Allow empty needle in grapheme_str*pos, grapheme_str*str

For consistency with str* and mb_str* functions.

Closes GH-6245.
Closes php/php-tasks#20.

diff --git a/ext/intl/grapheme/grapheme_string.c b/ext/intl/grapheme/grapheme_string.c
index e45330e15085ee33eb5257dcd08bb410ac0f6700..7d8df14d2e8074ce66b561025fed78f4bd2e2744 100644 (file)
--- a/ext/intl/grapheme/grapheme_string.c
+++ b/ext/intl/grapheme/grapheme_string.c
@@ -124,11 +124,6 @@ PHP_FUNCTION(grapheme_strpos)
 
        /* the offset is 'grapheme count offset' so it still might be invalid - we'll check it later */
 
-       if (needle_len == 0) {
-               zend_argument_value_error(2, "cannot be empty");
-               RETURN_THROWS();
-       }
-
        if (offset >= 0 && grapheme_ascii_check((unsigned char *)haystack, haystack_len) >= 0) {
                /* quick check to see if the string might be there
                 * I realize that 'offset' is 'grapheme count offset' but will work in spite of that
@@ -178,11 +173,6 @@ PHP_FUNCTION(grapheme_stripos)
 
        /* the offset is 'grapheme count offset' so it still might be invalid - we'll check it later */
 
-       if (needle_len == 0) {
-               zend_argument_value_error(2, "cannot be empty");
-               RETURN_THROWS();
-       }
-
        is_ascii = ( grapheme_ascii_check((unsigned char*)haystack, haystack_len) >= 0 );
 
        if ( is_ascii ) {
@@ -244,11 +234,6 @@ PHP_FUNCTION(grapheme_strrpos)
 
        /* the offset is 'grapheme count offset' so it still might be invalid - we'll check it later */
 
-       if (needle_len == 0) {
-               zend_argument_value_error(2, "cannot be empty");
-               RETURN_THROWS();
-       }
-
        is_ascii = grapheme_ascii_check((unsigned char *)haystack, haystack_len) >= 0;
 
        if ( is_ascii ) {
@@ -304,11 +289,6 @@ PHP_FUNCTION(grapheme_strripos)
 
        /* the offset is 'grapheme count offset' so it still might be invalid - we'll check it later */
 
-       if (needle_len == 0) {
-               zend_argument_value_error(2, "cannot be empty");
-               RETURN_THROWS();
-       }
-
        is_ascii = grapheme_ascii_check((unsigned char *)haystack, haystack_len) >= 0;
 
        if ( is_ascii ) {
@@ -572,11 +552,6 @@ static void strstr_common_handler(INTERNAL_FUNCTION_PARAMETERS, int f_ignore_cas
                RETURN_THROWS();
        }
 
-       if (needle_len == 0) {
-               zend_argument_value_error(2, "cannot be empty");
-               RETURN_THROWS();
-       }
-
        if ( !f_ignore_case ) {
 
                /* ASCII optimization: quick check to see if the string might be there */

diff --git a/ext/intl/grapheme/grapheme_util.c b/ext/intl/grapheme/grapheme_util.c
index 8633fddf5bc102ad4df751278c9ebd23c957ce33..2625a93263e87152601387b31b01c11f23e81346 100644 (file)
--- a/ext/intl/grapheme/grapheme_util.c
+++ b/ext/intl/grapheme/grapheme_util.c
@@ -126,6 +126,17 @@ int32_t grapheme_strpos_utf16(char *haystack, size_t haystack_len, char *needle,
        ubrk_setText(bi, uhaystack, uhaystack_len, &status);
        STRPOS_CHECK_STATUS(status, "Failed to set up iterator");
 
