static public int Freq(string file)
{
StreamReader sr = new StreamReader(file, System.Text.Encoding.Default);
string s = sr.ReadToEnd();
List <WordRu> lRoot = Analize(s);
SyntaxRu.RemovePrepPunc(lRoot);
List <string> lNorm = new List <string>();
List <int> lFreq = new List <int>();
Stack <WordRu> st = new Stack <WordRu>();
foreach (WordRu r in lRoot)
{
st.Push(r);
while (st.Count > 0)
{
WordRu w = st.Pop();
int index = lNorm.BinarySearch(w.norm);
if (index < 0)
{
lNorm.Insert(~index, w.norm);
lFreq.Insert(~index, 1);
}
else
{
lFreq[index]++;
}
foreach (WordRu c in w.lCh)
{
st.Push(c);
}
}
}
DSDict.FreqDataTable ft = new DSDict.FreqDataTable();
DSDictTableAdapters.FreqTableAdapter fta = new Semantics.DSDictTableAdapters.FreqTableAdapter();
int count = 0;
for (int i = 0; i < lNorm.Count; i++)
{
fta.FillByWord(ft, lNorm[i]);
if (ft.Rows.Count == 0)
{
fta.Insert(lNorm[i], lFreq[i]);
}
else
{
fta.UpdateByWord(lNorm[i], (int)ft[0]["freq"] + lFreq[i], lNorm[i]);
}
count++;
}
return(count);
}
static public int Syn(List <WordRu> lRoot1, List <WordRu> lRoot2)
{
List <string> lNorm1 = new List <string>(), lNorm2 = new List <string>(), lNorm = new List <string>();
List <int> lFreq1 = new List <int>(), lFreq2 = new List <int>();
List <List <WordRu> > llW1 = new List <List <WordRu> >(), llW2 = new List <List <WordRu> >();
List <List <string> > llSyn1 = new List <List <string> >(), llSyn2 = new List <List <string> >();
List <List <int> > llSFreq1 = new List <List <int> >(), llSFreq2 = new List <List <int> >();
List <bool> lMatch1 = new List <bool>(), lMatch2 = new List <bool>();
Stack <WordRu> st = new Stack <WordRu>();
DSDictTableAdapters.FreqTableAdapter fta = new Semantics.DSDictTableAdapters.FreqTableAdapter();
DSDictTableAdapters.SynTableAdapter sta = new Semantics.DSDictTableAdapters.SynTableAdapter();
DSDict.FreqDataTable fdt = new DSDict.FreqDataTable();
DSDict.SynDataTable sdt = new DSDict.SynDataTable();
foreach (WordRu r in lRoot1)
{
st.Push(r);
while (st.Count > 0)
{
WordRu w = st.Pop();
int index = lNorm1.BinarySearch(w.norm);
if (index >= 0)
{
llW1[index].Add(w);
continue;
}
index = ~index;
lNorm1.Insert(index, w.norm);
if (fta.FillByWord(fdt, w.norm) > 0)
{
lFreq1.Insert(index, (int)fdt[0].freq);
}
else
{
lFreq1.Insert(index, 0);
}
sta.Fill(sdt, w.norm);
List <string> lSyn = new List <string>();
List <int> lSFreq = new List <int>();
foreach (DSDict.SynRow sr in sdt)
{
lSyn.Add((string)sr.word);
if (sr.IsfreqNull())
{
lSFreq.Add(0);
}
else
{
lSFreq.Add((int)sr.freq);
}
}
llW1.Insert(index, new List <WordRu>());
llW1[index].Add(w);
llSyn1.Insert(index, lSyn);
llSFreq1.Insert(index, lSFreq);
index = lNorm.BinarySearch(w.norm);
if (index < 0)
{
index = ~index;
lNorm.Insert(index, w.norm);
lMatch1.Insert(index, true);
lMatch2.Insert(index, false);
}
foreach (string syn in lSyn)
{
index = lNorm.BinarySearch(syn);
if (index < 0)
{
index = ~index;
lNorm.Insert(index, syn);
lMatch1.Insert(index, true);
lMatch2.Insert(index, false);
}
}
}
}
foreach (WordRu r in lRoot2)
{
st.Push(r);
while (st.Count > 0)
{
WordRu w = st.Pop();
int index = lNorm2.BinarySearch(w.norm);
if (index >= 0)
{
llW2[index].Add(w);
continue;
}
index = ~index;
lNorm2.Insert(index, w.norm);
if (fta.FillByWord(fdt, w.norm) > 0)
{
lFreq2.Insert(index, (int)fdt[0].freq);
}
else
{
lFreq2.Insert(index, 0);
}
sta.Fill(sdt, w.norm);
List <string> lSyn = new List <string>();
List <int> lSFreq = new List <int>();
foreach (DSDict.SynRow sr in sdt)
{
lSyn.Add((string)sr.word);
if (sr.IsfreqNull())
{
lSFreq.Add(0);
}
else
{
lSFreq.Add((int)sr.freq);
}
}
llW2.Insert(index, new List <WordRu>());
llW2[index].Add(w);
llSyn2.Insert(index, lSyn);
llSFreq2.Insert(index, lSFreq);
index = lNorm.BinarySearch(w.norm);
if (index < 0)
{
index = ~index;
lNorm.Insert(index, w.norm);
lMatch1.Insert(index, true);
lMatch2.Insert(index, false);
}
else
{
lMatch2[index] = true;
}
foreach (string syn in lSyn)
{
index = lNorm.BinarySearch(syn);
if (index < 0)
{
index = ~index;
lNorm.Insert(index, syn);
lMatch1.Insert(index, true);
lMatch2.Insert(index, false);
}
else
{
lMatch2[index] = true;
}
}
}
}
int count = 0;
for (int i = 0; i < lNorm1.Count; i++)
{
int j;
for (j = 0; j < llSyn1[i].Count; j++)
{
int k = lNorm.BinarySearch(llSyn1[i][j]);
if (lMatch1[k] && lMatch2[k])
{
break;
}
}
if (j == llSyn1[i].Count)
{
continue;
}
int index = lNorm.BinarySearch(lNorm1[i]);
if (!lMatch1[index] || !lMatch2[index] || llSFreq1[i][j] > lFreq1[i])
{
count += llW1[i].Count;
foreach (WordRu w in llW1[i])
{
w.norm = llSyn1[i][j];
}
}
}
for (int i = 0; i < lNorm2.Count; i++)
{
int j;
for (j = 0; j < llSyn2[i].Count; j++)
{
int k = lNorm.BinarySearch(llSyn2[i][j]);
if (lMatch1[k] && lMatch2[k])
{
break;
}
}
if (j == llSyn2[i].Count)
{
continue;
}
int index = lNorm.BinarySearch(lNorm2[i]);
if (!lMatch1[index] || !lMatch2[index] || llSFreq2[i][j] > lFreq2[i])
{
count += llW2[i].Count;
foreach (WordRu w in llW2[i])
{
w.norm = llSyn2[i][j];
}
}
}
return(count);
}
