c
blob0fa93f241aa9261d84db3549482c466677674973
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 #ifdef HAVE_IEEEFP_H
21 #include <ieeefp.h>
22 #endif
24 #define BIT_DIGITS(N) (((N)*146)/485 + 1) /* log2(10) =~ 146/485 */
25 #define BITSPERDIG (SIZEOF_BDIGITS*CHAR_BIT)
26 #define EXTENDSIGN(n, l) (((~0 << (n)) >> (((n)*(l)) % BITSPERDIG)) & ~(~0 << (n)))
28 static void fmt_setup(char*,int,int,int,int);
30 static char*
31 remove_sign_bits(char *str, int base)
33 char *s, *t;
35 s = t = str;
37 if (base == 16) {
38 while (*t == 'f') {
39 t++;
42 else if (base == 8) {
43 *t |= EXTENDSIGN(3, strlen(t));
44 while (*t == '7') {
45 t++;
48 else if (base == 2) {
49 while (*t == '1') {
50 t++;
54 return t;
57 static char
58 sign_bits(int base, const char *p)
60 char c = '.';
62 switch (base) {
63 case 16:
64 if (*p == 'X') c = 'F';
65 else c = 'f';
66 break;
67 case 8:
68 c = '7'; break;
69 case 2:
70 c = '1'; break;
72 return c;
75 #define FNONE 0
76 #define FSHARP 1
77 #define FMINUS 2
78 #define FPLUS 4
79 #define FZERO 8
80 #define FSPACE 16
81 #define FWIDTH 32
82 #define FPREC 64
83 #define FPREC0 128
85 #define CHECK(l) do {\
86 while (blen + (l) >= bsiz) {\
87 bsiz*=2;\
89 rb_str_resize(result, bsiz);\
90 buf = RSTRING_PTR(result);\
91 } while (0)
93 #define PUSH(s, l) do { \
94 CHECK(l);\
95 memcpy(&buf[blen], s, l);\
96 blen += (l);\
97 } while (0)
99 #define FILL(c, l) do { \
100 CHECK(l);\
101 memset(&buf[blen], c, l);\
102 blen += (l);\
103 } while (0)
105 #define GETARG() (nextvalue != Qundef ? nextvalue : \
106 posarg < 0 ? \
107 (rb_raise(rb_eArgError, "unnumbered(%d) mixed with numbered", nextarg), 0) : \
108 (posarg = nextarg++, GETNTHARG(posarg)))
110 #define GETPOSARG(n) (posarg > 0 ? \
111 (rb_raise(rb_eArgError, "numbered(%d) after unnumbered(%d)", n, posarg), 0) : \
112 ((n < 1) ? (rb_raise(rb_eArgError, "invalid index - %d$", n), 0) : \
113 (posarg = -1, GETNTHARG(n))))
115 #define GETNTHARG(nth) \
116 ((nth >= argc) ? (rb_raise(rb_eArgError, "too few arguments"), 0) : argv[nth])
118 #define GETNUM(n, val) \
119 for (; p < end && rb_enc_isdigit(*p, enc); p++) { \
120 int next_n = 10 * n + (*p - '0'); \
121 if (next_n / 10 != n) {\
122 rb_raise(rb_eArgError, #val " too big"); \
124 n = next_n; \
126 if (p >= end) { \
127 rb_raise(rb_eArgError, "malformed format string - %%*[0-9]"); \
130 #define GETASTER(val) do { \
131 t = p++; \
132 n = 0; \
133 GETNUM(n, val); \
134 if (*p == '$') { \
135 tmp = GETPOSARG(n); \
137 else { \
138 tmp = GETARG(); \
139 p = t; \
141 val = NUM2INT(tmp); \
142 } while (0)
146 * call-seq:
147 * format(format_string [, arguments...] ) => string
148 * sprintf(format_string [, arguments...] ) => string
150 * Returns the string resulting from applying <i>format_string</i> to
151 * any additional arguments. Within the format string, any characters
152 * other than format sequences are copied to the result.
154 * The syntax of a format sequence is follows.
156 * %[flags][width][.precision]type
158 * A format
159 * sequence consists of a percent sign, followed by optional flags,
160 * width, and precision indicators, then terminated with a field type
161 * character. The field type controls how the corresponding
162 * <code>sprintf</code> argument is to be interpreted, while the flags
163 * modify that interpretation.
165 * The field type characters are:
167 * Field | Integer Format
168 * ------+--------------------------------------------------------------
169 * b | Convert argument as a binary number.
170 * | Negative numbers will be displayed as a two's complement
171 * | prefixed with `..1'.
