Merge pull request #20 from TeskaLabs/feature/c-api

Add a C API (initial version)

.gitignore

@@ -1,6 +1,8 @@
 .vscode
 build
 cysimdjson/cysimdjson.cpp
+cysimdjson/cysimdjson.h
+cysimdjson/cysimdjson_api.h
 cysimdjson.*.so
 __pycache__
 dist/

cysimdjson/cysimdjson.pyx

@@ -124,6 +124,8 @@ cdef extern from "jsoninter.h":
 
 	cdef const char * PyUnicode_AsUTF8AndSize(object, Py_ssize_t *)
 
+	cdef simdjson_element extract_element(void *)
+
 
 cdef class JSONObject:
 
@@ -313,6 +315,12 @@ cdef class JSONParser:
 		return get_active_implementation()
 
 
+# This method is used by C-level callers who want to wrap `simdjson::dom::element` into a cysimdjson object instance
+cdef public api object cysimdjson_wrap_element(void * element):
+	cdef simdjson_element v = extract_element(element)
+	return _wrap_element(v)
+
+
 cdef inline object _wrap_element(simdjson_element v):
 	cdef simdjson_element_type et = v.type()
 

cysimdjson/cysimdjsonc.cpp

@@ -0,0 +1,92 @@
+// We have to include "Python.h" b/c otherwise parser.parse() crashes
+// Assumption is that Python.h sets some definition that is used by `simdjson.h`
+// TODO: Find what `ifdef` is set and set that without `Python.h`
+#include "Python.h"
+
+#include "simdjson/simdjson.h"
+
+extern "C" {
+#include "cysimdjsonc.h"
+}
+
+void * cysimdjson_parser_new(void) {
+	simdjson::dom::parser * parser = new simdjson::dom::parser();
+	void * p = static_cast<void*>(parser);
+	return p;
+}
+
+void cysimdjson_parser_del(void * p) {
+	assert(p != NULL);
+	simdjson::dom::parser * parser = static_cast<simdjson::dom::parser *>(p);
+	delete parser;
+}
+
+
+size_t cysimdjson_element_sizeof(void) {
+	return sizeof(simdjson::dom::element);
+}
+
+
+bool cysimdjson_parser_parse(void * p, void * memory, const uint8_t * data, size_t datalen) {
+	simdjson::dom::parser * parser = static_cast<simdjson::dom::parser *>(p);
+
+	try {
+		// Initialize the element at the memory provided by a caller
+		// See: https://www.geeksforgeeks.org/placement-new-operator-cpp/
+		// `memory` is a pointer to a pre-allocated memory space with >= cysimdjson_element_sizeof() bytes
+		simdjson::dom::element * element = new(memory) simdjson::dom::element();
+
+		// Parse the JSON
+		auto err = parser->parse(
+			data, datalen,
+			true // Create a copy if needed (TODO: this may be optimized eventually to save data copy)
+		).get(*element);
+
+		if (err) {
+			// Likely syntax error in JSON
+			return true;
+		}
+
+	} catch (const std::bad_alloc& e) {
+		// Error when allocating memory
+		return true;
+	}
+	catch (...) {
+		return true;
+	}
+
+	// No error
+	return false;
+}
+
+
+bool cysimdjson_element_get_str(const char * attrname, size_t attrlen, void * e, const char ** output, size_t * outputlen) {
+	simdjson::dom::element * element = static_cast<simdjson::dom::element *>(e);
+	std::string_view pointer = std::string_view(attrname, attrlen);
+
+	std::string_view result;
+	auto err = element->at_pointer(pointer).get(result);
+	if (err) {
+		return true;
+	}
+
+	*output = result.data();
+	*outputlen = result.size();
+	return false;
+}
+
+
+// This is here for an unit test
+void cysimdjson_parser_test() {
+	printf("cysimdjson_parser_test started ...\n");
+
+	simdjson::dom::parser parser;
+	simdjson::dom::object object;
+
+	const char * jsond = R"({"key":"value"}     )";
+	const size_t jsond_len = std::strlen(jsond);
+
+	auto error = parser.parse(jsond, jsond_len).get(object);
+
+	printf("cysimdjson_parser_test OK!\n");
+}

cysimdjson/cysimdjsonc.h

@@ -0,0 +1,21 @@
+#ifndef CYSIMDJSONAPI_H
+#define CYSIMDJSONAPI_H
+
+// This is API for C (not C++) level
+// This header has to be C compliant
+
+#include <stddef.h>
+#include <stdint.h>
+#include <stdbool.h>
+
+void * cysimdjson_parser_new(void);
+void cysimdjson_parser_del(void * parser);
+
+size_t cysimdjson_element_sizeof(void);
+bool cysimdjson_parser_parse(void * parser, void * memory, const uint8_t * data, size_t datalen);
+
+bool cysimdjson_element_get_str(const char * attrname, size_t attrlen, void * element, const char ** output, size_t * outputlen);
+
+void cysimdjson_parser_test(void);
+
+#endif

cysimdjson/jsoninter.h

@@ -34,3 +34,9 @@ inline PyObject * element_to_py_string(dom::element & value) {
 inline std::string get_active_implementation() {
 	return simdjson::active_implementation->description();
 }
+
+
+inline dom::element extract_element(void * p) {
+	dom::element * element = static_cast<dom::element *>(p);
+	return *element;
+}

setup.py

@@ -12,10 +12,11 @@
 extensions = [
 	Extension(
 		"cysimdjson",
-		[
+		[			
 			'cysimdjson/cysimdjson.pyx',
 			'cysimdjson/simdjson/simdjson.cpp',
 			'cysimdjson/pysimdjson/errors.cpp',
+			'cysimdjson/cysimdjsonc.cpp',
 		],
 		language="c++",
 		extra_compile_args=[

test/__init__.py

@@ -1,2 +1,3 @@
 from .test_array import *
 from .test_document import *
+from .test_capi import *

test/test_capi/__init__.py

@@ -0,0 +1,27 @@
+import ctypes
+import unittest
+
+import cysimdjson
+
+
+class CySIMDJSONCAPITestCases(unittest.TestCase):
+
+
+	def test_capi_01(self):
+
+		ctypes.pythonapi.cysimdjson_parser_new.restype = ctypes.c_void_p
+		ctypes.pythonapi.cysimdjson_parser_del.argtypes = [ctypes.c_void_p]
+
+		parser = ctypes.pythonapi.cysimdjson_parser_new()
+
+		ctypes.pythonapi.cysimdjson_parser_del(parser)
+
+
+	def test_capi_02(self):
+		ctypes.pythonapi.cysimdjson_parser_new.restype = ctypes.c_int
+		element_sizeof = ctypes.pythonapi.cysimdjson_element_sizeof()
+		print("element_sizeof:", element_sizeof)
+
+
+	def test_capi_03(self):
+		ctypes.pythonapi.cysimdjson_parser_test()