/* find the example sentence references in the example sentence index file */
bool findExample()
{
bool retval = false;
StreamReader fp = new StreamReader(WNDB.path + "SENTIDX.VRB");
Byte[] b = System.Text.Encoding.Unicode.GetBytes(fp.ReadToEnd());
MemoryStream ms = new MemoryStream(b);
//StreamReader fp = new StreamReader(mms);
int wdnum = whichword - 1;
Lexeme lx = words[wdnum];
string tbuf = lx.word + "%" + pos.ident + ":" + fnum + ":" + lx.uniq + "::";
string str = WNDB.binSearch(tbuf, ms); //fp);
if (str != null)
{
str = str.Substring(lx.word.Length + 11);
StrTok st = new StrTok(str, ' ', ',', '\n');
string offset;
while ((offset = st.next()) != null)
{
getExample(offset, lx.word);
retval = true;
}
}
fp.Close();
return(retval);
}
} // for serialization
public SynSet(int off, PartOfSpeech p, string wd, Search sch, int sens)
{
pos = p;
hereiam = off;
search = sch;
sense = sens;
StreamReader f = WNDB.data(p);
f.DiscardBufferedData();
//f.BaseStream.Seek(off,SeekOrigin.Begin);
f.BaseStream.Position = off;
string rec = f.ReadLine();
//string str;
if (!rec.StartsWith(off.ToString("D8")))
{
Console.WriteLine("Error reading " + p.name + " file! " + off + ": " + rec);
WNDB.reopen(p);
f = WNDB.data(p);
f.DiscardBufferedData();
// f.BaseStream.Seek(off,SeekOrigin.Begin);
f.BaseStream.Position = off;
rec = f.ReadLine();
//if (!rec.StartsWith(off.ToString("D8")))
// str = "";
//else
// Console.WriteLine("Recovered...");
}
Parse(rec, pos, wd);
}
string morphword(string word)
{
string end = "";
string tmpbuf = "";
if (word == null)
{
return(null);
}
Exceptions e = new Exceptions(word, pos);
string tmp = e.next();
if (tmp != null)
{
return(tmp);
}
if (pos.name == "adverb")
{
return(null);
}
if (pos.name == "noun")
{
if (word.EndsWith("ful"))
{
tmpbuf = word.Substring(0, word.Length - 3);
end = "ful";
}
else if (word.EndsWith("ss") || word.Length <= 2)
{
return(null);
}
}
if (tmpbuf.Length == 0)
{
tmpbuf = word;
}
int offset = offsets[pos.ident];
int cnt = cnts[pos.ident];
for (int i = 0; i < cnt; i++)
{
if (tmpbuf.EndsWith(sufx[i + offset]))
{
// TDMS 11 Oct 2005 - bug fix - "word" substituted with "tmpbuf" as per
// wordnet code morph.c
//string retval = word.Substring(0,word.Length-sufx[i+offset].Length)+addr[i+offset];
string retval = tmpbuf.Substring(0, tmpbuf.Length - sufx[i + offset].Length) + addr[i + offset];
if (WNDB.is_defined(retval, pos).NonEmpty)
{
return(retval + end);
}
}
}
return(null);
}
static Exceptions()
{
excfps = new StreamReader[PartOfSpeech.parts.Count];
IDictionaryEnumerator d = PartOfSpeech.parts.GetEnumerator();
while (d.MoveNext())
{
PartOfSpeech p = (PartOfSpeech)(d.Value);
excfps[p.ident] = new StreamReader(WNDB.ExcFile(p));
}
}
void getExample(string off, string wd)
{
StreamReader fp = new StreamReader(WNDB.path + "SENTS.VRB");
Byte[] b = System.Text.Encoding.Unicode.GetBytes(fp.ReadToEnd());
MemoryStream ms = new MemoryStream(b);
string line = WNDB.binSearch(off, ms); //fp);
line = line.Substring(line.IndexOf(' ') + 1);
