Mercurial > hg > webaudioevaluationtool
changeset 1074:2b8c36924bfd
Merge from "dev_main". Added MUSHRA, PHP upload scripts, APE on Firefox, multi-screen support. Bug fixes #1370, #1298, #1391, #1300, #1389.
| author | Nicholas Jillings <n.g.r.jillings@se14.qmul.ac.uk> |
|---|---|
| date | Tue, 15 Sep 2015 10:20:50 +0100 |
| parents | 57708aa826bb (current diff) 522a2ed8afa2 (diff) |
| children | 056a8454500e |
| files | ape.css ape.js core.css core.js docs/ProjectSpecificationDocument.pdf docs/ProjectSpecificationDocument.tex example_eval/project.xml |
| diffstat | 9 files changed, 1104 insertions(+), 87 deletions(-) [+] |
line wrap: on
line diff
--- a/.hgignore Tue Sep 15 10:16:34 2015 +0100 +++ b/.hgignore Tue Sep 15 10:20:50 2015 +0100 @@ -30,4 +30,7 @@ saves/*.csv saves/*/*.csv saves/*/*.png -saves/*/*.xml \ No newline at end of file +saves/*/*.xml +saves/ratings/*.pdf +saves/timelines_movement/*.pdf +saves \ No newline at end of file
--- a/README.txt Tue Sep 15 10:16:34 2015 +0100 +++ b/README.txt Tue Sep 15 10:20:50 2015 +0100 @@ -35,6 +35,7 @@ QUICK START + Using the example project: 1. Make sure your system sample rate corresponds with the sample rate of the audio files, if the input XML file enforces the given sample rate. 2. Run pythonServer.py (make sure you have Python installed). @@ -44,6 +45,7 @@ LEGACY + The APE interface and most of the functionality of the interface is inspired by the APE toolbox for MATLAB [1]. See https://code.soundsoftware.ac.uk/projects/ape for the source code and corresponding paper. @@ -81,10 +83,16 @@ In Firefox, go to Tools>Web Developer>Web Console, or hit Cmd + Alt + K. +REMOTE TESTS + +As the test is browser-based, it can be run remotely from a web server without modification. To allow for remote storage of the output XML files (as opposed to saving them locally on the subject’s machine, which is the default if no ‘save’ path is specified or found), a PHP script on the server needs to accept the output XML files. An example of such script will be included in a future version. + + SCRIPTS -The tool comes with a few handy Python scripts for easy extraction of ratings or comments, and visualisation of ratings and timelines. See below for a quick guide on how to use them. All scripts written for Python 2.7. Visualisation requires the free matplotlib toolbox (http://matplotlib.org), numpy and scipy. -By default, the scripts can be run from the ‘scripts’ folder, with the result files in the ‘saves’ folder (the default location where result XMLs are stored). +The tool comes with a few handy Python (2.7) scripts for easy extraction of ratings or comments, and visualisation of ratings and timelines. See below for a quick guide on how to use them. All scripts written for Python 2.7. Visualisation requires the free matplotlib toolbox (http://matplotlib.org), numpy and scipy. +By default, the scripts can be run from the ‘scripts’ folder, with the result files in the ‘saves’ folder (the default location where result XMLs are stored). Each script takes the XML file folder as an argument, along with other arguments in some cases. +Note: to avoid all kinds of problems, please avoid using spaces in file and folder names (this may work on some systems, but others don’t like it). comment_parser.py Extracts comments from the output XML files corresponding with the different subjects found in ‘saves/’. It creates a folder per ‘audioholder’/page it finds, and stores a CSV file with comments for every ‘audioelement’/fragment within these respective ‘audioholders’/pages. In this CSV file, every line corresponds with a subject/output XML file. Depending on the settings, the first column containing the name of the corresponding XML file can be omitted (for anonymisation). @@ -93,6 +101,9 @@ evaluation_stats.py Shows a few statistics of tests in the ‘saves/‘ folder so far, mainly for checking for errors. Shows the number of files that are there, the audioholder IDs that were tested (and how many of each separate ID), the duration of each page, the duration of each complete test, the average duration per page, and the average duration in function of the page number. + generate_report.py + Similar to ‘evaluation_stats.py’, but generates a PDF report based on the output files in the ‘saves/‘ folder - or any folder specified as command line argument. Uses pdflatex to write a LaTeX document, then convert to a PDF. + score_parser.py Extracts rating values from the XML to CSV - necessary for running visualisation of ratings. Creates the folder ‘saves/ratings/‘ if not yet created, to which it writes a separate file for every ‘audioholder’/page in any of the output XMLs it finds in ‘saves/‘. Within each file, rows represent different subjects (output XML file names) and columns represent different ‘audioelements’/fragments. @@ -102,6 +113,9 @@ Requires the free matplotlib library. At this point, more than one subjects are needed for this script to work. + timeline_view_movement.py + Creates a timeline for every subject, for every ‘audioholder’/page, corresponding with any of the output XML files found in ‘/saves’. It shows the marker movements of the different fragments, along with when each fragment was played (red regions). Automatically takes fragment names, rating axis title, rating axis labels, and audioholder name from the XML file (if available). + timeline_view.py Creates a timeline for every subject, for every ‘audioholder’/page, corresponding with any of the output XML files found in ‘/saves’. It shows when and for how long the subject listened to each of the fragments.
