public void parse_twitter()
{
const string address = "http://twitter.com/skysportsfl/followers/";
var webClient = new WebClient();
webClient.Headers.Add("cookie",
"lang=en-gb; guest_id=v1%3A137337600960092466; __utma=43838368.517491664.1373376012.1375298648.1375367885.10; __utmz=43838368.1375298648.9.8.utmcsr=myxi.co|utmccn=(referral)|utmcmd=referral|utmcct=/; __utmv=43838368.lang%3A%20en; twll=l%3D1375367981; remember_checked=1; remember_checked_on=1; external_referer=\"OTZIBTkFw3v0VTPNOWsL6Q==|0\"; __utmb=43838368.1.10.1375367885; auth_token=4a3b1bfbf0a104e0e01d4f884eb5b86b68972515; secure_session=true; _twitter_sess=BAh7CiIKZmxhc2hJQzonQWN0aW9uQ29udHJvbGxlcjo6Rmxhc2g6OkZsYXNo%250ASGFzaHsABjoKQHVzZWR7ADoPY3JlYXRlZF9hdGwrCKCEVzpAAToOcmV0dXJu%250AX3RvIi5odHRwczovL3R3aXR0ZXIuY29tL1NreVNwb3J0c0ZML2ZvbGxvd2Vy%250AczoHaWQiJTMzOTMxODA3OTllOGRjYTI3OGRlZTg2MDg0OThmZDg0Ogxjc3Jm%250AX2lkIiUyNjk3MWZkODY5ODg4MzBiZmQyNWJlNzYxODA0M2U3MA%253D%253D--1e1015aadd109ffcb0d3848ea85e54b0843c94f7");
var webPage = webClient.DownloadString(address);
var htmlParser = HtmlParser.Create();
var html = htmlParser.Parse(webPage);
Assert.True(webPage.Contains("Followers"));
var names = html
.HasClass("stream-container")
.HasClass("username")
.Where(n => n.ToText() != "@");
foreach (var name in names)
{
Console.WriteLine(name.ToText());
}
}
public void br_gives_a_newline()
{
const string htmlString =
"<span>Line1<br>Line2</span>";
var htmlParser = HtmlParser.Create();
var html = htmlParser.Parse(htmlString);
var result = html.ToText();
Assert.Equal(2, result.Split('\n').Count());
}
public static void RegisterKeywordIndexing(IKeywordsIndexer indexer)
{
var htmlParser = HtmlParser.Create();
indexer
.Entity <BlogEntryData>()
.Index(entry => entry.Title)
.Index(entry => entry.Author)
.Index(entry => htmlParser.Parse(entry.Summary).ToText())
.Index(entry => htmlParser.Parse(entry.Content).ToText())
.ForEach(entry => entry.Tags,
tagBuilder => tagBuilder.Index(tag => tag.Title));
}
static void Exec(
string htmlString,
string name,
string value,
string expected)
{
var sut = HtmlParser.Create();
var html = sut.Parse(htmlString).ToArray();
html.Attr(name, value);
Assert.Equal(expected, html.ToHtml());
}
public void inline_elements_on_same_line()
{
const string htmlString =
"<span>Line1 <b>Line2</b></span>";
var htmlParser = HtmlParser.Create();
var html = htmlParser.Parse(htmlString);
var result = html.ToText();
Console.Write(result);
Assert.Equal(1, result.Split('\n').Count());
}
static IEnumerable <string> Exec(
string htmlString,
string attributeName,
params string[] expected)
{
var sut = HtmlParser.Create();
var html = sut.Parse(htmlString).ToArray();
var result = html.Attr(attributeName).ToArray();
Assert.Equal(expected, result);
return(result);
}
static IEnumerable <IHtmlNode> Exec(
string htmlString,
Func <HtmlElement, bool> isMatch,
string expected)
{
var sut = HtmlParser.Create();
var html = sut.Parse(htmlString).ToArray();
var result = html.Query(isMatch).ToArray();
Assert.Equal(expected, result.ToHtml());
return(result);
}
static IEnumerable <IHtmlNode> Exec(
string html,
int expectedCount = 0)
{
var sut = HtmlParser.Create();
var result = sut.Parse(html).ToArray();
Assert.NotNull(result);
Console.Write(result.ToHtml());
Assert.Equal(expectedCount, result.Count());
return(result);
}
