Fix CVE-2022-36033

(cherry picked from commit 9047b512382eba49dbfec8d738af83d2e4ceca52)

CVE-2022-36033.patch

@@ -0,0 +1,99 @@
+From 4ea768d96b3d232e63edef9594766d44597b3882 Mon Sep 17 00:00:00 2001
+From: Jonathan Hedley <jonathan@hedley.net>
+Date: Sun, 21 Aug 2022 14:04:56 +1000
+Subject: [PATCH] Strip control characters from URLs when resolving absolute
+ URLs
+
+---
+ .../java/org/jsoup/internal/StringUtil.java    | 10 +++++++++-
+ .../org/jsoup/internal/StringUtilTest.java     |  9 +++++++++
+ .../java/org/jsoup/safety/CleanerTest.java     | 18 ++++++++++++++++++
+ 3 files changed, 36 insertions(+), 1 deletion(-)
+
+diff --git a/src/main/java/org/jsoup/internal/StringUtil.java b/src/main/java/org/jsoup/internal/StringUtil.java
+index 0835225..608e96d 100644
+--- a/src/main/java/org/jsoup/internal/StringUtil.java
++++ b/src/main/java/org/jsoup/internal/StringUtil.java
+@@ -269,6 +269,7 @@ public final class StringUtil {
+      * @throws MalformedURLException if an error occurred generating the URL
+      */
+     public static URL resolve(URL base, String relUrl) throws MalformedURLException {
++        relUrl = stripControlChars(relUrl);
+         // workaround: java resolves '//path/file + ?foo' to '//path/?foo', not '//path/file?foo' as desired
+         if (relUrl.startsWith("?"))
+             relUrl = base.getPath() + relUrl;
+@@ -287,7 +288,9 @@ public final class StringUtil {
+      * @param relUrl the relative URL to resolve. (If it's already absolute, it will be returned)
+      * @return an absolute URL if one was able to be generated, or the empty string if not
+      */
+-    public static String resolve(final String baseUrl, final String relUrl) {
++    public static String resolve(String baseUrl, String relUrl) {
++        // workaround: java will allow control chars in a path URL and may treat as relative, but Chrome / Firefox will strip and may see as a scheme. Normalize to browser's view.
++        baseUrl = stripControlChars(baseUrl); relUrl = stripControlChars(relUrl);
+         try {
+             URL base;
+             try {
+@@ -306,6 +309,11 @@ public final class StringUtil {
+     }
+     private static final Pattern validUriScheme = Pattern.compile("^[a-zA-Z][a-zA-Z0-9+-.]*:");
+ 
++    private static final Pattern controlChars = Pattern.compile("[\\x00-\\x1f]*"); // matches ascii 0 - 31, to strip from url
++    private static String stripControlChars(final String input) {
++        return controlChars.matcher(input).replaceAll("");
++    }
++
+     private static final ThreadLocal<Stack<StringBuilder>> threadLocalBuilders = new ThreadLocal<Stack<StringBuilder>>() {
+         @Override
+         protected Stack<StringBuilder> initialValue() {
+diff --git a/src/test/java/org/jsoup/internal/StringUtilTest.java b/src/test/java/org/jsoup/internal/StringUtilTest.java
+index 1956084..9ffcec9 100644
+--- a/src/test/java/org/jsoup/internal/StringUtilTest.java
++++ b/src/test/java/org/jsoup/internal/StringUtilTest.java
+@@ -120,6 +120,15 @@ public class StringUtilTest {
+         assertEquals("http://example.com/b/c/g#s/../x", resolve("http://example.com/b/c/d;p?q", "g#s/../x"));
+     }
+ 
++    @Test void stripsControlCharsFromUrls() {
++        // should resovle to an absolute url:
++        assertEquals("foo:bar", resolve("\nhttps://\texample.com/", "\r\nfo\to:ba\br"));
++    }
++
++    @Test void allowsSpaceInUrl() {
++        assertEquals("https://example.com/foo bar/", resolve("HTTPS://example.com/example/", "../foo bar/"));
++    }
++
+     @Test
+     void isAscii() {
+         assertTrue(StringUtil.isAscii(""));
+diff --git a/src/test/java/org/jsoup/safety/CleanerTest.java b/src/test/java/org/jsoup/safety/CleanerTest.java
+index 3338054..0e62f17 100644
+--- a/src/test/java/org/jsoup/safety/CleanerTest.java
++++ b/src/test/java/org/jsoup/safety/CleanerTest.java
+@@ -309,6 +309,24 @@ public class CleanerTest {
+         assertEquals("<a rel=\"nofollow\">Clean</a>", clean);
+     }
+ 
++    @Test void dropsConcealedJavascriptProtocolWhenRelativesLinksEnabled() {
++        Safelist safelist = Safelist.basic().preserveRelativeLinks(true);
++        String html = "<a href=\"&#0013;ja&Tab;va&Tab;script&#0010;:alert(1)\">Link</a>";
++        String clean = Jsoup.clean(html, "https://", safelist);
++        assertEquals("<a rel=\"nofollow\">Link</a>", clean);
++
++        String colon = "<a href=\"ja&Tab;va&Tab;script&colon;alert(1)\">Link</a>";
++        String cleanColon = Jsoup.clean(colon, "https://", safelist);
++        assertEquals("<a rel=\"nofollow\">Link</a>", cleanColon);
++    }
++
++    @Test void dropsConcealedJavascriptProtocolWhenRelativesLinksDisabled() {
++        Safelist safelist = Safelist.basic().preserveRelativeLinks(false);
++        String html = "<a href=\"ja&Tab;vas&#0013;cript:alert(1)\">Link</a>";
++        String clean = Jsoup.clean(html, "https://", safelist);
++        assertEquals("<a rel=\"nofollow\">Link</a>", clean);
++    }
++
+     @Test public void handlesNoHrefAttribute() {
+         String dirty = "<a>One</a> <a href>Two</a>";
+         Safelist relaxedWithAnchor = Safelist.relaxed().addProtocols("a", "href", "#");
+-- 
+2.33.0
+

jsoup.spec

@@ -1,10 +1,12 @@
 Name:           jsoup
 Version:        1.14.2
-Release:        1
+Release:        2
 Summary:        Java HTML Parser
 License:        MIT
 URL:            http://jsoup.org/
 Source0:        https://github.com/jhy/jsoup/archive/refs/tags/jsoup-%{version}.tar.gz
+# https://github.com/jhy/jsoup/commit/4ea768d96b3d232e63edef9594766d44597b3882
+Patch0:         CVE-2022-36033.patch
 BuildArch:      noarch
 
 BuildRequires:  maven-local, mvn(org.apache.felix:maven-bundle-plugin)
@@ -34,6 +36,9 @@ for extracting and manipulating data, using the best of DOM, CSS, and jquery-lik
 %{_javadocdir}/%{name}/*
 
 %changelog
+* Mon Mar 04 2024 yaoxin <yao_xin001@hoperun.com> - 1.14.2-2
+- Fix CVE-2022-36033
+
 * Fri Sep 3 2021 houyingchao <houyingchao@huawei.com> - 1.14.2-1
 - Upgrade to 1.14.2