static mrb_encoding *
to_encoding(mrb_state *mrb, mrb_value enc)
{
int idx;
//StringValue(enc);
mrb_string_value(mrb, &enc);
if (!mrb_enc_asciicompat(mrb, mrb_enc_get(mrb, enc))) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "invalid name encoding (non ASCII)");
}
//idx = mrb_enc_find_index(StringValueCStr(enc));
idx = mrb_enc_find_index(mrb, mrb_string_value_cstr(mrb, &enc));
if (idx < 0) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "unknown encoding name - %s", RSTRING_PTR(enc));
}
return mrb_enc_from_index(mrb, idx);
}
int
mrb_to_encoding_index(mrb_state *mrb, mrb_value enc)
{
int idx;
idx = enc_check_encoding(mrb, enc);
if (idx >= 0) {
return idx;
}
else if (mrb_nil_p(enc = mrb_check_string_type(mrb, enc))) {
return -1;
}
if (!mrb_enc_asciicompat(mrb, mrb_enc_get(mrb, enc))) {
return -1;
}
//return mrb_enc_find_index(StringValueCStr(enc));
return mrb_enc_find_index(mrb, mrb_string_value_cstr(mrb, &enc));
}
mrb_value
mrb_str_format(mrb_state *mrb, int argc, const mrb_value *argv, mrb_value fmt)
{
const char *p, *end;
char *buf;
long blen, bsiz;
mrb_value result;
int width, prec, flags = FNONE;
int nextarg = 1;
int posarg = 0;
mrb_value nextvalue;
mrb_value tmp;
mrb_value str;
mrb_value hash = mrb_undef_value();
#define CHECK_FOR_WIDTH(f) \
if ((f) & FWIDTH) { \
mrb_raise(mrb, E_ARGUMENT_ERROR, "width given twice"); \
} \
if ((f) & FPREC0) { \
mrb_raise(mrb, E_ARGUMENT_ERROR, "width after precision"); \
}
#define CHECK_FOR_FLAGS(f) \
if ((f) & FWIDTH) { \
mrb_raise(mrb, E_ARGUMENT_ERROR, "flag after width"); \
} \
if ((f) & FPREC0) { \
mrb_raise(mrb, E_ARGUMENT_ERROR, "flag after precision"); \
}
++argc;
--argv;
mrb_string_value(mrb, &fmt);
fmt = mrb_str_new4(mrb, fmt);
p = RSTRING_PTR(fmt);
end = p + RSTRING_LEN(fmt);
blen = 0;
bsiz = 120;
result = mrb_str_buf_new(mrb, bsiz);
buf = RSTRING_PTR(result);
memset(buf, 0, bsiz);
for (; p < end; p++) {
const char *t;
int n;
mrb_sym id = 0;
for (t = p; t < end && *t != '%'; t++) ;
PUSH(p, t - p);
if (t >= end)
goto sprint_exit; /* end of fmt string */
p = t + 1; /* skip `%' */
width = prec = -1;
nextvalue = mrb_undef_value();
retry:
switch (*p) {
default:
mrb_raise(mrb, E_ARGUMENT_ERROR, "malformed format string - %%%c", *p);
break;
case ' ':
CHECK_FOR_FLAGS(flags);
flags |= FSPACE;
p++;
goto retry;
case '#':
CHECK_FOR_FLAGS(flags);
flags |= FSHARP;
p++;
goto retry;
case '+':
CHECK_FOR_FLAGS(flags);
flags |= FPLUS;
p++;
goto retry;
case '-':
CHECK_FOR_FLAGS(flags);
flags |= FMINUS;
p++;
goto retry;
case '0':
CHECK_FOR_FLAGS(flags);
flags |= FZERO;
p++;
goto retry;
case '1': case '2': case '3': case '4':
case '5': case '6': case '7': case '8': case '9':
n = 0;
GETNUM(n, width);
if (*p == '$') {
if (!UNDEF_P(nextvalue)) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "value given twice - %d$", n);
}
nextvalue = GETPOSARG(n);
p++;
goto retry;
