// TODO: this should use inputstreams from the loader, not File!
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: @Override public void inform(org.apache.lucene.analysis.util.ResourceLoader loader) throws java.io.IOException
public virtual void inform(ResourceLoader loader)
{
if (mapping != null)
{
IList<string> wlist = null;
File mappingFile = new File(mapping);
if (mappingFile.exists())
{
wlist = getLines(loader, mapping);
}
else
{
IList<string> files = splitFileNames(mapping);
wlist = new List<>();
foreach (string file in files)
{
IList<string> lines = getLines(loader, file.Trim());
wlist.AddRange(lines);
}
}
NormalizeCharMap.Builder builder = new NormalizeCharMap.Builder();
parseRules(wlist, builder);
normMap = builder.build();
if (normMap.map == null)
{
// if the inner FST is null, it means it accepts nothing (e.g. the file is empty)
// so just set the whole map to null
normMap = null;
}
}
}
//@Ignore("wrong finalOffset: https://issues.apache.org/jira/browse/LUCENE-3971")
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public void testRandomMaps() throws Exception
public virtual void testRandomMaps()
{
int numIterations = atLeast(3);
for (int i = 0; i < numIterations; i++)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final NormalizeCharMap map = randomMap();
NormalizeCharMap map = randomMap();
Analyzer analyzer = new AnalyzerAnonymousInnerClassHelper3(this, map);
int numRounds = 100;
checkRandomData(random(), analyzer, numRounds);
}
}
/// <summary>
/// Default constructor that takes a <seealso cref="Reader"/>. </summary>
public MappingCharFilter(NormalizeCharMap normMap, TextReader @in) : base(@in)
{
buffer.Reset(@in);
map = normMap.map;
cachedRootArcs = normMap.cachedRootArcs;
if (map != null)
{
fstReader = map.BytesReader;
}
else
{
fstReader = null;
}
}
//@Ignore("wrong finalOffset: https://issues.apache.org/jira/browse/LUCENE-3971")
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public void testFinalOffsetSpecialCase() throws Exception
public virtual void testFinalOffsetSpecialCase()
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final NormalizeCharMap.Builder builder = new NormalizeCharMap.Builder();
NormalizeCharMap.Builder builder = new NormalizeCharMap.Builder();
builder.add("t", "");
// even though this below rule has no effect, the test passes if you remove it!!
builder.add("tmakdbl", "c");
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final NormalizeCharMap map = builder.build();
NormalizeCharMap map = builder.build();
Analyzer analyzer = new AnalyzerAnonymousInnerClassHelper2(this, map);
string text = "gzw f quaxot";
checkAnalysisConsistency(random(), analyzer, false, text);
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: @Override public void setUp() throws Exception
public override void setUp()
{
base.setUp();
NormalizeCharMap.Builder builder = new NormalizeCharMap.Builder();
builder.add("aa", "a");
builder.add("bbb", "b");
builder.add("cccc", "cc");
builder.add("h", "i");
builder.add("j", "jj");
builder.add("k", "kkk");
builder.add("ll", "llll");
builder.add("empty", "");
// BMP (surrogate pair):
builder.add(UnicodeUtil.newString(new int[] {0x1D122}, 0, 1), "fclef");
builder.add("\uff01", "full-width-exclamation");
normMap = builder.build();
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: @Override public void setUp() throws Exception
public override void setUp()
{
base.setUp();
NormalizeCharMap.Builder builder = new NormalizeCharMap.Builder();
builder.add("aa", "a");
builder.add("bbb", "b");
builder.add("cccc", "cc");
builder.add("h", "i");
builder.add("j", "jj");
builder.add("k", "kkk");
builder.add("ll", "llll");
builder.add("empty", "");
// BMP (surrogate pair):
builder.add(UnicodeUtil.newString(new int[] { 0x1D122 }, 0, 1), "fclef");
builder.add("\uff01", "full-width-exclamation");
normMap = builder.build();
}
