Реализация чтения бинарных данных

Исправлена ошибка повторения последнего  блока в списке, возвращаемого командой `BlockReader.read_all_blocks`.

Добавлено поле `BlockBinary.raw_data`, в котором хранится бинарный код С++ структуры `BlockBinary` (обычно размером в 6344 байта).

К инициализации класса `BlockReader` добавлен именованный параметр `seek`, по умолчанию нулевой, позволяющий читать бинарный файл не с начала, а с нужной позиции.

setup.py

@@ -6,7 +6,7 @@ from pybind11 import get_cmake_dir
 from pybind11.setup_helpers import Pybind11Extension, build_ext
 from setuptools import setup
 
-__version__ = "0.1.2"
+__version__ = "0.1.3"
 
 ext_modules = [
     Pybind11Extension(

src/main.cpp

@@ -3,9 +3,14 @@
 #include <pybind11/stl.h>
 #include <vector>
 #include <fstream>
+#include <iostream>
 #include <ctime>
 #include <cmath>
 #include "stormpci.h"
+#include "string.h"
+
+namespace py = pybind11;
+
 
 #define STRINGIFY(x) #x
 #define MACRO_STRINGIFY(x) STRINGIFY(x)
@@ -43,10 +48,12 @@ double distance(double lat1, double lon1, double lat2, double lon2)
 class BlockReader
 {
 public:
-    BlockReader(const std::string &fn){
+    BlockReader(const std::string &fn, long int seek_position){
         f = std::ifstream(fn, std::fstream::binary);
+        pos = seek_position;
     }
     BlockBinary& read_block(){
+        f.seekg(pos);
         f.read(reinterpret_cast<char*>(&_single_block), sizeof(BlockBinary));
         if (not f){
             throw;
@@ -56,13 +63,16 @@ public:
         return _single_block;
     }
     std::vector<BlockBinary>& read_all_blocks(){
+        f.seekg(pos);
         bool read_successfully = true;
         while (read_successfully){
             f.read(reinterpret_cast<char*>(&_single_block), sizeof(BlockBinary));
-            if (not f){
+            if (f) {
+                _blocks_vector.push_back(_single_block);
+                pos += sizeof(BlockBinary);
+            } else {
                 read_successfully = false;
             }
-            _blocks_vector.push_back(_single_block);
         }
         return _blocks_vector;
     }
@@ -76,7 +86,6 @@ private:
     std::vector<BlockBinary> _blocks_vector;
 };
 
-namespace py = pybind11;
 
 PYBIND11_MODULE(stormpci_bin_reader, m) {
     m.doc() = R"pbdoc(
@@ -92,7 +101,7 @@ PYBIND11_MODULE(stormpci_bin_reader, m) {
 
 
     py::class_<BlockReader>(m, "BlockReader")
-        .def(py::init<const std::string &>())
+        .def(py::init<const std::string &, long int>(), py::arg("fn"), py::arg("seek")=int(0))
         .def("read_block", &BlockReader::read_block)
         .def("read_all_blocks", &BlockReader::read_all_blocks);
 
@@ -100,6 +109,9 @@ PYBIND11_MODULE(stormpci_bin_reader, m) {
         .def_readonly("strike", &BlockBinary::strike)
         .def_readonly("device_id", &BlockBinary::device_id)
         .def_readonly("experiment_id", &BlockBinary::experiment_id)
+        .def_property_readonly("raw_data", [](const BlockBinary bb) {
+            return py::bytes(reinterpret_cast<const char*>(&bb), sizeof(bb));
+        })
         .def("__repr__",
             [](const BlockBinary &bb) {
                 std::stringstream stream;

tests/test.py

@@ -1,7 +1,18 @@
 from stormpci_bin_reader import BlockReader
 
-br = BlockReader("2022-04-23-00:00:37.bin")
+br = BlockReader("2022-04-23-00:00:37.bin", seek=0)
 blocks = br.read_all_blocks()
 
-for block in blocks:
+# print(len(blocks))
+
+for j, block in enumerate(blocks):
+    if j == 0:
+        # print(len(block.raw_data))
+        with open("test.bin", "wb") as f:
+            f.write(block.raw_data)
     print(block)
+
+# for tests run in bash:
+#   head -c 6344 2022-04-23-00\:00\:37.bin > test2.bin
+# and
+#   diff test2.bin test.bin