search.buf += " EX: " + line.Replace("%s", wd);
fp.Close();
isDirty = true; // TDMS 19 July 2006 - attempt to tie the logic which
// populates buf to the logic that defines whether the
// synset is populated with relevant information
}
public Exceptions(string word, PartOfSpeech pos)
{
if (laststr != ((System.IO.FileStream)(excfps[pos.ident].BaseStream)).Name)
{
laststr = ((System.IO.FileStream)(excfps[pos.ident].BaseStream)).Name;
StreamReader fs = excfps[pos.ident];
Byte[] b = System.Text.Encoding.Unicode.GetBytes(fs.ReadToEnd());
excfps[pos.ident].BaseStream.Seek(0, SeekOrigin.Begin);
exc = new MemoryStream(b);
}
line = WNDB.binSearch(word, exc); //excfps[pos.ident]);
if (line != null)
{
endlp = line.IndexOf(' ');
}
}
public WNDBpart(PartOfSpeech p)
{
try
{
Console.WriteLine("WNDBpart");
index = new StreamReader(WNDB.IndexFile(p));
data = new StreamReader(WNDB.DataFile(p));
fps[p] = this;
}
catch
{
MessageBox.Show("Bad dict path");
// TODO: handle bad dict path
// ignore errors - as the user is locating the dictionary location
// wordnet classes are trying to instantiate based on an incorrect dict path
}
}
/* From search.c:
* Find word in index file and return parsed entry in data structure.
* Input word must be exact match of string in database. */
// From the WordNet Manual (http://wordnet.princeton.edu/man/wnsearch.3WN.html)
// index_lookup() finds searchstr in the index file for pos and returns a pointer
// to the parsed entry in an Index data structure. searchstr must exactly match the
// form of the word (lower case only, hyphens and underscores in the same places) in
// the index file. NULL is returned if a match is not found.
public static Index lookup(string word, PartOfSpeech pos)
{
int j;
if (word == "")
{
return(null);
}
// TDMS 14 Aug 2005 - changed to allow for numbers as well
// because the database contains searches that can start with
// numerals
//if (!char.IsLetter(word[0]))
if (!char.IsLetter(word[0]) && !char.IsNumber(word[0]))
{
return(null);
}
string line = WNDB.binSearch(word, pos);
if (line == null)
{
return(null);
}
Index idx = new Index();
StrTok st = new StrTok(line);
idx.wd = st.next(); /* the word */
idx.pos = PartOfSpeech.of(st.next()); /* the part of speech */
idx.sense_cnt = int.Parse(st.next()); /* collins count */
int ptruse_cnt = int.Parse(st.next()); /* number of pointers types */
idx.ptruse = new PointerType[ptruse_cnt];
for (j = 0; j < ptruse_cnt; j++)
{
idx.ptruse[j] = PointerType.of(st.next());
}
int off_cnt = int.Parse(st.next());
idx.offs = new int[off_cnt];
idx.tagsense_cnt = int.Parse(st.next());
for (j = 0; j < off_cnt; j++)
{
idx.offs[j] = int.Parse(st.next());
}
return(idx);
}
// From the WordNet Manual (http://wordnet.princeton.edu/man/wnsearch.3WN.html)
// findtheinfo() is the primary search algorithm for use with database interface
// applications. Search results are automatically formatted, and a pointer to the
// text buffer is returned. All searches listed in WNHOME/include/wnconsts.h can be
// done by findtheinfo().
