public FieldFragList CreateFieldFragList(FieldPhraseList fieldPhraseList, int fragCharSize)
{
if (fragCharSize < MIN_FRAG_CHAR_SIZE)
throw new ArgumentException("fragCharSize(" + fragCharSize + ") is too small. It must be " +
MIN_FRAG_CHAR_SIZE + " or higher.");
FieldFragList ffl = new FieldFragList(fragCharSize);
List<WeightedPhraseInfo> wpil = new List<WeightedPhraseInfo>();
LinkedList<WeightedPhraseInfo>.Enumerator ite = fieldPhraseList.phraseList.GetEnumerator();
WeightedPhraseInfo phraseInfo = null;
int startOffset = 0;
bool taken = false;
while (true)
{
if (!taken)
{
if (!ite.MoveNext()) break;
phraseInfo = ite.Current;
}
taken = false;
if (phraseInfo == null) break;
// if the phrase violates the border of previous fragment, discard it and try next phrase
if (phraseInfo.StartOffset < startOffset)
{
if(phraseInfo.EndOffset < startOffset)
continue;
startOffset = phraseInfo.StartOffset;
}
wpil.Clear();
wpil.Add(phraseInfo);
int st = phraseInfo.StartOffset - MARGIN < startOffset ?
startOffset : phraseInfo.StartOffset - MARGIN;
int en = st + fragCharSize;
if (phraseInfo.EndOffset > en)
en = phraseInfo.EndOffset;
startOffset = en;
while (true)
{
if (ite.MoveNext())
{
phraseInfo = ite.Current;
taken = true;
if (phraseInfo == null) break;
}
else
break;
if (phraseInfo.EndOffset <= en)
wpil.Add(phraseInfo);
else
break;
}
ffl.Add(st, en, wpil);
}
return ffl;
}
public void TestSmallerFragSizeThanTermQuery()
{
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl("abcdefghijklmnopqrs", "abcdefghijklmnopqrs"), SimpleFragListBuilder.MIN_FRAG_CHAR_SIZE);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(abcdefghijklmnopqrs((0,19)))/1.0(0,19)", ffl.fragInfos[0].ToString());
}
public void Test1TermIndex()
{
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl("a", "a"), 100);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(a((0,1)))/1.0(0,100)", ffl.fragInfos[0].ToString());
}
public void TestPhraseQuerySlop()
{
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl("\"a b\"~1", "a c b"), 20);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(ab((0,1)(4,5)))/1.0(0,20)", ffl.fragInfos[0].ToString());
}
public void TestSmallerFragSizeThanPhraseQuery()
{
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl("\"abcdefgh jklmnopqrs\"", "abcdefgh jklmnopqrs"), SimpleFragListBuilder.MIN_FRAG_CHAR_SIZE);
Assert.AreEqual(1, ffl.fragInfos.Count);
Console.WriteLine(ffl.fragInfos[0].ToString());
Assert.AreEqual("subInfos=(abcdefghjklmnopqrs((0,21)))/1.0(0,21)", ffl.fragInfos[0].ToString());
}
public void Test1TermIndex()
{
FieldFragList ffl = this.ffl("a", "a");
SimpleFragmentsBuilder sfb = new SimpleFragmentsBuilder();
Assert.AreEqual("<b>a</b>", sfb.CreateFragment(reader, 0, F, ffl));
// change tags
sfb = new SimpleFragmentsBuilder(new String[] { "[" }, new String[] { "]" });
Assert.AreEqual("[a]", sfb.CreateFragment(reader, 0, F, ffl));
}
public void Test3Frags()
{
FieldFragList ffl = this.ffl("a c", "a b b b b b b b b b b b a b a b b b b b c a a b b");
SimpleFragmentsBuilder sfb = new SimpleFragmentsBuilder();
String[] f = sfb.CreateFragments(reader, 0, F, ffl, 3);
Assert.AreEqual(3, f.Length);
Assert.AreEqual("<b>a</b> b b b b b b b b b ", f[0]);
Assert.AreEqual("b b <b>a</b> b <b>a</b> b b b b b ", f[1]);
Assert.AreEqual("<b>c</b> <b>a</b> <b>a</b> b b", f[2]);
}
public void Test2Frags()
{
FieldFragList ffl = this.ffl("a", "a b b b b b b b b b b b a b a b");
SimpleFragmentsBuilder sfb = new SimpleFragmentsBuilder();
String[] f = sfb.CreateFragments(reader, 0, F, ffl, 3);
// 3 snippets requested, but should be 2
Assert.AreEqual(2, f.Length);
