fix a bunch of stuff

src/render.cpp

@@ -372,7 +372,6 @@ void update_and_render(Arena *arena, App_State &app, f32 delta_time_ms)
         i64 cluster_to_highlight = app.viewer.highlighted_cluster;
         ImGui::PushItemWidth(100.f);
         if (ImGui::InputScalar("##highlighted_cluster", ImGuiDataType_S64, &cluster_to_highlight, &step_i64, &step_fast_i64, "%u")) {
-          app.viewer.highlight_cluster = true;
           viewer_jump_to_cluster(app, cluster_to_highlight);
         }
         ImGui::PopItemWidth();

src/rntuple.cpp

@@ -111,11 +111,15 @@ void gather_ntuple_metadata(Arena *arena, RMicroFileReader &reader, const RNTupl
         pinfo->range.start = page_info.fLocator.GetPosition<u64>(); 
         pinfo->range.len = page_info.fLocator.fBytesOnStorage + (page_info.fHasChecksum) * checksum_size;
         pinfo->n_elems = page_info.fHasChecksum ? -page_info.fNElements : page_info.fNElements;
+        // This assert is just here because we're using u32 for cluster_id to save memory.
+        // If in the future we get RNTuples with more than 4B clusters we can just change the type to u64.
+        assert(cluster_desc.GetId() <= UINT_MAX);
         pinfo->cluster_id = cluster_desc.GetId();
 
         Cluster_Info_Node &cluster = clusters[pinfo->cluster_id];
         if (!cluster.first_page || pinfo->range.start < cluster.first_page->range.start) {
           cluster.first_page = pinfo;
+          pinfo->is_first_in_cluster = true;
         }
 
         if (UNLIKELY(!pinfo_head)) {
@@ -288,7 +292,9 @@ TFile_Data get_tfile_data(const Inspected_File &file, String8 ntpl_name)
   Root_File_Info root_file_info = get_root_file_info(file.name.c(), ntpl_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;
+  u32 tfile_obj_nbytes;
+  memcpy(&tfile_obj_nbytes, file.mem + root_file_info.tfile_obj_seek, sizeof(tfile_obj_nbytes));
+  tfile_data.rng_root_file_obj.len = bswap(tfile_obj_nbytes);
 
   // parse compression
   u32 compression_be;
@@ -384,66 +390,119 @@ Section find_section(App_State &app, u64 off, i64 hilite_cluster = -1)
   const RNTuple_Data &rdata = app.rndata;
   const TFile_Data &tdata = app.tfile_data;
   u64 rblob_sz = rdata.rblob_header_size; // @Incomplete
-  b8 hilite = false;
+  Section sec {};
 
   // TFile starting sections
   u64 root_header_padding = tdata.rng_root_file_obj.start - tdata.root_file_header_size;
-  if (off < tdata.rng_root_file_obj.start)
+  if (off < tdata.rng_root_file_obj.start) {
-    return { Sec_TFile_Header, { 0, tdata.root_file_header_size + root_header_padding }, 0, root_header_padding, hilite };
+    sec.id = Sec_TFile_Header;
-
+    sec.range = { 0, tdata.root_file_header_size + root_header_padding };
-  if (tdata.rng_root_file_obj.start <= off && off < tdata.rng_root_file_obj.end())
+    sec.post_size = root_header_padding;
-    return { Sec_TFile_Object, tdata.rng_root_file_obj, 0, 0, hilite };
+    return sec;
-
-  // TFile ending sections
-  if (tdata.rng_root_file_info_header.start <= off && off < tdata.rng_root_file_info.end())
-    return { Sec_TFile_Info, tdata.rng_root_file_info, tdata.rng_root_file_info_header.len, 0, hilite };
-
-  if (tdata.rng_root_file_free_header.start <= off && off < tdata.rng_root_file_free.end()) {
-    // printf("%lu\n", tdata.rng_root_file_free_header.len);
-    return { Sec_TFile_FreeList, tdata.rng_root_file_free, tdata.rng_root_file_free_header.len, 0, hilite };
   }
 
