From c7785d8f54bce10dc36eb2e91fc7778ef9eadfd3 Mon Sep 17 00:00:00 2001
From: Alexander Kryklia <kryklia@gmail.com>
Date: Mon, 24 Feb 2014 15:19:31 +0200
Subject: [PATCH] Allow HTML in grader messages.

---
 common/lib/capa/capa/inputtypes.py            | 14 ++++-
 common/lib/capa/capa/responsetypes.py         | 16 ++++--
 common/lib/capa/capa/tests/test_inputtypes.py | 39 ++++++++++++++
 .../lib/capa/capa/tests/test_responsetypes.py | 53 +++++++++++++++++++
 4 files changed, 116 insertions(+), 6 deletions(-)

diff --git a/common/lib/capa/capa/inputtypes.py b/common/lib/capa/capa/inputtypes.py
index 790e93679d..d638b41350 100644
--- a/common/lib/capa/capa/inputtypes.py
+++ b/common/lib/capa/capa/inputtypes.py
@@ -46,6 +46,7 @@ import re
 import shlex  # for splitting quoted strings
 import sys
 import pyparsing
+import html5lib
 
 from .registry import TagRegistry
 from chem import chemcalc
@@ -286,7 +287,18 @@ class InputTypeBase(object):
         context = self._get_render_context()
 
         html = self.capa_system.render_template(self.template, context)
-        return etree.XML(html)
+
+        try:
+            output = etree.XML(html)
+        except etree.XMLSyntaxError as ex:
+            # If `html` contains attrs with no values, like `controls` in <audio controls src='smth'/>,
+            # XML parser will raise exception, so wee fallback to html5parser, which will set empty "" values for such attrs.
+            try:
+                output = html5lib.parseFragment(html, treebuilder='lxml', namespaceHTMLElements=False)[0]
+            except IndexError:
+                raise ex
+
+        return output
 
     def get_user_visible_answer(self, internal_answer):
         """
diff --git a/common/lib/capa/capa/responsetypes.py b/common/lib/capa/capa/responsetypes.py
index 4b06369599..01a1401091 100644
--- a/common/lib/capa/capa/responsetypes.py
+++ b/common/lib/capa/capa/responsetypes.py
@@ -14,6 +14,7 @@ import cgi
 import inspect
 import json
 import logging
+import html5lib
 import numbers
 import numpy
 import os
@@ -1761,17 +1762,22 @@ class CodeResponse(LoncapaResponse):
                           " tags: 'correct', 'score', 'msg'")
                 return fail
 
-        # Next, we need to check that the contents of the external grader message
-        #   is safe for the LMS.
+        # Next, we need to check that the contents of the external grader message is safe for the LMS.
         # 1) Make sure that the message is valid XML (proper opening/closing tags)
-        # 2) TODO: Is the message actually HTML?
+        # 2) If it is not valid XML, make sure it is valid HTML. Note: html5lib parser will try to repair any broken HTML
+        # For example: <aaa></bbb> will become <aaa/>.
         msg = score_result['msg']
+
         try:
             etree.fromstring(msg)
         except etree.XMLSyntaxError as _err:
-            log.error("Unable to parse external grader message as valid"
+            # If `html` contains attrs with no values, like `controls` in <audio controls src='smth'/>,
+            # XML parser will raise exception, so wee fallback to html5parser, which will set empty "" values for such attrs.
+            parsed = html5lib.parseFragment(msg, treebuilder='lxml', namespaceHTMLElements=False)
+            if not parsed:
+                log.error("Unable to parse external grader message as valid"
                       " XML: score_msg['msg']=%s", msg)
-            return fail
+                return fail
 
         return (True, score_result['correct'], score_result['score'], msg)
 
diff --git a/common/lib/capa/capa/tests/test_inputtypes.py b/common/lib/capa/capa/tests/test_inputtypes.py
index 866fe75f6e..c52b9763f9 100644
--- a/common/lib/capa/capa/tests/test_inputtypes.py
+++ b/common/lib/capa/capa/tests/test_inputtypes.py
@@ -20,6 +20,7 @@ TODO:
 import json
 from lxml import etree
 import unittest
+import textwrap
 import xml.sax.saxutils as saxutils
 
 from . import test_capa_system
@@ -583,6 +584,44 @@ class MatlabTest(unittest.TestCase):
         self.assertEqual(input_state['queuestate'], 'queued')
         self.assertFalse('queue_msg' in input_state)
 
