<?php
  require("sys.php");
include_once("jcode.php");
include_once("code_table.jis2ucs");
// Eroge RSS Cecker(http://kiisu.jpn.org/)の部分を書きかえること。
ini_set('user_agent', 'Eroge RSS Cecker(http://kiisu.jpn.org/)');
  $de = $_GET["de"];
  $rss_id00 = $_GET["rss_id00"];
  $start = time();
  $now = date("Y-m-d");
  $nowtime = date("Y-m-d H:i:s");
  $day7 = date("Y-m-d H:i:s" , strtotime("-7 day"));
  $lastmonth = date("Y-m-d H:i:s", strtotime("-1 month"));
  $month3 = date("Y-m-d H:i:s", strtotime("-3 month"));
  $month2 = date("Y-m-d H:i:s", strtotime("-2 month"));
//　何日前のデータまで（登録日＝データ元作成日）にするか
  $day3 = date("Y-m-d" , strtotime("-3 day"));
//　何時間前の更新まで、データ登録するか
  $lastcount = date("Y-m-d H:i:s", strtotime("-1 hours"));
  $cnt = 0;
  $con = mysql_connect($DBSERVER, $DBUSER, $DBPASSWORD);
  $selectdb = mysql_select_db($DBNAME, $con);

//　データ登録モードでない場合、データ取得
if(empty($de)) {
  $sql = "SELECT * FROM rss_list where rss_uri != ''";
if(!empty($rss_id00))   $sql .= " and rss_id = " . $rss_id00;
  $rst = mysql_query($sql, $con);
  while ($col = mysql_fetch_array($rst) and $cnt < 1) {
$number = $col["number"];
$rss_id = $col["rss_id"];
$entry_day = $col["entry_day"];
$renewal_day = $col["renewal_day"];
$renewal_day = strtotime("$renewal_day");


//　最終登録日が1週間以前、1、2、3ヶ月前以前の場合に更新チェックを一定の確率で行うようにする。
  $sql3 = "SELECT entry_time FROM rss_content where number = '$number' order by entry_time desc limit 1";
  $rst3 = mysql_query($sql3, $con);
print mysql_error();
  while ($col3 = mysql_fetch_array($rst3)) {
$lastdate = $col3["entry_time"];
}
srand(microtime()*1000000);
if($lastdate < $lastmonth) $randam = rand(1,5);
elseif($lastdate < $month2) $randam = rand(1,10);
elseif($lastdate < $month3) $randam = rand(1,50);
else $randam = 1;
//　乱数が「1」になった場合に更新チェック
if(empty($randam) or $randam == 1) {

//　「Last-Modified」で更新日を比較
$fp = fopen($col["rss_uri"], 'rb');
if(!empty($fp)) {
$header = stream_get_meta_data($fp);
fclose($fp);
$header = preg_grep("/^Last-Modified/",$header["wrapper_data"]);
$header = array_values($header);
$header = $header[0];
$header = substr($header,15);
}
$header = strtotime("$header");

//　更新されていると考えられる場合、データ取得
if($header > $renewal_day or $header == "-1" or $entry_day > $day3) {
$a = file_get_contents($col["rss_uri"]);
if(empty($a)) {
  $sql2 = "update rss_list set error_count = error_count +1 where rss_id = '$rss_id'";
  $rst2 = mysql_query($sql2, $con);
print mysql_error();
print "error";
}
else {
// $a = mb_convert_encoding($a,"UTF-8","SJIS,UTF-8,EUC-JP,ASCII,JIS");
$a = JcodeConvert($a, 0, 4);
//自動追加される広告コードやヘッダ部分の削除
preg_match("/\<\/head\>(.*?\<\/body\>)/is",$a,$matches);
if(!empty($matches[1])) $a = $matches[1];
$patterns = array ("/\<!-- geoguide start --\>.*?\<!-- geoguide end --\>/", "/\<!-- AAACAFE banner tag --\>.*?\<!-- AAACAFE banner tag --\>/");
$replace = array ("", "");
$a = preg_replace($patterns, $replace, $a);

$rssc = $col["rss_copy"];

//　現在のデータと比較して新しいか、失敗していない場合、データ登録
if($a != $rssc) {
$a = addslashes($a); 
  $sql2 = "update rss_list set rss_copy = '$a' , renewal_day = '$nowtime' , error_count = '0'  where rss_id = '$rss_id' ";
  $rst2 = mysql_query($sql2, $con);
print mysql_error();
}

