Make it possible read "old" podio files (#232)

* Make it possible read "old" podio files

* Make it possible to read old SIO files

include/podio/SIOBlock.h

@@ -60,10 +60,10 @@ namespace podio {
   class SIOCollectionIDTableBlock : public sio::block {
   public:
     SIOCollectionIDTableBlock() :
-      sio::block("CollectionIDs", sio::version::encode_version(0, 1)) {}
+      sio::block("CollectionIDs", sio::version::encode_version(0, 2)) {}
 
     SIOCollectionIDTableBlock(podio::EventStore* store) :
-      sio::block("CollectionIDs", sio::version::encode_version(0, 1)),
+      sio::block("CollectionIDs", sio::version::encode_version(0, 2)),
       _store(store), _table(store->getCollectionIDTable()) {}
 
     SIOCollectionIDTableBlock(const SIOCollectionIDTableBlock&) = delete;

src/ROOTReader.cc

@@ -144,12 +144,21 @@ namespace podio {
     m_table = new CollectionIDTable();
     metadatatree->SetBranchAddress("CollectionIDs", &m_table);
 
-    auto collectionInfo = new std::vector<std::tuple<int, std::string, bool>>;
-    metadatatree->SetBranchAddress("CollectionTypeInfo", &collectionInfo);
-    metadatatree->GetEntry(0);
-
-    createCollectionBranches(*collectionInfo);
-    delete collectionInfo;
+    // Check if the CollectionTypeInfo branch is there and assume that the file
+    // has been written with with podio pre #197 (<0.13.1) if that is not the case
+    if (auto* collInfoBranch = root_utils::getBranch(metadatatree, "CollectionTypeInfo")) {
+      auto collectionInfo = new std::vector<root_utils::CollectionInfoT>;
+      collInfoBranch->SetAddress(&collectionInfo);
+      metadatatree->GetEntry(0);
+      createCollectionBranches(*collectionInfo);
+      delete collectionInfo;
+    } else {
+      std::cout << "PODIO: Reconstructing CollectionTypeInfo branch from other sources in file: \'"
+                << m_chain->GetFile()->GetName() << "\'" << std::endl;
+      metadatatree->GetEntry(0);
+      const auto collectionInfo = root_utils::reconstructCollectionInfo(m_chain, *m_table);
+      createCollectionBranches(collectionInfo);
+    }
   }
 
   void ROOTReader::closeFile(){
@@ -196,7 +205,7 @@ namespace podio {
     m_inputs.clear();
   }
 
- void ROOTReader::createCollectionBranches(const std::vector<std::tuple<int, std::string, bool>>& collInfo) {
+ void ROOTReader::createCollectionBranches(const std::vector<root_utils::CollectionInfoT>& collInfo) {
     size_t collectionIndex{0};
 
     for (const auto& [collID, collType, isSubsetColl] : collInfo) {

src/ROOTWriter.cc

@@ -104,7 +104,7 @@ void ROOTWriter::setBranches(const std::vector<StoreCollection>& collections) {
     m_metadatatree->Branch("CollectionIDs", collIDTable);
 
     // collectionID, collection type, subset collection
-    std::vector<std::tuple<int, std::string, bool>> collectionInfo;
+    std::vector<root_utils::CollectionInfoT> collectionInfo;
     collectionInfo.reserve(m_collectionsToWrite.size());
     for (const auto& name : m_collectionsToWrite) {
       const auto collID = collIDTable->collectionID(name);

src/SIOBlock.cc

@@ -12,13 +12,15 @@
 #endif
 
 namespace podio {
-  void SIOCollectionIDTableBlock::read(sio::read_device& device, sio::version_type) {
+  void SIOCollectionIDTableBlock::read(sio::read_device& device, sio::version_type version) {
     std::vector<std::string> names;
     std::vector<int> ids;
     device.data(names);
     device.data(ids);
     device.data(_types);
-    device.data(_isSubsetColl);
+    if (version >= sio::version::encode_version(0, 2)) {
+      device.data(_isSubsetColl);
+    }
 
     _table = new CollectionIDTable(std::move(ids), std::move(names));
   }

src/SIOReader.cc

@@ -125,7 +125,8 @@ namespace podio {
     m_blocks.push_back(m_eventMetaData);
 
     for (size_t i = 0; i < m_typeNames.size(); ++i) {
-      auto blk = podio::SIOBlockFactory::instance().createBlock(m_typeNames[i], m_table->names()[i], m_subsetCollectionBits[i]);
+      const bool subsetColl = m_subsetCollectionBits.size() && m_subsetCollectionBits[i];
+      auto blk = podio::SIOBlockFactory::instance().createBlock(m_typeNames[i], m_table->names()[i], subsetColl);
       m_blocks.push_back(blk);
       m_inputs.emplace_back(blk->getCollection(), m_table->names()[i]);
     }

src/rootUtils.h

@@ -3,12 +3,16 @@
 
 #include "podio/CollectionBase.h"
 #include "podio/CollectionBranches.h"
+#include "podio/CollectionIDTable.h"
 
 #include "TBranch.h"
 #include "TClass.h"
 
+#include <iostream>
 #include <vector>
 #include <string>
+#include <tuple>
+#include <string_view>
 
 namespace podio::root_utils {
 // Workaround slow branch retrieval for 6.22/06 performance degradation
@@ -47,6 +51,43 @@ inline void setCollectionAddresses(podio::CollectionBase* collection, const Coll
   }
 }
 
+// A collection of additional information that describes the collection: the
+// collectionID, the collection (data) type, and whether it is a subset
+// collection
+using CollectionInfoT = std::tuple<int, std::string, bool>;
+
+/**
+ * reconstruct the collection info from information that is available from other
+ * trees in the file.
+ *
+ * NOTE: This function is only supposed to be called if there is no
+ * "CollectionTypeInfo" branch in the metadata tree, as it assumes that the file
+ * has been written with podio previous to #197 where there were no subset
+ * collections
+ */
+inline auto reconstructCollectionInfo(TTree* eventTree, podio::CollectionIDTable const& idTable) {
+  std::vector<CollectionInfoT> collInfo;
+
+  for (size_t iColl = 0; iColl < idTable.names().size(); ++iColl) {
+    const auto collID = idTable.ids()[iColl];
+    const auto& name = idTable.names()[iColl];
+
+    if (auto branch = getBranch(eventTree, name.c_str())) {
+      const std::string_view bufferClassName = branch->GetClassName();
+      // this comes with vector<...Data>, where we only care about the ...
+      std::string_view dataClass = bufferClassName;
+      dataClass.remove_suffix(5);
+      const auto collClass = std::string(dataClass.substr(7)) + "Collection";
+      // Assume that there are no subset collections in "old files"
+      collInfo.emplace_back(collID, std::move(collClass), false);
+    } else {
+      std::cerr << "Problems reconstructing collection info for collection: \'" << name << "\'\n";
+    }
+  }
+
+  return collInfo;
+}
+
 }
 
 #endif