Commit 8e8ac35d by kimth

restored responsetypes.py from master

parent 07de2d99
#
# File: courseware/capa/responsetypes.py
#
'''
Problem response evaluation. Handles checking of student responses, of a variety of types.
Used by capa_problem.py
'''
# standard library imports
import hashlib
import inspect
import json
import logging
import numbers
import numpy
import random
import re
import requests
import time
import traceback
import abc
# specific library imports
from calc import evaluator, UndefinedVariable
from correctmap import CorrectMap
from util import *
from lxml import etree
from lxml.html.soupparser import fromstring as fromstring_bs # uses Beautiful Soup!!! FIXME?
log = logging.getLogger('mitx.' + __name__)
#-----------------------------------------------------------------------------
# Exceptions
class LoncapaProblemError(Exception):
'''
Error in specification of a problem
'''
pass
class ResponseError(Exception):
'''
Error for failure in processing a response
'''
pass
class StudentInputError(Exception):
pass
#-----------------------------------------------------------------------------
#
# Main base class for CAPA responsetypes
class LoncapaResponse(object):
'''
Base class for CAPA responsetypes. Each response type (ie a capa question,
which is part of a capa problem) is represented as a subclass,
which should provide the following methods:
- get_score : evaluate the given student answers, and return a CorrectMap
- get_answers : provide a dict of the expected answers for this problem
Each subclass must also define the following attributes:
- response_tag : xhtml tag identifying this response (used in auto-registering)
In addition, these methods are optional:
- get_max_score : if defined, this is called to obtain the maximum score possible for this question
- setup_response : find and note the answer input field IDs for the response; called by __init__
- check_hint_condition : check to see if the student's answers satisfy a particular condition for a hint to be displayed
- render_html : render this Response as HTML (must return XHTML compliant string)
- __unicode__ : unicode representation of this Response
Each response type may also specify the following attributes:
- max_inputfields : (int) maximum number of answer input fields (checked in __init__ if not None)
- allowed_inputfields : list of allowed input fields (each a string) for this Response
- required_attributes : list of required attributes (each a string) on the main response XML stanza
- hint_tag : xhtml tag identifying hint associated with this response inside hintgroup
'''
__metaclass__ = abc.ABCMeta # abc = Abstract Base Class
response_tag = None
hint_tag = None
max_inputfields = None
allowed_inputfields = []
required_attributes = []
def __init__(self, xml, inputfields, context, system=None):
'''
Init is passed the following arguments:
- xml : ElementTree of this Response
- inputfields : ordered list of ElementTrees for each input entry field in this Response
- context : script processor context
- system : ModuleSystem instance which provides OS, rendering, and user context
'''
self.xml = xml
self.inputfields = inputfields
self.context = context
self.system = system
for abox in inputfields:
if abox.tag not in self.allowed_inputfields:
msg = "%s: cannot have input field %s" % (unicode(self), abox.tag)
msg += "\nSee XML source line %s" % getattr(xml, 'sourceline', '<unavailable>')
raise LoncapaProblemError(msg)
if self.max_inputfields and len(inputfields) > self.max_inputfields:
msg = "%s: cannot have more than %s input fields" % (unicode(self), self.max_inputfields)
msg += "\nSee XML source line %s" % getattr(xml, 'sourceline', '<unavailable>')
raise LoncapaProblemError(msg)
for prop in self.required_attributes:
if not xml.get(prop):
msg = "Error in problem specification: %s missing required attribute %s" % (unicode(self), prop)
msg += "\nSee XML source line %s" % getattr(xml, 'sourceline', '<unavailable>')
raise LoncapaProblemError(msg)
self.answer_ids = [x.get('id') for x in self.inputfields] # ordered list of answer_id values for this response
if self.max_inputfields == 1:
self.answer_id = self.answer_ids[0] # for convenience
self.default_answer_map = {} # dict for default answer map (provided in input elements)
for entry in self.inputfields:
answer = entry.get('correct_answer')
if answer:
self.default_answer_map[entry.get('id')] = contextualize_text(answer, self.context)
if hasattr(self, 'setup_response'):
self.setup_response()
def render_html(self, renderer):
'''
Return XHTML Element tree representation of this Response.
Arguments:
- renderer : procedure which produces HTML given an ElementTree
'''
tree = etree.Element('span') # render ourself as a <span> + our content
for item in self.xml:
item_xhtml = renderer(item) # call provided procedure to do the rendering
if item_xhtml is not None: tree.append(item_xhtml)
tree.tail = self.xml.tail
return tree
def evaluate_answers(self, student_answers, old_cmap):
'''
Called by capa_problem.LoncapaProblem to evaluate student answers, and to
generate hints (if any).
Returns the new CorrectMap, with (correctness,msg,hint,hintmode) for each answer_id.
'''
new_cmap = self.get_score(student_answers)
self.get_hints(student_answers, new_cmap, old_cmap)
# log.debug('new_cmap = %s' % new_cmap)
return new_cmap
def get_hints(self, student_answers, new_cmap, old_cmap):
'''
Generate adaptive hints for this problem based on student answers, the old CorrectMap,
and the new CorrectMap produced by get_score.
Does not return anything.
Modifies new_cmap, by adding hints to answer_id entries as appropriate.
'''
hintgroup = self.xml.find('hintgroup')
if hintgroup is None: return
# hint specified by function?
hintfn = hintgroup.get('hintfn')
if hintfn:
'''
Hint is determined by a function defined in the <script> context; evaluate that function to obtain
list of hint, hintmode for each answer_id.