--- a/scripts/comment_parser.py Tue Sep 15 10:16:34 2015 +0100 +++ b/scripts/comment_parser.py Tue Sep 15 10:20:50 2015 +0100 @@ -4,9 +4,35 @@ import xml.etree.ElementTree as ET import os import csv +import sys -# XML results files location (modify as needed): -folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder +# COMMAND LINE ARGUMENTS + +assert len(sys.argv)<3, "comment_parser takes at most 1 command line argument\n"+\ + "Use: python score_parser.py [rating_folder_location]" + +# XML results files location +if len(sys.argv) == 1: + folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder + print "Use: python comment_parser.py [XML_files_location]" + print "Using default path: " + folder_name +elif len(sys.argv) == 2: + folder_name = sys.argv[1] # First command line argument is folder + +# check if folder_name exists +if not os.path.exists(folder_name): + #the file is not there + print "Folder '"+folder_name+"' does not exist." + sys.exit() # terminate script execution +elif not os.access(os.path.dirname(folder_name), os.W_OK): + #the file does exist but write privileges are not given + print "No write privileges in folder '"+folder_name+"'." + + +# CODE + +# remember which files have been opened this time +file_history = [] # get every XML file in folder for file in os.listdir(folder_name): @@ -33,23 +59,28 @@ csv_name = folder_name +'/' + page_name+'/'+page_name+'-comments-'+audio_id+'.csv' - # append (!) to file [page_name]/[page_name]-comments-[id].csv - with open(csv_name, 'a') as csvfile: - writer = csv.writer(csvfile, - delimiter=',', - dialect="excel", - quoting=csv.QUOTE_ALL) - commentstr = audioelement.find("./comment/response").text + # If file hasn't been opened yet this time, empty + if csv_name not in file_history: + csvfile = open(csv_name, 'w') + file_history.append(csv_name) # remember this file has been written to this time around + else: + # append (!) to file [page_name]/[page_name]-comments-[id].csv + csvfile = open(csv_name, 'a') + writer = csv.writer(csvfile, + delimiter=',', + dialect="excel", + quoting=csv.QUOTE_ALL) + commentstr = audioelement.find("./comment/response").text - if commentstr is None: - commentstr = ''; - - # anonymous comments: - #writer.writerow([commentstr.encode("utf-8")]) - # comments with (file) name: - writer.writerow([file[:-4]] + [commentstr.encode("utf-8")]) + if commentstr is None: + commentstr = '' + + # anonymous comments: + #writer.writerow([commentstr.encode("utf-8")]) + # comments with (file) name: + writer.writerow([file[:-4]] + [commentstr.encode("utf-8")]) - #TODO Replace 'new line' in comment with something else? + #TODO Replace 'new line' in comment with something else? # PRO TIP: Change from csv to txt by running this in bash: # $ cd folder_where_csvs_are/
--- a/scripts/evaluation_stats.py Tue Sep 15 10:16:34 2015 +0100 +++ b/scripts/evaluation_stats.py Tue Sep 15 10:20:50 2015 +0100 @@ -4,9 +4,19 @@ import xml.etree.ElementTree as ET import os # for getting files from directory import operator # for sorting data with multiple keys +import sys # for accessing command line arguments -# XML results files location (modify as needed): -folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder +# Command line arguments +assert len(sys.argv)<3, "evaluation_stats takes at most 1 command line argument\n"+\ + "Use: python evaluation_stats.py [results_folder]" + +# XML results files location +if len(sys.argv) == 1: + folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder + print "Use: python evaluation_stats.py [results_folder]" + print "Using default path: " + folder_name +elif len(sys.argv) == 2: + folder_name = sys.argv[1] # First command line argument is folder # Turn number of seconds (int) to '[minutes] min [seconds] s' (string) def seconds2timestr(time_in_seconds): @@ -19,6 +29,8 @@ number_of_pages = 0 number_of_fragments = 0 total_empty_comments = 0 +total_not_played = 0 +total_not_moved = 0 time_per_page_accum = 0 # arrays