Backport a patch from upstream to fix handling unknown tags in AppStream metadata

Related: rhbz#1991688
Signed-off-by: Mohan Boddu <mboddu@redhat.com>

.libappstream-glib.metadata

@@ -0,0 +1 @@
+f01aa02454db76f5907b1d4e65d1f07944df28a9 appstream-glib-0.7.18.tar.xz

0001-Improve-handling-of-em-and-code-tags.patch

@@ -0,0 +1,243 @@
+diff -Nru appstream-glib-0.7.18/libappstream-glib/as-node.c appstream-glib-0.7.18/libappstream-glib/as-node.c
+--- appstream-glib-0.7.18/libappstream-glib/as-node.c	2020-09-07 11:20:43.894573000 +0100
++++ appstream-glib-0.7.18/libappstream-glib/as-node.c	2023-06-07 20:58:11.000000000 +0100
+@@ -555,6 +555,8 @@
+ 	AsNode			*current;
+ 	AsNodeFromXmlFlags	 flags;
+ 	const gchar * const	*locales;
++	guint8			 is_em_text:1;
++	guint8			 is_code_text:1;
+ } AsNodeToXmlHelper;
+ 
+ /**
+@@ -604,6 +606,16 @@
+ 	AsNode *current;
+ 	guint i;
+ 
++	/* do not create a child node for em and code tags */
++	if (g_strcmp0 (element_name, "em") == 0) {
++		helper->is_em_text = 1;
++		return;
++	}
++	if (g_strcmp0 (element_name, "code") == 0) {
++		helper->is_code_text = 1;
++		return;
++	}
++
+ 	/* check if we should ignore the locale */
+ 	data = g_slice_new0 (AsNodeData);
+ 
+@@ -662,6 +674,53 @@
+ 			GError             **error)
+ {
+ 	AsNodeToXmlHelper *helper = (AsNodeToXmlHelper *) user_data;
++	AsNodeData *data = helper->current->data;
++
++	/* do not create a child node for em and code tags */
++	if (g_strcmp0 (element_name, "em") == 0) {
++		helper->is_em_text = 0;
++		return;
++	}
++	if (g_strcmp0 (element_name, "code") == 0) {
++		helper->is_code_text = 0;
++		return;
++	}
++
++	if (data->cdata != NULL) {
++		/* split up into lines and add each with spaces stripped */
++		if ((helper->flags & AS_NODE_FROM_XML_FLAG_LITERAL_TEXT) == 0) {
++			AsRefString *cdata = data->cdata;
++			data->cdata = as_node_reflow_text (cdata, strlen (cdata));
++			as_ref_string_unref (cdata);
++		}
++
++		/* intern commonly duplicated tag values and save a bit of memory */
++		if (data->is_tag_valid) {
++			AsNode *root = g_node_get_root (helper->current);
++			switch (data->tag) {
++			case AS_TAG_CATEGORY:
++			case AS_TAG_COMPULSORY_FOR_DESKTOP:
++			case AS_TAG_CONTENT_ATTRIBUTE:
++			case AS_TAG_DEVELOPER_NAME:
++			case AS_TAG_EXTENDS:
++			case AS_TAG_ICON:
++			case AS_TAG_ID:
++			case AS_TAG_KUDO:
++			case AS_TAG_LANG:
++			case AS_TAG_METADATA_LICENSE:
++			case AS_TAG_MIMETYPE:
++			case AS_TAG_PROJECT_GROUP:
++			case AS_TAG_PROJECT_LICENSE:
++			case AS_TAG_SOURCE_PKGNAME:
++			case AS_TAG_URL:
++				as_node_cdata_to_intern (root, data);
++				break;
++			default:
++				break;
++			}
++		}
++	}
++
+ 	helper->current = helper->current->parent;
+ }
+ 
+@@ -693,8 +752,9 @@
+ 	if (i >= text_len)
+ 		return;
+ 
+-	/* split up into lines and add each with spaces stripped */
+-	if (data->cdata != NULL) {
++	if (data->cdata != NULL &&
++	    g_strcmp0 (as_tag_data_get_name (data), "p") != 0 &&
++	    g_strcmp0 (as_tag_data_get_name (data), "li") != 0) {
+ 		g_set_error (error,
+ 			     AS_NODE_ERROR,
+ 			     AS_NODE_ERROR_INVALID_MARKUP,
+@@ -703,37 +763,33 @@
+ 			     data->cdata, text);
+ 		return;
+ 	}
+-	if ((helper->flags & AS_NODE_FROM_XML_FLAG_LITERAL_TEXT) > 0) {
+-		data->cdata = as_ref_string_new_with_length (text, text_len + 1);
+-	} else {
+-		data->cdata = as_node_reflow_text (text, (gssize) text_len);
+-	}
+ 
+-	/* intern commonly duplicated tag values and save a bit of memory */
+-	if (data->is_tag_valid && data->cdata != NULL) {
+-		AsNode *root = g_node_get_root (helper->current);
+-		switch (data->tag) {
+-		case AS_TAG_CATEGORY:
+-		case AS_TAG_COMPULSORY_FOR_DESKTOP:
+-		case AS_TAG_CONTENT_ATTRIBUTE:
+-		case AS_TAG_DEVELOPER_NAME:
+-		case AS_TAG_EXTENDS:
+-		case AS_TAG_ICON:
+-		case AS_TAG_ID:
+-		case AS_TAG_KUDO:
+-		case AS_TAG_LANG:
+-		case AS_TAG_METADATA_LICENSE:
+-		case AS_TAG_MIMETYPE:
+-		case AS_TAG_PROJECT_GROUP:
+-		case AS_TAG_PROJECT_LICENSE:
+-		case AS_TAG_SOURCE_PKGNAME:
+-		case AS_TAG_URL:
+-			as_node_cdata_to_intern (root, data);
+-			break;
+-		default:
+-			break;
++	/* support em and code tags */
++	if (helper->is_em_text || helper->is_code_text || data->cdata != NULL) {
++		g_autoptr(GString) str = g_string_new (NULL);
++
++		if (data->cdata != NULL) {
++			g_string_append (str, data->cdata);
++			as_ref_string_unref (data->cdata);
+ 		}
++
++		if (helper->is_em_text)
++			g_string_append (str, "<em>");
++		if (helper->is_code_text)
++			g_string_append (str, "<code>");
++
++		g_string_append_len (str, text, text_len);
++
++		if (helper->is_code_text)
++			g_string_append (str, "</code>");
++		if (helper->is_em_text)
++			g_string_append (str, "</em>");
++
++		data->cdata = as_ref_string_new_with_length (str->str, str->len);
++		return;
+ 	}
++
++	data->cdata = as_ref_string_new_with_length (text, text_len);
+ }
+ 
+ static void
+@@ -790,7 +846,7 @@
+ 			   AsNodeFromXmlFlags flags,
+ 			   GError **error)
+ {
+-	AsNodeToXmlHelper helper;
++	AsNodeToXmlHelper helper = {0};
+ 	AsNode *root = NULL;
+ 	gboolean ret;
+ 	g_autoptr(GError) error_local = NULL;
+@@ -927,7 +983,7 @@
+ 		   GCancellable *cancellable,
+ 		   GError **error)
+ {
+-	AsNodeToXmlHelper helper;
++	AsNodeToXmlHelper helper = {0};
+ 	GError *error_local = NULL;
+ 	AsNode *root = NULL;
+ 	const gchar *content_type = NULL;
+diff -Nru appstream-glib-0.7.18/libappstream-glib/as-self-test.c appstream-glib-0.7.18/libappstream-glib/as-self-test.c
+--- appstream-glib-0.7.18/libappstream-glib/as-self-test.c	2020-09-07 11:20:43.896573000 +0100
++++ appstream-glib-0.7.18/libappstream-glib/as-self-test.c	2023-06-07 20:58:11.000000000 +0100
+@@ -2861,6 +2861,20 @@
+ 			     "<!-- this documents bar -->"
+ 			     "<bar key=\"value\">baz</bar>"
+ 			     "</foo>";
++	const gchar *valid_em_code = "<description>"
++			     "<p>"
++			     "It now also supports <em>em</em> and <code>code</code> tags."
++			     "</p>"
++			     "</description>";
++	const gchar *valid_em_code_2 = "<description>"
++			     "<p><em>Emphasis</em> at the start of the paragraph</p>"
++			     "</description>";
++	const gchar *valid_em_code_empty = "<description>"
++			     "<p><em></em></p>"
++			     "</description>";
++	const gchar *valid_em_code_empty_2 = "<description>"
++			     "<p>empty <em></em> emphasis</p>"
++			     "</description>";
+ 	GError *error = NULL;
+ 	AsNode *n2;
+ 	AsNode *root;
+@@ -2924,6 +2938,43 @@
+ 	g_string_free (xml, TRUE);
+ 	as_node_unref (root);
+ 
++	/* support em and code tags */
++	root = as_node_from_xml (valid_em_code, 0, &error);
++	g_assert_no_error (error);
++	g_assert (root != NULL);
++
++	n2 = as_node_find (root, "description/p");
++	g_assert (n2 != NULL);
++	g_assert_cmpstr (as_node_get_data (n2), ==, "It now also supports <em>em</em> and <code>code</code> tags.");
++	as_node_unref (root);
++
++	root = as_node_from_xml (valid_em_code_2, 0, &error);
++	g_assert_no_error (error);
++	g_assert (root != NULL);
++
++	n2 = as_node_find (root, "description/p");
++	g_assert (n2 != NULL);
++	g_assert_cmpstr (as_node_get_data (n2), ==, "<em>Emphasis</em> at the start of the paragraph");
++	as_node_unref (root);
++
++	root = as_node_from_xml (valid_em_code_empty, 0, &error);
++	g_assert_no_error (error);
++	g_assert (root != NULL);
++
++	n2 = as_node_find (root, "description/p");
++	g_assert (n2 != NULL);
++	g_assert_cmpstr (as_node_get_data (n2), ==, NULL);
++	as_node_unref (root);
++
++	root = as_node_from_xml (valid_em_code_empty_2, 0, &error);
++	g_assert_no_error (error);
++	g_assert (root != NULL);
++
++	n2 = as_node_find (root, "description/p");
++	g_assert (n2 != NULL);
++	g_assert_cmpstr (as_node_get_data (n2), ==, "empty  emphasis");
++	as_node_unref (root);
++
+ 	/* keep comments */
+ 	root = as_node_from_xml (valid,
+ 				 AS_NODE_FROM_XML_FLAG_KEEP_COMMENTS,

