Support em/code tags

Some appstream-data packages add <em> and </em> or <code> and </code> to the files. Not all package manager can handle that. An example would be pamac from Manjaro

2 files changed, 52 insertions, 0 deletions

diff --git a/libappstream-glib/as-node.c b/libappstream-glib/as-node.c
index aafb16a..b4159ea 100644
--- a/libappstream-glib/as-node.c
+++ b/libappstream-glib/as-node.c
@@ -555,6 +555,8 @@ typedef struct {
 	AsNode			*current;
 	AsNodeFromXmlFlags	 flags;
 	const gchar * const	*locales;
+	guint8			 is_em_text;
+	guint8			 is_code_text;
 } AsNodeToXmlHelper;
 
 /**
@@ -604,6 +606,16 @@ as_node_start_element_cb (GMarkupParseContext *context,
 	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 +674,16 @@ as_node_end_element_cb (GMarkupParseContext *context,
 			GError             **error)
 {
 	AsNodeToXmlHelper *helper = (AsNodeToXmlHelper *) user_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;
+	}
 	helper->current = helper->current->parent;
 }
 
@@ -695,6 +717,20 @@ as_node_text_cb (GMarkupParseContext *context,
 
 	/* split up into lines and add each with spaces stripped */
 	if (data->cdata != NULL) {
+		/* support em and code tags */
+		if (g_strcmp0 (as_tag_data_get_name (data), "p") == 0 ||
+			g_strcmp0 (as_tag_data_get_name (data), "li") == 0) {
+			g_autoptr(GString) str = g_string_new (data->cdata);
+			as_ref_string_unref (data->cdata);
+			if (helper->is_em_text)
+				g_string_append_printf (str, "<em>%s</em>", text);
+			else if (helper->is_code_text)
+				g_string_append_printf (str, "<code>%s</code>", text);
+			else
+				g_string_append (str, text);
+			data->cdata = as_ref_string_new_with_length (str->str, str->len);
+			return;
+		}
 		g_set_error (error,
 			     AS_NODE_ERROR,
 			     AS_NODE_ERROR_INVALID_MARKUP,
diff --git a/libappstream-glib/as-self-test.c b/libappstream-glib/as-self-test.c
index 2f0820f..eafb15e 100644
--- a/libappstream-glib/as-self-test.c
+++ b/libappstream-glib/as-self-test.c
@@ -2865,6 +2865,11 @@ as_test_node_xml_func (void)
 			     "<!-- 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>";
 	GError *error = NULL;
 	AsNode *n2;
 	AsNode *root;
@@ -2928,6 +2933,17 @@ as_test_node_xml_func (void)
 	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);
+	printf ("<%s>\n", as_node_get_data (n2));
+	g_assert_cmpstr (as_node_get_data (n2), ==, "It now also supports<em>em</em> and <code>code</code> tags.");
+	as_node_unref (root);
+
 	/* keep comments */
 	root = as_node_from_xml (valid,
 				 AS_NODE_FROM_XML_FLAG_KEEP_COMMENTS,