The function should take arguments (answer_ids, student_answers, new_cmap, old_cmap)
and it should modify new_cmap as appropriate.
We may extend this in the future to add another argument which provides a callback procedure
to a social hint generation system.
'''
if not hintfn in self.context:
msg = 'missing specified hint function %s in script context' % hintfn
msg += "\nSee XML source line %s" % getattr(self.xml, 'sourceline', '<unavailable>')
raise LoncapaProblemError(msg)
try:
self.context[hintfn](self.answer_ids, student_answers, new_cmap, old_cmap)
except Exception as err:
msg = 'Error %s in evaluating hint function %s' % (err, hintfn)
msg += "\nSee XML source line %s" % getattr(self.xml, 'sourceline', '<unavailable>')
raise ResponseError(msg)
return
# hint specified by conditions and text dependent on conditions (a-la Loncapa design)
# see http://help.loncapa.org/cgi-bin/fom?file=291
#
# Example:
#
# <formularesponse samples="x@-5:5#11" id="11" answer="$answer">
# <textline size="25" />
# <hintgroup>
# <formulahint samples="x@-5:5#11" answer="$wrongans" name="inversegrad"></formulahint>
# <hintpart on="inversegrad">
# <text>You have inverted the slope in the question. The slope is
# (y2-y1)/(x2 - x1) you have the slope as (x2-x1)/(y2-y1).</text>
# </hintpart>
# </hintgroup>
# </formularesponse>
if self.hint_tag is not None and hintgroup.find(self.hint_tag) is not None and hasattr(self, 'check_hint_condition'):
rephints = hintgroup.findall(self.hint_tag)
hints_to_show = self.check_hint_condition(rephints, student_answers)
hintmode = hintgroup.get('mode', 'always') # can be 'on_request' or 'always' (default)
for hintpart in hintgroup.findall('hintpart'):
if hintpart.get('on') in hints_to_show:
hint_text = hintpart.find('text').text
aid = self.answer_ids[-1] # make the hint appear after the last answer box in this response
new_cmap.set_hint_and_mode(aid, hint_text, hintmode)
log.debug('after hint: new_cmap = %s' % new_cmap)
@abc.abstractmethod
def get_score(self, student_answers):
'''
Return a CorrectMap for the answers expected vs given. This includes
(correctness, npoints, msg) for each answer_id.
Arguments:
- student_answers : dict of (answer_id,answer) where answer = student input (string)
- old_cmap : previous CorrectMap (may be empty); useful for analyzing or recording history of responses
'''
pass
@abc.abstractmethod
def get_answers(self):
'''
Return a dict of (answer_id,answer_text) for each answer for this question.
'''
pass
def check_hint_condition(self, hxml_set, student_answers):
'''
Return a list of hints to show.
- hxml_set : list of Element trees, each specifying a condition to be satisfied for a named hint condition
- student_answers : dict of student answers
Returns a list of names of hint conditions which were satisfied. Those are used to determine which hints are displayed.
'''
pass
def setup_response(self):
pass
def __unicode__(self):
return u'LoncapaProblem Response %s' % self.xml.tag
#-----------------------------------------------------------------------------
class ChoiceResponse(LoncapaResponse):
'''
This Response type is used when the student chooses from a discrete set of
choices. Currently, to be marked correct, all "correct" choices must be
supplied by the student, and no extraneous choices may be included.
This response type allows for two inputtypes: radiogroups and checkbox
groups. radiogroups are used when the student should select a single answer,
and checkbox groups are used when the student may supply 0+ answers.
Note: it is suggested to include a "None of the above" choice when no
answer is correct for a checkboxgroup inputtype; this ensures that a student
must actively mark something to get credit.
If two choices are marked as correct with a radiogroup, the student will
have no way to get the answer right.
TODO: Allow for marking choices as 'optional' and 'required', which would
not penalize a student for including optional answers and would also allow
for questions in which the student can supply one out of a set of correct
answers.This would also allow for survey-style questions in which all
answers are correct.
Example:
<choiceresponse>
<radiogroup>
<choice correct="false">
<text>This is a wrong answer.</text>
</choice>
<choice correct="true">
<text>This is the right answer.</text>
</choice>
<choice correct="false">
<text>This is another wrong answer.</text>
</choice>
</radiogroup>
</choiceresponse>
In the above example, radiogroup can be replaced with checkboxgroup to allow
the student to select more than one choice.
'''
response_tag = 'choiceresponse'
max_inputfields = 1
allowed_inputfields = ['checkboxgroup', 'radiogroup']
def setup_response(self):
self.assign_choice_names()
correct_xml = self.xml.xpath('//*[@id=$id]//choice[@correct="true"]',
id=self.xml.get('id'))
self.correct_choices = set([choice.get('name') for choice in correct_xml])
def assign_choice_names(self):
'''
Initialize name attributes in <choice> tags for this response.
