properly handle big tfiles

src/rntuple.cpp

@@ -234,15 +234,13 @@ TFile_Data get_tfile_data(const Inspected_File &file)
 { 
   TFile_Data tfile_data{};
 
-  Root_File_Info root_file_info = get_root_file_info(file.name.c());
-  tfile_data.root_file_header_size = root_file_info.tfile_header_nbytes;
-  tfile_data.rng_root_file_obj.start = root_file_info.tfile_obj_seek;
-  tfile_data.rng_root_file_obj.len = root_file_info.tfile_obj_nbytes;
-
   // parse root version
+  const u64 version_seek = 4;
   u32 version_be;  
-  memcpy(&version_be, file.mem + root_file_info.version_seek, sizeof(version_be));
+  memcpy(&version_be, file.mem + version_seek, sizeof(version_be));
   u32 version = bswap_32(version_be);
+  b32 is_big_file = version > 1000000;
+  version -= is_big_file * 1000000;
   u32 version_major = version / 10000;
   u32 version_minor = (version - version_major * 10000) / 100;
   u32 version_patch = (version - version_major * 10000 - version_minor * 100);
@@ -250,25 +248,42 @@ TFile_Data get_tfile_data(const Inspected_File &file)
   tfile_data.root_version_minor = (u16)version_minor;
   tfile_data.root_version_patch = (u16)version_patch;
   
+  Root_File_Info root_file_info = get_root_file_info(file.name.c(), is_big_file);
+  tfile_data.root_file_header_size = root_file_info.tfile_header_nbytes;
+  tfile_data.rng_root_file_obj.start = root_file_info.tfile_obj_seek;
+  tfile_data.rng_root_file_obj.len = root_file_info.tfile_obj_nbytes;
+
   // parse compression
   u32 compression_be;
   memcpy(&compression_be, file.mem + root_file_info.compression_seek, sizeof(compression_be));
   tfile_data.compression = bswap_32(compression_be);
 
   // parse info
-  u32 info_seek_be;
+  if (is_big_file) {
-  u32 info_nbytes_be;
+    u64 info_seek_be;
+    memcpy(&info_seek_be, file.mem + root_file_info.info_seek_seek, sizeof(info_seek_be));
+    tfile_data.rng_root_file_info.start = bswap_64(info_seek_be);
+  } else {
+    u32 info_seek_be;
     memcpy(&info_seek_be, file.mem + root_file_info.info_seek_seek, sizeof(info_seek_be));
-  memcpy(&info_nbytes_be, file.mem + root_file_info.info_nbytes_seek, sizeof(info_nbytes_be));
     tfile_data.rng_root_file_info.start = bswap_32(info_seek_be);
+  }
+  u32 info_nbytes_be;
+  memcpy(&info_nbytes_be, file.mem + root_file_info.info_nbytes_seek, sizeof(info_nbytes_be));
   tfile_data.rng_root_file_info.len = bswap_32(info_nbytes_be);
 
   // parse free
-  u32 free_seek_be;
+  if (is_big_file) {
-  u32 free_nbytes_be;
+    u64 free_seek_be;
+    memcpy(&free_seek_be, file.mem + root_file_info.free_seek_seek, sizeof(free_seek_be));
+    tfile_data.rng_root_file_free.start = bswap_64(free_seek_be);
+  } else {
+    u32 free_seek_be;
     memcpy(&free_seek_be, file.mem + root_file_info.free_seek_seek, sizeof(free_seek_be));
-  memcpy(&free_nbytes_be, file.mem + root_file_info.free_nbytes_seek, sizeof(free_nbytes_be));
     tfile_data.rng_root_file_free.start = bswap_32(free_seek_be);
+  }
+  u32 free_nbytes_be;
+  memcpy(&free_nbytes_be, file.mem + root_file_info.free_nbytes_seek, sizeof(free_nbytes_be));
   tfile_data.rng_root_file_free.len = bswap_32(free_nbytes_be);
   
   return tfile_data;

src/root/RMicroFileReader.cxx

@@ -1068,7 +1068,7 @@ static size_t ComputeNumChunks(size_t nbytes, size_t maxChunkSize)
 //                         PUBLIC INTERFACE
 // =========================================================================================
 
-Root_File_Info get_root_file_info(const char *fname)
+Root_File_Info get_root_file_info(const char *fname, bool is_big_file)
 {
    RTFString fileName { fname };
    Root_File_Info fileInfo {};
@@ -1076,13 +1076,11 @@ Root_File_Info get_root_file_info(const char *fname)
    fileInfo.tfile_obj_seek = 100; // kBEGIN
    fileInfo.tfile_obj_nbytes = sizeof(RTFFile) + fileName.GetSize() + RTFString{}.GetSize() + RTFUUID{}.GetSize();
    fileInfo.version_seek = offsetof(RTFHeader, fVersion);
-   fileInfo.compression_seek = offsetof(RTFHeader, fInfoShort.fCompress);
+   fileInfo.compression_seek = is_big_file ? offsetof(RTFHeader, fInfoLong.fCompress) : offsetof(RTFHeader, fInfoShort.fCompress);
-   fileInfo.info_seek_seek = offsetof(RTFHeader, fInfoShort.fSeekInfo);
+   fileInfo.info_seek_seek =   is_big_file ? offsetof(RTFHeader, fInfoLong.fSeekInfo) : offsetof(RTFHeader, fInfoShort.fSeekInfo);
-   fileInfo.info_seek_sizeof = 4; // TODO: check if big file
+   fileInfo.info_nbytes_seek = is_big_file ? offsetof(RTFHeader, fInfoLong.fNbytesInfo) : offsetof(RTFHeader, fInfoShort.fNbytesInfo);
-   fileInfo.info_nbytes_seek = offsetof(RTFHeader, fInfoShort.fNbytesInfo);
+   fileInfo.free_seek_seek =   is_big_file ? offsetof(RTFHeader, fInfoLong.fSeekFree) : offsetof(RTFHeader, fInfoShort.fSeekFree);
-   fileInfo.free_seek_seek = offsetof(RTFHeader, fInfoShort.fSeekFree);
+   fileInfo.free_nbytes_seek = is_big_file ? offsetof(RTFHeader, fInfoLong.fNbytesFree) : offsetof(RTFHeader, fInfoShort.fNbytesFree);
-   fileInfo.free_seek_sizeof = 4; // TODO: check if big file
-   fileInfo.free_nbytes_seek = offsetof(RTFHeader, fInfoShort.fNbytesFree);
    return fileInfo;
 }
 

src/root/RMicroFileReader.hxx

@@ -47,14 +47,12 @@ struct Root_File_Info {
   std::uint64_t version_seek;
   std::uint64_t compression_seek;
   std::uint64_t info_seek_seek;
-  std::uint64_t info_seek_sizeof;
   std::uint64_t info_nbytes_seek;
   std::uint64_t free_seek_seek;
-  std::uint64_t free_seek_sizeof;
   std::uint64_t free_nbytes_seek;
 };
 
-Root_File_Info get_root_file_info(const char *fname);
+Root_File_Info get_root_file_info(const char *fname, bool is_big_file);
 
 class RMicroFileReader final {
 public: