def prepare_for_processing(self,verbose=True):
if self.prepared:
return self.code_list
self.d = acedu.paths.corresponding()
# I need a pretty printer for debugging purposes
self.p = pprint.PrettyPrinter(indent=2)
self.homeworks = []
for homework_file in self.d:
if verbose:
print "parsing "+homework_file.path
try:
if homework_file.basename() != ".DS_Store":
self.homeworks.append(parse(homework_file.path))
except:
print "parsing error in "+homework_file.path
raise
self.assignments = []
for homework in self.homeworks:
homework_assignments = homework.getElementsByTagName("assignment")
if 0 == len(homework_assignments):
print "Warning, encountered homework that had no assignment."
print "outputing assignment"
print homework.toprettyxml()
sys.exit("Warning, encountered homework that had no assignment.")
for assignment in homework_assignments:
self.assignments.append(assignment)
for assignment in self.assignments:
attrs = attributesOf(assignment)
try:
assert(attrs["student"])
print (attrs["student"])
assert(attrs["book_edition"])
assert(attrs["chapter"])
problems = assignment.getElementsByTagName("problem")
for problem in problems:
attrs = attributesOf(problem)
assert(attrs["number"])
if verbose: print getText(problem.childNodes)
except:
print ("failed on "+str(attrs))
raise
self.assignments = map(Assignment,self.assignments)
self.book_editions = defaultdict(list)
for assignment in self.assignments:
self.book_editions[assignment.book_edition].append(assignment)
#print self.book_editions
for assignment in self.assignments:
if(assignment.chapter != int(sys.argv[1])):
print assignment.student
assert(assignment.chapter == int(sys.argv[1]))
num_probs = len(assignment.problems)
if verbose: print "num_probs was "+str(num_probs)+" for student "+assignment.student
self.problems = []
for assignment in self.assignments:
for problem in assignment.problems:
prob_id = (assignment.book_edition,assignment.chapter,problem.number)
prob_data = (assignment.student,problem.text)
prob = (prob_id,prob_data)
self.problems.append(prob)
if verbose: print len(self.problems)
self.problem_groups = defaultdict(list)
self.code_list = []
for problem in self.problems:
prob_id, prob_data = problem
assignment_student,problem_text = prob_data
prob_code_data = (assignment_student,prob_id)
prob_code = str(hash(prob_code_data))
self.code_list.append((prob_code,prob_code_data))
self.problem_groups[prob_id].append(prob_code+"\n"+("="*50)+"\n"+problem_text)
if verbose: self.p.pprint(self.code_list)
#print "by groups"
self.count_by_groups = 0
for group in self.problem_groups.values():
self.count_by_groups += len(group)
if verbose: print self.count_by_groups
self.prepared = True
return self.code_list
def __init__(self,theNode): attrs = attributesOf(theNode) self.number = int(attrs["number"]) self.text = getText(theNode.childNodes)