+    def test_get_html(self):
+        # usual output
+        output = self.the_input.get_html()
+        self.assertEqual(
+            etree.tostring(output),
+            """<div>{\'status\': \'queued\', \'button_enabled\': True, \'rows\': \'10\', \'queue_len\': \'3\'\
+, \'mode\': \'\', \'cols\': \'80\', \'STATIC_URL\': \'/dummy-static/\', \'linenumbers\': \'true\'\
+, \'queue_msg\': \'\', \'value\': \'print "good evening"\', \'msg\': u\'Submitted\
+. As soon as a response is returned, this message will be replaced by that feedback.\', \'hidden\': \'\'\
+, \'id\': \'prob_1_2\', \'tabsize\': 4}</div>"""
+        )
+
+        # test html, that is correct HTML5 html, but is not parsable by XML parser.
+        old_render_template = self.the_input.capa_system.render_template
+        self.the_input.capa_system.render_template = lambda *args: textwrap.dedent("""
+                <div class='matlabResponse'><div id='mwAudioPlaceHolder'>
+                <audio controls autobuffer autoplay src='data:audio/wav;base64='>Audio is not supported on this browser.</audio>
+                <div>Right click <a href=https://endpoint.mss-mathworks.com/media/filename.wav>here</a> and click \"Save As\" to download the file</div></div>
+                <div style='white-space:pre' class='commandWindowOutput'></div><ul></ul></div>
+            """).replace('\n', '')
+        output = self.the_input.get_html()
+        self.assertEqual(
+            etree.tostring(output),
+            textwrap.dedent("""
+            <div class='matlabResponse'><div id='mwAudioPlaceHolder'>
+            <audio src='data:audio/wav;base64=' autobuffer="" controls="" autoplay="">Audio is not supported on this browser.</audio>
+            <div>Right click <a href="https://endpoint.mss-mathworks.com/media/filename.wav">here</a> and click \"Save As\" to download the file</div></div>
+            <div style='white-space:pre' class='commandWindowOutput'/><ul/></div>
+            """).replace('\n', '').replace('\'', '\"')
+        )
+
+        # check that exception is raised during parsing for html.
+        self.the_input.capa_system.render_template = lambda *args: "<aaa"
+        with self.assertRaises(etree.XMLSyntaxError):
+            self.the_input.get_html()
+
+        self.the_input.capa_system.render_template = old_render_template
+
 
 class SchematicTest(unittest.TestCase):
     '''
diff --git a/common/lib/capa/capa/tests/test_responsetypes.py b/common/lib/capa/capa/tests/test_responsetypes.py
index 3a3a5c3451..770f6959ee 100644
--- a/common/lib/capa/capa/tests/test_responsetypes.py
+++ b/common/lib/capa/capa/tests/test_responsetypes.py
@@ -998,6 +998,59 @@ class CodeResponseTest(ResponseTest):
             self.assertEquals(answers_converted['1_3_1'], ['answer1', 'answer2', 'answer3'])
             self.assertEquals(answers_converted['1_4_1'], [fp.name, fp.name])
 
+    def test_parse_score_msg_of_responder(self):
+        """
+        Test whether LoncapaProblem._parse_score_msg correcly parses valid HTML5 html.
+        """
+        valid_grader_msgs = [
+            u'<span>MESSAGE</span>',  # Valid XML
+            textwrap.dedent("""
+                <div class='matlabResponse'><div id='mwAudioPlaceHolder'>
+                <audio controls autobuffer autoplay src='data:audio/wav;base64='>Audio is not supported on this browser.</audio>
+                <div>Right click <a href=https://endpoint.mss-mathworks.com/media/filename.wav>here</a> and click \"Save As\" to download the file</div></div>
+                <div style='white-space:pre' class='commandWindowOutput'></div><ul></ul></div>
+            """).replace('\n', ''),  # Valid HTML5 real case Matlab response, invalid XML
+            '<aaa></bbb>'  # Invalid XML, but will be parsed by html5lib to <aaa/>
+        ]
+
+        invalid_grader_msgs = [
+            '<audio',  # invalid XML and HTML5
+        ]
+
+        answer_ids = sorted(self.problem.get_question_answers())
+
+        # CodeResponse requires internal CorrectMap state. Build it now in the queued state
+        old_cmap = CorrectMap()
+        for i, answer_id in enumerate(answer_ids):
+            queuekey = 1000 + i
+            queuestate = CodeResponseTest.make_queuestate(queuekey, datetime.now(UTC))
+            old_cmap.update(CorrectMap(answer_id=answer_ids[i], queuestate=queuestate))
+
+        for grader_msg in valid_grader_msgs:
+            correct_score_msg = json.dumps({'correct': True, 'score': 1, 'msg': grader_msg})
+            incorrect_score_msg = json.dumps({'correct': False, 'score': 0, 'msg': grader_msg})
+            xserver_msgs = {'correct': correct_score_msg, 'incorrect': incorrect_score_msg, }
+
+            for i, answer_id in enumerate(answer_ids):
+                self.problem.correct_map = CorrectMap()
+                self.problem.correct_map.update(old_cmap)
+                output = self.problem.update_score(xserver_msgs['correct'], queuekey=1000 + i)
+                self.assertEquals(output[answer_id]['msg'], grader_msg)
+
+        for grader_msg in invalid_grader_msgs:
+            correct_score_msg = json.dumps({'correct': True, 'score': 1, 'msg': grader_msg})
+            incorrect_score_msg = json.dumps({'correct': False, 'score': 0, 'msg': grader_msg})
+            xserver_msgs = {'correct': correct_score_msg, 'incorrect': incorrect_score_msg, }
+
+            for i, answer_id in enumerate(answer_ids):
+                self.problem.correct_map = CorrectMap()
+                self.problem.correct_map.update(old_cmap)
+
+                output = self.problem.update_score(xserver_msgs['correct'], queuekey=1000 + i)
+                self.assertEquals(output[answer_id]['msg'], u'Invalid grader reply. Please contact the course staff.')
+
+
+
 
 class ChoiceResponseTest(ResponseTest):
     from capa.tests.response_xml_factory import ChoiceResponseXMLFactory