coyote/models.cpp

#include <ctime>
#include <regex>
#include <stdexcept>
#include <system_error>
#include <nlohmann/json.hpp>

#include "models.h"
#include "numberhelper.h"

#define DOMAIN_RE "(?:[a-zA-Z0-9-]+\\.)+[a-zA-Z]{2,}"
// https://docs.joinmastodon.org/methods/accounts/#422-unprocessable-entity
// https://akko.erincandescent.net/@postmaster-emeritus
#define USERNAME_RE "[a-zA-Z0-9\\-_]+"

using json = nlohmann::json;
static time_t parse_rfc3339(const std::string& str);


void from_json(const json& j, Emoji& emoji) {
    j.at("shortcode").get_to(emoji.shortcode);
    j.at("url").get_to(emoji.url);
}

void from_json(const json& j, AccountField& field) {
    j.at("name").get_to(field.name);
    field.value_html = j.at("value").get<std::string>();
    // https://social.kernel.org/@monsieuricon/Ac6oYwtLhess6uil1c
    // https://social.kernel.org/@monsieuricon/Ac8RXJRqxTfUrM1xQG
    if (!j.contains("verified_at") || j["verified_at"].is_null()) {
        field.verified_at = -1;
    } else {
        field.verified_at = parse_rfc3339(j["verified_at"].get_ref<const std::string&>());
    }
}

static std::regex host_regex("https?://(" DOMAIN_RE ")/.*", std::regex::ECMAScript | std::regex::icase);
void from_json(const json& j, Account& account) {
    using namespace std::string_literals;

    j.at("id").get_to(account.id);
    j.at("username").get_to(account.username);
    j.at("display_name").get_to(account.display_name);
    j.at("bot").get_to(account.bot);
    account.created_at = parse_rfc3339(j.at("created_at").get_ref<const std::string&>());
    account.note_html = j.at("note").get<std::string>();
    j.at("avatar").get_to(account.avatar);
    j.at("avatar_static").get_to(account.avatar_static);
    j.at("header").get_to(account.header);
    j.at("followers_count").get_to(account.followers_count);
    j.at("following_count").get_to(account.following_count);
    j.at("statuses_count").get_to(account.statuses_count);
    j.at("emojis").get_to(account.emojis);
    j.at("fields").get_to(account.fields);

    std::smatch sm;
    const std::string& url = j.at("url").get_ref<const std::string&>();
    if (!std::regex_match(url, sm, host_regex)) {
        throw std::runtime_error("failed to find host in url: "s + url);
    }
    account.server = sm.str(1);

    if (account.display_name.empty()) {
        // https://mastodonapp.uk/@probertd8
        account.display_name = account.username;
    }
}

void from_json(const json& j, Size& size) {
    j.at("width").get_to(size.width);
    j.at("height").get_to(size.height);
}

void from_json(const json& j, Media& media) {
    j.at("type").get_to(media.type);
    j.at("url").get_to(media.url);
    if (!j.at("preview_url").is_null()) {
        media.preview_url = j["preview_url"].get<std::string>();
    }
    if (!j.at("remote_url").is_null()) {
        media.remote_url = j["remote_url"].get<std::string>();
    }
    // .contains() check: https://social.platypus-sandbox.com/@ashten/AcPl5zVFxeGu7eSKSe
    if (j.contains("meta") && (media.type == "image" || media.type == "video" || media.type == "gifv")) {
        // .contains() checks: https://social.platypus-sandbox.com/@ashten/AcPlHNRnHGzjZMGX56
        const json& meta = j["meta"];
        if (meta.contains("original")) {
            media.size = meta["original"].get<Size>();
        }
        if (meta.contains("small")) {
            media.preview_size = meta["small"].get<Size>();
        }
    }
    if (!j.at("description").is_null()) {
        media.description = j["description"].get<std::string>();
    }
}

void from_json(const json& j, PollOption& option) {
    j.at("title").get_to(option.title);
    j.at("votes_count").get_to(option.votes_count);
}

void from_json(const json& j, Poll& poll) {
    if (!j.at("expires_at").is_null()) {
        poll.expires_at = parse_rfc3339(j["expires_at"].get_ref<const std::string&>());
    } else {
        poll.expires_at = -1;
    }
    j.at("expired").get_to(poll.expired);
    if (j.contains("voters_count") && !j["voters_count"].is_null()) {
        j.at("voters_count").get_to(poll.voters_count);
    } else {
        poll.voters_count = -1;
    }
    j.at("votes_count").get_to(poll.votes_count);
    j.at("options").get_to(poll.options);
    j.at("emojis").get_to(poll.emojis);
}

