X-Git-Url: https://vcs.fsf.org/?p=squirrelmail.git;a=blobdiff_plain;f=functions%2Fstrings.php;h=37e466fd6ca49ec1cfb8601a7d2ba046d5ddcbfd;hp=f62cf138585636db2ca5e79b86b1ca52e65acc15;hb=11e62dd0fef499029744a33795aa7ac53e51c3b0;hpb=01f013c1311bc8cc3972b4472611901238096775 diff --git a/functions/strings.php b/functions/strings.php index f62cf138..37e466fd 100644 --- a/functions/strings.php +++ b/functions/strings.php @@ -6,7 +6,7 @@ * This code provides various string manipulation functions that are * used by the rest of the SquirrelMail code. * - * @copyright © 1999-2007 The SquirrelMail Project Team + * @copyright 1999-2020 The SquirrelMail Project Team * @license http://opensource.org/licenses/gpl-license.php GNU Public License * @version $Id$ * @package squirrelmail @@ -349,12 +349,12 @@ function sqWordWrap(&$line, $wrap, $charset='') { } } - ereg("^([\t >]*)([^\t >].*)?$", $line, $regs); + preg_match('/^([\t >]*)([^\t >].*)?$/', $line, $regs); $beginning_spaces = $regs[1]; if (isset($regs[2])) { $words = explode(' ', $regs[2]); } else { - $words = ''; + $words = array(); } $i = 0; @@ -474,12 +474,7 @@ function get_location () { $is_secure_connection, $sq_ignore_http_x_forwarded_headers; /* Get the path, handle virtual directories */ - if(strpos(php_self(), '?')) { - $path = substr(php_self(), 0, strpos(php_self(), '?')); - } else { - $path = php_self(); - } - $path = substr($path, 0, strrpos($path, '/')); + $path = substr(php_self(FALSE), 0, strrpos(php_self(FALSE), '/')); // proto+host+port are already set in config: if ( !empty($config_location_base) ) { @@ -655,18 +650,23 @@ function OneTimePadCreate ($length=100) { } /** - * Returns a string showing the size of the message/attachment. - * - * @param int $bytes the filesize in bytes - * @return string the filesize in human readable format - * @since 1.0 - */ -function show_readable_size($bytes) { - $bytes /= 1024; + * Returns a string showing a byte size figure in + * a more easily digested (readable) format + * + * @param int $bytes the size in bytes + * @param int $filesize_divisor the divisor we'll use (OPTIONAL; default 1024) + * + * @return string The size in human readable format + * + * @since 1.0 + * + */ +function show_readable_size($bytes, $filesize_divisor=1024) { + $bytes /= $filesize_divisor; $type = _("KiB"); - if ($bytes / 1024 > 1) { - $bytes /= 1024; + if ($bytes / $filesize_divisor > 1) { + $bytes /= $filesize_divisor; $type = _("MiB"); } @@ -678,7 +678,8 @@ function show_readable_size($bytes) { settype($bytes, 'integer'); } - return $bytes . ' ' . $type; + global $nbsp; + return $bytes . $nbsp . $type; } /** @@ -726,7 +727,7 @@ function GenerateRandomString($size, $chars, $flags = 0) { * @since 1.0.3 */ function quoteimap($str) { - return preg_replace("/([\"\\\\])/", "\\\\$1", $str); + return str_replace(array('\\', '"'), array('\\\\', '\\"'), $str); } /** @@ -1094,7 +1095,16 @@ function sq_str_pad($string, $width, $pad, $padtype, $charset='') { * @link http://www.php.net/substr * @link http://www.php.net/mb_substr */ -function sq_substr($string,$start,$length,$charset='auto') { +function sq_substr($string,$start,$length=NULL,$charset='auto') { + + // if $length is NULL, use the full string length... + // we have to do this to mimick the use of substr() + // where $length is not given + // + if (is_null($length)) + $length = sq_strlen($length); + + // use automatic charset detection, if function call asks for it static $charset_auto, $bUse_mb; @@ -1130,6 +1140,125 @@ function sq_substr($string,$start,$length,$charset='auto') { return substr($string,$start,$length); } +/** + * This is a replacement for PHP's substr_replace() that is + * multibyte-aware. + * + * @param string $string The string to operate upon + * @param string $replacement The string to be inserted + * @param int $start The offset at which to begin substring replacement + * @param int $length The number of characters after $start to remove + * NOTE that if you need to specify a charset but + * want to achieve normal substr_replace() behavior + * where $length is not specified, use NULL (OPTIONAL; + * default from $start to end of string) + * @param string $charset The charset of the given string. A value of NULL + * here will force the use of PHP's standard substr(). + * (OPTIONAL; default is "auto", which indicates that + * the user's current charset should be used). + * + * @return string The manipulated string + * + * Of course, you can use more advanced (e.g., negative) values + * for $start and $length as needed - see the PHP manual for more + * information: http://www.php.net/manual/function.substr-replace.php + * + */ +function sq_substr_replace($string, $replacement, $start, $length=NULL, + $charset='auto') +{ + + // NULL charset? Just use substr_replace() + // + if (is_null($charset)) + return