mirrors/php-src
0
0
Fork 0
mirror of https://github.com/php/php-src.git synced 2024-11-29 21:04:10 +08:00
Code Issues Packages Projects Releases Wiki Activity

- Fix strrpos() logic (which was broken even in PHP 5).

Browse Source 
- Optimizations for a couple of functions.
This commit is contained in:
Andrei Zmievski 2006-10-02 19:18:14 +00:00
parent 640c76969e
commit 0b34032e49
2 changed files with 91 additions and 104 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

193
ext/standard/string.c
View File

@ -2372,14 +2372,8 @@ PHP_FUNCTION(stristr)
php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)"); php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)");
RETURN_FALSE; RETURN_FALSE;
} }
if (U_IS_BMP(Z_LVAL_PP(needle))) { needle_len = zend_codepoint_to_uchar((UChar32)Z_LVAL_PP(needle), u_needle_char);
u_needle_char[needle_len++] = (UChar)Z_LVAL_PP(needle); u_needle_char[needle_len] = 0;
u_needle_char[needle_len] = 0;
} else {
u_needle_char[needle_len++] = (UChar)U16_LEAD(Z_LVAL_PP(needle));
u_needle_char[needle_len++] = (UChar)U16_TRAIL(Z_LVAL_PP(needle));
u_needle_char[needle_len] = 0;
}
target.u = u_needle_char; target.u = u_needle_char;
} else { } else {
needle_char[needle_len++] = (char)Z_LVAL_PP(needle); needle_char[needle_len++] = (char)Z_LVAL_PP(needle);
@ -2426,7 +2420,7 @@ PHP_FUNCTION(stristr)
Finds first occurrence of a string within another */ Finds first occurrence of a string within another */
PHP_FUNCTION(strstr) PHP_FUNCTION(strstr)
{ {
void *haystack; zstr haystack;
int haystack_len; int haystack_len;
zend_uchar haystack_type; zend_uchar haystack_type;
zval **needle; zval **needle;
@ -2450,16 +2444,16 @@ PHP_FUNCTION(strstr)
/* haystack type determines the needle type */ /* haystack type determines the needle type */
if (haystack_type == IS_UNICODE) { if (haystack_type == IS_UNICODE) {
convert_to_unicode_ex(needle); convert_to_unicode_ex(needle);
found = zend_u_memnstr((UChar*)haystack, found = zend_u_memnstr(haystack.u,
Z_USTRVAL_PP(needle), Z_USTRVAL_PP(needle),
Z_USTRLEN_PP(needle), Z_USTRLEN_PP(needle),
(UChar*)haystack + haystack_len); haystack.u + haystack_len);
} else { } else {
convert_to_string_ex(needle); convert_to_string_ex(needle);
found = php_memnstr((char*)haystack, found = php_memnstr(haystack.s,
Z_STRVAL_PP(needle), Z_STRVAL_PP(needle),
Z_STRLEN_PP(needle), Z_STRLEN_PP(needle),
(char*)haystack + haystack_len); haystack.s + haystack_len);
} }
} else { } else {
convert_to_long_ex(needle); convert_to_long_ex(needle);
@ -2468,39 +2462,33 @@ PHP_FUNCTION(strstr)
php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)"); php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)");
RETURN_FALSE; RETURN_FALSE;
} }
/* supplementary codepoint values may require 2 UChar's */
if (U_IS_BMP(Z_LVAL_PP(needle))) {
u_needle_char[n_len++] = (UChar) Z_LVAL_PP(needle);
u_needle_char[n_len] = 0;
} else {
u_needle_char[n_len++] = (UChar) U16_LEAD(Z_LVAL_PP(needle));
u_needle_char[n_len++] = (UChar) U16_TRAIL(Z_LVAL_PP(needle));
u_needle_char[n_len] = 0;
}
found = zend_u_memnstr((UChar*)haystack, n_len = zend_codepoint_to_uchar((UChar32)Z_LVAL_PP(needle), u_needle_char);
u_needle_char[n_len] = 0;
found = zend_u_memnstr(haystack.u,
u_needle_char, u_needle_char,
n_len, n_len,
(UChar*)haystack + haystack_len); haystack.u + haystack_len);
} else { } else {
