pcgamedb/crawler/igdb.go

490 lines
12 KiB
Go
Raw Normal View History

2024-09-24 06:17:11 -04:00
package crawler
import (
"bytes"
2024-09-24 06:17:11 -04:00
"encoding/json"
"errors"
"fmt"
"net/url"
"runtime/debug"
"strconv"
"strings"
2024-11-26 10:14:15 -05:00
"time"
2024-11-15 02:02:45 -05:00
2024-11-20 06:09:04 -05:00
"pcgamedb/cache"
"pcgamedb/config"
"pcgamedb/constant"
"pcgamedb/db"
"pcgamedb/model"
"pcgamedb/utils"
"github.com/PuerkitoBio/goquery"
2024-12-02 03:17:01 -05:00
"github.com/go-resty/resty/v2"
2024-09-24 06:17:11 -04:00
)
2024-11-26 10:14:15 -05:00
type twitchToken struct {
}
2024-09-24 06:17:11 -04:00
var token = twitchToken{}
2024-11-26 10:14:15 -05:00
func (t *twitchToken) getToken() (string, error) {
if val, exist := cache.Get("twitch_token"); exist {
return val, nil
}
token, expires, err := loginTwitch()
if err != nil {
return "", fmt.Errorf("failed to login twitch: %w", err)
2024-09-24 06:17:11 -04:00
}
_ = cache.SetWithExpire("twitch_token", token, expires)
return token, nil
2024-11-26 10:14:15 -05:00
}
func loginTwitch() (string, time.Duration, error) {
2024-11-26 10:14:15 -05:00
baseURL, _ := url.Parse(constant.TwitchAuthURL)
params := url.Values{}
params.Add("client_id", config.Config.Twitch.ClientID)
params.Add("client_secret", config.Config.Twitch.ClientSecret)
params.Add("grant_type", "client_credentials")
baseURL.RawQuery = params.Encode()
2024-12-02 03:17:01 -05:00
resp, err := utils.Request().SetHeader("User-Agent", "").Post(baseURL.String())
2024-11-26 10:14:15 -05:00
if err != nil {
return "", 0, err
2024-11-26 10:14:15 -05:00
}
data := struct {
AccessToken string `json:"access_token"`
ExpiresIn int64 `json:"expires_in"`
TokenType string `json:"token_type"`
}{}
2024-12-02 03:17:01 -05:00
err = json.Unmarshal(resp.Body(), &data)
2024-11-26 10:14:15 -05:00
if err != nil {
return "", 0, err
2024-11-26 10:14:15 -05:00
}
return data.AccessToken, time.Second * time.Duration(data.ExpiresIn), nil
2024-11-26 10:14:15 -05:00
}
2024-12-02 03:17:01 -05:00
func igdbRequest(URL string, dataBody any) (*resty.Response, error) {
2024-11-26 10:14:15 -05:00
t, err := token.getToken()
if err != nil {
return nil, err
}
2024-12-02 03:17:01 -05:00
resp, err := utils.Request().SetBody(dataBody).SetHeaders(map[string]string{
"Client-ID": config.Config.Twitch.ClientID,
"Authorization": "Bearer " + t,
"User-Agent": "",
"Content-Type": "text/plain",
}).Post(URL)
2024-11-26 10:14:15 -05:00
if err != nil {
return nil, err
2024-09-24 06:17:11 -04:00
}
2024-11-26 10:14:15 -05:00
return resp, nil
}
func getIGDBID(name string) (int, error) {
var err error
2024-12-02 03:17:01 -05:00
resp, err := igdbRequest(constant.IGDBSearchURL, fmt.Sprintf(`search "%s"; fields *; limit 50; where game.platforms = [6] | game.platforms=[130] | game.platforms=[384] | game.platforms=[163];`, name))
2024-09-24 06:17:11 -04:00
if err != nil {
return 0, err
}
2024-12-02 03:17:01 -05:00
if string(resp.Body()) == "[]" {
resp, err = igdbRequest(constant.IGDBSearchURL, fmt.Sprintf(`search "%s"; fields *; limit 50;`, name))
if err != nil {
return 0, err
}
2024-11-26 10:14:15 -05:00
}
2024-09-24 06:17:11 -04:00
var data model.IGDBSearches
2024-12-02 03:17:01 -05:00
if err = json.Unmarshal(resp.Body(), &data); err != nil {
2024-09-24 06:17:11 -04:00
return 0, fmt.Errorf("failed to unmarshal: %w, %s", err, debug.Stack())
}
if len(data) == 1 {
return GetIGDBAppParent(data[0].Game)
2024-09-24 06:17:11 -04:00
}
maxSimilairty := 0.0
maxSimilairtyIndex := 0
for i, item := range data {
2024-09-24 06:17:11 -04:00
if strings.EqualFold(item.Name, name) {
return item.Game, nil
}
if sim := utils.Similarity(name, item.Name); sim >= 0.8 {
if sim > maxSimilairty {
maxSimilairty = sim
maxSimilairtyIndex = i
}
2024-09-24 06:17:11 -04:00
}
detail, err := GetIGDBAppDetail(item.Game)
2024-09-24 06:17:11 -04:00
if err != nil {
return 0, err
}
for _, alternativeNames := range detail.AlternativeNames {
if sim := utils.Similarity(alternativeNames.Name, name); sim >= 0.8 {
if sim > maxSimilairty {
maxSimilairty = sim
maxSimilairtyIndex = i
}
2024-09-24 06:17:11 -04:00
}
}
}
if maxSimilairty >= 0.8 {
return GetIGDBAppParent(data[maxSimilairtyIndex].Game)
}
2024-09-24 06:17:11 -04:00
return 0, fmt.Errorf("IGDB ID not found: %s", name)
}
func getIGDBIDBySteamSearch(name string) (int, error) {
baseURL, _ := url.Parse(constant.SteamSearchURL)
params := url.Values{}
params.Add("term", name)
baseURL.RawQuery = params.Encode()
2024-12-02 03:17:01 -05:00
resp, err := utils.Request().Get(baseURL.String())
if err != nil {
return 0, err
}
2024-12-02 03:17:01 -05:00
doc, err := goquery.NewDocumentFromReader(bytes.NewReader(resp.Body()))
if err != nil {
return 0, err
}
type searchResult struct {
ID int
Type string
Name string
}
var items []searchResult
doc.Find(".search_result_row").Each(func(i int, s *goquery.Selection) {
if itemKey, exist := s.Attr("data-ds-itemkey"); exist {
if strings.HasPrefix(itemKey, "App_") {
id, err := strconv.Atoi(itemKey[4:])
if err != nil {
return
}
name := s.Find(".title").Text()
items = append(items, searchResult{
ID: id,
Type: "App",
Name: name,
})
}
if strings.HasPrefix(itemKey, "Bundle_") {
id, err := strconv.Atoi(itemKey[7:])
if err != nil {
return
}
name := s.Find(".title").Text()
items = append(items, searchResult{
ID: id,
Type: "Bundle",
Name: name,
})
}
}
})
maxSim := 0.0
var maxSimItem searchResult
for _, item := range items {
if strings.EqualFold(strings.TrimSpace(item.Name), strings.TrimSpace(name)) {
maxSimItem = item
break
} else {
sim := utils.Similarity(item.Name, name)
if sim >= 0.8 && sim > maxSim {
maxSim = sim
maxSimItem = item
}
}
}
if maxSim != 0 {
if maxSimItem.Type == "App" {
return GetIGDBIDBySteamAppID(maxSimItem.ID)
}
if maxSimItem.Type == "Bundle" {
return GetIGDBIDBySteamBundleID(maxSimItem.ID)
}
}
return 0, fmt.Errorf("steam ID not found: %s", name)
}
// GetIGDBAppParent returns the parent of the game, if no parent return itself
func GetIGDBAppParent(id int) (int, error) {
key := fmt.Sprintf("igdb_parent:%d", id)
val, exist := cache.Get(key)
if exist {
id, err := strconv.Atoi(val)
if err != nil {
return 0, err
}
return id, nil
}
detail, err := GetIGDBAppDetail(id)
if err != nil {
return 0, err
}
2024-11-22 13:27:53 -05:00
hasParent := false
for detail.VersionParent != 0 {
hasParent = true
detail, err = GetIGDBAppDetail(detail.VersionParent)
if err != nil {
return 0, err
}
}
2024-11-22 13:27:53 -05:00
if hasParent {
return detail.ID, nil
}
_ = cache.Set(key, id)
2024-11-26 10:14:15 -05:00
return id, nil
}
// GetIGDBID returns the IGDB ID of the game, try directly IGDB api first, then steam search
2024-09-24 06:17:11 -04:00
func GetIGDBID(name string) (int, error) {
key := fmt.Sprintf("igdb_id:%s", name)
val, exist := cache.Get(key)
if exist {
return strconv.Atoi(val)
}
2024-09-24 06:17:11 -04:00
name1 := name
name2 := FormatName(name)
names := []string{name1}
if name1 != name2 {
names = append(names, name2)
}
for _, name := range names {
id, err := getIGDBID(name)
2024-09-24 06:17:11 -04:00
if err == nil {
