utils_public.hpp

Go to the documentation of this file.

#ifndef TAKANE_UTILS_PUBLIC_HPP
#define TAKANE_UTILS_PUBLIC_HPP
 
#include <string>
#include <filesystem>
#include <unordered_map>
#include <functional>
 
#include "H5Cpp.h"
 
#include "millijson/millijson.hpp"
#include "byteme/byteme.hpp"
#include "chihaya/chihaya.hpp"
#include "utils_json.hpp"
 
namespace takane {
 
struct ObjectMetadata {
    std::string type;
    
    std::unordered_map<std::string, std::shared_ptr<millijson::Base> > other;
};
 
inline ObjectMetadata reformat_object_metadata(millijson::Base* raw) {
    if (raw->type() != millijson::OBJECT) {
        throw std::runtime_error("metadata should be a JSON object");
    }
 
    ObjectMetadata output;
    output.other = std::move(reinterpret_cast<millijson::Object*>(raw)->values);
 
    auto tIt = output.other.find("type");
    if (tIt == output.other.end()) {
        throw std::runtime_error("metadata should have a 'type' property");
    }
 
    const auto& tval = tIt->second;
    if (tval->type() != millijson::STRING) {
        throw std::runtime_error("metadata should have a 'type' string");
    }
 
    output.type = std::move(reinterpret_cast<millijson::String*>(tval.get())->value);
    output.other.erase(tIt);
    return output;
}
 
inline ObjectMetadata read_object_metadata(const std::filesystem::path& path) try {
    std::shared_ptr<millijson::Base> obj = internal_json::parse_file(path / "OBJECT");
    return reformat_object_metadata(obj.get());
} catch (std::exception& e) {
    throw std::runtime_error("failed to read the OBJECT file at '" + path.string() + "'; " + std::string(e.what()));
}
 
struct Options {
    bool parallel_reads = true;
    
    hsize_t hdf5_buffer_size = 10000;
 
public:
    std::unordered_map<std::string, std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&)> > custom_validate;
 
   std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&)> custom_global_validate;
 
public:
    std::unordered_map<std::string, std::function<std::vector<size_t>(const std::filesystem::path&, const ObjectMetadata&, Options&)> > custom_dimensions;
 
    std::unordered_map<std::string, std::function<size_t(const std::filesystem::path&, const ObjectMetadata& m, Options&)> > custom_height;
 
public:
    std::unordered_map<std::string, std::unordered_set<std::string> > custom_derived_from;
 
    std::unordered_map<std::string, std::unordered_set<std::string> > custom_satisfies_interface;
 
public:
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&)> bam_file_strict_check;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&)> bcf_file_strict_check;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&, bool)> bed_file_strict_check;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&)> bigbed_file_strict_check;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&)> bigwig_file_strict_check;
 
    std::function<bool(const std::filesystem::path&, const ObjectMetadata&, Options& options)> data_frame_factor_any_duplicated;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&, bool)> fasta_file_strict_check;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&, bool)> fastq_file_strict_check;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&, bool)> gff_file_strict_check;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&)> gmt_file_strict_check;
 
    std::function<void(const std::filesystem::path&, const ObjectMetadata&, Options&)> rds_file_strict_check;
 
    chihaya::Options delayed_array_options;
};
 
}
 
#endif