/// <summary>
/// ReadSqlStatement
/// </summary>
/// <param name="name"></param>
/// <param name="id"></param>
/// <param name="sqlStmt"></param>
/// <param name="owner"></param>
/// <returns></returns>
public static bool ReadSqlStatement(
string name,
int version,
out string sqlStmt,
out string keyColName,
out string owner)
{
string sql = @"
select
sql,
key_col_name,
ownr_id
from dev_mbs_owner.mbs_spotfire_sql
where name = '" + name + "' and " +
"version = " + version;
DbCommandMx cmd = new DbCommandMx();
cmd.Prepare(sql);
cmd.ExecuteReader();
sqlStmt = keyColName = owner = null;
bool exists = cmd.Read();
if (exists)
{
sqlStmt = cmd.GetClob(0);
keyColName = cmd.GetString(1);
owner = cmd.GetString(2);
}
cmd.CloseReader();
cmd.Dispose();
return(exists);
}
static ICdkMol CdkMolUtil => StaticCdkMol.I; // static molecule shortcut for utility methods
/// <summary>
/// UpdateCorpFingerprintDatabaseMx
/// ///////////////////////////////////////////////////////
/// Syntax: Update FingerprintDatabaseMx [Corp | ChEMBL] [MACCS | ECFP4] [Load | ByCidRange | SinceLastCheckpoint | LoadMissing | <SingleCorpId>]
///
/// Corp Examples:
/// Update FingerprintDatabaseMx Corp MACCS Load
/// Update FingerprintDatabaseMx Corp MACCS LoadMissing
/// Update FingerprintDatabaseMx Corp MACCS SinceLastCheckpoint
///
/// Update FingerprintDatabaseMx Corp ECFP4 Load
/// Update FingerprintDatabaseMx Corp ECFP4 LoadMissing
/// Update FingerprintDatabaseMx Corp ECFP4 SinceLastCheckpoint
///
/// ChEMBL Examples:
/// Update FingerprintDatabaseMx Chembl MACCS Load
/// Update FingerprintDatabaseMx Chembl MACCS LoadMissing
///
/// Update FingerprintDatabaseMx Chembl ECFP4 Load
/// Update FingerprintDatabaseMx Chembl ECFP4 LoadMissing
/// ///////////////////////////////////////////////////////
/// </summary>
/// <param name="argString"></param>
/// <returns></returns>
static public string Update(
string argString)
{
MoleculeMx mol;
double mw;
string chime, smiles, molString, molFile = "";
string msg = "", sql = "", chemblId, cid = "", maxCorpIdSql, maxIdSql2, mf, missingFixCriteria = "", CorpIdList = "";
int storeChunkCount = 0, CorpId, molregno, molId, lowId = 0, highId = 0, maxDestId = 0, maxSrcId = 0;
int readCount = 0, storeCount = 0;
ByCheckpoint = ByCidRange = ByCidList = LoadIfMissing = false;
ReadChunkSize = DefaultReadChunkSize;
WriteChunkSize = DefaultWriteChunkSize;
Failures = new Dictionary <string, string>();
NewUndefinedStructureCids = new List <string>();
LastFailure = "";
FailureCount = 0;
// global try loop
try
{
////////////////////////
/// Parse Parameters ///
////////////////////////
// See which database
argString = argString.Trim();
if (Lex.StartsWith(argString, "Corp"))
{
Database = "Corp";
argString = argString.Substring(5).Trim();
}
else if (Lex.StartsWith(argString, "Chembl"))
{
Database = "ChEMBL";
argString = argString.Substring(6).Trim();
}
else
{
return(SyntaxMsg);
}
// See which fingerprint type
FingerprintType = FingerprintType.MACCS; // default to MACCS if type not defined
if (Lex.TryReplace(ref argString, "MACCS", ""))
{
FingerprintType = FingerprintType.MACCS;
argString = argString.Trim();
}
else if (Lex.TryReplace(ref argString, "ECFP4", ""))
{
FingerprintType = FingerprintType.Circular; // (i.e. ECFP4)
