Unicode support for *printf()

18 years ago by Antony Dovgal — view source

unread

Hello all.

Attached is the patch which adds Unicode support to *printf() functions stack.
We (Andrei and me) made several assumptions that are worth mentioning:

sprintf() and vsprintf():

use runtime_encoding when dealing with Unicode data.

printf() and vprintf():

the result data is converted to output_encoding when formatting is done;
return number of bytes outputted.

fprintf() and vfprintf():

use runtime_encoding, as all conversions are done by underlying streams API;
both functions return the number returned by streams API (which seems to be number of bytes).

I did not run any benchmarks yet, but I don't expect it to cause any major slowdown.
I would like to hear your comments before applying the patch, so don't hesitate to post them.

--
Wbr,
Antony Dovgal

18 years ago by Andrei Zmievski — view source

unread

I know we had discussion about *fprintf() on IRC, but I'm still not
clear on some stuff.

Sara, could you please explain again why UG(unicode) should not be
used as the selection for php_formatted_print/php_u_formatted_print? I
bet a few of us are hazy on the streams Unicode/binary details, so
let's please clarify this and make sure we're on the same page.

Thanks,

-Andrei

Hello all.

Attached is the patch which adds Unicode support to *printf()
functions stack.
We (Andrei and me) made several assumptions that are worth mentioning:

sprintf() and vsprintf(): - use runtime_encoding when dealing with
Unicode data.

printf() and vprintf():

the result data is converted to output_encoding when formatting is
done;

return number of bytes outputted.

fprintf() and vfprintf():

use runtime_encoding, as all conversions are done by underlying
streams API;

both functions return the number returned by streams API (which
seems to be number of bytes).

I did not run any benchmarks yet, but I don't expect it to cause any
major slowdown.
I would like to hear your comments before applying the patch, so don't
hesitate to post them.

--
Wbr, Antony Dovgal
Index: ext/standard/formatted_print.c

RCS file: /repository/php-src/ext/standard/formatted_print.c,v
retrieving revision 1.88
diff -u -p -d -r1.88 formatted_print.c
--- ext/standard/formatted_print.c 7 Dec 2006 20:45:21 -0000 1.88
+++ ext/standard/formatted_print.c 11 Dec 2006 22:44:21 -0000
@@ -40,6 +40,9 @@
#define MAX_FLOAT_DIGITS 38
#define MAX_FLOAT_PRECISION 40

+#define PHP_OUTPUT 0
+#define PHP_RUNTIME 1

#if 0
/* trick to control varargs functions through cpp */

define PRINTF_DEBUG(arg) php_printf arg

@@ -50,7 +53,10 @@
static char hexchars[] = "0123456789abcdef";
static char HEXCHARS[] = "0123456789ABCDEF";

+static UChar u_hexchars[] = {0x30, 0x31, 0x32, 0x33, 0x34, 0x35,
0x36, 0x37, 0x38, 0x39, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66};
+static UChar u_HEXCHARS[] = {0x30, 0x31, 0x32, 0x33, 0x34, 0x35,
0x36, 0x37, 0x38, 0x39, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46};

+/* php_sprintf_appendchar() {{{ */
inline static void
php_sprintf_appendchar(char **buffer, int *pos, int *size, char add
TSRMLS_DC)
{
@@ -62,8 +68,21 @@ php_sprintf_appendchar(char **buffer, in
PRINTF_DEBUG(("sprintf: appending '%c', pos=\n", add, *pos));
(*buffer)[(pos)++] = add;
}
+/ }}} */

+/* php_u_sprintf_appendchar() {{{ */
+inline static void
+php_u_sprintf_appendchar(UChar **buffer, int *pos, int *size, UChar
add TSRMLS_DC)
+{
if ((*pos + 1) >= *size) {
*size <<= 1;
*buffer = eurealloc(*buffer, *size);
}

(*buffer)[(pos)++] = add;
+}
+/ }}} */
+/* php_sprintf_appendstring() {{{ */
inline static void
php_sprintf_appendstring(char **buffer, int *pos, int *size, char
*add,
int min_width, int max_width, char padding,
@@ -112,10 +131,57 @@ php_sprintf_appendstring(char *buffer,
}
}
}
+/ }}} */

