/// <summary>
/// Adds a suffix to a word
/// </summary>
/// <param name="word" type="string">
/// <para>
/// The word to get the suffix added to
/// </para>
/// </param>
/// <param name="rule" type="NetSpell.SpellChecker.Dictionary.Affix.AffixRule">
/// <para>
/// The AffixRule to use when adding the suffix
/// </para>
/// </param>
/// <returns>
/// The word with the suffix added
/// </returns>
public static string AddSuffix(string word, AffixRule rule)
{
foreach (AffixEntry entry in rule.AffixEntries)
{
// check that this entry is valid
if (word.Length >= entry.ConditionCount)
{
int passCount = 0;
for (int i = 0; i < entry.ConditionCount; i++)
{
int charCode = word[word.Length - (entry.ConditionCount - i)];
if ((entry.Condition[charCode] & (1 << i)) == (1 << i))
{
passCount++;
}
else
{
break;
}
}
if (passCount == entry.ConditionCount)
{
int tempLen = word.Length - entry.StripCharacters.Length;
string tempWord = word.Substring(0, tempLen);
tempWord += entry.AddCharacters;
return(tempWord);
}
}
}
return(word);
}
public static string AddPrefix(string word, AffixRule rule)
{
foreach (AffixEntry entry in rule.AffixEntries)
{
if (word.Length >= entry.ConditionCount)
{
int passCount = 0;
for (int i = 0; i < entry.ConditionCount; i++)
{
int charCode = (int)word[i];
if ((entry.Condition[charCode] & (1 << i)) == (1 << i))
{
passCount++;
}
else
{
break;
}
}
if (passCount == entry.ConditionCount)
{
string tempWord = word.Substring(entry.StripCharacters.Length);
tempWord = entry.AddCharacters + tempWord;
return(tempWord);
}
}
}
return(word);
}
/// <summary>
/// Informs whether the newRule can be combined with appliedRule (to a word).
/// </summary>
/// <param name="newRule"></param>
/// <param name="appliedRule"></param>
/// <returns></returns>
private bool CanCombineRules(AffixRule newRule, AffixRule appliedRule)
{
if (newRule == null || appliedRule == null)
{
return(true);
}
// Only 1 rule of each type (SuffixRule of PrefixRule) can be combined.
if (newRule is SuffixRule && appliedRule is SuffixRule)
{
return(false);
}
if (newRule is PrefixRule && appliedRule is PrefixRule)
{
return(false);
}
// SuffixRule and PrefixRule can be comined if they are both combinable (CanCombine= true).
if (newRule.CanCombine && appliedRule.CanCombine)
{
return(true);
}
return(false);
}
/// <summary>
/// Adds a suffix to a word
/// </summary>
/// <param name="word" type="string">
/// <para>
/// The word to get the suffix added to
/// </para>
/// </param>
/// <param name="rule" type="NetSpell.SpellChecker.Dictionary.Affix.AffixRule">
/// <para>
/// The AffixRule to use when adding the suffix
/// </para>
/// </param>
/// <returns>
/// The word with the suffix added
/// </returns>
public static string AddSuffix(string word, AffixRule rule)
{
foreach (AffixEntry entry in rule.AffixEntries)
{
// check that this entry is valid
if (word.Length >= entry.ConditionCount)
{
int passCount = 0;
for (int i = 0; i < entry.ConditionCount; i++)
{
int charCode = (int)word[word.Length - (entry.ConditionCount - i)];
if ((entry.Condition[charCode] & (1 << i)) == (1 << i))
{
passCount++;
}
else
{
break;
}
}
if (passCount == entry.ConditionCount)
{
int tempLen = word.Length - entry.StripCharacters.Length;
string tempWord = word.Substring(0, tempLen);
tempWord += entry.AddCharacters;
return tempWord;
}
}
}
return word;