argString = argString.Trim();
}
FpDao = new FingerprintDao(Database, FingerprintType);
List <FingerprintRec> fpRecList = new List <FingerprintRec>();
string args = argString.Trim();
string initialMsg = "Update FingerprintDatabase started: " + args;
CidList = new List <string>(); // init empty list
//////////////////////
/// Corp Database ///
//////////////////////
if (CorpDatabase)
{
if (Lex.Eq(args, "Load"))
{
ByCidRange = true;
ShowProgress("Getting range of CorpIds to insert...");
maxCorpIdSql = SelectMaxCorpId; // get highest id in source db
maxSrcId = SelectSingleValueDao.SelectInt(maxCorpIdSql);
if (maxSrcId < 0)
{
maxSrcId = 0;
}
maxDestId = GetMaxDestMolId();
//maxIdSql2 = "select max(src_compound_id_nbr) from dev_mbs_owner.corp_uc_xref where src_id = 0"; // get highest id in UniChemDb db
//highCorpId = SelectSingleValueDao.SelectInt(maxIdSql2);
if (maxDestId < 0)
{
maxDestId = 0;
}
}
else if (Lex.Eq(args, "SinceLastCheckpoint"))
{
ByCheckpoint = true;
ShowProgress("Getting list of CorpIds updated since last checkpoint...");
CidList = GetNewAndModifiedCorpIdList(out CidUpdateDateDict);
//CidUpdateList = new List<string>(); // debug with single cmpd
//CidUpdateList.Add("03435269");
if (CidList.Count == 0)
{
return("There have been no updates since the last checkpoint");
}
initialMsg += ", CorpIds to add/update: " + CidList.Count;
}
else if (Lex.StartsWith(args, "ByCorpIdList"))
{
ByCidList = true;
CorpIdList = args.Substring("ByCorpIdList".Length).Trim();
if (Lex.IsUndefined(CorpIdList))
{
throw new Exception("Undefined CorpId list");
}
}
else if (Lex.StartsWith(args, "LoadMissing"))
{
LoadIfMissing = true;
if (args.Contains(" "))
{
missingFixCriteria = args.Substring("LoadMissing".Length).Trim();
}
ShowProgress("Getting list of missing CorpIds...");
CidList = GetMissingCidList();
if (CidList.Count == 0)
{
return("There are no missing CorpIds");
}
initialMsg += ", Missing CorpIds: " + CidList.Count;
}
else if (int.TryParse(args, out maxSrcId)) // single CorpId
{
ByCidRange = true;
maxDestId = maxSrcId - 1; // say 1 less is the max we have
}
else
{
return(SyntaxMsg);
}
}
///////////////////////
/// ChEMBL Database ///
///////////////////////
else if (ChemblDatabase)
{
if (Lex.Eq(args, "Load"))
{
ByCidRange = true;
ShowProgress("Getting range of MolRegNos to insert...");
sql = "select max(molregno) from chembl_owner.compound_struct_xxxxxx";
maxSrcId = SelectSingleValueDao.SelectInt(sql);
if (maxSrcId < 0)
{
maxSrcId = 0;
}
maxDestId = GetMaxDestMolId();
if (maxDestId < 0)
{
maxDestId = 0;
}
}
else if (Lex.StartsWith(args, "LoadMissing"))
{
LoadIfMissing = true;
ShowProgress("Getting list of missing ChEMBL Ids...");
CidList = GetMissingCidList();
if (CidList.Count == 0)
{
return("There are no missing Ids");
}
initialMsg += ", Missing Chembl Ids: " + CidList.Count;
}
else
{
return(SyntaxMsg);
}
}
else
{
return(SyntaxMsg);
}
CidListOriginalCount = CidList.Count;
Log(initialMsg);
/////////////////////////////
// Loop over chunks of data
/////////////////////////////
for (int chunk = 1; ; chunk++)
{
//////////////////////
/// Corp Database ///
//////////////////////
if (CorpDatabase)
{
if (ByCheckpoint) // single chunk
{
string cidList = GetNextListChunk();
if (Lex.IsUndefined(cidList))