'''
for index, choice in enumerate(self.xml.xpath('//*[@id=$id]//choice',
id=self.xml.get('id'))):
choice.set("name", "choice_" + str(index))
def get_score(self, student_answers):
student_answer = student_answers.get(self.answer_id, [])
if not isinstance(student_answer, list):
student_answer = [student_answer]
student_answer = set(student_answer)
required_selected = len(self.correct_choices - student_answer) == 0
no_extra_selected = len(student_answer - self.correct_choices) == 0
correct = required_selected & no_extra_selected
if correct:
return CorrectMap(self.answer_id, 'correct')
else:
return CorrectMap(self.answer_id, 'incorrect')
def get_answers(self):
return {self.answer_id: self.correct_choices}
#-----------------------------------------------------------------------------
class MultipleChoiceResponse(LoncapaResponse):
# TODO: handle direction and randomize
snippets = [{'snippet': '''<multiplechoiceresponse direction="vertical" randomize="yes">
<choicegroup type="MultipleChoice">
<choice location="random" correct="false"><span>`a+b`<br/></span></choice>
<choice location="random" correct="true"><span><math>a+b^2</math><br/></span></choice>
<choice location="random" correct="false"><math>a+b+c</math></choice>
<choice location="bottom" correct="false"><math>a+b+d</math></choice>
</choicegroup>
</multiplechoiceresponse>
'''}]
response_tag = 'multiplechoiceresponse'
max_inputfields = 1
allowed_inputfields = ['choicegroup']
def setup_response(self):
self.mc_setup_response() # call secondary setup for MultipleChoice questions, to set name attributes
# define correct choices (after calling secondary setup)
xml = self.xml
cxml = xml.xpath('//*[@id=$id]//choice[@correct="true"]', id=xml.get('id'))
self.correct_choices = [choice.get('name') for choice in cxml]
def mc_setup_response(self):
'''
Initialize name attributes in <choice> stanzas in the <choicegroup> in this response.
'''
i = 0
for response in self.xml.xpath("choicegroup"):
rtype = response.get('type')
if rtype not in ["MultipleChoice"]:
response.set("type", "MultipleChoice") # force choicegroup to be MultipleChoice if not valid
for choice in list(response):
if choice.get("name") is None:
choice.set("name", "choice_" + str(i))
i += 1
else:
choice.set("name", "choice_" + choice.get("name"))
def get_score(self, student_answers):
'''
grade student response.
'''
# log.debug('%s: student_answers=%s, correct_choices=%s' % (unicode(self),student_answers,self.correct_choices))
if self.answer_id in student_answers and student_answers[self.answer_id] in self.correct_choices:
return CorrectMap(self.answer_id, 'correct')
else:
return CorrectMap(self.answer_id, 'incorrect')
def get_answers(self):
return {self.answer_id: self.correct_choices}
class TrueFalseResponse(MultipleChoiceResponse):
response_tag = 'truefalseresponse'
def mc_setup_response(self):
i = 0
for response in self.xml.xpath("choicegroup"):
response.set("type", "TrueFalse")
for choice in list(response):
if choice.get("name") is None:
choice.set("name", "choice_" + str(i))
i += 1
else:
choice.set("name", "choice_" + choice.get("name"))
def get_score(self, student_answers):
correct = set(self.correct_choices)
answers = set(student_answers.get(self.answer_id, []))
if correct == answers:
return CorrectMap(self.answer_id, 'correct')
return CorrectMap(self.answer_id, 'incorrect')
#-----------------------------------------------------------------------------
class OptionResponse(LoncapaResponse):
'''
TODO: handle direction and randomize
'''
snippets = [{'snippet': '''<optionresponse direction="vertical" randomize="yes">
<optioninput options="('Up','Down')" correct="Up"><text>The location of the sky</text></optioninput>
<optioninput options="('Up','Down')" correct="Down"><text>The location of the earth</text></optioninput>
</optionresponse>'''}]
response_tag = 'optionresponse'
hint_tag = 'optionhint'
allowed_inputfields = ['optioninput']
def setup_response(self):
self.answer_fields = self.inputfields
def get_score(self, student_answers):
# log.debug('%s: student_answers=%s' % (unicode(self),student_answers))
cmap = CorrectMap()
amap = self.get_answers()
for aid in amap:
if aid in student_answers and student_answers[aid] == amap[aid]:
cmap.set(aid, 'correct')
else:
cmap.set(aid, 'incorrect')
return cmap
def get_answers(self):
amap = dict([(af.get('id'), af.get('correct')) for af in self.answer_fields])
# log.debug('%s: expected answers=%s' % (unicode(self),amap))
return amap
#-----------------------------------------------------------------------------
class NumericalResponse(LoncapaResponse):
response_tag = 'numericalresponse'
hint_tag = 'numericalhint'
allowed_inputfields = ['textline']
required_attributes = ['answer']
max_inputfields = 1
def setup_response(self):
xml = self.xml
context = self.context
self.correct_answer = contextualize_text(xml.get('answer'), context)
try:
self.tolerance_xml = xml.xpath('//*[@id=$id]//responseparam[@type="tolerance"]/@default',
id=xml.get('id'))[0]
self.tolerance = contextualize_text(self.tolerance_xml, context)
except Exception:
self.tolerance = 0
try:
self.answer_id = xml.xpath('//*[@id=$id]//textline/@id',
id=xml.get('id'))[0]
except Exception:
self.answer_id = None
def get_score(self, student_answers):
'''Grade a numeric response '''
student_answer = student_answers[self.answer_id]
try:
correct = compare_with_tolerance(evaluator(dict(), dict(), student_answer), complex(self.correct_answer), self.tolerance)