-  /// Handle pages
+  if (tdata.rng_root_file_obj.start <= off && off < tdata.rng_root_file_obj.end()) {
+    sec.id = Sec_TFile_Object;
+    sec.range = tdata.rng_root_file_obj;
+    return sec;
+  }
+
+  // TFile ending sections
+  if (tdata.rng_root_file_info_header.start <= off && off < tdata.rng_root_file_info.end()) {
+    sec.id = Sec_TFile_Info;
+    sec.range = tdata.rng_root_file_info;
+    sec.pre_size = tdata.rng_root_file_info_header.len;
+    return sec;
+  }
+
+  if (tdata.rng_root_file_free_header.start <= off && off < tdata.rng_root_file_free.end()) {
+    sec.id = Sec_TFile_FreeList;
+    sec.range = tdata.rng_root_file_free;
+    sec.pre_size = tdata.rng_root_file_free_header.len;
+    return sec;
+  }
+
+  /// Page fast lookup (relative to app.last_pinfo)
   {
     // fast case: `off` is in the same page info as previous `off`.
-    if (app.last_pinfo->range.start - rblob_sz < off && off < app.last_pinfo->range.end()) {
+    u64 pre_size = app.last_pinfo->is_first_in_cluster ? rblob_sz : 0;
-      hilite = hilite_cluster >= 0 && app.last_pinfo->cluster_id == (u64)hilite_cluster;
+    if (app.last_pinfo->range.start - pre_size < off && off < app.last_pinfo->range.end()) {
-      return { Sec_Page, app.last_pinfo->range, rblob_sz, app.last_pinfo->checksum_size(), hilite };
+      sec.id = Sec_Page;
+      sec.range = app.last_pinfo->range;
+      sec.pre_size = pre_size;
+      sec.post_size = app.last_pinfo->checksum_size();
+      sec.highlighted = hilite_cluster >= 0 && app.last_pinfo->cluster_id == (u64)hilite_cluster;
+      return sec;
     }
 
     // still fast case: `off is in the next page info as the previous.
     if (app.last_pinfo->next) // don't check if it's checksum, since it's the first byte of the page
       app.last_pinfo = app.last_pinfo->next;
-    if (app.last_pinfo && app.last_pinfo->range.start - rblob_sz <= off && off < app.last_pinfo->range.end()) {
+    if (app.last_pinfo) {
-      hilite = hilite_cluster >= 0 && app.last_pinfo->cluster_id == (u64)hilite_cluster;
+      u64 pre_size = app.last_pinfo->is_first_in_cluster ? rblob_sz : 0;
-      return { Sec_Page, app.last_pinfo->range, rblob_sz, app.last_pinfo->checksum_size(), hilite };
+      if (app.last_pinfo->range.start - pre_size <= off && off < app.last_pinfo->range.end()) {
+        sec.id = Sec_Page;
+        sec.range = app.last_pinfo->range;
+        sec.pre_size = pre_size;
+        sec.post_size = app.last_pinfo->checksum_size();
+        sec.highlighted = hilite_cluster >= 0 && app.last_pinfo->cluster_id == (u64)hilite_cluster;
+        return sec;
+      }
     }
   }
 
-  if (rdata.rng_anchor_key.start <= off && off < rdata.rng_anchor.end())
+  if (rdata.rng_anchor_key.start <= off && off < rdata.rng_anchor.end()) {
-    return { Sec_RNTuple_Anchor, rdata.rng_anchor, rdata.rng_anchor_key.len, 8, hilite };
+    sec.id = Sec_RNTuple_Anchor;
+    sec.range = rdata.rng_anchor;
+    sec.pre_size = rdata.rng_anchor_key.len;
+    sec.post_size = 8;
+    return sec;
+  }
 
-  if (rdata.rng_header.start - rblob_sz <= off && off < rdata.rng_header.end()) 
+  if (rdata.rng_header.start - rblob_sz <= off && off < rdata.rng_header.end()) {
-    return { Sec_RNTuple_Header, rdata.rng_header, rblob_sz, 8, hilite };
+    sec.id = Sec_RNTuple_Header;
+    sec.range = rdata.rng_header;
+    sec.pre_size = rblob_sz;
+    sec.post_size = 8;
+    return sec;
+  }
 
-  if (rdata.rng_footer.start - rblob_sz <= off && off < rdata.rng_footer.end())
+  if (rdata.rng_footer.start - rblob_sz <= off && off < rdata.rng_footer.end()) {
-    return { Sec_RNTuple_Footer, rdata.rng_footer, rblob_sz, 8, hilite };
+    sec.id = Sec_RNTuple_Footer;
+    sec.range = rdata.rng_footer;
+    sec.pre_size = rblob_sz;
+    sec.post_size = 8;
+    return sec;
+  }
 
-  if (rdata.rng_tkeys_list.start <= off && off < rdata.rng_tkeys_list.end()) 
+  if (rdata.rng_tkeys_list.start <= off && off < rdata.rng_tkeys_list.end()) {
-    return { Sec_TKey_List, rdata.rng_tkeys_list, rblob_sz, 0, hilite };
+    sec.id = Sec_TKey_List;
+    sec.range = rdata.rng_tkeys_list;
+    sec.pre_size = rblob_sz;
+    return sec;
+  }
 
