public long Write()
{
if (_committed)
{
return(_ix.VersionId);
}
var trieBuilder = new TrieBuilder();
var docTimer = Stopwatch.StartNew();
var upsert = new DocumentUpsertCommand(_writeSession, _analyzer, trieBuilder);
foreach (var doc in _documents.ReadSource())
{
doc.Id = _count++;
upsert.Write(doc);
}
Log.InfoFormat("stored {0} documents in {1}", _count, docTimer.Elapsed);
var posTimer = Stopwatch.StartNew();
var tries = trieBuilder.GetTries();
foreach (var trie in tries)
{
// decode into a list of words and set postings address
foreach (var node in trie.Value.EndOfWordNodes())
{
node.PostingsAddress = _postingsWriter.Write(node.Postings);
}
if (Log.IsDebugEnabled)
{
foreach (var word in trie.Value.Words())
{
Log.DebugFormat("{0}\t{1}", word.Value, word.Postings.Count);
}
}
}
Log.InfoFormat(
"stored postings refs in trees in {0}",
posTimer.Elapsed);
var treeTimer = Stopwatch.StartNew();
_ix.FieldOffsets = SerializeTries(tries, _compoundFile);
Log.InfoFormat("serialized trees in {0}", treeTimer.Elapsed);
_ix.PostingsOffset = _compoundFile.Position;
_postingsWriter.Stream.Flush();
_postingsWriter.Stream.Position = 0;
_postingsWriter.Stream.CopyTo(_compoundFile);
_ix.DocumentCount = _count;
return(_ix.VersionId);
}
public void when_()
{
var trieBuilder = new TrieBuilder <string>(
rootNodeDataFactory: () => "",
nodeDataMerger: nodeDatas => nodeDatas.First(),
nodeDataComparor: (first, second) => first.CompareTo(second),
nodeDataHasher: nodeData => nodeData.GetHashCode(),
canParentAcceptChildChecker: (parent, child) => child.StartsWith(parent),
canNodeHaveChildrenChecker: nodeData => true);
var datas = new[]
{
"abi",
"efg*",
"a",
"abc",
"abc*",
"abcd",
};
var expectedTree = new TestNode
{
Key = "",
Children = new[]
{
new TestNode
{
Key = "a",
Children = new[]
{
new TestNode
{
Key = "abc",
Children = new[]
{
new TestNode {
Key = "abc*"
},
new TestNode {
Key = "abcd"
}
}
},
new TestNode {
Key = "abi"
}
}
},
new TestNode {
Key = "efg*"
}
},
};
var trie = trieBuilder.CreateTrie(datas);
AssertTree(expectedTree, trie);
}
public void Write(
Document document,
IDocumentStoreWriter storeWriter,
IAnalyzer analyzer,
TrieBuilder trieBuilder)
{
var analyzed = analyzer.AnalyzeDocument(document);
foreach (var word in analyzed.Words)
{
var field = word.Term.Field;
var token = word.Term.Word.Value;
var posting = word.Posting;
trieBuilder.Add(new WordInfo(field, token, posting));
}
storeWriter.Write(document);
}
public void Puzzle3Test()
{
//Arrange
IDictionaryRepository dictionaryRepository = new DictionaryRepository();
ITrieBuilder trieBuilder = new TrieBuilder(dictionaryRepository);
BoggleWordSearchService boggleWordSearchService = new BoggleWordSearchService(trieBuilder);
var board = "FVKOWQIMEZTLRPBE";
var expectedResults = Solution3Results();
//Act
HashSet <string> result = boggleWordSearchService.FindWordsInBoggle(board, 4, 4);
//only care that tries look the same.
var resultString = Newtonsoft.Json.JsonConvert.SerializeObject(result.OrderBy(keys => keys));
var expectedString = Newtonsoft.Json.JsonConvert.SerializeObject(expectedResults);
//Assert
Assert.AreEqual(expectedString, resultString);
}
public void Puzzle5By5Test()
{
//Arrange
Mock <IDictionaryRepository> dictionaryRepositoryMock = new Mock <IDictionaryRepository>();
dictionaryRepositoryMock.Setup(x => x.GetWords()).Returns(Solution5By5Results().ToList);
ITrieBuilder trieBuilder = new TrieBuilder(dictionaryRepositoryMock.Object);
BoggleWordSearchService boggleWordSearchService = new BoggleWordSearchService(trieBuilder);
var board = "sabtebralnuetoenrkeseqnss";
var expectedResults = Solution5By5Results();
//Act
HashSet <string> result = boggleWordSearchService.FindWordsInBoggle(board, 5, 5);
//only care that tries look the same.