void findtheinfo()
{
SynSet cursyn = null;
Indexes ixs = new Indexes(word, pos);
Index idx = null;
int depth = sch.rec ? 1 : 0;
senses = new SynSetList();
switch (sch.ptp.mnemonic)
{
case "OVERVIEW":
WNOverview();
break;
case "FREQ":
if (countSenses == null)
{
countSenses = new ArrayList();
}
while ((idx = ixs.next()) != null)
{
countSenses.Add(idx.offs.Length);
buf += "Sense " + countSenses.Count + ": " +
idx.offs.Length;
}
break;
case "WNGREP":
strings = WNDB.wngrep(word, pos);
for (int wi = 0; wi < strings.Count; wi++)
{
buf += (string)strings[wi] + "\n";
}
break;
case "VERBGROUP":
goto case "RELATIVES";
case "RELATIVES":
while ((idx = ixs.next()) != null)
{
relatives(idx);
}
break;
default:
/* look at all spellings of word */
while ((idx = ixs.next()) != null)
{
/* Print extra sense msgs if looking at all senses */
if (whichsense == ALLSENSES)
{
buf += "\n";
}
/* Go through all of the searchword's senses in the
* database and perform the search requested. */
for (int sense = 0; sense < idx.offs.Length; sense++)
{
if (whichsense == ALLSENSES || whichsense == sense + 1)
{
prflag = false;
/* Determine if this synset has already been done
* with a different spelling. If so, skip it. */
for (int j = 0; j < senses.Count; j++)
{
SynSet ss = (SynSet)senses[j];
if (ss.hereiam == idx.offs[sense])
{
goto skipit;
}
}
cursyn = new SynSet(idx, sense, this);
//TODO: moved senses.add(cursyn) from here to each case and handled it differently according to search - this handling needs to be verified to ensure the filter is not to limiting
switch (sch.ptp.mnemonic)
{
case "ANTPTR":
if (pos.name == "adj")
{
cursyn.traceAdjAnt();
}
else
{
cursyn.tracePtrs(sch.ptp, pos, depth);
}
if (cursyn.isDirty)
{ // TDMS 25 Oct 2005 - restrict to relevant values
cursyn.frames.Clear(); // TDMS 03 Jul 2006 - frames get added in wordnet.cs after filtering
senses.Add(cursyn);
}
/*
* if (cursyn.senses != null )
* if (cursyn.senses.isDirty)
* { // TDMS 25 Oct 2005 - restrict to relevant values
* cursyn.frames.Clear(); // TDMS 03 Jul 2006 - frames get added in wordnet.cs after filtering
* senses.Add(cursyn);
* }
*/
// perform the senses restrictions based upon pos
/*
* switch(pos.name) {
* case "verb":
* if (cursyn.senses != null) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
* break;
*
* default:
* if (cursyn.senses != null && cursyn.sense != 0) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
* break;
* }
*/
break;
case "COORDS":
//eg. search for 'car', select Noun -> 'Coordinate Terms'
cursyn.traceCoords(PointerType.of("HYPOPTR"), pos, depth);
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.senses != null )
* if (cursyn.senses.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
break;
case "FRAMES":
//eg. search for 'right', select Verb -> 'Sample Sentences'
cursyn.strFrame(true);
// TDMS 03 JUL 2006 fixed relevancy check if (cursyn.sense != 0) // TDMS 25 Oct 2005 - restrict to relevant values
if (cursyn.isDirty)
{
senses.Add(cursyn);
}
/*
* if (cursyn.frames.Count != 0) // TDMS 03 Jul 2006 - only add frame if there are any retrieved
* senses.Add(cursyn);
*/
break;
case "MERONYM":
//eg. search for 'car', select Noun -> 'Meronym'
senses.isDirty = false;
cursyn.tracePtrs(PointerType.of("HASMEMBERPTR"), pos, depth);
cursyn.tracePtrs(PointerType.of("HASSTUFFPTR"), pos, depth);
cursyn.tracePtrs(PointerType.of("HASPARTPTR"), pos, depth);
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.senses != null )
