// Copyright: Copyright (C) 2002-2007 Pristine Communications
// This file is in UTF-8 萬國碼
// Set the include path
if (!defined("INCPATH_SET")) {
require_once dirname(__FILE__) . "/incpath.inc.php";
}
// Referenced subroutines
require_once "monica/htmlchar.inc.php";
require_once "monica/markabbr.inc.php";
require_once "monica/urlregex.inc.php";
// The URL regular-expression pattern
define("_A2HTML_URLREGEX_COMBINED", "(?:" . URLREGEX_EMAIL . "|" . URLREGEX_FTP
. "|" . URLREGEX_FILE . "|" . URLREGEX_HTTP . "|" . URLREGEX_HTTPS
. "|" . URLREGEX_GOPHER . "|" . URLREGEX_MAILTO . "|" . URLREGEX_NEWS
. "|" . URLREGEX_NNTP . "|" . URLREGEX_TELNET . "|" . URLREGEX_WAIS
. "|" . URLREGEX_PROSPERO . ")");
// The substitution ID configuration
define("_A2HTML_SUBST_ID_LEN", 5);
define("_A2HTML_SUBST_ID_MIN", pow(10, _A2HTML_SUBST_ID_LEN - 1));
define("_A2HTML_SUBST_ID_MAX", pow(10, _A2HTML_SUBST_ID_LEN) - 1);
// The punctuation information
$_A2HTML_QUOTES = array(
"\"" => "\"",
"'" => "'",
"`" => "'",
"(" => ")",
"[" => "]",
"{" => "}",
"<" => ">",
"「" => "」",
"『" => "』",
"〈" => "〉",
"《" => "》",
"【" => "】",
"〔" => "〕",
"”" => "”",
"“" => "”",
"‘" => "’",
"‵" => "′",
);
$_A2HTML_DOTS = array(
"...", ".", ",", ":", ";", "!", "?",
"。", "，", "：", "；", "、", "…", "！", "？",
);
// a2html: Convert a textarea input into HTML content
function a2html($source)
{
// The punctuation information
global $_A2HTML_QUOTES, $_A2HTML_DOTS;
$result = $source;
// Find and record all the URL's
// Process line by line. PHP PCRE has serial bugs that crashes
// randomly when matching RFC-822 e-mails. The reason is not clear.
// We try to avoid it by reducing the text piece size as possible
// while keeping the integrity of the text. This is not a solution,
// though.
$lines = explode("\n", $result);
for ($i = 0; $i < count($lines); $i++) {
$temp = $lines[$i];
$lines[$i] = "";
if (substr($temp, 0, 1) == " ") {
$lines[$i] .= " ";
$temp = substr($temp, 1);
}
$substs = array();
while ( preg_match("/^(.*?)(" . _A2HTML_URLREGEX_COMBINED . ")(.*)$/s",
$temp, $m)) {
// De-quote
foreach (array_keys($_A2HTML_QUOTES) as $start) {
$end = $_A2HTML_QUOTES[$start];
// In a quotation
if ( substr($m[1], -strlen($start)) == $start
&& substr($m[2], -strlen($end)) == $end) {
$m[2] = substr($m[2], 0, -strlen($end));
$m[3] = $end . $m[3];
break;
}
}
// Remove trailing dots
foreach ($_A2HTML_DOTS as $dot) {
// End in this dot
if (substr($m[2], -strlen($dot)) == $dot) {
$m[2] = substr($m[2], 0, -strlen($dot));
$m[3] = $dot . $m[3];
break;
}
}
// Still matched after the above removal
if (preg_match("/^" . _A2HTML_URLREGEX_COMBINED . "$/", $m[2])) {
$lines[$i] .= _a2html_plain_line($m[1]) . _a2html_url2link($m[2]);
} else {
$lines[$i] .= _a2html_plain_line($m[1] . $m[2]);
}
$temp = $m[3];
}
// Add the remains
$lines[$i] .= _a2html_plain_line($temp);
}
$result = implode("\n", $lines);
// Add line breaks
$result = preg_replace("/(\r?\n)/", "
\\1", $result);
return $result;
}
// _a2html_plain_line: Convert plain line content to HTML
function _a2html_plain_line($line)
{
// Escape the HTML characters and mark the abbreviation
$line = h_abbr($line);
// Tag non-breaking spaces
$line = str_replace(" ", " ", $line);
return $line;
}
// _a2html_url2link: Convert an URL to a link
function _a2html_url2link($url)
{
// There is a scheme
if (preg_match("/^" . URLREGEX_SCHEME . ":/", $url)) {
return "" . h($url) . "";
// Schemeless - assumed to be an e-mail address
} else {
return "" . h($url) . "";
}
}
?>