public void TestGetLastMessageText_NoPrevious()
{
string original = @"<html>
<body>
This is a boring email.
</body>
</html>";
// Note: it's acceptable to not preserve whitespace / insert empty HTML tags because we're
// manipulating HTML, not plain text. As long as the rendered page isn't impacted, all is well
string expected = @"<html><head></head><body>
This is a boring email.
</body></html>";
string actual = EmailBodyProcessingUtils.GetLastMessageText_Html(original);
Assert.AreEqual(Normalize(expected), Normalize(actual));
}
public void TestGetLastMessageText_Previous_FromColon()
{
string original = @"<html>
<body>
<div class=""wrapppingBothMessageAndReply"">
<p class=""customStyling"">This is random text with some custom styling and outlook-generated elements.<o:p></o:p></p>
<div>
<div style=""border:none;border-top:solid"">
<p class=""customStyling""><b>From:</b> someAddress;<br><b>Subject:</b> RE: Build error<o:p></o:p>
</p>
</div>
</div>
<p class=""customStyling"">
<o:p> </o:p>
</p>
<p class=""customStyling"">text of the reply
</p>
</div>
<div>Something after the containing div</div>
</body>
</html>";
// Note: it's acceptable to not preserve whitespace because it's
// manipulating HTML, not plain text. As long as the rendered page isn't impacted, all is well
// Note that we expect that both
// 1. Elements following the latest message are removed
// 2. Anything in the same element as the latest message but after the start of the previous should be cleared out
string expected = @"<html><head></head><body>
<div class=""wrapppingBothMessageAndReply"">
<p class=""customStyling"">This is random text with some custom styling and outlook-generated elements.<o:p></o:p></p>
<div>
<div style=""border:none;border-top:solid"">
<p class=""customStyling""><b></b></p></div></div></div></body></html>";
string actual = EmailBodyProcessingUtils.GetLastMessageText_Html(original);
Assert.AreEqual(Normalize(expected), Normalize(actual));
}
public void TestGetLastMessageText_EmailClientsSchemas()
{
const string schemasFolder = "LastMessageSchemas";
foreach (var originalFilename in Directory.GetFiles(schemasFolder, "*.orig"))
{
Trace.WriteLine(string.Format("Processing email schema file {0}", originalFilename));
var baseFilename = Path.GetFileNameWithoutExtension(originalFilename);
var expectedFilename = Path.Combine(schemasFolder, baseFilename + ".expected");
var original = File.ReadAllText(originalFilename);
var expected = Normalize(File.ReadAllText(expectedFilename));
var actual = Normalize(EmailBodyProcessingUtils.GetLastMessageText_Html(original));
// Note: it's acceptable to not preserve whitespace because it's
// manipulating HTML, not plain text. As long as the rendered page isn't impacted, all is well
// Note that we expect that both
// 1. Elements following the latest message are removed
// 2. Anything in the same element as the latest message but after the start of the previous should be cleared out
Assert.AreEqual(expected, actual);
}
}