make ffi work, add default tight_bounds

ffi/cffi-test.c

@@ -7,9 +7,15 @@
 void visit_callback(void* engine_context,
                     KernelStringSlice path,
                     int64_t size,
+                    const Stats* stats,
                     const DvInfo* dv_info,
                     const CStringMap* partition_values) {
-  printf("file: %.*s\n", (int)path.len, path.ptr);
+  printf("file: %.*s (size: %" PRId64 ", num_records:", (int)path.len, path.ptr, size);
+  if (stats) {
+    printf("%" PRId64 ")\n", stats->num_records);
+  } else {
+    printf(" [no stats])\n");
+  }
 }
 
 void visit_data(void* engine_context,

ffi/src/scan.rs

@@ -271,10 +271,17 @@ pub unsafe extern "C" fn kernel_scan_data_free(data: Handle<SharedScanDataIterat
     data.drop_handle();
 }
 
+#[repr(C)]
+pub struct Stats {
+    num_records: u64,
+    tight_bounds: bool,
+}
+
 type CScanCallback = extern "C" fn(
     engine_context: NullableCvoid,
     path: KernelStringSlice,
     size: i64,
+    stats: Option<&Stats>,
     dv_info: &DvInfo,
     partition_map: &CStringMap,
 );
@@ -336,16 +343,22 @@ fn rust_callback(
     context: &mut ContextWrapper,
     path: &str,
     size: i64,
+    kernel_stats: Option<delta_kernel::scan::state::Stats>,
     dv_info: DvInfo,
     partition_values: HashMap<String, String>,
 ) {
     let partition_map = CStringMap {
         values: partition_values,
     };
+    let stats = kernel_stats.map(|ks| Stats {
+        num_records: ks.num_records,
+        tight_bounds: ks.tight_bounds,
+    });
     (context.callback)(
         context.engine_context,
         path.into(),
         size,
+        stats.as_ref(),
         &dv_info,
         &partition_map,
     );

kernel/src/scan/state.rs

@@ -33,11 +33,16 @@ pub struct DvInfo {
     deletion_vector: Option<DeletionVectorDescriptor>,
 }
 
+fn default_false() -> bool {
+    false
+}
+
 /// Give engines an easy way to consume stats
 #[derive(Deserialize)]
 #[serde(rename_all = "camelCase")]
 pub struct Stats {
     pub num_records: u64,
+    #[serde(default = "default_false")]
     pub tight_bounds: bool,
 }