// Get hash representing pair of files
private string GetHash(string fileName1, string fileName2, CloneSearchParameters searchParameters)
{
byte[] hash1;
byte[] hash2;
byte[] hash;
using (MD5 md5 = MD5.Create())
{
// Get hash of file 1 contents
using (FileStream fs = File.OpenRead(fileName1))
{
hash1 = md5.ComputeHash(fs);
}
// Get hash of file 2 contents
using (FileStream fs = File.OpenRead(fileName2))
{
hash2 = md5.ComputeHash(fs);
}
// Combine hashes and min clone length data into single hash
byte[] comb = hash1.Concat(hash2)
.Concat(BitConverter.GetBytes(searchParameters.MinLines))
.Concat(BitConverter.GetBytes(searchParameters.MinTokens))
.Concat(BitConverter.GetBytes(searchParameters.MinPercentMatch))
.ToArray();
hash = md5.ComputeHash(comb);
}
// Convert hash to hex
return(string.Concat(hash.Select(b => b.ToString("X2"))));
}
public CloneFinder(TokenList tokenList1, TokenList tokenList2, CloneSearchParameters searchParameters)
{
this.FileName1 = tokenList1.FileName;
this.TokenList1 = tokenList1.Tokens;
this.FileName2 = tokenList2.FileName;
this.TokenList2 = tokenList2.Tokens;
this.MinLines = searchParameters.MinLines;
this.MinTokens = searchParameters.MinTokens;
this.MatchThreshold = searchParameters.MinPercentMatch / 100;
}
// Check table for entry matching given filenames
public List <Clone> TryGet(string fileName1, string fileName2, CloneSearchParameters searchParameters)
{
string key = GetHash(fileName1, fileName2, searchParameters);
if (table.ContainsKey(key))
{
return(table[key]);
}
else
{
return(null);
}
}
// Add entry to table
public void AddEntry(string fileName1, string fileName2, CloneSearchParameters searchParameters, List <Clone> clones)
{
table.Add(GetHash(fileName1, fileName2, searchParameters), clones);
}
