lix/lix/libutil/tarfile.cc

#include <archive.h>
#include <archive_entry.h>
#include <kj/async.h>

#include "async-io.hh"
#include "file-descriptor.hh"
#include "lix/libutil/c-calls.hh"
#include "lix/libutil/charptr-cast.hh"
#include "lix/libutil/file-system.hh"
#include "lix/libutil/logging.hh"
#include "lix/libutil/serialise.hh"
#include "result.hh"
#include "lix/libutil/tarfile.hh"

namespace nix {

static int callback_open(struct archive *, void * self)
{
    return ARCHIVE_OK;
}

static ssize_t callback_read(struct archive * archive, void * _self, const void ** buffer)
{
    auto self = static_cast<TarArchive *>(_self);
    *buffer = self->buffer.data();

    try {
        return self->source->read(charptr_cast<char *>(self->buffer.data()), self->buffer.size());
    } catch (EndOfFile &) {
        return 0;
    } catch (std::exception & err) { // NOLINT(lix-foreign-exceptions)
        // NOLINTNEXTLINE(lix-unsafe-c-calls): what() is a c string
        archive_set_error(archive, EIO, "Source threw exception: %s", err.what());
        return -1;
    }
}

static int callback_close(struct archive *, void * self)
{
    return ARCHIVE_OK;
}

void TarArchive::check(int err, const std::string & reason)
{
    if (err == ARCHIVE_EOF) {
        throw EndOfFile("reached end of archive");
    } else if (err != ARCHIVE_OK) {
        throw ArchiveError(reason, archive_error_string(this->archive.get()));
    }
}

TarArchive::TarArchive(Source & source, bool raw)
    : archive{archive_read_new()}
    , source(&source)
    , buffer(65536)
{
    if (!raw) {
        archive_read_support_filter_all(archive.get());
        archive_read_support_format_all(archive.get());
    } else {
        archive_read_support_filter_all(archive.get());
        archive_read_support_format_raw(archive.get());
        archive_read_support_format_empty(archive.get());
    }
    archive_read_set_option(archive.get(), nullptr, "mac-ext", nullptr);
    check(
        archive_read_open(
            archive.get(), (void *) this, callback_open, callback_read, callback_close
        ),
        "Failed to open archive (%s)"
    );
}

TarArchive::TarArchive(const Path & path) : archive{archive_read_new()}
{
    archive_read_support_filter_all(archive.get());
    archive_read_support_format_all(archive.get());
    archive_read_set_option(archive.get(), nullptr, "mac-ext", nullptr);
    check(
        archive_read_open_filename(archive.get(), requireCString(path), 16384),
        "failed to open archive: %s"
    );
}

void TarArchive::close()
{
    check(archive_read_close(this->archive.get()), "Failed to close archive (%s)");
}

static void extract_archive(TarArchive & archive, const Path & destDir)
{
    requireCString(destDir);

    int flags =
        ARCHIVE_EXTRACT_TIME | ARCHIVE_EXTRACT_SECURE_SYMLINKS | ARCHIVE_EXTRACT_SECURE_NODOTDOT;

    for (;;) {
        struct archive_entry * entry;
        int r = archive_read_next_header(archive.archive.get(), &entry);
        if (r == ARCHIVE_EOF) {
            break;
        }
        auto name = archive_entry_pathname(entry);
        if (!name) {
            throw Error(
                "cannot get archive member name: %s", archive_error_string(archive.archive.get())
            );
        }
        if (r == ARCHIVE_WARN) {
            printTaggedWarning("%1%", Uncolored(archive_error_string(archive.archive.get())));
        } else {
            archive.check(r);
        }

        // NOLINTNEXTLINE(lix-unsafe-c-calls): destDir is checked, name is a c string
        archive_entry_copy_pathname(entry, (destDir + "/" + name).c_str());

        // sources can and do contain dirs with no rx bits
        if (archive_entry_filetype(entry) == AE_IFDIR && (archive_entry_mode(entry) & 0500) != 0500)
        {
            archive_entry_set_mode(entry, archive_entry_mode(entry) | 0500);
        }

        // Patch hardlink path
        const char * original_hardlink = archive_entry_hardlink(entry);
        if (original_hardlink) {
            // NOLINTNEXTLINE(lix-unsafe-c-calls): destDir is checked, name is a c string
            archive_entry_copy_hardlink(entry, (destDir + "/" + original_hardlink).c_str());
        }

        archive.check(archive_read_extract(archive.archive.get(), entry, flags));
    }

    archive.close();
}

kj::Promise<Result<void>> unpackTarfile(AsyncInputStream & source, const Path & destDir)
try {
    Pipe pipe;
    pipe.create();

    auto thr = std::async(
        std::launch::async,
        [&](AutoCloseFD fd) {
            FdSource source(fd.get());
            auto archive = TarArchive(source);

            createDirs(destDir);
            extract_archive(archive, destDir);
        },
        std::move(pipe.readSide)
    );

    AsyncFdIoStream sink{std::move(pipe.writeSide)};
    TRY_AWAIT(source.drainInto(sink));
    thr.get();
    co_return result::success();
} catch (...) {
    co_return result::current_exception();
}

void unpackTarfile(const Path & tarFile, const Path & destDir)
{
    auto archive = TarArchive(tarFile);

    createDirs(destDir);
    extract_archive(archive, destDir);
}

}