# We should catch this explicitly.
# I think this is just pyparsing.ParseException, calc.UndefinedVariable:
# But we'd need to confirm
except:
raise StudentInputError('Invalid input -- please use a number only')
if correct:
return CorrectMap(self.answer_id, 'correct')
else:
return CorrectMap(self.answer_id, 'incorrect')
# TODO: add check_hint_condition(self,hxml_set,student_answers)
def get_answers(self):
return {self.answer_id: self.correct_answer}
#-----------------------------------------------------------------------------
class StringResponse(LoncapaResponse):
response_tag = 'stringresponse'
hint_tag = 'stringhint'
allowed_inputfields = ['textline']
required_attributes = ['answer']
max_inputfields = 1
def setup_response(self):
self.correct_answer = contextualize_text(self.xml.get('answer'), self.context).strip()
def get_score(self, student_answers):
'''Grade a string response '''
student_answer = student_answers[self.answer_id].strip()
correct = self.check_string(self.correct_answer, student_answer)
return CorrectMap(self.answer_id, 'correct' if correct else 'incorrect')
def check_string(self, expected, given):
if self.xml.get('type') == 'ci': return given.lower() == expected.lower()
return given == expected
def check_hint_condition(self, hxml_set, student_answers):
given = student_answers[self.answer_id].strip()
hints_to_show = []
for hxml in hxml_set:
name = hxml.get('name')
correct_answer = contextualize_text(hxml.get('answer'), self.context).strip()
if self.check_string(correct_answer, given): hints_to_show.append(name)
log.debug('hints_to_show = %s' % hints_to_show)
return hints_to_show
def get_answers(self):
return {self.answer_id: self.correct_answer}
#-----------------------------------------------------------------------------
class CustomResponse(LoncapaResponse):
'''
Custom response. The python code to be run should be in <answer>...</answer>
or in a <script>...</script>
'''
snippets = [{'snippet': '''<customresponse>
<text>
<br/>
Suppose that \(I(t)\) rises from \(0\) to \(I_S\) at a time \(t_0 \neq 0\)
In the space provided below write an algebraic expression for \(I(t)\).
<br/>
<textline size="5" correct_answer="IS*u(t-t0)" />
</text>
<answer type="loncapa/python">
correct=['correct']
try:
r = str(submission[0])
except ValueError:
correct[0] ='incorrect'
r = '0'
if not(r=="IS*u(t-t0)"):
correct[0] ='incorrect'
</answer>
</customresponse>'''},
{'snippet': '''<script type="loncapa/python"><![CDATA[
def sympy_check2():
messages[0] = '%s:%s' % (submission[0],fromjs[0].replace('<','&lt;'))
#messages[0] = str(answers)
correct[0] = 'correct'
]]>
</script>
<customresponse cfn="sympy_check2" type="cs" expect="2.27E-39" dojs="math" size="30" answer="2.27E-39">
<textline size="40" dojs="math" />
<responseparam description="Numerical Tolerance" type="tolerance" default="0.00001" name="tol"/>
</customresponse>'''}]
response_tag = 'customresponse'
allowed_inputfields = ['textline', 'textbox']
def setup_response(self):
xml = self.xml
# if <customresponse> has an "expect" (or "answer") attribute then save that
self.expect = xml.get('expect') or xml.get('answer')
self.myid = xml.get('id')
log.debug('answer_ids=%s' % self.answer_ids)
# the <answer>...</answer> stanza should be local to the current <customresponse>. So try looking there first.
self.code = None
answer = None
try:
answer = xml.xpath('//*[@id=$id]//answer', id=xml.get('id'))[0]
except IndexError:
# print "xml = ",etree.tostring(xml,pretty_print=True)
# if we have a "cfn" attribute then look for the function specified by cfn, in the problem context
# ie the comparison function is defined in the <script>...</script> stanza instead
cfn = xml.get('cfn')
if cfn:
log.debug("cfn = %s" % cfn)
if cfn in self.context:
self.code = self.context[cfn]
else:
msg = "%s: can't find cfn %s in context" % (unicode(self), cfn)
msg += "\nSee XML source line %s" % getattr(self.xml, 'sourceline', '<unavailable>')
raise LoncapaProblemError(msg)
if not self.code:
if answer is None:
# raise Exception,"[courseware.capa.responsetypes.customresponse] missing code checking script! id=%s" % self.myid
log.error("[courseware.capa.responsetypes.customresponse] missing code checking script! id=%s" % self.myid)
self.code = ''
else:
answer_src = answer.get('src')
if answer_src is not None:
self.code = self.system.filesystem.open('src/' + answer_src).read()
else:
self.code = answer.text
def get_score(self, student_answers):
'''
student_answers is a dict with everything from request.POST, but with the first part
of each key removed (the string before the first "_").
