| public Exec ( string a, string b ) : string | ||
| a | string | 1st string |
| b | string | 2nd string |
| return | string | |
public string Exec(string a, string b)
{
if (a == null || b == null)
{
return null;
}
int x;
int y;
int maxx;
int maxy;
int[] go = new int[4];
const int X = 1;
const int Y = 2;
const int R = 3;
const int D = 0;
/*
* setup memory if needed => processing speed up
*/
maxx = a.Length + 1;
maxy = b.Length + 1;
if ((maxx >= sizex) || (maxy >= sizey))
{
sizex = maxx + 8;
sizey = maxy + 8;
net = RectangularArrays.ReturnRectangularIntArray(sizex, sizey);
way = RectangularArrays.ReturnRectangularIntArray(sizex, sizey);
}
/*
* clear the network
*/
for (x = 0; x < maxx; x++)
{
for (y = 0; y < maxy; y++)
{
net[x][y] = 0;
}
}
/*
* set known persistent values
*/
for (x = 1; x < maxx; x++)
{
net[x][0] = x;
way[x][0] = X;
}
for (y = 1; y < maxy; y++)
{
net[0][y] = y;
way[0][y] = Y;
}
for (x = 1; x < maxx; x++)
{
for (y = 1; y < maxy; y++)
{
go[X] = net[x - 1][y] + DELETE;
// way on x costs 1 unit
go[Y] = net[x][y - 1] + INSERT;
// way on y costs 1 unit
go[R] = net[x - 1][y - 1] + REPLACE;
go[D] = net[x - 1][y - 1]
+ ((a[x - 1] == b[y - 1]) ? NOOP : 100);
// diagonal costs 0, when no change
ushort min = (ushort)D;
if (go[min] >= go[X])
{
min = (ushort)X;
}
if (go[min] > go[Y])
{
min = (ushort)Y;
}
if (go[min] > go[R])
{
min = (ushort)R;
}
way[x][y] = min;
net[x][y] = (ushort)go[min];
}
}
// read the patch string
StringBuilder result = new StringBuilder();
char @base = (char)('a' - 1);
char deletes = @base;
char equals = @base;
for (x = maxx - 1, y = maxy - 1; x + y != 0;)
{
switch (way[x][y])
{
case X:
if (equals != @base)
{
result.Append("-" + (equals));
equals = @base;
}
deletes++;
x--;
break;
// delete
case Y:
if (deletes != @base)
{
result.Append("D" + (deletes));
deletes = @base;
}
if (equals != @base)
{
result.Append("-" + (equals));
equals = @base;
}
result.Append('I');
result.Append(b[--y]);
break;
// insert
case R:
if (deletes != @base)
{
result.Append("D" + (deletes));
deletes = @base;
}
if (equals != @base)
{
result.Append("-" + (equals));
equals = @base;
}
result.Append('R');
result.Append(b[--y]);
x--;
break;
// replace
case D:
if (deletes != @base)
{
result.Append("D" + (deletes));
deletes = @base;
}
equals++;
x--;
y--;
break;
// no change
}
}
if (deletes != @base)
{
result.Append("D" + (deletes));
deletes = @base;
}
return result.ToString();
}
} /// <summary>
/// Entry point to the DiffIt application.
/// <para>
/// This application takes one argument, the path to a file containing a
/// stemmer table. The program reads the file and generates the patch commands
/// for the stems.
/// </para>
/// </summary>
/// <param name="args">the path to a file containing a stemmer table</param>
public static void Main(string[] args)
{
int ins = Get(0, args[0]);
int del = Get(1, args[0]);
int rep = Get(2, args[0]);
int nop = Get(3, args[0]);
// LUCENENET specific - reformatted with :
string charset = SystemProperties.GetProperty("egothor:stemmer:charset", "UTF-8");
var stemmerTables = new List <string>();
// LUCENENET specific
// command line argument overrides environment variable or default, if supplied
for (int i = 1; i < args.Length; i++)
{
if ("-e".Equals(args[i], StringComparison.Ordinal) || "--encoding".Equals(args[i], StringComparison.Ordinal))
{
charset = args[i];
}
else
{
stemmerTables.Add(args[i]);
}
}
foreach (var stemmerTable in stemmerTables)
{
// System.out.println("[" + args[i] + "]");
Diff diff = new Diff(ins, del, rep, nop);
using (TextReader input = new StreamReader(new FileStream(stemmerTable, FileMode.Open, FileAccess.Read), Encoding.GetEncoding(charset)))
{
string line;
while ((line = input.ReadLine()) != null)
{
try
{
line = line.ToLowerInvariant();
StringTokenizer st = new StringTokenizer(line);
st.MoveNext();
string stem = st.Current;
Console.WriteLine(stem + " -a");
while (st.MoveNext())
{
string token = st.Current;
if (token.Equals(stem, StringComparison.Ordinal) == false)
{
Console.WriteLine(stem + " " + diff.Exec(token, stem));
}
}
}
catch (InvalidOperationException /*x*/)
{
// no base token (stem) on a line
}
}
}
}
}