172 * B | Equivalent to `b', but uses an uppercase 0B for prefix
173 * | in the alternative format by #.
174 * d | Convert argument as a decimal number.
175 * i | Identical to `d'.
176 * o | Convert argument as an octal number.
177 * | Negative numbers will be displayed as a two's complement
178 * | prefixed with `..7'.
179 * u | Identical to `d'.
180 * x | Convert argument as a hexadecimal number.
181 * | Negative numbers will be displayed as a two's complement
182 * | prefixed with `..f' (representing an infinite string of
183 * | leading 'ff's).
184 * X | Equivalent to `x', but uses uppercase letters.
186 * Field | Float Format
187 * ------+--------------------------------------------------------------
188 * e | Convert floating point argument into exponential notation
189 * | with one digit before the decimal point as [-]d.dddddde[+-]dd.
190 * | The precision specifies the number of digits after the decimal
191 * | point (defaulting to six).
192 * E | Equivalent to `e', but uses an uppercase E to indicate
193 * | the exponent.
194 * f | Convert floating point argument as [-]ddd.dddddd,
195 * | where the precision specifies the number of digits after
196 * | the decimal point.
197 * g | Convert a floating point number using exponential form
198 * | if the exponent is less than -4 or greater than or
199 * | equal to the precision, or in dd.dddd form otherwise.
200 * | The precision specifies the number of significant digits.
201 * G | Equivalent to `g', but use an uppercase `E' in exponent form.
203 * Field | Other Format
204 * ------+--------------------------------------------------------------
205 * c | Argument is the numeric code for a single character or
206 * | a single character string itself.
207 * p | The valuing of argument.inspect.
208 * s | Argument is a string to be substituted. If the format
209 * | sequence contains a precision, at most that many characters
210 * | will be copied.
211 * % | A percent sign itself will be displayed. No argument taken.
213 * The flags modifies the behavior of the formats.
214 * The flag characters are:
216 * Flag | Applies to | Meaning
217 * ---------+---------------+-----------------------------------------
218 * space | bBdiouxX | Leave a space at the start of
219 * | eEfgG | non-negative numbers.
220 * | (numeric fmt) | For `o', `x', `X', `b' and `B', use
221 * | | a minus sign with absolute value for
222 * | | negative values.
223 * ---------+---------------+-----------------------------------------
224 * (digit)$ | all | Specifies the absolute argument number
225 * | | for this field. Absolute and relative
226 * | | argument numbers cannot be mixed in a
227 * | | sprintf string.
228 * ---------+---------------+-----------------------------------------
229 * # | bBoxX | Use an alternative format.
230 * | eEfgG | For the conversions `o', increase the precision
231 * | | until the first digit will be `0' if
232 * | | it is not formatted as complements.
233 * | | For the conversions `x', `X', `b' and `B'
234 * | | on non-zero, prefix the result with ``0x'',
235 * | | ``0X'', ``0b'' and ``0B'', respectively.
236 * | | For `e', `E', `f', `g', and 'G',
237 * | | force a decimal point to be added,
238 * | | even if no digits follow.
239 * | | For `g' and 'G', do not remove trailing zeros.
240 * ---------+---------------+-----------------------------------------
241 * + | bBdiouxX | Add a leading plus sign to non-negative
242 * | eEfgG | numbers.
243 * | (numeric fmt) | For `o', `x', `X', `b' and `B', use
244 * | | a minus sign with absolute value for
245 * | | negative values.
246 * ---------+---------------+-----------------------------------------
247 * - | all | Left-justify the result of this conversion.
248 * ---------+---------------+-----------------------------------------
249 * 0 (zero) | bBdiouxX | Pad with zeros, not spaces.
250 * | eEfgG | For `o', `x', `X', `b' and `B', radix-1
251 * | (numeric fmt) | is used for negative numbers formatted as
252 * | | complements.
253 * ---------+---------------+-----------------------------------------
254 * * | all | Use the next argument as the field width.
255 * | | If negative, left-justify the result. If the
256 * | | asterisk is followed by a number and a dollar
257 * | | sign, use the indicated argument as the width.
259 * Examples of flags:
261 * # `+' and space flag specifies the sign of non-negative numbers.
262 * sprintf("%d", 123) #=> "123"
263 * sprintf("%+d", 123) #=> "+123"
264 * sprintf("% d", 123) #=> " 123"
266 * # `#' flag for `o' increases number of digits to show `0'.
267 * # `+' and space flag changes format of negative numbers.