'''
log.debug('%s: student_answers=%s' % (unicode(self), student_answers))
idset = sorted(self.answer_ids) # ordered list of answer id's
try:
submission = [student_answers[k] for k in idset] # ordered list of answers
except Exception as err:
msg = '[courseware.capa.responsetypes.customresponse] error getting student answer from %s' % student_answers
msg += '\n idset = %s, error = %s' % (idset, err)
log.error(msg)
raise Exception(msg)
# global variable in context which holds the Presentation MathML from dynamic math input
dynamath = [student_answers.get(k + '_dynamath', None) for k in idset] # ordered list of dynamath responses
# if there is only one box, and it's empty, then don't evaluate
if len(idset) == 1 and not submission[0]:
return CorrectMap(idset[0], 'incorrect', msg='<font color="red">No answer entered!</font>')
correct = ['unknown'] * len(idset)
messages = [''] * len(idset)
# put these in the context of the check function evaluator
# note that this doesn't help the "cfn" version - only the exec version
self.context.update({'xml': self.xml, # our subtree
'response_id': self.myid, # my ID
'expect': self.expect, # expected answer (if given as attribute)
'submission': submission, # ordered list of student answers from entry boxes in our subtree
'idset': idset, # ordered list of ID's of all entry boxes in our subtree
'dynamath': dynamath, # ordered list of all javascript inputs in our subtree
'answers': student_answers, # dict of student's responses, with keys being entry box IDs
'correct': correct, # the list to be filled in by the check function
'messages': messages, # the list of messages to be filled in by the check function
'options': self.xml.get('options'), # any options to be passed to the cfn
'testdat': 'hello world',
})
# pass self.system.debug to cfn
self.context['debug'] = self.system.DEBUG
# exec the check function
if type(self.code) == str:
try:
exec self.code in self.context['global_context'], self.context
except Exception as err:
print "oops in customresponse (code) error %s" % err
print "context = ", self.context
print traceback.format_exc()
else: # self.code is not a string; assume its a function
# this is an interface to the Tutor2 check functions
fn = self.code
ret = None
log.debug(" submission = %s" % submission)
try:
answer_given = submission[0] if (len(idset) == 1) else submission
# handle variable number of arguments in check function, for backwards compatibility
# with various Tutor2 check functions
args = [self.expect, answer_given, student_answers, self.answer_ids[0]]
argspec = inspect.getargspec(fn)
nargs = len(argspec.args) - len(argspec.defaults or [])
kwargs = {}
for argname in argspec.args[nargs:]:
kwargs[argname] = self.context[argname] if argname in self.context else None
log.debug('[customresponse] answer_given=%s' % answer_given)
log.debug('nargs=%d, args=%s, kwargs=%s' % (nargs, args, kwargs))
ret = fn(*args[:nargs], **kwargs)
except Exception as err:
log.error("oops in customresponse (cfn) error %s" % err)
# print "context = ",self.context
log.error(traceback.format_exc())
raise Exception("oops in customresponse (cfn) error %s" % err)
log.debug("[courseware.capa.responsetypes.customresponse.get_score] ret = %s" % ret)
if type(ret) == dict:
correct = ['correct'] * len(idset) if ret['ok'] else ['incorrect'] * len(idset)
msg = ret['msg']
if 1:
# try to clean up message html
msg = '<html>' + msg + '</html>'
msg = msg.replace('&#60;', '&lt;')
#msg = msg.replace('&lt;','<')
msg = etree.tostring(fromstring_bs(msg, convertEntities=None), pretty_print=True)
#msg = etree.tostring(fromstring_bs(msg),pretty_print=True)
msg = msg.replace('&#13;', '')
#msg = re.sub('<html>(.*)</html>','\\1',msg,flags=re.M|re.DOTALL) # python 2.7
msg = re.sub('(?ms)<html>(.*)</html>', '\\1', msg)
messages[0] = msg
else:
correct = ['correct'] * len(idset) if ret else ['incorrect'] * len(idset)
# build map giving "correct"ness of the answer(s)
correct_map = CorrectMap()
for k in range(len(idset)):
correct_map.set(idset[k], correct[k], msg=messages[k])
return correct_map
def get_answers(self):
'''
Give correct answer expected for this response.
use default_answer_map from entry elements (eg textline),
when this response has multiple entry objects.
but for simplicity, if an "expect" attribute was given by the content author
ie <customresponse expect="foo" ...> then that.
'''
if len(self.answer_ids) > 1:
return self.default_answer_map
if self.expect:
return {self.answer_ids[0]: self.expect}
return self.default_answer_map
#-----------------------------------------------------------------------------
class SymbolicResponse(CustomResponse):
"""
Symbolic math response checking, using symmath library.
"""
snippets = [{'snippet': '''<problem>
<text>Compute \[ \exp\left(-i \frac{\theta}{2} \left[ \begin{matrix} 0 & 1 \\ 1 & 0 \end{matrix} \right] \right) \]
and give the resulting \(2\times 2\) matrix: <br/>
<symbolicresponse answer="">
<textline size="40" math="1" />
</symbolicresponse>
<br/>
Your input should be typed in as a list of lists, eg <tt>[[1,2],[3,4]]</tt>.
</text>
</problem>'''}]
response_tag = 'symbolicresponse'
def setup_response(self):
self.xml.set('cfn', 'symmath_check')
code = "from symmath import *"
exec code in self.context, self.context
CustomResponse.setup_response(self)
#-----------------------------------------------------------------------------
class CodeResponse(LoncapaResponse):
'''
Grade student code using an external server, called 'xqueue'
In contrast to ExternalResponse, CodeResponse has following behavior:
1) Goes through a queueing system
2) Does not do external request for 'get_answers'
'''
response_tag = 'coderesponse'
allowed_inputfields = ['textline', 'textbox']
max_inputfields = 1
def setup_response(self):
xml = self.xml
self.url = xml.get('url', "http://ec2-50-16-59-149.compute-1.amazonaws.com/xqueue/submit/") # FIXME -- hardcoded url
answer = xml.find('answer')
if answer is not None:
answer_src = answer.get('src')
if answer_src is not None:
self.code = self.system.filesystem.open('src/' + answer_src).read()
else:
self.code = answer.text