+/* php_u_sprintf_appendstring() {{{ */
+inline static void
+php_u_sprintf_appendstring(UChar **buffer, int *pos, int *size, UChar
*add,
				   int min_width, int max_width, UChar padding,
				   int alignment, int len, int neg, int expprec, int  
always_sign)
+{
register int npad;

int req_size;

int copy_len;

copy_len = (expprec ? MIN(max_width, len) : len);

npad = min_width - copy_len;

if (npad < 0) {
npad = 0;
}

req_size = *pos + MAX(min_width, copy_len) + 1;

if (req_size > *size) {
while (req_size > *size) {
	*size <<= 1;
}
*buffer = eurealloc(*buffer, *size);
}

if (alignment == ALIGN_RIGHT) {
if ((neg || always_sign) && padding == 0x30 /* '0' */) {
	(*buffer)[(*pos)++] = (neg) ? 0x2D /* '-' */ : 0x2B /* '+' */;
	add++;
	len--;
	copy_len--;
}
while (npad-- > 0) {
	(*buffer)[(*pos)++] = padding;
}
}

u_memcpy(&(*buffer)[*pos], add, copy_len + 1);

*pos += copy_len;

if (alignment == ALIGN_LEFT) {
while (npad--) {
	(*buffer)[(*pos)++] = padding;
}
}
+}
+/* }}} */
+/* php_sprintf_appendint() {{{ */
inline static void
-php_sprintf_appendint(char **buffer, int *pos, int *size, long number,
+php_sprintf_appendint(char **buffer, int *pos, int *size, long number,
int width, char padding, int alignment,
int always_sign)
{
@@ -155,7 +221,49 @@ php_sprintf_appendint(char *buffer, int
padding, alignment, (NUM_BUF_SIZE - 1) - i,
neg, 0, always_sign);
}
+/ }}} */

+/* php_u_sprintf_appendint() {{{ */
+inline static void
+php_u_sprintf_appendint(UChar **buffer, int *pos, int *size, long
number,
				int width, UChar padding, int alignment,
				int always_sign)
+{
UChar numbuf[NUM_BUF_SIZE];

register unsigned long magn, nmagn;

register unsigned int i = NUM_BUF_SIZE - 1, neg = 0;

if (number < 0) {
neg = 1;
magn = ((unsigned long) -(number + 1)) + 1;
} else {
magn = (unsigned long) number;
}

/* Can't right-pad 0's on integers */

if (alignment==0 && padding== 0x30 /* '0' /) padding = 0x20 / ' '
*/;

numbuf[i] = 0x0A /* '\0' */;

do {
nmagn = magn / 10;
numbuf[--i] = (UChar)(magn - (nmagn * 10)) + 0x30 /* '0' */;
magn = nmagn;
}

while (magn > 0 && i > 0);

if (neg) {
numbuf[--i] = 0x2D /* '-' */;
} else if (always_sign) {
numbuf[--i] = 0x2B /* '+' */;
}

php_u_sprintf_appendstring(buffer, pos, size, &numbuf[i], width, 0,
					 padding, alignment, (NUM_BUF_SIZE - 1) - i,
					 neg, 0, always_sign);
+}
+/* }}} */

+/* php_sprintf_appenduint() {{{ */
inline static void
php_sprintf_appenduint(char **buffer, int *pos, int *size,
unsigned long number,
@@ -185,7 +293,38 @@ php_sprintf_appenduint(char *buffer, in
php_sprintf_appendstring(buffer, pos, size, &numbuf[i], width, 0,
padding, alignment, (NUM_BUF_SIZE - 1) - i, 0, 0, 0);
}
+/ }}} */

+/* php_u_sprintf_appenduint() {{{ */
+inline static void
+php_u_sprintf_appenduint(UChar **buffer, int *pos, int *size,
			   unsigned long number,
			   int width, UChar padding, int alignment)
+{
UChar numbuf[NUM_BUF_SIZE];

register unsigned long magn, nmagn;

register unsigned int i = NUM_BUF_SIZE - 1;

magn = (unsigned int) number;

