Add CDvInfo struct

Author: nicklan

ffi/examples/read-table/CMakeLists.txt

@@ -14,8 +14,10 @@ include(CTest)
 set(TestRunner "../../../tests/read-table-testing/run_test.sh")
 set(DatPath "../../../../acceptance/tests/dat/out/reader_tests/generated")
 set(ExpectedPath "../../../tests/read-table-testing/expected-data")
+set(KernelTestPath "../../../../kernel/tests/data")
 add_test(NAME read_and_print_all_prim COMMAND ${TestRunner} ${DatPath}/all_primitive_types/delta/ ${ExpectedPath}/all-prim-types.expected)
 add_test(NAME read_and_print_basic_partitioned COMMAND ${TestRunner} ${DatPath}/basic_partitioned/delta/ ${ExpectedPath}/basic-partitioned.expected)
+add_test(NAME read_and_print_with_dv_small COMMAND ${TestRunner} ${KernelTestPath}/table-with-dv-small/ ${ExpectedPath}/table-with-dv-small.expected)
 
 if(WIN32)
   set(CMAKE_C_FLAGS_DEBUG "/MT")

ffi/examples/read-table/read_table.c

@@ -49,7 +49,7 @@ void scan_row_callback(
   KernelStringSlice path,
   int64_t size,
   const Stats* stats,
-  const DvInfo* dv_info,
+  const CDvInfo* cdv_info,
   const Expression* transform,
   const CStringMap* partition_values)
 {
@@ -62,18 +62,25 @@ void scan_row_callback(
     print_diag(" [no stats])\n");
   }
   KernelStringSlice table_root_slice = { context->table_root, strlen(context->table_root) };
-  ExternResultKernelBoolSlice selection_vector_res =
-    selection_vector_from_dv(dv_info, context->engine, table_root_slice);
-  if (selection_vector_res.tag != OkKernelBoolSlice) {
-    printf("Could not get selection vector from kernel\n");
-    exit(-1);
-  }
-  KernelBoolSlice selection_vector = selection_vector_res.ok;
-  if (selection_vector.len > 0) {
-    print_diag("  Selection vector for this file:\n");
-    print_selection_vector("    ", &selection_vector);
+  KernelBoolSlice selection_vector;
+
+  if (cdv_info->has_vector) {
+    ExternResultKernelBoolSlice selection_vector_res =
+      selection_vector_from_dv(cdv_info->info, context->engine, table_root_slice);
+    if (selection_vector_res.tag != OkKernelBoolSlice) {
+      printf("Could not get selection vector from kernel\n");
+      exit(-1);
+    }
+    selection_vector = selection_vector_res.ok;
+    if (selection_vector.len > 0) {
+      print_diag("  Selection vector for this file:\n");
+      print_selection_vector("    ", &selection_vector);
+    } else {
+      print_diag("  No selection vector for this file\n");
+    }
   } else {
     print_diag("  No selection vector for this file\n");
+    selection_vector.len = 0;
   }
   context->partition_values = partition_values;
   print_partition_info(context, partition_values);

ffi/src/scan.rs

@@ -275,13 +275,23 @@ pub struct Stats {
     pub num_records: u64,
 }
 
+/// Contains information that can be used to get a selection vector. If `has_vector` is false, that
+/// indicates there is no selection vector to consider. It is always possible to get a vector out of
+/// a `DvInfo`, but if `has_vector` is false it will just be an empty vector (indicating all
+/// selected).
+#[repr(C)]
+pub struct CDvInfo<'a> {
+    info: &'a DvInfo,
+    has_vector: bool,
+}
+
 /// This callback will be invoked for each valid file that needs to be read for a scan.
 ///
 /// The arguments to the callback are:
 /// * `context`: a `void*` context this can be anything that engine needs to pass through to each call
 /// * `path`: a `KernelStringSlice` which is the path to the file
 /// * `size`: an `i64` which is the size of the file
-/// * `dv_info`: a [`DvInfo`] struct, which allows getting the selection vector for this file
+/// * `dv_info`: a [`CDvInfo`] struct, which allows getting the selection vector for this file
 /// * `transform`: An optional expression that, if not `NULL`, _must_ be applied to physical data to
 ///   convert it to the correct logical format. If this is `NULL`, no transform is needed.
 /// * `partition_values`: [DEPRECATED] a `HashMap<String, String>` which are partition values
@@ -290,7 +300,7 @@ type CScanCallback = extern "C" fn(
     path: KernelStringSlice,
     size: i64,
     stats: Option<&Stats>,
-    dv_info: &DvInfo,
+    dv_info: &CDvInfo,
     transform: Option<&Expression>,
     partition_map: &CStringMap,
 );
@@ -430,12 +440,16 @@ fn rust_callback(
     let stats = kernel_stats.map(|ks| Stats {
         num_records: ks.num_records,
     });
+    let cdv_info = CDvInfo {
+        info: &dv_info,
+        has_vector: dv_info.has_vector(),
+    };
     (context.callback)(
         context.engine_context,
         kernel_string_slice!(path),
         size,
         stats.as_ref(),
-        &dv_info,
+        &cdv_info,
         transform.as_ref(),
         &partition_map,
     );

ffi/tests/read-table-testing/expected-data/table-with-dv-small.expected

@@ -0,0 +1,16 @@
+Reading table at ../../../../kernel/tests/data/table-with-dv-small/
+version: 1
+
+Schema:
+└─ value: integer
+
+value:  [
+  1,
+  2,
+  3,
+  4,
+  5,
+  6,
+  7,
+  8
+]