libappstream-glib.spec

@@ -6,11 +6,14 @@
 Summary:   Library for AppStream metadata
 Name:      libappstream-glib
 Version:   0.7.18
-Release:   3%{?dist}
+Release:   5%{?dist}
 License:   LGPLv2+
 URL:       http://people.freedesktop.org/~hughsient/appstream-glib/
 Source0:   http://people.freedesktop.org/~hughsient/appstream-glib/releases/appstream-glib-%{version}.tar.xz
 
+# backported from upstream
+Patch0:    0001-Improve-handling-of-em-and-code-tags.patch
+
 BuildRequires: glib2-devel >= %{glib2_version}
 BuildRequires: docbook-utils
 BuildRequires: gtk-doc
@@ -131,6 +134,14 @@ from a directory of packages.
 %{_mandir}/man1/appstream-builder.1.gz
 
 %changelog
+* Fri Apr 26 2024 Richard Hughes <richard@hughsie.com> 0.7.18-5
+- Backport a patch from upstream to fix handling unknown tags in AppStream metadata
+- Resolves: #RHEL-28856
+
+* Mon Aug 09 2021 Mohan Boddu <mboddu@redhat.com> - 0.7.18-4
+- Rebuilt for IMA sigs, glibc 2.34, aarch64 flags
+  Related: rhbz#1991688
+
 * Fri Apr 16 2021 Mohan Boddu <mboddu@redhat.com> - 0.7.18-3
 - Rebuilt for RHEL 9 BETA on Apr 15th 2021. Related: rhbz#1947937