Changeset 16553

Timestamp:

2008-07-25T16:37:50+12:00 (16 years ago)

Author:

ak19

Message:

Added method check_is_utf8 that will return 1/true if the given string is utf8 and 0/false if it isn't. Does not modify the string parameter. Based on ensure_utf8.

File:

• gsdl/trunk/perllib/unicode.pm (modified) (1 diff)

gsdl/trunk/perllib/unicode.pm

@@ -545 +545 @@
 
 
+# Returns true (1) if the given string is utf8 and false (0) if it isn't.
+# Does not modify the string parameter.
+sub check_is_utf8 {
+    my $value=shift;
+
+    if (!defined($value)) {
+    return 0; # not utf8 because it is undefined
+    }
+
+    $value =~ m/^/g; # to set \G
+    while ($value =~ m!\G.*?([\x80-\xff]+)!sg) {
+    my $highbytes=$1;
+    # make sure this block of high bytes is utf-8
+    $highbytes =~ /^/g; # set pos()
+    while ($highbytes =~
+        m!\G (?: [\xc0-\xdf][\x80-\xbf]   | # 2 byte utf-8
+            [\xe0-\xef][\x80-\xbf]{2} | # 3 byte
+            [\xf0-\xf7][\x80-\xbf]{3} | # 4 byte
+            [\xf8-\xfb][\x80-\xbf]{4} | # 5 byte
+            [\xfc-\xfd][\x80-\xbf]{5} | # 6 byte
+            )*([\x80-\xff])? !xg
+        ) {
+        my $badbyte=$1;
+        if (defined $badbyte) { # not end of string
+        return 0; # non-utf8 found
+        } 
+    }
+    }
+    
+    return 1;
+}
+
 sub substr
 {