268 * sprintf("%o", 123) #=> "173"
269 * sprintf("%#o", 123) #=> "0173"
270 * sprintf("%+o", -123) #=> "-173"
271 * sprintf("%o", -123) #=> "..7605"
272 * sprintf("%#o", -123) #=> "..7605"
274 * # `#' flag for `x' add a prefix `0x' for non-zero numbers.
275 * # `+' and space flag disables complements for negative numbers.
276 * sprintf("%x", 123) #=> "7b"
277 * sprintf("%#x", 123) #=> "0x7b"
278 * sprintf("%+x", -123) #=> "-7b"
279 * sprintf("%x", -123) #=> "..f85"
280 * sprintf("%#x", -123) #=> "0x..f85"
281 * sprintf("%#x", 0) #=> "0"
283 * # `#' for `X' uses the prefix `0X'.
284 * sprintf("%X", 123) #=> "7B"
285 * sprintf("%#X", 123) #=> "0X7B"
287 * # `#' flag for `b' add a prefix `0b' for non-zero numbers.
288 * # `+' and space flag disables complements for negative numbers.
289 * sprintf("%b", 123) #=> "1111011"
290 * sprintf("%#b", 123) #=> "0b1111011"
291 * sprintf("%+b", -123) #=> "-1111011"
292 * sprintf("%b", -123) #=> "..10000101"
293 * sprintf("%#b", -123) #=> "0b..10000101"
294 * sprintf("%#b", 0) #=> "0"
296 * # `#' for `B' uses the prefix `0B'.
297 * sprintf("%B", 123) #=> "1111011"
298 * sprintf("%#B", 123) #=> "0B1111011"
300 * # `#' for `e' forces to show the decimal point.
301 * sprintf("%.0e", 1) #=> "1e+00"
302 * sprintf("%#.0e", 1) #=> "1.e+00"
304 * # `#' for `f' forces to show the decimal point.
305 * sprintf("%.0f", 1234) #=> "1234"
306 * sprintf("%#.0f", 1234) #=> "1234."
308 * # `#' for `g' forces to show the decimal point.
309 * # It also disables stripping lowest zeros.
310 * sprintf("%g", 123.4) #=> "123.4"
311 * sprintf("%#g", 123.4) #=> "123.400"
312 * sprintf("%g", 123456) #=> "123456"
313 * sprintf("%#g", 123456) #=> "123456."
315 * The field width is an optional integer, followed optionally by a
316 * period and a precision. The width specifies the minimum number of
317 * characters that will be written to the result for this field.
319 * Examples of width:
321 * # padding is done by spaces, width=20
322 * # 0 or radix-1. <------------------>
323 * sprintf("%20d", 123) #=> " 123"
324 * sprintf("%+20d", 123) #=> " +123"
325 * sprintf("%020d", 123) #=> "00000000000000000123"
326 * sprintf("%+020d", 123) #=> "+0000000000000000123"
327 * sprintf("% 020d", 123) #=> " 0000000000000000123"
328 * sprintf("%-20d", 123) #=> "123 "
329 * sprintf("%-+20d", 123) #=> "+123 "
330 * sprintf("%- 20d", 123) #=> " 123 "
331 * sprintf("%020x", -123) #=> "..ffffffffffffffff85"
333 * For
334 * numeric fields, the precision controls the number of decimal places
335 * displayed. For string fields, the precision determines the maximum
336 * number of characters to be copied from the string. (Thus, the format
337 * sequence <code>%10.10s</code> will always contribute exactly ten
338 * characters to the result.)