}
/// <summary>
///
/// </summary>
/// <param name="path"></param>
private void LoadAffixesFile(string path)
{
StreamReader f = File.OpenText(path);
string s;
char[] sepChars = new char[] { ' ', '\t' };
AffixRule curAffixRule = null;
affixRules = new AffixRulesCollection();
affixRulesLookup = new AffixRule['Z' + 1];
tryCharacters = "";
replacePatterns = new List <ReplacePattern> ();
while ((s = f.ReadLine()) != null)
{
// ignore comment
if (s.StartsWith("#"))
{
continue;
}
string[] data = s.Split(sepChars, StringSplitOptions.RemoveEmptyEntries);
if (data.Length == 0)
{
continue; // empty string
}
if (data[0] == "PFX" || data[0] == "SFX")
{
if (data.Length == 4)
{
// add new rule
curAffixRule = AffixRule.Parse(data);
affixRules.Add(curAffixRule);
affixRulesLookup[curAffixRule.Name] = curAffixRule;
}
else if (curAffixRule != null)
{
// add affix
curAffixRule.ParseAndAddAffix(data);
}
}
else
{
if (data[0] == "TRY")
{
tryCharacters = data[1];
}
else if (data[0] == "REP" && data.Length == 3)
{
replacePatterns.Add(new ReplacePattern(data[1], data[2]));
}
curAffixRule = null;
}
}
f.Close();
}
/// <summary>
/// Expands an affix compressed base word
/// </summary>
/// <param name="word" type="GuruComponents.Netrix.SpellChecker.NetSpell.Dictionary.Word">
/// <para>
/// The word to expand
/// </para>
/// </param>
/// <returns>
/// A System.Collections.ArrayList of words expanded from base word
/// </returns>
public ArrayList ExpandWord(Word word)
{
ArrayList suffixWords = new ArrayList();
ArrayList words = new ArrayList();
suffixWords.Add(word.Text);
string prefixKeys = "";
// check suffix keys first
foreach (char key in word.AffixKeys)
{
if (_suffixRules.ContainsKey(key.ToString(CultureInfo.CurrentUICulture)))
{
AffixRule rule = _suffixRules[key.ToString(CultureInfo.CurrentUICulture)];
string tempWord = AffixUtility.AddSuffix(word.Text, rule);
if (tempWord != word.Text)
{
if (rule.AllowCombine)
{
suffixWords.Add(tempWord);
}
else
{
words.Add(tempWord);
}
}
}
else if (_prefixRules.ContainsKey(key.ToString(CultureInfo.CurrentUICulture)))
{
prefixKeys += key.ToString(CultureInfo.CurrentUICulture);
}
}
// apply prefixes
foreach (char key in prefixKeys)
{
AffixRule rule = _prefixRules[key.ToString(CultureInfo.CurrentUICulture)];
// apply prefix to all suffix words
foreach (string suffixWord in suffixWords)
{
string tempWord = AffixUtility.AddPrefix(suffixWord, rule);
if (tempWord != suffixWord)
{
words.Add(tempWord);
}
}
}
words.AddRange(suffixWords);
//System.Diagnostics.Debug.WriteLine("Word Expanded: {0}; Child Words: {1}", word.Text, words.Count);
return(words);
}
/// <summary>
/// Expands an affix compressed base word
/// </summary>
/// <param name="word" type="NetSpell.SpellChecker.Dictionary.Word">
/// <para>
/// The word to expand
/// </para>
/// </param>
/// <returns>
/// A System.Collections.Generic.List<string> of words expanded from base word
/// </returns>
public List <string> ExpandWord(Word word)
{
List <string> suffixWords = new List <string>();
List <string> words = new List <string>();
suffixWords.Add(word.Text);
string prefixKeys = "";
// check suffix keys first
foreach (char key in word.AffixKeys)
{
if (_suffixRules.ContainsKey(key.ToString(CultureInfo.CurrentUICulture)))
{
AffixRule rule = _suffixRules[key.ToString(CultureInfo.CurrentUICulture)];