{
break;
}
sql = SelectByCorpIdCriteria;
sql = Lex.Replace(sql, "<CorpIdCriteria>", "in (" + cidList + ")");
string matchString = "order by m.corp_nbr";
if (!Lex.Contains(sql, matchString))
{
throw new Exception(matchString + " not found");
}
sql = Lex.Replace(sql, matchString, "order by m.molecule_date");
msg = "Processing " + CidListOriginalCount + " updates since " + CheckpointDateTime;
// + " (" + Mobius.Data.CidList.FormatCidListForDisplay(null, chunkCidList) + ")";
}
else if (ByCidRange) // by CorpId range
{
if (maxDestId >= maxSrcId)
{
break; // done
}
lowId = maxDestId + 1; // start of next chunk
highId = lowId + ReadChunkSize;
maxDestId = highId;
//lowCorpId = highCorpId = 12345; // debug
if (highId >= maxSrcId)
{
highId = maxSrcId;
}
sql = SelectByCorpIdCriteria;
sql = Lex.Replace(sql, "<CorpIdCriteria>", "between " + lowId + " and " + highId);
msg = "Processing CorpId range: " + lowId + " to " + highId;
}
else if (ByCidList) // by single user-supplied CorpId list
{
if (chunk > 1)
{
break; // break 2nd time through
}
sql = SelectByCorpIdCriteria;
sql = Lex.Replace(sql, "<CorpIdCriteria>", "in (" + CorpIdList + ")");
msg = "Processing CorpId list: " + CorpIdList;
}
else if (LoadIfMissing)
{
string cidList = GetNextListChunk();
if (Lex.IsUndefined(cidList))
{
break; // all done
}
sql = SelectByCorpIdCriteria;
sql = Lex.Replace(sql, "<CorpIdCriteria>", "in (" + cidList + ")");
msg = "Processing missing CorpId Chunk: " + Mobius.Data.CidList.FormatAbbreviatedCidListForDisplay(null, cidList) +
", Total Ids: " + CidListOriginalCount;
Log(msg);
}
else
{
return(SyntaxMsg);
}
}
///////////////////////
/// ChEMBL Database ///
///////////////////////
else if (ChemblDatabase)
{
if (ByCidRange) // by CID range
{
if (maxDestId >= maxSrcId)
{
break; // done
}
lowId = maxDestId + 1; // start of next chunk
highId = lowId + ReadChunkSize;
maxDestId = highId;
//lowId = highId = 12345; // debug
if (maxDestId >= maxSrcId)
{
maxDestId = maxSrcId;
}
sql = SelectChemblSql;
sql = Lex.Replace(sql, "<molregnoCriteria>", "between " + lowId + " and " + highId);
msg = "Processing ChEMBL MolRegNo range: " + lowId + " to " + highId;
}
else if (LoadIfMissing)
{
string cidList = GetNextListChunk();
if (Lex.IsUndefined(cidList))
{
break; // all done
}
sql = SelectByCorpIdCriteria;
sql = Lex.Replace(sql, "<CorpIdCriteria>", "in (" + cidList + ")");
msg = "Processing missing ChEMBL Id Chunk: " + Mobius.Data.CidList.FormatAbbreviatedCidListForDisplay(null, cidList) +
", Total Ids: " + CidListOriginalCount;
}
else
{
return(SyntaxMsg);
}
}
else
{
return(SyntaxMsg);
}
ShowProgress(msg);
// Execute the SQL to get the rows for the chunk
DbCommandMx rdr = DbCommandMx.PrepareAndExecuteReader(sql);
DateTime lastShowProgressTime = DateTime.MinValue;
///////////////////////////////////////////
/// Loop over rows in the current chunk ///
///////////////////////////////////////////
while (true)
{
// Update progress display
if (DateTime.Now.Subtract(lastShowProgressTime).TotalSeconds > 1) // show progress
{
int storeTotalCount = storeCount + storeChunkCount;
string msg2 = msg + "\r\n" +
"Reads: " + readCount + "\r\n" +
"Undefined: " + NewUndefinedStructureCids.Count + "\r\n" +
"Insert/Updates: " + storeTotalCount + "\r\n" +
