lyuehh
1/21/2014 - 9:55 AM

a.js

#!/bin/sh

for i in `seq 2003 2014`
do
    u="http://www.3dcp.cn/zs/gonggao.php?type=ssq&year=$i"
    echo $u
    curl -s $u | iconv -f gbk -t utf8  | hquery -p -r '$("table").eq(13).parent().html()' > $i.html
done
#!/bin/sh

for i in `seq 2003 2014`
do
    echo $i
    cat ${i}.html | hquery -p -f a.js | sed '$d' > json/${i}.json
done
a = _.reduce($("tr"), function(memo, v, i) {
    var $el = $(v);
    var obj = {};
    obj.id = $.trim($el.find('.end').text());
    obj.date = $el.find('td').eq(1).text();
    obj.red = $el.find('.STYLE13').html() && $el.find('.STYLE13').html().replace(/ /g, ' ');
    obj.blue = $el.find('.STYLE12').text();
    if (obj.id !== '') {
        memo.push(obj);
    }
    return memo;
}, []);
console.log(JSON.stringify(a, null, 2));