<?php set_time_limit(0); //转存本地地址 define( 'STORE_PATH', dirname(__FILE__) . '/../../../../temp_data/test/' ); define('LIST_PATH', STORE_PATH . 'list/'); define('CHAPTER_PATH', LIST_PATH . 'chapter/'); define('DETAIL_PATH', LIST_PATH . 'detail/'); createFolder(STORE_PATH); createFolder(LIST_PATH); createFolder(CHAPTER_PATH); createFolder(DETAIL_PATH); $ftp_server = "ip"; $ftp_user = "anonymous"; $ftp_pass = "anonymous"; $conn_id = ftp_connect ( $ftp_server ) or die ( "Couldn't connect to $ftp_server" ); if (@ftp_login ( $conn_id, $ftp_user, $ftp_pass )) { echo "Connected as $ftp_user@$ftp_server\n"; } else { echo "Couldn't connect as $ftp_user\n"; } ftp_pasv($conn_id, true); /** * 创建文件并写入内容 * * @param string $path path/ * @param string $fileName filename * @param string $data content * * @return string 返回路径 */ function createFile ($path, $fileName, $data) { if (empty($data)) { return false; } file_put_contents($path . $fileName, $data); return $path . $fileName; } /** * 创建目录 * * @param string $path path * * @return no */ function createFolder ($path) { if (! file_exists($path)) { createFolder(dirname($path)); if (mkdir($path, 0777)) { echo "\n dir not find ,make dir " . dirname($path) . " is ok!"; } else { echo "\n dir not find ,make dir " . dirname($path) . " is fail!"; } } } /** * 抓取电子书籍的类别 * * @param string $xmlUrl xml地址 * @param string $saveFileName 文件名称 * * @return string 返回路径 */ function getClassXml ($conn_id,$categoryDir, $saveFileName = 'category.xml') { //获取顶级栏目 $clist = getDirListFormFtp($conn_id, $categoryDir); $itemTPL = '<Item><ID>%s</ID><Name>%s</Name></Item>'; $items=''; foreach ($clist as $key=>$value){ $items .=sprintf($itemTPL,$value['name'],$value['detail']); } $bodyTPL='<Tofo><Class>%s</Class></Tofo>'; $data = sprintf($bodyTPL,$items); echo "\n".createFile(STORE_PATH, $saveFileName, $data); } set_time_limit(0); $start_time = time(); $bookCache=array(); $categoryDir = "\\tingshu\\web"; getClassXml($conn_id,$categoryDir); //获取顶级栏目 $clist = getDirListFormFtp($conn_id, $categoryDir); //获取二级栏目分类 foreach ($clist as $key=>$value){ $_secondCateoryDir = $categoryDir.'\\'.$value['name']; $_secondCateorylist = getDirListFormFtp($conn_id, $_secondCateoryDir); $listData=''; $listTPL='<Tofo><Class><ID>%s</ID><Name>%s</Name><Books>%s</Books></Class></Tofo>'; $items ='<Item><ID>%s</ID><Name>%s</Name></Item>'; $listItemString=''; //获取详细书籍章节列表 foreach ($_secondCateorylist as $key=>$book){ $listItemString.=sprintf($items,$book['name'], $book['detail']); $_bookListDir = $_secondCateoryDir.'\\'.$book['name']; $chapters = getBook($conn_id,$_bookListDir); //生产book章节html getBookofChapter ($book,$chapters); //缓存bookid和章节信息 $bookCache[$book['name']] = array('category'=>$value['name'],'chapters'=>$chapters); } //生成二级栏目列表页 $listData = sprintf($listTPL, $value['name'], $value['detail'],$listItemString); $list_save_path = $value['name'] . '.xml'; echo "\n".createFile(LIST_PATH, $list_save_path, $listData); } $arrString = "<?php \n \$bookCache=".var_export($bookCache, true).";\n?