"Failures: " + FailureCount + "\r\n" +
"Failure Types: " + Failures.Count + "\r\n" +
"Last Failure: " + LastFailure;
ShowProgress(msg2);
lastShowProgressTime = DateTime.Now;
}
// Read and process next compound
bool readOk = rdr.Read();
if (readOk)
{
readCount++;
try
{
double t1 = 0, t2 = 0, t3 = 0, t4 = 0;
DateTime t0 = DateTime.Now;
mol = null;
//t2 = TimeOfDay.Delta(ref t0);
//////////////////////
/// Corp Database ///
//////////////////////
if (CorpDatabase)
{
CorpId = rdr.GetInt(0); // corp_nbr
//Log("CorpId: " + CorpId); // debug
molId = CorpId;
cid = CorpId.ToString();
cid = CompoundId.NormalizeForDatabase(cid);
if (!rdr.IsNull(1)) // be sure chime field isn't null
{
chime = rdr.GetClob(1);
if (Lex.IsDefined(chime))
{
molFile = MoleculeMx.ChimeStringToMolfileString(chime); // convert Chime to MolFile
mol = new MoleculeMx(MoleculeFormat.Molfile, molFile);
}
}
MoleculeDateTime = rdr.GetDateTimeByName("Molecule_Date"); // Date molecule was updated in the CorpDB cartridge DB
}
///////////////////////
/// ChEMBL Database ///
///////////////////////
else // chembl
{
molId = molregno = rdr.GetInt(0);
cid = chemblId = rdr.GetString(1);
smiles = rdr.GetString(2);
if (Lex.IsDefined(smiles))
{
mol = new MoleculeMx(MoleculeFormat.Smiles, smiles);
}
}
if (MoleculeMx.IsUndefined(mol) || mol.AtomCount <= 1)
{
NewUndefinedStructureCids.Add(cid);
continue;
//mol = new AtomContainer(); // write empty structure
}
bool includeOverallFingerprint = true;
List <BitSetFingerprint> fps = CdkMol.BuildBitSetFingerprints(mol.MolfileString, includeOverallFingerprint, FingerprintType);
//t3 = TimeOfDay.Delta(ref t0);
foreach (BitSetFingerprint fp in fps)
{
FingerprintRec fpr = new FingerprintRec();
fpr.molId = molId;
fpr.SrcId = SrcDbId;
fpr.Cid = cid;
fpr.Cardinality = fp.cardinality();
fpr.Fingerprint = fp.asBitSet().toLongArray();
fpRecList.Add(fpr);
}
//t4 = TimeOfDay.Delta(ref t0);
t4 = t4;
}
catch (Exception ex)
{
if (!Failures.ContainsKey(ex.Message))
{
Failures.Add(ex.Message, cid);
}
else
{
Failures[ex.Message] += ", " + cid;
}
LastFailure = "Cid: " + cid + " - " + ex.Message;
Log(LastFailure);
//ShowProgress(ex.Message + "\r\n" + ex.StackTrace.ToString()); // debug
FailureCount++;
continue;
}
storeChunkCount++;
}
bool commitTransaction = (storeChunkCount >= WriteChunkSize || (!readOk && storeChunkCount > 0));
if (commitTransaction) // end of chunk of data to store?
{
// if updating by CheckPoint date range then merge existing data with new/updated data
if (ByCheckpoint)
{
if (readCount > 0 && (storeCount > 0 || FailureCount == 0)) // make sure not everything has failed)
{
MergeRecordsIntoFiles(fpRecList);
}
}
// Simple append of records to files
else
{
FpDao.OpenWriters("bin", FileMode.Append); // open bin files for append
foreach (FingerprintRec fpr in fpRecList) // write out buffered recs
{
FpDao.WriteFingerprintRec(fpr);
}
FpDao.CloseWriters();
int cnt = fpRecList.Count;
if (cnt > 0)
{
string cid1 = fpRecList[0].Cid;
string cid2 = fpRecList[cnt - 1].Cid;
Log("Records Appended: " + cnt + ", CIDS: " + cid1 + " - " + cid2);
}
else
{
Log("Records Appended: 0");
}
}
fpRecList.Clear();
storeCount += storeChunkCount;
storeChunkCount = 0;
}
if (!readOk)
{
break;
}
} // end of read loop for rows in a chunk