340 * Examples of precisions:
342 * # precision for `d', 'o', 'x' and 'b' is
343 * # minimum number of digits <------>
344 * sprintf("%20.8d", 123) #=> " 00000123"
345 * sprintf("%20.8o", 123) #=> " 00000173"
346 * sprintf("%20.8x", 123) #=> " 0000007b"
347 * sprintf("%20.8b", 123) #=> " 01111011"
348 * sprintf("%20.8d", -123) #=> " -00000123"
349 * sprintf("%20.8o", -123) #=> " ..777605"
350 * sprintf("%20.8x", -123) #=> " ..ffff85"
351 * sprintf("%20.8b", -11) #=> " ..110101"
353 * # "0x" and "0b" for `#x' and `#b' is not counted for
354 * # precision but "0" for `#o' is counted. <------>
355 * sprintf("%#20.8d", 123) #=> " 00000123"
356 * sprintf("%#20.8o", 123) #=> " 00000173"
357 * sprintf("%#20.8x", 123) #=> " 0x0000007b"
358 * sprintf("%#20.8b", 123) #=> " 0b01111011"
359 * sprintf("%#20.8d", -123) #=> " -00000123"
360 * sprintf("%#20.8o", -123) #=> " ..777605"
361 * sprintf("%#20.8x", -123) #=> " 0x..ffff85"
362 * sprintf("%#20.8b", -11) #=> " 0b..110101"
364 * # precision for `e' is number of
365 * # digits after the decimal point <------>
366 * sprintf("%20.8e", 1234.56789) #=> " 1.23456789e+03"
368 * # precision for `f' is number of
369 * # digits after the decimal point <------>
370 * sprintf("%20.8f", 1234.56789) #=> " 1234.56789000"
372 * # precision for `g' is number of
373 * # significant digits <------->
374 * sprintf("%20.8g", 1234.56789) #=> " 1234.5679"
376 * # <------->
377 * sprintf("%20.8g", 123456789) #=> " 1.2345679e+08"
379 * # precision for `s' is
380 * # maximum number of characters <------>
381 * sprintf("%20.8s", "string test") #=> " string t"
383 * Examples:
385 * sprintf("%d %04x", 123, 123) #=> "123 007b"
386 * sprintf("%08b '%4s'", 123, 123) #=> "01111011 ' 123'"
387 * sprintf("%1$*2$s %2$d %1$s", "hello", 8) #=> " hello 8 hello"
388 * sprintf("%1$*2$s %2$d", "hello", -8) #=> "hello -8"
389 * sprintf("%+g:% g:%-g", 1.23, 1.23, 1.23) #=> "+1.23: 1.23:1.23"
390 * sprintf("%u", -123) #=> "-123"
393 VALUE
394 rb_f_sprintf(int argc, const VALUE *argv)
396 return rb_str_format(argc - 1, argv + 1, GETNTHARG(0));
399 VALUE
400 rb_str_format(int argc, const VALUE *argv, VALUE fmt)
402 rb_encoding *enc;
403 const char *p, *end;
404 char *buf;
405 int blen, bsiz;
406 VALUE result;
408 int width, prec, flags = FNONE;
409 int nextarg = 1;
410 int posarg = 0;
411 int tainted = 0;
412 VALUE nextvalue;
413 VALUE tmp;
414 VALUE str;
416 #define CHECK_FOR_WIDTH(f) \
417 if ((f) & FWIDTH) { \
418 rb_raise(rb_eArgError, "width given twice"); \
420 if ((f) & FPREC0) { \
421 rb_raise(rb_eArgError, "width after precision"); \
423 #define CHECK_FOR_FLAGS(f) \
424 if ((f) & FWIDTH) { \
425 rb_raise(rb_eArgError, "flag after width"); \
427 if ((f) & FPREC0) { \
428 rb_raise(rb_eArgError, "flag after precision"); \
431 ++argc;
432 --argv;
433 if (OBJ_TAINTED(fmt)) tainted = 1;
434 StringValue(fmt);
435 enc = rb_enc_get(fmt);
436 fmt = rb_str_new4(fmt);
437 p = RSTRING_PTR(fmt);
438 end = p + RSTRING_LEN(fmt);
439 blen = 0;
440 bsiz = 120;
441 result = rb_str_buf_new(bsiz);
442 rb_enc_copy(result, fmt);
443 buf = RSTRING_PTR(result);
444 memset(buf, 0, bsiz);
446 for (; p < end; p++) {
447 const char *t;
448 int n;
450 for (t = p; t < end && *t != '%'; t++) ;
451 PUSH(p, t - p);