string tempWord = AffixUtility.AddSuffix(word.Text, rule);
if (tempWord != word.Text)
{
if (rule.AllowCombine)
{
suffixWords.Add(tempWord);
}
else
{
words.Add(tempWord);
}
}
}
else if (_prefixRules.ContainsKey(key.ToString(CultureInfo.CurrentUICulture)))
{
prefixKeys += key.ToString(CultureInfo.CurrentUICulture);
}
}
// apply prefixes
foreach (char key in prefixKeys)
{
AffixRule rule = _prefixRules[key.ToString(CultureInfo.CurrentUICulture)];
// apply prefix to all suffix words
foreach (string suffixWord in suffixWords)
{
string tempWord = AffixUtility.AddPrefix(suffixWord, rule);
if (tempWord != suffixWord)
{
words.Add(tempWord);
}
}
}
words.AddRange(suffixWords);
TraceWriter.TraceVerbose("Word Expanded: {0}; Child Words: {1}", word.Text, words.Count);
return(words);
}
/// <summary>
///
/// </summary>
/// <param name="word"></param>
/// <returns></returns>
private bool CheckWordInternal(string word)
{
if (words.ContainsKey(word))
{
return(true);
}
Dictionary <string, AffixRule> wordForms = new Dictionary <string, AffixRule> ();
wordForms.Add(word, null);
while (wordForms.Count > 0)
{
string checkingForm = wordForms.Keys.First();
AffixRule appliedAffixRule = wordForms[checkingForm];
for (int i = 0; i < affixRules.Count; ++i)
{
var affixRule = affixRules[i];
if (!CanCombineRules(affixRule, appliedAffixRule))
{
continue;
}
foreach (Affix affix in affixRule.Affixes)
{
string wordBase = affix.ApplyReverse(checkingForm);
if (wordBase != null)
{
// check the word presents in dictionary
SpellWord spellWord;
if (words.TryGetValue(wordBase, out spellWord))
{
if (spellWord.CanApplyRules(appliedAffixRule, affixRule))
{
return(true);
}
}
// add new word form
if (appliedAffixRule == null && !wordForms.ContainsKey(wordBase))
{
wordForms.Add(wordBase, affixRule);
}
}
}
}
// remove checked word
wordForms.Remove(checkingForm);
}
return(false);
}
/// <summary>
///
/// </summary>
/// <param name="rule"></param>
/// <returns></returns>
public bool CanApplyRule(AffixRule rule)
{
int n = AffixRules.Length;
char checkingRuleName = rule.Name;
for (int i = 0; i < n; ++i)
{
if (AffixRules[i].Name == checkingRuleName)
{
return(true);
}
}
return(false);
}
/// <summary>
///
/// </summary>
/// <param name="path"></param>
private void LoadWordsFile(string path)
{
StreamReader f = File.OpenText(path);
words = new Dictionary <string, SpellWord> ();
string word;
f.ReadLine(); // read 1st line with words count
while ((word = f.ReadLine()) != null)
{
int p = word.IndexOf('/');
SpellWord spellWord = new SpellWord();
AffixRulesCollection wordRules = new AffixRulesCollection();
if (p != -1)
{
int n = word.Length;
for (int i = p + 1; i < n; ++i)
{
char c = word[i];
if (c < affixRulesLookup.Length)
{
AffixRule rule = affixRulesLookup[c];
if (rule != null)
{
wordRules.Add(rule);
}
}
}
word = word.Substring(0, p);
}
spellWord.AffixRules = wordRules.ToArray <AffixRule>();
try
{
words.Add(word, spellWord);
}
catch (Exception)
{
}
}
f.Close();
}
/// <summary>
/// Initializes the dictionary by loading and parsing the
/// dictionary file and the user file.
/// </summary>
public void Initialize()
{
// clean up data first
_baseWords.Clear();
_replaceCharacters.Clear();
_prefixRules.Clear();
_suffixRules.Clear();
_phoneticRules.Clear();
_tryCharacters = "";
//******Below Lines Added*********
if (_dictionaryFile.StartsWith("."))