_ = cache.Set(key, id)
2024-09-24 06:17:11 -04:00
return id, nil
}
}
for _, name := range names {
id, err := getIGDBIDBySteamSearch(name)
if err == nil {
_ = cache.Set(key, id)
return id, nil
}
}
2024-09-24 06:17:11 -04:00
return 0, errors.New("IGDB ID not found")
}
func GetIGDBAppDetail(id int) (*model.IGDBGameDetail, error) {
key := fmt.Sprintf("igdb_game:%v", id)
val, exist := cache.Get(key)
if exist {
var data model.IGDBGameDetail
if err := json.Unmarshal([]byte(val), &data); err != nil {
return nil, err
2024-09-24 06:17:11 -04:00
}
return &data, nil
2024-09-24 06:17:11 -04:00
}
var err error
2024-12-02 03:17:01 -05:00
resp, err := igdbRequest(constant.IGDBGameURL, fmt.Sprintf(`where id=%v ;fields *,alternative_names.name,language_supports.language,language_supports.language_support_type,screenshots.url,cover.url,involved_companies.company,involved_companies.developer,involved_companies.publisher;`, id))
2024-11-26 10:14:15 -05:00
2024-09-24 06:17:11 -04:00
if err != nil {
return nil, err
}
var data model.IGDBGameDetails
2024-12-02 03:17:01 -05:00
if err = json.Unmarshal(resp.Body(), &data); err != nil {
2024-09-24 06:17:11 -04:00
return nil, err
}
if len(data) == 0 {
return nil, errors.New("IGDB App not found")
}
if data[0].Name == "" {
return GetIGDBAppDetail(id)
}
jsonBytes, err := json.Marshal(data[0])
if err == nil {
_ = cache.Set(key, string(jsonBytes))
2024-09-24 06:17:11 -04:00
}
return data[0], nil
2024-09-24 06:17:11 -04:00
}
func GetIGDBCompany(id int) (string, error) {
key := fmt.Sprintf("igdb_companies:%v", id)
val, exist := cache.Get(key)
if exist {
return val, nil
}
2024-09-24 06:17:11 -04:00
var err error
2024-12-02 03:17:01 -05:00
resp, err := igdbRequest(constant.IGDBCompaniesURL, fmt.Sprintf(`where id=%v; fields *;`, id))
2024-09-24 06:17:11 -04:00
if err != nil {
return "", err
}
var data model.IGDBCompanies
2024-12-02 03:17:01 -05:00
if err = json.Unmarshal(resp.Body(), &data); err != nil {
2024-09-24 06:17:11 -04:00
return "", err
}
if len(data) == 0 {
return "", errors.New("not found")
2024-09-24 06:17:11 -04:00
}
if data[0].Name == "" {
return GetIGDBCompany(id)
}
_ = cache.Set(key, data[0].Name)
return data[0].Name, nil
2024-09-24 06:17:11 -04:00
}
func GenerateIGDBGameInfo(id int) (*model.GameInfo, error) {
item := &model.GameInfo{}
detail, err := GetIGDBAppDetail(id)
2024-09-24 06:17:11 -04:00
if err != nil {
return nil, err
}
item.IGDBID = id
item.Name = detail.Name
item.Description = detail.Summary
item.Cover = strings.Replace(detail.Cover.URL, "t_thumb", "t_original", 1)
item.InfoUpdatedAt = time.Now()
2024-09-24 06:17:11 -04:00
for _, lang := range detail.LanguageSupports {
if lang.LanguageSupportType == 3 {
l, exist := constant.IGDBLanguages[lang.Language]
if !exist {
continue
}
item.Languages = append(item.Languages, l.Name)
}
}
for _, screenshot := range detail.Screenshots {
item.Screenshots = append(item.Screenshots, strings.Replace(screenshot.URL, "t_thumb", "t_original", 1))
}
for _, alias := range detail.AlternativeNames {
item.Aliases = append(item.Aliases, alias.Name)
}
for _, company := range detail.InvolvedCompanies {
if company.Developer || company.Publisher {
companyName, err := GetIGDBCompany(company.Company)
2024-09-24 06:17:11 -04:00
if err != nil {
continue
}
if company.Developer {
item.Developers = append(item.Developers, companyName)
}
if company.Publisher {
item.Publishers = append(item.Publishers, companyName)
}
}
}
return item, nil
}
// OrganizeGameItemWithIGDB Will add GameItem.ID to the newly added GameInfo.GameIDs
