Modify the functions in FileFromDirac

To fulfill the requirements in #9, modify FileFromDirac

• Filter certain sites
• Sort the sites for MC

python/PRConfig/FilesFromDirac.py

@@ -111,9 +111,9 @@ def get_access_urls_data(bkkpath, sites_to_remove=[], max_files=500):
 
     # ensure files are always in same order
     file_list.sort()
-    if any(file.endswith("raw") for file in file_list):
-        raise NotImplementedError(
-            "File ending with 'raw' found, please write a script manually")
+#   if any(file.endswith("raw") for file in file_list):
+#       raise NotImplementedError(
+#           "File ending with 'raw' found, please write a script manually")
 
     print(
         "#### Checking output of . /cvmfs/lhcb.cern.ch/lib/LbEnv --quiet; lb-dirac dirac-dms-lfn-accessURL --Terminal"
@@ -139,7+139,7 @@
             if successful:
                 m = re.match(r' +([^: ]*) *: *(.*)', line)
                 if m:
                     urls[m.group(1)].append(m.group(2))
 
     # Get the first URL (if more than one) for each LFN, while skipping
     # LFNs for which we couldn't find an URL (e.g. a site was down).
@@ -151,7+151,7 @@
         lfn for lfn in lfns_tmp if not any(site in lfn for site in excluded)
     ]
     lfns = sorted(lfns, key=lambda str: not "eoslhcb.cern.ch" in str)
+    if any(file.endswith("raw") for file in file_list):
+        lfns = [lfn for lfn in lfns if "eoslhcb.cern.ch" in lfn]
 
     return lfns
 
     # TODO warn if some of the first N files was not resolved to a URL
     # since then one would get numerically different results.
+
+if __name__ == "__main__":
+    # bkk_path = ("/MC/Upgrade/Beam7000GeV-Upgrade-MagDown-Nu7.6-25ns-Pythia8"
+    # "/Sim10-Up08/Digi15-Up04")
+    # evt_type = "30000000"
+    bkk_path_for_data = (
+        "/LHCb/Collision18/Beam6500GeV-VeloClosed-MagDown/Real\ Data/Reco18/Stripping34/90000000/BHADRON.MDST"
+    )
+    # for url in get_access_urls_mc(bkk_path, evt_type, ['XDIGI']):
+    for url in get_access_urls_data(bkk_path_for_data, max_files=10):
+        print(url)