From 326d005981b2214b52db3f709fc2cc4c6efab199 Mon Sep 17 00:00:00 2001
From: Dario Berzano <dario.berzano@cern.ch>
Date: Fri, 29 Nov 2013 01:35:45 +0100
Subject: [PATCH] Improved AliEn dataset retrieval with run ranges
Since run ranges usually contain many "holes", yet they are more
convenient to use than specifying a list of runs manually, a
preventive check to prune invalid run numbers is performed: AliEn
queries on non-existing runs are consequently avoided.
---
proof/proof/src/TDataSetManagerAliEn.cxx | 72 +++++++++++++++++++-----
1 file changed, 57 insertions(+), 15 deletions(-)
diff --git a/proof/proof/src/TDataSetManagerAliEn.cxx b/proof/proof/src/TDataSetManagerAliEn.cxx
index fc7ea0e735a..08432954052 100644
--- a/proof/proof/src/TDataSetManagerAliEn.cxx
+++ b/proof/proof/src/TDataSetManagerAliEn.cxx
@@ -451,43 +451,85 @@ TList *TDataSetManagerAliEn::GetFindCommandsFromUri(TString &uri,
findCommands = new TList();
findCommands->SetOwner(kTRUE);
- TString basePathSim;
+ TString basePathRun;
+
+ if (!gGrid) {
+ TGrid::Connect("alien:");
+ if (!gGrid) {
+ delete findCommands;
+ delete runList;
+ return NULL;
+ }
+ }
if (sim) {
// Montecarlo init.
// Check whether this period is in /alice/sim/<period> or in
// /alice/sim/<year>/<period> and act properly, since naming convention
// is unclear!
- if (!gGrid) {
- TGrid::Connect("alien:");
- if (!gGrid) {
- delete findCommands;
- delete runList;
- return NULL;
- }
- }
// Check once for all
- basePathSim.Form("/alice/sim/%s", lhcPeriod.Data()); // no year
- if (!gGrid->Cd(basePathSim.Data())) {
- basePathSim.Form("/alice/sim/%d/%s", year, lhcPeriod.Data());
+ basePathRun.Form("/alice/sim/%s", lhcPeriod.Data()); // no year
+ if (!gGrid->Cd(basePathRun.Data())) {
+ basePathRun.Form("/alice/sim/%d/%s", year, lhcPeriod.Data());
}
}
else {
// Real data init.
// Parse the pass string: if it starts with a number, prepend "pass"
if ((pass[0] >= '0') && (pass[0] <= '9')) pass.Prepend("pass");
+ basePathRun.Form("/alice/data/%d/%s", year, lhcPeriod.Data());
+ }
+
+ // Form a list of valid runs (to avoid unnecessary queries when run ranges
+ // are specified)
+ std::vector<Int_t> validRuns;
+ {
+ TGridResult *validRunDirs = gGrid->Ls( basePathRun.Data() );
+ if (!validRunDirs) return NULL;
+
+ TIter nrd(validRunDirs);
+ TMap *dir;
+ TObjString *os;
+ validRuns.resize( (size_t)(validRunDirs->GetEntries()) );
+
+ while (( dir = dynamic_cast<TMap *>(nrd()) ) != NULL) {
+ os = dynamic_cast<TObjString *>( dir->GetValue("name") );
+ if (!os) continue;
+ Int_t run = (os->String()).Atoi();
+ if (run > 0) validRuns.push_back(run);
+ }
}
for (UInt_t i=0; i<runList->size(); i++) {
+ // Check if current run is valid
+ Bool_t valid = kFALSE;
+ for (UInt_t j=0; j<validRuns.size(); j++) {
+ if (validRuns[j] == (*runList)[i]) {
+ valid = kTRUE;
+ break;
+ }
+ }
+ if (!valid) {
+ //if (gDebug >=1) {
+ Warning("TDataSetManagerAliEn::GetFindCommandsFromUri",
+ "Avoiding unnecessary find on run %d: not found", (*runList)[i]);
+ //}
+ continue;
+ }
+ else {
+ Info("TDataSetManagerAliEn::GetFindCommandsFromUri",
+ "Run found: %d", (*runList)[i]);
+ }
+
// Here we need to assemble the find string
TString basePath, fileName, temp;
if (sim) {
// Montecarlo
temp.Form("/%06d", runList->at(i));
- basePath = basePathSim + temp;
+ basePath = basePathRun + temp;
if (!esd) {
temp.Form("/AOD%03d", aodNum);
@@ -496,8 +538,8 @@ TList *TDataSetManagerAliEn::GetFindCommandsFromUri(TString &uri,
}
else {
// Real data
- basePath.Form("/alice/data/%d/%s/%09d/ESDs/%s", year,
- lhcPeriod.Data(), runList->at(i), pass.Data());
+ temp.Form("/%09d/ESDs/%s", runList->at(i), pass.Data());
+ basePath = basePathRun + temp;
if (esd) {
basePath.Append("/*.*");
}
--
GitLab