rdr.Dispose();
} // end for loop of chunks
DeleteTempFiles();
if (LoadIfMissing) // update list of cids with missing structures
{
ExistingUndefinedStructureCids.UnionWith(NewUndefinedStructureCids);
FpDao.WriteUndefinedStructuresCids(ExistingUndefinedStructureCids);
}
msg = "*** Update Complete ***\r\n\r\n" + msg;
ShowProgress(msg);
System.Threading.Thread.Sleep(100);
string logMsg = "UpdateFingerprintDb - CIDs stored: " + storeCount + ", Undefined structures: " + NewUndefinedStructureCids.Count + ", failures: " + FailureCount + "\r\n";
foreach (string key in Failures.Keys)
{
logMsg += key + " - CIDs: " + Failures[key] + "\r\n";
}
Log(logMsg);
return(logMsg);
} // end of main try loop
catch (Exception ex)
{
Log(DebugLog.FormatExceptionMessage(ex));
throw new Exception(ex.Message, ex);
}
}
/// <summary>
/// Retrieve any existing SVG for the list of supplied molecules
/// The Id column should contain the CorpId
/// </summary>
/// <param name="molList"></param>
public static int SelectMoleculeListSvg(
List <MoleculeMx> molList)
{
MoleculeMx mol;
int corpId, molSvgsFetchedCount = 0;
string corpIdString, molString, svg;
const string sql = @"
SELECT
corp_nbr,
molstructure svgString
FROM
mbs_owner.corp_moltable_mx
WHERE
corp_nbr in (<list>)
and molstructure is not null
" ;
if (!Security.UserInfo.Privileges.CanRetrieveStructures) // structures allowed?
{
return(0);
}
//if (DebugMx.True) return 0; // debug, don't use existing values
List <string> lsnList = new List <string>();
Dictionary <string, MoleculeMx> molDict = new Dictionary <string, MoleculeMx>();
foreach (MoleculeMx mol0 in molList) // set up a dict keyed by cid with mol values
{
if (mol0.PrimaryFormat != MoleculeFormat.Helm || Lex.IsUndefined(mol0.PrimaryValue))
{
continue;
}
if (int.TryParse(mol0.Id, out corpId))
{
molDict[mol0.Id] = mol0;
}
}
if (molDict.Count == 0)
{
return(0);
}
DbCommandMx cmd = new DbCommandMx();
cmd.PrepareListReader(sql, DbType.String);
cmd.ExecuteListReader(new List <string>(molDict.Keys));
while (cmd.Read())
{
corpId = cmd.GetInt(0);
if (!cmd.IsNull(1)) // molstructure
{
molString = cmd.GetClob(1);
if (!SvgUtil.IsSvgString(molString))
{
continue; // skip if not SVG
}
svg = molString; // should be compressed format SVG
corpIdString = CompoundId.Normalize(corpId.ToString());
if (Lex.IsDefined(svg) && molDict.ContainsKey(corpIdString))
{
mol = molDict[corpIdString];
mol.SvgString = svg;
molSvgsFetchedCount++;
}
}
}
cmd.CloseReader();
return(molSvgsFetchedCount);
}
/// <summary>
/// Sync the Mobius CorpMoltable replicate used to retrieve Smiles
/// Syntax: UpdateCorpDbMoltableMx [ ByDateRange | ByCorpIdRange | LoadMissing | <singleCorpId>]
/// </summary>
/// <returns></returns>
public static string UpdateCorpDbMoltableMx(
string args)
{
DateTime moleculeDateTime = DateTime.MinValue;
double mw;
string msg = "", sql = "", maxCorpIdSql, mf = "", chime = "", smiles = "", checkPointDate = "", helm = "", sequence = "", svg = "";
object[][] pva = null;
int pvaCount = 0, CorpId, lowCorpId = 0, highCorpId = 0, srcMaxCorpId = 0;
int SelectChunkSize = 20; // small chunks
int InsertBufferSize = 10;
//int SelectChunkSize = 100000; // big chunks
//int InsertBufferSize = 1000;
// Select data from corp_moltable by CorpId range