452 if (t >= end) {
453 /* end of fmt string */
454 goto sprint_exit;
456 p = t + 1; /* skip `%' */
458 width = prec = -1;
459 nextvalue = Qundef;
460 retry:
461 switch (*p) {
462 default:
463 if (rb_enc_isprint(*p, enc))
464 rb_raise(rb_eArgError, "malformed format string - %%%c", *p);
465 else
466 rb_raise(rb_eArgError, "malformed format string");
467 break;
469 case ' ':
470 CHECK_FOR_FLAGS(flags);
471 flags |= FSPACE;
472 p++;
473 goto retry;
475 case '#':
476 CHECK_FOR_FLAGS(flags);
477 flags |= FSHARP;
478 p++;
479 goto retry;
481 case '+':
482 CHECK_FOR_FLAGS(flags);
483 flags |= FPLUS;
484 p++;
485 goto retry;
487 case '-':
488 CHECK_FOR_FLAGS(flags);
489 flags |= FMINUS;
490 p++;
491 goto retry;
493 case '0':
494 CHECK_FOR_FLAGS(flags);
495 flags |= FZERO;
496 p++;
497 goto retry;
499 case '1': case '2': case '3': case '4':
500 case '5': case '6': case '7': case '8': case '9':
501 n = 0;
502 GETNUM(n, width);
503 if (*p == '$') {
504 if (nextvalue != Qundef) {
505 rb_raise(rb_eArgError, "value given twice - %d$", n);
507 nextvalue = GETPOSARG(n);
508 p++;
509 goto retry;
511 CHECK_FOR_WIDTH(flags);
512 width = n;
513 flags |= FWIDTH;
514 goto retry;
516 case '*':
517 CHECK_FOR_WIDTH(flags);
518 flags |= FWIDTH;
519 GETASTER(width);
520 if (width < 0) {
521 flags |= FMINUS;
522 width = -width;
524 p++;
525 goto retry;
527 case '.':
528 if (flags & FPREC0) {
529 rb_raise(rb_eArgError, "precision given twice");
531 flags |= FPREC|FPREC0;
533 prec = 0;
534 p++;
535 if (*p == '*') {
536 GETASTER(prec);
537 if (prec < 0) { /* ignore negative precision */
538 flags &= ~FPREC;
540 p++;
541 goto retry;
544 GETNUM(prec, precision);
545 goto retry;
547 case '\n':
548 case '\0':
549 p--;
550 case '%':
551 if (flags != FNONE) {
552 rb_raise(rb_eArgError, "invalid format character - %%");
554 PUSH("%", 1);
555 break;
557 case 'c':
559 VALUE val = GETARG();
560 VALUE tmp;
561 int c, n;
563 tmp = rb_check_string_type(val);
564 if (!NIL_P(tmp)) {
565 if (rb_enc_strlen(RSTRING_PTR(tmp),RSTRING_END(tmp),enc) != 1) {
566 rb_raise(rb_eArgError, "%%c requires a character");
568 c = rb_enc_codepoint(RSTRING_PTR(tmp), RSTRING_END(tmp), enc);
570 else {
571 c = NUM2INT(val);
573 n = rb_enc_codelen(c, enc);
574 if (n == 0) {
575 rb_raise(rb_eArgError, "invalid character");
577 if (!(flags & FWIDTH)) {
578 CHECK(n);
579 rb_enc_mbcput(c, &buf[blen], enc);
580 blen += n;
582 else if ((flags & FMINUS)) {
583 CHECK(n);
584 rb_enc_mbcput(c, &buf[blen], enc);
585 blen += n;
586 FILL(' ', width-1);
588 else {
589 FILL(' ', width-1);
590 CHECK(n);
591 rb_enc_mbcput(c, &buf[blen], enc);
592 blen += n;
595 break;
597 case 's':
598 case 'p':
600 VALUE arg = GETARG();
601 long len, slen;
603 if (*p == 'p') arg = rb_inspect(arg);
604 str = rb_obj_as_string(arg);
605 if (OBJ_TAINTED(str)) tainted = 1;
606 len = RSTRING_LEN(str);
607 enc = rb_enc_check(result, str);
608 if (flags&(FPREC|FWIDTH)) {
609 slen = rb_enc_strlen(RSTRING_PTR(str),RSTRING_END(str),enc);
610 if (slen < 0) {
611 rb_raise(rb_eArgError, "invalid mbstring sequence");
613 if ((flags&FPREC) && (prec < slen)) {
614 char *p = rb_enc_nth(RSTRING_PTR(str), RSTRING_END(str),
615 prec, enc);
616 slen = prec;
617 len = p - RSTRING_PTR(str);
619 /* need to adjust multi-byte string pos */