needle_char[0] = (char) Z_LVAL_PP(needle); needle_char[0] = (char) Z_LVAL_PP(needle);
needle_char[1] = 0; needle_char[1] = 0;
found = php_memnstr((char*)haystack, found = php_memnstr(haystack.s,
needle_char, needle_char,
1, 1,
(char*)haystack + haystack_len); haystack.s + haystack_len);
} }
} }
if (found) { if (found) {
switch (haystack_type) { switch (haystack_type) {
case IS_UNICODE: case IS_UNICODE:
found_offset = (UChar*)found - (UChar*)haystack; found_offset = (UChar*)found - haystack.u;
if (part) { if (part) {
UChar *ret; UChar *ret;
ret = eumalloc(found_offset + 1); ret = eumalloc(found_offset + 1);
u_strncpy(ret, haystack, found_offset); u_strncpy(ret, haystack.u, found_offset);
ret[found_offset] = '\0'; ret[found_offset] = '\0';
RETURN_UNICODEL(ret , found_offset, 0); RETURN_UNICODEL(ret , found_offset, 0);
} else { } else {
@ -2509,11 +2497,11 @@ PHP_FUNCTION(strstr)
break; break;
case IS_STRING: case IS_STRING:
found_offset = (char *)found - (char *)haystack; found_offset = (char *)found - haystack.s;
if (part) { if (part) {
char *ret; char *ret;
ret = emalloc(found_offset + 1); ret = emalloc(found_offset + 1);
strncpy(ret, haystack, found_offset); strncpy(ret, haystack.s, found_offset);
ret[found_offset] = '\0'; ret[found_offset] = '\0';
RETURN_STRINGL(ret , found_offset, 0); RETURN_STRINGL(ret , found_offset, 0);
} else { } else {
@ -2535,7 +2523,7 @@ PHP_FUNCTION(strstr)
Finds position of first occurrence of a string within another */ Finds position of first occurrence of a string within another */
PHP_FUNCTION(strpos) PHP_FUNCTION(strpos)
{ {
void *haystack; zstr haystack;
int haystack_len; int haystack_len;
zend_uchar haystack_type; zend_uchar haystack_type;
zval **needle; zval **needle;
@ -2544,6 +2532,7 @@ PHP_FUNCTION(strpos)
char needle_char[2]; char needle_char[2];
UChar u_needle_char[3]; UChar u_needle_char[3];
int n_len = 0; int n_len = 0;
int32_t cu_offset = 0;
if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "tZ|l", &haystack, if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "tZ|l", &haystack,
&haystack_len, &haystack_type, &needle, &offset) == FAILURE) { &haystack_len, &haystack_type, &needle, &offset) == FAILURE) {
@ -2551,7 +2540,7 @@ PHP_FUNCTION(strpos)
} }
/* /*
* Unicode note: it's okay to not convert offset to codepoint offset here. * Unicode note: it's okay to not convert offset to code unit offset here.
* We'll just do a rough check that the offset does not exceed length in * We'll just do a rough check that the offset does not exceed length in
* code units, and leave the rest to zend_u_memnstr(). * code units, and leave the rest to zend_u_memnstr().
*/ */
@ -2568,25 +2557,23 @@ PHP_FUNCTION(strpos)
/* haystack type determines the needle type */ /* haystack type determines the needle type */
if (haystack_type == IS_UNICODE) { if (haystack_type == IS_UNICODE) {
int32_t cp_offset = 0;
convert_to_unicode_ex(needle); convert_to_unicode_ex(needle);
/* locate the codepoint at the specified offset */ /* calculate code unit offset */
U16_FWD_N((UChar*)haystack, cp_offset, haystack_len, offset); U16_FWD_N(haystack.u, cu_offset, haystack_len, offset);
found = zend_u_memnstr((UChar*)haystack + cp_offset, found = zend_u_memnstr(haystack.u + cu_offset,
Z_USTRVAL_PP(needle), Z_USTRVAL_PP(needle),
Z_USTRLEN_PP(needle), Z_USTRLEN_PP(needle),
