pixwhile/pixivclient.cpp

#include <FastHash.h>

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

using namespace std::string_literals;

static const constexpr char* touch_user_agent = "Mozilla/5.0 (Android 12; Mobile; rv:97.0) Gecko/97.0 Firefox/97.0";
static const constexpr char* desktop_user_agent = "Mozilla/5.0 (Windows NT 10.0; rv:111.0) Gecko/20100101 Firefox/111.0";
static void to_lower(std::string& str);

PixivClient::PixivClient(Redis* redis) : _redis(redis) {
    this->_www_pixiv_net_client.set_keep_alive(true);
    this->_www_pixiv_net_client.set_default_headers({
        {"Cookie", "webp_available=1"}
    });
    this->_i_pximg_net_client.set_keep_alive(true);
    this->_i_pximg_net_client.set_default_headers({
        {"Referer", "https://www.pixiv.net/"}
    });
}

User PixivClient::get_user(uint64_t user_id) {
    nlohmann::json j = this->_call_api("pixwhile:user:"s + std::to_string(user_id), std::nullopt, 24 * 60 * 60,
        "/touch/ajax/user/details", {{"lang", "en"}, {"id", std::to_string(user_id)}}, {{"User-Agent", touch_user_agent}});
    return j.at("user_details").get<User>();
}

Illusts PixivClient::get_illusts(uint64_t user_id, size_t page) {
    httplib::Params params = {{"lang", "en"}, {"id", std::to_string(user_id)}};
    if (page != 0) {
        params.insert({"p", std::to_string(page + 1)});
    }

    nlohmann::json j = this->_call_api("pixwhile:illusts:"s + std::to_string(user_id), std::to_string(page), 60 * 60,
        "/touch/ajax/user/illusts", std::move(params), {{"User-Agent", touch_user_agent}});
    return j.get<Illusts>();
}

Illust PixivClient::get_illust(uint64_t illust_id) {
    nlohmann::json j = this->_call_api("pixwhile:illust:"s + std::to_string(illust_id), std::nullopt, 24 * 60 * 60,
        "/touch/ajax/illust/details", {{"lang", "en"}, {"illust_id", std::to_string(illust_id)}}, {{"User-Agent", touch_user_agent}});
    return j.get<Illust>();
}

SearchResults PixivClient::search_illusts(std::string query, size_t page, const std::string& order) {
    to_lower(query);
    httplib::Params params = {
        {"lang", "en"},
        {"mode", "all"},
        {"p", std::to_string(page + 1)},
        {"s_mode", "s_tag_full"},
        {"type", "illust_and_ugoira"},
        {"order", order},
        {"word", query}
    };

    std::string cache_key = "pixwhile:search:"s + order + ':' + std::to_string(FastHash(query.data(), query.size(), 0));
    nlohmann::json j = this->_call_api(std::move(cache_key), std::to_string(page), 60 * 60,
        "/ajax/search/illustrations/"s + blankie::murl::escape(std::move(query)), std::move(params), {{"User-Agent", desktop_user_agent}});
    return j.get<SearchResults>();
}

std::vector<SearchSuggestion> PixivClient::get_search_suggestions(std::string query) {
    to_lower(query);
    std::string body = [&]() {
        std::string cache_key = "pixwhile:searchsugg:"s + std::to_string(FastHash(query.data(), query.size(), 0));
        std::optional<std::string> cached_body = this->_redis->get(cache_key);
        if (cached_body) {
            return std::move(*cached_body);
        }

        httplib::Result res = this->_www_pixiv_net_client.Get("/rpc/cps.php", {
            {"lang", "en"}, {"keyword", query}
        }, {{"User-Agent", desktop_user_agent}, {"Referer", "https://www.pixiv.net/"}});
        if (!res) {
            throw HTTPLibException(res.error());
        }

        this->_redis->set(std::move(cache_key), res->body, 24 * 60 * 60);
        return std::move(res->body);
    }();
    nlohmann::json j = nlohmann::json::parse(std::move(body)).at("candidates");

    std::vector<SearchSuggestion> search_suggestions;
    search_suggestions.reserve(j.size());

    for (const nlohmann::json& i : j) {
        SearchSuggestion search_suggestion = i.get<SearchSuggestion>();
        if (search_suggestion.tag != query) {
            search_suggestions.push_back(std::move(search_suggestion));
        }
    }

    return search_suggestions;
}

nlohmann::json PixivClient::_call_api(std::string cache_key, std::optional<std::string> cache_field, time_t expiry,
        std::string path, httplib::Params params, httplib::Headers headers) {
    std::optional<std::string> success_body = !cache_field
        ? this->_redis->get(cache_key)
        : this->_redis->hget(cache_key, *cache_field);
    if (success_body) {
        return nlohmann::json::parse(std::move(*success_body));
    }

    httplib::Result res = this->_www_pixiv_net_client.Get(std::move(path), std::move(params), std::move(headers));
    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>());
    }

    j = j.at("body");
    // trim data sent to redis without making our own serialization of our objects
    j.erase("ads");
    if (!cache_field) {
        this->_redis->set(std::move(cache_key), j.dump(), expiry);
    } else {
        this->_redis->hset(cache_key, std::move(*cache_field), j.dump());
        this->_redis->expire_nx(std::move(cache_key), expiry);
    }
    return j;
}

bool PixivClient::i_pximg_url_valid(const std::string& path) {
    httplib::Result res = this->_i_pximg_net_client.Head(path, {{"User-Agent", touch_user_agent}});
    if (!res) {
        throw HTTPLibException(res.error());
    }
    return res->status >= 200 && res->status <= 200;
}


static void to_lower(std::string& str) {
    for (size_t i = 0; i < str.size(); i++) {
        if (str[i] >= 'A' && str[i] <= 'Z') {
            str[i] = str[i] - 'A' + 'a';
        }
    }
}