private void sort_dedupe_file(RECORD_TYPE type, string input, string output, DEDUP_SORT_ORDER order)
{
Item ix = null;
switch (type)
{
case RECORD_TYPE.FINGERPRINT_RECORD_CLOG:
ix = new fingerprintCLOG(order);
break;
case RECORD_TYPE.FINGERPRINT_RECORD_FPDB:
ix = new fingerprintFPDB(order);
break;
case RECORD_TYPE.FINGERPRINT_RECORD_MSG:
ix = new fingerprintDMSG(order);
break;
}
SortAPI sapi = new SortAPI(input, output, ix);
sapi.do_chunk_sort();
sapi.do_merge_work();
sapi.close_streams();
}
private void MSG_AGGREGATE(fingerprintDMSG msg)
{
if (msg == null)
{
if (counter > 0)
{
REDDY.ptrIFSDMux.DoDedupeBatch(msglist);
}
DEFS.DEBUGYELLOW("BATCH", "DoDedupeBatch, counter = " + counter);
for (int i = 0; i < 1024; i++)
{
msglist[i] = null;
}
counter = 0;
top = -1;
return;
}
else if ((top == -1) || (msg.fsid == msglist[top].fsid && msg.inode == msglist[top].inode &&
OPS.SomeFBNToStartFBN(1, msg.fbn) == OPS.SomeFBNToStartFBN(1, msglist[top].fbn)))
{
int idx = (int)(msg.fbn % 1024);
DEFS.ASSERT(msglist[idx] == null, "This cannot be populated already");
msglist[idx] = DUP(msg);
top = idx;
counter++;
}
else
{
//send msg here.
MSG_AGGREGATE(null);
MSG_AGGREGATE(msg);
return;
}
}
int IComparer.Compare(object obj1, object obj2)
{
switch (((Item)obj1).get_itemtype())
{
case RECORD_TYPE.FINGERPRINT_RECORD_CLOG:
{
fingerprintCLOG c1 = (fingerprintCLOG)obj1;
fingerprintCLOG c2 = (fingerprintCLOG)obj2;
if (c1.dbn < c2.dbn)
{
return(-1);
}
else if (c1.dbn > c2.dbn)
{
return(1);
}
return(0);
}
//break; unreachable
case RECORD_TYPE.FINGERPRINT_RECORD_FPDB:
{
fingerprintFPDB c1 = (fingerprintFPDB)obj1;
fingerprintFPDB c2 = (fingerprintFPDB)obj2;
if (c1.dbn < c2.dbn)
{
return(-1);
}
else if (c1.dbn > c2.dbn)
{
return(1);
}
return(0);
}
case RECORD_TYPE.FINGERPRINT_RECORD_MSG:
{
fingerprintDMSG c1 = (fingerprintDMSG)obj1;
fingerprintDMSG c2 = (fingerprintDMSG)obj2;
if (c1.sourcedbn < c2.sourcedbn)
{
return(-1);
}
else if (c1.sourcedbn > c2.sourcedbn)
{
return(1);
}
return(0);
}
//break;
}
DEFS.ASSERT(false, "Shouldnt have come here 3423423");
return(0);
}
private fingerprintDMSG DUP(fingerprintDMSG msg2)
{
fingerprintDMSG obj = new fingerprintDMSG(DEDUP_SORT_ORDER.UNDEFINED_PLACEHOLDER);
obj.fsid = msg2.fsid;
obj.inode = msg2.inode;
obj.fbn = msg2.fbn;
obj.sourcedbn = msg2.sourcedbn;
obj.destinationdbn = msg2.destinationdbn;
//obj.fp
return(obj);
}
private void print_file(RECORD_TYPE type, DEDUP_SORT_ORDER order, string fpath, string txtfile)
{
Item record = null;
switch (type)
{
case RECORD_TYPE.FINGERPRINT_RECORD_CLOG:
record = new fingerprintCLOG(order);
break;
case RECORD_TYPE.FINGERPRINT_RECORD_FPDB:
record = new fingerprintFPDB(order);
break;
case RECORD_TYPE.FINGERPRINT_RECORD_MSG:
record = new fingerprintDMSG(order);
break;
}
FileStream fsrc = new FileStream(fpath, FileMode.Open);
FileStream fdest = new FileStream(txtfile, FileMode.Create);
StreamWriter log = new StreamWriter(fdest);
int count = (int)(fsrc.Length / record.get_size());
byte[] buffer = new byte[record.get_size()];
for (int i = 0; i < count; i++)
{
fsrc.Read(buffer, 0, record.get_size());
record.parse_bytes(buffer, 0);
log.WriteLine(record.get_string_rep());
}
log.Flush();
log.Close();
fsrc.Close();
}
/*
* Send the messages and get work done.