(UChar*)haystack + haystack_len); haystack.u + haystack_len);
} else { } else {
convert_to_string_ex(needle); convert_to_string_ex(needle);
found = php_memnstr((char*)haystack + offset, found = php_memnstr(haystack.s + offset,
Z_STRVAL_PP(needle), Z_STRVAL_PP(needle),
Z_STRLEN_PP(needle), Z_STRLEN_PP(needle),
(char*)haystack + haystack_len); haystack.s + haystack_len);
} }
} else { } else {
convert_to_long_ex(needle); convert_to_long_ex(needle);
if (haystack_type == IS_UNICODE) { if (haystack_type == IS_UNICODE) {
int32_t cp_offset = 0;
if (Z_LVAL_PP(needle) < 0 || Z_LVAL_PP(needle) > 0x10FFFF) { if (Z_LVAL_PP(needle) < 0 || Z_LVAL_PP(needle) > 0x10FFFF) {
php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)"); php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)");
RETURN_FALSE; RETURN_FALSE;
@ -2594,30 +2581,33 @@ PHP_FUNCTION(strpos)
n_len += zend_codepoint_to_uchar(Z_LVAL_PP(needle), u_needle_char); n_len += zend_codepoint_to_uchar(Z_LVAL_PP(needle), u_needle_char);
u_needle_char[n_len] = 0; u_needle_char[n_len] = 0;
/* locate the codepoint at the specified offset */ /* calculate code unit offset */
U16_FWD_N((UChar*)haystack, cp_offset, haystack_len, offset); U16_FWD_N(haystack.u, cu_offset, haystack_len, offset);
found = zend_u_memnstr((UChar*)haystack + cp_offset, found = zend_u_memnstr(haystack.u + cu_offset,
u_needle_char, u_needle_char,
n_len, n_len,
(UChar*)haystack + haystack_len); haystack.u + haystack_len);
} else { } else {
needle_char[0] = (char) Z_LVAL_PP(needle); needle_char[0] = (char) Z_LVAL_PP(needle);
needle_char[1] = 0; needle_char[1] = 0;
found = php_memnstr((char*)haystack + offset, found = php_memnstr(haystack.s + offset,
needle_char, needle_char,
1, 1,
(char*)haystack + haystack_len); haystack.s + haystack_len);
} }
} }
if (found) { if (found) {
if (haystack_type == IS_UNICODE) { if (haystack_type == IS_UNICODE) {
/* simple subtraction will not suffice, since there may be /* Simple subtraction will not suffice, since there may be
supplementary codepoints */ supplementary codepoints. We count how many codepoints there are
RETURN_LONG(u_countChar32(haystack, ((char *)found - (char *)haystack)/sizeof(UChar))); between the starting offset and the found location and add them
to the starting codepoint offset. */
RETURN_LONG(offset + u_countChar32(haystack.u + cu_offset,
(UChar*)found - (haystack.u + cu_offset)));
} else { } else {
RETURN_LONG((char *)found - (char *)haystack); RETURN_LONG((char *)found - haystack.s);
} }
} else { } else {
RETURN_FALSE; RETURN_FALSE;
@ -2636,7 +2626,7 @@ PHP_FUNCTION(stripos)
void *haystack_dup = NULL, *needle_dup = NULL; void *haystack_dup = NULL, *needle_dup = NULL;
char needle_char[2]; char needle_char[2];
char c = 0; char c = 0;
UChar u_needle_char[8]; UChar u_needle_char[3];
void *found = NULL; void *found = NULL;
int cu_offset = 0; int cu_offset = 0;
@ -2647,6 +2637,12 @@ PHP_FUNCTION(stripos)
if (Z_TYPE_PP(haystack) != IS_UNICODE && Z_TYPE_PP(haystack) != IS_STRING) { if (Z_TYPE_PP(haystack) != IS_UNICODE && Z_TYPE_PP(haystack) != IS_STRING) {
convert_to_text_ex(haystack); convert_to_text_ex(haystack);
} }
/*
* Unicode note: it's okay to not convert offset to code unit offset here.
* We'll just do a rough check that the offset does not exceed length in
* code units, and leave the rest to zend_u_memnstr().
