pixwhile/pixivclient.cpp

#include <regex>

#include "blankie/murl.h"
#include "pixivclient.h"

static inline std::optional<std::string> get_1920x960_cover_image(blankie::murl::Url url);
static inline std::optional<std::string> get_original_cover_image(blankie::murl::Url url);
static inline std::optional<std::string> get_original_profile_picture(blankie::murl::Url url);
static inline uint64_t to_ull(const std::string& str);

PixivClient::PixivClient() {
    this->_www_pixiv_net_client.set_keep_alive(true);
    this->_www_pixiv_net_client.set_default_headers({
        {"User-Agent", "Mozilla/5.0 (Android 12; Mobile; rv:97.0) Gecko/97.0 Firefox/97.0"}
    });
}

User PixivClient::get_user(uint64_t user_id) {
    httplib::Result res = this->_www_pixiv_net_client.Get("/touch/ajax/user/details", {
        {"lang", "en"}, {"id", std::to_string(user_id)}
    }, httplib::Headers());
    return this->_handle_result(std::move(res)).at("user_details").get<User>();
}

std::vector<uint64_t> PixivClient::get_illusts(uint64_t user_id) {
    httplib::Result res = this->_www_pixiv_net_client.Get("/touch/ajax/illust/user_illusts", {
        {"lang", "en"}, {"user_id", std::to_string(user_id)}
    }, httplib::Headers());

    nlohmann::json user_illust_ids = this->_handle_result(std::move(res)).at("user_illust_ids");
    std::vector<uint64_t> ret;
    ret.reserve(user_illust_ids.size());

    for (auto &[_, i] : user_illust_ids.items()) {
        ret.push_back(to_ull(std::move(i)));
    }

    return ret;
}

nlohmann::json PixivClient::_handle_result(httplib::Result res) {
    if (!res) {
        throw HTTPLibException(res.error());
    }

    nlohmann::json j = nlohmann::json::parse(std::move(res->body));
    if (j.at("error")) {
        throw PixivException(res->status, j.at("message").get<std::string>());
    }
    return j.at("body");
}

void from_json(const nlohmann::json& j, User& user) {
    j.at("user_account").get_to(user.username);
    j.at("user_name").get_to(user.display_name);
    user.user_id = to_ull(j.at("user_id").get_ref<const nlohmann::json::string_t&>());

    if (j.contains("cover_image") && j["cover_image"].is_object()) {
        nlohmann::json cover_image = j["cover_image"];
        std::string c_720x360 = cover_image.at("profile_cover_image").at("720x360").get<std::string>();
        std::optional<std::string> original = get_original_cover_image(c_720x360);
        std::optional<std::string> c_1920x960 = get_1920x960_cover_image(c_720x360);

        user.cover_images = {std::move(original), {std::move(c_720x360)}};
        if (c_1920x960) {
            user.cover_images->thumbnails.push_back(std::move(*c_1920x960));
        }
    }

    nlohmann::json profile_img = j.at("profile_img");
    if (profile_img.contains("main_s")) {
        user.profile_pictures.thumbnails.push_back(profile_img["main_s"].get<std::string>());
    }
    user.profile_pictures.thumbnails.push_back(profile_img.at("main").get<std::string>());
    user.profile_pictures.original = get_original_profile_picture(user.profile_pictures.thumbnails.back());

    std::string user_webpage = j.at("user_webpage").get<std::string>();
    if (!user_webpage.empty()) {
        user.links.push_back({"Webpage", std::move(user_webpage)});
    }
    auto add_social_as_needed = [&](const char* key, const char* public_name) {
        nlohmann::json social = j.at("social");
        if (!social.is_object()) {
            return;
        }
        if (!social.contains(key)) {
            return;
        }

        std::string url = social[key].at("url").get<std::string>();
        user.links.push_back({public_name, std::move(url)});
    };
    add_social_as_needed("twitter", "Twitter");
    add_social_as_needed("instagram", "Instagram");
    add_social_as_needed("tumblr", "Tumblr");
    add_social_as_needed("facebook", "Facebook");
    add_social_as_needed("circlems", "Circle.ms");
    add_social_as_needed("pawoo", "Pawoo");
}

static std::regex resolution_path_regex("/c/(\\d+x\\d+)(.+)");
static inline std::optional<std::string> get_1920x960_cover_image(blankie::murl::Url url) {
    std::smatch sm;
    if (!std::regex_match(url.path, sm, resolution_path_regex)) {
        return std::nullopt;
    }
    if (sm[1] == "1920x960") {
        return std::nullopt;
    }
    url.path = "/c/1920x960" + sm.str(2);
    return url.to_string();
}

static std::regex thumbnail_path_regex("/c/[^/]+/(.+)_master\\d+(\\.\\w{3,4})?");
static inline std::optional<std::string> get_original_cover_image(blankie::murl::Url url) {
    std::smatch sm;
    if (!std::regex_match(url.path, sm, thumbnail_path_regex)) {
        return std::nullopt;
    }
    url.path = sm.str(1) + sm.str(2);
    return url.to_string();
}

static std::regex profile_picture_thumbnail_path_regex("(/.+)_\\d{2,}(\\.\\w{3,4})");
static inline std::optional<std::string> get_original_profile_picture(blankie::murl::Url url) {
    std::smatch sm;
    if (!std::regex_match(url.path, sm, profile_picture_thumbnail_path_regex)) {
        return std::nullopt;
    }
    url.path = sm.str(1) + sm.str(2);
    return url.to_string();
}

static inline uint64_t to_ull(const std::string& str) {
    char* endptr;

    errno = 0;
    unsigned long long res = strtoull(str.c_str(), &endptr, 10);
    if (endptr[0] != '\0') {
        throw std::invalid_argument(str + " contains trailing data");
    } else if (res == ULLONG_MAX && errno == ERANGE) {
        throw std::overflow_error(str + " overflows uint64_t");
    }

    return res;
}