用groovy写抓票程序

用groovy写抓票程序 - 疯狂的菠菜 - ITeye技术网站

用groovy写抓票程序

博客分类：
• Groovy

GroovyGoogleHTML

年底了能买到火车票是非常幸运的事儿, 比如我同事, 通过电话就订到了车票, 而我死活都没打进那个电话.



于是用groovy写了个程序, 用来抓取火车票信息, 网上相关的程序还不少, 我只是用groovy来练练手而已, 本来可以完善一下, 像这个(http://www.cnblogs.com/guozili/archive/2011/01/19/1939157.html)可以从多个网站抓取, 像这个(http://www.notedit.com/2010/11/%E6%8A%A2%E7%81%AB%E8%BD%A6%E7%A5%A8%E7%9A%84%E7%A8%8B%E5%BA%8F/)可以定时抓取, 本来我想通过定时抓取发消息的, 后来搞到了票, 就这样吧.

class GetTicket {
    final static String host = "http://hz.58.com/huochepiao/?StartStation=%25u676D%25u5DDE&EndStation=%25u5B9C%25u660C"
    // 最早发车时间
    final static int earliest = 120
    // 已经确认无票的过滤掉
    final static List filterList = [
    "http://hz.58.com/huochepiao/4538967059457x.shtml",
    "http://hz.58.com/huochepiao/4536633437697x.shtml"
    ]

    def void get() {
        def htmlSource =  new Http().get(host).source.toString()
        int i = 0
        LinkedList<Entry> list = [] as LinkedList<Entry>;
        htmlSource.eachLine{
            if (i > 0 && i <= 4) {
                switch(i) {
                    case 1:
                        list[list.size()-1].location = it.trim()
                        break;
                    case 2:
                        list[list.size()-1].number = it.trim()
                        break;
                    case 3:
                        list[list.size()-1].type = it.trim()
                        break;
                    case 4:
                        def matcher = it.trim() =~ /(.+)<\/a>/
                        def pair = matcher[0][1].split(" ")
                        pair[1] = pair[1].replaceAll(/月|日/, "")
                        list[list.size()-1].count = pair[0]
                        list[list.size()-1].date = pair[1]
                        if (Integer.valueOf(pair[1]) < earliest) {
                            list.removeLast()
                        }
                        break;
                }
                i++
                return;
            }else {
                i = 0;
            }

            if (it ==~ /^ +<a href="http:\/\/hz\.58\.com\/huochepiao.+/){
                def matcher = it =~ /"(http:\/\/hz\.58\.com\/huochepiao.+?)"/
                def url = matcher[0][1].trim()
                if (filterList.contains(url)) {
                    return;
                }
                Entry entry = [:] as Entry
                entry.url = url
                list << entry
                i++

                // 临近站信息
                matcher = it =~ /.+>(.+)$/
                if (matcher.matches()) {
                    entry.location = matcher[0][1].trim()
                    i++
                }
            }
        }

        list = list.sort()

        list.each{ println "${it.date}\t${it.count}\t ${it.type}\t ${it.number}\t ${it.location}\t ${it.url}" }
    }
}
class Entry implements Comparable{
    def url
    def location
    def number
    def type
    def count
    def date
    int compareTo( def other) {
        return Integer.valueOf(other.date) - Integer.valueOf(date)
    }

    @Override
    public String toString() {
        return ToStringBuilder.reflectionToString(this);
    }
}

class GetTicket {
    final static String host = "http://hz.58.com/huochepiao/?StartStation=%25u676D%25u5DDE&EndStation=%25u5B9C%25u660C"
    // 最早发车时间
    final static int earliest = 120
    // 已经确认无票的过滤掉
    final static List filterList = [
    "http://hz.58.com/huochepiao/4538967059457x.shtml",
    "http://hz.58.com/huochepiao/4536633437697x.shtml"
    ]

    def void get() {
        def htmlSource =  new Http().get(host).source.toString()
        int i = 0
        LinkedList<Entry> list = [] as LinkedList<Entry>;
        htmlSource.eachLine{
            if (i > 0 && i <= 4) {
                switch(i) {
                    case 1:
                        list[list.size()-1].location = it.trim()
                        break;
                    case 2:
                        list[list.size()-1].number = it.trim()
                        break;
                    case 3:
                        list[list.size()-1].type = it.trim()
                        break;
                    case 4:
                        def matcher = it.trim() =~ /(.+)<\/a>/
                        def pair = matcher[0][1].split(" ")
                        pair[1] = pair[1].replaceAll(/月|日/, "")
                        list[list.size()-1].count = pair[0]
                        list[list.size()-1].date = pair[1]
                        if (Integer.valueOf(pair[1]) < earliest) {
                            list.removeLast()
                        }
                        break;
                }
                i++
                return;
            }else {
                i = 0;
            }

            if (it ==~ /^ +<a href="http:\/\/hz\.58\.com\/huochepiao.+/){
                def matcher = it =~ /"(http:\/\/hz\.58\.com\/huochepiao.+?)"/
                def url = matcher[0][1].trim()
                if (filterList.contains(url)) {
                    return;
                }
                Entry entry = [:] as Entry
                entry.url = url
                list << entry
                i++

                // 临近站信息
                matcher = it =~ /.+>(.+)$/
                if (matcher.matches()) {
                    entry.location = matcher[0][1].trim()
                    i++
                }
            }
        }

        list = list.sort()

        list.each{ println "${it.date}\t${it.count}\t ${it.type}\t ${it.number}\t ${it.location}\t ${it.url}" }
    }
}
class Entry implements Comparable{
    def url
    def location
    def number
    def type
    def count
    def date
    int compareTo( def other) {
        return Integer.valueOf(other.date) - Integer.valueOf(date)
    }

    @Override
    public String toString() {
        return ToStringBuilder.reflectionToString(this);
    }
}

注: 解析html用到了com.googlecode.groovyhttp.Http 这个lib(http://code.google.com/p/groovy-http/)



输出结果:

引用

131 1张 硬卧 K529 杭州 - 恩施 http://hz.58.com/huochepiao/4555300451203x.shtml

129 1张 硬卧 K529 杭州 - 荆门 http://hz.58.com/huochepiao/4547173468803x.shtml

128 1张 K529 杭州 - 恩施 http://hz.58.com/huochepiao/4547524308355x.shtml

124 1张 硬座 K253 杭州 - 宜昌 http://hz.58.com/huochepiao/4557214747137x.shtml

123 1张 硬卧 K253 杭州南 - 宜昌 http://hz.58.com/huochepiao/4557440945411x.shtml

123 1张 站票 K529 杭州 - 宜昌 http://hz.58.com/huochepiao/4532977245187x.shtml

122 1张 硬座 K253 杭州南 - 宜昌 http://hz.58.com/huochepiao/4557377085571x.shtml

122 1张 硬座 K253 杭州南 - 宜昌 http://hz.58.com/huochepiao/4544944871170x.shtml