Added MD5 file hash searching support to the scraper when using ScreenScraper

es-app/src/guis/GuiScraperMenu.cpp

@@ -145,6 +145,12 @@ GuiScraperMenu::GuiScraperMenu(std::string title)
     setSize(mMenu.getSize());
 
     setPosition((Renderer::getScreenWidth() - mSize.x) / 2.0f, Renderer::getScreenHeight() * 0.13f);
+
+    // Make sure that the hash searching max file size is within the allowed range.
+    if (Settings::getInstance()->getInt("ScraperSearchFileHashMaxSize") < 32)
+        Settings::getInstance()->setInt("ScraperSearchFileHashMaxSize", 32);
+    else if (Settings::getInstance()->getInt("ScraperSearchFileHashMaxSize") > 800)
+        Settings::getInstance()->setInt("ScraperSearchFileHashMaxSize", 800);
 }
 
 GuiScraperMenu::~GuiScraperMenu()
@@ -902,6 +908,33 @@ void GuiScraperMenu::openOtherOptions()
             ->setOpacity(DISABLED_OPACITY);
     }
 
+    // Maximum file size for non-interactive mode file hash searching.
+    auto scraperSearchFileHashMaxSize =
+        std::make_shared<SliderComponent>(32.0f, 800.0f, 32.0f, "MiB");
+    scraperSearchFileHashMaxSize->setValue(
+        static_cast<float>(Settings::getInstance()->getInt("ScraperSearchFileHashMaxSize")));
+    s->addWithLabel("HASH SEARCHES MAX FILE SIZE", scraperSearchFileHashMaxSize);
+    s->addSaveFunc([scraperSearchFileHashMaxSize, s] {
+        if (scraperSearchFileHashMaxSize->getValue() !=
+            static_cast<float>(Settings::getInstance()->getInt("ScraperSearchFileHashMaxSize"))) {
+            Settings::getInstance()->setInt(
+                "ScraperSearchFileHashMaxSize",
+                static_cast<int>(scraperSearchFileHashMaxSize->getValue()));
+            s->setNeedsSaving();
+        }
+    });
+
+    // File hash searching is not supported by TheGamesDB, so gray out the option if this scraper
+    // is selected. Also gray it out for ScreenScraper if file hash searching has been disabled.
+    if (Settings::getInstance()->getString("Scraper") == "thegamesdb" ||
+        !Settings::getInstance()->getBool("ScraperSearchFileHash")) {
+        scraperSearchFileHashMaxSize->setEnabled(false);
+        scraperSearchFileHashMaxSize->setOpacity(DISABLED_OPACITY);
+        scraperSearchFileHashMaxSize->getParent()
+            ->getChild(scraperSearchFileHashMaxSize->getChildIndex() - 1)
+            ->setOpacity(DISABLED_OPACITY);
+    }
+
     // Overwrite files and data.
     auto scraperOverwriteData = std::make_shared<SwitchComponent>();
     scraperOverwriteData->setState(Settings::getInstance()->getBool("ScraperOverwriteData"));
@@ -929,6 +962,29 @@ void GuiScraperMenu::openOtherOptions()
         }
     });
 
+    // Search using file hashes for non-interactive mode.
+    auto scraperSearchFileHash = std::make_shared<SwitchComponent>();
+    scraperSearchFileHash->setState(Settings::getInstance()->getBool("ScraperSearchFileHash"));
+    s->addWithLabel("SEARCH USING FILE HASHES (NON-INTERACTIVE MODE)", scraperSearchFileHash);
+    s->addSaveFunc([scraperSearchFileHash, s] {
+        if (scraperSearchFileHash->getState() !=
+            Settings::getInstance()->getBool("ScraperSearchFileHash")) {
+            Settings::getInstance()->setBool("ScraperSearchFileHash",
+                                             scraperSearchFileHash->getState());
+            s->setNeedsSaving();
+        }
+    });
+
+    // File hash searching is not supported by TheGamesDB, so gray out the option if this scraper
+    // is selected.
+    if (Settings::getInstance()->getString("Scraper") == "thegamesdb") {
+        scraperSearchFileHash->setEnabled(false);
+        scraperSearchFileHash->setOpacity(DISABLED_OPACITY);
+        scraperSearchFileHash->getParent()
+            ->getChild(scraperSearchFileHash->getChildIndex() - 1)
+            ->setOpacity(DISABLED_OPACITY);
+    }
+
     // Search using metadata names.
     auto scraperSearchMetadataName = std::make_shared<SwitchComponent>();
     scraperSearchMetadataName->setState(
@@ -1105,6 +1161,23 @@ void GuiScraperMenu::openOtherOptions()
     };
 
