Reject invalid characters in XML strings (Bug#41094).

* lisp/xml.el (xml-escape-string): Search for invalid characters.
(xml-invalid-character): New error symbol.

* test/lisp/xml-tests.el (xml-print-invalid-cdata): New unit test.

* etc/NEWS: Document new behavior.

diff --git a/etc/NEWS b/etc/NEWS
index 32b59cb76fcb911e1474bd58e26100cd9c655b73..efad273da6c4ff0f8167d6405ffc619ba67a250f 100644 (file)
--- a/etc/NEWS
+++ b/etc/NEWS
@@ -393,6 +393,13 @@ component are now rejected by 'json-read' and friends.  This makes
 them more compliant with the JSON specification and consistent with
 the native JSON parsing functions.
 
+** xml.el
+
+*** XML serialization functions now reject invalid characters.
+Previously 'xml-print' would produce invalid XML when given a string
+with characters that are not valid in XML (see
+https://www.w3.org/TR/xml/#charsets).  Now it rejects such strings.
+
 \f
 * New Modes and Packages in Emacs 28.1
 

diff --git a/lisp/xml.el b/lisp/xml.el
index dc774a202cfc37cbd16c089ba3252baa0d688df9..767cf04284634d1673809d071b5184b6d8210418 100644 (file)
--- a/lisp/xml.el
+++ b/lisp/xml.el
@@ -1023,9 +1023,17 @@ entity references (e.g., replace each & with &).
 XML character data must not contain & or < characters, nor the >
 character under some circumstances.  The XML spec does not impose
 restriction on \" or \\=', but we just substitute for these too
-\(as is permitted by the spec)."
+\(as is permitted by the spec).
+
+If STRING contains characters that are invalid in XML (as defined
+by https://www.w3.org/TR/xml/#charsets), signal an error of type
+`xml-invalid-character'."
   (with-temp-buffer
     (insert string)
+    (goto-char (point-min))
+    (when (re-search-forward
+           "[^\u0009\u000A\u000D\u0020-\uD7FF\uE000-\uFFFD\U00010000-\U0010FFFF]")
+      (signal 'xml-invalid-character (list (char-before) (match-beginning 0))))
     (dolist (substitution '(("&" . "&amp;")
                            ("<" . "&lt;")
                            (">" . "&gt;")
@@ -1036,6 +1044,9 @@ restriction on \" or \\=', but we just substitute for these too
        (replace-match (cdr substitution) t t nil)))
     (buffer-string)))
 
+(define-error 'xml-invalid-character "Invalid XML character"
+  'wrong-type-argument)
+
 (defun xml-debug-print-internal (xml indent-string)
   "Outputs the XML tree in the current buffer.
 The first line is indented with INDENT-STRING."

diff --git a/test/lisp/xml-tests.el b/test/lisp/xml-tests.el
index 57e685cd34793744ad53d5d9c44e1cd3aab1d47c..72c78d00e3ed4bf9b333c7f9304e93966725bfb7 100644 (file)
--- a/test/lisp/xml-tests.el
+++ b/test/lisp/xml-tests.el
@@ -164,6 +164,16 @@ Parser is called with and without 'symbol-qnames argument.")
     (should (equal (cdr xml-parse-test--namespace-attribute-qnames)
                    (xml-parse-region nil nil nil nil 'symbol-qnames)))))
 
+(ert-deftest xml-print-invalid-cdata ()
+  "Check that Bug#41094 is fixed."
+  (with-temp-buffer
+    (should (equal (should-error (xml-print '((foo () "\0")))
+                                 :type 'xml-invalid-character)
+                   '(xml-invalid-character 0 1)))
+    (should (equal (should-error (xml-print '((foo () "\u00FF \xFF")))
+                                 :type 'xml-invalid-character)
+                   '(xml-invalid-character #x3FFFFF 3)))))
+
 ;; Local Variables:
 ;; no-byte-compile: t
 ;; End: