PHP getHtmlData Examples

Example #1

File: index.php Project: BigBin/web_crawler

<?php

require_once '../apps/conf_ini.php';
$oParser = new HtmlParse();
// アクセスする URL を指定
$URL = 'http://perldoc.jp/docs/modules/DBD-mysql-2.1026/DBD/mysql/INSTALL.pod';
$sHtml = getHtmlData($URL);
//UTF-8にエンコード
$enc = mb_detect_encoding($sHtml);
$sHtml = mb_convert_encoding($sHtml, "UTF-8", $enc);
//解析
$rtn = $oParser->execHtmlParse($sHtml);
if ($rtn !== false) {
    list($aSubject, $aUrl, $aNonTagHtml) = $rtn;
    $fp = fopen("C:/body.txt", 'w');
    $sLine = implode('', $aNonTagHtml);
    fwrite($fp, mb_convert_encoding($sLine, "SJIS", "UTF-8"));
    $flag = fclose($fp);
}

Example #2

File: get_html_data.php Project: BigBin/web_crawler

<?php

require_once '../apps/conf_ini.php';
require_once 'apps/class/Process/CollectHtml/CollectHtmlDao_cls.php';
// 初期化
$oDb = new CollectHtmlDao($oDbConnMng);
// HTML取得先URLリスト取得
$aUrlHeader = $oDb->getUrl();
while ($aUrlHeader && count($aUrlHeader) != 0) {
    $aHtml = array();
    $aOkUrlId = array();
    $aErrUrlId = array();
    for ($iCnt = 0; $iCnt < count($aUrlHeader); $iCnt++) {
        // アクセスする URL を指定
        echo $aUrlHeader[$iCnt]['url'] . "\n";
        $sHtml = getHtmlData($aUrlHeader[$iCnt]['url']);
        if (!$sHtml) {
            $aErrUrlId[] = $aUrlHeader[$iCnt]['url_id'];
            continue;
        }
        // UTF-8にエンコード
        $enc = mb_detect_encoding($sHtml);
        $sHtml = mb_convert_encoding($sHtml, "UTF-8", $enc);
        //
        $aHtml[] = array('url_id' => $aUrlHeader[$iCnt]['url_id'], 'html' => $sHtml);
        $aOkUrlId[] = $aUrlHeader[$iCnt]['url_id'];
    }
    $oDb->setHtml($aHtml);
    $oDb->updateUrlStatus($aOkUrlId, URL_STATUS_END);
    if (count($aErrUrlId) != 0) {
        $oDb->updateUrlStatus($aErrUrlId, URL_STATUS_FAILED);

Example #3

File: action.php Project: smc0210/michang

      $add_data_arr['meeting'] = $meeting;
      $add_data_arr['fuelcharge'] = $fuelcharge;
      $add_data_arr['airport'] = $airport;
      $add_data_arr['insurance'] = $insurance;
      $add_data_arr['passportvisa'] = $passportvisa;
      
      $file_path = $baseDir."board/boardSkin/s_nowcroatia/";
      $schedule = '<TABLE cellSpacing=0 cellPadding=0 width=780 border=0>';
      $schedule .= '<TBODY><TR><TD><IMG ';
      $schedule .= 'src="/board/boardSkin/s_nowcroatia/images/view/head_sch.gif"';
      $schedule .= 'width=780 height=31></TD></TR></TBODY></TABLE>';
      
      for($i=1; $i<=$days; $i++)
              $schedule .= getHtmlData($file_path."schedule.htm");
      
      $check_point = getHtmlData($file_path."check_point.htm");
      
      $add_data_arr['schedule'] = $schedule;
      $add_data_arr['check_point'] = $check_point;
      
      $board_control->update_add_data($insertId, $add_data_arr, $title_en);
 }
 elseif($boardSetting['template'] == 'event')
 {
      $add_data_arr['start_date'] = $start_date;
      $add_data_arr['end_date'] = $end_date;
      
      $board_control->update_add_data($insertId, $add_data_arr, $title_en);
 }
 elseif($boardSetting['template'] == 'calendar')
 {