Lucene.Net.Analysis.Synonym.SynonymMap.Builder.Add C# (CSharp) Метод

            internal virtual void Add(CharsRef input, int numInputWords, CharsRef output, int numOutputWords, bool includeOrig)
            {
                // first convert to UTF-8
                if (numInputWords <= 0)
                {
                    throw new System.ArgumentException("numInputWords must be > 0 (got " + numInputWords + ")");
                }
                if (input.Length <= 0)
                {
                    throw new System.ArgumentException("input.length must be > 0 (got " + input.Length + ")");
                }
                if (numOutputWords <= 0)
                {
                    throw new System.ArgumentException("numOutputWords must be > 0 (got " + numOutputWords + ")");
                }
                if (output.Length <= 0)
                {
                    throw new System.ArgumentException("output.length must be > 0 (got " + output.Length + ")");
                }

                Debug.Assert(!HasHoles(input), "input has holes: " + input);
                Debug.Assert(!HasHoles(output), "output has holes: " + output);

                //System.out.println("fmap.add input=" + input + " numInputWords=" + numInputWords + " output=" + output + " numOutputWords=" + numOutputWords);
                UnicodeUtil.UTF16toUTF8(output.Chars, output.Offset, output.Length, utf8Scratch);
                // lookup in hash
                int ord = words.Add(utf8Scratch);
                if (ord < 0)
                {
                    // already exists in our hash
                    ord = (-ord) - 1;
                    //System.out.println("  output=" + output + " old ord=" + ord);
                }
                else
                {
                    //System.out.println("  output=" + output + " new ord=" + ord);
                }

                MapEntry e = workingSet.ContainsKey(input) ? workingSet[input] : null;
                if (e == null)
                {
                    e = new MapEntry();
                    workingSet[CharsRef.DeepCopyOf(input)] = e; // make a copy, since we will keep around in our map
                }

                e.ords.Add(ord);
                e.includeOrig |= includeOrig;
                maxHorizontalContext = Math.Max(maxHorizontalContext, numInputWords);
                maxHorizontalContext = Math.Max(maxHorizontalContext, numOutputWords);
            }

        public virtual void TestMaxPosition3WithSynomyms()
        {
            foreach (bool consumeAll in new bool[] { true, false })
            {
                MockTokenizer tokenizer = new MockTokenizer(new StringReader("one two three four five"), MockTokenizer.WHITESPACE, false);
                // if we are consuming all tokens, we can use the checks, otherwise we can't
                tokenizer.EnableChecks = consumeAll;

                SynonymMap.Builder builder = new SynonymMap.Builder(true);
                builder.Add(new CharsRef("one"), new CharsRef("first"), true);
                builder.Add(new CharsRef("one"), new CharsRef("alpha"), true);
                builder.Add(new CharsRef("one"), new CharsRef("beguine"), true);
                CharsRef multiWordCharsRef = new CharsRef();
                SynonymMap.Builder.Join(new string[] { "and", "indubitably", "single", "only" }, multiWordCharsRef);
                builder.Add(new CharsRef("one"), multiWordCharsRef, true);
                SynonymMap.Builder.Join(new string[] { "dopple", "ganger" }, multiWordCharsRef);
                builder.Add(new CharsRef("two"), multiWordCharsRef, true);
                SynonymMap synonymMap = builder.Build();
                TokenStream stream = new SynonymFilter(tokenizer, synonymMap, true);
                stream = new LimitTokenPositionFilter(stream, 3, consumeAll);

                // "only", the 4th word of multi-word synonym "and indubitably single only" is not emitted, since its position is greater than 3.
                AssertTokenStreamContents(stream, new string[] { "one", "first", "alpha", "beguine", "and", "two", "indubitably", "dopple", "three", "single", "ganger" }, new int[] { 1, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0 });
            }
        }