/// <summary>
/// Gets all synsets for a word, optionally restricting the returned synsets to one or more parts of speech. This
/// method does not perform any morphological analysis to match up the given word.
/// </summary>
/// <param name="word">Word to get SynSets for.</param>
/// <param name="pos">Part-of-speech to search.</param>
/// <returns>A readonly collection of SynSets that contain the requested word.</returns>
public IReadOnlyCollection <SynSet> GetSynSets(string word, WordNetPos pos)
{
word = word.ToLowerInvariant().Replace(' ', '_');
var indexLine = index[pos].Search(word);
if (indexLine == null)
{
return(new ReadOnlyCollection <SynSet>(new SynSet[0]));
}
SynSet mostCommonSynset;
var synsets = GetSynSetShells(indexLine, pos, out mostCommonSynset, wordNet);
foreach (var synset in synsets)
{
synset.Instantiate(this);
}
// we only need to set this flag if there is more than one synset for the word-pos pair
if (synsets.Count > 1)
{
mostCommonSynset.SetAsMostCommonSynsetFor(word);
}
return(new ReadOnlyCollection <SynSet>(synsets));
}
/// <summary>
/// Gets all synsets for a word, optionally restricting the returned synsets to one or more parts of speech. This
/// method does not perform any morphological analysis to match up the given word.
/// </summary>
/// <param name="word">Word to get SynSets for.</param>
/// <param name="pos">Part-of-speech to search.</param>
/// <returns>A readonly collection of SynSets that contain the requested word.</returns>
public IReadOnlyCollection <SynSet> GetSynSets(string word, WordNetPos pos)
{
List <SynSet> list;
return(posWordSynSets[pos].TryGetValue(word, out list)
? new ReadOnlyCollection <SynSet>(list)
: new ReadOnlyCollection <SynSet>(new SynSet[0]));
}
/// <summary>
/// Constructor. Creates the shell of a SynSet without any actual information.
/// To gain access to SynSet words, gloss, and related SynSets, call SynSet.Instantiate.
/// </summary>
/// <param name="pos">POS of SynSet</param>
/// <param name="offset">Byte location of SynSet definition within data file</param>
/// <param name="wordnet">WordNet engine used to instantiate this synset.</param>
internal SynSet(WordNetPos pos, int offset, WordNet wordnet)
{
Id = string.Format("{0}:{1}", pos, offset);
Pos = pos;
Offset = offset;
Instantiated = false;
wordNet = wordnet;
}
/// <summary>
/// Gets the relation for a given POS and symbol
/// </summary>
/// <param name="pos">POS to get relation for</param>
/// <param name="symbol">Symbol to get relation for</param>
/// <returns>SynSet relation</returns>
public static SynSetRelation GetSynSetRelation(WordNetPos pos, string symbol)
{
if (pos == WordNetPos.None)
{
throw new ArgumentException(@"The pos argument must not be None", "pos");
}
return(symbolRelation[pos][symbol]);
}
/// <summary>
/// Gets the most common synset for a given word/pos pair. This is only available for memory-based
/// engines (see constructor).