else: # no <answer> stanza; get code from <script>
self.code = self.context['script_code']
if not self.code:
msg = '%s: Missing answer script code for coderesponse' % unicode(self)
msg += "\nSee XML source line %s" % getattr(self.xml, 'sourceline', '<unavailable>')
raise LoncapaProblemError(msg)
self.tests = xml.get('tests')
# Extract 'answer' and 'initial_display' from XML. Note that the code to be exec'ed here is:
# (1) Internal edX code, i.e. NOT student submissions, and
# (2) The code should only define the strings 'initial_display', 'answer', 'preamble', 'test_program'
# following the 6.01 problem definition convention
penv = {}
penv['__builtins__'] = globals()['__builtins__']
try:
exec(self.code, penv, penv)
except Exception as err:
log.error('Error in CodeResponse %s: Error in problem reference code' % err)
raise Exception(err)
try:
self.answer = penv['answer']
self.initial_display = penv['initial_display']
except Exception as err:
log.error("Error in CodeResponse %s: Problem reference code does not define 'answer' and/or 'initial_display' in <answer>...</answer>" % err)
raise Exception(err)
def get_score(self, student_answers):
try:
submission = [student_answers[self.answer_id]]
except Exception as err:
log.error('Error in CodeResponse %s: cannot get student answer for %s; student_answers=%s' % (err, self.answer_id, student_answers))
raise Exception(err)
self.context.update({'submission': submission})
extra_payload = {'edX_student_response': json.dumps(submission)}
r, queuekey = self._send_to_queue(extra_payload) # TODO: Perform checks on the xqueue response
# Non-null CorrectMap['queuekey'] indicates that the problem has been submitted
cmap = CorrectMap()
cmap.set(self.answer_id, queuekey=queuekey, msg='Submitted to queue')
return cmap
def update_score(self, score_msg, oldcmap, queuekey):
# Parse 'score_msg' as XML
try:
rxml = etree.fromstring(score_msg)
except Exception as err:
msg = 'Error in CodeResponse %s: cannot parse response from xworker r.text=%s' % (err, score_msg)
raise Exception(err)
# The following process is lifted directly from ExternalResponse
ad = rxml.find('awarddetail').text
admap = {'EXACT_ANS': 'correct', # TODO: handle other loncapa responses
'WRONG_FORMAT': 'incorrect',
}
self.context['correct'] = ['correct']
if ad in admap:
self.context['correct'][0] = admap[ad]
# Replace 'oldcmap' with new grading results if queuekey matches.
# If queuekey does not match, we keep waiting for the score_msg that will match
if oldcmap.is_right_queuekey(self.answer_id, queuekey):
msg = rxml.find('message').text.replace('&nbsp;', '&#160;')
oldcmap.set(self.answer_id, correctness=self.context['correct'][0], msg=msg, queuekey=None) # Queuekey is consumed
else:
log.debug('CodeResponse: queuekey %d does not match for answer_id=%s.' % (queuekey, self.answer_id))
return oldcmap
# CodeResponse differentiates from ExternalResponse in the behavior of 'get_answers'. CodeResponse.get_answers
# does NOT require a queue submission, and the answer is computed (extracted from problem XML) locally.
def get_answers(self):
anshtml = '<font color="blue"><span class="code-answer"><br/><pre>%s</pre><br/></span></font>' % self.answer
return {self.answer_id: anshtml}
def get_initial_display(self):
return {self.answer_id: self.initial_display}
# CodeResponse._send_to_queue implements the same interface as defined for ExternalResponse's 'get_score'
def _send_to_queue(self, extra_payload):
# Prepare payload
xmlstr = etree.tostring(self.xml, pretty_print=True)
header = {'return_url': self.system.xqueue_callback_url}
# Queuekey generation
h = hashlib.md5()
h.update(str(self.system.seed))
h.update(str(time.time()))
queuekey = int(h.hexdigest(), 16)
header.update({'queuekey': queuekey})
payload = {'xqueue_header': json.dumps(header), # TODO: 'xqueue_header' should eventually be derived from a config file
'xml': xmlstr,
'edX_cmd': 'get_score',
'edX_tests': self.tests,
'processor': self.code,
}
payload.update(extra_payload)
# Contact queue server
try:
r = requests.post(self.url, data=payload)
except Exception as err:
msg = "Error in CodeResponse %s: cannot connect to queue server url=%s" % (err, self.url)
log.error(msg)
raise Exception(msg)
return r, queuekey
#-----------------------------------------------------------------------------
class ExternalResponse(LoncapaResponse):
'''
Grade the students input using an external server.
Typically used by coding problems.
'''
snippets = [{'snippet': '''<externalresponse tests="repeat:10,generate">
<textbox rows="10" cols="70" mode="python"/>
<answer><![CDATA[
initial_display = """
def inc(x):
"""
answer = """
def inc(n):
return n+1
"""
preamble = """
import sympy
"""
test_program = """
import random
def testInc(n = None):
if n is None:
n = random.randint(2, 20)
print 'Test is: inc(%d)'%n
return str(inc(n))
def main():
f = os.fdopen(3,'w')
test = int(sys.argv[1])
rndlist = map(int,os.getenv('rndlist').split(','))
random.seed(rndlist[0])
if test == 1: f.write(testInc(0))
elif test == 2: f.write(testInc(1))
else: f.write(testInc())
f.close()
main()
"""
]]>
</answer>
</externalresponse>'''}]
response_tag = 'externalresponse'
allowed_inputfields = ['textline', 'textbox']
def setup_response(self):
xml = self.xml
self.url = xml.get('url') or "http://eecs1.mit.edu:8889/pyloncapa" # FIXME - hardcoded URL
# answer = xml.xpath('//*[@id=$id]//answer',id=xml.get('id'))[0] # FIXME - catch errors
answer = xml.find('answer')
if answer is not None:
answer_src = answer.get('src')
if answer_src is not None:
self.code = self.system.filesystem.open('src/' + answer_src).read()
else:
self.code = answer.text
else: # no <answer> stanza; get code from <script>
self.code = self.context['script_code']
if not self.code:
msg = '%s: Missing answer script code for externalresponse' % unicode(self)
msg += "\nSee XML source line %s" % getattr(self.xml, 'sourceline', '<unavailable>')
raise LoncapaProblemError(msg)
self.tests = xml.get('tests')
def do_external_request(self, cmd, extra_payload):
'''
Perform HTTP request / post to external server.
cmd = remote command to perform (str)
extra_payload = dict of extra stuff to post.
