Function collect_files

common/hf-cache.cpp:614–660

}

static bool collect_files(const fs::path    & old_cache,
                          const std::string & owner,
                          const std::string & repo,
                          const nl::json    & node,
                          const hf_files    & files,
                          migrate_files     & to_migrate) {

    if (!node.contains("rfilename") ||
        !node.contains("lfs")       ||
        !node["lfs"].contains("sha256")) {
        return true;
    }

    std::string path = node["rfilename"];
    std::string sha256 = node["lfs"]["sha256"];

    auto split = get_gguf_split_info(path);

    if (split.count <= 1) {
        return collect_file(old_cache, owner, repo, path, sha256, files, to_migrate);
    }

    std::vector<std::pair<std::string, std::string>> splits;

    for (const auto & f : files) {
        auto split_f = get_gguf_split_info(f.path);
        if (split_f.count == split.count && split_f.prefix == split.prefix) {
            // sadly the manifest only provides the sha256 of the first file (index == 1)
            // the rest will be verified using the size...
            std::string f_sha256 = (split_f.index == 1) ? sha256 : "";
            splits.emplace_back(f.path, f_sha256);
        }
    }

    if ((int)splits.size() != split.count) {
        LOG_WRN("%s: expected %d split files but found %d in repo\n", __func__, split.count, (int)splits.size());
        return false;
    }

    for (const auto & [f_path, f_sha256] : splits) {
        if (!collect_file(old_cache, owner, repo, f_path, f_sha256, files, to_migrate)) {
            return false;
        }
    }

    return true;
}

static bool migrate_file(const migrate_file & file) {
    std::error_code ec;