Provide feature test for issue #510.

issue.features/issue0510.feature

@@ -0,0 +1,51 @@
+@issue
+@junit
+@wip
+Feature: Issue #510 -- JUnit XML output is not well-formed (in some cases)
+
+  . Special control characters in JUnit stdout/stderr sections
+  . are directly written to CDATA XML sections.
+  .
+  . According to the XML charset specification only the following unicode
+  . codepoints (characters) are allowed in a CDATA section:
+  .
+  .   Char ::=  #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]
+  .   /* any Unicode character, excluding the surrogate blocks, FFFE, and FFFF. */
+  .
+  . [XML-charsets] "The normative reference is XML 1.0 (Fifth Edition),
+  .     section 2.2, https://www.w3.org/TR/REC-xml/#charsets
+
+
+  @use.with_xmllint=yes
+  @xfail
+  Scenario:
+    Given a new working directory
+    And a file named "features/steps/special_char_steps.py" with:
+      """
+      # -*- coding: UTF-8 -*-
+      from __future__ import print_function
+      from behave import step
+
+      @step(u'we print ^D')
+      def step_print_special_char_control_d(context):
+          print(u"\004")
+      """
+    And a file named "features/special_char.feature" with:
+      """
+      Feature: An illegal char
+        Scenario: Control-D
+          When we print ^D
+      """
+    When I run "behave --junit features/special_char.feature"
+    Then it should pass with:
+      """
+      1 scenario passed, 0 failed, 0 skipped
+      """
+    When I run "xmllint reports/TESTS-special_char.xml"
+    Then it should pass
+    And the command output should not contain "parser error"
+    And the command output should not contain:
+      """
+      reports/TESTS-special_char.xml:12: parser error : PCDATA invalid Char value 4
+      """
+    And note that "xmllint reports additional correlated errors"