fix bug in RMicroFileReader

Makefile

@@ -32,6 +32,9 @@ $(ROOT_IFACE): src/root/RMicroFileReader.cxx
 d: $(ROOT_IFACE_DBG) build/imgui.o
 	$(MOLD) $(CXX) -DDEBUG -g -O0 -DENABLE_ASAN $(CFLAGS) $(INC) $(ROOTFLAGS) -o rntviewer src/rntviewer.cpp build/imgui.o -lasan $(ROOT_IFACE_DBG) $(LIBS) $(ROOTLIBS)
 	
+asan: $(ROOT_IFACE_DBG) build/imgui.o
+	$(MOLD) $(CXX) -DDEBUG -g -O0 $(CFLAGS) -fsanitize=address $(INC) $(ROOTFLAGS) -o rntviewer src/rntviewer.cpp build/imgui.o $(ROOT_IFACE_DBG) $(LIBS) $(ROOTLIBS)
+
 noasan: $(ROOT_IFACE_DBG) build/imgui.o
 	$(MOLD) $(CXX) -DDEBUG -g -O0 $(CFLAGS) -fsanitize=undefined $(INC) $(ROOTFLAGS) -o rntviewer src/rntviewer.cpp build/imgui.o $(ROOT_IFACE_DBG) $(LIBS) $(ROOTLIBS)
 

src/mem.cpp

@@ -120,24 +120,7 @@ void *arena_push_impl(Arena *arena, u64 size)
 
 template <typename T>
 internal
-T *arena_push(Arena *arena)
-{
-  void *mem = arena_push_impl(arena, sizeof(T));
-  return (T *)mem;
-}
-
-template <typename T>
-internal
-T *arena_push_zeroed(Arena *arena)
-{
-  T *mem = arena_push<T>(arena);
-  memset(mem, 0, sizeof(T));
-  return mem;
-}
-
-template <typename T>
-internal
-T *arena_push_array_no_zero(Arena *arena, u64 count)
+T *arena_push_array_nozero(Arena *arena, u64 count)
 {
   void *mem = arena_push_impl(arena, sizeof(T) * count);
   return (T *)mem;
@@ -147,11 +130,25 @@ template <typename T>
 internal
 T *arena_push_array(Arena *arena, u64 count)
 {
-  T *ary = arena_push_array_no_zero<T>(arena, count);
+  T *ary = arena_push_array_nozero<T>(arena, count);
   memset(ary, 0, sizeof(T) * count);
   return ary;
 }
 
+template <typename T>
+internal
+T *arena_push(Arena *arena)
+{
+  return arena_push_array<T>(arena, 1);
+}
+
+template <typename T>
+internal
+T *arena_push_nozero(Arena *arena)
+{
+  return arena_push_array_nozero<T>(arena, 1);
+}
+
 internal
 u8 *arena_push_contiguous(Arena *arena, u64 size)
 {

src/render.cpp

@@ -47,6 +47,17 @@ u32 mem_edit_bg_color_fn(const u8 *, u64 off, const void *user_data)
   if (rdata.rng_anchor.start <= off && off <= rdata.rng_anchor.end()) return COL(app->vsettings.col_anchor);
   if (rdata.rng_header.start <= off && off <= rdata.rng_header.end()) return COL(app->vsettings.col_header);
   if (rdata.rng_footer.start <= off && off <= rdata.rng_footer.end()) return COL(app->vsettings.col_footer);
+  // @Speed!
+  for (u64 group_idx = 0; group_idx < rdata.n_page_groups; ++group_idx) {
+    const Page_Info_Group &group = rdata.page_groups[group_idx];
+    if (off < group.range.start || off > group.range.end()) 
+      continue;
+
+    for (Page_Info_Node *pinfo = group.first; pinfo; pinfo = pinfo->next) {
+      if (pinfo->range.start == off) return COL(app->vsettings.col_page_start);
+      if (pinfo->range.start < off && off <= pinfo->range.end()) return COL(app->vsettings.col_page);
+    }
+  }
 #undef COL
 