* if (cursyn.senses.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
break;
case "HOLONYM":
//eg. search for 'car', select Noun -> 'Holonyms'
cursyn.tracePtrs(PointerType.of("ISMEMBERPTR"), pos, depth);
cursyn.tracePtrs(PointerType.of("ISSTUFFPTR"), pos, depth);
cursyn.tracePtrs(PointerType.of("ISPARTPTR"), pos, depth);
// if (cursyn.senses != null && cursyn.sense != 0) // TDMS 25 Oct 2005 - restrict to relevant values
// senses.Add(cursyn);
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.senses != null )
* if (cursyn.senses.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
break;
case "HMERONYM":
//eg. search for 'car', select Noun -> 'Meronyms Tree'
cursyn.partsAll(sch.ptp);
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
// senses.Add(SearchTrack.ssParent);
/*
* if (cursyn.senses != null )
* if (cursyn.senses.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
// if (cursyn.sense != 0) // TDMS 25 Oct 2005 - restrict to relevant values
// senses.Add(cursyn);
break;
case "HHOLONYM":
cursyn.partsAll(sch.ptp);
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.senses != null) // && cursyn.sense != 0) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
break;
case "SEEALSOPTR":
cursyn.seealso();
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.sense != 0) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
break;
case "SIMPTR":
goto case "HYPERPTR";
case "SYNS":
goto case "HYPERPTR";
case "HYPERPTR":
//eg. search for 'car', select Noun -> 'Synonyms/Hypernyms, ordered by estimated frequency'
wordsFrom(cursyn);
cursyn.strsns(sense + 1);
prflag = true;
cursyn.tracePtrs(sch.ptp, pos, depth);
if (pos.name == "adj")
{
cursyn.tracePtrs(PointerType.of("PERTPTR"), pos, depth);
cursyn.tracePtrs(PointerType.of("PPLPTR"), pos, depth);
}
else if (pos.name == "adv")
{
cursyn.tracePtrs(PointerType.of("PERTPTR"), pos, depth);
}
if (pos.name == "verb")
{
cursyn.strFrame(false);
}
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
// senses.Add(cursyn);
break;
case "NOMINALIZATIONS": // 26/8/05 - changed "DERIVATION" to "NOMINALIZATIONS" - this needs to be verified
// derivation - TDMS
cursyn.tracenomins(sch.ptp);
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.sense != 0) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
break;
case "CLASSIFICATION":
goto case "CLASS";
case "CLASS":
//eg. search for 'car', select Noun -> 'Domain Terms'
cursyn.traceclassif(sch.ptp, new SearchType(false, sch.ptp));
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.senses != null )
* if (cursyn.senses.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
break;
case "HYPOPTR":
//eg. search for 'car', select Noun -> 'Hyponyms'
cursyn.tracePtrs(sch.ptp, pos, depth);
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.senses != null )
* if (cursyn.senses.isDirty)
* { // TDMS 25 Oct 2005 - restrict to relevant values
* cursyn.frames.Clear(); // TDMS 03 Jul 2006 - frames get added in wordnet.cs after filtering
* senses.Add(cursyn);
* }
*/
break;
default:
cursyn.tracePtrs(sch.ptp, pos, depth);
if (cursyn.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
{
senses.Add(cursyn);
}
/*
* if (cursyn.senses != null )
* if (cursyn.senses.isDirty) // TDMS 25 Oct 2005 - restrict to relevant values
* senses.Add(cursyn);
*/
break;
}
skipit :;
}
}
}
break;
}
}
public string next()
{
string word, tmp;
int prep, cnt, st_idx = 0, end_idx = 0, end_idx1, end_idx2;
string append = "";
/* first time through for this string */