Assert.AreEqual("<b>a</b> b b b b b b b b b ", f[0]);
Assert.AreEqual("b b <b>a</b> b <b>a</b> b", f[1]);
}
public void Test3Frags()
{
FieldFragList ffl = this.Ffl("a c", "a b b b b b b b b b b b a b a b b b b b c a a b b");
ScoreOrderFragmentsBuilder sofb = new ScoreOrderFragmentsBuilder();
String[] f = sofb.CreateFragments(reader, 0, F, ffl, 3, null);
Assert.AreEqual(3, f.Length);
// check score order
Assert.AreEqual("<b>c</b> <b>a</b> <b>a</b> b b", f[0]);
Assert.AreEqual("b b <b>a</b> b <b>a</b> b b b b b ", f[1]);
Assert.AreEqual("<b>a</b> b b b b b b b b b ", f[2]);
}
public void Test1PhraseLongMVB()
{
MakeIndexLongMVB();
FieldQuery fq = new FieldQuery(PqF("sp", "pe", "ee", "ed"), true, true); // "speed" -(2gram)-> "sp","pe","ee","ed"
FieldTermStack stack = new FieldTermStack(reader, 0, F, fq);
FieldPhraseList fpl = new FieldPhraseList(stack, fq);
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl, 100);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(sppeeeed((88,93)))/1.0(82,182)", ffl.fragInfos[0].ToString());
}
public void Test1PhraseLongMV()
{
MakeIndexLongMV();
FieldQuery fq = new FieldQuery(PqF("search", "engines"), true, true);
FieldTermStack stack = new FieldTermStack(reader, 0, F, fq);
FieldPhraseList fpl = new FieldPhraseList(stack, fq);
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl, 100);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(searchengines((102,116))searchengines((157,171)))/2.0(96,196)", ffl.fragInfos[0].ToString());
}
public void Test1PhraseShortMV()
{
MakeIndexShortMV();
FieldQuery fq = new FieldQuery(Tq("d"), true, true);
FieldTermStack stack = new FieldTermStack(reader, 0, F, fq);
FieldPhraseList fpl = new FieldPhraseList(stack, fq);
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl, 100);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(d((6,7)))/1.0(0,100)", ffl.fragInfos[0].ToString());
}
public void Test1PhraseShortMV()
{
MakeIndexShortMV();
FieldQuery fq = new FieldQuery(Tq("d"), true, true);
FieldTermStack stack = new FieldTermStack(reader, 0, F, fq);
FieldPhraseList fpl = new FieldPhraseList(stack, fq);
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl, 100);
SimpleFragmentsBuilder sfb = new SimpleFragmentsBuilder();
Assert.AreEqual("a b c <b>d</b> e", sfb.CreateFragment(reader, 0, F, ffl));
}
public void TestUnstoredField()
{
MakeUnstoredIndex();
FieldQuery fq = new FieldQuery(Tq("aaa"), true, true);
FieldTermStack stack = new FieldTermStack(reader, 0, F, fq);
FieldPhraseList fpl = new FieldPhraseList(stack, fq);
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl, 100);
SimpleFragmentsBuilder sfb = new SimpleFragmentsBuilder();
Assert.IsNull(sfb.CreateFragment(reader, 0, F, ffl));
}
public void Test1PhraseLongMVB()
{
MakeIndexLongMVB();
FieldQuery fq = new FieldQuery(PqF("sp", "pe", "ee", "ed"), true, true); // "speed" -(2gram)-> "sp","pe","ee","ed"
FieldTermStack stack = new FieldTermStack(reader, 0, F, fq);
FieldPhraseList fpl = new FieldPhraseList(stack, fq);
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl, 100);
SimpleFragmentsBuilder sfb = new SimpleFragmentsBuilder();
Assert.AreEqual("ssing <b>speed</b>, the", sfb.CreateFragment(reader, 0, F, ffl));
}
public void Test1PhraseLongMV()
{
MakeIndexLongMV();
FieldQuery fq = new FieldQuery(PqF("search", "engines"), true, true);
FieldTermStack stack = new FieldTermStack(reader, 0, F, fq);
FieldPhraseList fpl = new FieldPhraseList(stack, fq);
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl, 100);
SimpleFragmentsBuilder sfb = new SimpleFragmentsBuilder();
