From 6e863a82ecdf472c224447d726d6d88dad44bc93 Mon Sep 17 00:00:00 2001
From: Leon Styhre <leon@leonstyhre.com>
Date: Sat, 9 Sep 2023 09:48:56 +0200
Subject: [PATCH] Added conversion of some additional HTML character codes when
 scraping using ScreenScraper

---
 es-app/src/guis/GuiScraperSearch.cpp  |  5 -----
 es-app/src/scrapers/ScreenScraper.cpp | 18 +++++++++++++-----
 2 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/es-app/src/guis/GuiScraperSearch.cpp b/es-app/src/guis/GuiScraperSearch.cpp
index c30992c6e..52351a2bc 100644
--- a/es-app/src/guis/GuiScraperSearch.cpp
+++ b/es-app/src/guis/GuiScraperSearch.cpp
@@ -1104,11 +1104,6 @@ bool GuiScraperSearch::saveMetadata(const ScraperSearchResult& result,
             metadata.set(key, result.mdl.get(key));
             metadataUpdated = true;
         }
-
-        // For the description, expand any escaped HTML quotation marks to literal
-        // quotation marks.
-        if (key == "desc" && metadataUpdated)
-            metadata.set(key, Utils::String::replace(metadata.get(key), "&quot;", "\""));
     }
 
     return metadataUpdated;
diff --git a/es-app/src/scrapers/ScreenScraper.cpp b/es-app/src/scrapers/ScreenScraper.cpp
index 2892832f6..51f8f4380 100644
--- a/es-app/src/scrapers/ScreenScraper.cpp
+++ b/es-app/src/scrapers/ScreenScraper.cpp
@@ -382,12 +382,15 @@ void ScreenScraperRequest::processGame(const pugi::xml_document& xmldoc,
                                                            {region, "wor", "us", "ss", "eu", "jp"})
                                   .text()
                                   .get()};
-        // Game names sometimes contain ampersands encoded as hex codes.
+
+        // Translate some HTML character codes to UTF-8 characters for the game name.
+        gameName = Utils::String::replace(gameName, "&nbsp;", " ");
         gameName = Utils::String::replace(gameName, "&#x26;", "&");
+
         // In some very rare cases game names contain newline characters that we need to remove.
         result.mdl.set("name", Utils::String::replace(gameName, "\n", ""));
-        LOG(LogDebug) << "ScreenScraperRequest::processGame(): Name: " << result.mdl.get("name");
 
+        LOG(LogDebug) << "ScreenScraperRequest::processGame(): Name: " << result.mdl.get("name");
         LOG(LogDebug) << "ScreenScraperRequest::processGame(): Game ID: " << result.gameID;
 
         pugi::xml_node system {game.child("systeme")};
@@ -431,10 +434,15 @@ void ScreenScraperRequest::processGame(const pugi::xml_document& xmldoc,
                                      .text()
                                      .get()};
 
-        // Translate some HTML character codes to UTF-8 characters.
+        // Translate some HTML character codes to UTF-8 characters for the description.
+        // This does not capture all such characters in the ScreenScraper database but these
+        // are the most common ones.
         if (!description.empty()) {
-            result.mdl.set("desc", Utils::String::replace(description, "&nbsp;", " "));
-            result.mdl.set("desc", Utils::String::replace(description, "&copy;", "©"));
+            description = Utils::String::replace(description, "&nbsp;", " ");
+            description = Utils::String::replace(description, "&quot;", "\"");
+            description = Utils::String::replace(description, "&copy;", "©");
+            description = Utils::String::replace(description, "&#039;", "'");
+            result.mdl.set("desc", description);
         }
 
         // Get the date proper. The API returns multiple 'date' children nodes to the 'dates'