andrei Mon Jul 10 21:18:01 2006 UTC
Modified files:
/php-src/ext/unicode unicode_iterators.c
Log:
Implement following() for combining sequences.
http://cvs.php.net/viewvc.cgi/php-src/ext/unicode/unicode_iterators.c?r1=1.35&r2=1.36&diff_format=u
Index: php-src/ext/unicode/unicode_iterators.c
diff -u php-src/ext/unicode/unicode_iterators.c:1.35
php-src/ext/unicode/unicode_iterators.c:1.36
--- php-src/ext/unicode/unicode_iterators.c:1.35 Mon Jul 10 20:14:12 2006
+++ php-src/ext/unicode/unicode_iterators.c Mon Jul 10 21:18:01 2006
@@ -14,7 +14,7 @@
+----------------------------------------------------------------------+
*/
-/* $Id: unicode_iterators.c,v 1.35 2006/07/10 20:14:12 andrei Exp $ */
+/* $Id: unicode_iterators.c,v 1.36 2006/07/10 21:18:01 andrei Exp $ */
/*
* TODO
@@ -424,12 +424,12 @@
if (flags & ITER_REVERSE) {
text_iter_helper_move(0, object->text, object->text_len,
&object->u.cs.start,
&object->u.cs.start_cp_offset);
- object->u.cs.end = object->u.cs.start;
} else {
text_iter_helper_move(1, object->text, object->text_len,
&object->u.cs.start,
&object->u.cs.start_cp_offset);
- object->u.cs.end = object->u.cs.start;
}
+ object->u.cs.end = object->u.cs.start;
+ object->u.cs.end_cp_offset = object->u.cs.start_cp_offset;
object->u.cs.index++;
}
@@ -446,6 +446,64 @@
object->u.cs.index = 0; /* because _next increments index */
}
+static void text_iter_cs_following(text_iter_obj *object, int32_t offset, long
flags TSRMLS_DC)
+{
+ int32_t k, tmp;
+
+ if (offset < 0) {
+ offset = 0;
+ }
+
+ /*
+ * On invalid iterator we always want to start looking for the code unit
+ * offset from the beginning of the string.
+ */
+ if (object->u.cs.start_cp_offset == UBRK_DONE) {
+ object->u.cs.start_cp_offset = 0;
+ object->u.cs.start = 0;
+ }
+
+ /*
+ * Try to locate the code unit position relative to the last known
codepoint
+ * offset.
+ */
+ k = object->u.cs.start;
+ if (offset > object->u.cs.start_cp_offset) {
+ U16_FWD_N(object->text, k, object->text_len, offset -
object->u.cs.start_cp_offset);
+ } else {
+ U16_BACK_N(object->text, 0, k, object->u.cs.start_cp_offset -
offset);
+ }
+
+ /*
+ * Locate the actual boundary.
+ */
+ if (flags & ITER_REVERSE) {
+ /*
+ * If offset was at or beyond the length of text, we need to
adjust it
+ * to the number of codepoints in the text.
+ */
+ if (k == object->text_len) {
+ offset = u_countChar32(object->text, object->text_len);
+ }
+ text_iter_helper_move(0, object->text, object->text_len, &k,
&offset);
+ } else {
+ text_iter_helper_move(1, object->text, object->text_len, &k,
&offset);
+ }
+
+ if (k == object->u.cs.start) {
+ return;
+ }
+
+ object->u.cs.start = k;
+ object->u.cs.start_cp_offset = offset;
+ object->u.cs.end = object->u.cs.start;
+}
+
+static zend_bool text_iter_cs_isBoundary(text_iter_obj *object, int32_t
offset, long flags TSRMLS_DC)
+{
+ return 1;
+}
+
static text_iter_ops text_iter_cs_ops = {
text_iter_cs_valid,
text_iter_cs_current,
@@ -453,6 +511,8 @@
text_iter_cs_offset,
text_iter_cs_next,
text_iter_cs_rewind,
+ text_iter_cs_following,
+ text_iter_cs_isBoundary,
};
@@ -598,7 +658,6 @@
} else {
object->u.brk.bound = ubrk_following(object->u.brk.iter, k);
}
- object->u.brk.next = object->u.brk.bound;
/*
* If boundary is the same one as where we were at before, simply
return.
@@ -607,6 +666,8 @@
return;
}
+ object->u.brk.next = object->u.brk.bound;
+
/*
* Adjust the internal codepoint offset based on how far we've moved.
*/
--
PHP CVS Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php
