public async Task CoordTest()
{
using (Index index = new Index())
{
index.WorkPath = "CoordTest\\";
int N = 1000;
index.AddDataSource(new TestDataSource(N));
index.Stemmers.Clear();
await index.CreateAsync();
Assert.False(index.IsCreating);
Assert.True(index.CanSearch);
string[] words = { "and", "tupman", "everybody", "old" };
// coordinate test
foreach (string word in words)
{
List <int> pos = new List <int>();
foreach (Match m in Regex.Matches(TestText1.ToLower(), @"\b" + word + @"\b"))
{
pos.Add(m.Index);
}
Index.SearchResult res = index.Search(word);
Assert.Equal(2 * N, res.foundPages.Count);
Assert.Equal(pos.Count, res.foundPages[0].pos.Count);
foreach (var p in res.foundPages)
{
Assert.True(pos.SequenceEqual(p.pos));
}
}
}
Directory.Delete("CoordTest\\", true);
}
public void BuilderTest()
{
Vocab voc = new Vocab();
voc.Add("and", 1);
voc.Add("end", 3);
voc.Add("old", 2);
voc.Add("the", 6);
voc.Add("them", 5);
voc.Add("then", 4);
voc.Range = ('a', 'z');
voc.Name = "en";
Index.Builder bldr = new Index.Builder("BuilderTest").AddVoc(voc);
bldr.AddDoc("A", "");
string[] words = { "and", "tupman", "everybody", "old" };
List <int>[] pos = new List <int> [words.Length];
for (int q = 0; q < pos.Length; q++)
{
pos[q] = new List <int>();
}
foreach (Match match in Regex.Matches(TestText1.ToLower(), @"\b\w+\b"))
{
for (int q = 0; q < pos.Length; q++)
{
if (match.Value.Equals(words[q]))
{
pos[q].Add(match.Index);
}
}
bldr.AddWord(match.Value, (ulong)match.Index);
}
bldr.EndPage("1");
using (Index index = bldr.Build())
{
for (int q = 0; q < pos.Length; q++)
{
Index.SearchResult res = index.Search(words[q]);
Assert.NotNull(res);
Assert.Equal(1, res.foundPages.Count);
Assert.True(pos[q].SequenceEqual(res.foundPages[0].pos));
}
}
Directory.Delete("BuilderTest", true);
}
public async Task PageTextTest()
{
int N = 10;
SamePageDataSource ds = new SamePageDataSource(N);
using (Index index = new Index("PageTextTest"))
{
index.AddDataSource(ds);
await index.CreateAsync();
Index.SearchResult res = index.Search("\"old lady\"");
Assert.Contains(Index.SearchResult.BEGIN_MATCHED_SYMBOL + "old" + Index.SearchResult.END_MATCHED_SYMBOL, res.foundPages[0].text);
res = index.Search("Test");
Assert.Contains(Index.SearchResult.BEGIN_MATCHED_SYMBOL + "Test" + Index.SearchResult.END_MATCHED_SYMBOL, res.foundDocs.First().headers["Name"]);
}
Directory.Delete("PageTextTest", true);
}
public async Task RequestSyntaxTest()
{
using (Index index = new Index())
{
index.WorkPath = "RequestSyntaxTest\\";
//index.Stemmers.Clear();
int Npages = 100;
index.AddDataSource(new TestDataSource(Npages));
await index.CreateAsync();
Assert.False(index.IsCreating);
Assert.True(index.CanSearch);
Index.SearchResult res = index.Search("and (tupman|old)" /*, new Index.SearchOptions(){ dist = 20}*/);
Assert.Equal(2, res.foundDocs.Count);
Assert.Equal(Npages, res.foundDocs.First().pages.Count);
Assert.Equal(Npages, res.foundDocs.Last().pages.Count);
res = index.Search("and (tupman|old) {Name=Dump}" /*, new Index.SearchOptions(){ dist = 20}*/);
Assert.Equal(1, res.foundDocs.Count);
Assert.Equal(Npages, res.foundDocs.First().pages.Count);
Assert.Equal(res.foundPages[0].pos.Count, res.foundPages[1].pos.Count);
for (int q = 0; q < 2 * Npages; q++)
{
Assert.Equal(42, res.foundPages[q].pos.Count);
}
Assert.True(Enumerable.SequenceEqual(res.foundPages[0].pos, res.foundPages[1].pos));
res = index.Search("lady old", new Index.SearchOptions()
{
dist = 40
});
foreach (int pos in res.foundPages[0].pos)
{
Console.WriteLine("" + pos + ":" + TestText1.Substring(pos, 10));
}
Assert.Equal(10, res.foundPages[0].pos.Count);
res = index.Search("\"lady\" old", new Index.SearchOptions()
{
dist = 40
});
Assert.Equal(8, res.foundPages[0].pos.Count);
res = index.Search("\"old lady\"", new Index.SearchOptions()
{
dist = 40
});
Assert.Equal(8, res.foundPages[0].pos.Count);
res = index.Search("\"lady old\"", new Index.SearchOptions()
{
dist = 40
});
Assert.Equal(0, res.foundPages.Count);
Assert.Equal(12, index.Search("lady (old | young)", new Index.SearchOptions()
{
dist = 40
}).foundPages[0].pos.Count);
Assert.Equal(3, index.Search("\"old ladies were\"", new Index.SearchOptions()
{
dist = 40
}).foundPages[0].pos.Count);
res = index.Search("\"old lady were\"", new Index.SearchOptions()
{
dist = 40
});
Assert.Equal(0, res.foundPages.Count);
res = index.Search("\"old (lady|ladies) (who|were|looked)\"", new Index.SearchOptions()
{
dist = 40
});
//foreach (int pos in res.foundPages[0].pos)
// Console.WriteLine(""+pos+":"+TestText1.Substring(pos,10));
Assert.Equal(9, res.foundPages[0].pos.Count);
res = index.Search("?an?", new Index.SearchOptions()
{
dist = 40
});
Assert.Equal(Regex.Matches(TestText1.ToLower(), @"\w*an\w*").Count, res.foundPages[0].pos.Count);
}
Directory.Delete("RequestSyntaxTest\\", true);
}