/* Can't right-pad 0's on integers */

if (alignment == 0 && padding == 0x30 /* '0' /) padding = 0x20 / '
' */;

numbuf[i] = 0x0A /* '\0' */;

do {
nmagn = magn / 10;
numbuf[--i] = (UChar)(magn - (nmagn * 10)) + 0x30 /* '0' */;
magn = nmagn;
} while (magn > 0 && i > 0);

php_u_sprintf_appendstring(buffer, pos, size, &numbuf[i], width, 0,
					 padding, alignment, (NUM_BUF_SIZE - 1) - i, 0, 0, 0);
+}
+/* }}} */

+/* php_sprintf_appenddouble() {{{ */
inline static void
php_sprintf_appenddouble(char **buffer, int *pos,
int *size, double number,
@@ -231,7 +370,7 @@ php_sprintf_appenddouble(char **buffer,
}
case 'E':
case 'f':
	s = ap_php_conv_fp(fmt, number, 0, precision,
	s = ap_php_conv_fp(fmt, number, NO, precision,
				&is_negative, &num_buf[1], &s_len);
	if (is_negative) {
		num_buf[0] = '-';
@@ -272,8 +411,122 @@ php_sprintf_appenddouble(char *buffer,
php_sprintf_appendstring(buffer, pos, size, s, width, 0, padding,
alignment, s_len, is_negative, 0, always_sign);
}
+/ }}} */

+/* php_u_sprintf_appenddouble() {{{ */
+inline static void
+php_u_sprintf_appenddouble(UChar **buffer, int *pos,
				 int *size, double number,
				 int width, UChar padding,
				 int alignment, int precision,
				 int adjust, UChar fmt,
				 int always_sign
				 TSRMLS_DC)
+{
char num_buf[NUM_BUF_SIZE];

char *s = NULL, *q, s_fmt;

UChar *uni_s;

int s_len = 0, is_negative = 0;

if ((adjust & ADJ_PRECISION) == 0) {
precision = FLOAT_PRECISION;
} else if (precision > MAX_FLOAT_PRECISION) {
precision = MAX_FLOAT_PRECISION;
}

if (zend_isnan(number)) {
UChar *nan = USTR_MAKE("NaN");
is_negative = (number<0);
php_u_sprintf_appendstring(buffer, pos, size, nan, 3, 0, padding,
						 alignment, precision, is_negative, 0, always_sign);
efree(nan);
return;
}

if (zend_isinf(number)) {
UChar *inf = USTR_MAKE("INF");
is_negative = (number<0);
php_u_sprintf_appendstring(buffer, pos, size, inf, 3, 0, padding,
						 alignment, precision, is_negative, 0, always_sign);
efree(inf);
return;
}

switch (fmt) {
case 0x66 /* 'f' */:
case 0x46 /* 'F' */:
	s_fmt = 'f';
	break;
case 0x65 /* 'e' */:
	s_fmt = 'e';
	break;
case 0x45 /* 'E' */:
	s_fmt = 'E';
	break;
case 0x67 /* 'g' */:
	s_fmt = 'g';
	break;
case 0x47 /* 'G' */:
	s_fmt = 'G';
	break;
}

switch (fmt) {
case 0x46 /* 'F' */:
	/* break is missing */
case 0x65 /* 'e' */:
	if (precision) {
		precision--;
	}
case 0x45 /* 'E' */:
case 0x66 /* 'f' */:
	s = ap_php_conv_fp(s_fmt, number, NO, precision,
				&is_negative, &num_buf[1], &s_len);
	if (is_negative) {
		num_buf[0] = '-';
		s = num_buf;
		s_len++;
	} else if (always_sign) {
		num_buf[0] = '+';
		s = num_buf;
		s_len++;
	}
	s[s_len] = '\0';
	break;
case 0x67 /* 'g' */:
case 0x47 /* 'G' */:
	if (precision == 0)
		precision = 1;
	/*
	 * * We use &num_buf[ 1 ], so that we have room for the sign
	 */
	s = bsd_gcvt(number, precision, &num_buf[1]);
	is_negative = 0;
	if (*s == '-') {
		is_negative = 1;
		s = &num_buf[1];
	} else if (always_sign) {
		num_buf[0] = '+';
		s = num_buf;
	}
	s_len = strlen(s);
	if (fmt == 0x47 /* 'G' */ && (q = strchr(s, 'e')) != NULL) {
		*q = 'E';
	}
	break;
}

uni_s = zend_ascii_to_unicode(s, s_len + 1 ZEND_FILE_LINE_CC);

php_u_sprintf_appendstring(buffer, pos, size, uni_s, width, 0,
padding,
					 alignment, s_len, is_negative, 0, always_sign);
efree(uni_s);
+}
+/* }}} */
+/* php_sprintf_append2n() {{{ */
inline static void
php_sprintf_append2n(char **buffer, int *pos, int *size, long number,
int width, char padding, int alignment, int n,
@@ -302,8 +555,35 @@ php_sprintf_append2n(char *buffer, int
padding, alignment, (NUM_BUF_SIZE - 1) - i,
0, expprec, 0);
}
+/ }}} */

+/* php_u_sprintf_append2n() {{{ */
+inline static void
+php_u_sprintf_append2n(UChar **buffer, int *pos, int *size, long
number,
			 int width, UChar padding, int alignment, int n,
			 UChar *chartable, int expprec)
+{
UChar numbuf[NUM_BUF_SIZE];

register unsigned long num;

register unsigned int i = NUM_BUF_SIZE - 1;

register int andbits = (1 << n) - 1;

num = (unsigned long) number;

numbuf[i] = '\0';

do {
numbuf[--i] = chartable[(num & andbits)];
num >>= n;
}

while (num > 0);

php_u_sprintf_appendstring(buffer, pos, size, &numbuf[i], width, 0,
					 padding, alignment, (NUM_BUF_SIZE - 1) - i,
					 0, expprec, 0);
+}
+/* }}} */

+/* php_sprintf_getnumber() {{{ */
inline static long
php_sprintf_getnumber(char *buffer, int *pos)
{
@@ -318,8 +598,25 @@ php_sprintf_getnumber(char *buffer, int
pos += i;
return num;
}
+/ }}} */

-/* {{{ php_formatted_print
+/* php_u_sprintf_getnumber() {{{ */
+inline static long
+php_u_sprintf_getnumber(UChar *buffer, int *pos)
+{
UChar *endptr;

register long num = zend_u_strtol(&buffer[*pos], &endptr, 10);

register int i = 0;

if (endptr != NULL) {
i = (endptr - &buffer[*pos]);
}

*pos += i;

return num;
+}
+/* }}} */
+/* {{{ php_formatted_print()

New sprintf implementation for PHP.

Modifiers:
@@ -343,8 +640,7 @@ php_sprintf_getnumber(char *buffer, int

"X" integer argument is printed as uppercase hexadecimal

*/
-static char *
-php_formatted_print(int ht, int *len, int use_array, int
format_offset TSRMLS_DC)
+static char * php_formatted_print(int ht, int *len, int use_array,
int format_offset, int type TSRMLS_DC)
{
zval ***args, **z_format;
int argc, size = 240, inpos = 0, outpos = 0, temppos;
@@ -518,9 +814,26 @@ php_formatted_print(int ht, int *len, in
switch (format[inpos]) {
case 's': {
zval *var, var_copy;
			int use_copy;
			int use_copy = 0;
			zend_make_printable_zval(tmp, &var_copy, &use_copy);
			if (Z_TYPE_P(tmp) != IS_UNICODE) {
				zend_make_printable_zval(tmp, &var_copy, &use_copy);
			} else {
				var_copy = *tmp;
				zval_copy_ctor(&var_copy);
				INIT_PZVAL(&var_copy);
				use_copy = 1;
				switch (type) {
					case PHP_OUTPUT:
						convert_to_string_with_converter(&var_copy,  
ZEND_U_CONVERTER(UG(output_encoding_conv)));
						break;
					case PHP_RUNTIME:
					default:
						convert_to_string_with_converter(&var_copy,  
ZEND_U_CONVERTER(UG(runtime_encoding_conv)));
						break;
				}
			}
			if (use_copy) {
				var = &var_copy;
			} else {
@@ -559,7 +872,6 @@ php_formatted_print(int ht, int *len, in
case 'E':
case 'f':
case 'F':
			/* XXX not done */
			convert_to_double(tmp);
			php_sprintf_appenddouble(&result, &outpos, &size,
									 Z_DVAL_P(tmp),
@@ -630,17 +942,323 @@ php_formatted_print(int ht, int len, in
}
/ }}} */

+/* php_u_formatted_print() {{{ */
+static zstr php_u_formatted_print(int ht, int *len, int use_array,
int format_offset, int type TSRMLS_DC)
+{
zval ***args, **z_format;

int argc, size = 240, inpos = 0, outpos = 0, temppos;

int alignment, width, precision, currarg, adjusting, argnum;

UChar *format, *result, padding;

int always_sign;

zstr result_str;

argc = ZEND_NUM_ARGS();

/* verify the number of args */

if ((use_array && argc != (2 + format_offset))
	|| (!use_array && argc < (1 + format_offset))) {
WRONG_PARAM_COUNT_WITH_RETVAL(NULL_ZSTR);
}

args = (zval ***)safe_emalloc(argc, sizeof(zval *), 0);

if (zend_get_parameters_array_ex(argc, args) == FAILURE) {
efree(args);
WRONG_PARAM_COUNT_WITH_RETVAL(NULL_ZSTR);
}

if (use_array) {
int i = 1;
zval ***newargs;
zval **array;
z_format = args[format_offset];
array = args[1 + format_offset];
SEPARATE_ZVAL(array);
convert_to_array_ex(array);
argc = 1 + zend_hash_num_elements(Z_ARRVAL_PP(array));
newargs = (zval ***)safe_emalloc(argc, sizeof(zval *), 0);
newargs[0] = z_format;
for (zend_hash_internal_pointer_reset(Z_ARRVAL_PP(array));
	 zend_hash_get_current_data(Z_ARRVAL_PP(array), (void  
**)&newargs[i++]) == SUCCESS;
	 zend_hash_move_forward(Z_ARRVAL_PP(array)));
efree(args);
args = newargs;
format_offset = 0;
}

convert_to_unicode_ex(args[format_offset]);

format = Z_USTRVAL_PP(args[format_offset]);

result = eumalloc(size);

currarg = 1;