Return XML tree of response (from response body)
'''
xmlstr = etree.tostring(self.xml, pretty_print=True)
payload = {'xml': xmlstr,
'edX_cmd': cmd,
'edX_tests': self.tests,
'processor': self.code,
}
payload.update(extra_payload)
try:
r = requests.post(self.url, data=payload) # call external server
except Exception as err:
msg = 'Error %s - cannot connect to external server url=%s' % (err, self.url)
log.error(msg)
raise Exception(msg)
if self.system.DEBUG: log.info('response = %s' % r.text)
if (not r.text) or (not r.text.strip()):
raise Exception('Error: no response from external server url=%s' % self.url)
try:
rxml = etree.fromstring(r.text) # response is XML; prase it
except Exception as err:
msg = 'Error %s - cannot parse response from external server r.text=%s' % (err, r.text)
log.error(msg)
raise Exception(msg)
return rxml
def get_score(self, student_answers):
idset = sorted(self.answer_ids)
cmap = CorrectMap()
try:
submission = [student_answers[k] for k in idset]
except Exception as err:
log.error('Error %s: cannot get student answer for %s; student_answers=%s' % (err, self.answer_ids, student_answers))
raise Exception(err)
self.context.update({'submission': submission})
extra_payload = {'edX_student_response': json.dumps(submission)}
try:
rxml = self.do_external_request('get_score', extra_payload)
except Exception as err:
log.error('Error %s' % err)
if self.system.DEBUG:
cmap.set_dict(dict(zip(sorted(self.answer_ids), ['incorrect'] * len(idset))))
cmap.set_property(self.answer_ids[0], 'msg', '<font color="red" size="+2">%s</font>' % str(err).replace('<', '&lt;'))
return cmap
ad = rxml.find('awarddetail').text
admap = {'EXACT_ANS': 'correct', # TODO: handle other loncapa responses
'WRONG_FORMAT': 'incorrect',
}
self.context['correct'] = ['correct']
if ad in admap:
self.context['correct'][0] = admap[ad]
# create CorrectMap
for key in idset:
idx = idset.index(key)
msg = rxml.find('message').text.replace('&nbsp;', '&#160;') if idx == 0 else None
cmap.set(key, self.context['correct'][idx], msg=msg)
return cmap
def get_answers(self):
'''
Use external server to get expected answers
'''
try:
rxml = self.do_external_request('get_answers', {})
exans = json.loads(rxml.find('expected').text)
except Exception as err:
log.error('Error %s' % err)
if self.system.DEBUG:
msg = '<font color=red size=+2>%s</font>' % str(err).replace('<', '&lt;')
exans = [''] * len(self.answer_ids)
exans[0] = msg
if not (len(exans) == len(self.answer_ids)):
log.error('Expected %d answers from external server, only got %d!' % (len(self.answer_ids), len(exans)))
raise Exception('Short response from external server')
return dict(zip(self.answer_ids, exans))
#-----------------------------------------------------------------------------
class FormulaResponse(LoncapaResponse):
'''
Checking of symbolic math response using numerical sampling.
'''
snippets = [{'snippet': '''<problem>
<script type="loncapa/python">
I = "m*c^2"
</script>
<text>
<br/>
Give an equation for the relativistic energy of an object with mass m.
</text>
<formularesponse type="cs" samples="m,c@1,2:3,4#10" answer="$I">
<responseparam description="Numerical Tolerance" type="tolerance"
default="0.00001" name="tol" />
<textline size="40" math="1" />
</formularesponse>
</problem>'''}]
response_tag = 'formularesponse'
hint_tag = 'formulahint'
allowed_inputfields = ['textline']
required_attributes = ['answer']
max_inputfields = 1
def setup_response(self):
xml = self.xml
context = self.context
self.correct_answer = contextualize_text(xml.get('answer'), context)
self.samples = contextualize_text(xml.get('samples'), context)
try:
self.tolerance_xml = xml.xpath('//*[@id=$id]//responseparam[@type="tolerance"]/@default',
id=xml.get('id'))[0]
self.tolerance = contextualize_text(self.tolerance_xml, context)
except Exception:
self.tolerance = '0.00001'
ts = xml.get('type')
if ts is None:
typeslist = []
else:
typeslist = ts.split(',')
if 'ci' in typeslist: # Case insensitive
self.case_sensitive = False
elif 'cs' in typeslist: # Case sensitive
self.case_sensitive = True
else: # Default
self.case_sensitive = False
def get_score(self, student_answers):
given = student_answers[self.answer_id]
correctness = self.check_formula(self.correct_answer, given, self.samples)
return CorrectMap(self.answer_id, correctness)
def check_formula(self, expected, given, samples):
variables = samples.split('@')[0].split(',')
numsamples = int(samples.split('@')[1].split('#')[1])
sranges = zip(*map(lambda x: map(float, x.split(",")),
samples.split('@')[1].split('#')[0].split(':')))
ranges = dict(zip(variables, sranges))
for i in range(numsamples):
instructor_variables = self.strip_dict(dict(self.context))
student_variables = dict()
for var in ranges: # ranges give numerical ranges for testing
value = random.uniform(*ranges[var])
instructor_variables[str(var)] = value
student_variables[str(var)] = value
#log.debug('formula: instructor_vars=%s, expected=%s' % (instructor_variables,expected))
instructor_result = evaluator(instructor_variables, dict(), expected, cs=self.case_sensitive)
try:
#log.debug('formula: student_vars=%s, given=%s' % (student_variables,given))
student_result = evaluator(student_variables,
dict(),
given,
cs=self.case_sensitive)
except UndefinedVariable as uv:
log.debug('formularesponse: undefined variable in given=%s' % given)
raise StudentInputError(uv.message + " not permitted in answer")
except Exception as err:
#traceback.print_exc()
log.debug('formularesponse: error %s in formula' % err)
raise StudentInputError("Error in formula")
if numpy.isnan(student_result) or numpy.isinf(student_result):
return "incorrect"
if not compare_with_tolerance(student_result, instructor_result, self.tolerance):
return "incorrect"
return "correct"
def strip_dict(self, d):
''' Takes a dict. Returns an identical dict, with all non-word
keys and all non-numeric values stripped out. All values also
converted to float. Used so we can safely use Python contexts.
