抓取国家统计局网站上的最新县及县以上行政区划代码（apache camel版），apachecamel,private stat

抓取国家统计局网站上的最新县及县以上行政区划代码（apache camel版），apachecamel,private stat

private static Pattern pattern = Pattern.compile("'(t\\\\d{8}_\\\\d{8,}\\\\.htm)'",Pattern.CASE_INSENSITIVE);    private static Pattern pattern1 = Pattern.compile("<span class=\\"content\\">(.*?)</span>",Pattern.CASE_INSENSITIVE);from("timer://foo?fixedRate=true&delay=0&period=864000")//每隔24小时                .to("http4://www.stats.gov.cn/tjbz/xzqhdm/")//从那个地方                .process(new Processor() {//处理下结果，从中regex出最新数据的url                    @Override                    public void process(Exchange exchange) throws Exception {                        Matcher m = pattern.matcher(exchange.getIn().getBody(String.class));                        if(m.find()){                            in.setHeader("newurl", "http4://www.stats.gov.cn/tjbz/xzqhdm/" + m.group(1));                        }                    }                })                .setHeader(Exchange.HTTP_URI,header("newurl"))                .setProperty(Exchange.CHARSET_NAME, constant("GBK"))                .to("http4://will.be.override")//从最新的url或者最新的数据//              .setBody().javaScript("'abc'")                .process(new Processor() {//再处理下                    @Override                    public void process(Exchange exchange) throws Exception {                        Matcher m = pattern1.matcher(exchange.getIn().getBody(String.class));                        if(m.find()){                            String c = m.group(1);                            c = c.replaceAll("(( )|\\\\s)+", ",");                            String[] ccc = c.split("<BR>");                            StringBuffer sb = new StringBuffer();                            for(String s : ccc){                                sb.append(s);                                sb.append("\\n");                            }                            in.setBody(sb.toString());                        }                    }                })                .setHeader(Exchange.FILE_NAME, constant("m3958.html"))                .to("file:target/m3958");//到文件中//该片段来自于http://byrx.net