     // Switch callbacks.
+    auto hashSearchToggleFunc = [scraperSearchFileHashMaxSize]() {
+        if (scraperSearchFileHashMaxSize->getEnabled()) {
+            scraperSearchFileHashMaxSize->setEnabled(false);
+            scraperSearchFileHashMaxSize->setOpacity(DISABLED_OPACITY);
+            scraperSearchFileHashMaxSize->getParent()
+                ->getChild(scraperSearchFileHashMaxSize->getChildIndex() - 1)
+                ->setOpacity(DISABLED_OPACITY);
+        }
+        else {
+            scraperSearchFileHashMaxSize->setEnabled(true);
+            scraperSearchFileHashMaxSize->setOpacity(1.0f);
+            scraperSearchFileHashMaxSize->getParent()
+                ->getChild(scraperSearchFileHashMaxSize->getChildIndex() - 1)
+                ->setOpacity(1.0f);
+        }
+    };
+
     auto interactiveToggleFunc = [scraperSemiautomatic]() {
         if (scraperSemiautomatic->getEnabled()) {
             scraperSemiautomatic->setEnabled(false);
@@ -1140,6 +1213,7 @@ void GuiScraperMenu::openOtherOptions()
     };
 
     mScraperRetryOnErrorCount->setCallback(scraperRetryCountFunc);
+    scraperSearchFileHash->setCallback(hashSearchToggleFunc);
     scraperInteractive->setCallback(interactiveToggleFunc);
     scraperRespectExclusions->setCallback(excludeRecursivelyToggleFunc);
 

es-app/src/guis/GuiScraperSearch.cpp

@@ -40,6 +40,8 @@ GuiScraperSearch::GuiScraperSearch(SearchType type, unsigned int scrapeCount, in
     , mSearchType {type}
     , mRowCount {rowCount}
     , mScrapeCount {scrapeCount}
+    , mNextSearch {false}
+    , mHashSearch {false}
     , mRefinedSearch {false}
     , mBlockAccept {false}
     , mAcceptedResult {false}
@@ -164,14 +166,17 @@ GuiScraperSearch::~GuiScraperSearch()
     // scraping when the miximage was getting generated.
     if (Settings::getInstance()->getBool("MiximageGenerate") &&
         mMiximageGeneratorThread.joinable()) {
-        mScrapeResult.savedNewMedia = true;
         // We always let the miximage generator thread complete.
         mMiximageGeneratorThread.join();
         mMiximageGenerator.reset();
+        mScrapeResult.savedNewMedia = true;
         TextureResource::manualUnload(mLastSearch.game->getMiximagePath(), false);
         ViewController::getInstance()->onFileChanged(mLastSearch.game, true);
     }
 
+    if (mCalculateMD5HashThread.joinable())
+        mCalculateMD5HashThread.join();
+
     mWindow->setAllowTextScrolling(false);
 }
 
@@ -335,6 +340,7 @@ void GuiScraperSearch::updateView()
 
 void GuiScraperSearch::search(ScraperSearchParams& params)
 {
+    mHashSearch = false;
     mBlockAccept = true;
     mAcceptedResult = false;
     mMiximageResult = false;
@@ -358,8 +364,30 @@ void GuiScraperSearch::search(ScraperSearchParams& params)
     else
         params.automaticMode = false;
 
+    params.md5Hash = "";
+    if (!Utils::FileSystem::isDirectory(params.game->getPath()))
+        params.fileSize = Utils::FileSystem::getFileSize(params.game->getPath());
+
+    // Only use MD5 file hash searching when in non-interactive mode.
+    if (mSearchType == ALWAYS_ACCEPT_FIRST_RESULT &&
+        Settings::getInstance()->getBool("ScraperSearchFileHash") &&
+        Settings::getInstance()->getString("Scraper") == "screenscraper" && params.fileSize != 0 &&
+        params.fileSize <=
+            Settings::getInstance()->getInt("ScraperSearchFileHashMaxSize") * 1024 * 1024) {
+
+        // Run the MD5 hash calculation in a separate thread as it may take a long time to
+        // complete and we don't want to freeze the UI in the meanwhile.
+        std::promise<bool>().swap(mMD5HashPromise);
+        mMD5HashFuture = mMD5HashPromise.get_future();
+
+        mHashSearch = true;
+        mCalculateMD5HashThread =
+            std::thread(&GuiScraperSearch::calculateMD5Hash, this, params.game->getPath());
+    }
+
     mLastSearch = params;
-    mSearchHandle = startScraperSearch(params);
+    mSearchHandle = nullptr;
+    mNextSearch = true;
 }
 