620 if ((flags&FWIDTH) && (width > slen)) {
621 width -= slen;
622 if (!(flags&FMINUS)) {
623 CHECK(width);
624 while (width--) {
625 buf[blen++] = ' ';
628 CHECK(len);
629 memcpy(&buf[blen], RSTRING_PTR(str), len);
630 blen += len;
631 if (flags&FMINUS) {
632 CHECK(width);
633 while (width--) {
634 buf[blen++] = ' ';
637 rb_enc_associate(result, enc);
638 break;
641 PUSH(RSTRING_PTR(str), len);
642 rb_enc_associate(result, enc);
644 break;
646 case 'd':
647 case 'i':
648 case 'o':
649 case 'x':
650 case 'X':
651 case 'b':
652 case 'B':
653 case 'u':
655 volatile VALUE tmp1;
656 volatile VALUE val = GETARG();
657 char fbuf[32], nbuf[64], *s;
658 const char *prefix = 0;
659 int sign = 0, dots = 0;
660 char sc = 0;
661 long v = 0;
662 int base, bignum = 0;
663 int len, pos;
665 switch (*p) {
666 case 'd':
667 case 'i':
668 case 'u':
669 sign = 1; break;
670 case 'o':
671 case 'x':
672 case 'X':
673 case 'b':
674 case 'B':
675 if (flags&(FPLUS|FSPACE)) sign = 1;
676 break;
678 if (flags & FSHARP) {
679 switch (*p) {
680 case 'o':
681 prefix = "0"; break;
682 case 'x':
683 prefix = "0x"; break;
684 case 'X':
685 prefix = "0X"; break;
686 case 'b':
687 prefix = "0b"; break;
688 case 'B':
689 prefix = "0B"; break;
693 bin_retry:
694 switch (TYPE(val)) {
695 case T_FLOAT:
696 if (FIXABLE(RFLOAT_VALUE(val))) {
697 val = LONG2FIX((long)RFLOAT_VALUE(val));
698 goto bin_retry;
700 val = rb_dbl2big(RFLOAT_VALUE(val));
701 if (FIXNUM_P(val)) goto bin_retry;
702 bignum = 1;
703 break;
704 case T_STRING:
705 val = rb_str_to_inum(val, 0, Qtrue);
706 goto bin_retry;
707 case T_BIGNUM:
708 bignum = 1;
709 break;
710 case T_FIXNUM:
711 v = FIX2LONG(val);
712 break;
713 default:
714 val = rb_Integer(val);
715 goto bin_retry;
718 switch (*p) {
719 case 'o':
720 base = 8; break;
721 case 'x':
722 case 'X':
723 base = 16; break;
724 case 'b':
725 case 'B':
726 base = 2; break;
727 case 'u':
728 case 'd':
729 case 'i':
730 default:
731 base = 10; break;
734 if (!bignum) {
735 if (base == 2) {
736 val = rb_int2big(v);
737 goto bin_retry;
739 if (sign) {
740 char c = *p;
741 if (c == 'i') c = 'd'; /* %d and %i are identical */
742 if (v < 0) {
743 v = -v;
744 sc = '-';
745 width--;
747 else if (flags & FPLUS) {
748 sc = '+';
749 width--;
751 else if (flags & FSPACE) {
752 sc = ' ';
753 width--;
755 sprintf(fbuf, "%%l%c", c);
756 sprintf(nbuf, fbuf, v);
757 s = nbuf;
759 else {
760 s = nbuf;
761 if (v < 0) {
762 dots = 1;
764 sprintf(fbuf, "%%l%c", *p == 'X' ? 'x' : *p);
765 sprintf(++s, fbuf, v);
766 if (v < 0) {
767 char d = 0;
769 s = remove_sign_bits(s, base);
770 switch (base) {
771 case 16:
772 d = 'f'; break;
773 case 8:
774 d = '7'; break;
776 if (d && *s != d) {
777 *--s = d;
782 else {
783 if (sign) {
784 tmp = rb_big2str(val, base);
785 s = RSTRING_PTR(tmp);
786 if (s[0] == '-') {
787 s++;
788 sc = '-';
789 width--;
791 else if (flags & FPLUS) {
792 sc = '+';
793 width--;
795 else if (flags & FSPACE) {
796 sc = ' ';
797 width--;
800 else {
801 if (!RBIGNUM_SIGN(val)) {
802 val = rb_big_clone(val);
803 rb_big_2comp(val);
805 tmp1 = tmp = rb_big2str0(val, base, RBIGNUM_SIGN(val));
806 s = RSTRING_PTR(tmp);
807 if (*s == '-') {
808 dots = 1;
809 if (base == 10) {
810 rb_warning("negative number for %%u specifier");