}
CHECK_FOR_WIDTH(flags);
width = n;
flags |= FWIDTH;
goto retry;
case '<':
case '{':
{
const char *start = p;
char term = (*p == '<') ? '>' : '}';
mrb_value symname;
for (; p < end && *p != term; )
p++;
if (id) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "name%.*s after <%s>",
(int)(p - start + 1), start, mrb_sym2name(mrb, id));
}
symname = mrb_str_new(mrb, start + 1, p - start - 1);
id = mrb_intern(mrb, RSTRING_PTR(symname));
nextvalue = GETNAMEARG(mrb_symbol_value(id), start, (int)(p - start + 1));
if (UNDEF_P(nextvalue)) {
mrb_raise(mrb, E_KEY_ERROR, "key%.*s not found", (int)(p - start + 1), start);
}
if (term == '}') goto format_s;
p++;
goto retry;
}
case '*':
CHECK_FOR_WIDTH(flags);
flags |= FWIDTH;
GETASTER(width);
if (width < 0) {
flags |= FMINUS;
width = -width;
}
p++;
goto retry;
case '.':
if (flags & FPREC0) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "precision given twice");
}
flags |= FPREC|FPREC0;
prec = 0;
p++;
if (*p == '*') {
GETASTER(prec);
if (prec < 0) { /* ignore negative precision */
flags &= ~FPREC;
}
p++;
goto retry;
}
GETNUM(prec, precision);
goto retry;
case '\n':
case '\0':
p--;
case '%':
if (flags != FNONE) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "invalid format character - %%");
}
PUSH("%", 1);
break;
case 'c':
{
mrb_value val = GETARG();
mrb_value tmp;
unsigned int c;
int n;
#ifdef INCLUDE_ENCODING
mrb_encoding *enc = mrb_enc_get(mrb, fmt);
#endif //INCLUDE_ENCODING
tmp = mrb_check_string_type(mrb, val);
if (!mrb_nil_p(tmp)) {
if (RSTRING_LEN(tmp) != 1 ) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "%%c requires a character");
}
#ifdef INCLUDE_ENCODING
c = mrb_enc_codepoint_len(mrb, RSTRING_PTR(tmp), RSTRING_END(tmp), &n, enc);
#else
c = RSTRING_PTR(tmp)[0];
n = 1;
#endif //INCLUDE_ENCODING
}
else {
c = mrb_fixnum(val);
n = mrb_enc_codelen(mrb, c, enc);
}
if (n <= 0) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "invalid character");
}
if (!(flags & FWIDTH)) {
CHECK(n);
mrb_enc_mbcput(c, &buf[blen], enc);
blen += n;
}
else if ((flags & FMINUS)) {
CHECK(n);
mrb_enc_mbcput(c, &buf[blen], enc);
blen += n;
FILL(' ', width-1);
}
else {
FILL(' ', width-1);
CHECK(n);
mrb_enc_mbcput(c, &buf[blen], enc);
blen += n;
}
}
break;
case 's':
case 'p':
format_s:
{
mrb_value arg = GETARG();
long len, slen;
#ifdef INCLUDE_ENCODING
mrb_encoding *enc = mrb_enc_get(mrb, fmt);
#endif //INCLUDE_ENCODING
if (*p == 'p') arg = mrb_inspect(mrb, arg);
str = mrb_obj_as_string(mrb, arg);
len = RSTRING_LEN(str);
mrb_str_set_len(mrb, result, blen);
if (flags&(FPREC|FWIDTH)) {
slen = RSTRING_LEN(str);
if (slen < 0) {
mrb_raise(mrb, E_ARGUMENT_ERROR, "invalid mbstring sequence");
}
if ((flags&FPREC) && (prec < slen)) {
#ifdef INCLUDE_ENCODING
char *p = mrb_enc_nth(mrb, RSTRING_PTR(str), RSTRING_END(str),prec, enc);
#else
char *p = RSTRING_PTR(str) + prec;
#endif //INCLUDE_ENCODING
slen = prec;
len = p - RSTRING_PTR(str);
}
/* need to adjust multi-byte string pos */
if ((flags&FWIDTH) && (width > slen)) {
width -= (int)slen;
if (!(flags&FMINUS)) {
CHECK(width);
while (width--) {
buf[blen++] = ' ';
}
}
CHECK(len);
memcpy(&buf[blen], RSTRING_PTR(str), len);
blen += len;
if (flags&FMINUS) {
CHECK(width);
while (width--) {
buf[blen++] = ' ';
}
}
mrb_enc_associate(mrb, result, enc);
break;
}
}
PUSH(RSTRING_PTR(str), len);
mrb_enc_associate(mrb, result, enc);
}
break;
case 'd':
case 'i':
case 'o':
case 'x':
case 'X':
case 'b':
case 'B':
case 'u':
{
mrb_value val = GETARG();
char fbuf[32], nbuf[64], *s;
const char *prefix = 0;
int sign = 0, dots = 0;
char sc = 0;
long v = 0, org_v = 0;
int base;
int len;
switch (*p) {
case 'd':
case 'i':
case 'u':
sign = 1; break;
case 'o':
case 'x':
case 'X':
case 'b':
case 'B':
if (flags&(FPLUS|FSPACE)) sign = 1;
break;
}
if (flags & FSHARP) {
switch (*p) {
case 'o': prefix = "0"; break;
case 'x': prefix = "0x"; break;
case 'X': prefix = "0X"; break;
case 'b': prefix = "0b"; break;
case 'B': prefix = "0B"; break;
}
}
bin_retry:
switch (mrb_type(val)) {
case MRB_TT_FLOAT:
if (FIXABLE(mrb_float(val))) {
val = mrb_fixnum_value((mrb_int)mrb_float(val));
goto bin_retry;
}
val = mrb_flt2big(mrb, mrb_float(val));
if (FIXNUM_P(val)) goto bin_retry;
break;
case MRB_TT_STRING:
val = mrb_str_to_inum(mrb, val, 0, TRUE);
goto bin_retry;
case MRB_TT_FIXNUM:
v = (long)mrb_fixnum(val);
break;
default:
val = mrb_Integer(mrb, val);
goto bin_retry;
}
switch (*p) {
case 'o':
base = 8; break;
case 'x':
case 'X':
base = 16; break;
case 'b':
case 'B':
base = 2; break;
case 'u':
case 'd':
case 'i':
default:
base = 10; break;
}
if (base == 2) {
org_v = v;
if ( v < 0 && !sign ) {
val = mrb_fix2binstr(mrb, mrb_fixnum_value(v), base);
dots = 1;
}
else {
val = mrb_fix2str(mrb, mrb_fixnum_value(v), base);
}
v = mrb_fixnum(mrb_str_to_inum(mrb, val, 10, 0/*Qfalse*/));
}
if (sign) {
char c = *p;
if (c == 'i') c = 'd'; /* %d and %i are identical */
if (base == 2) c = 'd';
if (v < 0) {
v = -v;
sc = '-';
width--;
}
else if (flags & FPLUS) {
sc = '+';
width--;
}
else if (flags & FSPACE) {
sc = ' ';
width--;
}
snprintf(fbuf, sizeof(fbuf), "%%l%c", c);
snprintf(nbuf, sizeof(nbuf), fbuf, v);
s = nbuf;
}
else {
char c = *p;
if (c == 'X') c = 'x';
if (base == 2) c = 'd';
s = nbuf;
if (v < 0) {
dots = 1;
}
snprintf(fbuf, sizeof(fbuf), "%%l%c", c);
snprintf(++s, sizeof(nbuf) - 1, fbuf, v);
if (v < 0) {
char d = 0;
s = remove_sign_bits(s, base);
switch (base) {
case 16: d = 'f'; break;
case 8: d = '7'; break;
case 2: d = '1'; break;
}
if (d && *s != d) {
*--s = d;
}
}
}
len = (int)strlen(s);
if (dots) {
prec -= 2;
width -= 2;
}
if (*p == 'X') {
char *pp = s;
int c;
#ifdef INCLUDE_ENCODING
mrb_encoding *enc = mrb_enc_get(mrb, fmt);
#endif //INCLUDE_ENCODING
while ((c = (int)(unsigned char)*pp) != 0) {
#ifdef INCLUDE_ENCODING
*pp = mrb_enc_toupper(c, enc);
#else
*pp = toupper(c);
#endif //INCLUDE_ENCODING
pp++;
}
}
if (prefix && !prefix[1]) { /* octal */
if (dots) {
prefix = 0;
}
else if (len == 1 && *s == '0') {
len = 0;
if (flags & FPREC) prec--;
}
else if ((flags & FPREC) && (prec > len)) {
prefix = 0;
}
}
else if (len == 1 && *s == '0') {
prefix = 0;
}
if (prefix) {
width -= (int)strlen(prefix);
}
if ((flags & (FZERO|FMINUS|FPREC)) == FZERO) {
prec = width;
width = 0;
}
else {
if (prec < len) {
if (!prefix && prec == 0 && len == 1 && *s == '0') len = 0;
prec = len;
}
width -= prec;
}
if (!(flags&FMINUS)) {
CHECK(width);
while (width-- > 0) {
buf[blen++] = ' ';
}
}
if (sc) PUSH(&sc, 1);
if (prefix) {
int plen = (int)strlen(prefix);
PUSH(prefix, plen);
}
CHECK(prec - len);
if (dots) PUSH("..", 2);
if (v < 0 || (base == 2 && org_v < 0)) {
char c = sign_bits(base, p);
while (len < prec--) {
buf[blen++] = c;
}
}
else if ((flags & (FMINUS|FPREC)) != FMINUS) {
char c = '0';
while (len < prec--) {
buf[blen++] = c;
}
}
PUSH(s, len);
CHECK(width);
while (width-- > 0) {
buf[blen++] = ' ';
}
}
break;
case 'f':
case 'g':
case 'G':
case 'e':
case 'E':
case 'a':
case 'A':
{
mrb_value val = GETARG();
double fval;
int i, need = 6;
char fbuf[32];
fval = mrb_float(mrb_Float(mrb, val));
if (isnan(fval) || isinf(fval)) {
const char *expr;
if (isnan(fval)) {
expr = "NaN";
}
else {
expr = "Inf";
}
need = (int)strlen(expr);
if ((!isnan(fval) && fval < 0.0) || (flags & FPLUS))
need++;
if ((flags & FWIDTH) && need < width)
need = width;
CHECK(need + 1);
snprintf(&buf[blen], need + 1, "%*s", need, "");
if (flags & FMINUS) {
if (!isnan(fval) && fval < 0.0)
buf[blen++] = '-';
else if (flags & FPLUS)
buf[blen++] = '+';
else if (flags & FSPACE)
blen++;
memcpy(&buf[blen], expr, strlen(expr));
}
else {
if (!isnan(fval) && fval < 0.0)
buf[blen + need - strlen(expr) - 1] = '-';
else if (flags & FPLUS)
buf[blen + need - strlen(expr) - 1] = '+';
else if ((flags & FSPACE) && need > width)
blen++;
memcpy(&buf[blen + need - strlen(expr)], expr,
strlen(expr));
}
blen += strlen(&buf[blen]);
break;
}
fmt_setup(fbuf, sizeof(fbuf), *p, flags, width, prec);
need = 0;
if (*p != 'e' && *p != 'E') {
i = INT_MIN;
frexp(fval, &i);
if (i > 0)
need = BIT_DIGITS(i);
}
need += (flags&FPREC) ? prec : 6;
if ((flags&FWIDTH) && need < width)
need = width;
need += 20;
CHECK(need);
snprintf(&buf[blen], need, fbuf, fval);
blen += strlen(&buf[blen]);
}
break;
}
flags = FNONE;
}
sprint_exit:
/* XXX - We cannot validate the number of arguments if (digit)$ style used.
*/
if (posarg >= 0 && nextarg < argc) {
const char *mesg = "too many arguments for format string";
if (RTEST(ruby_debug)) mrb_raise(mrb, E_ARGUMENT_ERROR, "%s", mesg);
if (RTEST(ruby_verbose)) mrb_warn("%s", mesg);
}
mrb_str_resize(mrb, result, blen);
return result;
}