if (firsttime)
{
firsttime = false;
cnt = str.Split('_').Length;
svprep = 0;
/* first try exception list */
e = new Exceptions(str, pos);
if ((tmp = e.next()) != null && tmp != str)
{
svcnt = 1; /* force next time to pass NULL */
return(tmp);
}
/* then try simply morph on original string */
if (pos.name != "verb" && ((tmp = morphword(str)) != null) && str != tmp)
{
return(tmp);
}
if (pos.name == "verb" && cnt > 1 && (prep = hasprep(str, cnt)) != 0)
{
svprep = prep;
return(morphprep(str));
}
else
{
svcnt = cnt = str.Split('_').Length;
while (--cnt > 0)
{
end_idx1 = str.Substring(st_idx).IndexOf('_') + st_idx;
end_idx2 = str.Substring(st_idx).IndexOf('-') + st_idx;
if (end_idx1 >= st_idx && end_idx2 >= st_idx)
{
if (end_idx1 < end_idx2)
{
end_idx = end_idx1;
append = "_";
}
else
{
end_idx = end_idx2;
append = "-";
}
}
else if (end_idx1 >= st_idx)
{
end_idx = end_idx1;
append = "_";
}
else
{
end_idx = end_idx2;
append = "-";
}
if (end_idx < 0)
{
return(null);
}
word = str.Substring(st_idx, end_idx - st_idx);
if ((tmp = morphword(word)) != null)
{
searchstr += tmp;
}
else
{
searchstr += word;
}
searchstr += append;
st_idx = end_idx + 1;
}
word = str.Substring(st_idx);
if ((tmp = morphword(word)) != null)
{
searchstr += tmp;
}
else
{
searchstr += word;
}
if (searchstr != str && WNDB.is_defined(searchstr, pos).NonEmpty)
{
return(searchstr);
}
else
{
return(null);
}
}
}
else // not firsttime
{
if (svprep > 0)
{
svprep = 0;
return(null);
}
else if (svcnt == 1)
{
return(e.next());
}
else
{
svcnt = 1;
e = new Exceptions(str, pos);
if ((tmp = e.next()) != null && tmp != str)
{
return(tmp);
}
return(null);
}
}
}
string morphprep(string s)
{
string excWord, lastwd = null;
int i, offset, cnt, rest, last;
string word, end, retval;
/* Assume that the verb is the first word in the phrase. Strip it
* off, check for validity, then try various morphs with the
* rest of the phrase tacked on, trying to find a match. */
rest = s.IndexOf('_');
last = s.LastIndexOf('_');
end = "";
if (rest != last)
{ // more than 2 words
lastwd = morphword(s.Substring(last + 1));
if (lastwd != null)
{
end = s.Substring(rest, last - rest + 1) + lastwd;
}
}
word = s.Substring(0, rest);
for (i = 0; i < word.Length; i++)
{
if (!char.IsLetterOrDigit(word[i]))
{
return(null);
}
}
offset = offsets[PartOfSpeech.of("verb").ident];
cnt = cnts[PartOfSpeech.of("verb").ident];
/* First try to find the verb in the exception list */
Exceptions e = new Exceptions(word, PartOfSpeech.of("verb"));
while ((excWord = e.next()) != null && excWord != word)
{
retval = excWord + s.Substring(rest);
if (WNDB.is_defined(retval, PartOfSpeech.of("verb")).NonEmpty)
{
return(retval);
}
else if (lastwd != null)
{
retval = excWord + end;
if (WNDB.is_defined(retval, PartOfSpeech.of("verb")).NonEmpty)
{
return(retval);
}
}
}
for (i = 0; i < cnt; i++)
{
if ((excWord = wordbase(word, i + offset)) != null && excWord != word) // ending is different
{
retval = excWord + s.Substring(rest);
if (WNDB.is_defined(retval, PartOfSpeech.of("verb")).NonEmpty)
{
return(retval);
}
else if (lastwd != null)
{
retval = excWord + end;
if (WNDB.is_defined(retval, PartOfSpeech.of("verb")).NonEmpty)
{
return(retval);
}
}
}
}
retval = word + s.Substring(rest);
if (s != retval)
{
return(retval);
}
if (lastwd != null)
{
retval = word + end;
if (s != retval)
{
return(retval);
}
}
return(null);
}
public static string binSearch(string word, PartOfSpeech pos)
{
return(binSearch(word, WNDB.index(pos)));
}