is_null($length) ? substr_replace($string, $replacement, $start) + : substr_replace($string, $replacement, $start, $length); + + + // use current character set? + // + if ($charset == 'auto') + { +//FIXME: is there any reason why this cannot be a global flag used by all string wrapper functions? + static $auto_charset; + if (!isset($auto_charset)) + { + global $default_charset; +//FIXME - do we need this? +global $squirrelmail_language; + set_my_charset(); + $auto_charset = $default_charset; +//FIXME - do we need this? +if ($squirrelmail_language == 'ja_JP') $auto_charset = 'euc-jp'; + } + $charset = $auto_charset; + } + + + // standardize character set name + // + $charset = strtolower($charset); + + +/* ===== FIXME: this list is not used in 1.5.x, but if we need it, unless this differs between all our string function wrappers, we should store this info in the session + // only use mbstring with the following character sets + // + $sq_substr_replace_mb_charsets = array( + 'utf-8', + 'big5', + 'gb2312', + 'gb18030', + 'euc-jp', + 'euc-cn', + 'euc-tw', + 'euc-kr' + ); + + + // now we can use our own implementation using + // mb_substr() and mb_strlen() if needed + // + if (in_array($charset, $sq_substr_replace_mb_charsets) + && in_array($charset, sq_mb_list_encodings())) +===== */ +//FIXME: is there any reason why this cannot be a global array used by all string wrapper functions? + if (in_array($charset, sq_mb_list_encodings())) + { + + $string_length = mb_strlen($string, $charset); + + if ($start < 0) + $start = max(0, $string_length + $start); + + else if ($start > $string_length) + $start = $string_length; + + if ($length < 0) + $length = max(0, $string_length - $start + $length); + + else if (is_null($length) || $length > $string_length) + $length = $string_length; + + if ($start + $length > $string_length) + $length = $string_length - $start; + + return mb_substr($string, 0, $start, $charset) + . $replacement + . mb_substr($string, + $start + $length, + $string_length, // FIXME: I can't see why this is needed: - $start - $length, + $charset); + + } + + + // else use normal substr_replace() + // + return is_null($length) ? substr_replace($string, $replacement, $start) + : substr_replace($string, $replacement, $start, $length); + +} + /** * Wrapper that is used to switch between vanilla and multibyte strpos * functions. @@ -1245,3 +1374,393 @@ function sq_count8bit($string) { function sq_trim_value ( &$value ) { $value = trim($value); } + +/** + * Truncates the given string so that it has at + * most $max_chars characters. NOTE that a "character" + * may be a multibyte character, or (optionally), an + * HTML entity , so this function is different than + * using substr() or mb_substr(). + * + * NOTE that if $elipses is given and used, the returned + * number of characters will be $max_chars PLUS the + * length of $elipses + * + * @param string $string The string to truncate + * @param int $max_chars The maximum allowable characters + * @param string $elipses A string that will be added to + * the end of the truncated string + * (ONLY if it is truncated) (OPTIONAL; + * default not used) + * @param boolean $html_entities_as_chars Whether or not to keep + * HTML entities together + * (OPTIONAL; default ignore + * HTML entities) + * + * @return string The truncated string + * + * @since 1.4.20 and 1.5.2 (replaced truncateWithEntities()) + * + */ +function sm_truncate_string($string, $max_chars, $elipses='', + $html_entities_as_chars=FALSE) +{ + + // if the length of the string is less than + // the allowable number of characters, just + // return it as is (even if it contains any + // HTML entities, that would just make the + // actual length even smaller) + // + $actual_strlen = sq_strlen($string, 'auto'); + if ($max_chars <= 0 || $actual_strlen <= $max_chars) + return $string; + + + // if needed, count the number of HTML entities in + // the string up to the maximum character limit, + // pushing that limit up for each entity found + // + $adjusted_max_chars = $max_chars; + if ($html_entities_as_chars) + { + + // $loop_count is needed to prevent an endless loop + // which is caused by buggy mbstring versions that + // return 0 (zero) instead of FALSE in some rare + // cases. Thanks, PHP. + // see: http://bugs.php.net/bug.php?id=52731 + // also: tracker $3053349 + // + $loop_count = 0; + $entity_pos = $entity_end_pos = -1; + while ($entity_end_pos + 1 < $actual_strlen + && ($entity_pos = sq_strpos($string, '&', $entity_end_pos + 1)) !== FALSE + && ($entity_end_pos = sq_strpos($string, ';', $entity_pos)) !