Assert.AreEqual(" most <b>search engines</b> use only one of these methods. Even the <b>search engines</b> that says they can use t",
sfb.CreateFragment(reader, 0, F, ffl));
}
public void Test2TermsQuery()
{
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl("a b", "c d e"), 20);
Assert.AreEqual(0, ffl.fragInfos.Count);
ffl = sflb.CreateFieldFragList(fpl("a b", "d b c"), 20);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(b((2,3)))/1.0(0,20)", ffl.fragInfos[0].ToString());
ffl = sflb.CreateFieldFragList(fpl("a b", "a b c"), 20);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(a((0,1))b((2,3)))/2.0(0,20)", ffl.fragInfos[0].ToString());
}
public void Test2TermsIndex1Frag()
{
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl("a", "a a"), 100);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(a((0,1))a((2,3)))/2.0(0,100)", ffl.fragInfos[0].ToString());
ffl = sflb.CreateFieldFragList(fpl("a", "a b b b b b b b b a"), 20);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(a((0,1))a((18,19)))/2.0(0,20)", ffl.fragInfos[0].ToString());
ffl = sflb.CreateFieldFragList(fpl("a", "b b b b a b b b b a"), 20);
Assert.AreEqual(1, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(a((8,9))a((18,19)))/2.0(2,22)", ffl.fragInfos[0].ToString());
}
public virtual String[] CreateFragments(IndexReader reader, int docId, String fieldName, FieldFragList fieldFragList, int maxNumFragments)
{
if (maxNumFragments < 0)
throw new ArgumentException("maxNumFragments(" + maxNumFragments + ") must be positive number.");
List<WeightedFragInfo> fragInfos = GetWeightedFragInfoList(fieldFragList.fragInfos);
List<String> fragments = new List<String>(maxNumFragments);
Field[] values = GetFields(reader, docId, fieldName);
if (values.Length == 0) return null;
StringBuilder buffer = new StringBuilder();
int[] nextValueIndex = { 0 };
for (int n = 0; n < maxNumFragments && n < fragInfos.Count; n++)
{
WeightedFragInfo fragInfo = fragInfos[n];
fragments.Add(MakeFragment(buffer, nextValueIndex, values, fragInfo));
}
return fragments.ToArray();
}
public void Test2TermsIndex2Frags()
{
SimpleFragListBuilder sflb = new SimpleFragListBuilder();
FieldFragList ffl = sflb.CreateFieldFragList(fpl("a", "a b b b b b b b b b b b b b a"), 20);
Assert.AreEqual(2, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(a((0,1)))/1.0(0,20)", ffl.fragInfos[0].ToString());
Assert.AreEqual("subInfos=(a((28,29)))/1.0(22,42)", ffl.fragInfos[1].ToString());
ffl = sflb.CreateFieldFragList(fpl("a", "a b b b b b b b b b b b b a"), 20);
Assert.AreEqual(2, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(a((0,1)))/1.0(0,20)", ffl.fragInfos[0].ToString());
Assert.AreEqual("subInfos=(a((26,27)))/1.0(20,40)", ffl.fragInfos[1].ToString());
ffl = sflb.CreateFieldFragList(fpl("a", "a b b b b b b b b b a"), 20);
Assert.AreEqual(2, ffl.fragInfos.Count);
Assert.AreEqual("subInfos=(a((0,1)))/1.0(0,20)", ffl.fragInfos[0].ToString());
Assert.AreEqual("subInfos=(a((20,21)))/1.0(20,40)", ffl.fragInfos[1].ToString());
}
public virtual String CreateFragment(IndexReader reader, int docId, String fieldName, FieldFragList fieldFragList)
{
String[] fragments = CreateFragments(reader, docId, fieldName, fieldFragList, 1);
if (fragments == null || fragments.Length == 0) return null;
return fragments[0];
}
public virtual string[] CreateFragments(IndexReader reader, int docId, string fieldName, FieldFragList fieldFragList, int maxNumFragments, int fragCharSize, IState state)
{
if (maxNumFragments < 0)
{
throw new ArgumentException("maxNumFragments(" + maxNumFragments + ") must be positive number.");
}