   // @Speed
   for (u64 cg_idx = 0; cg_idx < rdata.n_cluster_groups; ++cg_idx) {
     Cluster_Group_Info &cg_info = rdata.cluster_groups[cg_idx];
-    if (cg_info.rng_page_list.start - rblob_sz <= off && off < cg_info.rng_page_list.end())
+    if (cg_info.rng_page_list.start - rblob_sz <= off && off < cg_info.rng_page_list.end()) {
-      return { Sec_Page_List, cg_info.rng_page_list, rblob_sz, 8, hilite };
+      sec.id = Sec_Page_List;
+      sec.range = cg_info.rng_page_list;
+      sec.pre_size = rblob_sz;
+      sec.post_size = 8;
+      return sec;
+    }
   }
 
   // Slow page group lookup, ideally only done once per render when last_pinfo is invalid.
   for (Page_Info_Chunk *chunk = rdata.page_chunks; chunk; chunk = chunk->next) {
     // If we're at the start of a chunk, return a fake Sec_Page used to highlight the RBlob header bytes. 
-    if (chunk->range.start - rblob_sz <= off && off < chunk->range.start)
+    if (chunk->range.start - rblob_sz <= off && off < chunk->range.start) {
-      return { Sec_Page, { chunk->range.start, 0 }, rblob_sz, 0, hilite };
+      sec.id = Sec_Page;
+      sec.range = { chunk->range.start, 0 };
+      sec.pre_size = rblob_sz;
+      return sec;
+    }
 
     if (chunk->range.start <= off && off < chunk->range.end()) {
       for (u64 group_idx = chunk->first_group; group_idx < rdata.n_page_groups; ++group_idx) {
@@ -452,10 +511,15 @@ Section find_section(App_State &app, u64 off, i64 hilite_cluster = -1)
           continue;
 
         for (Page_Info_Node *pinfo = group.first; pinfo; pinfo = pinfo->next) {
-          if (pinfo->range.start <= off && off < pinfo->range.end()) {
+          u64 pre_size = pinfo->is_first_in_cluster ? rblob_sz : 0;
+          if (pinfo->range.start - pre_size <= off && off < pinfo->range.end()) {
             app.last_pinfo = pinfo;
-            hilite = hilite_cluster >= 0 && pinfo->cluster_id == (u64)hilite_cluster;
+            sec.id = Sec_Page;
-            return { Sec_Page, pinfo->range, rblob_sz, pinfo->checksum_size(), hilite };
+            sec.range = pinfo->range;
+            sec.pre_size = pre_size;
+            sec.post_size = pinfo->checksum_size();
+            sec.highlighted = hilite_cluster >= 0 && pinfo->cluster_id == (u64)hilite_cluster;
+            return sec;
           }
         }
       }
@@ -467,7 +531,7 @@ Section find_section(App_State &app, u64 off, i64 hilite_cluster = -1)
   }
 
   return {};
-}
+}  
 
 struct Sec_Hover_Info {
   Byte_Range rng;
@@ -674,6 +738,14 @@ b8 hover_try_key(const Try_Sec_Hover_Fn &try_sec_hover, const u8 *data, u64 star
   }
 }
 
+internal
+b8 hover_try_object(const Try_Sec_Hover_Fn &try_sec_hover)
+{
+  return try_sec_hover.field<u16>("Version: %u")
+      || try_sec_hover.field<u32>("Unique ID: %u")
+      || try_sec_hover.field<u32>("Bits: %u");
+}
+
 // `off` is the absolute offset into `data`.
 internal
 Sec_Hover_Info get_section_hover_info(Arena *arena, Section section, u64 off, const u8 *data)
@@ -781,6 +853,8 @@ Sec_Hover_Info get_section_hover_info(Arena *arena, Section section, u64 off, co
           || try_sec_hover.field<u64>("Seek Dir: 0x%lX") 
           || try_sec_hover.field<u64>("Seek Parent: 0x%lX") 
           || try_sec_hover.field<u64>("Seek Keys: 0x%lX") 
+          || try_sec_hover.field<u16>("UUID Vers.Class: %u")
+          || try_sec_hover.field<u16>("UUID: %u", hover_display_val_le)
           ;
         } else {
           ok = ok || try_sec_hover.field<u16>("Version: %u") 
@@ -791,20 +865,18 @@ Sec_Hover_Info get_section_hover_info(Arena *arena, Section section, u64 off, co
           || try_sec_hover.field<u32>("Seek Dir: 0x%lX") 
           || try_sec_hover.field<u32>("Seek Parent: 0x%lX") 
           || try_sec_hover.field<u32>("Seek Keys: 0x%lX") 
+          || try_sec_hover.field<u16>("UUID Vers.Class: %u")
+          || try_sec_hover.field<u16>("UUID: %u", hover_display_val_le)
+          || try_sec_hover.range("Padding", 3 * sizeof(u32))
           ;
         }
-        ok = ok || try_sec_hover.field<u16>("UUID Vers.Class: %u")
-        || try_sec_hover.field<u16>("UUID: %u", hover_display_val_le)
-        || try_sec_hover.range("Padding", 3 * sizeof(u32))
-        ;
       }
     }
   } break;
 