func OrganizeGameItemWithIGDB(game *model.GameItem) (*model.GameInfo, error) {
id, err := GetIGDBID(game.Name)
if err != nil {
return nil, err
2024-09-24 06:17:11 -04:00
}
d, err := db.GetGameInfoByPlatformID("igdb", id)
if err == nil {
d.GameIDs = append(d.GameIDs, game.ID)
d.GameIDs = utils.Unique(d.GameIDs)
return d, nil
}
info, err := GenerateGameInfo("igdb", id)
if err != nil {
return nil, err
}
info.GameIDs = append(info.GameIDs, game.ID)
info.GameIDs = utils.Unique(info.GameIDs)
return info, nil
}
func GetIGDBIDBySteamAppID(id int) (int, error) {
key := fmt.Sprintf("igdb_id_by_steam_app_id:%v", id)
val, exist := cache.Get(key)
if exist {
return strconv.Atoi(val)
}
2024-09-24 06:17:11 -04:00
var err error
2024-12-02 03:17:01 -05:00
resp, err := igdbRequest(constant.IGDBWebsitesURL, fmt.Sprintf(`where url = "https://store.steampowered.com/app/%v" | url = "https://store.steampowered.com/app/%v/"*; fields *; limit 500;`, id, id))
2024-09-24 06:17:11 -04:00
if err != nil {
return 0, err
}
var data []struct {
Game int `json:"game"`
}
2024-12-02 03:17:01 -05:00
if err = json.Unmarshal(resp.Body(), &data); err != nil {
2024-09-24 06:17:11 -04:00
return 0, err
}
if len(data) == 0 {
return 0, errors.New("not found")
2024-09-24 06:17:11 -04:00
}
if data[0].Game == 0 {
return GetIGDBIDBySteamAppID(id)
}
_ = cache.Set(key, strconv.Itoa(data[0].Game))
return GetIGDBAppParent(data[0].Game)
}
func GetIGDBIDBySteamBundleID(id int) (int, error) {
key := fmt.Sprintf("igdb_id_by_steam_bundle_id:%v", id)
val, exist := cache.Get(key)
if exist {
return strconv.Atoi(val)
}
var err error
2024-12-02 03:17:01 -05:00
resp, err := igdbRequest(constant.IGDBWebsitesURL, fmt.Sprintf(`where url = "https://store.steampowered.com/bundle/%v" | url = "https://store.steampowered.com/bundle/%v/"*; fields *; limit 500;`, id, id))
if err != nil {
return 0, err
}
var data []struct {
Game int `json:"game"`
}
2024-12-02 03:17:01 -05:00
if err = json.Unmarshal(resp.Body(), &data); err != nil {
return 0, err
}
if len(data) == 0 {
return 0, errors.New("not found")
}
if data[0].Game == 0 {
return GetIGDBIDBySteamBundleID(id)
2024-09-24 06:17:11 -04:00
}
_ = cache.Set(key, strconv.Itoa(data[0].Game))
2024-09-24 06:17:11 -04:00
return GetIGDBAppParent(data[0].Game)
2024-09-24 06:17:11 -04:00
}
// GetIGDBPopularGameIDs get IGDB popular game IDs
// popularity_type = 1 IGDB Visits: Game page visits on IGDB.com.
// popularity_type = 2 IGDB Want to Play: Additions to IGDB.com users “Want to Play” lists.
// popularity_type = 3 IGDB Playing: Additions to IGDB.com users “Playing” lists.
// popularity_type = 4 IGDB Played: Additions to IGDB.com users “Played” lists.
func GetIGDBPopularGameIDs(popularityType int, offset int, limit int) ([]int, error) {
var err error
2024-12-02 03:17:01 -05:00
resp, err := igdbRequest(constant.IGDBPopularityURL, fmt.Sprintf("fields game_id,value,popularity_type; sort value desc; limit %v; offset %v; where popularity_type = %v;", limit, offset, popularityType))
if err != nil {
return nil, err
}
2024-11-26 10:14:15 -05:00
type IgdbPopularity struct {
GameID int `json:"game_id"`
Value float64 `json:"value"`
}
2024-11-26 10:14:15 -05:00
var data []IgdbPopularity
2024-12-02 03:17:01 -05:00
if err = json.Unmarshal(resp.Body(), &data); err != nil {
return nil, err
}
ret := make([]int, 0)
for _, d := range data {
pid, err := GetIGDBAppParent(d.GameID)
if err != nil {
ret = append(ret, d.GameID)
continue
}
ret = append(ret, pid)
}
return ret, nil
}