/// </summary>
/// <param name="word">Word to get SynSets for. This method will replace all spaces with underscores and
/// will call String.ToLower to normalize case.</param>
/// <param name="pos">Part of speech to find</param>
/// <returns>Most common synset for given word/pos pair</returns>
public SynSet GetMostCommonSynSet(string word, WordNetPos pos)
{
// all words are lower case and space-replaced...we need to do this here, even though it gets done in GetSynSets (we use it below)
word = word.ToLower().Replace(' ', '_');
// get synsets for word-pos pair
var synsets = GetSynSets(word, pos);
// get most common synset
SynSet mostCommon = null;
if (synsets.Count == 1)
{
return(synsets.First());
}
if (synsets.Count <= 1)
{
return(null);
}
// one (and only one) of the synsets should be flagged as most common
foreach (var synset in synsets)
{
if (synset.IsMostCommonSynsetFor(word))
{
if (mostCommon == null)
{
mostCommon = synset;
}
else
{
throw new Exception("Multiple most common synsets found");
}
}
}
if (mostCommon == null)
{
throw new NullReferenceException("Failed to find most common synset");
}
return(mostCommon);
}
/// <summary>
/// Gets the definition for a synset
/// </summary>
/// <param name="pos">Part-of-speech to get definition for.</param>
/// <param name="offset">Offset or a index into data file.</param>
public string GetSynSetDefinition(WordNetPos pos, int offset)
{
data[pos].DiscardBufferedData();
data[pos].BaseStream.Position = offset;
// read synset definition
var synSetDefinition = data[pos].ReadLine();
if (string.IsNullOrEmpty(synSetDefinition))
{
return(null);
}
// make sure file positions line up
if (int.Parse(synSetDefinition.Substring(0, synSetDefinition.IndexOf(' '))) != offset)
{
throw new Exception("Position mismatch: passed " + offset + " and got definition line \"" + synSetDefinition + "\"");
}
return(synSetDefinition);
}
/// <summary>
/// Gets synset shells from a word index line. A synset shell is an instance of SynSet with only the POS and Offset
/// members initialized. These members are enough to look up the full synset within the corresponding data file. This
/// method is static to prevent inadvertent references to a current WordNetEngine, which should be passed via the
/// corresponding parameter.
/// </summary>
/// <param name="wordIndexLine">Word index line from which to get synset shells</param>
/// <param name="pos">POS of the given index line</param>
/// <param name="mostCommonSynSet">Returns the most common synset for the word</param>
/// <param name="wordNet">The WordNet instance</param>
/// <returns>Synset shells for the given index line</returns>
/// <exception cref="System.Exception">Failed to get most common synset</exception>
internal static List <SynSet> GetSynSetShells(string wordIndexLine, WordNetPos pos, out SynSet mostCommonSynSet, WordNet wordNet)
{
var synsets = new List <SynSet>();
mostCommonSynSet = null;
// get number of synsets
var parts = wordIndexLine.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
var numSynSets = int.Parse(parts[2]);
// grab each synset shell, from last to first
int firstOffsetIndex = parts.Length - numSynSets;
for (int i = parts.Length - 1; i >= firstOffsetIndex; --i)
{
// create synset
int offset = int.Parse(parts[i]);
// add synset to collection
var synset = new SynSet(pos, offset, wordNet);
synsets.Add(synset);
// if this is the last synset offset to get (since we grabbed them in reverse order), record it as the most common synset
if (i == firstOffsetIndex)
{
mostCommonSynSet = synset;
}
}
if (mostCommonSynSet == null)
{
throw new Exception("Failed to get most common synset");
}
return(synsets);
}
/// <summary>
/// Gets the relation for a given POS and symbol
/// </summary>
/// <param name="pos">POS to get relation for</param>
/// <param name="symbol">Symbol to get relation for</param>
/// <returns>SynSet relation</returns>
public static SynSetRelation GetSynSetRelation(WordNetPos pos, string symbol) {
if (pos == WordNetPos.None)
throw new ArgumentException(@"The pos argument must not be None", "pos");
return symbolRelation[pos][symbol];
}
/// <summary>
/// Gets all synsets for a word, optionally restricting the returned synsets to one or more parts of speech. This
/// method does not perform any morphological analysis to match up the given word.
/// </summary>
/// <param name="word">Word to get SynSets for.</param>
/// <param name="pos">Part-of-speech to search.</param>
/// <returns>A readonly collection of SynSets that contain the requested word.</returns>
public IReadOnlyCollection<SynSet> GetSynSets(string word, WordNetPos pos) {
List<SynSet> list;
return posWordSynSets[pos].TryGetValue(word, out list)
? new ReadOnlyCollection<SynSet>(list)
: new ReadOnlyCollection<SynSet>(new SynSet[0]);
}
/// <summary>
/// Gets all words with the specified part-of-speech.