== FALSE + && $entity_pos <= $adjusted_max_chars + && $loop_count++ < $max_chars) + { + $adjusted_max_chars += $entity_end_pos - $entity_pos; + } + + + // this isn't necessary because sq_substr() would figure this + // out anyway, but we can avoid a sq_substr() call and we + // know that we don't have to add an elipses (this is now + // an accurate comparison, since $adjusted_max_chars, like + // $actual_strlen, does not take into account HTML entities) + // + if ($actual_strlen <= $adjusted_max_chars) + return $string; + + } + + + // get the truncated string + // + $truncated_string = sq_substr($string, 0, $adjusted_max_chars); + + + // return with added elipses + // + return $truncated_string . $elipses; + +} + +/** + * Gathers the list of secuirty tokens currently + * stored in the user's preferences and optionally + * purges old ones from the list. + * + * @param boolean $purge_old Indicates if old tokens + * should be purged from the + * list ("old" is 2 days or + * older unless the administrator + * overrides that value using + * $max_token_age_days in + * config/config_local.php) + * (OPTIONAL; default is to always + * purge old tokens) + * + * @return array The list of tokens + * + * @since 1.4.19 and 1.5.2 + * + */ +function sm_get_user_security_tokens($purge_old=TRUE) +{ + + global $data_dir, $username, $max_token_age_days, + $use_expiring_security_tokens; + + $tokens = getPref($data_dir, $username, 'security_tokens', ''); + if (($tokens = unserialize($tokens)) === FALSE || !is_array($tokens)) + $tokens = array(); + + // purge old tokens if necessary + // + if ($purge_old) + { + if (empty($max_token_age_days)) $max_token_age_days = 2; + $now = time(); + $discard_token_date = $now - ($max_token_age_days * 86400); + $cleaned_tokens = array(); + foreach ($tokens as $token => $timestamp) + if ($timestamp >= $discard_token_date) + $cleaned_tokens[$token] = $timestamp; + $tokens = $cleaned_tokens; + } + + return $tokens; + +} + +/** + * Generates a security token that is then stored in + * the user's preferences with a timestamp for later + * verification/use (although session-based tokens + * are not stored in user preferences). + * + * NOTE: By default SquirrelMail will use a single session-based + * token, but if desired, user tokens can have expiration + * dates associated with them and become invalid even during + * the same login session. When in that mode, the note + * immediately below applies, otherwise it is irrelevant. + * To enable that mode, the administrator must add the + * following to config/config_local.php: + * $use_expiring_security_tokens = TRUE; + * + * NOTE: The administrator can force SquirrelMail to generate + * a new token every time one is requested (which may increase + * obscurity through token randomness at the cost of some + * performance) by adding the following to + * config/config_local.php: $do_not_use_single_token = TRUE; + * Otherwise, only one token will be generated per user which + * will change only after it expires or is used outside of the + * validity period specified when calling sm_validate_security_token() + * + * WARNING: If the administrator has turned the token system + * off by setting $disable_security_tokens to TRUE in + * config/config.php or the configuration tool, this + * function will not store tokens in the user + * preferences (but it will still generate and return + * a random string). + * + * @param boolean $force_generate_new When TRUE, a new token will + * always be created even if current + * configuration dictates otherwise + * (OPTIONAL; default FALSE) + * + * @return string A security token + * + * @since 1.4.19 and 1.5.2 + * + */ +function sm_generate_security_token($force_generate_new=FALSE) +{ + + global $data_dir, $username, $disable_security_tokens, $do_not_use_single_token, + $use_expiring_security_tokens; + $max_generation_tries = 1000; + + // if we're using session-based tokens, just return + // the same one every time (generate it if it's not there) + // + if (!$use_expiring_security_tokens) + { + if (sqgetGlobalVar('sm_security_token', $token, SQ_SESSION)) + return $token; + + // create new one since there was none in session + $token = GenerateRandomString(12, '', 7); + sqsession_register($token, 'sm_security_token'); + return $token; + } + + $tokens = sm_get_user_security_tokens(); + + if (!$force_generate_new && !$do_not_use_single_token && !empty($tokens)) + return key($tokens); + + $new_token = GenerateRandomString(12, '', 7); + $count = 0; + while (isset($tokens[$new_token])) + { + $new_token = GenerateRandomString(12, '', 7); + if (++$count > $max_generation_tries) + { + logout_error(_("Fatal token generation error; please contact your system administrator or the SquirrelMail Team")); + exit; + } + } + + // is the token system enabled? CAREFUL! + // + if (!