*/
private void ACTUAL_DEDUPE(string script)
{
FileStream fsrc = new FileStream(script, FileMode.Open);
Item record = new fingerprintDMSG(DEDUP_SORT_ORDER.UNDEFINED_PLACEHOLDER);
int count = (int)(fsrc.Length / record.get_size());
byte[] buffer = new byte[record.get_size()];
for (int i = 0; i < count; i++)
{
fsrc.Read(buffer, 0, record.get_size());
record.parse_bytes(buffer, 0);
fingerprintDMSG msg = (fingerprintDMSG)record;
//REDDY.ptrIFSDMux.DoDedupe(msg.fsid, msg.inode, msg.fbn, msg.sourcedbn, msg.destinationdbn);
MSG_AGGREGATE(msg);
int progress = (i * 100) / count;
ui.Update_DedupeUI(m_stage, progress);
}
MSG_AGGREGATE(null);
fsrc.Close();
}
/*
* The crux.
*/
private void GenDedupeScript(string fpdbdata, string clogdata, string mergedfile, string scriptfilepath)
{
int existing_dbn_counter = 0;
DEFS.DEBUG("DEDUPE", "starting dedupe op genscript");
FileStream fsrcfpdb = null;
if (File.Exists(fpdbdata) == true)
{
fsrcfpdb = new FileStream(fpdbdata, FileMode.Open);
}
FileStream fsrcclog = new FileStream(clogdata, FileMode.Open);
FileStream fdest = new FileStream(mergedfile, FileMode.Create);
FileStream scriptfile = new FileStream(scriptfilepath, FileMode.Create);
fingerprintCLOG fpclog = new fingerprintCLOG(DEDUP_SORT_ORDER.UNDEFINED_PLACEHOLDER);
fingerprintFPDB fpfpdb = new fingerprintFPDB(DEDUP_SORT_ORDER.UNDEFINED_PLACEHOLDER);
fingerprintDMSG fpmsg = new fingerprintDMSG(DEDUP_SORT_ORDER.UNDEFINED_PLACEHOLDER);
int clogrecsize = ((Item)fpclog).get_size();
int fpdbrecsize = ((Item)fpfpdb).get_size();
int fpmsgrecsize = ((Item)fpmsg).get_size();
int clogcnt = (int)(fsrcclog.Length / clogrecsize);
byte[] currfp = new byte[16];
int currdbn = -1;
byte[] buffer1 = new byte[clogrecsize];
byte[] buffer2 = new byte[fpdbrecsize];
byte[] buffer3 = new byte[fpmsgrecsize];
while (clogcnt-- > 0)
{
fsrcclog.Read(buffer1, 0, clogrecsize);
((Item)fpclog).parse_bytes(buffer1, 0);
if (compare_fp(currfp, fpclog.fp) != 0)
{
for (int i = 0; i < 16; i++)
{
currfp[i] = fpclog.fp[i];
}
currdbn = fpclog.dbn;
//check if this fp is thre in fpdb, if yes get the dbn.
bool foundinfpdb = false;
if (fsrcfpdb != null && fsrcfpdb.Position < fsrcfpdb.Length)
{
do
{
fsrcfpdb.Read(buffer2, 0, fpdbrecsize);
((Item)fpfpdb).parse_bytes(buffer2, 0);
fdest.Write(buffer2, 0, fpdbrecsize);
//Console.WriteLine("Read FPDB file : " + fsrcfpdb.Position + " : " + ((Item)fpfpdb).get_string_rep() + " curr=" + currdbn);
if (compare_fp(currfp, fpfpdb.fp) == 0)
{
currdbn = fpfpdb.dbn; //let dedupe to old block preferably
foundinfpdb = true;
existing_dbn_counter++;
break;
}
else if ((compare_fp(currfp, fpfpdb.fp) > 0))
{
fsrcfpdb.Position -= fpdbrecsize;
break;
}
} while (fsrcfpdb.Position < fsrcfpdb.Length);
}
if (foundinfpdb == false)
{
//write to new fpdb, which was encounted from newly written data.
fpfpdb.dbn = currdbn;
for (int i = 0; i < 16; i++)
{
fpfpdb.fp[i] = currfp[i];
}
((Item)fpfpdb).get_bytes(buffer2, 0);
fdest.Write(buffer2, 0, fpdbrecsize);
}
}
//dont have to copy the same duplicates. i.e the first dbn which we saw from some file
//need not be deduped to the same file right?