/// </summary>
/// <param name="pos">The part-of-speech to get words for.</param>
/// <returns>A readonly collection containing all the words with the specified part-of-speech tag.</returns>
public IReadOnlyCollection <string> GetAllWords(WordNetPos pos)
{
return(pos == WordNetPos.None ? null : index[pos].GetAllWords());
}
/// <summary>
/// Gets all words with the specified part-of-speech.
/// </summary>
/// <param name="pos">The part-of-speech to get words for.</param>
/// <returns>A readonly collection containing all the words with the specified part-of-speech tag.</returns>
public IReadOnlyCollection <string> GetAllWords(WordNetPos pos)
{
return(new ReadOnlyCollection <string>(posWordSynSets[pos].Keys.ToList()));
}
/// <summary>
/// Gets the definition for a synset
/// </summary>
/// <param name="pos">Part-of-speech to get definition for.</param>
/// <param name="offset">Offset or a index into data file.</param>
public string GetSynSetDefinition(WordNetPos pos, int offset)
{
throw new NotSupportedException();
}
/// <summary>
/// Gets the definition for a synset
/// </summary>
/// <param name="pos">Part-of-speech to get definition for.</param>
/// <param name="offset">Offset or a index into data file.</param>
public string GetSynSetDefinition(WordNetPos pos, int offset) {
throw new NotSupportedException();
}
/// <summary>
/// Gets definition line for synset from data file
/// </summary>
/// <param name="pos">POS to get definition for</param>
/// <param name="offset">Offset into data file</param>
internal string GetSynSetDefinition(WordNetPos pos, long offset) {
// set data file to synset location
data[pos].DiscardBufferedData();
data[pos].BaseStream.Position = offset;
// read synset definition
var synSetDefinition = data[pos].ReadLine();
if (string.IsNullOrEmpty(synSetDefinition))
return null;
// make sure file positions line up
if (int.Parse(synSetDefinition.Substring(0, synSetDefinition.IndexOf(' '))) != offset)
throw new Exception("Position mismatch: passed " + offset + " and got definition line \"" + synSetDefinition + "\"");
return synSetDefinition;
}
/// <summary>
/// Gets the most common synset for a given word/pos pair. This is only available for memory-based
/// engines (see constructor).
/// </summary>
/// <param name="word">Word to get SynSets for. This method will replace all spaces with underscores and
/// will call String.ToLower to normalize case.</param>
/// <param name="pos">Part of speech to find</param>
/// <returns>Most common synset for given word/pos pair</returns>
public SynSet GetMostCommonSynSet(string word, WordNetPos pos) {
// all words are lower case and space-replaced...we need to do this here, even though it gets done in GetSynSets (we use it below)
word = word.ToLower().Replace(' ', '_');
// get synsets for word-pos pair
var synsets = GetSynSets(word, pos);
// get most common synset
SynSet mostCommon = null;
if (synsets.Count == 1)
return synsets.First();
if (synsets.Count <= 1)
return null;
// one (and only one) of the synsets should be flagged as most common
foreach (var synset in synsets)
if (synset.IsMostCommonSynsetFor(word))
if (mostCommon == null)
mostCommon = synset;
else
throw new Exception("Multiple most common synsets found");
if (mostCommon == null)
throw new NullReferenceException("Failed to find most common synset");
return mostCommon;
}
/// <summary>
/// Gets synset shells from a word index line. A synset shell is an instance of SynSet with only the POS and Offset
/// members initialized. These members are enough to look up the full synset within the corresponding data file. This
/// method is static to prevent inadvertent references to a current WordNetEngine, which should be passed via the
/// corresponding parameter.
