djmoffat@1099: #!/usr/bin/python djmoffat@1099: djmoffat@1099: import xml.etree.ElementTree as ET djmoffat@1099: import os djmoffat@1099: import sys djmoffat@1099: import csv djmoffat@1099: djmoffat@1099: # COMMAND LINE ARGUMENTS djmoffat@1099: djmoffat@1099: assert len(sys.argv)<3, "score_parser takes at most 1 command line argument\n"+\ djmoffat@1099: "Use: python score_parser.py [rating_folder_location]" djmoffat@1099: djmoffat@1099: # XML results files location djmoffat@1099: if len(sys.argv) == 1: djmoffat@1099: folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder djmoffat@1099: print "Use: python score_parser.py [rating_folder_location]" djmoffat@1099: print "Using default path: " + folder_name djmoffat@1099: elif len(sys.argv) == 2: djmoffat@1099: folder_name = sys.argv[1] # First command line argument is folder djmoffat@1099: djmoffat@1099: # check if folder_name exists djmoffat@1099: if not os.path.exists(folder_name): djmoffat@1099: #the file is not there djmoffat@1099: print "Folder '"+folder_name+"' does not exist." djmoffat@1099: sys.exit() # terminate script execution djmoffat@1099: elif not os.access(os.path.dirname(folder_name), os.W_OK): djmoffat@1099: #the file does exist but write privileges are not given djmoffat@1099: print "No write privileges in folder '"+folder_name+"'." djmoffat@1099: djmoffat@1099: djmoffat@1099: # CODE djmoffat@1099: djmoffat@1099: # remember which files have been opened this time djmoffat@1099: file_history = [] djmoffat@1099: djmoffat@1099: # get every XML file in folder djmoffat@1099: for file_name in os.listdir(folder_name): djmoffat@1099: if file_name.endswith(".xml"): djmoffat@1099: tree = ET.parse(folder_name + '/' + file_name) djmoffat@1099: root = tree.getroot() djmoffat@1099: djmoffat@1099: # get subject ID from XML file djmoffat@1099: subject_id = file_name[:-4] # file name (without extension) as subject ID djmoffat@1099: djmoffat@1099: # get list of all pages this subject evaluated n@1102: for audioholder in root.findall("./page"): # iterate over pages djmoffat@1099: page_name = audioholder.get('id') # get page name djmoffat@1099: djmoffat@1099: if page_name is None: # ignore 'empty' audio_holders djmoffat@1099: print "WARNING: " + file_name + " contains empty audio holder. (score_parser.py)" djmoffat@1099: break djmoffat@1099: djmoffat@1099: file_name = folder_name+'/ratings/'+page_name+'-ratings.csv' # score file name djmoffat@1099: djmoffat@1099: # create folder 'ratings' if not yet created djmoffat@1099: if not os.path.exists(folder_name + '/ratings'): djmoffat@1099: os.makedirs(folder_name + '/ratings') djmoffat@1099: djmoffat@1099: # header: fragment IDs in 'alphabetical' order djmoffat@1099: # go to fragment column, or create new column if it doesn't exist yet djmoffat@1099: djmoffat@1099: # get array of audio elements and number of audio elements djmoffat@1099: audiolist = audioholder.findall("./audioelement") djmoffat@1099: n_fragments = len(audiolist) djmoffat@1099: djmoffat@1099: # get alphabetical array of fragment IDs from this subject's XML djmoffat@1099: fragmentnamelist = [] # make empty list djmoffat@1099: for audioelement in audiolist: # iterate over all audioelements djmoffat@1099: fragmentnamelist.append(audioelement.get('id')) # add to list djmoffat@1099: djmoffat@1099: djmoffat@1099: # if file exists, get header and add any 'new' fragments not yet in the header djmoffat@1099: if os.path.isfile(file_name): djmoffat@1099: with open(file_name, 'r') as readfile: djmoffat@1099: filereader = csv.reader(readfile, delimiter=',') djmoffat@1099: headerrow = filereader.next() djmoffat@1099: djmoffat@1099: # If file hasn't been opened yet this time, remove all rows except header djmoffat@1099: if file_name not in file_history: djmoffat@1099: with open(file_name, 'w') as writefile: djmoffat@1099: filewriter = csv.writer(writefile, delimiter=',') djmoffat@1099: headerrow = sorted(headerrow) djmoffat@1099: filewriter.writerow(headerrow) djmoffat@1099: file_history.append(file_name) djmoffat@1099: djmoffat@1099: # Which of the fragments are in fragmentnamelist but not in headerrow? djmoffat@1099: newfragments = list(set(fragmentnamelist)-set(headerrow)) djmoffat@1099: newfragments = sorted(newfragments) # new fragments in alphabetical order djmoffat@1099: # If not empty, read file and rewrite adding extra columns djmoffat@1099: if newfragments: # if not empty djmoffat@1099: with open('temp.csv', 'w') as writefile: djmoffat@1099: filewriter = csv.writer(writefile, delimiter=',') djmoffat@1099: filewriter.writerow(headerrow + newfragments) # write new header djmoffat@1099: with open(file_name, 'r') as readfile: djmoffat@1099: filereader = csv.reader(readfile, delimiter=',') djmoffat@1099: filereader.next() # skip header djmoffat@1099: for row in filereader: # rewrite row plus empty cells for every new fragment name djmoffat@1099: filewriter.writerow(row + ['']*len(newfragments)) djmoffat@1099: os.rename('temp.csv', file_name) # replace old file with temp file djmoffat@1099: headerrow = headerrow + newfragments djmoffat@1099: djmoffat@1099: djmoffat@1099: # if file does not exist yet, create file and make header djmoffat@1099: else: djmoffat@1099: headerrow = sorted(fragmentnamelist) # sort alphabetically djmoffat@1099: headerrow.insert(0,'') djmoffat@1099: fragmentnamelist = fragmentnamelist[1:] #HACKY FIX inserting in firstrow also affects fragmentnamelist djmoffat@1099: with open(file_name, 'w') as writefile: djmoffat@1099: filewriter = csv.writer(writefile, delimiter=',') djmoffat@1099: filewriter.writerow(headerrow) djmoffat@1099: file_history.append(file_name) djmoffat@1099: djmoffat@1099: # open file to write for this page djmoffat@1099: writefile = open(file_name, 'a') djmoffat@1099: filewriter = csv.writer(writefile, delimiter=',') djmoffat@1099: djmoffat@1099: # prepare row to be written for this subject for this page djmoffat@1099: ratingrow = [subject_id] djmoffat@1099: djmoffat@1099: # get scores related to fragment [id] djmoffat@1099: for fragmentname in headerrow[1:]: # iterate over fragments in header (skip first empty column) djmoffat@1099: elementvalue = audioholder.find("./audioelement/[@id='" djmoffat@1099: + fragmentname djmoffat@1099: + "']/value") djmoffat@1099: if hasattr(elementvalue, 'text'): # if rating for this fragment exists djmoffat@1099: ratingrow.append(elementvalue.text) # add to rating row djmoffat@1099: else: # if this subject has not rated this fragment djmoffat@1099: ratingrow.append('') # append empty cell djmoffat@1099: djmoffat@1099: # write row: [subject ID, rating fragment ID 1, ..., rating fragment ID M] djmoffat@1099: if any(ratingrow[1:]): # append to file if row non-empty (except subject name) djmoffat@1099: filewriter.writerow(ratingrow)