$disable_security_tokens) + { + $tokens[$new_token] = time(); + setPref($data_dir, $username, 'security_tokens', serialize($tokens)); + } + + return $new_token; + +} + +/** + * Validates a given security token and optionally remove it + * from the user's preferences if it was valid. If the token + * is too old but otherwise valid, it will still be rejected. + * + * "Too old" is 2 days or older unless the administrator + * overrides that value using $max_token_age_days in + * config/config_local.php + * + * Session-based tokens of course are always reused and are + * valid for the lifetime of the login session. + * + * WARNING: If the administrator has turned the token system + * off by setting $disable_security_tokens to TRUE in + * config/config.php or the configuration tool, this + * function will always return TRUE. + * + * @param string $token The token to validate + * @param int $validity_period The number of seconds tokens are valid + * for (set to zero to remove valid tokens + * after only one use; set to -1 to allow + * indefinite re-use (but still subject to + * $max_token_age_days - see elsewhere); + * use 3600 to allow tokens to be reused for + * an hour) (OPTIONAL; default is to only + * allow tokens to be used once) + * NOTE this is unrelated to $max_token_age_days + * or rather is an additional time constraint on + * tokens that allows them to be re-used (or not) + * within a more narrow timeframe + * @param boolean $show_error Indicates that if the token is not + * valid, this function should display + * a generic error, log the user out + * and exit - this function will never + * return in that case. + * (OPTIONAL; default FALSE) + * + * @return boolean TRUE if the token validated; FALSE otherwise + * + * @since 1.4.19 and 1.5.2 + * + */ +function sm_validate_security_token($token, $validity_period=0, $show_error=FALSE) +{ + + global $data_dir, $username, $max_token_age_days, + $use_expiring_security_tokens, + $disable_security_tokens; + + // bypass token validation? CAREFUL! + // + if ($disable_security_tokens) return TRUE; + + // if we're using session-based tokens, just compare + // the same one every time + // + if (!$use_expiring_security_tokens) + { + if (!sqgetGlobalVar('sm_security_token', $session_token, SQ_SESSION)) + { + if (!$show_error) return FALSE; + logout_error(_("Fatal security token error; please log in again")); + exit; + } + if ($token !== $session_token) + { + if (!$show_error) return FALSE; + logout_error(_("The current page request appears to have originated from an untrusted source.")); + exit; + } + return TRUE; + } + + // don't purge old tokens here because we already + // do it when generating tokens + // + $tokens = sm_get_user_security_tokens(FALSE); + + // token not found? + // + if (empty($tokens[$token])) + { + if (!$show_error) return FALSE; + logout_error(_("This page request could not be verified and appears to have expired.")); + exit; + } + + $now = time(); + $timestamp = $tokens[$token]; + + // whether valid or not, we want to remove it from + // user prefs if it's old enough (unless requested to + // bypass this (in which case $validity_period is -1)) + // + if ($validity_period >= 0 + && $timestamp < $now - $validity_period) + { + unset($tokens[$token]); + setPref($data_dir, $username, 'security_tokens', serialize($tokens)); + } + + // reject tokens that are too old + // + if (empty($max_token_age_days)) $max_token_age_days = 2; + $old_token_date = $now - ($max_token_age_days * 86400); + if ($timestamp < $old_token_date) + { + if (!$show_error) return FALSE; + logout_error(_("The current page request appears to have originated from an untrusted source.")); + exit; + } + + // token OK! + // + return TRUE; + +} + +/** + * Wrapper for PHP's htmlspecialchars() that + * attempts to add the correct character encoding + * + * @param string $string The string to be converted + * @param int $flags A bitmask that controls the behavior of htmlspecialchars() + * (See http://php.net/manual/function.htmlspecialchars.php ) + * (OPTIONAL; default ENT_COMPAT, ENT_COMPAT | ENT_SUBSTITUTE for PHP >=5.4) + * @param string $encoding The character encoding to use in the conversion + * (OPTIONAL; default automatic detection) + * @param boolean $double_encode Whether or not to convert entities that are + * already in the string (only supported in + * PHP 5.2.3+) (OPTIONAL; default TRUE) + * + * @return string The converted text + * + */ +function sm_encode_html_special_chars($string, $flags=ENT_COMPAT, + $encoding=NULL, $double_encode=TRUE) +{ + if (!$encoding) + { + global $default_charset; + if ($default_charset == 'iso-2022-jp') + $default_charset = 'EUC-JP'; + $encoding = $default_charset; + } + + if (check_php_version(5, 2, 3)) { + // Replace invalid characters with a symbol instead of returning + // empty string for the entire to be encoded string. + if (check_php_version(5, 4, 0) && $flags == ENT_COMPAT) { + $flags = $flags | ENT_SUBSTITUTE; + } + return htmlspecialchars($string, $flags, $encoding, $double_encode); + } + + return htmlspecialchars($string, $flags, $encoding); +} +