Merge pull request #41 from slsdetectorgroup/numpy-refactor

Numpy refactor
This commit is contained in:
Bechir Braham 2024-04-02 15:25:18 +02:00 committed by GitHub
commit f6e2d358f9
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
7 changed files with 126 additions and 174 deletions

View File

@ -1,56 +0,0 @@
#pragma once
#include "aare/Frame.hpp"
#include <filesystem>
namespace aare {
class FileInterface {
public:
// options:
// - r reading
// - w writing (overwrites existing file)
// - a appending (appends to existing file)
// TODO! do we need to support w+, r+ and a+?
FileInterface(std::filesystem::path& fname, const char* opts="r" ){};
// write one frame
virtual void write(Frame& frame) = 0;
// write n_frames frames
virtual void write(std::vector<Frame>& frames) = 0;
// read one frame
virtual Frame read() = 0;
// read n_frames frames
virtual std::vector<Frame> read(size_t n_frames) = 0; //Is this the right interface?
// read one frame into the provided buffer
virtual void read_into(std::byte* image_buf) = 0;
// read n_frames frame into the provided buffer
virtual void read_into(std::byte* image_buf, size_t n_frames) = 0;
// read the frame number on position frame_index
virtual size_t frame_number(size_t frame_index) = 0;
// size of one frame, important fro teh read_into function
virtual size_t bytes_per_frame() const = 0;
// goto frame number
virtual void seek(size_t frame_number) = 0;
// return the position of the file pointer (in number of frames)
virtual size_t tell() = 0;
// total number of frames in the file
virtual size_t total_frames() = 0;
virtual size_t rows() const = 0;
virtual size_t cols() const = 0;
// function to query the data type of the file
/*virtual DataType dtype = 0; */
virtual ~FileInterface() = 0;
};
} // namespace aare

View File

@ -83,7 +83,7 @@ class FileInterface {
public:
std::string mode;
std::filesystem::path m_fname;
// std::filesystem::path m_fname;
std::filesystem::path m_base_path;
std::string m_base_name, m_ext;
int m_findex;

View File

@ -4,15 +4,31 @@
#include "aare/defs.hpp"
#include <iostream>
#include <numeric>
#include <filesystem>
namespace aare{
class NumpyFile : public FileInterface {
FILE *fp = nullptr;
size_t initial_header_len = 0;
size_t current_frame{};
std::filesystem::path m_fname;
uint32_t header_len{};
uint8_t header_len_size{};
ssize_t header_size{};
header_t m_header;
uint8_t major_ver_{};
uint8_t minor_ver_{};
void load_metadata();
void get_frame_into(size_t, std::byte *);
Frame get_frame(size_t frame_number);
public:
NumpyFile(const std::filesystem::path& fname);
NumpyFile(FileConfig, header_t);
void write(Frame &frame) override;
Frame read() override { return get_frame(this->current_frame++); }
@ -24,26 +40,12 @@ class NumpyFile : public FileInterface {
size_t pixels() override;
void seek(size_t frame_number) override { this->current_frame = frame_number; }
size_t tell() override { return this->current_frame; }
size_t total_frames() const override { return header.shape[0]; }
ssize_t rows() const override { return header.shape[1]; }
ssize_t cols() const override { return header.shape[2]; }
ssize_t bitdepth() const override { return header.dtype.bitdepth(); }
NumpyFile(std::filesystem::path fname);
NumpyFile(FileConfig, header_t);
header_t header;
uint8_t major_ver_{};
uint8_t minor_ver_{};
uint32_t header_len{};
uint8_t header_len_size{};
ssize_t header_size{};
size_t total_frames() const override { return m_header.shape[0]; }
ssize_t rows() const override { return m_header.shape[1]; }
ssize_t cols() const override { return m_header.shape[2]; }
ssize_t bitdepth() const override { return m_header.dtype.bitdepth(); }
~NumpyFile();
private:
size_t current_frame{};
void get_frame_into(size_t, std::byte *);
Frame get_frame(size_t frame_number);
};
} // namespace aare

View File

@ -13,7 +13,7 @@ class NumpyFileFactory : public FileFactory {
public:
NumpyFileFactory(std::filesystem::path fpath);
void parse_metadata(FileInterface *_file) override;
void parse_metadata(FileInterface *_file) override{/*TODO! remove after refactor*/};
NumpyFile* load_file_read() override;
NumpyFile* load_file_write(FileConfig) override;
void parse_fname(FileInterface*)override{};

View File

@ -9,8 +9,10 @@ namespace aare {
class RawFile : public FileInterface {
using config = RawFileConfig;
public:
std::filesystem::path m_fname; //TO be made private!
void write(Frame &frame) override{};
Frame read() override { return get_frame(this->current_frame++); };
std::vector<Frame> read(size_t n_frames) override;

