feat(duckdb): add distributed scan fragments

vortex-duckdb/cpp/include/table_function.hpp

@@ -64,6 +64,18 @@ struct VortexBindData final : FunctionData {
         }
     };
 
+    struct DistributedFragment {
+        idx_t file_index;
+        idx_t row_start;
+        idx_t row_end;
+        idx_t estimated_bytes;
+
+        bool operator==(const DistributedFragment &other) const {
+            return file_index == other.file_index && row_start == other.row_start &&
+                   row_end == other.row_end && estimated_bytes == other.estimated_bytes;
+        }
+    };
+
     struct PortableSnapshot {
         string portable_bind;
         vector<DistributedFile> distributed_files;
@@ -78,7 +90,7 @@ struct VortexBindData final : FunctionData {
     bool explicit_split_mode = false;
     bool splits_applied = false;
     vector<idx_t> eligible_file_indexes;
-    vector<idx_t> assigned_file_indexes;
+    vector<DistributedFragment> assigned_fragments;
 #endif
 };
 

vortex-duckdb/include/vortex.h

@@ -13,6 +13,15 @@
 
 #define COUNT_STAR_PROJ_IDX UINT64_MAX
 
+#if defined(VORTEX_VANE_DISTRIBUTED)
+typedef struct {
+  uint64_t file_index;
+  uint64_t row_start;
+  uint64_t row_end;
+  uint64_t estimated_bytes;
+} VortexDistributedFragmentView;
+#endif
+
 #if defined(VORTEX_VANE_DISTRIBUTED)
 typedef struct {
   const uint8_t *name;
@@ -119,6 +128,27 @@ const uint8_t *duckdb_table_function_distributed_bind_bytes(const void *portable
                                                             size_t *size_out);
 #endif
 
+#if defined(VORTEX_VANE_DISTRIBUTED)
+extern
+duckdb_vx_data duckdb_table_function_distributed_plan_fragments(const uint8_t *portable_bind,
+                                                                size_t portable_bind_size,
+                                                                const uint64_t *selected_file_indexes,
+                                                                size_t selected_file_count,
+                                                                size_t target_fragment_count,
+                                                                duckdb_vx_error *error_out);
+#endif
+
+#if defined(VORTEX_VANE_DISTRIBUTED)
+extern size_t duckdb_table_function_distributed_fragment_count(const void *fragment_plan);
+#endif
+
+#if defined(VORTEX_VANE_DISTRIBUTED)
+extern
+bool duckdb_table_function_distributed_fragment_at(const void *fragment_plan,
+                                                   size_t index,
+                                                   VortexDistributedFragmentView *fragment_out);
+#endif
+
 #if defined(VORTEX_VANE_DISTRIBUTED)
 extern size_t duckdb_table_function_distributed_file_count(const void *portable_bind);
 #endif
@@ -158,8 +188,8 @@ bool duckdb_table_function_distributed_file_is_selected(duckdb_vx_table_filter_s
 extern
 duckdb_vx_data duckdb_table_function_init_global_distributed(const uint8_t *portable_bind,
                                                              size_t portable_bind_size,
-                                                             const uint64_t *assigned_file_indexes,
-                                                             size_t assigned_file_count,
+                                                             const VortexDistributedFragmentView *assigned_fragments,
+                                                             size_t assigned_fragment_count,
                                                              bool ignore_optional_filters,
                                                              const duckdb_vx_tfunc_init_input *init_input,
                                                              duckdb_vx_error *error_out);