diff options
author | Brion Vibber <brion@pobox.com> | 2010-03-01 16:59:40 -0800 |
---|---|---|
committer | Brion Vibber <brion@pobox.com> | 2010-03-01 16:59:40 -0800 |
commit | 3c0729786de07b25cba4c62d71079d6b96f71994 (patch) | |
tree | 5e27620994df88cc8ac6fe15c1f71cf603425253 | |
parent | e30af99873ceda83bb4e9a2e221da894e8363489 (diff) | |
parent | 0ca7aa68d3e6f86d9b5388fd33baee7b2719f5d9 (diff) |
Merge branch 'testing' of git@gitorious.org:statusnet/mainline into testing
-rw-r--r-- | lib/util.php | 24 |
1 files changed, 22 insertions, 2 deletions
diff --git a/lib/util.php b/lib/util.php index d12a7920d..7a170a5f5 100644 --- a/lib/util.php +++ b/lib/util.php @@ -809,8 +809,28 @@ function common_shorten_links($text) function common_xml_safe_str($str) { - // Neutralize control codes and surrogates - return preg_replace('/[\p{Cc}\p{Cs}]/u', '*', $str); + // Replace common eol and extra whitespace input chars + $unWelcome = array( + "\t", // tab + "\n", // newline + "\r", // cr + "\0", // null byte eos + "\x0B" // vertical tab + ); + + $replacement = array( + ' ', // single space + ' ', + '', // nothing + '', + ' ' + ); + + $str = str_replace($unWelcome, $replacement, $str); + + // Neutralize any additional control codes and UTF-16 surrogates + // (Twitter uses '*') + return preg_replace('/[\p{Cc}\p{Cs}]/u', '*', $str); } function common_tag_link($tag) |