<?php
/**
* SjisUtil
* 文字列処理関数集。
* $Id: SjisUtil.inc,v 1.1 2015/10/08 11:18:56 wanggb Exp $
* @author iimuro
* @access public
* @package jp.aimslib2.util
*/
class SjisUtil {
/**
* 文字化けしている漢字がきたら、%5Cを追加する処理
* @param String 文字列
* @return String %5Cを追加された文字列
*/
public function escapePostgresLike5C($str) {
// rawurlencode("—"),rawurlencode("ソ"),rawurlencode("Ы"),rawurlencode("�\")
// rawurlencode("噂"),rawurlencode("浬"),rawurlencode("欺"),rawurlencode("圭")
// rawurlencode("構"),rawurlencode("蚕"),rawurlencode("十"),rawurlencode("申")
// rawurlencode("曾"),rawurlencode("箪"),rawurlencode("貼"),rawurlencode("能")
// rawurlencode("表"),rawurlencode("暴"),rawurlencode("予"),rawurlencode("禄")
// rawurlencode("兔"),rawurlencode("喀"),rawurlencode("媾"),rawurlencode("彌")
// rawurlencode("拿"),rawurlencode("杤"),rawurlencode("歃"),rawurlencode("濬")
// rawurlencode("畚"),rawurlencode("秉"),rawurlencode("綵"),rawurlencode("臀")
// rawurlencode("藹"),rawurlencode("觸"),rawurlencode("軆"),rawurlencode("鐔")
// rawurlencode("饅"),rawurlencode("鷭"),rawurlencode("�x"),rawurlencode("�x")
$search_array = array("%81%5C","%83%5C","%84%5C","%87%5C",
"%89%5C","%8A%5C","%8B%5C","%8C%5C",
"%8D%5C","%8E%5C","%8F%5C","%90%5C",
"%91%5C","%92%5C","%93%5C","%94%5C",
"%95%5C","%96%5C","%97%5C","%98%5C",
"%99%5C","%9A%5C","%9B%5C","%9C%5C",
"%9D%5C","%9E%5C","%9F%5C","%E0%5C",
"%E1%5C","%E2%5C","%E3%5C","%E4%5C",
"%E5%5C","%E6%5C","%E7%5C","%E8%5C",
"%E9%5C","%EA%5C","%FA%78","%FB%78"
);
$replace_array = array("%81%5C%5C","%83%5C%5C","%84%5C%5C","%87%5C%5C",
"%89%5C%5C","%8A%5C%5C","%8B%5C%5C","%8C%5C%5C",
"%8D%5C%5C","%8E%5C%5C","%8F%5C%5C","%90%5C%5C",
"%91%5C%5C","%92%5C%5C","%93%5C%5C","%94%5C%5C",
"%95%5C%5C","%96%5C%5C","%97%5C%5C","%98%5C%5C",
"%99%5C%5C","%9A%5C%5C","%9B%5C%5C","%9C%5C%5C",
"%9D%5C%5C","%9E%5C%5C","%9F%5C%5C","%E0%5C%5C",
"%E1%5C%5C","%E2%5C%5C","%E3%5C%5C","%E4%5C%5C",
"%E5%5C%5C","%E6%5C%5C","%E7%5C%5C","%E8%5C%5C",
"%E9%5C%5C","%EA%5C%5C","%FA%78%5C","%FB%78%5C"
);
$encoded = rawurlencode($str);
$tmp = str_replace($search_array, $replace_array, $encoded);
$raw = rawurldecode($tmp);
return $raw;
}
/**
* 文字化けしている漢字がきたら、%5Cを追加する処理のつもり
* とりあえずは何もしない。ダミー関数。
* @param String 文字列
* @return String 文字列
*/
public function escapeMysqlLike5C($str) {
// rawurlencode("—"),rawurlencode("ソ"),rawurlencode("Ы"),rawurlencode("�\")
// rawurlencode("噂"),rawurlencode("浬"),rawurlencode("欺"),rawurlencode("圭")
// rawurlencode("構"),rawurlencode("蚕"),rawurlencode("十"),rawurlencode("申")
// rawurlencode("曾"),rawurlencode("箪"),rawurlencode("貼"),rawurlencode("能")
// rawurlencode("表"),rawurlencode("暴"),rawurlencode("予"),rawurlencode("禄")
// rawurlencode("兔"),rawurlencode("喀"),rawurlencode("媾"),rawurlencode("彌")
// rawurlencode("拿"),rawurlencode("杤"),rawurlencode("歃"),rawurlencode("濬")
// rawurlencode("畚"),rawurlencode("秉"),rawurlencode("綵"),rawurlencode("臀")
// rawurlencode("藹"),rawurlencode("觸"),rawurlencode("軆"),rawurlencode("鐔")