public AnalyzerAnonymousInnerClassHelper3(TestMappingCharFilter outerInstance, NormalizeCharMap map)
{
this.outerInstance = outerInstance;
this.map = map;
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public void testRandomMaps2() throws Exception
public virtual void testRandomMaps2()
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final java.util.Random random = random();
Random random = random();
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int numIterations = atLeast(3);
int numIterations = atLeast(3);
for (int iter = 0; iter < numIterations; iter++)
{
if (VERBOSE)
{
Console.WriteLine("\nTEST iter=" + iter);
}
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final char endLetter = (char) org.apache.lucene.util.TestUtil.nextInt(random, 'b', 'z');
char endLetter = (char)TestUtil.Next(random, 'b', 'z');
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final java.util.Map<String,String> map = new java.util.HashMap<>();
IDictionary <string, string> map = new Dictionary <string, string>();
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final NormalizeCharMap.Builder builder = new NormalizeCharMap.Builder();
NormalizeCharMap.Builder builder = new NormalizeCharMap.Builder();
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int numMappings = atLeast(5);
int numMappings = atLeast(5);
if (VERBOSE)
{
Console.WriteLine(" mappings:");
}
while (map.Count < numMappings)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final String key = org.apache.lucene.util.TestUtil.randomSimpleStringRange(random, 'a', endLetter, 7);
string key = TestUtil.randomSimpleStringRange(random, 'a', endLetter, 7);
if (key.Length != 0 && !map.ContainsKey(key))
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final String value = org.apache.lucene.util.TestUtil.randomSimpleString(random);
string value = TestUtil.randomSimpleString(random);
map[key] = value;
builder.add(key, value);
if (VERBOSE)
{
Console.WriteLine(" " + key + " -> " + value);
}
}
}
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final NormalizeCharMap charMap = builder.build();
NormalizeCharMap charMap = builder.build();
if (VERBOSE)
{
Console.WriteLine(" test random documents...");
}
for (int iter2 = 0; iter2 < 100; iter2++)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final String content = org.apache.lucene.util.TestUtil.randomSimpleStringRange(random, 'a', endLetter, atLeast(1000));
string content = TestUtil.randomSimpleStringRange(random, 'a', endLetter, atLeast(1000));
if (VERBOSE)
{
Console.WriteLine(" content=" + content);
}
// Do stupid dog-slow mapping:
// Output string:
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final StringBuilder output = new StringBuilder();
StringBuilder output = new StringBuilder();
// Maps output offset to input offset:
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final java.util.List<Integer> inputOffsets = new java.util.ArrayList<>();
IList <int?> inputOffsets = new List <int?>();
int cumDiff = 0;
int charIdx = 0;
while (charIdx < content.Length)
{
int matchLen = -1;
string matchRepl = null;
foreach (KeyValuePair <string, string> ent in map.SetOfKeyValuePairs())
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final String match = ent.getKey();
string match = ent.Key;
if (charIdx + match.Length <= content.Length)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int limit = charIdx+match.length();
int limit = charIdx + match.Length;
bool matches = true;
for (int charIdx2 = charIdx; charIdx2 < limit; charIdx2++)
{
if (match[charIdx2 - charIdx] != content[charIdx2])
{
matches = false;
break;
}
}
if (matches)
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final String repl = ent.getValue();
string repl = ent.Value;
if (match.Length > matchLen)
{
// Greedy: longer match wins
matchLen = match.Length;
matchRepl = repl;
}
}
}
}
if (matchLen != -1)
{
// We found a match here!