 void GuiScraperSearch::stop()
@@ -691,6 +719,27 @@ void GuiScraperSearch::returnResult(ScraperSearchResult result)
 
 void GuiScraperSearch::update(int deltaTime)
 {
+    // The only purpose of calling startScraperSearch() here instead of in search() is because
+    // the optional MD5 hash calculation needs to run in a separate thread to not lock the UI.
+    if (mNextSearch && mHashSearch) {
+        if (mMD5HashFuture.valid()) {
+            // Only wait one millisecond as this update() function runs very frequently.
+            if (mMD5HashFuture.wait_for(std::chrono::milliseconds(1)) ==
+                std::future_status::ready) {
+                if (mCalculateMD5HashThread.joinable())
+                    mCalculateMD5HashThread.join();
+                mLastSearch.md5Hash = mMD5Hash;
+                mSearchHandle = startScraperSearch(mLastSearch);
+                mMD5Hash = "";
+                mNextSearch = false;
+            }
+        }
+    }
+    else if (mNextSearch) {
+        mSearchHandle = startScraperSearch(mLastSearch);
+        mNextSearch = false;
+    }
+
     GuiComponent::update(deltaTime);
 
     if (mBlockAccept)
@@ -789,6 +838,7 @@ void GuiScraperSearch::update(int deltaTime)
         if (mGeneratorFuture.wait_for(std::chrono::milliseconds(1)) == std::future_status::ready) {
             mMDResolveHandle.reset();
             // We always let the miximage generator thread complete.
+            if (mMiximageGeneratorThread.joinable())
                 mMiximageGeneratorThread.join();
             if (!mGeneratorFuture.get())
                 mScrapeResult.savedNewMedia = true;

es-app/src/guis/GuiScraperSearch.h

@@ -114,6 +114,12 @@ private:
 
     int getSelectedIndex();
 
+    void calculateMD5Hash(std::string path)
+    {
+        mMD5Hash = Utils::Math::md5Hash(path, true);
+        mMD5HashPromise.set_value(true);
+    }
+
     // For TheGamesDB, retrieve URLs for the additional metadata assets
     // that need to be downloaded.
     void retrieveMediaURLs(ScraperSearchResult result);
@@ -166,6 +172,8 @@ private:
     std::function<void()> mRefineCallback;
     int mRowCount;
     unsigned int mScrapeCount;
+    bool mNextSearch;
+    bool mHashSearch;
     bool mRefinedSearch;
     bool mBlockAccept;
     bool mAcceptedResult;
@@ -183,6 +191,11 @@ private:
     std::vector<ScraperSearchResult> mScraperResults;
     std::map<std::string, std::unique_ptr<HttpReq>> mThumbnailReqMap;
 
+    std::string mMD5Hash;
+    std::thread mCalculateMD5HashThread;
+    std::promise<bool> mMD5HashPromise;
+    std::future<bool> mMD5HashFuture;
+
     std::unique_ptr<MiximageGenerator> mMiximageGenerator;
     std::thread mMiximageGeneratorThread;
     std::promise<bool> mGeneratorPromise;

es-app/src/scrapers/Scraper.h

@@ -36,12 +36,15 @@ enum downloadStatus {
 struct ScraperSearchParams {
     SystemData* system;
     FileData* game;
+    std::string md5Hash;
+    long fileSize;
 
     std::string nameOverride;
     bool automaticMode;
 
     ScraperSearchParams()
-        : automaticMode {false}
+        : fileSize {0}
+        , automaticMode {false}
     {
     }
 };

es-app/src/scrapers/ScreenScraper.cpp

@@ -194,7 +194,8 @@ void screenscraper_generate_scraper_requests(const ScraperSearchParams& params,
     if (params.nameOverride == "") {
         if (Settings::getInstance()->getBool("ScraperSearchMetadataName")) {
             path = ssConfig.getGameSearchUrl(
-                Utils::String::removeParenthesis(params.game->metadata.get("name")));
+                Utils::String::removeParenthesis(params.game->metadata.get("name")), params.md5Hash,
+                params.fileSize);
         }
         else {
             std::string cleanName;
@@ -208,11 +209,11 @@ void screenscraper_generate_scraper_requests(const ScraperSearchParams& params,
                 cleanName = params.game->getCleanName();
             }
 
-            path = ssConfig.getGameSearchUrl(cleanName);
+            path = ssConfig.getGameSearchUrl(cleanName, params.md5Hash, params.fileSize);
         }
     }
     else {
-        path = ssConfig.getGameSearchUrl(params.nameOverride);
+        path = ssConfig.getGameSearchUrl(params.nameOverride, params.md5Hash, params.fileSize);
     }
 
     auto& platforms = params.system->getPlatformIds();
@@ -697,44 +698,16 @@ bool ScreenScraperRequest::processMedia(ScraperSearchResult& result,
     return regionFallback;
 }
 