if (currdbn != fpclog.dbn)
{
//push this to the messagequeue scriptfile
fpmsg.fsid = fpclog.fsid;
fpmsg.inode = fpclog.inode;
fpmsg.fbn = fpclog.fbn;
fpmsg.sourcedbn = fpclog.dbn;
fpmsg.destinationdbn = currdbn;
for (int i = 0; i < 16; i++)
{
fpmsg.fp[i] = fpclog.fp[i];
}
((Item)fpmsg).get_bytes(buffer3, 0);
scriptfile.Write(buffer3, 0, fpmsgrecsize);
}
}
if (fsrcfpdb != null)
{
fsrcfpdb.Close();
}
fsrcclog.Close();
fdest.Flush();
fdest.Close();
scriptfile.Flush();
scriptfile.Close();
DEFS.DEBUG("DEDUPE", "finishing dedupe op Genscript : EXISTING : " + existing_dbn_counter);
}
int IComparer.Compare(object obj1, object obj2)
{
switch (((Item)obj1).get_itemtype())
{
case RECORD_TYPE.FINGERPRINT_RECORD_CLOG:
{
fingerprintCLOG c1 = (fingerprintCLOG)obj1;
fingerprintCLOG c2 = (fingerprintCLOG)obj2;
for (int i = 0; i < 16; i++)
{
if (c1.fp[i] < c2.fp[i])
{
return(-1);
}
else if (c1.fp[i] > c2.fp[i])
{
return(1);
}
}
return(0);
}
//break; unreachable
case RECORD_TYPE.FINGERPRINT_RECORD_FPDB:
{
fingerprintFPDB c1 = (fingerprintFPDB)obj1;
fingerprintFPDB c2 = (fingerprintFPDB)obj2;
for (int i = 0; i < 16; i++)
{
if (c1.fp[i] < c2.fp[i])
{
return(-1);
}
else if (c1.fp[i] > c2.fp[i])
{
return(1);
}
}
return(0);
}
//break; unreachable
case RECORD_TYPE.FINGERPRINT_RECORD_MSG:
{
fingerprintDMSG c1 = (fingerprintDMSG)obj1;
fingerprintDMSG c2 = (fingerprintDMSG)obj2;
for (int i = 0; i < 16; i++)
{
if (c1.fp[i] < c2.fp[i])
{
return(-1);
}
else if (c1.fp[i] > c2.fp[i])
{
return(1);
}
}
return(0);
}
//break; unreachable
}
DEFS.ASSERT(false, "Shouldnt have come here wewrwr2");
return(0);
}
int IComparer.Compare(object obj1, object obj2)
{
switch (((Item)obj1).get_itemtype())
{
case RECORD_TYPE.FINGERPRINT_RECORD_CLOG:
{
fingerprintCLOG c1 = (fingerprintCLOG)obj1;
fingerprintCLOG c2 = (fingerprintCLOG)obj2;
if (c1.fsid < c2.fsid)
{
return(-1);
}
else if (c1.fsid > c2.fsid)
{
return(1);
}
else
{
if (c1.inode < c2.inode)
{
return(-1);
}
else if (c1.inode > c2.inode)
{
return(1);
}
else
{
if (c1.fbn < c2.fbn)
{
return(-1);
}
else if (c1.fbn > c2.fbn)
{
return(1);
}
else
{
if (c1.cnt > c2.cnt)
{
return(-1);
}
else if (c1.cnt < c2.cnt)
{
return(1);
}
else
{
return(0); //can actually assert!
}
}
}
}
}
//break; unreachable
case RECORD_TYPE.FINGERPRINT_RECORD_MSG:
{
fingerprintDMSG c1 = (fingerprintDMSG)obj1;
fingerprintDMSG c2 = (fingerprintDMSG)obj2;
if (c1.fsid < c2.fsid)
{
return(-1);
}
else if (c1.fsid > c2.fsid)
{
return(1);
}
else
{
if (c1.inode < c2.inode)
{
return(-1);
}
else if (c1.inode > c2.inode)
{
return(1);
}
else
{
if (c1.fbn < c2.fbn)
{
return(-1);
}
else if (c1.fbn > c2.fbn)
{
return(1);
}
else
{
return(0);
}
}
}
}
//break; unreachable
}
DEFS.ASSERT(false, "Shouldnt have come here 34234a23");
return(0);
}