pixwhile/pixivmodels.cpp

#include <regex>

#include "blankie/murl.h"

#include "pixivmodels.h"
#include "numberhelper.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 Images get_profile_pictures(const nlohmann::json& j);
static Images get_illust_image(const nlohmann::json& j);

const std::string& Images::original_or_thumbnail() const {
    if (this->original) {
        return *this->original;
    }
    if (!this->thumbnails.empty()) {
        return this->thumbnails.back();
    }
    throw std::runtime_error("Images does not contain any images");
}

const std::string& Images::thumbnail_or_original() const {
    if (!this->thumbnails.empty()) {
        return this->thumbnails.back();
    }
    if (this->original) {
        return *this->original;
    }
    throw std::runtime_error("Images does not contain any images");
}

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));
        }
    }

    user.profile_pictures = get_profile_pictures(j.at("profile_img"));

    if (j.contains("user_webpage")) {
        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["social"];
        if (!social.contains(key)) {
            return;
        }

        std::string url = social[key].at("url").get<std::string>();
        user.links.push_back({public_name, std::move(url)});
    };
    if (j.contains("social") && j["social"].is_object()) {
        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");
    }
}

void from_json(const nlohmann::json& j, Tag& tag) {
    j.at("tag").get_to(tag.japanese);
    if (j.contains("romaji")) {
        tag.romaji = j["romaji"].get<std::string>();
    }
    if (j.contains("translation")) {
        tag.english = j["translation"].get<std::string>();
    }
}

void from_json(const nlohmann::json& j, Illust& illust) {
    bool full_data = j.contains("illust_details");
    const nlohmann::json& author_details = j.at("author_details");
    const nlohmann::json& illust_details = full_data ? j.at("illust_details") : j;

    author_details.at("user_account").get_to(illust.username);
    author_details.at("user_name").get_to(illust.display_name);
    illust.user_id = to_ull(author_details.at("user_id").get_ref<const nlohmann::json::string_t&>());
    if (full_data) {
        illust.user_profile_pictures = get_profile_pictures(author_details.at("profile_img"));
    }

    illust.illust_id = to_ull(illust_details.at("id").get_ref<const nlohmann::json::string_t&>());
    illust_details.at("title").get_to(illust.title);
    illust.ai_generated = illust_details.at("ai_type").get<int>() != 0;
    illust_details.at("upload_timestamp").get_to(illust.upload_time);

    std::string comment_html = illust_details.at("comment_html").get<std::string>();
    if (!comment_html.empty()) {
        illust.comment_html = std::move(comment_html);
    }
    if (illust_details.contains("display_tags")) {
        illust_details["display_tags"].get_to(illust.tags);
    }

    if (illust_details.contains("manga_a")) {
        const nlohmann::json& manga_a = illust_details["manga_a"];
        illust.images.reserve(manga_a.size());
        for (auto &[_, i] : manga_a.items()) {
            illust.images.push_back(get_illust_image(i));
        }
    } else {
        illust.images = {get_illust_image(illust_details)};
    }
}

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 Images get_profile_pictures(const nlohmann::json& j) {
    Images images;

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

    return images;
}

static Images get_illust_image(const nlohmann::json& j) {
    Images images;

    auto add_if_exists = [&](const char* key) {
        if (j.contains(key) && j[key].is_string()) {
            images.thumbnails.push_back(j[key].get<std::string>());
        }
    };
    add_if_exists("url_ss");
    add_if_exists("url_placeholder");
    add_if_exists("url_s");
    add_if_exists("url");
    if (j.contains("url_big") && j["url_big"].is_string()) {
        std::string url_big = j["url_big"].get<std::string>();
        if (url_big.starts_with("/img-original/")) {
            images.original = std::move(url_big);
        } else {
            images.thumbnails.push_back(std::move(url_big));
        }
    }

    return images;
}