if (VERBOSE)
{
Console.WriteLine(" match=" + content.Substring(charIdx, matchLen) + " @ off=" + charIdx + " repl=" + matchRepl);
}
output.Append(matchRepl);
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int minLen = Math.min(matchLen, matchRepl.length());
int minLen = Math.Min(matchLen, matchRepl.Length);
// Common part, directly maps back to input
// offset:
for (int outIdx = 0; outIdx < minLen; outIdx++)
{
inputOffsets.Add(output.Length - matchRepl.Length + outIdx + cumDiff);
}
cumDiff += matchLen - matchRepl.Length;
charIdx += matchLen;
if (matchRepl.Length < matchLen)
{
// Replacement string is shorter than matched
// input: nothing to do
}
else if (matchRepl.Length > matchLen)
{
// Replacement string is longer than matched
// input: for all the "extra" chars we map
// back to a single input offset:
for (int outIdx = matchLen; outIdx < matchRepl.Length; outIdx++)
{
inputOffsets.Add(output.Length + cumDiff - 1);
}
}
else
{
// Same length: no change to offset
}
Debug.Assert(inputOffsets.Count == output.Length, "inputOffsets.size()=" + inputOffsets.Count + " vs output.length()=" + output.Length);
}
else
{
inputOffsets.Add(output.Length + cumDiff);
output.Append(content[charIdx]);
charIdx++;
}
}
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final String expected = output.toString();
string expected = output.ToString();
if (VERBOSE)
{
Console.Write(" expected:");
for (int charIdx2 = 0; charIdx2 < expected.Length; charIdx2++)
{
Console.Write(" " + expected[charIdx2] + "/" + inputOffsets[charIdx2]);
}
Console.WriteLine();
}
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final MappingCharFilter mapFilter = new MappingCharFilter(charMap, new java.io.StringReader(content));
MappingCharFilter mapFilter = new MappingCharFilter(charMap, new StringReader(content));
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final StringBuilder actualBuilder = new StringBuilder();
StringBuilder actualBuilder = new StringBuilder();
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final java.util.List<Integer> actualInputOffsets = new java.util.ArrayList<>();
IList <int?> actualInputOffsets = new List <int?>();
// Now consume the actual mapFilter, somewhat randomly:
while (true)
{
if (random.nextBoolean())
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int ch = mapFilter.read();
int ch = mapFilter.read();
if (ch == -1)
{
break;
}
actualBuilder.Append((char)ch);
}
else
{
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final char[] buffer = new char[org.apache.lucene.util.TestUtil.nextInt(random, 1, 100)];
char[] buffer = new char[TestUtil.Next(random, 1, 100)];
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int off = buffer.length == 1 ? 0 : random.nextInt(buffer.length-1);
int off = buffer.Length == 1 ? 0 : random.Next(buffer.Length - 1);
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final int count = mapFilter.read(buffer, off, buffer.length-off);
int count = mapFilter.read(buffer, off, buffer.Length - off);
if (count == -1)
{
break;
}
else
{
actualBuilder.Append(buffer, off, count);
}
}
if (random.Next(10) == 7)
{
// Map offsets
while (actualInputOffsets.Count < actualBuilder.Length)
{
actualInputOffsets.Add(mapFilter.correctOffset(actualInputOffsets.Count));
}
}
}
// Finish mappping offsets
while (actualInputOffsets.Count < actualBuilder.Length)
{
actualInputOffsets.Add(mapFilter.correctOffset(actualInputOffsets.Count));
}
//JAVA TO C# CONVERTER WARNING: The original Java variable was marked 'final':
//ORIGINAL LINE: final String actual = actualBuilder.toString();
string actual = actualBuilder.ToString();
// Verify:
assertEquals(expected, actual);
assertEquals(inputOffsets, actualInputOffsets);
}
}
}
public AnalyzerAnonymousInnerClassHelper3(TestMappingCharFilter outerInstance, NormalizeCharMap map)
{
this.outerInstance = outerInstance;
this.map = map;
}
protected internal virtual void parseRules(IList<string> rules, NormalizeCharMap.Builder builder)
{
foreach (string rule in rules)
{
Matcher m = p.matcher(rule);
if (!m.find())
{
throw new System.ArgumentException("Invalid Mapping Rule : [" + rule + "], file = " + mapping);
}
builder.add(parseString(m.group(1)), parseString(m.group(2)));
}
}