while (inpos<Z_USTRLEN_PP(args[format_offset])) {
int expprec = 0, multiuse = 0;
zval *tmp;
if (format[inpos] != 0x25 /* '%' */) {
	php_u_sprintf_appendchar(&result, &outpos, &size, format[inpos++]  
TSRMLS_CC);
} else if (format[inpos + 1] == 0x25 /* '%' */) {
	php_u_sprintf_appendchar(&result, &outpos, &size, 0x25 /* '%' */  
TSRMLS_CC);
	inpos += 2;
} else {
	/* starting a new format specifier, reset variables */
	alignment = ALIGN_RIGHT;
	adjusting = 0;
	padding = 0x20 /* ' ' */;
	always_sign = 0;
	inpos++;			/* skip the '%' */
	if ((format[inpos] < 0x7f) && !u_isalpha(format[inpos])) {
		/* first look for argnum */
		temppos = inpos;
		while (format[temppos] >= 0x30 /* '0' */ && format[temppos] <=  
0x39 /* '9' */) temppos++;
		if (format[temppos] == 0x24 /* '$' */) {
			argnum = php_u_sprintf_getnumber(format, &inpos);
			if (argnum == 0) {
				efree(result);
				efree(args);
				php_error_docref(NULL TSRMLS_CC, E_WARNING, "Zero is not a  
valid argument number");
				return NULL_ZSTR;
			}
			multiuse = 1;
			inpos++;  /* skip the '$' */
		} else {
			argnum = currarg++;
		}
		argnum += format_offset;
		/* after argnum comes modifiers */
		for (;; inpos++) {
			if (u_isspace(format[inpos]) || format[inpos] == 0x30 /* '0' */)  
{
				padding = format[inpos];
			} else if (format[inpos] == 0x2D /* '-' */) {
				alignment = ALIGN_LEFT;
				/* space padding, the default */
			} else if (format[inpos] == 0x2B /* '+' */) {
				always_sign = 1;
			} else if (format[inpos] == 0x27 /* '\'' */) {
				padding = format[++inpos];
			} else {
				break;
			}
		}
		/* after modifiers comes width */
		if (isdigit((int)format[inpos])) {
			width = php_u_sprintf_getnumber(format, &inpos);
			adjusting |= ADJ_WIDTH;
		} else {
			width = 0;
		}
		/* after width and argnum comes precision */
		if (format[inpos] == 0x2E /* '.' */) {
			inpos++;
			if (format[inpos] >= 0x30 /* '0' */ && format[inpos] <= 0x39 /*  
'9' */) {
				precision = php_u_sprintf_getnumber(format, &inpos);
				adjusting |= ADJ_PRECISION;
				expprec = 1;
			} else {
				precision = 0;
			}
		} else {
			precision = 0;
		}
	} else {
		width = precision = 0;
		argnum = currarg++ + format_offset;
	}
	if (argnum >= argc) {
		efree(result);
		efree(args);
		php_error_docref(NULL TSRMLS_CC, E_WARNING, "Too few arguments");
		return NULL_ZSTR;
	}
	if (format[inpos] == 0x6C /* 'l' */) {
		inpos++;
	}
	/* now we expect to find a type specifier */
	if (multiuse) {
		MAKE_STD_ZVAL(tmp);
		*tmp = **(args[argnum]);
		INIT_PZVAL(tmp);
		zval_copy_ctor(tmp);
	} else {
		SEPARATE_ZVAL(args[argnum]);
		tmp = *(args[argnum]);
	}
	switch (format[inpos]) {
		case 0x73 /* 's' */: {
			zval *var, var_copy;
			int use_copy;
			zend_make_unicode_zval(tmp, &var_copy, &use_copy);
			if (use_copy) {
				var = &var_copy;
			} else {
				var = tmp;
			}
			php_u_sprintf_appendstring(&result, &outpos, &size,
									 Z_USTRVAL_P(var),
									 width, precision, padding,
									 alignment,
									 Z_USTRLEN_P(var),
									 0, expprec, 0);
			if (use_copy) {
				zval_dtor(&var_copy);
			}
			break;
		}
		case 0x64 /* 'd' */:
			convert_to_long(tmp);
			php_u_sprintf_appendint(&result, &outpos, &size,
								  Z_LVAL_P(tmp),
								  width, padding, alignment,
								  always_sign);
			break;
		case 0x75 /* 'u' */:
			convert_to_long(tmp);
			php_u_sprintf_appenduint(&result, &outpos, &size,
								  Z_LVAL_P(tmp),
								  width, padding, alignment);
			break;
		case 0x67 /* 'g' */:
		case 0x47 /* 'G' */:
		case 0x65 /* 'e' */:
		case 0x45 /* 'E' */:
		case 0x66 /* 'f' */:
		case 0x46 /* 'F' */:
			convert_to_double(tmp);
			php_u_sprintf_appenddouble(&result, &outpos, &size,
									 Z_DVAL_P(tmp),
									 width, padding, alignment,
									 precision, adjusting,
									 format[inpos], always_sign
									 TSRMLS_CC);
			break;
		case 0x63 /* 'c' */:
			convert_to_long(tmp);
			php_u_sprintf_appendchar(&result, &outpos, &size,
								(char) Z_LVAL_P(tmp) TSRMLS_CC);
			break;
		case 0x6F /* 'o' */:
			convert_to_long(tmp);
			php_u_sprintf_append2n(&result, &outpos, &size,
								 Z_LVAL_P(tmp),
								 width, padding, alignment, 3,
								 u_hexchars, expprec);
			break;
		case 0x78 /* 'x' */:
			convert_to_long(tmp);
			php_u_sprintf_append2n(&result, &outpos, &size,
								 Z_LVAL_P(tmp),
								 width, padding, alignment, 4,
								 u_hexchars, expprec);
			break;
		case 0x58 /* 'X' */:
			convert_to_long(tmp);
			php_u_sprintf_append2n(&result, &outpos, &size,
								 Z_LVAL_P(tmp),
								 width, padding, alignment, 4,
								 u_HEXCHARS, expprec);
			break;
		case 0x62 /* 'b' */:
			convert_to_long(tmp);
			php_u_sprintf_append2n(&result, &outpos, &size,
								 Z_LVAL_P(tmp),
								 width, padding, alignment, 1,
								 u_hexchars, expprec);
			break;
		case 0x25 /* '%' */:
			php_u_sprintf_appendchar(&result, &outpos, &size, 0x25 /* '%' */  
TSRMLS_CC);
			break;
		default:
			break;
	}
	if (multiuse) {
		zval_ptr_dtor(&tmp);
	}
	inpos++;
}
}

