public void TestForever()
{
String docs = "<DOC>\r\n" +
"<DOCNO>TEST-000</DOCNO>\r\n" +
//"<docno>TEST-000</docno>\r\n" +
"<DOCHDR>\r\n" +
"http://lucene.apache.org.trecdocmaker.test\r\n" +
"HTTP/1.1 200 OK\r\n" +
"Date: Sun, 11 Jan 2009 08:00:00 GMT\r\n" +
"Server: Apache/1.3.27 (Unix)\r\n" +
"Last-Modified: Sun, 11 Jan 2009 08:00:00 GMT\r\n" +
"Content-Length: 614\r\n" +
"Connection: close\r\n" +
"Content-Type: text/html\r\n" +
"</DOCHDR>\r\n" +
"<html>\r\n" +
"\r\n" +
"<head>\r\n" +
"<title>\r\n" +
"TEST-000 title\r\n" +
"</title>\r\n" +
"</head>\r\n" +
"\r\n" +
"<body>\r\n" +
"TEST-000 text\r\n" +
"\r\n" +
"</body>\r\n" +
"\r\n" +
"</DOC>";
StringableTrecSource source = new StringableTrecSource(docs, true);
source.SetConfig(null);
DocData dd = source.GetNextDocData(new DocData());
assertDocData(dd, "TEST-000_0", "TEST-000 title", "TEST-000 text", source
.ParseDate("Sun, 11 Jan 2009 08:00:00 GMT"));
// same document, but the second iteration changes the name.
dd = source.GetNextDocData(dd);
assertDocData(dd, "TEST-000_1", "TEST-000 title", "TEST-000 text", source
.ParseDate("Sun, 11 Jan 2009 08:00:00 GMT"));
source.Dispose();
// Don't test that NoMoreDataException is thrown, since the forever flag is
// turned on.
}
public void TestOneDocument()
{
String docs = "<DOC>\r\n" +
"<DOCNO>TEST-000</DOCNO>\r\n" +
"<DOCHDR>\r\n" +
"http://lucene.apache.org.trecdocmaker.test\r\n" +
"HTTP/1.1 200 OK\r\n" +
"Date: Sun, 11 Jan 2009 08:00:00 GMT\r\n" +
"Server: Apache/1.3.27 (Unix)\r\n" +
"Last-Modified: Sun, 11 Jan 2009 08:00:00 GMT\r\n" +
"Content-Length: 614\r\n" +
"Connection: close\r\n" +
"Content-Type: text/html\r\n" +
"</DOCHDR>\r\n" +
"<html>\r\n" +
"\r\n" +
"<head>\r\n" +
"<title>\r\n" +
"TEST-000 title\r\n" +
"</title>\r\n" +
"</head>\r\n" +
"\r\n" +
"<body>\r\n" +
"TEST-000 text\r\n" +
"\r\n" +
"</body>\r\n" +
"\r\n" +
"</DOC>";
StringableTrecSource source = new StringableTrecSource(docs, false);
source.SetConfig(null);
DocData dd = source.GetNextDocData(new DocData());
assertDocData(dd, "TEST-000_0", "TEST-000 title", "TEST-000 text", source
.ParseDate("Sun, 11 Jan 2009 08:00:00 GMT"));
assertNoMoreDataException(source);
}