From 7a913522ac398bc4e477a4102b465ae678dd63d7 Mon Sep 17 00:00:00 2001
From: Ashot Vardanian <ashvardanian@gmail.com>
Date: Sun, 16 Oct 2022 21:41:02 +0400
Subject: [PATCH] Add: Placeholder for bulk import/export tools

---
 include/ukv/db.h  |  3 ++
 tools/dataset.cpp |  1 +
 tools/dataset.h   | 99 +++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 103 insertions(+)
 create mode 100644 tools/dataset.cpp
 create mode 100644 tools/dataset.h

diff --git a/include/ukv/db.h b/include/ukv/db.h
index 3f81d58b8..0a5088b74 100644
--- a/include/ukv/db.h
+++ b/include/ukv/db.h
@@ -106,6 +106,9 @@ typedef void* ukv_arena_t;
 typedef uint8_t* ukv_bytes_ptr_t;
 typedef uint8_t const* ukv_bytes_cptr_t;
 
+typedef void* ukv_callback_payload_t;
+typedef void (*ukv_callback_t)(ukv_callback_payload_t);
+
 typedef enum {
 
     ukv_options_default_k = 0,
diff --git a/tools/dataset.cpp b/tools/dataset.cpp
new file mode 100644
index 000000000..8b1378917
--- /dev/null
+++ b/tools/dataset.cpp
@@ -0,0 +1 @@
+
diff --git a/tools/dataset.h b/tools/dataset.h
new file mode 100644
index 000000000..7223fe270
--- /dev/null
+++ b/tools/dataset.h
@@ -0,0 +1,99 @@
+
+#pragma once
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <ukv/db.h>
+
+typedef struct ukv_graph_import_t {
+
+    ukv_database_t db;
+    ukv_error_t* error;
+    ukv_arena_t* arena = NULL;
+    ukv_options_t options = ukv_options_default_k;
+
+    ukv_collection_t collection = ukv_collection_main_k;
+    ukv_str_view_t paths_pattern = ".*\.(csv|ndjson|parquet)";
+    ukv_size_t max_batch_size = 1024ul * 1024ul * 1024ul;
+    ukv_callback_t callback = NULL;
+    ukv_callback_payload_t callback_payload = NULL;
+
+    ukv_str_view_t source_id_field = "source";
+    ukv_str_view_t target_id_field = "target";
+    ukv_str_view_t edge_id_field = "edge";
+
+} ukv_graph_import_t;
+
+void ukv_graph_import(ukv_graph_import_t*);
+
+typedef struct ukv_graph_export_t {
+
+    ukv_database_t db;
+    ukv_error_t* error;
+    ukv_arena_t* arena = NULL;
+    ukv_options_t options = ukv_options_default_k;
+
+    ukv_collection_t collection = ukv_collection_main_k;
+    ukv_str_view_t paths_extension = ".parquet";
+    ukv_size_t max_batch_size = 1024ul * 1024ul * 1024ul;
+    ukv_callback_t callback = NULL;
+    ukv_callback_payload_t callback_payload = NULL;
+
+    ukv_str_view_t source_id_field = "source";
+    ukv_str_view_t target_id_field = "target";
+    ukv_str_view_t edge_id_field = "edge";
+
+} ukv_graph_export_t;
+
+void ukv_graph_export(ukv_graph_export_t*);
+
+typedef struct ukv_docs_import_t {
+
+    ukv_database_t db;
+    ukv_error_t* error;
+    ukv_arena_t* arena = NULL;
+    ukv_options_t options = ukv_options_default_k;
+
+    ukv_collection_t collection = ukv_collection_main_k;
+    ukv_str_view_t paths_pattern = ".*\.(csv|ndjson|parquet)";
+    ukv_size_t max_batch_size = 1024ul * 1024ul * 1024ul;
+    ukv_callback_t callback = NULL;
+    ukv_callback_payload_t callback_payload = NULL;
+
+    ukv_size_t fields_count = 0;
+    ukv_str_view_t const* fields = NULL;
+    ukv_size_t fields_stride = 0;
+
+    ukv_str_view_t id_field = "_id";
+
+} ukv_docs_import_t;
+
+void ukv_docs_import(ukv_docs_import_t*);
+
+typedef struct ukv_docs_export_t {
+
+    ukv_database_t db;
+    ukv_error_t* error;
+    ukv_arena_t* arena = NULL;
+    ukv_options_t options = ukv_options_default_k;
+
+    ukv_collection_t collection = ukv_collection_main_k;
+    ukv_str_view_t paths_extension = ".parquet";
+    ukv_size_t max_batch_size = 1024ul * 1024ul * 1024ul;
+    ukv_callback_t callback = NULL;
+    ukv_callback_payload_t callback_payload = NULL;
+
+    ukv_size_t fields_count = 0;
+    ukv_str_view_t const* fields = NULL;
+    ukv_size_t fields_stride = 0;
+
+} ukv_docs_export_t;
+
+void ukv_docs_export(ukv_docs_export_t*);
+
+#ifdef __cplusplus
+} /* end extern "C" */
+#endif
+