// rawurlencode("饅"),rawurlencode("鷭"),rawurlencode("�x"),rawurlencode("�x")
// $search_array = array("%81%5C","%83%5C","%84%5C","%87%5C",
// "%89%5C","%8A%5C","%8B%5C","%8C%5C",
// "%8D%5C","%8E%5C","%8F%5C","%90%5C",
// "%91%5C","%92%5C","%93%5C","%94%5C",
// "%95%5C","%96%5C","%97%5C","%98%5C",
// "%99%5C","%9A%5C","%9B%5C","%9C%5C",
// "%9D%5C","%9E%5C","%9F%5C","%E0%5C",
// "%E1%5C","%E2%5C","%E3%5C","%E4%5C",
// "%E5%5C","%E6%5C","%E7%5C","%E8%5C",
// "%E9%5C","%EA%5C","%FA%78","%FB%78"
// );
// $replace_array = array("%81%5C%5C","%83%5C%5C","%84%5C%5C","%87%5C%5C",
// "%89%5C%5C","%8A%5C%5C","%8B%5C%5C","%8C%5C%5C",
// "%8D%5C%5C","%8E%5C%5C","%8F%5C%5C","%90%5C%5C",
// "%91%5C%5C","%92%5C%5C","%93%5C%5C","%94%5C%5C",
// "%95%5C%5C","%96%5C%5C","%97%5C%5C","%98%5C%5C",
// "%99%5C%5C","%9A%5C%5C","%9B%5C%5C","%9C%5C%5C",
// "%9D%5C%5C","%9E%5C%5C","%9F%5C%5C","%E0%5C%5C",
// "%E1%5C%5C","%E2%5C%5C","%E3%5C%5C","%E4%5C%5C",
// "%E5%5C%5C","%E6%5C%5C","%E7%5C%5C","%E8%5C%5C",
// "%E9%5C%5C","%EA%5C%5C","%FA%78%5C","%FB%78%5C"
// );
// $encoded = rawurlencode($str);
// $tmp = str_replace($search_array, $replace_array, $encoded);
// $raw = rawurldecode($tmp);
// return $raw;
return $str;
}
public static function stripMysqlSlashesLike5C($str) {
if (defined("MYSQL_USE_5C_ESCAPE") && (!MYSQL_USE_5C_ESCAPE)) {
return $str;
}
$search_array = array("%81%5C%5C","%83%5C%5C","%84%5C%5C","%87%5C%5C",
"%89%5C%5C","%8A%5C%5C","%8B%5C%5C","%8C%5C%5C",
"%8D%5C%5C","%8E%5C%5C","%8F%5C%5C","%90%5C%5C",
"%91%5C%5C","%92%5C%5C","%93%5C%5C","%94%5C%5C",
"%95%5C%5C","%96%5C%5C","%97%5C%5C","%98%5C%5C",
"%99%5C%5C","%9A%5C%5C","%9B%5C%5C","%9C%5C%5C",
"%9D%5C%5C","%9E%5C%5C","%9F%5C%5C","%E0%5C%5C",
"%E1%5C%5C","%E2%5C%5C","%E3%5C%5C","%E4%5C%5C",
"%E5%5C%5C","%E6%5C%5C","%E7%5C%5C","%E8%5C%5C",
"%E9%5C%5C","%EA%5C%5C","%FA%78%5C","%FB%78%5C"
);
$replace_array = array("%81%5C","%83%5C","%84%5C","%87%5C",
"%89%5C","%8A%5C","%8B%5C","%8C%5C",
"%8D%5C","%8E%5C","%8F%5C","%90%5C",
"%91%5C","%92%5C","%93%5C","%94%5C",
"%95%5C","%96%5C","%97%5C","%98%5C",
"%99%5C","%9A%5C","%9B%5C","%9C%5C",
"%9D%5C","%9E%5C","%9F%5C","%E0%5C",
"%E1%5C","%E2%5C","%E3%5C","%E4%5C",
"%E5%5C","%E6%5C","%E7%5C","%E8%5C",
"%E9%5C","%EA%5C","%FA%78","%FB%78"
);
$encoded = rawurlencode($str);
$tmp = str_replace($search_array, $replace_array, $encoded);
$raw = rawurldecode($tmp);
return $raw;
}
/**
* 文字コードがSJIS対応のDBの場合において、文字化け文字をエスケープせずに、
* バックスラッシュとクォートだけをエスケープします。
*/
public function escapeMySQL5CwithCP932($str) {
$search_array = array("%81%5C","%83%5C","%84%5C","%87%5C",
"%89%5C","%8A%5C","%8B%5C","%8C%5C",
"%8D%5C","%8E%5C","%8F%5C","%90%5C",
"%91%5C","%92%5C","%93%5C","%94%5C",
"%95%5C","%96%5C","%97%5C","%98%5C",
"%99%5C","%9A%5C","%9B%5C","%9C%5C",
"%9D%5C","%9E%5C","%9F%5C","%E0%5C",
"%E1%5C","%E2%5C","%E3%5C","%E4%5C",
"%E5%5C","%E6%5C","%E7%5C","%E8%5C",