View File

@ -3,6 +3,33 @@
namespace aare{
NumpyFile::NumpyFile(const std::filesystem::path& fname) {
//TODO! add opts to constructor
m_fname = fname;
fp = fopen(m_fname.c_str(), "rb");
if (!fp) {
throw std::runtime_error(fmt::format("Could not open: {} for reading", m_fname.c_str()));
}
load_metadata();
}
NumpyFile::NumpyFile(FileConfig config, header_t header) {
mode = "w";
m_fname = config.fname;
m_bitdepth = config.dtype.bitdepth();
m_rows = config.rows;
m_cols = config.cols;
m_header = header;
m_header.shape = {0, config.rows, config.cols};
fp = fopen(m_fname.c_str(), "wb");
if (!fp) {
throw std::runtime_error(fmt::format("Could not open: {} for reading", m_fname.c_str()));
}
initial_header_len =
aare::NumpyHelpers::write_header(std::filesystem::path(m_fname.c_str()), header);
}
void NumpyFile::write(Frame &frame) {
if (fp == nullptr) {
throw std::runtime_error("File not open");
@ -14,30 +41,10 @@ void NumpyFile::write(Frame &frame) {
fwrite(frame._get_data(), frame.size(), 1, fp);
}
NumpyFile::NumpyFile(std::filesystem::path fname_) {
this->m_fname = fname_;
fp = fopen(this->m_fname.c_str(), "rb");
}
NumpyFile::NumpyFile(FileConfig config, header_t header) {
this->mode = "w";
this->m_fname = config.fname;
this->m_bitdepth = config.dtype.bitdepth();
this->m_rows = config.rows;
this->m_cols = config.cols;
this->header = header;
this->header.shape = {0, config.rows, config.cols};
fp = fopen(this->m_fname.c_str(), "wb");
if (!fp) {
throw std::runtime_error(fmt::format("Could not open: {} for reading", this->m_fname.c_str()));
}
this->initial_header_len =
aare::NumpyHelpers::write_header(std::filesystem::path(this->m_fname.c_str()), this->header);
}
Frame NumpyFile::get_frame(size_t frame_number) {
Frame frame(header.shape[1], header.shape[2], header.dtype.bitdepth());
Frame frame(m_header.shape[1], m_header.shape[2], m_header.dtype.bitdepth());
get_frame_into(frame_number, frame._get_data());
return frame;
}
@ -45,7 +52,7 @@ void NumpyFile::get_frame_into(size_t frame_number, std::byte *image_buf) {
if (fp == nullptr) {
throw std::runtime_error("File not open");
}
if (frame_number > header.shape[0]) {
if (frame_number > m_header.shape[0]) {
throw std::runtime_error("Frame number out of range");
}
fseek(fp, header_size + frame_number * bytes_per_frame(), SEEK_SET);
@ -53,24 +60,24 @@ void NumpyFile::get_frame_into(size_t frame_number, std::byte *image_buf) {
}
size_t NumpyFile::pixels() {
return std::accumulate(header.shape.begin() + 1, header.shape.end(), 1, std::multiplies<uint64_t>());
return std::accumulate(m_header.shape.begin() + 1, m_header.shape.end(), 1, std::multiplies<uint64_t>());
};
size_t NumpyFile::bytes_per_frame() { return header.dtype.bitdepth() / 8 * pixels(); };
size_t NumpyFile::bytes_per_frame() { return m_header.dtype.bitdepth() / 8 * pixels(); };
std::vector<Frame> NumpyFile::read(size_t n_frames) {
// TODO: implement this in a more efficient way
std::vector<Frame> frames;
for (size_t i = 0; i < n_frames; i++) {
frames.push_back(this->get_frame(this->current_frame));
this->current_frame++;
frames.push_back(get_frame(current_frame));
current_frame++;
}
return frames;
}
void NumpyFile::read_into(std::byte *image_buf, size_t n_frames) {
// TODO: implement this in a more efficient way
for (size_t i = 0; i < n_frames; i++) {
this->get_frame_into(this->current_frame++, image_buf);
image_buf += this->bytes_per_frame();
get_frame_into(current_frame++, image_buf);
image_buf += bytes_per_frame();
}
}
@ -82,11 +89,11 @@ NumpyFile::~NumpyFile() {
size_t data_size = file_size - initial_header_len;
size_t n_frames = data_size / bytes_per_frame();
// update number of frames in header (first element of shape)
this->header.shape[0] = n_frames;
m_header.shape[0] = n_frames;
fseek(fp, 0, SEEK_SET);
// create string stream to contain header
std::stringstream ss;
aare::NumpyHelpers::write_header(ss, this->header);
aare::NumpyHelpers::write_header(ss, m_header);
std::string header_str = ss.str();
// write header
fwrite(header_str.c_str(), header_str.size(), 1, fp);
@ -100,4 +107,69 @@ NumpyFile::~NumpyFile() {
}
}
void NumpyFile::load_metadata(){
// read magic number
std::array<char, 6> tmp{};
fread(tmp.data(), tmp.size(), 1, fp);
if (tmp != aare::NumpyHelpers::magic_str) {
for (auto item : tmp)
fmt::print("{}, ", int(item));
fmt::print("\n");
throw std::runtime_error("Not a numpy file");
}
// read version
fread(reinterpret_cast<char *>(&major_ver_), sizeof(major_ver_), 1,fp);
fread(reinterpret_cast<char *>(&minor_ver_), sizeof(minor_ver_), 1,fp);
if (major_ver_ == 1) {
header_len_size = 2;
} else if (major_ver_ == 2) {
header_len_size = 4;
} else {
throw std::runtime_error("Unsupported numpy version");
}
// read header length
fread(reinterpret_cast<char *>(&header_len), header_len_size,1, fp);
header_size = aare::NumpyHelpers::magic_string_length + 2 + header_len_size + header_len;
if (header_size % 16 != 0) {
fmt::print("Warning: header length is not a multiple of 16\n");
}
// read header
auto buf_v = std::vector<char>(header_len);
fread(buf_v.data(), header_len,1,fp);
std::string header(buf_v.data(), header_len);
// parse header
std::vector<std::string> keys{"descr", "fortran_order", "shape"};
aare::logger::debug("original header: \"header\"");
auto dict_map = aare::NumpyHelpers::parse_dict(header, keys);
if (dict_map.size() == 0)
throw std::runtime_error("invalid dictionary in header");
std::string descr_s = dict_map["descr"];
std::string fortran_s = dict_map["fortran_order"];
std::string shape_s = dict_map["shape"];
std::string descr = aare::NumpyHelpers::parse_str(descr_s);
aare::DType dtype = aare::NumpyHelpers::parse_descr(descr);
// convert literal Python bool to C++ bool
bool fortran_order = aare::NumpyHelpers::parse_bool(fortran_s);
// parse the shape tuple
auto shape_v = aare::NumpyHelpers::parse_tuple(shape_s);
shape_t shape;
for (auto item : shape_v) {
auto dim = static_cast<unsigned long>(std::stoul(item));
shape.push_back(dim);
}
m_header = {dtype, fortran_order, shape};
}
} // namespace aare