var resultString = Newtonsoft.Json.JsonConvert.SerializeObject(result.OrderBy(keys => keys));
var expectedString = Newtonsoft.Json.JsonConvert.SerializeObject(expectedResults.OrderBy(keys => keys));
//Assert
Assert.AreEqual(expectedString, resultString);
}
public long Commit()
{
var ts = new List <Task>();
var trieBuilder = new TrieBuilder();
using (var documentsToAnalyze = new BlockingCollection <Document>())
{
ts.Add(Task.Run(() =>
{
Log.Info("serializing documents");
var count = 0;
var docFileName = Path.Combine(_directory, _indexVersionId + ".rdoc");
var docAddressFn = Path.Combine(_directory, _indexVersionId + ".da");
var readTimer = new Stopwatch();
readTimer.Start();
using (var docAddressWriter = new DocumentAddressWriter(new FileStream(docAddressFn, FileMode.Create, FileAccess.Write)))
using (var docWriter = new DocumentWriter(new FileStream(docFileName, FileMode.Create, FileAccess.Write), _compression))
{
foreach (var doc in ReadSourceAndAssignIdentifiers())
{
documentsToAnalyze.Add(doc);
var adr = docWriter.Write(doc);
docAddressWriter.Write(adr);
count++;
}
}
documentsToAnalyze.CompleteAdding();
Log.InfoFormat("serialized {0} documents in {1}", count, readTimer.Elapsed);
}));
ts.Add(Task.Run(() =>
{
var analyzeTimer = new Stopwatch();
analyzeTimer.Start();
Log.Info("analyzing");
var count = 0;
try
{
while (true)
{
var doc = documentsToAnalyze.Take();
var analyzed = _analyzer.AnalyzeDocument(doc);
foreach (var term in analyzed.Words.GroupBy(t => t.Term.Field))
{
trieBuilder.Add(term.Key, term.Select(t =>
{
var field = t.Term.Field;
var token = t.Term.Word.Value;
var posting = t.Posting;
return(new WordInfo(field, token, posting));
}).ToList());
}
count++;
}
}
catch (InvalidOperationException)
{
// Done
trieBuilder.CompleteAdding();
}
Log.InfoFormat("analyzed {0} documents in {1}", count, analyzeTimer.Elapsed);
}));
Task.WaitAll(ts.ToArray());
}
var tries = trieBuilder.GetTries();
var tasks = new List <Task>
{
Task.Run(() =>
{
var postingsTimer = new Stopwatch();
postingsTimer.Start();
Log.Info("serializing postings");
var posFileName = Path.Combine(_directory, string.Format("{0}.{1}", _indexVersionId, "pos"));
using (var postingsWriter = new PostingsWriter(new FileStream(posFileName, FileMode.Create, FileAccess.Write, FileShare.None)))
{
foreach (var trie in tries)
{
foreach (var node in trie.Value.EndOfWordNodes())
{
node.PostingsAddress = postingsWriter.Write(node.Postings);
}
if (Log.IsDebugEnabled)
{
foreach (var word in trie.Value.Words())
{
Log.Debug(word);
}
}
}
}
Log.InfoFormat("serialized postings in {0}", postingsTimer.Elapsed);
}),
Task.Run(() =>
{
var trieTimer = new Stopwatch();
trieTimer.Start();
Log.Info("serializing tries");
SerializeTries(tries);
Log.InfoFormat("serialized tries in {0}", trieTimer.Elapsed);
}),
Task.Run(() =>
{
var docHasTimer = new Stopwatch();
docHasTimer.Start();
Log.Info("serializing doc hashes");
var docHashesFileName = Path.Combine(_directory, string.Format("{0}.{1}", _indexVersionId, "pk"));
_primaryKeys.Keys.Select(h => new DocHash(h)).Serialize(docHashesFileName);
Log.InfoFormat("serialized doc hashes in {0}", docHasTimer.Elapsed);
})
};
Task.WaitAll(tasks.ToArray());
CreateIxInfo().Serialize(Path.Combine(_directory, _indexVersionId + ".ix"));
if (_compression > 0)
{
Log.Info("compression: true");
}
else
{
Log.Info("compression: false");
}
return(_indexVersionId);
}
public long Write()
{
if (_committed)
{
return(_indexVersionId);
}
var trieBuilder = new TrieBuilder();
var posFileName = Path.Combine(
_directory, string.Format("{0}.{1}", _indexVersionId, "pos"));
var docTimer = Stopwatch.StartNew();
foreach (var doc in _documents.ReadSource())
{
doc.Id = _count++;
new DocumentUpsertOperation().Write(
doc,
_storeWriter,
_analyzer,
trieBuilder);
}
Log.InfoFormat("stored {0} documents in {1}", _count + 1, docTimer.Elapsed);
var posTimer = Stopwatch.StartNew();
var tries = trieBuilder.GetTries();
using (var postingsWriter = new PostingsWriter(
new FileStream(posFileName, FileMode.Append, FileAccess.Write, FileShare.ReadWrite)))
{
foreach (var trie in tries)
{
foreach (var node in trie.Value.EndOfWordNodes())
{
node.PostingsAddress = postingsWriter.Write(node.Postings);