{
_dictionaryFile = Thread.CurrentThread.CurrentCulture.Name;
}
// add the default file extension
if (!_dictionaryFile.EndsWith(".dic"))
{
_dictionaryFile += ".dic";
}
//******Above Lines Added*********
// the following is used to split a line by white space
Regex _spaceRegx = new Regex(@"[^\s]+", RegexOptions.Compiled);
MatchCollection partMatches;
string currentSection = "";
AffixRule currentRule = null;
// Simplify our lives by assuming the default path
if (_dictionaryFolder == null)
{
_dictionaryFolder = Path.Combine(Path.GetDirectoryName(System.Reflection.Assembly.GetExecutingAssembly().Location), "Dictionary");
}
string dictionaryPath = Path.Combine(_dictionaryFolder, _dictionaryFile);
System.Diagnostics.Debug.WriteLine("Loading Dictionary:{0}", dictionaryPath);
// open dictionary file
if (!File.Exists(dictionaryPath))
{
throw new FileNotFoundException("Cannot find dictonary with location " + dictionaryPath);
}
//BinaryReader reader = new BinaryReader(File.Open("foo.txt", FileMode.Open),
// Encoding.GetEncoding(1252));
FileStream fs = new FileStream(dictionaryPath, FileMode.Open, FileAccess.Read, FileShare.Read);
StreamReader sr = new StreamReader(fs, true);
// read line by line
while (sr.Peek() >= 0)
{
string tempLine = sr.ReadLine().Trim();
if (tempLine.Length > 0)
{
// check for section flag
switch (tempLine)
{
case "[Copyright]":
case "[Try]":
case "[Replace]":
case "[Prefix]":
case "[Suffix]":
case "[Phonetic]":
case "[Words]":
// set current section that is being parsed
currentSection = tempLine;
break;
default:
// parse line and place in correct object
switch (currentSection)
{
case "[Copyright]":
this.Copyright += tempLine + "\r\n";
break;
case "[Try]": // ISpell try chars
this.TryCharacters += tempLine;
break;
case "[Replace]": // ISpell replace chars
this.ReplaceCharacters.Add(tempLine);
break;
case "[Prefix]": // MySpell prefix rules
case "[Suffix]": // MySpell suffix rules
// split line by white space
partMatches = _spaceRegx.Matches(tempLine);
// if 3 parts, then new rule
if (partMatches.Count == 3)
{
currentRule = new AffixRule();
// part 1 = affix key
currentRule.Name = partMatches[0].Value;
// part 2 = combine flag
if (partMatches[1].Value == "Y")
{
currentRule.AllowCombine = true;
}
// part 3 = entry count, not used
if (currentSection == "[Prefix]")
{
// add to prefix collection
this.PrefixRules.Add(currentRule.Name, currentRule);
}
else
{
// add to suffix collection
this.SuffixRules.Add(currentRule.Name, currentRule);
}
}
//if 4 parts, then entry for current rule
else if (partMatches.Count == 4)
{
// part 1 = affix key
if (currentRule.Name == partMatches[0].Value)
{
AffixEntry entry = new AffixEntry();
// part 2 = strip char
if (partMatches[1].Value != "0")
{
entry.StripCharacters = partMatches[1].Value;
}
// part 3 = add chars
if (partMatches[2].Value != "0")
{
entry.AddCharacters = partMatches[2].Value;
}
// part 4 = conditions
AffixUtility.EncodeConditions(partMatches[3].Value, entry);
currentRule.AffixEntries.Add(entry);
}
}
break;
case "[Phonetic]": // ASpell phonetic rules
// split line by white space
partMatches = _spaceRegx.Matches(tempLine);
if (partMatches.Count >= 2)
{
PhoneticRule rule = new PhoneticRule();
PhoneticUtility.EncodeRule(partMatches[0].Value, ref rule);
rule.ReplaceString = partMatches[1].Value;
_phoneticRules.Add(rule);
}
break;
case "[Words]": // dictionary word list
// splits word into its parts
string[] parts = tempLine.Split('/');
Word tempWord = new Word();
// part 1 = base word
tempWord.Text = parts[0];
// part 2 = affix keys
if (parts.Length >= 2)
{
tempWord.AffixKeys = parts[1];
}
// part 3 = phonetic code
if (parts.Length >= 3)
{
tempWord.PhoneticCode = parts[2];
}
this.BaseWords.Add(tempWord.Text, tempWord);
break;
} // currentSection swith
break;
} //tempLine switch
} // if templine
} // read line
// close files
sr.Close();
fs.Close();
//System.Diagnostics.Debug.WriteLine("Dictionary Loaded BaseWords:{0}; PrefixRules:{1}; SuffixRules:{2}; PhoneticRules:{3}",
// this.BaseWords.Count, this.PrefixRules.Count, this.SuffixRules.Count, this.PhoneticRules.Count);
this.LoadUserFile();
_initialized = true;
}
/// <summary>
/// Initializes the dictionary by loading and parsing the
/// dictionary file and the user file.