View File

@ -4,78 +4,10 @@
namespace aare {
NumpyFileFactory::NumpyFileFactory(std::filesystem::path fpath) { this->m_fpath = fpath; }
void NumpyFileFactory::parse_metadata(FileInterface *_file) {
auto file = dynamic_cast<NumpyFile *>(_file);
// open ifsteam to file
f = std::ifstream(file->m_fname, std::ios::binary);
// check if file exists
if (!f.is_open()) {
throw std::runtime_error(fmt::format("Could not open: {} for reading", file->m_fname.c_str()));
}
// read magic number
std::array<char, 6> tmp{};
f.read(tmp.data(), tmp.size());
if (tmp != aare::NumpyHelpers::magic_str) {
for (auto item : tmp)
fmt::print("{}, ", int(item));
fmt::print("\n");
throw std::runtime_error("Not a numpy file");
}
// read version
f.read(reinterpret_cast<char *>(&file->major_ver_), 1);
f.read(reinterpret_cast<char *>(&file->minor_ver_), 1);
if (file->major_ver_ == 1) {
file->header_len_size = 2;
} else if (file->major_ver_ == 2) {
file->header_len_size = 4;
} else {
throw std::runtime_error("Unsupported numpy version");
}
// read header length
f.read(reinterpret_cast<char *>(&file->header_len), file->header_len_size);
file->header_size = aare::NumpyHelpers::magic_string_length + 2 + file->header_len_size + file->header_len;
if (file->header_size % 16 != 0) {
fmt::print("Warning: header length is not a multiple of 16\n");
}
// read header
auto buf_v = std::vector<char>(file->header_len);
f.read(buf_v.data(), file->header_len);
std::string header(buf_v.data(), file->header_len);
// parse header
std::vector<std::string> keys{"descr", "fortran_order", "shape"};
aare::logger::debug("original header: \"header\"");
auto dict_map = aare::NumpyHelpers::parse_dict(header, keys);
if (dict_map.size() == 0)
throw std::runtime_error("invalid dictionary in header");
std::string descr_s = dict_map["descr"];
std::string fortran_s = dict_map["fortran_order"];
std::string shape_s = dict_map["shape"];
std::string descr = aare::NumpyHelpers::parse_str(descr_s);
aare::DType dtype = aare::NumpyHelpers::parse_descr(descr);
// convert literal Python bool to C++ bool
bool fortran_order = aare::NumpyHelpers::parse_bool(fortran_s);
// parse the shape tuple
auto shape_v = aare::NumpyHelpers::parse_tuple(shape_s);
shape_t shape;
for (auto item : shape_v) {
auto dim = static_cast<unsigned long>(std::stoul(item));
shape.push_back(dim);
}
file->header = {dtype, fortran_order, shape};
}
NumpyFile *NumpyFileFactory::load_file_read() {
NumpyFile *file = new NumpyFile(this->m_fpath);
parse_metadata(file);
return file;
};