+       if (uneedle_len == 0) {
+               offset_pos = grapheme_get_haystack_offset(bi, last && offset >= 0 ? uhaystack_len : offset);
+               if (offset_pos == -1) {
+                       zend_argument_value_error(3, "must be contained in argument #1 ($haystack)");
+                       ret_pos = -1;
+                       goto finish;
+               }
+               ret_pos = offset_pos;
+               goto finish;
+       }
+
        status = U_ZERO_ERROR;
        src = usearch_open(uneedle, uneedle_len, uhaystack, uhaystack_len, "", bi, &status);
        STRPOS_CHECK_STATUS(status, "Error creating search object");

diff --git a/ext/intl/tests/grapheme_empty.phpt b/ext/intl/tests/grapheme_empty.phpt
index 31b3d8cc4393f6045df46f7487ccda986d824c81..873abf624e2f467e37868b97312732c927eab9a1 100644 (file)
--- a/ext/intl/tests/grapheme_empty.phpt
+++ b/ext/intl/tests/grapheme_empty.phpt
@@ -7,82 +7,54 @@ Test grapheme_strpos-alike functions with empty needle
 
 ini_set("intl.error_level", E_WARNING);
 
-try {
-    var_dump(grapheme_strpos("abc", "", -1));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_strpos("abc", ""));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_strpos("abc", "", -1));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_stripos("abc", ""));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_stripos("abc", "", -1));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_strrpos("abc", ""));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_strrpos("abc", "", -1));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_strripos("abc", ""));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_strripos("abc", "", 1));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_strstr("abc", ""));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
-
-try {
-    var_dump(grapheme_stristr("abc", ""));
-} catch (ValueError $exception) {
-    echo $exception->getMessage() . "\n";
-}
+var_dump(grapheme_strpos("abc", ""));
+var_dump(grapheme_strpos("abc", "", -1));
+var_dump(grapheme_stripos("abc", ""));
+var_dump(grapheme_stripos("abc", "", -1));
+var_dump(grapheme_strrpos("abc", ""));
+var_dump(grapheme_strrpos("abc", "", -1));
+var_dump(grapheme_strripos("abc", ""));
+var_dump(grapheme_strripos("abc", "", 1));
+var_dump(grapheme_strstr("abc", ""));
+var_dump(grapheme_strstr("abc", "", true));
+var_dump(grapheme_stristr("abc", ""));
+var_dump(grapheme_stristr("abc", "", true));
+var_dump(grapheme_strpos("äbc", ""));
+var_dump(grapheme_strpos("äbc", "", -1));
+var_dump(grapheme_stripos("äbc", ""));
+var_dump(grapheme_stripos("äbc", "", -1));
+var_dump(grapheme_strrpos("äbc", ""));
+var_dump(grapheme_strrpos("äbc", "", -1));
+var_dump(grapheme_strripos("äbc", ""));
+var_dump(grapheme_strripos("äbc", "", 1));
+var_dump(grapheme_strstr("äbc", ""));
+var_dump(grapheme_strstr("äbc", "", true));
+var_dump(grapheme_stristr("äbc", ""));
+var_dump(grapheme_stristr("äbc", "", true));
 
 ?>
 --EXPECT--
-grapheme_strpos(): Argument #2 ($needle) cannot be empty
-grapheme_strpos(): Argument #2 ($needle) cannot be empty
-grapheme_strpos(): Argument #2 ($needle) cannot be empty
-grapheme_stripos(): Argument #2 ($needle) cannot be empty
-grapheme_stripos(): Argument #2 ($needle) cannot be empty
-grapheme_strrpos(): Argument #2 ($needle) cannot be empty
-grapheme_strrpos(): Argument #2 ($needle) cannot be empty
-grapheme_strripos(): Argument #2 ($needle) cannot be empty
-grapheme_strripos(): Argument #2 ($needle) cannot be empty
-grapheme_strstr(): Argument #2 ($needle) cannot be empty
-grapheme_stristr(): Argument #2 ($needle) cannot be empty
+int(0)
+int(2)
+int(0)
+int(2)
+int(3)
+int(2)
+int(3)
+int(3)
+string(3) "abc"
+string(0) ""
+string(3) "abc"
+string(0) ""
+int(0)
+int(2)
+int(0)
+int(2)
+int(3)
+int(2)
+int(3)
+int(3)
+string(4) "äbc"
+string(0) ""
+string(4) "äbc"
+string(0) ""