   return IM_COL32(0, 0, 0, 0);
@@ -74,6 +85,8 @@ Viewer_Settings make_viewer_settings()
   COL(col_footer, 50, 0, 150);
   COL(col_key,    0, 100, 50);
   COL(col_tfile,  90, 90, 90);
+  COL(col_page,   125, 0, 125);
+  COL(col_page_start, 75, 0, 75);
 #undef COL
   return settings;
 }
@@ -91,7 +104,7 @@ void update_and_render(Arena *arena, App_State &app, f32 delta_time_ms)
 
   const u64 inspected_max_size = 2048;
   u64 text_buf_size = min(app.inspected_file_size * 2, inspected_max_size);
-  char *text_buf = arena_push_array_no_zero<char>(scratch.arena, text_buf_size + 1);
+  char *text_buf = arena_push_array_nozero<char>(scratch.arena, text_buf_size + 1);
   // Convert file content to human readable
   // @Speed: maybe do this only when the file changes
   for (u64 i = 0; i < text_buf_size / 2; ++i)
@@ -160,6 +173,10 @@ void update_and_render(Arena *arena, App_State &app, f32 delta_time_ms)
       ImGui::SameLine(); 
       if (ImGui::Button("TKey Header"))    {} // TODO app.vsettings.base_display_addr = app.rndatarng_footer.start;
 
+      ImGui::Separator();
+      ImGui::Text("Num pages: %lu", app.rndata.n_pages);
+      ImGui::Text("Num elements: %lu", app.rndata.n_elems);
+
       ImGui::EndTable();
     }
 

src/render.h

@@ -4,6 +4,8 @@ struct Viewer_Settings {
   float col_footer[3];
   float col_key[3];
   float col_tfile[3];
+  float col_page[3];
+  float col_page_start[3];
 
   u64 base_display_addr;
 };

src/rntuple.cpp

@@ -63,26 +63,80 @@ ROOT::Experimental::RNTupleDescriptor create_descriptor(Arena *arena, RMicroFile
 }
 
 internal
-void gather_metadata(Arena *arena, RMicroFileReader &reader, const RNTuple_File_Info &info)
+void gather_metadata(Arena *arena, RMicroFileReader &reader, const RNTuple_File_Info &info, RNTuple_Data &rndata)
 {
   using namespace ROOT::Experimental;
   using namespace ROOT::Experimental::Internal;
 
   RNTupleDescriptor descriptor = create_descriptor(arena, reader, info);
-
+  u64 n_pages = 0;
+  u64 n_elems = 0;
+  Page_Info_Node *pinfo_head = nullptr, *pinfo_tail = nullptr;
+  
   // for all clusters ...
-  DescriptorId_t cluster_id = descriptor.FindClusterId(0, 0);
-  while (cluster_id != kInvalidDescriptorId) {
-    const RClusterDescriptor &cluster_desc = descriptor.GetClusterDescriptor(cluster_id);
+  for (const RClusterDescriptor &cluster_desc : descriptor.GetClusterIterable()) {
     for (const auto &[col_id, col_range] : cluster_desc.GetColumnRangeIterable()) {
       // TODO gather column metadata
       // TODO gather page metadata
+      // fprintf(stderr, "col_id: %d\n", col_id);
+
+      // TODO!! insert page_info sorted by byte range!
       const auto &page_range = cluster_desc.GetPageRange(col_id);
       for (const auto &page_info : page_range.fPageInfos) {
-        
+        Page_Info_Node *pinfo = arena_push<Page_Info_Node>(arena);
+        pinfo->range.start = page_info.fLocator.GetPosition<u64>(); 
+        pinfo->range.len = page_info.fLocator.fBytesOnStorage;
+        pinfo->n_elems = page_info.fNElements;
+        if (pinfo_head) {
+          assert(pinfo_tail);
+          if (pinfo->range.start < pinfo_head->range.start) {
+            pinfo->next = pinfo_head;
+            pinfo_head = pinfo;
+          } else {
+            assert(pinfo->range.start > pinfo_tail->range.end());
+            pinfo_tail->next = pinfo;
+            pinfo_tail = pinfo;
+          }
+        } else {
+          assert(!pinfo_tail);
+          pinfo_head = pinfo_tail = pinfo;
+        }
+        ++n_pages;
+        n_elems += page_info.fNElements;
       }
     }
   }
+
+  fprintf(stderr, "Loaded %lu pages\n", n_pages);
+
+  // Create page groups
+  const u64 GROUP_SIZE = 1000;
+  Page_Info_Group *groups = arena_push_array_nozero<Page_Info_Group>(arena, n_pages / GROUP_SIZE + 1);
+  Page_Info_Group *cur_group = groups, *prev_group = nullptr;
+  Page_Info_Node *last = nullptr;
+  u64 idx = 0;
+  u64 n_groups = 0;
+  for (Page_Info_Node *pinfo = pinfo_head; pinfo; pinfo = pinfo->next) {
+    if (idx++ % GROUP_SIZE == 0) {
+      ++n_groups;
+      cur_group->first = pinfo;
+      cur_group->range.start = pinfo->range.start;
+      if (prev_group)
+        prev_group->range.len = cur_group->first - prev_group->first;
+      prev_group = cur_group;
+      printf("group %lu -> %lu\n", cur_group->range.start, cur_group->range.end());
+      ++cur_group;
+    }
+    last = pinfo;
+  }
+  if (last)
+    prev_group->range.len = last->range.end() - prev_group->range.start;
+
+  rndata.pages = pinfo_head;
+  rndata.page_groups = groups;
+  rndata.n_page_groups = n_groups;
+  rndata.n_pages = n_pages;
+  rndata.n_elems = n_elems;
 }
 