-// Currently not used in this module.
-void ScreenScraperRequest::processList(const pugi::xml_document& xmldoc,
-                                       std::vector<ScraperSearchResult>& results)
+std::string ScreenScraperRequest::ScreenScraperConfig::getGameSearchUrl(const std::string& gameName,
+                                                                        const std::string& md5Hash,
+                                                                        const long fileSize) const
 {
-    assert(mRequestQueue != nullptr);
-
-    LOG(LogDebug) << "ScreenScraperRequest::processList(): Processing a list of results";
-
-    pugi::xml_node data {xmldoc.child("Data")};
-    pugi::xml_node game {data.child("jeu")};
-
-    if (!game) {
-        LOG(LogDebug) << "ScreenScraperRequest::processList(): Found nothing";
+    if (md5Hash != "") {
+        LOG(LogDebug)
+            << "ScreenScraper::getGameSearchUrl(): Performing MD5 file hash search using digest \""
+            << md5Hash << "\"";
     }
 
-    ScreenScraperRequest::ScreenScraperConfig ssConfig;
-
-    // Limit the number of results per platform, not in total.
-    // Otherwise if the first platform returns >= 7 games
-    // but the second platform contains the relevant game,
-    // the relevant result would not be shown.
-    for (int i {0}; game && i < MAX_SCRAPER_RESULTS; ++i) {
-        std::string id {game.child("id").text().get()};
-        std::string name {game.child("nom").text().get()};
-        std::string platformId {game.child("systemeid").text().get()};
-        std::string path {ssConfig.getGameSearchUrl(name) + "&systemeid=" + platformId +
-                          "&gameid=" + id};
-
-        mRequestQueue->push(
-            std::unique_ptr<ScraperRequest>(new ScreenScraperRequest(results, path)));
-
-        game = game.next_sibling("jeu");
-    }
-}
-
-std::string ScreenScraperRequest::ScreenScraperConfig::getGameSearchUrl(
-    const std::string gameName) const
-{
     std::string searchName {gameName};
     bool singleSearch {false};
 
@@ -811,6 +784,12 @@ std::string ScreenScraperRequest::ScreenScraperConfig::getGameSearchUrl(
             .append("&output=xml")
             .append("&romnom=")
             .append(HttpReq::urlEncode(searchName));
+        if (md5Hash != "") {
+            screenScraperURL.append("&md5=")
+                .append(md5Hash)
+                .append("&romtaille=")
+                .append(std::to_string(fileSize));
+        }
     }
     else {
         screenScraperURL.append(API_URL_BASE)

es-app/src/scrapers/ScreenScraper.h

@@ -41,7 +41,9 @@ public:
 
     // Settings for the scraper.
     static const struct ScreenScraperConfig {
-        std::string getGameSearchUrl(const std::string gameName) const;
+        std::string getGameSearchUrl(const std::string& gameName,
+                                     const std::string& md5Hash,
+                                     const long fileSize) const;
 
         // Access to the API.
         const std::string API_DEV_U = {15, 21, 39, 22, 42, 40};
@@ -119,7 +121,6 @@ protected:
     void process(const std::unique_ptr<HttpReq>& req,
                  std::vector<ScraperSearchResult>& results) override;
 
-    void processList(const pugi::xml_document& xmldoc, std::vector<ScraperSearchResult>& results);
     void processGame(const pugi::xml_document& xmldoc, std::vector<ScraperSearchResult>& results);
     bool processMedia(ScraperSearchResult& result,
                       const pugi::xml_node& media_list,

es-core/src/Settings.cpp

@@ -142,8 +142,10 @@ void Settings::setDefaults()
     mStringMap["ScraperLanguage"] = {"en", "en"};
     mIntMap["ScraperRetryOnErrorCount"] = {3, 3};
     mIntMap["ScraperRetryOnErrorTimer"] = {3, 3};
+    mIntMap["ScraperSearchFileHashMaxSize"] = {128, 128};
     mBoolMap["ScraperOverwriteData"] = {true, true};
     mBoolMap["ScraperHaltOnInvalidMedia"] = {true, true};
+    mBoolMap["ScraperSearchFileHash"] = {true, true};
     mBoolMap["ScraperSearchMetadataName"] = {true, true};
     mBoolMap["ScraperIncludeFolders"] = {true, true};
     mBoolMap["ScraperInteractive"] = {false, false};