fasta_file.hpp

Go to the documentation of this file.

#ifndef TAKANE_FASTA_FILE_HPP
#define TAKANE_FASTA_FILE_HPP
 
#include "utils_files.hpp"
#include "ritsuko/ritsuko.hpp"
#include "byteme/byteme.hpp"
 
#include <filesystem>
#include <stdexcept>
#include <string>
 
namespace takane {
 
namespace fasta_file {
 
inline void validate(const std::filesystem::path& path, const ObjectMetadata& metadata, Options& options) {
    const std::string type_name = "fasta_file"; // use a separate variable to avoid dangling reference warnings from GCC.
    const auto& famap = internal_json::extract_typed_object_from_metadata(metadata.other, type_name);
 
    const std::string version_name = "version"; // again, avoid dangling reference warnings.
    const std::string& vstring = internal_json::extract_string_from_typed_object(famap, version_name, type_name);
    auto version = ritsuko::parse_version_string(vstring.c_str(), vstring.size(), /* skip_patch = */ true);
    if (version.major != 1) {
        throw std::runtime_error("unsupported version string '" + vstring + "'");
    }
 
    internal_files::check_sequence_type(famap, type_name.c_str());
 
    // Check if it's indexed.
    bool indexed = internal_files::is_indexed(famap);
    auto fpath = path / "file.fasta.";
    if (indexed) {
        fpath += "bgz";
    } else {
        fpath += "gz";
    }
 
    internal_files::check_gzip_signature(fpath);
    {
        auto reader = internal_other::open_reader<byteme::GzipFileReader>(fpath, byteme::GzipFileReaderOptions());
        char first_char;
        if (reader->read(reinterpret_cast<unsigned char*>(&first_char), 1) == 0 || first_char != '>') {
            throw std::runtime_error("FASTA file does not start with '>'");
        }
    }
 
    if (indexed) {
        auto fixpath = path / "file.fasta.fai";
        if (!std::filesystem::exists(fixpath)) {
            throw std::runtime_error("missing FASTA index file");
        }
 
        auto ixpath = fpath;
        ixpath += ".gzi";
        if (!std::filesystem::exists(ixpath)) {
            throw std::runtime_error("missing BGZF index file");
        }
    }
 
    if (options.fasta_file_strict_check) {
        options.fasta_file_strict_check(path, metadata, options, indexed);
    }
}
 
}
 
}
 
#endif