/**
* Get all the lesson pages in a term.
* @param $term_no represent the term number (with lesson.)
* @return true;
*/
function get_term_class_pages($term_no)
{
//require_once(dirname(__FILE__)."\inc.php");
set_time_limit(0);
//设置时间无超时停止
get_college_page($term_no);
//抓取学院开课页面
foreach ($college_list as $college) {
$filepl = "college_%s/%s.txt";
$file = sprintf($filepl, $term_no, $college);
$file_name = "resource/" . $file;
$output = file_get_contents($file_name);
//获取学院开课页面源码
$mode = array(FIRST => "tr", SECOND => "td", LAST => "");
$list = get_attr_lists($output, $mode);
//根据学院开课页面源码提取出各种属性
$col_count = count($list[0]);
$row_count = count($list);
//echo $row_count." ".$col_count;
$row = 0;
for ($row = 0; $row < $row_count; $row++) {
//行数
$col = 0;
if ($row == 0 || $row == $row_count - 1) {
continue;
}
for ($col = 0; $col < $col_count; $col++) {
//列数
if ($col == lesson_no) {
//echo $list[$row][$col]." ";
get_class_page($term_no, $list[$row][$col]);
//根据学期号，课程号获取课程页面
break;
}
}
}
}
return true;
}
<?php
if (!defined('ROOT_PATH')) {
define("ROOT_PATH", substr(dirname(__FILE__), 0, -7));
//获取根目录
}
//第一步:先爬取学院网页
require_once ROOT_PATH . "\\Spider\\include\\spider.func.php";
set_time_limit(0);
get_college_page(20131);
//第二步:先将学院_课程数据插入数据库。
/*set_time_limit(0);
insert_college_lesson_in_term("20132");*/
//第三步:将全部课程_学生数据插入数据库。
require_once ROOT_PATH . "\\Spider\\include\\inc.php";
require_once ROOT_PATH . "\\Spider\\include\\spider.func.php";
require_once ROOT_PATH . "\\Spider\\class\\db\\lesson_spider_db.php";
require_once ROOT_PATH . "\\Spider\\class\\db\\database.php";
//使用这种方式包括文件才不会出错
set_time_limit(0);
$db = new cDatabase(DATABASE_INFO_CONSTANTS::$db_info);
$college_lesson_db = new college_lesson_spider_db($db);
//注意这里如果没有关闭的话，后面会出现错误
$lesson_db = new lesson_student_spider_db($db);
//打开课程学生数据库操作类
$log = new myLog();
$log->setMode(0);
//设置日志不输出
$my_dir = "resource/all_info_20132/";
$handler = opendir($my_dir);
while (($filename = readdir($handler)) !== false) {