}
if (Log.IsDebugEnabled)
{
foreach (var word in trie.Value.Words())
{
Log.DebugFormat("{0}\t{1}", word.Value, word.Count);
}
}
}
}
Log.InfoFormat(
"stored postings refs in trees and wrote postings file in {0}",
posTimer.Elapsed);
var treeTimer = new Stopwatch();
treeTimer.Start();
SerializeTries(tries);
Log.InfoFormat("serialized trees in {0}", treeTimer.Elapsed);
return(_indexVersionId);
}
public long Commit()
{
var trieBuilder = new TrieBuilder();
var docAddresses = new List <BlockInfo>();
var analyzed = _analyzer.AnalyzeDocument(_document);
foreach (var term in analyzed.Words.GroupBy(t => t.Term.Field))
{
trieBuilder.Add(term.Key, term.Select(t =>
{
var field = t.Term.Field;
var token = t.Term.Word.Value;
var posting = t.Posting;
return(new WordInfo(field, token, posting));
}).ToList());
}
trieBuilder.CompleteAdding();
var indexVersionId = Util.GetChronologicalFileId();
var docFileName = Path.Combine(_directory, indexVersionId + ".rdoc");
var docAddressesFn = Path.Combine(_directory, indexVersionId + ".da");
BlockInfo adr;
using (var docWriter = new DocumentWriter(
new FileStream(docFileName, FileMode.Create, FileAccess.Write, FileShare.None, 4096, true), _compression))
{
adr = docWriter.Write(_document);
}
using (var docAddressWriter = new DocumentAddressWriter(
new FileStream(docAddressesFn, FileMode.Create, FileAccess.Write, FileShare.None)))
{
docAddressWriter.Write(adr);
}
var tries = trieBuilder.GetTries();
var tasks = new List <Task>
{
Task.Run(() =>
{
var posFileName = Path.Combine(_directory, string.Format("{0}.{1}", indexVersionId, "pos"));
using (var postingsWriter = new PostingsWriter(new FileStream(posFileName, FileMode.Create, FileAccess.Write, FileShare.None)))
{
foreach (var trie in tries)
{
foreach (var node in trie.Value.EndOfWordNodes())
{
node.PostingsAddress = postingsWriter.Write(node.Postings);
}
if (Log.IsDebugEnabled)
{
foreach (var word in trie.Value.Words())
{
Log.Debug(word);
}
}
}
}
}),
Task.Run(() =>
{
SerializeTries(tries, indexVersionId);
}),
Task.Run(() =>
{
var docHashesFileName = Path.Combine(_directory, string.Format("{0}.{1}", indexVersionId, "pk"));
new DocHash[] { new DocHash(_primaryKeyHash) }.Serialize(docHashesFileName);
})
};
Task.WaitAll(tasks.ToArray());
new IxInfo
{
VersionId = indexVersionId,
DocumentCount = 1,
Compression = _compression
}.Serialize(Path.Combine(_directory, indexVersionId + ".ix"));
if (_compression > 0)
{
Log.Info("compression: true");
}
else
{
Log.Info("compression: false");
}
return(indexVersionId);
}
public DocumentUpsertCommand(IWriteSession writeSession, IAnalyzer analyzer, TrieBuilder treeBuilder)
{
_writeSession = writeSession;
_analyzer = analyzer;
_treeBuilder = treeBuilder;
}
public void BuildSmallTrieIsValid()
{
//Arrange
var smallWordList = new List <string> {
"Apple", "Apples", "Ape", "Cat"
};
Mock <IDictionaryRepository> dictionaryRepo = new Mock <IDictionaryRepository>();
dictionaryRepo.Setup(x => x.GetWords()).Returns(smallWordList);
TrieBuilder trieBuilder = new TrieBuilder(dictionaryRepo.Object);
var expectedResults = Factory.CreateTrieList();
expectedResults.Add(new Trie
{
Value = 'A',
ValidWord = false,
Children = new List <ITrie>
{
new Trie
{
Value = 'P',
ValidWord = false,
Children = new List <ITrie>
{
new Trie
{
Value = 'P',
ValidWord = false,
Children = new List <ITrie>
{
new Trie
{
Value = 'L',
ValidWord = false,
Children = new List <ITrie>
{
new Trie
{
Value = 'E',
ValidWord = true,
Children = new List <ITrie>
{
new Trie
{
Value = 'S',
ValidWord = true,
Children = new List <ITrie>()
}
}
}
}
}
}
},
new Trie
{
Value = 'E',
ValidWord = true,
Children = new List <ITrie>()
}
}
}
}
});
expectedResults.Add(new Trie
{
Value = 'C',
ValidWord = false,
Children = new List <ITrie>
{
new Trie
{
Value = 'A',
ValidWord = false,
Children = new List <ITrie>
{
new Trie
{
Value = 'T',
ValidWord = true,
Children = new List <ITrie>()
}
}
}
}
});
//Act
var result = trieBuilder.BuildTrie();
//only care that tries look the same.
var resultString = Newtonsoft.Json.JsonConvert.SerializeObject(result);
var expectedString = Newtonsoft.Json.JsonConvert.SerializeObject(expectedResults);
//Assert
Assert.AreEqual(expectedString, resultString);
}