static std::regex akkoma_status_url_regex("https?://(" DOMAIN_RE ")/(?:@|users/)(" USERNAME_RE ")/.+");
void from_json(const json& j, Post& post) {
    j.at("id").get_to(post.id);
    post.created_at = parse_rfc3339(j.at("created_at").get_ref<const std::string&>());
    if (!j.at("in_reply_to_id").is_null()) {
        post.in_reply_to_id = j["in_reply_to_id"].get<std::string>();
    }
    if (!j.at("in_reply_to_account_id").is_null()) {
        post.in_reply_to_account_id = j["in_reply_to_account_id"].get<std::string>();
    }
    j.at("sensitive").get_to(post.sensitive);
    j.at("spoiler_text").get_to(post.spoiler_text);
    j.at("replies_count").get_to(post.replies_count);
    j.at("reblogs_count").get_to(post.reblogs_count);
    j.at("favourites_count").get_to(post.favorites_count);
    // https://social.kernel.org/@monsieuricon/Ac6oYwtLhess6uil1c
    if (j.contains("edited_at") && !j["edited_at"].is_null()) {
        post.edited_at = parse_rfc3339(j["edited_at"].get_ref<const std::string&>());
    } else {
        post.edited_at = -1;
    }
    post.content = j.at("content").get<std::string>();
    if (!j.at("reblog").is_null()) {
        post.reblog = std::make_unique<Post>();
        from_json(j["reblog"].get<json>(), *post.reblog.get());
    }
    j.at("media_attachments").get_to(post.media_attachments);
    j.at("emojis").get_to(post.emojis);
    // https://social.kernel.org/@monsieuricon/Ac6oYwtLhess6uil1c
    if (j.contains("poll") && !j["poll"].is_null()) {
        post.poll = j["poll"].get<Poll>();
    }

    // empty account with username accessible: https://dlx.pink/notice/AbtdJkjioOo8ZSdDhw
    // empty account with username inaccesible: https://dlx.pink/notice/AbD2kgNviafFEsebqq
    if (j.at("account").size()) {
        j["account"].get_to(post.account);
    } else {
        std::smatch sm;
        const std::string& url = j.at("url").get_ref<const std::string&>();
        if (!std::regex_match(url, sm, akkoma_status_url_regex)) {
            return;
        }

        post.account = {
            .username = sm.str(2),
            .server = sm.str(1),
            .display_name = sm.str(2),
        };
    }
}

void from_json(const json& j, PostContext& context) {
    j.at("ancestors").get_to(context.ancestors);
    j.at("descendants").get_to(context.descendants);
}

void from_json(const json& j, Instance& instance) {
    j.at("title").get_to(instance.title);
    j.at("description").get_to(instance.description);
    j.at("thumbnail").at("url").get_to(instance.thumbnail);
    j.at("contact").at("email").get_to(instance.contact_email);
    j.at("contact").at("account").get_to(instance.contact_account);

    json rules = j.at("rules");
    instance.rules.reserve(rules.size());
    for (const json& i : rules) {
        instance.rules.push_back(i.at("text").get<std::string>());
    }
}


static std::regex rfc3339_re(R"EOF((\d{4})-(\d{2})-(\d{2})T(\d{2}):(\d{2}):(\d{2})(?:\.\d+)?(?:(Z)|([+-]\d{2}):(\d{2})))EOF", std::regex::ECMAScript | std::regex::icase);
time_t parse_rfc3339(const std::string& str) {
    using namespace std::string_literals;

    std::smatch sm;
    if (!std::regex_match(str, sm, rfc3339_re)) {
        throw std::invalid_argument("unknown date format: "s + str);
    }

    struct tm tm = {
        .tm_sec    = to_int(sm.str(6)),
        .tm_min    = to_int(sm.str(5)),
        .tm_hour   = to_int(sm.str(4)),

        .tm_mday   = to_int(sm.str(3)),
        .tm_mon    = to_int(sm.str(2)) - 1,
        .tm_year   = to_int(sm.str(1)) - 1900,

        .tm_gmtoff = !sm.str(7).empty() ? 0 : to_int(sm.str(8)) * 60 * 60 + to_int(sm.str(9)) * 60,
    };
    time_t time = mktime(&tm);
    if (time == -1) {
        throw std::system_error(errno, std::generic_category(), "mktime()");
    }
    return time;
}