812 s = remove_sign_bits(++s, base);
813 switch (base) {
814 case 16:
815 if (s[0] != 'f') *--s = 'f'; break;
816 case 8:
817 if (s[0] != '7') *--s = '7'; break;
818 case 2:
819 if (s[0] != '1') *--s = '1'; break;
825 pos = -1;
826 len = strlen(s);
827 if (dots) {
828 prec -= 2;
829 width -= 2;
832 if (*p == 'X') {
833 char *pp = s;
834 int c;
835 while ((c = (int)(unsigned char)*pp) != 0) {
836 *pp = rb_enc_toupper(c, enc);
837 pp++;
840 if (prefix && !prefix[1]) { /* octal */
841 if (dots) {
842 prefix = 0;
844 else if (len == 1 && *s == '0') {
845 len = 0;
846 if (flags & FPREC) prec--;
848 else if ((flags & FPREC) && (prec > len)) {
849 prefix = 0;
852 else if (len == 1 && *s == '0') {
853 prefix = 0;
855 if (prefix) {
856 width -= strlen(prefix);
858 if ((flags & (FZERO|FMINUS|FPREC)) == FZERO) {
859 prec = width;
860 width = 0;
862 else {
863 if (prec < len) {
864 if (!prefix && prec == 0 && len == 1 && *s == '0') len = 0;
865 prec = len;
867 width -= prec;
869 if (!(flags&FMINUS)) {
870 CHECK(width);
871 while (width-- > 0) {
872 buf[blen++] = ' ';
875 if (sc) PUSH(&sc, 1);
876 if (prefix) {
877 int plen = strlen(prefix);
878 PUSH(prefix, plen);
880 CHECK(prec - len);
881 if (dots) PUSH("..", 2);
882 if (!bignum && v < 0) {
883 char c = sign_bits(base, p);
884 while (len < prec--) {
885 buf[blen++] = c;
888 else if ((flags & (FMINUS|FPREC)) != FMINUS) {
889 char c;
891 if (!sign && bignum && !RBIGNUM_SIGN(val))
892 c = sign_bits(base, p);
893 else
894 c = '0';
895 while (len < prec--) {
896 buf[blen++] = c;
899 PUSH(s, len);
900 CHECK(width);
901 while (width-- > 0) {
902 buf[blen++] = ' ';
905 break;
907 case 'f':
908 case 'g':
909 case 'G':
910 case 'e':
911 case 'E':
913 VALUE val = GETARG();
914 double fval;
915 int i, need = 6;
916 char fbuf[32];
918 fval = RFLOAT_VALUE(rb_Float(val));
919 if (isnan(fval) || isinf(fval)) {
920 const char *expr;
922 if (isnan(fval)) {
923 expr = "NaN";
925 else {
926 expr = "Inf";
928 need = strlen(expr);
929 if ((!isnan(fval) && fval < 0.0) || (flags & FPLUS))
930 need++;
931 if ((flags & FWIDTH) && need < width)
932 need = width;
934 CHECK(need);
935 sprintf(&buf[blen], "%*s", need, "");
936 if (flags & FMINUS) {
937 if (!isnan(fval) && fval < 0.0)
938 buf[blen++] = '-';
939 else if (flags & FPLUS)
940 buf[blen++] = '+';
941 else if (flags & FSPACE)
942 blen++;
943 strncpy(&buf[blen], expr, strlen(expr));
945 else {
946 if (!isnan(fval) && fval < 0.0)
947 buf[blen + need - strlen(expr) - 1] = '-';
948 else if (flags & FPLUS)
949 buf[blen + need - strlen(expr) - 1] = '+';
950 else if ((flags & FSPACE) && need > width)
951 blen++;
952 strncpy(&buf[blen + need - strlen(expr)], expr,
953 strlen(expr));
955 blen += strlen(&buf[blen]);
956 break;
959 fmt_setup(fbuf, *p, flags, width, prec);
960 need = 0;
961 if (*p != 'e' && *p != 'E') {
962 i = INT_MIN;
963 frexp(fval, &i);
964 if (i > 0)
965 need = BIT_DIGITS(i);
967 need += (flags&FPREC) ? prec : 6;
968 if ((flags&FWIDTH) && need < width)
969 need = width;
970 need += 20;
972 CHECK(need);
973 sprintf(&buf[blen], fbuf, fval);
974 blen += strlen(&buf[blen]);
976 break;
978 flags = FNONE;
981 sprint_exit:
982 /* XXX - We cannot validate the number of arguments if (digit)$ style used.
