extract urls

This commit is contained in:
Nicolas Constant 2020-08-01 13:56:59 -04:00
parent 16d310a37e
commit d33cbbfb1c
No known key found for this signature in database
GPG Key ID: 1E9F677FB01A5688
2 changed files with 108 additions and 0 deletions

View File

@ -18,6 +18,7 @@ namespace BirdsiteLive.Domain.Tools
private readonly Regex _mentionRegex = new Regex(@"\W(\@[a-zA-Z0-9_ー]+\b)(?!;)");
//private readonly Regex _mentionRegex = new Regex(@"(?<=[\s>]|^)@(\w*[a-zA-Z0-9_ー]+\w*)\b(?!;)");
//private readonly Regex _mentionRegex = new Regex(@"(?<=[\s>]|^)@(\w*[a-zA-Z0-9_ー]+)\b(?!;)");
private readonly Regex _urlRegex = new Regex(@"((http|ftp|https):\/\/[\w\-_]+(\.[\w\-_]+)+([\w\-\.,@?^=%&amp;:/~\+#]*[\w\-\@?^=%&amp;/~\+#])?)");
private readonly InstanceSettings _instanceSettings;
#region Ctor
@ -32,6 +33,38 @@ namespace BirdsiteLive.Domain.Tools
var tags = new List<Tag>();
messageContent = $" {messageContent} ";
// Extract Urls
var urlMatch = _urlRegex.Matches(messageContent);
foreach (var m in urlMatch)
{
var url = m.ToString().Replace("\n", string.Empty).Trim();
var protocol = "https://";
if (url.StartsWith("http://")) protocol = "http://";
else if (url.StartsWith("ftp://")) protocol = "ftp://";
var truncatedUrl = url.Replace(protocol, string.Empty);
if (truncatedUrl.StartsWith("www."))
{
protocol += "www.";
truncatedUrl = truncatedUrl.Replace("www.", string.Empty);
}
var firstPart = truncatedUrl;
var secondPart = string.Empty;
if (truncatedUrl.Length > 30)
{
firstPart = truncatedUrl.Substring(0, 30);
secondPart = truncatedUrl.Substring(30);
}
messageContent = Regex.Replace(messageContent, m.ToString(),
$@" <a href=""{url}"" rel=""nofollow noopener noreferrer"" target=""_blank""><span class=""invisible"">{protocol}</span><span class=""ellipsis"">{firstPart}</span><span class=""invisible"">{secondPart}</span></a>");
}
// Extract Hashtags
var hashtagMatch = _hastagRegex.Matches(messageContent);
foreach (var m in hashtagMatch)
{
@ -49,6 +82,7 @@ namespace BirdsiteLive.Domain.Tools
$@" <a href=""{url}"" class=""mention hashtag"" rel=""tag"">#<span>{tag}</span></a>");
}
// Extract Mentions
var mentionMatch = _mentionRegex.Matches(messageContent);
foreach (var m in mentionMatch)
{

View File

@ -22,6 +22,80 @@ namespace BirdsiteLive.Domain.Tests.Tools
}
#endregion
[TestMethod]
public void Extract_FormatUrl_Test()
{
#region Stubs
var message = $"Bla!{Environment.NewLine}https://t.co/L8BpyHgg25";
#endregion
var service = new StatusExtractor(_settings);
var result = service.ExtractTags(message);
#region Validations
Assert.AreEqual(0, result.tags.Length);
Assert.IsTrue(result.content.Contains("Bla!"));
Assert.IsTrue(result.content.Contains(@"<a href=""https://t.co/L8BpyHgg25"" rel=""nofollow noopener noreferrer"" target=""_blank""><span class=""invisible"">https://</span><span class=""ellipsis"">t.co/L8BpyHgg25</span><span class=""invisible""></span></a>"));
#endregion
}
[TestMethod]
public void Extract_FormatUrl_Long_Test()
{
#region Stubs
var message = $"Bla!{Environment.NewLine}https://www.eff.org/deeplinks/2020/07/pact-act-not-solution-problem-harmful-online-content";
#endregion
var service = new StatusExtractor(_settings);
var result = service.ExtractTags(message);
#region Validations
Assert.AreEqual(0, result.tags.Length);
Assert.IsTrue(result.content.Contains("Bla!"));
Assert.IsTrue(result.content.Contains(@"<a href=""https://www.eff.org/deeplinks/2020/07/pact-act-not-solution-problem-harmful-online-content"" rel=""nofollow noopener noreferrer"" target=""_blank""><span class=""invisible"">https://www.</span><span class=""ellipsis"">eff.org/deeplinks/2020/07/pact</span><span class=""invisible"">-act-not-solution-problem-harmful-online-content</span></a>"));
#endregion
}
[TestMethod]
public void Extract_FormatUrl_Exact_Test()
{
#region Stubs
var message = $"Bla!{Environment.NewLine}https://www.eff.org/deeplinks/2020/07/pact";
#endregion
var service = new StatusExtractor(_settings);
var result = service.ExtractTags(message);
#region Validations
Assert.AreEqual(0, result.tags.Length);
Assert.IsTrue(result.content.Contains("Bla!"));
Assert.IsTrue(result.content.Contains(@"<a href=""https://www.eff.org/deeplinks/2020/07/pact"" rel=""nofollow noopener noreferrer"" target=""_blank""><span class=""invisible"">https://www.</span><span class=""ellipsis"">eff.org/deeplinks/2020/07/pact</span><span class=""invisible""></span></a>"));
#endregion
}
[TestMethod]
public void Extract_MultiUrls__Test()
{
#region Stubs
var message = $"https://t.co/L8BpyHgg25 Bla!{Environment.NewLine}https://www.eff.org/deeplinks/2020/07/pact-act-not-solution-problem-harmful-online-content";
#endregion
var service = new StatusExtractor(_settings);
var result = service.ExtractTags(message);
#region Validations
Assert.AreEqual(0, result.tags.Length);
Assert.IsTrue(result.content.Contains("Bla!"));
Assert.IsTrue(result.content.Contains(@"<a href=""https://t.co/L8BpyHgg25"" rel=""nofollow noopener noreferrer"" target=""_blank""><span class=""invisible"">https://</span><span class=""ellipsis"">t.co/L8BpyHgg25</span><span class=""invisible""></span></a>"));
Assert.IsTrue(result.content.Contains(@"<a href=""https://www.eff.org/deeplinks/2020/07/pact-act-not-solution-problem-harmful-online-content"" rel=""nofollow noopener noreferrer"" target=""_blank""><span class=""invisible"">https://www.</span><span class=""ellipsis"">eff.org/deeplinks/2020/07/pact</span><span class=""invisible"">-act-not-solution-problem-harmful-online-content</span></a>"));
#endregion
}
[TestMethod]
public void Extract_SingleHashTag_Test()
{