const string SelectByCorpIdRange = @"
SELECT
m.corp_nbr,
chime(m.ctab),
m.molformula,
m.molweight,
null molsmiles,
s.helm_txt,
s.sequence_txt,
m.molecule_date
FROM
corp_owner.corp_moltable m,
corp_owner.corp_substance s
where
m.corp_nbr > 0
and s.corp_nbr = m.corp_nbr
and (s.status_code is null or s.status_code = 'A')
ORDER BY corp_nbr";
// Select data from corp_moltable by date range comparing to corp_moltable_mx
const string SelectByDateRange = @"
select
m.corp_nbr,
chime(m.ctab),
m.molformula,
m.molweight,
null molsmiles,
s.helm_txt,
s.sequence_txt,
m.molecule_date,
m2.molecule_date
from
corp_owner.corp_moltable m,
corp_owner.corp_substance s,
corp_moltable_mx m2
where
m.molecule_date > to_date('1-jan-1900 000000','DD-MON-YYYY HH24MISS')
and s.corp_nbr = M.CORP_NBR
and (s.status_code is null or s.status_code = 'A')
and m2.corp_nbr (+) = m.corp_nbr
and m2.molecule_date (+) != m.molecule_date
order by m.molecule_date" ;
// Select for missing smiles strings, ex: Update CorpDbMoltableMx LoadMissing mx.molecule_date > '1-jan-2014'
const string SelectMissingSmilesFix = @"
select /* check for CorpIds in corp_moltable not in corp_moltable_mx */
corp_nbr,
chime(ctab),
molformula,
molweight,
null molsmiles,
helm_txt,
sequence_txt,
molecule_date
from
(
select
m.*,
s.helm_txt,
s.sequence_txt,
mx.molsmiles molsmiles_mx
from
corp_owner.corp_moltable m,
corp_owner.corp_substance s,
corp_moltable_mx mx
where
s.corp_nbr = M.CORP_NBR
and (s.status_code is null or s.status_code = 'A')
and mx.corp_nbr (+) = m.corp_nbr
and 1=1 /* condition to substitute */
) m
where molsmiles_mx is null /* extra condition */
order by corp_nbr" ;
// Insert missing helm info
const string SelectMissingHelmFix = @"
select /* check for CorpIds in corp_moltable not in corp_moltable_mx */
corp_nbr,
chime(ctab),
molformula,
molweight,
null molsmiles,
helm_txt,
sequence_txt,
molecule_date
from
(
select
m.*,
s.helm_txt,
s.sequence_txt,
mx.molsmiles molsmiles_mx
from
corp_owner.corp_moltable m,
corp_owner.corp_substance s,
corp_moltable_mx mx
where
s.corp_nbr = M.CORP_NBR
and (s.status_code is null or s.status_code = 'A')
and mx.corp_nbr (+) = m.corp_nbr
and 1=1 /* condition to substitute */
) m
where length(helm_txt) > 0 /* extra condition */
order by corp_nbr" ;
// Secondary "large" structure table (~5k mols)
const string SelectLargeMols = @"
select
corp_nbr,
to_clob(molstructure),
to_clob(molformula),
molweight,
molsmiles,
null helm_txt,
null sequence_txt,
molecule_date
from
(select
corp_srl_nbr corp_nbr,
'CompoundId=' || corp_srl_nbr molstructure,
null ctab,
mlclr_frml_txt molformula,
mlclr_wgt molweight,
null molsmiles,
null molecule_date
from rdm_owner.rdm_sbstnc
where rdw_src_cd = 'LRG'" ;
// Insert statement
const string InsertSql = @"
insert into mbs_owner.corp_moltable_mx (
corp_nbr,
molstructure,
molformula,
molweight,
molsmiles,
molecule_date)
values (:0, :1, :2, :3, :4, :5)" ;
// Build select sql
bool byDateRange = false, byCorpIdRange = false, missingFix = true, deleteExisting = true;
string missingFixCriteria = "";
if (Lex.IsUndefined(args) || Lex.Eq(args, "ByDateRange"))
{
byDateRange = true;
}
else if (Lex.Eq(args, "ByCorpIdRange"))
{
byCorpIdRange = true;
Progress.Show("Getting range of CorpIds to insert...");