984 if (posarg >= 0 && nextarg < argc) {
985 const char *mesg = "too many arguments for format string";
986 if (RTEST(ruby_debug)) rb_raise(rb_eArgError, "%s", mesg);
987 if (RTEST(ruby_verbose)) rb_warn("%s", mesg);
989 rb_str_resize(result, blen);
991 if (tainted) OBJ_TAINT(result);
992 return result;
995 static void
996 fmt_setup(char *buf, int c, int flags, int width, int prec)
998 *buf++ = '%';
999 if (flags & FSHARP) *buf++ = '#';
1000 if (flags & FPLUS) *buf++ = '+';
1001 if (flags & FMINUS) *buf++ = '-';
1002 if (flags & FZERO) *buf++ = '0';
1003 if (flags & FSPACE) *buf++ = ' ';
1005 if (flags & FWIDTH) {
1006 sprintf(buf, "%d", width);
1007 buf += strlen(buf);
1010 if (flags & FPREC) {
1011 sprintf(buf, ".%d", prec);
1012 buf += strlen(buf);
1015 *buf++ = c;
1016 *buf = '\0';
1019 #undef FILE
1020 #define FILE rb_printf_buffer
1021 #define __sbuf rb_printf_sbuf
1022 #define __sFILE rb_printf_sfile
1023 #undef feof
1024 #undef ferror
1025 #undef clearerr
1026 #undef fileno
1027 #if SIZEOF_LONG < SIZEOF_VOIDP
1028 # if SIZEOF_LONG_LONG == SIZEOF_VOIDP
1029 # define _HAVE_SANE_QUAD_
1030 # define _HAVE_LLP64_
1031 # define quad_t LONG_LONG
1032 # define u_quad_t unsigned LONG_LONG
1033 # endif
1034 #endif
1035 #undef vsnprintf
1036 #undef snprintf
1037 #include "missing/vsnprintf.c"
1039 static int
1040 ruby__sfvwrite(register rb_printf_buffer *fp, register struct __suio *uio)
1042 struct __siov *iov;
1043 VALUE result = (VALUE)fp->_bf._base;
1044 char *buf = (char*)fp->_p;
1045 size_t len, n;
1046 size_t blen = buf - RSTRING_PTR(result), bsiz = fp->_w;
1048 if (RBASIC(result)->klass) {
1049 rb_raise(rb_eRuntimeError, "rb_vsprintf reentered");
1051 if ((len = uio->uio_resid) == 0)
1052 return 0;
1053 CHECK(len);
1054 buf += blen;
1055 fp->_w = bsiz;
1056 for (iov = uio->uio_iov; len > 0; ++iov) {
1057 MEMCPY(buf, iov->iov_base, char, n = iov->iov_len);
1058 buf += n;
1059 len -= n;
1061 fp->_p = (unsigned char *)buf;
1062 return 0;
1065 VALUE
1066 rb_enc_vsprintf(rb_encoding *enc, const char *fmt, va_list ap)
1068 rb_printf_buffer f;
1069 VALUE result;
1071 f._flags = __SWR | __SSTR;
1072 f._bf._size = 0;
1073 f._w = 120;
1074 result = rb_str_buf_new(f._w);
1075 if (enc) rb_enc_associate(result, enc);
1076 f._bf._base = (unsigned char *)result;
1077 f._p = (unsigned char *)RSTRING_PTR(result);
1078 RBASIC(result)->klass = 0;
1079 f.vwrite = ruby__sfvwrite;
1080 BSD_vfprintf(&f, fmt, ap);
1081 RBASIC(result)->klass = rb_cString;
1082 rb_str_resize(result, (char *)f._p - RSTRING_PTR(result));
1084 return result;
1087 VALUE
1088 rb_enc_sprintf(rb_encoding *enc, const char *format, ...)
1090 VALUE result;
1091 va_list ap;
1093 va_start(ap, format);
1094 result = rb_enc_vsprintf(enc, format, ap);
1095 va_end(ap);
1097 return result;
1100 VALUE
1101 rb_vsprintf(const char *fmt, va_list ap)
1103 return rb_enc_vsprintf(NULL, fmt, ap);
1106 VALUE
1107 rb_sprintf(const char *format, ...)
1109 VALUE result;
1110 va_list ap;
1112 va_start(ap, format);
1113 result = rb_vsprintf(format, ap);
1114 va_end(ap);
1116 return result;
1119 VALUE
1120 rb_str_vcatf(VALUE str, const char *fmt, va_list ap)
1122 rb_printf_buffer f;
1123 VALUE klass;
1125 StringValue(str);
1126 rb_str_modify(str);
1127 f._flags = __SWR | __SSTR;
1128 f._bf._size = 0;
1129 f._w = rb_str_capacity(str);
1130 f._bf._base = (unsigned char *)str;
1131 f._p = (unsigned char *)RSTRING_END(str);
1132 klass = RBASIC(str)->klass;
1133 RBASIC(str)->klass = 0;
1134 f.vwrite = ruby__sfvwrite;
1135 BSD_vfprintf(&f, fmt, ap);
1136 RBASIC(str)->klass = klass;
1137 rb_str_resize(str, (char *)f._p - RSTRING_PTR(str));
1139 return str;
1142 VALUE
1143 rb_str_catf(VALUE str, const char *format, ...)
1145 va_list ap;
1147 va_start(ap, format);
1148 str = rb_str_vcatf(str, format, ap);
1149 va_end(ap);
1151 return str;