   case Sec_RNTuple_Header: 
   case Sec_RNTuple_Footer: 
-  case Sec_Page_List: 
+  case Sec_Page_List: {
-  case Sec_Page: {
     hover_try_key(try_sec_hover, data, start)
     || try_sec_hover.maybe_rootzip(data, start)
     || try_sec_hover.range("Payload", section.range.len - section.post_size) // TODO: improve
@@ -812,8 +884,26 @@ Sec_Hover_Info get_section_hover_info(Arena *arena, Section section, u64 off, co
     ;
   } break;
 
+  case Sec_Page: {
+    b8 ok = section.pre_size && hover_try_key(try_sec_hover, data, start);
+    ok = ok || try_sec_hover.maybe_rootzip(data, start)
+    || try_sec_hover.range("Payload", section.range.len - section.post_size) // TODO: improve
+    || try_sec_hover.field<u64>("Checksum: 0x%lX", hover_display_val_le)
+    ;
+  } break;
+
   case Sec_TFile_Info: {
     hover_try_key(try_sec_hover, data, start)
+    || try_sec_hover.maybe_rootzip(data, start)
+    // || try_sec_hover.field<u32>("Byte Count: %u", [] (Arena *arena, String8_Node *prev, const char *fmt, u32 x) { 
+    //    x = bswap(x);
+    //    x -= 0x400000000; 
+    //    return push_str8_node_child(arena, prev, fmt, x);
+    // })
+    // || try_sec_hover.field<u16>("Version: %u")
+    // || hover_try_object(try_sec_hover)
+    // || try_sec_hover.field<u8>("Name: %u")
+    // || try_sec_hover.field<u32>("N Objects: %u")
     || try_sec_hover.range("Payload", section.range.len) // TODO: improve
     ;
   } break;

src/rntuple.h

@@ -15,7 +15,8 @@ struct Page_Info_Node {
 
   Byte_Range range; // len includes checksum
   i32 n_elems; // negative = page has checksum
-  u64 cluster_id;
+  u32 cluster_id;
+  b8 is_first_in_cluster;
 
   u64 checksum_size() const {
     return (n_elems < 0) * 8;

src/root/RMicroFileReader.cxx

@@ -1081,7 +1081,6 @@ Root_File_Info get_root_file_info(const char *fname, const char *ntplName, bool
      100, 0, RTFString{"TFile"}, fileNameStr, RTFString{},
      sizeof(RTFFile) + fileNameStr.GetSize() + RTFString{}.GetSize() + RTFUUID{}.GetSize()
    };
-   fileInfo.tfile_obj_nbytes = fileHeader.GetSize();
    fileInfo.version_seek = offsetof(RTFHeader, fVersion);
    fileInfo.compression_seek = is_big_file ? offsetof(RTFHeader, fInfoLong.fCompress) : offsetof(RTFHeader, fInfoShort.fCompress);
    fileInfo.info_seek_seek   = is_big_file ? offsetof(RTFHeader, fInfoLong.fSeekInfo) : offsetof(RTFHeader, fInfoShort.fSeekInfo);
@@ -1217,9 +1216,11 @@ RMicroFileReader::GetNTupleProper(const char *ntupleName)
    fileInfo.anchor_nbytes = objNbytes;
    {
       // @Incomplete: each section has a differently-sized RBlob, we need to account for that!
+      RTFString blobName { kBlobClassName };
       RTFKey dummy;
-      dummy.MakeBigKey();
+      if (fileHeader.IsBigFile())
-      fileInfo.rblob_key_header_nbytes = dummy.GetHeaderSize();
+         dummy.MakeBigKey();
+      fileInfo.rblob_key_header_nbytes = dummy.GetHeaderSize() + blobName.GetSize() + 2 * RTFString{}.GetSize();
    }
    // @----
    ReadBuffer(ntuple, objNbytes, offset);

src/root/RMicroFileReader.hxx

@@ -47,7 +47,6 @@ struct RNTuple_File_Info {
 struct Root_File_Info {
   std::uint64_t tfile_header_nbytes;
   std::uint64_t tfile_obj_seek;
-  std::uint64_t tfile_obj_nbytes;
   std::uint64_t version_seek;
   std::uint64_t compression_seek;
   std::uint64_t info_seek_seek;