Returning links inside multiple-levels iframe usin

2019-07-31 03:24发布

I am trying to get all links under multiple level iframes in CasperJS. There was a solution to resolve the case that there is one-level iframe. I am trying to put getLinksFromIFrame inside the getLinksFromIfram to do recurive traverse but failed.

For this code, how should I do it for multiple-level iframes?

function getLinksFromIframes( callback ) {
    var links = [];

    var iframes = this.evaluate( function() {
        var iframes = [];
        [].forEach.call(document.querySelectorAll("iframe"), function(iframe, i) { iframes.push( i ); });
        return iframes;
    });

    iframes.forEach( function( index ) {
        this.withFrame(index, function() {
            links = links.concat( this.getElementsAttribute( 'a', 'href' ) );
            console.log("works: " + links);
        });
    }, this);

    this.then(function(){
        callback.call(this, links);
    });
}

casper.start(url, function () {
    getLinksFromIframes.call(this, function(links){
        thelinks = links;
        console.log("Links: " + thelinks);
    });
})
.then(function(){
    console.log("Links later: " + thelinks);
})
.run();

1条回答
Summer. ? 凉城
2楼-- · 2019-07-31 04:22

Maybe like this:

var casper = require("casper").create({
        //    verbose: true,
        //    logLevel: "debug",
        webSecurityEnabled : false
    });
//page.onConsoleMessage = function(msg) {console.log(msg);};
casper.on('remote.message', function (message) {
    this.echo(message);
});

casper.start("http://domu-test-2/node/1", function () {
    this.evaluate(function () {
        var i,
        x = document.querySelector("iframe#test") //First iframe
            .contentDocument.querySelector("iframe#test2") //Second iframe in the first
            .contentDocument.querySelectorAll("a"); //Links
        for (i = 0; i < x.length; i++) {
            console.log(x[i].href)
        }
    })
}).wait(1000).run();

It was hard, but i've created this script:

var casper = require("casper").create({
        //  verbose: true,
        //    logLevel: "debug",
        webSecurityEnabled : false
    });
var links = [];
function get_links(obj) {
    return obj.evaluate(function () {
        var i,
            l = document.querySelectorAll("a"),
            l2 = [];
        for (i = 0; i < l.length; i++) {
            l2[i] = l[i].href;
        }
        return l2
    });
}
function unique(arr) {
    var obj = {};
    for (var i = 0; i < arr.length; i++) {
        if (/http(.*)?/.test(arr[i])) {
            var str = arr[i];
            obj[str] = true;
        }
    }
    return Object.keys(obj);
}

function getLinksFromIframes(callback) {
    this.echo("Here we come: " + this.getCurrentUrl() + "\n");
    function to_frame(obj) {
        var iframes = to_evaluate(obj);
        iframes.forEach(function (index) {
            this.withFrame(index, function () {
                this.echo("We are here: " + this.getCurrentUrl());
                var l = unique(get_links(this));
                var i;
                for (i = 0; i < l.length; i++) {
                    console.log(l[i]);
                    links.push(l[i])
                }
                links = unique(links);
                console.log("");
                to_frame(this) //multi lvl
            }); //The first iframe
        }, obj);
    }
    function to_evaluate(obj) {
        return obj.evaluate(function () {
            var iframes = [];
            [].forEach.call(document.querySelectorAll("iframe"), function (iframe, i) {
                iframes.push(i);
            });
            return iframes;
        })
    }
    to_frame(this);
    this.then(function () {
        callback.call(this);
    });
}

casper.start("http://domu-test-2/node/1", function () {
    getLinksFromIframes.call(this, function () {
        console.log("Done!\n");
        var i;
        for (i = 0; i < links.length; i++) {
            console.log(links[i]);
        }
    });
}).then(function () {}).run();

Note:
Now we have a full multi lvl.

./casperjs test.js >>/dev/stdout
Here we come: http://domu-test-2/node/1

We are here: http://domu-test-2/node/2
http://link_1_inside_iframe(1.1)_from_main_frame

We are here: http://domu-test-2/node/3
http://link_1_inside_iframe(2.1)_from_1.1

We are here: http://domu-test-2/node/5
http://link_1_inside_iframe(2.2)_from_1.1

We are here: http://domu-test-2/node/4
http://link_1_inside_iframe(1.2)_from_main_frame

We are here: http://domu-test-2/node/6
http://link_1_inside_iframe(2.1)_from_1.2
查看更多
登录 后发表回答