*/
if (offset < 0 || offset > Z_UNILEN_PP(haystack)) { if (offset < 0 || offset > Z_UNILEN_PP(haystack)) {
php_error_docref(NULL TSRMLS_CC, E_WARNING, "Offset not contained in string."); php_error_docref(NULL TSRMLS_CC, E_WARNING, "Offset not contained in string.");
RETURN_FALSE; RETURN_FALSE;
@ -2670,7 +2666,7 @@ PHP_FUNCTION(stripos)
} }
needle_len = Z_UNILEN_PP(needle); needle_len = Z_UNILEN_PP(needle);
if (Z_TYPE_PP(haystack) == IS_UNICODE) { if (Z_TYPE_PP(haystack) == IS_UNICODE) {
/* calculate codeunit offset */ /* calculate code unit offset */
U16_FWD_N(Z_USTRVAL_PP(haystack), cu_offset, haystack_len, offset); U16_FWD_N(Z_USTRVAL_PP(haystack), cu_offset, haystack_len, offset);
found = php_u_stristr(Z_USTRVAL_PP(haystack) + cu_offset, Z_USTRVAL_PP(needle), haystack_len, needle_len TSRMLS_CC); found = php_u_stristr(Z_USTRVAL_PP(haystack) + cu_offset, Z_USTRVAL_PP(needle), haystack_len, needle_len TSRMLS_CC);
} else { } else {
@ -2683,44 +2679,20 @@ PHP_FUNCTION(stripos)
(char *)haystack_dup + haystack_len); (char *)haystack_dup + haystack_len);
} }
} else { } else {
switch (Z_TYPE_PP(needle)) { convert_to_long_ex(needle);
case IS_LONG:
case IS_BOOL:
if (Z_TYPE_PP(haystack) == IS_UNICODE) {
if (Z_LVAL_PP(needle) < 0 || Z_LVAL_PP(needle) > 0x10FFFF) {
php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)");
RETURN_FALSE;
}
needle_len = zend_codepoint_to_uchar((UChar32)Z_LVAL_PP(needle), u_needle_char);
} else {
c = tolower((char)Z_LVAL_PP(needle));
}
break;
case IS_DOUBLE:
if (Z_TYPE_PP(haystack) == IS_UNICODE) {
if ((UChar32)Z_DVAL_PP(needle) < 0 || (UChar32)Z_DVAL_PP(needle) > 0x10FFFF) {
php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)");
RETURN_FALSE;
}
needle_len = zend_codepoint_to_uchar((UChar32)Z_DVAL_PP(needle), u_needle_char);
} else {
c = tolower((char)Z_DVAL_PP(needle));
}
break;
default:
php_error_docref(NULL TSRMLS_CC, E_WARNING, "needle is not a string or an integer.");
RETURN_FALSE;
break;
}
if (Z_TYPE_PP(haystack) == IS_UNICODE) { if (Z_TYPE_PP(haystack) == IS_UNICODE) {
/* calculate codeunit offset */ if (Z_LVAL_PP(needle) < 0 || Z_LVAL_PP(needle) > 0x10FFFF) {
U16_FWD_N(Z_USTRVAL_PP(haystack), cu_offset, haystack_len, offset); php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)");
RETURN_FALSE;
}
needle_len = zend_codepoint_to_uchar((UChar32)Z_LVAL_PP(needle), u_needle_char);
u_needle_char[needle_len] = 0; u_needle_char[needle_len] = 0;
/* calculate code unit offset */
U16_FWD_N(Z_USTRVAL_PP(haystack), cu_offset, haystack_len, offset);
found = php_u_stristr(Z_USTRVAL_PP(haystack) + cu_offset, found = php_u_stristr(Z_USTRVAL_PP(haystack) + cu_offset,
u_needle_char, haystack_len, needle_len TSRMLS_CC); u_needle_char, haystack_len, needle_len TSRMLS_CC);
} else { } else {
c = tolower((char)Z_LVAL_PP(needle));
needle_char[0] = c; needle_char[0] = c;
needle_char[1] = '\0'; needle_char[1] = '\0';
haystack_dup = estrndup(Z_STRVAL_PP(haystack), haystack_len); haystack_dup = estrndup(Z_STRVAL_PP(haystack), haystack_len);
@ -2767,6 +2739,7 @@ PHP_FUNCTION(strrpos)
long offset = 0; long offset = 0;
char *p, *e, ord_needle[2]; char *p, *e, ord_needle[2];
UChar *pos, *u_p, *u_e, u_ord_needle[3]; UChar *pos, *u_p, *u_e, u_ord_needle[3];
int cu_offset = 0;
if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ZZ|l", if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ZZ|l",
&zhaystack, &zneedle, &offset) == FAILURE) { &zhaystack, &zneedle, &offset) == FAILURE) {
@ -2790,14 +2763,8 @@ PHP_FUNCTION(strrpos)