efree(args);

/* possibly, we have to make sure we have room for the terminating
null? */

result[outpos] = 0;

*len = outpos;

result_str.u = result;

switch (type) {
case PHP_OUTPUT:
{
	UErrorCode status = U_ZERO_ERROR;
	char *s;
	int s_len;
zend_unicode_to_string_ex(ZEND_U_CONVERTER(UG(output_encoding_conv)),
&s, &s_len, result, outpos, &status);
	if(U_FAILURE(status)) {
		efree(s);
		efree(result);
		return NULL_ZSTR;
	}
	efree(result_str.v);
	result_str.s = s;
	*len = s_len;
	break;
}
case PHP_RUNTIME:
default:
	/* nothing to be done */
	break;
}

return result_str;
+}
+/* }}} */
/* {{{ proto string sprintf(string format [, mixed arg1 [, mixed
...]])
Return a formatted string */
PHP_FUNCTION(user_sprintf)
{
char *result;
int len;

if ((result=php_formatted_print(ht, &len, 0, 0 TSRMLS_CC))==NULL) {
RETURN_FALSE;
zstr result;

if (!UG(unicode)) {
if ((result.s = php_formatted_print(ht, &len, 0, 0, PHP_RUNTIME  
TSRMLS_CC))==NULL) {
	RETURN_FALSE;
}
RETVAL_STRINGL(result.s, len, 0);
} else {
result = php_u_formatted_print(ht, &len, 0, 0, PHP_RUNTIME  
TSRMLS_CC);
if (result.v == NULL) {
	RETURN_FALSE;
}
RETVAL_UNICODEL(result.u, len, 0);
}
RETVAL_STRINGL(result, len, 0);
}
/* }}} */

@@ -648,13 +1266,21 @@ PHP_FUNCTION(user_sprintf)
Return a formatted string */
PHP_FUNCTION(vsprintf)
{
char *result;
int len;

if ((result=php_formatted_print(ht, &len, 1, 0 TSRMLS_CC))==NULL) {
RETURN_FALSE;
zstr result;

if (!UG(unicode)) {
if ((result.s = php_formatted_print(ht, &len, 1, 0, PHP_RUNTIME  
TSRMLS_CC))==NULL) {
	RETURN_FALSE;
}
RETVAL_STRINGL(result.s, len, 0);
} else {
result = php_u_formatted_print(ht, &len, 1, 0, PHP_RUNTIME  
TSRMLS_CC);
if (result.v == NULL) {
	RETURN_FALSE;
}
RETVAL_UNICODEL(result.u, len, 0);
}
RETVAL_STRINGL(result, len, 0);
}
/* }}} */

@@ -662,14 +1288,22 @@ PHP_FUNCTION(vsprintf)
Output a formatted string */
PHP_FUNCTION(user_printf)
{
char *result;
int len;

if ((result=php_formatted_print(ht, &len, 0, 0 TSRMLS_CC))==NULL) {
RETURN_FALSE;
zstr result;

if (!UG(unicode)) {
if ((result.s = php_formatted_print(ht, &len, 0, 0, PHP_OUTPUT  
TSRMLS_CC))==NULL) {
	RETURN_FALSE;
}
} else {
result = php_u_formatted_print(ht, &len, 0, 0, PHP_OUTPUT  
TSRMLS_CC);
if (result.v == NULL) {
	RETURN_FALSE;
}
}
PHPWRITE(result, len);

efree(result);

