From 8b13f717b4e01dc87c4f7e7a6a139c3525ae7eec Mon Sep 17 00:00:00 2001
From: Abe Jellinek <jellinek@berkeley.edu>
Date: Wed, 13 Mar 2024 12:20:59 -0400
Subject: [PATCH] SAXXMLReader: Handle non-UTF-8 encodings (#3846)

---
 resource/feeds/SAXXMLReader.js      | 10 ++++++++-
 test/tests/data/feedWindows1252.rss | 15 +++++++++++++
 test/tests/feedReaderTest.js        | 35 +++++++++++++++++++++++++++++
 3 files changed, 59 insertions(+), 1 deletion(-)
 create mode 100644 test/tests/data/feedWindows1252.rss

diff --git a/resource/feeds/SAXXMLReader.js b/resource/feeds/SAXXMLReader.js
index b1cf4413eb..eaba91509a 100644
--- a/resource/feeds/SAXXMLReader.js
+++ b/resource/feeds/SAXXMLReader.js
@@ -62,7 +62,15 @@ class SAXXMLReader {
 		if (!response.ok) {
 			throw new Error("Unable to fetch data");
 		}
-		this._data = await response.text();
+		let buf = await response.arrayBuffer();
+		// We should use NetUtil.parseResponseContentType, but we don't have access to it here
+		let charset = response.headers.get("Content-Type")
+			?.match(/charset=([^;]+)/)
+			?.[1];
+		if (!charset) {
+			charset = 'utf-8';
+		}
+		this._data = new TextDecoder(charset).decode(buf);
 		this._parseAndNotify();
 	}
 	
diff --git a/test/tests/data/feedWindows1252.rss b/test/tests/data/feedWindows1252.rss
new file mode 100644
index 0000000000..f4281f19c3
--- /dev/null
+++ b/test/tests/data/feedWindows1252.rss
@@ -0,0 +1,15 @@
+<?xml version="1.0" encoding="windows-1252"?>
+<rss xmlns:dc="http://purl.org/dc/elements/1.1/" version="2.0">
+  <channel>
+    <title>Stortinget: Helse- og omsorgsministeren</title>
+    <link>https://www.stortinget.no</link>
+    <description>https://www.stortinget.no</description>
+    <language>no-NO</language>
+    <item>
+      <title>Skriftlig spørsmål fra Tage Pettersen (H) til helse- og omsorgsministeren. Til behandling</title>
+      <link>https://www.stortinget.no/no/Saker-og-publikasjoner/Sporsmal/Skriftlige-sporsmal-og-svar/Skriftlig-sporsmal/?qid=98117&amp;utm_medium=rss&amp;utm_source=www.stortinget.no&amp;utm_campaign=Helse- og omsorgsministeren</link>
+      <description>Hva vil helseministeren foreta seg på kort sikt for å sikre rekruttering av helsepersonell og at stortingets vedtak om å tilby en tverrfaglig helsekartlegging av barn som flyttes ut av hjemmet etterleves?</description>
+      <dc:date>2024-03-13</dc:date>
+    </item>
+  </channel>
+</rss>
\ No newline at end of file
diff --git a/test/tests/feedReaderTest.js b/test/tests/feedReaderTest.js
index 84cddf6d26..0256e17248 100644
--- a/test/tests/feedReaderTest.js
+++ b/test/tests/feedReaderTest.js
@@ -251,4 +251,39 @@ describe("Zotero.FeedReader", function () {
 			assert.equal(item.enclosedItems[0].url, "https://static01.nyt.com/images/2021/06/16/world/16biden-photos1/16biden-photos1-moth.jpg");
 		});
 	});
+
+	describe("Legacy text encodings", function () {
+		var httpd;
+		var port = 16213;
+		var baseURL = `http://127.0.0.1:${port}/`;
+
+		before(function () {
+			Cu.import("resource://zotero-unit/httpd.js");
+			httpd = new HttpServer();
+			httpd.start(port);
+
+			httpd._handler._mimeMappings.rss = "text/xml; charset=ISO-8859-1";
+
+			httpd.registerPathHandler("/feedWindows1252.rss", {
+				handle(request, response) {
+					response.setStatusLine(null, 200, 'OK');
+					let file = getTestDataDirectory();
+					file.append("feedWindows1252.rss");
+					httpd._handler._writeFileResponse(request, file, response, 0, file.fileSize);
+				}
+			});
+		});
+		
+		after(async function () {
+			await new Promise(resolve => httpd.stop(resolve));
+		});
+		
+		it("should handle an ISO-8859-1 (windows-1252) feed", async function () {
+			let fr = new Zotero.FeedReader(baseURL + "feedWindows1252.rss");
+			await fr.process();
+			let itemIterator = new fr.ItemIterator();
+			let item = await itemIterator.next().value;
+			assert.equal(item.title, "Skriftlig spørsmål fra Tage Pettersen (H) til helse- og omsorgsministeren. Til behandling");
+		});
+	});
 })