//　取得データに削除コード「erc_erase」が記述されている場合、全データ削除、アドレス変更
if(eregi("erc_erase",$a)) {
$number = $col["number"];
if(eregi("erc_erase:http",$a)) {
 preg_match ("/erc_erase:(.*?):(.*?):/is", $a, $iten);
  $sql2 = "update rss_list set rss_uri = '$iten[2]' , site_address = '$iten[1]' where rss_id = '$rss_id' ";
  $rst2 = mysql_query($sql2, $con);
}
else {
  $sql2 = "delete FROM rss_list where rss_id = '$rss_id'";
  $rst2 = mysql_query($sql2, $con);
//  $sql2 = "delete FROM rss_content where number = '$number'";
//  $rst2 = mysql_query($sql2, $con);
}
print mysql_error();
}
}
}
//　データ取得に「$ec」回失敗した場合、全データ削除
if($col["error_count"] >= $ec) {
  $sql2 = "delete from rss_list where rss_id = '$number'";
  $rst2 = mysql_query($sql2, $con);
  $sql2 = "delete from rss_content where rss_id = '$number'";
  $rst2 = mysql_query($sql2, $con);
}
}
}
mysql_free_result($rst);
}


if($de == "on") $limit = 700;

//データ登録
  $sql = "SELECT rss_copy,number,site_name,site_address,entry_day,renewal_day,error_count,title_marker_start,title_marker_end,link_marker_start,link_marker_end,description_marker_start,description_marker_end,subject_marker_start,subject_marker_end,date_marker_start,date_marker_end,category,rss_id FROM rss_list where renewal_day >= '$lastcount'";
if(!empty($rss_id00))   $sql .= " and rss_id = " . $rss_id00;
  $rst = mysql_query($sql, $con);
print mysql_error();
  while ($col = mysql_fetch_array($rst)) {
$number = $col["number"];
if(!empty($limit2)) {
$limit /= 2;
unset($limit2);
}

//　RSS2.0の場合
if(strpos($a,"rss version=\"2.0\"") != 0 and empty($col["site_name"])) {
  $sql2 = "update rss_list set date_marker_start = '<pubDate>' , date_marker_end = '</pubDate>' , subject_marker_start = '<category>' , subject_marker_end = '</category>'  where number = '$number' ";
  $rst2 = mysql_query($sql2, $con);
}

$a = $col["rss_copy"];
$a = str_replace ("\t", "", $a);
$a = str_replace ("\n", "", $a);
$a = str_replace ("\r", "", $a);
$a = preg_replace("/\40+/"," ",$a);

// 開始・終了記号を正規表現で使用出来るように加工
if(strpos($col[title_marker_start],"[#on]") !== false)  $tms = str_replace ("[#on]", "", $col[title_marker_start]);
else $tms = preg_quote ($col[title_marker_start],"/"); 
if(strpos($col[title_marker_end],"[#on]") !== false) $tme = str_replace ("[#on]", "", $col[title_marker_end]);
else $tme = preg_quote ($col[title_marker_end],"/");
if(strpos($col[link_marker_start],"[#on]") !== false) $lms = str_replace ("[#on]", "", $col[link_marker_start]);
else $lms = preg_quote ($col[link_marker_start],"/");
if(strpos($col[link_marker_end],"[#on]") !== false) $lme = str_replace ("[#on]", "", $col[link_marker_end]);
else $lme = preg_quote ($col[link_marker_end],"/");
if(strpos($col[date_marker_start],"[#on]") !== false) $dms = str_replace ("[#on]", "", $col[date_marker_start]);
else $dms = preg_quote ($col[date_marker_start],"/");
if(strpos($col[date_marker_end],"[#on]") !== false) $dme = str_replace ("[#on]", "", $col[date_marker_end]);
else $dme = preg_quote ($col[date_marker_end],"/");
if(strpos($col[subject_marker_start],"[#on]") !== false) $sms = str_replace ("[#on]", "", $col[subject_marker_start]);
else $sms = preg_quote ($col[subject_marker_start],"/");
if(strpos($col[subject_marker_end],"[#on]") !== false) $sme = str_replace ("[#on]", "", $col[subject_marker_end]);
else $sme = preg_quote ($col[subject_marker_end],"/");
if(strpos($col[description_marker_start],"[#on]") !== false) $dems = str_replace ("[#on]", "", $col[description_marker_start]);
else $dems = preg_quote ($col[description_marker_start],"/");
if(strpos($col[description_marker_end],"[#on]") !== false) $deme = str_replace ("[#on]", "", $col[description_marker_end]);
else $deme = preg_quote ($col[description_marker_end],"/");

//　サイト名がない場合（RSS登録の場合）、サイト情報を取得、登録件数を2倍に
if(empty($col["site_name"])) {
preg_match ( "/$tms(.+?)$tme/is", $a, $title);
if(!empty($dems)) preg_match ( "/$dems(.+?)$deme/is", $a, $description);
if(!empty($lms)) preg_match ( "/$lms(.+?)$lme/is", $a, $link);
  $sql2 = "insert into homepagelist(url,title) values('$link[1]','$title[1]')";
  $rst2 = mysql_query($sql2, $con);
  $sql2 = "select id from homepagelist where title = '$title[1]'";
  $rst2 = mysql_query($sql2,$con);
  while ($col2 = mysql_fetch_array($rst2)) {
$number2 = $col2["id"];
}
  $sql2 = "update rss_list set site_address = '$link[1]' , site_name = '$title[1]' , site_summary = '$description[1]', number = '$number2'  where rss_id = '$col[rss_id]' ";
  $rst2 = mysql_query($sql2, $con);
$limit2 = $limit;
$limit *= 2;
}
if($col["entry_day"] == $now) {
$limit2 = $limit;
$limit *= 2;
}

//　RSSの場合の諸作業
 if(eregi("<item>",$a)) {
$b = strpos($a,"<item>");
$a = substr($a,$b + 5 );
}
 if(eregi("<item rdf",$a)) {
$b = strpos($a,"<item rdf");
$a = substr($a,$b + 9 );
}
 if(strpos($a,"<![CDATA[") != "0") {
$a = str_replace("<![CDATA[","",$a);
$a = str_replace("]]>","",$a);
}

//　各要素を取得
//　負荷回避のため、正規表現を使う場合と使わない場合で場合分け
if(!empty($a)) {
if(strpos($col[title_marker_start],"[#on]") === false and strpos($col[title_marker_end],"[#on]") === false ) {
$tms = $col["title_marker_start"];
$tme = $col["title_marker_end"];
$ts = mb_strlen($tms);
$ts2 = mb_strlen($tme);
  $z = $a;
  $cnt = 0;
  while ($cnt < $limit) {
//　「mb_strpos」英字の大文字、小文字を区別してしまう。難有り。
$b = mb_strpos($z,$tms);
if(!is_int($b))  break;
$z = mb_substr($z,$b + $ts );
$c = mb_strpos($z,$tme);
$title[1][$cnt] = mb_substr($z,0,$c);
$z = mb_substr($z,$ts2 +$c);
  $cnt++;
}
}
else {
preg_match_all ( "/$tms(.+?)$tme/is", $a, $title);
}

if(!empty($dems)) {
if(strpos($col[description_marker_start],"[#on]") === false and strpos($col[description_marker_end],"[#on]") === false ) {
  $cnt = 0;
  $z = $a;
$dems = $col["description_marker_start"];
$deme = $col["description_marker_end"];
$ts = mb_strlen($dems);
$ts2 = mb_strlen($deme);
  while ($cnt < $limit) {
$b = mb_strpos($z,$dems);
if(!is_int($b))  break;
$z = mb_substr($z,$b + $ts );
$c = mb_strpos($z,$deme);
$description[1][$cnt] = mb_substr($z,0,$c );
$z = mb_substr($z,$ts2 +$c);
  $cnt++;
}
}
else{
preg_match_all ( "/$dems(.*?)$deme/is", $a, $description);
}
}
if(!empty($sms)) {
if(strpos($col[subject_marker_start],"[#on]") === false and strpos($col[subject_marker_end],"[#on]") === false ) {
  $cnt = 0;
  $z = $a;
$sms = $col["subject_marker_start"];
$sme = $col["subject_marker_end"];
$ts = mb_strlen($sms);
$t2s = mb_strlen($sme);
  while ($cnt < $limit) {
$b = mb_strpos($z,$sms);
if(!is_int($b))  break;
$z = mb_substr($z,$b + $ts );
$c = mb_strpos($z,$sme);
$subject[1][$cnt] = mb_substr($z,0,$c );
$z = mb_substr($z,$ts2 +$c);
  $cnt++;
}
}
else {
preg_match_all ( "/$sms(.*?)$sme/is", $a, $subject);
}
}