maxCorpIdSql = "select max(corp_nbr) from corp_owner.corp_moltable"; // get highest CorpId in source db
srcMaxCorpId = SelectSingleValueDao.SelectInt(maxCorpIdSql);
if (srcMaxCorpId < 0)
{
srcMaxCorpId = 0;
}
maxCorpIdSql = "select max(corp_nbr) from mbs_owner.corp_moltable_mx"; // get highest CorpId in dest db
highCorpId = SelectSingleValueDao.SelectInt(maxCorpIdSql);
if (highCorpId < 0)
{
highCorpId = 0;
}
}
else if (Lex.StartsWith(args, "LoadMissing"))
{
missingFix = true;
if (args.Contains(" "))
{
missingFixCriteria = args.Substring(10).Trim();
}
}
else if (int.TryParse(args, out srcMaxCorpId)) // single CorpId
{
byCorpIdRange = true;
highCorpId = srcMaxCorpId - 1; // say 1 less is the max we have
}
else
{
return("Syntax: UpdateCorpDbMoltableMx [ ByDateRange | ByCorpIdRange | LoadMissing | <singleCorpId>]");
}
Log("UpdateCorpDbMoltableMx started: " + args);
int readCount = 0, insCount = 0, insertCount = 0, updateCount = 0, undefinedStructures = 0, smilesSuccess = 0, smilesFails = 0, helmStructures = 0;
List <string> CorpIdList = new List <string>();
for (int chunk = 1; ; chunk++) // loop over chunks
{
if (byDateRange) // single chunk
{
if (chunk > 1)
{
break; // break 2nd time through
}
checkPointDate = UserObjectDao.GetUserParameter("MOBIUS", "UpdateCorpDbMoltableMxCheckpointDate", "01-sep-2013 000000");
//UserObjectDao.SetUserParameter("MOBIUS", "UpdateCorpDbMoltableMxCheckpointDate", checkPointDate);
sql = Lex.Replace(SelectByDateRange, "1-jan-1900 000000", checkPointDate);
msg = "Reading where date >= " + checkPointDate;
}
else if (byCorpIdRange) // by CorpId range
{
if (highCorpId >= srcMaxCorpId)
{
break; // done
}
lowCorpId = highCorpId + 1; // start of next chunk
highCorpId = lowCorpId + SelectChunkSize;
if (highCorpId >= srcMaxCorpId)
{
highCorpId = srcMaxCorpId;
}
sql = Lex.Replace(SelectByCorpIdRange, "corp_nbr > 0", "corp_nbr between " + lowCorpId + " and " + highCorpId);
msg = "Reading: " + lowCorpId + " to " + highCorpId + ", Reads: " + readCount + ", Inserts: " + insertCount;
}
else if (missingFix)
{
if (chunk > 1)
{
break; // break 2nd time through
}
sql = SelectMissingHelmFix;
if (Lex.IsDefined(missingFixCriteria)) // substitute any criteria
{
sql = Lex.Replace(sql, "1=1", missingFixCriteria);
}
msg = "Fixing missing data";
}
Progress.Show(msg);
DbCommandMx readCmd = new DbCommandMx();
readCmd.MxConn = DbConnectionMx.GetConnection("prd123");
readCmd.PrepareUsingDefinedConnection(sql, null);
DbDataReader rdr = readCmd.ExecuteReader();
DbCommandMx insertCmd = new DbCommandMx();
OracleDbType[] pta = new OracleDbType[6];
pta[0] = OracleDbType.Int32; // corp_nbr
pta[1] = OracleDbType.Clob; // molstructure
pta[2] = OracleDbType.Clob; // molformula
pta[3] = OracleDbType.Double; // molweight
pta[4] = OracleDbType.Clob; // smiles
pta[5] = OracleDbType.Date; // molecule_date
insertCmd.Prepare(InsertSql, pta);
insertCmd.BeginTransaction(); // be sure we have a transaction going
pva = DbCommandMx.NewObjectArrayArray(6, InsertBufferSize); // alloc insert row array
object[] vo = new object[6];
while (true)
{
bool readOk = rdr.Read();
if (readOk)
{
rdr.GetValues(vo);
CorpId = readCmd.GetInt(0); // corp_nbr
vo[0] = CorpId;
CorpIdList.Add(CorpId.ToString());
if (!readCmd.IsNull(1)) // molstructure
{