/// </summary>
public void Initialize()
{
// clean up data first
_baseWords.Clear();
_replaceCharacters.Clear();
_prefixRules.Clear();
_suffixRules.Clear();
_phoneticRules.Clear();
_tryCharacters = "";
// the following is used to split a line by white space
Regex _spaceRegx = new Regex(@"[^\s]+", RegexOptions.Compiled);
MatchCollection partMatches;
string currentSection = "";
AffixRule currentRule = null;
string dictionaryPath = Path.Combine(_dictionaryFolder, _dictionaryFile);
TraceWriter.TraceInfo("Loading Dictionary:{0}", dictionaryPath);
// open dictionary file
FileStream fs = null;
try
{
fs = new FileStream(dictionaryPath, FileMode.Open, FileAccess.Read, FileShare.Read);
using (var sr = new StreamReader(fs, Encoding.UTF8))
{
fs = null;
// read line by line
while (sr.Peek() >= 0)
{
string tempLine = sr.ReadLine().Trim();
if (tempLine.Length <= 0)
{
continue;
}
// check for section flag
if (tempLine.StartsWith("[") && tempLine.EndsWith("]"))
{
// set current section that is being parsed
currentSection = tempLine;
continue;
}
// parse line and place in correct object
switch (currentSection)
{
case "[Copyright]":
Copyright += tempLine + "\r\n";
break;
case "[Try]": // ISpell try chars
TryCharacters += tempLine;
break;
case "[Replace]": // ISpell replace chars
ReplaceCharacters.Add(tempLine);
break;
case "[Prefix]": // MySpell prefix rules
case "[Suffix]": // MySpell suffix rules
// split line by white space
partMatches = _spaceRegx.Matches(tempLine);
// if 3 parts, then new rule
if (partMatches.Count == 3)
{
currentRule = new AffixRule();
// part 1 = affix key
currentRule.Name = partMatches[0].Value;
// part 2 = combine flag
if (partMatches[1].Value == "Y")
{
currentRule.AllowCombine = true;
}
// part 3 = entry count, not used
if (currentSection == "[Prefix]")
{
// add to prefix collection
PrefixRules.Add(currentRule.Name, currentRule);
}
else
{
// add to suffix collection
SuffixRules.Add(currentRule.Name, currentRule);
}
}
//if 4 parts, then entry for current rule
else if (partMatches.Count == 4)
{
// part 1 = affix key
if (currentRule.Name == partMatches[0].Value)
{
AffixEntry entry = new AffixEntry();
// part 2 = strip char
if (partMatches[1].Value != "0")
{
entry.StripCharacters = partMatches[1].Value;
}
// part 3 = add chars
entry.AddCharacters = partMatches[2].Value;
// part 4 = conditions
AffixUtility.EncodeConditions(partMatches[3].Value, entry);
currentRule.AffixEntries.Add(entry);
}
}
break;
case "[Phonetic]": // ASpell phonetic rules
// split line by white space
partMatches = _spaceRegx.Matches(tempLine);
if (partMatches.Count >= 2)
{
PhoneticRule rule = new PhoneticRule();
PhoneticUtility.EncodeRule(partMatches[0].Value, ref rule);
rule.ReplaceString = partMatches[1].Value;
_phoneticRules.Add(rule);
}
break;
case "[Words]": // dictionary word list
// splits word into its parts
string[] parts = tempLine.Split('/');
Word tempWord = new Word();
// part 1 = base word
tempWord.Text = parts[0];
// part 2 = affix keys
if (parts.Length >= 2)
{
tempWord.AffixKeys = parts[1];
}
// part 3 = phonetic code
if (parts.Length >= 3)
{
tempWord.PhoneticCode = parts[2];
}
BaseWords.Add(tempWord.Text, tempWord);
break;
} // currentSection switch
} // read line
// close files
}
}
finally
{
if (fs != null)
{
fs.Dispose();
}
}
TraceWriter.TraceInfo("Dictionary Loaded BaseWords:{0}; PrefixRules:{1}; SuffixRules:{2}; PhoneticRules:{3}",
BaseWords.Count, PrefixRules.Count, SuffixRules.Count, PhoneticRules.Count);
LoadUserFile();
_initialized = true;
}
/// <summary>
/// Initializes the dictionary by loading and parsing the
/// dictionary file and the user file.
/// </summary>
public WordDictionary(TextReader inputDictionary)
{
for (byte i = 0; i < 128; i++)
{
_encodeTable[i] = i;
_decodeTable[i] = (char)i;
}
MatchCollection partMatches;
string currentSection = "";
AffixRule currentRule = null;
while (inputDictionary.Peek() >= 0)
{
string tempLine = encode(inputDictionary.ReadLine().Trim());
if (tempLine.Length > 0)
{
// check for section flag
switch (tempLine)
{
case "[Copyright]":
case "[Try]":
case "[Replace]":
case "[Prefix]":
case "[Suffix]":
case "[Phonetic]":
case "[Words]":
// set current section that is being parsed
currentSection = tempLine;
break;
default:
// parse line and place in correct object
switch (currentSection)
{
case "[Copyright]":
//this.Copyright += tempLine + "\r\n";
break;
case "[Try]": // ISpell try chars
_tryCharacters += tempLine;
break;
case "[Replace]": // ISpell replace chars
_replaceCharacters.Add(tempLine);
break;
case "[Prefix]": // MySpell prefix rules
case "[Suffix]": // MySpell suffix rules
// split line by white space
partMatches = _spaceRegx.Matches(tempLine);
// if 3 parts, then new rule
if (partMatches.Count == 3)
{
currentRule = new AffixRule();
// part 1 = affix key
currentRule.Name = partMatches[0].Value;
// part 2 = combine flag
if (partMatches[1].Value == "Y")
{
currentRule.AllowCombine = true;
}
// part 3 = entry count, not used
if (currentSection == "[Prefix]")
{
// add to prefix collection
_prefixRules.Add(currentRule.Name, currentRule);
}
else
{
// add to suffix collection
_suffixRules.Add(currentRule.Name, currentRule);
}
}
//if 4 parts, then entry for current rule
else if (partMatches.Count == 4)
{
// part 1 = affix key
if (currentRule.Name == partMatches[0].Value)
{
AffixEntry entry = new AffixEntry();
// part 2 = strip char
if (partMatches[1].Value != "0")
{
entry.StripCharacters = partMatches[1].Value;
}
// part 3 = add chars
entry.AddCharacters = partMatches[2].Value;
// part 4 = conditions
AffixUtility.EncodeConditions(partMatches[3].Value, entry);
currentRule.AffixEntries.Add(entry);
}
}
break;
case "[Phonetic]": // ASpell phonetic rules
// split line by white space
partMatches = _spaceRegx.Matches(tempLine);
if (partMatches.Count >= 2)
{
PhoneticRule rule = new PhoneticRule();
PhoneticUtility.EncodeRule(partMatches[0].Value, ref rule);
rule.ReplaceString = partMatches[1].Value;
_phoneticRules.Add(rule);
}
break;
case "[Words]": // dictionary word list
// splits word into its parts
string[] parts = tempLine.Split('/');
Word tempWord = new Word();
// part 1 = base word
tempWord.Text = parts[0];
// part 2 = affix keys
if (parts.Length >= 2)
{
tempWord.AffixKeys = parts[1];
}
// part 3 = phonetic code
if (parts.Length >= 3)
{
tempWord.PhoneticCode = parts[2];
}
_baseWords[tempWord.Text] = tempWord;
break;
}
break;
}
}
}
}
public void ToString(String affixClass, char affixClassFlag, String strippingCharacters, String affix, String condition, String morphologicalDescription, String expectedLineString)
{
AffixRule affixRule = new AffixRule(affixClassFlag, strippingCharacters, affix, condition, morphologicalDescription);
Assert.AreEqual(expectedLineString, affixRule.ToString(AffixClass.FromString(affixClass)));
}