/// </summary>
/// <param name="wordIndexLine">Word index line from which to get synset shells</param>
/// <param name="pos">POS of the given index line</param>
/// <param name="mostCommonSynSet">Returns the most common synset for the word</param>
/// <param name="wordNet">The WordNet instance</param>
/// <returns>Synset shells for the given index line</returns>
/// <exception cref="System.Exception">Failed to get most common synset</exception>
internal static List<SynSet> GetSynSetShells(string wordIndexLine, WordNetPos pos, out SynSet mostCommonSynSet, WordNet wordNet) {
var synsets = new List<SynSet>();
mostCommonSynSet = null;
// get number of synsets
var parts = wordIndexLine.Split(new[] { ' ' }, StringSplitOptions.RemoveEmptyEntries);
var numSynSets = int.Parse(parts[2]);
// grab each synset shell, from last to first
int firstOffsetIndex = parts.Length - numSynSets;
for (int i = parts.Length - 1; i >= firstOffsetIndex; --i) {
// create synset
int offset = int.Parse(parts[i]);
// add synset to collection
var synset = new SynSet(pos, offset, wordNet);
synsets.Add(synset);
// if this is the last synset offset to get (since we grabbed them in reverse order), record it as the most common synset
if (i == firstOffsetIndex)
mostCommonSynSet = synset;
}
if (mostCommonSynSet == null)
throw new Exception("Failed to get most common synset");
return synsets;
}
/// <summary>
/// Gets all synsets for a word, optionally restricting the returned synsets to one or more parts of speech. This
/// method does not perform any morphological analysis to match up the given word.
/// </summary>
/// <param name="word">Word to get SynSets for.</param>
/// <param name="pos">Part-of-speech to search.</param>
/// <returns>A readonly collection of SynSets that contain the requested word.</returns>
public IReadOnlyCollection<SynSet> GetSynSets(string word, WordNetPos pos) {
word = word.ToLowerInvariant().Replace(' ', '_');
var indexLine = index[pos].Search(word);
if (indexLine == null)
return new ReadOnlyCollection<SynSet>(new SynSet[0]);
SynSet mostCommonSynset;
var synsets = GetSynSetShells(indexLine, pos, out mostCommonSynset, wordNet);
foreach (var synset in synsets) {
synset.Instantiate(this);
}
// we only need to set this flag if there is more than one synset for the word-pos pair
if (synsets.Count > 1)
mostCommonSynset.SetAsMostCommonSynsetFor(word);
return new ReadOnlyCollection<SynSet>(synsets);
}
/// <summary>
/// Gets all words with the specified part-of-speech.
/// </summary>
/// <param name="pos">The part-of-speech to get words for.</param>
/// <returns>A readonly collection containing all the words with the specified part-of-speech tag.</returns>
public IReadOnlyCollection<string> GetAllWords(WordNetPos pos) {
return pos == WordNetPos.None ? null : index[pos].GetAllWords();
}
/// <summary>
/// Constructor. Creates the shell of a SynSet without any actual information.
/// To gain access to SynSet words, gloss, and related SynSets, call SynSet.Instantiate.
/// </summary>
/// <param name="pos">POS of SynSet</param>
/// <param name="offset">Byte location of SynSet definition within data file</param>
/// <param name="wordnet">WordNet engine used to instantiate this synset.</param>
internal SynSet(WordNetPos pos, int offset, WordNet wordnet) {
Id = string.Format("{0}:{1}", pos, offset);
Pos = pos;
Offset = offset;
Instantiated = false;
wordNet = wordnet;
}
/// <summary>
/// Gets all words with the specified part-of-speech.
/// </summary>
/// <param name="pos">The part-of-speech to get words for.</param>
/// <returns>A readonly collection containing all the words with the specified part-of-speech tag.</returns>
public IReadOnlyCollection<string> GetAllWords(WordNetPos pos) {
return new ReadOnlyCollection<string>(posWordSynSets[pos].Keys.ToList());
}