chime = readCmd.GetClob(1);
chime = OracleMx.ClearStringIfExceedsMaxStringSize(chime);
vo[1] = chime;
}
else
{
chime = "";
}
if (!readCmd.IsNull(2)) // molformula
{
mf = readCmd.GetClob(2);
mf = OracleMx.ClearStringIfExceedsMaxStringSize(mf);
vo[2] = mf;
}
if (!readCmd.IsNull(3)) // molweight
{
mw = readCmd.GetDouble(3);
vo[3] = mw;
}
if (Lex.IsDefined(chime)) // molsmiles - calculate from chime string
{
MoleculeMx cs = new MoleculeMx(MoleculeFormat.Chime, chime);
if (cs.AtomCount > 1) // need more than one atom
{
MoleculeMx cs2 = cs.ConvertTo(MoleculeFormat.Smiles);
smiles = cs2.GetSmilesString();
if (Lex.IsDefined(smiles))
{
smilesSuccess++;
}
else
{
Log("Smiles conversion failure for CorpId: " + CorpId);
smilesFails++;
}
smiles = OracleMx.ClearStringIfExceedsMaxStringSize(smiles);
vo[4] = smiles;
}
else
{
undefinedStructures++;
}
}
else
{
undefinedStructures++;
}
if (!readCmd.IsNull(5))
{
helm = readCmd.GetClob(5);
if (Lex.IsDefined(helm))
{
svg = HelmControl.GetSvg(helm);
vo[1] = SvgUtil.CompressSvgString(svg); // store compressed svg in molstructure column for now
helmStructures++;
}
}
if (!readCmd.IsNull(6))
{
sequence = readCmd.GetClob(6);
if (Lex.IsDefined(sequence))
{
// nothing yet
}
}
moleculeDateTime = DateTime.MinValue;
if (!readCmd.IsNull(7)) // molecule_date
{
moleculeDateTime = readCmd.GetDateTime(7);
vo[5] = moleculeDateTime;
}
for (int pi = 0; pi < 6; pi++) // invert for insert
{
pva[pi][pvaCount] = vo[pi];
}
if (Debug)
{
msg = String.Format("CorpId: {0}, mf: {1}, chime: {2}, smiles: {3}", CorpId.ToString(), mf.Length, chime.Length, smiles.Length);
Log(msg);
}
pvaCount++;
}
if (pvaCount >= InsertBufferSize || (!readOk && pvaCount > 0)) // write if buffer full or at end
{
try
{
if (deleteExisting)
{
int delCount = DoDeletes(CorpIdList);
updateCount += delCount; // count deletes as updates
insertCount -= delCount; // subtract from inserts
}
CorpIdList.Clear();
insCount = insertCmd.ExecuteArrayNonReader(pva, ref pvaCount);
insertCmd.Commit();
insertCmd.BeginTransaction();
insertCount += insCount;
}
catch (Exception ex)
{
throw new Exception(ex.Message, ex);
}
if (byDateRange)
{
string checkPointDate2 = String.Format("{0:dd-MMM-yyyy HHmmss}", moleculeDateTime); // format date time that will work with oracle
UserObjectDao.SetUserParameter("MOBIUS", "UpdateCorpDbMoltableMxCheckpointDate", checkPointDate2);
msg = "Processing where date >= " + checkPointDate + ", Reads: " + readCount + ", Inserts: " + insertCount + ", Updates: " + updateCount;
}
else if (byCorpIdRange) // CorpId range
{
msg = "Processing: " + lowCorpId + " to " + highCorpId + ", Reads: " + readCount + ", Inserts: " + insertCount;
}
else if (missingFix)
{
msg = "Fixing missing smiles, Updates: " + updateCount;
}
msg += String.Format(", Undefined structures: {0} , Smiles failures: {1}, Helms: {2}", undefinedStructures, smilesFails, helmStructures);
Progress.Show(msg);
}
if (!readOk)
{
break;
}
readCount++;
}
readCmd.Dispose();
insertCmd.Dispose();
} // end for select chunk
msg = "UpdateCorpDbMoltableMx - Inserts: " + insertCount + ", Updates: " + updateCount;
msg += String.Format(", Undefined structures: {0} , Smiles failures: {1}, Helms: {2}", undefinedStructures, smilesFails, helmStructures);
Log(msg);
return(msg);
}