nodejs 扒页面 api-docs
var continuation = require('../continuation');var http = require("http");
var fs = require('fs');
var url = require('url');
var ps = require("path");
//var request = require("request");
function main_entry() {
var $baseurl = "http://directwebremoting.org/dwr/javadoc/";
var $basepath = "E:\\dwrdoc\\";
var $lefttopurl = "http://directwebremoting.org/dwr/javadoc/overview-frame.html";
var getLocal = function(str) {
str = str.replace($baseurl , "");
str = str.replace(/^\// , "");
if(/^http\:/.test(str)) {
return $basepath;
}
var local = ps.resolve($basepath , str);
var dirname = ps.dirname(local);
console.log(local + "," + dirname);
mkdir(dirname);
return local;
};
var mkdir = function(str) {
var arr = str.split(/\\/g);
var step = "";
while(arr.length) {
var shift = arr.shift();
step += shift + "\\";
if(!fs.existsSync(step)) {
fs.mkdirSync(step , 0777);
}
}
};
var getLinks = function(str , $base) {
$base = $base || $baseurl;
var reg = /<a.*?href="(.+?)".*?>.*?<\/a>/gi , result;
var list = [];
while((result = reg.exec(str)) != null) {
var location = url.resolve($base , result).replace(/\?.*$/ , "");
if(!/#.*$/.test(location)) {
list.push(location);
}
}
var reg = /<frame.*?src="(.+?)".*?>/gi;
while((result = reg.exec(str)) != null) {
var location = url.resolve($base , result).replace(/\?.*$/ , "");
if(!/^#/.test(location)) {
list.push(location);
}
}
/*
var reg = /<img.*?src="(.+?)".*?>/gi;
while((result = reg.exec(str)) != null) {
var location = url.resolve($base , result).replace(/\?.*$/ , "");
var local = getLocal(location);
console.log(location + "............");
console.log(local + "...............");
http.request(location).pipe(fs.createWriteStream(local));
}
*/
return list;
};
var $httpget = function(url , next) {
//console.log('正在处理 ' + url);
if(url.indexOf("mailto:") == 0) {
return "";
}
var result = [];
http.get(url , cont(res));
res.on("data" , function(data) {
result.push(data.toString());
});
res.on("end" , cont(callback));
next(null , result.join(''));
};
$indexpath = $basepath + "index.html";
if(fs.existsSync($indexpath)) {
return;
}
$httpget($baseurl , obtain(result));
fs.writeFileSync($indexpath , result , "utf-8");
var listO = getLinks(result);
listO.forEach(function(item) {
var localPathO = getLocal(item);
if(fs.existsSync(localPathO)) {
return;
}
$httpget(item , obtain(resultO));
fs.writeFileSync(localPathO , resultO , "utf-8");
});
$httpget($lefttopurl , obtain(result1));
var list1 = getLinks(result1);
list1.forEach(function(item) {
var localPath = getLocal(item);
if(fs.existsSync(localPath)) {
return;
}
$httpget(item , obtain(result1));
fs.writeFileSync(localPath , result1 , "utf-8");
var $basedir = url.resolve(item , "./");
var list2 = getLinks(result1 , $basedir);
list2.forEach(function(item1) {
var localPath1 = getLocal(item1);
if(fs.existsSync(localPath1)) {
return;
}
$httpget(item1 , obtain(result2));
fs.writeFileSync(localPath1 , result2 , 'utf-8');
return;
/*
var list3 = getLinks(result2 , $basedir);
list3.forEach(function(item2) {
var localPath2 = getLocal(item2);
if(fs.existsSync(localPath2)) {
return;
}
$httpget(item2 , obtain(result3));
fs.writeFileSync(localPath2 , result3 , 'utf-8');
});
*/
});
});
};
var code = main_entry.toString();
var compiledCode = continuation.compile(code);
eval(compiledCode);
main_entry();
页:
[1]