public void TestSplitRR()
{
MultiPassIndexSplitter splitter = new MultiPassIndexSplitter();
Directory[] dirs = new Directory[] {
NewDirectory(),
NewDirectory(),
NewDirectory()
};
try
{
splitter.Split(TEST_VERSION_CURRENT, input, dirs, false);
Document doc;
TermsEnum te;
IndexReader ir;
using (ir = DirectoryReader.Open(dirs[0]))
{
assertTrue(ir.NumDocs - NUM_DOCS / 3 <= 1); // rounding error
doc = ir.Document(0);
assertEquals("0", doc.Get("id"));
te = MultiFields.GetTerms(ir, "id").GetIterator(null);
assertEquals(TermsEnum.SeekStatus.NOT_FOUND, te.SeekCeil(new BytesRef("1")));
assertNotSame("1", te.Term.Utf8ToString());
}
using (ir = DirectoryReader.Open(dirs[1]))
{
assertTrue(ir.NumDocs - NUM_DOCS / 3 <= 1);
doc = ir.Document(0);
assertEquals("1", doc.Get("id"));
te = MultiFields.GetTerms(ir, "id").GetIterator(null);
assertEquals(TermsEnum.SeekStatus.NOT_FOUND, te.SeekCeil(new BytesRef("0")));
assertNotSame("0", te.Term.Utf8ToString());
}
using (ir = DirectoryReader.Open(dirs[2]))
{
assertTrue(ir.NumDocs - NUM_DOCS / 3 <= 1);
doc = ir.Document(0);
assertEquals("2", doc.Get("id"));
te = MultiFields.GetTerms(ir, "id").GetIterator(null);
assertEquals(TermsEnum.SeekStatus.NOT_FOUND, te.SeekCeil(new BytesRef("1")));
assertNotSame("1", te.Term);
assertEquals(TermsEnum.SeekStatus.NOT_FOUND, te.SeekCeil(new BytesRef("0")));
assertNotSame("0", te.Term.Utf8ToString());
}
}
finally
{
foreach (Directory d in dirs)
{
d.Dispose();
}
}
}
public void TestSplitSeq()
{
MultiPassIndexSplitter splitter = new MultiPassIndexSplitter();
Directory[] dirs = new Directory[] {
NewDirectory(),
NewDirectory(),
NewDirectory()
};
try
{
splitter.Split(TEST_VERSION_CURRENT, input, dirs, true);
Document doc;
int start;
IndexReader ir;
using (ir = DirectoryReader.Open(dirs[0]))
{
assertTrue(ir.NumDocs - NUM_DOCS / 3 <= 1);
doc = ir.Document(0);
assertEquals("0", doc.Get("id"));
start = ir.NumDocs;
}
using (ir = DirectoryReader.Open(dirs[1]))
{
assertTrue(ir.NumDocs - NUM_DOCS / 3 <= 1);
doc = ir.Document(0);
assertEquals(start + "", doc.Get("id"));
start += ir.NumDocs;
}
using (ir = DirectoryReader.Open(dirs[2]))
{
assertTrue(ir.NumDocs - NUM_DOCS / 3 <= 1);
doc = ir.Document(0);
assertEquals(start + "", doc.Get("id"));
// make sure the deleted doc is not here
TermsEnum te = MultiFields.GetTerms(ir, "id").GetIterator(null);
Term t = new Term("id", (NUM_DOCS - 1) + "");
assertEquals(TermsEnum.SeekStatus.NOT_FOUND, te.SeekCeil(new BytesRef(t.Text())));
assertNotSame(t.Text(), te.Term.Utf8ToString());
}
}
finally
{
foreach (Directory d in dirs)
{
d.Dispose();
}
}
}
public static void Main(string[] args)
{
if (args.Length < 5)
{
Console.Error.WriteLine("Usage: MultiPassIndexSplitter -out <outputDir> -num <numParts> [-seq] <inputIndex1> [<inputIndex2 ...]");
Console.Error.WriteLine("\tinputIndex\tpath to input index, multiple values are ok");
Console.Error.WriteLine("\t-out ouputDir\tpath to output directory to contain partial indexes");
Console.Error.WriteLine("\t-num numParts\tnumber of parts to produce");
Console.Error.WriteLine("\t-seq\tsequential docid-range split (default is round-robin)");
Environment.Exit(-1);
}
List <IndexReader> indexes = new List <IndexReader>();
string outDir = null;
int numParts = -1;
bool seq = false;
for (int i = 0; i < args.Length; i++)
{
if (args[i].Equals("-out"))
{
outDir = args[++i];
}
else if (args[i].Equals("-num"))
{
numParts = Convert.ToInt32(args[++i]);
}
else if (args[i].Equals("-seq"))
{
seq = true;
}
else
{
DirectoryInfo file = new DirectoryInfo(args[i]);
if (!file.Exists)
{
Console.Error.WriteLine("Invalid input path - skipping: " + file);
continue;
}
Store.Directory dir = FSDirectory.Open(new DirectoryInfo(args[i]));
try
{
if (!DirectoryReader.IndexExists(dir))
{
Console.Error.WriteLine("Invalid input index - skipping: " + file);
continue;
}
}
catch (Exception)
{
Console.Error.WriteLine("Invalid input index - skipping: " + file);
continue;
}
indexes.Add(DirectoryReader.Open(dir));
}
}
if (outDir == null)
{
throw new Exception("Required argument missing: -out outputDir");
}
if (numParts < 2)
{
throw new Exception("Invalid value of required argument: -num numParts");
}
if (indexes.Count == 0)
{
throw new Exception("No input indexes to process");
}
DirectoryInfo @out = new DirectoryInfo(outDir);
@out.Create();
if (!new DirectoryInfo(outDir).Exists)
{
throw new Exception("Can't create output directory: " + @out);
}
Store.Directory[] dirs = new Store.Directory[numParts];
for (int i = 0; i < numParts; i++)
{
dirs[i] = FSDirectory.Open(new DirectoryInfo(Path.Combine(@out.FullName, "part-" + i)));
}
MultiPassIndexSplitter splitter = new MultiPassIndexSplitter();
IndexReader input;
if (indexes.Count == 1)
{
input = indexes[0];
}
else
{
input = new MultiReader(indexes.ToArray());
}
#pragma warning disable 612, 618
splitter.Split(LuceneVersion.LUCENE_CURRENT, input, dirs, seq);
#pragma warning restore 612, 618
}