PHPWRITE(result.s, len);

efree(result.v);
RETURN_LONG(len);
}
/* }}} */
@@ -678,14 +1312,22 @@ PHP_FUNCTION(user_printf)
Output a formatted string */
PHP_FUNCTION(vprintf)
{

char *result;
int len;

zstr result;
if ((result=php_formatted_print(ht, &len, 1, 0 TSRMLS_CC))==NULL) {
RETURN_FALSE;
if (!UG(unicode)) {
if ((result.s = php_formatted_print(ht, &len, 1, 0, PHP_OUTPUT  
TSRMLS_CC))==NULL) {
	RETURN_FALSE;
}
} else {
result = php_u_formatted_print(ht, &len, 1, 0, PHP_OUTPUT  
TSRMLS_CC);
if (result.v == NULL) {
	RETURN_FALSE;
}
}
PHPWRITE(result, len);

efree(result);

PHPWRITE(result.s, len);

efree(result.v);
RETURN_LONG(len);
}
/* }}} */
@@ -696,8 +1338,8 @@ PHP_FUNCTION(fprintf)
{
php_stream *stream;
zval **arg1;

char *result;

int len;

zstr result;

int len, ret;

if (ZEND_NUM_ARGS() < 2) {
WRONG_PARAM_COUNT;
@@ -709,15 +1351,22 @@ PHP_FUNCTION(fprintf)

php_stream_from_zval(stream, arg1);
if ((result=php_formatted_print(ht, &len, 0, 1 TSRMLS_CC))==NULL) {
RETURN_FALSE;
if (!UG(unicode)) {
if ((result.s = php_formatted_print(ht, &len, 0, 1, PHP_RUNTIME  
TSRMLS_CC))==NULL) {
	RETURN_FALSE;
}
ret = php_stream_write(stream, result.s, len);
} else {
result = php_u_formatted_print(ht, &len, 0, 1, PHP_RUNTIME  
TSRMLS_CC);
if (result.v == NULL) {
	RETURN_FALSE;
}
ret = php_stream_write_unicode(stream, result.u, len);
}
efree(result.v);
php_stream_write(stream, result, len);

efree(result);

RETURN_LONG(len);

RETURN_LONG(ret);
}
/* }}} */

@@ -727,8 +1376,8 @@ PHP_FUNCTION(vfprintf)
{
php_stream *stream;
zval **arg1;

char *result;

int len;

zstr result;

int len, ret;

if (ZEND_NUM_ARGS() != 3) {
WRONG_PARAM_COUNT;
@@ -740,15 +1389,22 @@ PHP_FUNCTION(vfprintf)

php_stream_from_zval(stream, arg1);
if ((result=php_formatted_print(ht, &len, 1, 1 TSRMLS_CC))==NULL) {
RETURN_FALSE;
if (!UG(unicode)) {
if ((result.s = php_formatted_print(ht, &len, 1, 1, PHP_RUNTIME  
TSRMLS_CC))==NULL) {
	RETURN_FALSE;
}
ret = php_stream_write(stream, result.s, len);
} else {
result = php_u_formatted_print(ht, &len, 1, 1, PHP_RUNTIME  
TSRMLS_CC);
if (result.v == NULL) {
	RETURN_FALSE;
}
ret = php_stream_write_unicode(stream, result.u, len);
}
php_stream_write(stream, result, len);

efree(result);

efree(result.v);

RETURN_LONG(len);

RETURN_LONG(ret);
}
/* }}} */

Unicode support for *printf()

-- Wbr, Antony Dovgal Index: ext/standard/formatted_print.c

define PRINTF_DEBUG(arg) php_printf arg

--
Wbr, Antony Dovgal
Index: ext/standard/formatted_print.c