>"; echo "\n建立缓存文件：".createFile(STORE_PATH, 'bookcache.php', $arrString); /** * 抓取书籍章节信息 * * @param array $bookIds 书籍章节信息 * * @return boolean 返回是否抓取成功 */ function getBookofChapter ($bookinfo,$chapters) { if (! is_array($chapters)) { return false; } $bookData = ''; $bookTPL = '<Tofo><Books><ID>%s</ID><Name>%s</Name><Volumes>%s</Volumes></Books></Tofo>'; $bookItemTPL='<Item><ID>%s</ID><Name>%s</Name></Item>'; //<Play>%s</Play> //<Download>%s</Download> //,$item['downurl'],$item['downurl'] $chapterString=''; foreach ($chapters as $key=>$item){ $chapterString.=sprintf($bookItemTPL,$item['name'],$item['detail']); } //生成二级栏目列表页 $bookData = sprintf($bookTPL, $bookinfo['name'], $bookinfo['detail'],$chapterString); $book_chapter_save_path = $bookinfo['name'] . '.xml'; if (! empty($bookData)) { echo "\n".createFile(CHAPTER_PATH, $book_chapter_save_path, $bookData); } return true; } function getBook($conn_id,$_bookListDir){ $chapter=array(); $buff = ftp_nlist ( $conn_id, $_bookListDir ); if(is_array($buff)){ $resourceArray = array(); foreach ($buff as $key=>$value){ if(strstr( $value, '.txt' )){ }else{ $resourcesName = str_replace($_bookListDir.'\\', "", $value); $temp = preg_split ( '/\./',$resourcesName); $resourceArray[trim($temp[0])]=$resourcesName; } } foreach ($buff as $key=>$value){ if(strstr( $value, '.txt' )){ $name = trim(str_replace(".txt","",str_replace($_bookListDir.'\\', "", $value))); $chapter[$name] = array ( 'name' => $name, 'detail' => getFileContentsFormFtp ( $conn_id, $value ), 'downurl' =>$resourceArray[$name] ); } } } ksort($chapter , SORT_NUMERIC); echo "\n"; var_dump('chapter index :'.implode(array_keys ($chapter), ',')); echo "\n"; return $chapter; } function getDirListFormFtp($conn_id, $categoryDir) { $categoryArray = array (); $dirs = getDirNameFormFtp ( $conn_id, $categoryDir ); //获取分类描述 foreach ( $dirs as $key => $value ) { $path = $categoryDir . '\\' . $value . '.txt'; $categoryArray [] = array ('name' => $value, 'detail' => getFileContentsFormFtp ( $conn_id, $path ) ); } return $categoryArray; } function getFileContentsFormFtp($conn_id, $server_file) { $_tempFileName = STORE_PATH.'temp.tmp'; $content = ''; try { if (ftp_get ( $conn_id, $_tempFileName, $server_file, FTP_ASCII )) { $content = file_get_contents ( $_tempFileName ); } } catch (Exception $e) { var_dump('error timeout:-----'); global $ftp_server; $conn_id = ftp_connect ( $ftp_server ); if (@ftp_login ( $conn_id, $ftp_user, $ftp_pass )) { echo "Connected as $ftp_user@$ftp_server\n"; } else { echo "Couldn't connect as $ftp_user\n"; } ftp_pasv($conn_id, true); if (ftp_get ( $conn_id, $_tempFileName, $server_file, FTP_ASCII )) { $content = file_get_contents ( $_tempFileName ); } } $content = iconv ( "GBK", "UTF-8//IGNORE", trim($content) ); echo "\n"; var_dump('file name :'.$server_file.';content:'.$content); return $content; } function getDirNameFormFtp($conn_id, $dirString) { $buff = ftp_rawlist ( $conn_id, $dirString ); $dirs = array_filter ( $buff, "dirfilter" ); foreach ( $dirs as $key => $value ) { $temp = preg_split ( '/<DIR>/', $value ); $dirs [$key] = trim ( $temp [1] ); } asort($dirs , SORT_NUMERIC); return $dirs; } function dirfilter($var) { return (strstr ( $var, '<DIR>' )); } ftp_close ( $conn_id );