php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)"); php_error(E_WARNING, "Needle argument codepoint value out of range (0 - 0x10FFFF)");
RETURN_FALSE; RETURN_FALSE;
} }
if (U_IS_BMP(Z_LVAL_PP(zneedle))) { needle_len = zend_codepoint_to_uchar((UChar32)Z_LVAL_PP(zneedle), u_ord_needle);
u_ord_needle[needle_len++] = (UChar)Z_LVAL_PP(zneedle); u_ord_needle[needle_len] = 0;
u_ord_needle[needle_len] = 0;
} else {
u_ord_needle[needle_len++] = (UChar)U16_LEAD(Z_LVAL_PP(zneedle));
u_ord_needle[needle_len++] = (UChar)U16_TRAIL(Z_LVAL_PP(zneedle));
u_ord_needle[needle_len] = 0;
}
needle.u = u_ord_needle; needle.u = u_ord_needle;
} else { } else {
convert_to_long_ex(zneedle); convert_to_long_ex(zneedle);
@ -2810,40 +2777,60 @@ PHP_FUNCTION(strrpos)
haystack = Z_UNIVAL_PP(zhaystack); haystack = Z_UNIVAL_PP(zhaystack);
haystack_len = Z_UNILEN_PP(zhaystack); haystack_len = Z_UNILEN_PP(zhaystack);
if ((haystack_len == 0) || (needle_len == 0)) { if ((haystack_len == 0) || (needle_len == 0) || needle_len > haystack_len) {
RETURN_FALSE; RETURN_FALSE;
} }
if (Z_TYPE_PP(zhaystack) == IS_UNICODE) { if (Z_TYPE_PP(zhaystack) == IS_UNICODE) {
if (offset >= 0) { if (offset >= 0) {
u_p = haystack.u + offset; U16_FWD_N(haystack.u, cu_offset, haystack_len, offset);
if (cu_offset > haystack_len - needle_len) {
RETURN_FALSE;
}
u_p = haystack.u + cu_offset;
u_e = haystack.u + haystack_len - needle_len; u_e = haystack.u + haystack_len - needle_len;
} else { } else {
u_p = haystack.u; u_p = haystack.u;
if (-offset > haystack_len) { if (-offset > haystack_len) {
u_e = haystack.u - needle_len; RETURN_FALSE;
} else if (needle_len > -offset) {
u_e = haystack.u + haystack_len - needle_len;
} else { } else {
u_e = haystack.u + haystack_len + offset; cu_offset = haystack_len;
U16_BACK_N(haystack.u, 0, cu_offset, -offset);
if (cu_offset == 0) {
RETURN_FALSE;
}
if (needle_len > haystack_len - cu_offset) {
u_e = haystack.u + haystack_len - needle_len;
} else {
u_e = haystack.u + cu_offset;
}
} }
} }
pos = u_strFindLast(u_p, u_e-u_p+needle_len, needle.u, needle_len); pos = u_strFindLast(u_p, u_e-u_p+needle_len, needle.u, needle_len);
if (pos) { if (pos) {
RETURN_LONG(pos - haystack.u); if (offset > 0) {
RETURN_LONG(offset + u_countChar32(u_p, (UChar*)pos - u_p));
} else {
RETURN_LONG(u_countChar32(haystack.u, (UChar*)pos - haystack.u));
}
} else { } else {
RETURN_FALSE; RETURN_FALSE;
} }
} else { } else {
if (offset >= 0) { if (offset >= 0) {
if (offset > haystack_len) {
RETURN_FALSE;
}
p = haystack.s + offset; p = haystack.s + offset;
e = haystack.s + haystack_len - needle_len; e = haystack.s + haystack_len - needle_len;
} else { } else {
p = haystack.s;
if (-offset > haystack_len) { if (-offset > haystack_len) {
e = haystack.s - needle_len; RETURN_FALSE;
} else if (needle_len > -offset) { }
p = haystack.s;
if (needle_len > -offset) {
e = haystack.s + haystack_len - needle_len; e = haystack.s + haystack_len - needle_len;
} else { } else {
e = haystack.s + haystack_len + offset; e = haystack.s + haystack_len + offset;

2
unicode-progress.txt
View File

@ -26,7 +26,6 @@ ext/standard
sscanf() sscanf()
Params API. Rest - no idea yet. Params API. Rest - no idea yet.
stristr()
strripos() strripos()
str_replace() str_replace()
stri_replace() stri_replace()
@ -158,6 +157,7 @@ ext/standard
stripcslashes() stripcslashes()
stripslashes() stripslashes()
stripos() stripos()
stristr()
strpbrk() strpbrk()
strpos() strpos()
strrchr() strrchr()