n@1106: #!/usr/bin/python n@1106: n@1106: import xml.etree.ElementTree as ET n@1106: import os n@1106: import sys n@1106: import csv n@1106: n@1106: # COMMAND LINE ARGUMENTS n@1106: n@1106: assert len(sys.argv)<3, "score_parser takes at most 1 command line argument\n"+\ n@1106: "Use: python score_parser.py [rating_folder_location]" n@1106: n@1106: # XML results files location n@1106: if len(sys.argv) == 1: n@1106: folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder n@1106: print "Use: python score_parser.py [rating_folder_location]" n@1106: print "Using default path: " + folder_name n@1106: elif len(sys.argv) == 2: n@1106: folder_name = sys.argv[1] # First command line argument is folder n@1106: n@1106: # check if folder_name exists n@1106: if not os.path.exists(folder_name): n@1106: #the file is not there n@1106: print "Folder '"+folder_name+"' does not exist." n@1106: sys.exit() # terminate script execution n@1106: elif not os.access(os.path.dirname(folder_name), os.W_OK): n@1106: #the file does exist but write privileges are not given n@1106: print "No write privileges in folder '"+folder_name+"'." n@1106: n@1106: n@1106: # CODE n@1106: n@1106: # remember which files have been opened this time n@1106: file_history = [] n@1106: n@1106: # get every XML file in folder n@1106: for file_name in os.listdir(folder_name): n@1106: if file_name.endswith(".xml"): n@1106: tree = ET.parse(folder_name + '/' + file_name) n@1106: root = tree.getroot() n@1106: n@1106: # get subject ID from XML file n@1106: subject_id = file_name[:-4] # file name (without extension) as subject ID n@1106: n@1106: # get list of all pages this subject evaluated n@1109: for audioholder in root.findall("./page"): # iterate over pages n@1106: page_name = audioholder.get('id') # get page name n@1106: n@1106: if page_name is None: # ignore 'empty' audio_holders n@1106: print "WARNING: " + file_name + " contains empty audio holder. (score_parser.py)" n@1106: break n@1106: n@1106: file_name = folder_name+'/ratings/'+page_name+'-ratings.csv' # score file name n@1106: n@1106: # create folder 'ratings' if not yet created n@1106: if not os.path.exists(folder_name + '/ratings'): n@1106: os.makedirs(folder_name + '/ratings') n@1106: n@1106: # header: fragment IDs in 'alphabetical' order n@1106: # go to fragment column, or create new column if it doesn't exist yet n@1106: n@1106: # get array of audio elements and number of audio elements n@1106: audiolist = audioholder.findall("./audioelement") n@1106: n_fragments = len(audiolist) n@1106: n@1106: # get alphabetical array of fragment IDs from this subject's XML n@1106: fragmentnamelist = [] # make empty list n@1106: for audioelement in audiolist: # iterate over all audioelements n@1106: fragmentnamelist.append(audioelement.get('id')) # add to list n@1106: n@1106: n@1106: # if file exists, get header and add any 'new' fragments not yet in the header n@1106: if os.path.isfile(file_name): n@1106: with open(file_name, 'r') as readfile: n@1106: filereader = csv.reader(readfile, delimiter=',') n@1106: headerrow = filereader.next() n@1106: n@1106: # If file hasn't been opened yet this time, remove all rows except header n@1106: if file_name not in file_history: n@1106: with open(file_name, 'w') as writefile: n@1106: filewriter = csv.writer(writefile, delimiter=',') n@1106: headerrow = sorted(headerrow) n@1106: filewriter.writerow(headerrow) n@1106: file_history.append(file_name) n@1106: n@1106: # Which of the fragments are in fragmentnamelist but not in headerrow? n@1106: newfragments = list(set(fragmentnamelist)-set(headerrow)) n@1106: newfragments = sorted(newfragments) # new fragments in alphabetical order n@1106: # If not empty, read file and rewrite adding extra columns n@1106: if newfragments: # if not empty n@1106: with open('temp.csv', 'w') as writefile: n@1106: filewriter = csv.writer(writefile, delimiter=',') n@1106: filewriter.writerow(headerrow + newfragments) # write new header n@1106: with open(file_name, 'r') as readfile: n@1106: filereader = csv.reader(readfile, delimiter=',') n@1106: filereader.next() # skip header n@1106: for row in filereader: # rewrite row plus empty cells for every new fragment name n@1106: filewriter.writerow(row + ['']*len(newfragments)) n@1106: os.rename('temp.csv', file_name) # replace old file with temp file n@1106: headerrow = headerrow + newfragments n@1106: n@1106: n@1106: # if file does not exist yet, create file and make header n@1106: else: n@1106: headerrow = sorted(fragmentnamelist) # sort alphabetically n@1106: headerrow.insert(0,'') n@1106: fragmentnamelist = fragmentnamelist[1:] #HACKY FIX inserting in firstrow also affects fragmentnamelist n@1106: with open(file_name, 'w') as writefile: n@1106: filewriter = csv.writer(writefile, delimiter=',') n@1106: filewriter.writerow(headerrow) n@1106: file_history.append(file_name) n@1106: n@1106: # open file to write for this page n@1106: writefile = open(file_name, 'a') n@1106: filewriter = csv.writer(writefile, delimiter=',') n@1106: n@1106: # prepare row to be written for this subject for this page n@1106: ratingrow = [subject_id] n@1106: n@1106: # get scores related to fragment [id] n@1106: for fragmentname in headerrow[1:]: # iterate over fragments in header (skip first empty column) n@1106: elementvalue = audioholder.find("./audioelement/[@id='" n@1106: + fragmentname n@1106: + "']/value") n@1106: if hasattr(elementvalue, 'text'): # if rating for this fragment exists n@1106: ratingrow.append(elementvalue.text) # add to rating row n@1106: else: # if this subject has not rated this fragment n@1106: ratingrow.append('') # append empty cell n@1106: n@1106: # write row: [subject ID, rating fragment ID 1, ..., rating fragment ID M] n@1106: if any(ratingrow[1:]): # append to file if row non-empty (except subject name) n@1106: filewriter.writerow(ratingrow)