'''
d = dict([(k, numpy.complex(d[k])) for k in d if type(k) == str and \
k.isalnum() and \
isinstance(d[k], numbers.Number)])
return d
def check_hint_condition(self, hxml_set, student_answers):
given = student_answers[self.answer_id]
hints_to_show = []
for hxml in hxml_set:
samples = hxml.get('samples')
name = hxml.get('name')
correct_answer = contextualize_text(hxml.get('answer'), self.context)
try:
correctness = self.check_formula(correct_answer, given, samples)
except Exception:
correctness = 'incorrect'
if correctness == 'correct':
hints_to_show.append(name)
log.debug('hints_to_show = %s' % hints_to_show)
return hints_to_show
def get_answers(self):
return {self.answer_id: self.correct_answer}
#-----------------------------------------------------------------------------
class SchematicResponse(LoncapaResponse):
response_tag = 'schematicresponse'
allowed_inputfields = ['schematic']
def setup_response(self):
xml = self.xml
answer = xml.xpath('//*[@id=$id]//answer', id=xml.get('id'))[0]
answer_src = answer.get('src')
if answer_src is not None:
self.code = self.system.filestore.open('src/' + answer_src).read() # Untested; never used
else:
self.code = answer.text
def get_score(self, student_answers):
from capa_problem import global_context
submission = [json.loads(student_answers[k]) for k in sorted(self.answer_ids)]
self.context.update({'submission': submission})
exec self.code in global_context, self.context
cmap = CorrectMap()
cmap.set_dict(dict(zip(sorted(self.answer_ids), self.context['correct'])))
return cmap
def get_answers(self):
# use answers provided in input elements
return self.default_answer_map
#-----------------------------------------------------------------------------
class ImageResponse(LoncapaResponse):
"""
Handle student response for image input: the input is a click on an image,
which produces an [x,y] coordinate pair. The click is correct if it falls
within a region specified. This region is nominally a rectangle.
Lon-CAPA requires that each <imageresponse> has a <foilgroup> inside it. That
doesn't make sense to me (Ike). Instead, let's have it such that <imageresponse>
should contain one or more <imageinput> stanzas. Each <imageinput> should specify
a rectangle, given as an attribute, defining the correct answer.
"""
snippets = [{'snippet': '''<imageresponse>
<imageinput src="image1.jpg" width="200" height="100" rectangle="(10,10)-(20,30)" />
<imageinput src="image2.jpg" width="210" height="130" rectangle="(12,12)-(40,60)" />
</imageresponse>'''}]
response_tag = 'imageresponse'
allowed_inputfields = ['imageinput']
def setup_response(self):
self.ielements = self.inputfields
self.answer_ids = [ie.get('id') for ie in self.ielements]
def get_score(self, student_answers):
correct_map = CorrectMap()
expectedset = self.get_answers()
for aid in self.answer_ids: # loop through IDs of <imageinput> fields in our stanza
given = student_answers[aid] # this should be a string of the form '[x,y]'
# parse expected answer
# TODO: Compile regexp on file load
m = re.match('[\(\[]([0-9]+),([0-9]+)[\)\]]-[\(\[]([0-9]+),([0-9]+)[\)\]]', expectedset[aid].strip().replace(' ', ''))
if not m:
msg = 'Error in problem specification! cannot parse rectangle in %s' % (etree.tostring(self.ielements[aid],
pretty_print=True))
raise Exception('[capamodule.capa.responsetypes.imageinput] ' + msg)
(llx, lly, urx, ury) = [int(x) for x in m.groups()]
# parse given answer
m = re.match('\[([0-9]+),([0-9]+)]', given.strip().replace(' ', ''))
if not m:
raise Exception('[capamodule.capa.responsetypes.imageinput] error grading %s (input=%s)' % (aid, given))
(gx, gy) = [int(x) for x in m.groups()]
# answer is correct if (x,y) is within the specified rectangle
if (llx <= gx <= urx) and (lly <= gy <= ury):
correct_map.set(aid, 'correct')
else:
correct_map.set(aid, 'incorrect')
return correct_map
def get_answers(self):
return dict([(ie.get('id'), ie.get('rectangle')) for ie in self.ielements])
#-----------------------------------------------------------------------------
# TEMPORARY: List of all response subclasses
# FIXME: To be replaced by auto-registration
__all__ = [CodeResponse, NumericalResponse, FormulaResponse, CustomResponse, SchematicResponse, ExternalResponse, ImageResponse, OptionResponse, SymbolicResponse, StringResponse, ChoiceResponse, MultipleChoiceResponse, TrueFalseResponse]
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment