public virtual List <string> GetSpellingVars(string separator)
{
List <string> spellingVars = new List <string>();
if (lexRecordObjs_ != null)
{
for (int i = 0; i < lexRecordObjs_.Count; i++)
{
LexRecord temp = (LexRecord)lexRecordObjs_[i];
string eui = temp.GetEui();
string category = temp.GetCategory();
spellingVars.Add(temp.GetBase() + separator + eui + separator + category);
List <string> tempSpellVars = temp.GetSpellingVars();
for (int j = 0; j < tempSpellVars.Count; j++)
{
string tempSpellVar = (string)tempSpellVars[j];
spellingVars.Add(tempSpellVar + separator + eui + separator + category);
}
}
}
return(spellingVars);
}
private static bool CheckGlreg(LexRecord lexRecord)
{
string @base = lexRecord.GetBase();
bool validFlag = CheckEntry.CheckGlreg(lexRecord, @base, 1);
return(validFlag);
}
public static string GetLexRecordInfo(LexRecord lexRecord)
{
string eui = lexRecord.GetEui();
string cat = lexRecord.GetCategory();
string @base = lexRecord.GetBase();
string info = eui + "|" + @base + "|" + cat;
return info;
}
public LexRecordNomObj(LexRecord lexRecord)
{
if (lexRecord != null)
{
base_ = lexRecord.GetBase();
eui_ = lexRecord.GetEui();
category_ = lexRecord.GetCategory();
nominalizations_ = lexRecord.GetNominalizations();
}
}
public static string GetItemFromLexRecord(LexRecord lexRecord, int contentType)
{
string outItem = "";
switch (contentType)
{
case 1:
outItem = lexRecord.GetBase();
break;
}
return outItem;
}
public virtual List <string> GetBases()
{
List <string> bases = new List <string>();
if (lexRecordObjs_ != null)
{
for (int i = 0; i < lexRecordObjs_.Count; i++)
{
LexRecord temp = (LexRecord)lexRecordObjs_[i];
bases.Add(temp.GetBase());
}
}
return(bases);
}
public virtual List <string> GetSpellingVars()
{
List <string> spellingVars = new List <string>();
if (lexRecordObjs_ != null)
{
for (int i = 0; i < lexRecordObjs_.Count; i++)
{
LexRecord temp = (LexRecord)lexRecordObjs_[i];
spellingVars.Add(temp.GetBase());
spellingVars.AddRange(temp.GetSpellingVars());
}
}
return(spellingVars);
}
private static bool CheckOrder(LexRecord lexRecord)
{
bool validFlag = true;
int contentType = 1;
string citation = lexRecord.GetBase();
List <string> spVars = lexRecord.GetSpellingVars();
List <string> bases = new List <string>();
bases.Add(citation);
foreach (string spVar in spVars)
{
bases.Add(spVar);
}
BaseComparator <string> bc = new BaseComparator <string>();
bases.Sort(bc);
if (!citation.Equals(bases[0]))
{
lexRecord.SetBase((string)bases[0]);
validFlag = false;
}
bases.RemoveAt(0);
lexRecord.SetSpellingVars(bases);
if (!validFlag)
{
ErrMsgUtilLexRecord.AddContentErrMsg(contentType, 6, citation, lexRecord);
}
return(validFlag);
}
private static bool CheckDuplicates(LexRecord lexRecord)
{
bool validFlag = true;
int contentType = 2;
string @base = lexRecord.GetBase();
List <string> svList = lexRecord.GetSpellingVars();
List <string> uSvList = new List <string>();
for (int i = 0; i < svList.Count; i++)
{
string sv = (string)svList[i];
if ((sv.Equals(@base) == true) || (uSvList.Contains(sv) == true))
{
validFlag = false;
ErrMsgUtilLexRecord.AddContentErrMsg(contentType, 2, sv, lexRecord);
}
else
{
uSvList.Add(sv);
}
}
if (!validFlag)
{
lexRecord.SetSpellingVars(uSvList);
}
return(validFlag);
}
//JAVA TO C# CONVERTER WARNING: Method 'throws' clauses are not available in .NET:
//ORIGINAL LINE: public static boolean Check(java.util.Vector<gov.nih.nlm.nls.lexCheck.Lib.LexRecord> lexRecords, java.util.Hashtable<String, java.util.HashSet<String>> dupRecExpList, java.io.BufferedWriter out, boolean verbose) throws java.io.IOException
public static bool Check(List <LexRecord> lexRecords, Dictionary <string, HashSet <string> > dupRecExpList,
System.IO.StreamWriter @out, bool verbose)
{
bool validFlag = true;
int recSize = lexRecords.Count;
for (int i = 0; i < recSize; i++)
{
LexRecord lexRecord = (LexRecord)lexRecords[i];
string cit = lexRecord.GetBase();
string cat = lexRecord.GetCategory();
string eui = lexRecord.GetEui();
string citCat = cit + "|" + cat;
bool dupFlag = AddToCitCatEuisTable(citCat, eui);
validFlag = (validFlag) && (dupFlag);
if (verbose == true)
{
if ((i % 100000 == 0) && (i > 0))
{
Console.WriteLine("- Loaded " + i + " lexRecords to table");
}
}
}
if (verbose == true)
{
Console.WriteLine("- Complete loaded " + recSize + " lexRecords to hash table for dupRec check");
}
if (!validFlag)
{
IEnumerator <string> keys = citCatEuisTable_.Keys.GetEnumerator();
while (keys.MoveNext() == true)
{
string key = (string)keys.Current;
HashSet <string> euis = (HashSet <string>)citCatEuisTable_[key];
bool dupFlag = false;
if (euis.Count > 1)
{
IEnumerator <string> it = euis.GetEnumerator();
while (it.MoveNext() == true)
{
string eui = (string)it.Current;
HashSet <string> dupRecExpEuis = (HashSet <string>)dupRecExpList[key];
if ((dupRecExpEuis == null) || (!dupRecExpEuis.Contains(eui)))
{
dupFlag = true;
}
}
if (dupFlag == true)
{
string errMsg = key;
IEnumerator <string> it2 = euis.GetEnumerator();
while (it2.MoveNext() == true)
{
errMsg = errMsg + "|" + (string)it2.Current;
}
@out.Write(errMsg + "|");
@out.WriteLine();
ErrMsgUtilLexicon.AddContentErrMsg(2, 2, errMsg);
}
}
}
}
return(validFlag);
}
/***********************************************************************************/
// The following methods map codes in the NIH Specialist Lexicon
// into the codes used in simplenlg
/***********************************************************************************/
/**
* get the simplenlg LexicalCategory of a record
*
* @param cat
* @return
*/
private LexicalCategory getSimplenlgCategory(LexRecord record)
{
string cat = record.GetCategory();
if (cat == null)
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.ANY));
}
else if (cat.Equals("noun", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.NOUN));
}
else if (cat.Equals("verb", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.VERB));
}
else if (cat.Equals("aux", StringComparison.OrdinalIgnoreCase) && string.Equals(record.GetBase(), "be", StringComparison.CurrentCultureIgnoreCase))
{ // return aux "be" as a VERB
// not needed for other aux "have" and "do", they have a verb entry
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.VERB));
}
else if (cat.Equals("adj", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.ADJECTIVE));
}
else if (cat.Equals("adv", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.ADVERB));
}
else if (cat.Equals("pron", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.PRONOUN));
}
else if (cat.Equals("det", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.DETERMINER));
}
else if (cat.Equals("prep", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.PREPOSITION));
}
else if (cat.Equals("conj", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.CONJUNCTION));
}
else if (cat.Equals("compl", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.COMPLEMENTISER));
}
else if (cat.Equals("modal", StringComparison.OrdinalIgnoreCase))
{
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.MODAL));
}
else
{ // return ANY for other cats
return(new LexicalCategory(LexicalCategory.LexicalCategoryEnum.ANY));
}
}
/**
* make a WordElement from a lexical record. Currently just specifies basic
* params and inflections Should do more in the future!
*
* @param record
* @return
*/
private WordElement makeWord(LexRecord record) // LexRecord
{
// get basic data
String baseForm = record.GetBase();
LexicalCategory category = record.GetSimpleNLGCategory(record);
String id = record.GetEui();
// create word class
WordElement wordElement = new WordElement(baseForm, category, id);
// now add type information
switch (category.GetLexicalCategory())
{
case LexicalCategory.LexicalCategoryEnum.ADJECTIVE:
addAdjectiveInfo(wordElement, record.GetCatEntry().GetAdjEntry());
break;
case LexicalCategory.LexicalCategoryEnum.ADVERB:
addAdverbInfo(wordElement, record.GetCatEntry().GetAdvEntry());
break;
case LexicalCategory.LexicalCategoryEnum.NOUN:
addNounInfo(wordElement, record.GetCatEntry().GetNounEntry());
break;
case LexicalCategory.LexicalCategoryEnum.VERB:
addVerbInfo(wordElement, record.GetCatEntry().GetVerbEntry());
break;
// ignore closed class words
}
Inflection?defaultInfl = wordElement.getDefaultInflectionalVariant();
// now add inflected forms
// if (keepStandardInflections || !standardInflections(record,
// category)) {
foreach (InflVar inflection in record.GetInflVarsAndAgreements().GetInflValues())
{
String simplenlgInflection = getSimplenlgInflection(inflection
.GetInflection());
if (simplenlgInflection != null)
{
String inflectedForm = inflection.GetVar();
Inflection?inflType = Inflection.REGULAR.getInflCode(inflection.GetType());
// store all inflectional variants, except for regular ones
// unless explicitly set
if (inflType != null &&
!(Inflection.REGULAR == inflType && !keepStandardInflections))
{
wordElement.addInflectionalVariant((Inflection)inflType,
simplenlgInflection, inflectedForm);
}
// if the infl variant is the default, also set this feature on
// the word
if (defaultInfl == null ||
(defaultInfl.Equals(inflType) && !(Inflection.REGULAR.Equals(inflType) && !keepStandardInflections)))
{
wordElement.setFeature(simplenlgInflection, inflectedForm);
}
// wordElement
// .setFeature(simplenlgInflection, inflection.GetVar());
}
}
// }
// add acronym info
addAcronymInfo(wordElement, record);
// now add spelling variants
addSpellingVariants(wordElement, record);
return(wordElement);
}
/***********************************************************************************/
// The following methods map codes in the NIH Specialist Lexicon
// into the codes used in simplenlg
/***********************************************************************************/
/**
* get the simplenlg LexicalCategory of a record
*
* @param cat
* @return
*/
private LexicalCategory getSimplenlgCategory(LexRecord record)
{
string cat = record.GetCategory();
if (cat == null)
{
return(new LexicalCategory_ANY());
}
else if (cat.equalsIgnoreCase("noun"))
{
return(new LexicalCategory_NOUN());
}
else if (cat.equalsIgnoreCase("verb"))
{
return(new LexicalCategory_VERB());
}
else if (cat.equalsIgnoreCase("aux") &&
record.GetBase().equalsIgnoreCase("be")) // return aux "be"
// as a VERB
// not needed for other aux "have" and "do", they have a verb entry
{
return(new LexicalCategory_VERB());
}
else if (cat.equalsIgnoreCase("adj"))
{
return(new LexicalCategory_ADJECTIVE());
}
else if (cat.equalsIgnoreCase("adv"))
{
return(new LexicalCategory_ADVERB());
}
else if (cat.equalsIgnoreCase("pron"))
{
return(new LexicalCategory_PRONOUN());
}
else if (cat.equalsIgnoreCase("det"))
{
return(new LexicalCategory_DETERMINER());
}
else if (cat.equalsIgnoreCase("prep"))
{
return(new LexicalCategory_PREPOSITION());
}
else if (cat.equalsIgnoreCase("conj"))
{
return(new LexicalCategory_CONJUNCTION());
}
else if (cat.equalsIgnoreCase("compl"))
{
return(new LexicalCategory_COMPLEMENTISER());
}
else if (cat.equalsIgnoreCase("modal"))
{
return(new LexicalCategory_MODAL());
}
// return ANY for other cats
else
{
return(new LexicalCategory_ANY());
}
}
private static bool CheckIrreg(LexRecord lexRecord, HashSet <string> irregExpEuiList)
{
bool validFlag = true;
List <string> variants = lexRecord.GetVariants();
string citation = lexRecord.GetBase();
List <string> svList = lexRecord.GetSpellingVars();
HashSet <string> baseList = new HashSet <string>(svList);
baseList.Add(citation);
HashSet <string> irregBases = new HashSet <string>();
string variant;
for (System.Collections.IEnumerator localIterator = variants.GetEnumerator(); localIterator.MoveNext();)
{
variant = (string)localIterator.Current;
if ((variant.StartsWith("irreg|")) || (variant.StartsWith("group(irreg|")))
{
string irregBase = GetIrregBase(variant);
if (!baseList.Contains(irregBase))
{
validFlag = false;
ErrMsgUtilLexRecord.AddContentErrMsg(5, 8, variant, lexRecord);
}
else
{
irregBases.Add(irregBase);
}
}
}
if (!validFlag)
{
return(validFlag);
}
if (irregBases.Count > 0)
{
string eui = lexRecord.GetEui();
if ((baseList.Count != irregBases.Count) && (!irregExpEuiList.Contains(eui)))
{
validFlag = false;
foreach (string @base in baseList)
{
if (!irregBases.Contains(@base))
{
ErrMsgUtilLexRecord.AddContentErrMsg(5, 9, @base, lexRecord);
}
}
}
}
return(validFlag);
}