"%E9%5C","%EA%5C","%FA%78","%FB%78"
);
$replace_array = array("%81%FF","%83%FF","%84%FF","%87%FF",
"%89%FF","%8A%FF","%8B%FF","%8C%FF",
"%8D%FF","%8E%FF","%8F%FF","%90%FF",
"%91%FF","%92%FF","%93%FF","%94%FF",
"%95%FF","%96%FF","%97%FF","%98%FF",
"%99%FF","%9A%FF","%9B%FF","%9C%FF",
"%9D%FF","%9E%FF","%9F%FF","%E0%FF",
"%E1%FF","%E2%FF","%E3%FF","%E4%FF",
"%E5%FF","%E6%FF","%E7%FF","%E8%FF",
"%E9%FF","%EA%FF","%FA%78","%FB%78"
);
$encoded = rawurlencode(trim($str));
$encoded = str_replace($search_array, $replace_array, $encoded);
//ErrorLogger::doOutput("0000000000000:" . $encoded, 0);
// この時点で、5Cなのは純粋なバックスラッシュのみ。
// 文字化け回避した文字を最後の文字とする文字列に分割
if (strpos($encoded, "%FF") !== false) {
$tmp_list = explode("%FF", $encoded);
$tmp_list2 = array();
for ($x = 0; $x < count($tmp_list); $x++) {
if ($x == count($tmp_list) - 1) {
$tmp = $tmp_list[$x];
} else {
$tmp = $tmp_list[$x] . "%FF";
}
array_push($tmp_list2, $tmp);
//ErrorLogger::doOutput("X" . $tmp, 0);
}
} else {
$tmp_list2 = array($encoded);
}
$count = 0;
$tmp_result = "";
foreach ($tmp_list2 as $tmp) {
//ErrorLogger::doOutput("X" . $count . ":" . $tmp, 0);
$count++;
$tail = "";
// 最後がバックスラッシュだったら、とりあえず除外。
while (true) {
if ((strlen($tmp) >= 6)
&& ((substr($tmp, -3) == "%5C") || (substr($tmp, -3) == "%27"))){
$tail = substr($tmp, -3) . $tail;
$tmp = substr($tmp, 0, strlen($tmp) - 3);
} else {
break;
}
}
//ErrorLogger::doOutput("MAIN:" . $tmp, 0);
//ErrorLogger::doOutput("TAIL:" . $tail, 0);
// 置換された最後の文字が、5C単独なのか、２バイト文字なのかを判別
$add_escape = false;
if ((strlen($tmp) >= 6) && (substr($tmp, -3) == "%FF")) {
$x = -3;
$ok_count = 0;
while (true) {
$chr = substr($tmp, $x -3, $x);
if ($chr == "") {
break;
}
if (!SjisUtil::isSJISFirstByte($chr)) {
break;
} else {
$ok_count++;
}
$x -= 3;
}
// 文字化け対象として、FFに置換された文字が、その前のSJIS2バイト文字の一部であるかを、
// SJISの先頭文字である「81〜9F」と「E0〜EF」を２バイト文字が使われていないところから、
// 奇数であるか偶数であるかで判断する。
if (($ok_count > 0) && ($ok_count % 2 == 1 )) {
// 奇数はそのまま→文字の一部
//ErrorLogger::doOutput("ODD", 0);
} else {
// 偶数なら、１こ余計にエスケープ→たまたまエスケープ文字と同じ形になったバックスラッシュ
//ErrorLogger::doOutput("EVEN", 0);
$add_escape = true;
}
} else {
// 2バイト文字の長さが無い
}
// バックスラッシュだけをエスケープ
$tmp = str_replace("%5C", "%5C%5C", $tmp);
$tail = str_replace("%5C", "%5C%5C", $tail);
$tmp = str_replace("%27", "%5C%27", $tmp);
$tail = str_replace("%27", "%5C%27", $tail);
// 元に戻す。
$tmp = str_replace($replace_array, $search_array, $tmp);
// 間違ってFFエスケープされてしまった5Cをエスケープ
if ($add_escape) {
$tmp .= "%5C";
}
// 切り取った部分を元に戻す。
$tmp_result .= $tmp . $tail;
}
// やっと元に戻れた
$raw = rawurldecode($tmp_result);
//ErrorLogger::doOutput("4444444444444:" . $raw, 0);
return $raw;
}
/**
* 対象の文字列が、SJIS文字の頭の2文字である「81〜9F」と「E0〜EF」に含まれるかどうか？
*/
public function isSJISFirstByte($str) {
if (preg_match("/^%[89EF][0-9A-F]$/", $str)) {
//ErrorLogger::doOutput("OK:" . $str, 0);
return true;
}
//ErrorLogger::doOutput("NOT FIRST BYTE:" . $str, 0);
return false;
}
}
?>