[df] Fix interaction between RSampleInfo and redirected EOS paths

When the input files are paths to FUSE-mounted EOS files, during the event loop
these paths will be redirected to the corresponding xroot EOS URL, in
TFile::Open.

This was causing a bad interaction with the sample metadata and the subsequent
usage in the event loop, e.g. through DefinePerSample. Specifically, we fill a
map with the metadata at construction time, which includes the input file paths
(without redirection). These will then be irretrievable during the event loop
since the map key will not correspond to the redirected path. Fix this by also
adding the redirected map during the filling of the map in ChangeSpec.

Author: vepadulano

tree/dataframe/src/RLoopManager.cxx

@@ -42,6 +42,25 @@
 #include "ROOT/RNTupleDS.hxx"
 #endif
 
+#ifdef R__UNIX
+// Functions needed to perform EOS XRootD redirection in ChangeSpec
+#include <optional>
+#include "TEnv.h"
+#include "TSystem.h"
+#ifndef R__FBSD
+#include <sys/xattr.h>
+#else
+#include <sys/extattr.h>
+#endif
+#ifdef R__MACOSX
+/* On macOS getxattr takes two extra arguments that should be set to 0 */
+#define getxattr(path, name, value, size) getxattr(path, name, value, size, 0u, 0)
+#endif
+#ifdef R__FBSD
+#define getxattr(path, name, value, size) extattr_get_file(path, EXTATTR_NAMESPACE_USER, name, value, size)
+#endif
+#endif
+
 #include <algorithm>
 #include <atomic>
 #include <cassert>
@@ -403,6 +422,38 @@ RLoopManager::RLoopManager(ROOT::RDF::Experimental::RDatasetSpec &&spec)
    ChangeSpec(std::move(spec));
 }
 
+#ifdef R__UNIX
+namespace {
+std::optional<std::string> GetRedirectedSampleId(std::string_view path, std::string_view datasetName)
+{
+   // Mimick the redirection done in TFile::Open to see if the path points to a FUSE-mounted EOS path.
+   // If so, we create a redirected sample ID with the full xroot URL.
+   TString expandedUrl(path.data());
+   gSystem->ExpandPathName(expandedUrl);
+   if (gEnv->GetValue("TFile.CrossProtocolRedirects", 1) == 1) {
+      TUrl fileurl(expandedUrl, /* default is file */ kTRUE);
+      if (strcmp(fileurl.GetProtocol(), "file") == 0) {
+         ssize_t len = getxattr(fileurl.GetFile(), "eos.url.xroot", nullptr, 0);
+         if (len > 0) {
+            std::string xurl(len, 0);
+            std::string fileNameFromUrl{fileurl.GetFile()};
+            if (getxattr(fileNameFromUrl.c_str(), "eos.url.xroot", &xurl[0], len) == len) {
+               // Sometimes the `getxattr` call may return an invalid URL due
+               // to the POSIX attribute not being yet completely filled by EOS.
+               if (auto baseName = fileNameFromUrl.substr(fileNameFromUrl.find_last_of("/") + 1);
+                   std::equal(baseName.crbegin(), baseName.crend(), xurl.crbegin())) {
+                  return xurl + '/' + datasetName.data();
+               }
+            }
+         }
+      }
+   }
+
+   return std::nullopt;
+}
+} // namespace
+#endif
+
 /**
  * @brief Changes the internal TTree held by the RLoopManager.
  *
@@ -441,6 +492,11 @@ void RLoopManager::ChangeSpec(ROOT::RDF::Experimental::RDatasetSpec &&spec)
          // is exposed to users via RSampleInfo and DefinePerSample).
          const auto sampleId = files[i] + '/' + trees[i];
          fSampleMap.insert({sampleId, &sample});
+#ifdef R__UNIX
+         // Also add redirected EOS xroot URL when available
+         if (auto redirectedSampleId = GetRedirectedSampleId(files[i], trees[i]))
+            fSampleMap.insert({redirectedSampleId.value(), &sample});
+#endif
       }
    }
    SetTree(std::move(chain));