 internal
@@ -109,7 +163,7 @@ RNTuple_Data get_rntuple_data(Arena *arena, const char *fname, const char *ntpl_
     rndata.rblob_header_size = file_info.rblob_key_header_nbytes;
     rndata.root_file_header_size = file_info.tfile_header_nbytes;
 
-    gather_metadata(arena, file_reader, file_info);
+    gather_metadata(arena, file_reader, file_info, rndata);
   }
 
   return rndata;

src/rntuple.h

@@ -5,6 +5,18 @@ struct Byte_Range {
   u64 end() const { return start + len; }
 };
 
+struct Page_Info_Node {
+  Page_Info_Node *next;
+
+  Byte_Range range;
+  u64 n_elems;
+};
+
+struct Page_Info_Group {
+  Byte_Range range;
+  Page_Info_Node *first;
+};
+
 struct RNTuple_Data {
   struct {
     u16 epoch, major, minor, patch;
@@ -16,4 +28,12 @@ struct RNTuple_Data {
   Byte_Range rng_anchor_key;
   Byte_Range rng_header;
   Byte_Range rng_footer;
+
+  Page_Info_Node *pages;
+  u64 n_pages;
+  // total number of elements of all pages
+  u64 n_elems;
+
+  Page_Info_Group *page_groups;
+  u64 n_page_groups;
 };

src/root/RMicroFileReader.cxx

@@ -1070,7 +1070,7 @@ static size_t ComputeNumChunks(size_t nbytes, size_t maxChunkSize)
 
 struct RMicroFileReader::Impl {
   std::unique_ptr<ROOT::Internal::RRawFile> fRawFile;
-  std::uint64_t fMaxBlobSize;
+  std::uint64_t fMaxBlobSize = 0;
 };
 
 RMicroFileReader::RMicroFileReader(const char *fname) {
@@ -1221,6 +1221,7 @@ void RMicroFileReader::ReadBuffer(void *buffer, size_t nbytes, std::uint64_t off
 {
    size_t nread;
    uint64_t maxBlobSize = impl->fMaxBlobSize;
+   assert(maxBlobSize != -1);
    if (maxBlobSize == 0 || nbytes <= maxBlobSize) {
       // Fast path: read single blob
       nread = impl->fRawFile->ReadAt(buffer, nbytes, offset);

src/str.cpp

@@ -5,7 +5,7 @@ String8 push_str8fv(Arena *arena, const char *fmt, va_list args)
   va_copy(args2, args);
   u32 needed_bytes = vsnprintf(0, 0, fmt, args) + 1;
   String8 result = {};
-  result.str = arena_push_array_no_zero<u8>(arena, needed_bytes);
+  result.str = arena_push_array_nozero<u8>(arena, needed_bytes);
   result.size = vsnprintf((char*)result.str, needed_bytes, fmt, args2);
   result.str[result.size] = 0;
   va_end(args2);