List <WeightedFragInfo> fragInfos = GetWeightedFragInfoList(fieldFragList.fragInfos);
List <String> fragments = new List <String>(maxNumFragments);
Field[] values = GetFields(reader, docId, fieldName, state);
if (values.Length == 0)
{
return(null);
}
StringBuilder buffer = new StringBuilder();
int[] nextValueIndex = { 0 };
for (int n = 0; n < maxNumFragments && n < fragInfos.Count; n++)
{
WeightedFragInfo fragInfo = fragInfos[n];
fragments.Add(MakeFragment(buffer, nextValueIndex, values, fragInfo, fragCharSize, state));
}
return(fragments.ToArray());
}
public virtual string CreateFragment(IndexReader reader, int docId, string fieldName, FieldFragList fieldFragList, int fragCharSize, IState state)
{
String[] fragments = CreateFragments(reader, docId, fieldName, fieldFragList, 1, fragCharSize, state);
if (fragments == null || fragments.Length == 0)
{
return(null);
}
return(fragments[0]);
}
public virtual String CreateFragment(IndexReader reader, int docId, String fieldName, FieldFragList fieldFragList)
{
String[] fragments = CreateFragments(reader, docId, fieldName, fieldFragList, 1);
if (fragments == null || fragments.Length == 0)
{
return(null);
}
return(fragments[0]);
}
public FieldFragList CreateFieldFragList(FieldPhraseList fieldPhraseList, int fragCharSize)
{
if (fragCharSize < MIN_FRAG_CHAR_SIZE)
{
throw new ArgumentException("fragCharSize(" + fragCharSize + ") is too small. It must be " +
MIN_FRAG_CHAR_SIZE + " or higher.");
}
FieldFragList ffl = new FieldFragList(fragCharSize);
List <WeightedPhraseInfo> wpil = new List <WeightedPhraseInfo>();
LinkedList <WeightedPhraseInfo> .Enumerator ite = fieldPhraseList.phraseList.GetEnumerator();
WeightedPhraseInfo phraseInfo = null;
int startOffset = 0;
bool taken = false;
while (true)
{
if (!taken)
{
if (!ite.MoveNext())
{
break;
}
phraseInfo = ite.Current;
}
taken = false;
if (phraseInfo == null)
{
break;
}
// if the phrase violates the border of previous fragment, discard it and try next phrase
if (phraseInfo.StartOffset < startOffset)
{
if (phraseInfo.EndOffset < startOffset)
{
continue;
}
startOffset = phraseInfo.StartOffset;
}
wpil.Clear();
wpil.Add(phraseInfo);
int st = phraseInfo.StartOffset - MARGIN < startOffset ?
startOffset : phraseInfo.StartOffset - MARGIN;
int en = st + fragCharSize;
if (phraseInfo.EndOffset > en)
{
en = phraseInfo.EndOffset;
}
startOffset = en;
while (true)
{
if (ite.MoveNext())
{
phraseInfo = ite.Current;
taken = true;
if (phraseInfo == null)
{
break;
}
}
else
{
break;
}
if (phraseInfo.EndOffset <= en)
{
wpil.Add(phraseInfo);
}
else
{
break;
}
}
ffl.Add(st, en, wpil);
}
return(ffl);
}
/// <summary>
/// return the best fragments.
/// </summary>
/// <param name="fieldQuery">FieldQuery object</param>
/// <param name="reader">IndexReader of the index</param>
/// <param name="docId">document id to be highlighted</param>
/// <param name="fieldName">field of the document to be highlighted</param>
/// <param name="fragCharSize">the length (number of chars) of a fragment</param>
/// <param name="maxNumFragments">maximum number of fragments</param>
/// <param name="fragCharSize1"></param>
/// <returns>created fragments or null when no fragments created. Size of the array can be less than maxNumFragments</returns>
public string[] GetBestFragments(FieldQuery fieldQuery, IndexReader reader, int docId, string fieldName, int fragCharSize, int maxNumFragments, IState state)
{
FieldFragList fieldFragList = GetFieldFragList(fieldQuery, reader, docId, fieldName, fragCharSize, state);
return(fragmentsBuilder.CreateFragments(reader, docId, fieldName, fieldFragList, maxNumFragments, fragCharSize, state));
}