if(!empty($dms)) {
if(strpos($col[date_marker_start],"[#on]") === false and strpos($col[date_marker_end],"[#on]") === false ) {
  $cnt = 0;
  $z = $a;
$dms = $col["date_marker_start"];
$dme = $col["date_marker_end"];
$ts = mb_strlen($dms);
$ts2 = mb_strlen($dme);
  while ($cnt < $limit) {
$b = mb_strpos($z,$dms);
if(!is_int($b))  break;
$z = mb_substr($z,$b + $ts );
$c = mb_strpos($z,$dme);
$date[1][$cnt] = mb_substr($z,0,$c );
$z = mb_substr($z,$ts2 +$c);
  $cnt++;
}
}
else {
preg_match_all ( "/$dms(.*?)$dme/is", $a, $date);
}
}
//　リンク部分の取得、加工
if(!empty($lms)) {
preg_match_all ( "/$lms(.+?)$lme/is", $a, $link);
$count_link = count ($link[1]);
$cnt_link = 0;
if(eregi("href",$link[1][$cnt_link])) {
preg_match_all ( "/href.?\=.?\"?\'?(.+?)\"?\'?\>/im", $a, $link);
}
  while ($count_link > $cnt_link) {
if(eregi("http://",$link[1][$cnt_link]))  { $cnt_link++; }
else {
$col["site_address"] = preg_replace("/\/[^\/]+$/", "/", $col["site_address"]);
$link[1][$cnt_link] = $col["site_address"] . $link[1][$cnt_link];
if(substr_count($link[1][$cnt_link],"/") == 4) $link[1][$cnt_link] = preg_replace("/([^\:\:])\/\.*?\//", "\\1/", $link[1][$cnt_link]);
else  $link[1][$cnt_link] = preg_replace("/[^http\:\:\/]\/[^\/]+\/\.*?\//", "/", $link[1][$cnt_link]);
$cnt_link++;
}
}
}
}
$address = $col["site_address"];
$category = $col["category"];
$entry_day = $col["entry_day"];
$cnt = 0;
//　各要素の補足、加工
  while ($cnt < $limit and !empty($title[1][$cnt])) {
$title[1][$cnt] = strip_tags($title[1][$cnt]);
$title[0][$cnt] = strip_tags($title[0][$cnt]);
$title[1][$cnt] = str_replace ("\n", "", $title[1][$cnt]);
$title[1][$cnt] = str_replace ("\r", "", $title[1][$cnt]);
$title[1][$cnt] = str_replace ("&nbsp;", " ", $title[1][$cnt]);
$title[1][$cnt] = mb_convert_kana ($title[1][$cnt], "as");
$title[1][$cnt] = preg_replace("/\40+/"," ",$title[1][$cnt]);
$title[1][$cnt] = addslashes ($title[1][$cnt]);
$title[1][$cnt] = trim($title[1][$cnt]);

//　文字数制限
if(strlen($description[1][$cnt]) > 500) unset($description[1][$cnt]);
if(strlen($link[1][$cnt]) > 500) unset($link[1][$cnt]);
if(strlen($title[1][$cnt]) > 500) unset($title[1][$cnt]);
if(strlen($subject[1][$cnt]) > 500) unset($subject[1][$cnt]);
if(strlen($date[1][$cnt]) > 500) unset($date[1][$cnt]);

$title2 = $title[1][$cnt];
//　分類の捕捉、加工
$sub = "";
$cc1 = count($category1);
if(!empty($subject[1][$cnt])) {
$cc2 = 0;
  while ($cc1 > $cc2) {
if(eregi($category1[$cc2],$subject[1][$cnt])) { 
if(!empty($sub)) $sub .= ",";
$sub .= $category2[$cc2];
$cc2++;
}
else {$cc2++;}
}
}
$subject[1][$cnt] = $sub;
if(!empty($subject[1][$cnt]) and !empty($category)) $subject[1][$cnt] = $category . "," . $subject[1][$cnt];
if(empty($subject[1][$cnt]) and !empty($category)) $subject[1][$cnt] = $category;
if(empty($subject[1][$cnt]) and empty($category)) {
$cc1 = count($category1);
if(!empty($description[1][$cnt])) {
$cc2 = 0;
  while ($cc1 > $cc2) {
if(eregi($category1[$cc2],$description[1][$cnt])) { 
if(!empty($subject[1][$cnt])) $subject[1][$cnt] .= ",";
$subject[1][$cnt] .= $category2[$cc2];
$cc2++;
}
else {$cc2++;}
}
}

if(!empty($title[0][$cnt]) and empty($subject[1][$cnt])) {
$cc2 = 0;
  while ($cc1 > $cc2) {
if(eregi($category1[$cc2],$title[0][$cnt])) {
if(!empty($subject[1][$cnt])) $subject[1][$cnt] .= ",";
$subject[1][$cnt] .= $category2[$cc2];
}
$cc2++;
}
}
}
$category3 = array_unique($category2);
$category3 = array_values ($category3);
$cc1 = count($category3);
$cc2 = 0;
$sub = "";
  while ($cc1 > $cc2) {
$sub2 = mb_substr_count($subject[1][$cnt], $category3[$cc2]);
if($sub2 >= 1) {
if(!empty($sub)) $sub .= ",";
$sub .= $category3[$cc2];
}
$cc2++;
}

$subject[1][$cnt] = $sub;
//　既に登録されているかチェック

$subject2 = $subject[1][$cnt];
$subject2 = addslashes ($subject2);
  $sql2 = "SELECT number,title,subject,description,date FROM rss_content where (number = '$number' or (description = '$number' and number = 0)) and title = '$title2' and subject = '$subject2'";
  $rst2 = mysql_query($sql2, $con);
  $recmax = mysql_num_rows($rst2);
print mysql_error();

// if(empty($link[1][$cnt])) $link[1][$cnt] = $address;
if($link[1][$cnt] == $address) $link[1][$cnt] = "";

//　英文形式の日付を「0000-00-00」形式の日付に変換。難有り。
if(strpos($date[1][$cnt],"+0900") != "0") $date[1][$cnt] = date("Y-m-d",strtotime($date[1][$cnt]));
if(strpos($date[1][$cnt],"JST") != "0") $date[1][$cnt] = date("Y-m-d",strtotime($date[1][$cnt]));

if(empty($date[1][$cnt]) and $day3 > $entry_day and $limit == "15") $date[1][$cnt] = $now;
$description[1][$cnt] = addslashes ($description[1][$cnt]);
$link[1][$cnt] = addslashes ($link[1][$cnt]);
$date2 = $date[1][$cnt];
$description[1][$cnt] = strip_tags($description[1][$cnt]);
$description2 = $description[1][$cnt];
$link2 = $link[1][$cnt];
if($description2 == $subject2) unset($description2);

//　データ修正
if($recmax == 1) {
  while ($col2 = mysql_fetch_array($rst2)) {
$des2 = $col2["description"];
$det2 = $col2["date"];
}
if(!empty($description2) and strpos($des2, $description2) === false and $det2 <= $date2) {
$des3 = $description2 . " 《" . $det2 . " ⇒ " . $des2;
  $sql2 = "update rss_content set description = \"$des3\" , date = \"$date2\" , entry_time = \"$nowtime\" where number = \"$number\" and title = \"$title2\" and subject = \"$subject2\"";
if($number == 109) $sql2 = "";
  $rst2 = mysql_query($sql2, $con);
print mysql_error();
}
$cnt++;}

//　データ登録
elseif($recmax == 0) {
if(!empty($title2) and !empty($subject2)) {
  $sql2 = "insert into rss_content (title,link,description,subject,date,number,entry_time) values (\"$title2\",\"$link2\",\"$description2\",\"$subject2\",\"$date2\",\"$number\",\"$nowtime\")";
  $rst2 = mysql_query($sql2, $con);
print mysql_error();
}
$cnt++;
}
else {
if(!empty($description[1][$cnt])) {
print "error";
}
$cnt++;}
}
unset($title);
unset($link);
unset($description);
unset($subject);
unset($date);
}
	mysql_close($con);

//　ログファイルに書きこみ
$end = time();
$nowtime = date("Y-m-d H:i:s") . ($end - $start) . "秒\n";
$logfile = "exec.log";
	$fp = fopen($logfile,"a+");
	flock($fp,2);
	fputs($fp,$nowtime);
	flock($fp,3);
	fclose($fp);
?>