initialisation @@ -52,6 +64,14 @@ number_of_comments = 0 # for this page number_of_missing_comments = 0 # for this page + not_played = 0 # for this page + not_moved = 0 # for this page + + # 'testTime' keeps total duration: subtract time so far for duration of this audioholder + duration = float(audioholder.find("./metric/metricresult[@id='testTime']").text) - total_duration + + # total duration of test + total_duration += duration # number of audio elements audioelements = audioholder.findall("./audioelement") # get audioelements @@ -60,24 +80,39 @@ # number of comments (interesting if comments not mandatory) for audioelement in audioelements: response = audioelement.find("./comment/response") + was_played = audioelement.find("./metric/metricresult/[@name='elementFlagListenedTo']") + was_moved = audioelement.find("./metric/metricresult/[@name='elementFlagMoved']") if response.text is not None and len(response.text) > 1: number_of_comments += 1 else: number_of_missing_comments += 1 - + if was_played is not None and was_played.text == 'false': + not_played += 1 + if was_moved is not None and was_moved.text == 'false': + not_moved += 1 + + # update global counters total_empty_comments += number_of_missing_comments - - # 'testTime' keeps total duration: subtract time so far for duration of this audioholder - duration = float(audioholder.find("./metric/metricresult[@id='testTime']").text) - total_duration - - # total duration of test - total_duration += duration + total_not_played += not_played + total_not_moved += not_moved # print audioholder id and duration print " " + page_name + ": " + seconds2timestr(duration) + ", "\ + str(number_of_comments)+"/"\ +str(number_of_comments+number_of_missing_comments)+" comments" + # number of audio elements not played + if not_played > 1: + print 'ATTENTION: '+str(not_played)+' fragments were not listened to!' + if not_played == 1: + print 'ATTENTION: one fragment was not listened to!' + + # number of audio element markers not moved + if not_moved > 1: + print 'ATTENTION: '+str(not_moved)+' markers were not moved!' + if not_moved == 1: + print 'ATTENTION: one marker was not moved!' + # keep track of duration in function of page index if len(duration_order)>page_number: duration_order[page_number].append(duration) @@ -124,8 +159,15 @@ print "Number of XML files: " + str(number_of_XML_files) print "Number of pages: " + str(number_of_pages) print "Number of fragments: " + str(number_of_fragments) -print "Number of empty comments: " + str(total_empty_comments) +print "Number of empty comments: " + str(total_empty_comments) +\ + " (" + str(round(100.0*total_empty_comments/number_of_fragments,2)) + "%)" +print "Number of unplayed fragments: " + str(total_not_played) +\ + " (" + str(round(100.0*total_not_played/number_of_fragments,2)) + "%)" +print "Number of unmoved markers: " + str(total_not_moved) +\ + " (" + str(round(100.0*total_not_moved/number_of_fragments,2)) + "%)" print "Average time per page: " + seconds2timestr(time_per_page_accum/number_of_pages) + +# Pages and number of times tested page_count_strings = list(str(x) for x in page_count) count_list = page_names + page_count_strings count_list[::2] = page_names @@ -133,8 +175,9 @@ print "Pages tested: " + str(count_list) # Average duration for first, second, ... page +print "Average duration per page:" for page_number in range(len(duration_order)): - print "Average duration page " + str(page_number+1) + ": " +\ + print " page " + str(page_number+1) + ": " +\ seconds2timestr(sum(duration_order[page_number])/len(duration_order[page_number])) +\ " ("+str(len(duration_order[page_number]))+" subjects)" @@ -153,8 +196,9 @@ combined_list = sorted(zip(*combined_list), key=operator.itemgetter(1, 2)) # sort # Show average duration for all songs +print "Average duration per audioholder:" for page_index in range(len(page_names)): - print "Average duration audioholder " + combined_list[page_index][0] + ": " \ + print " "+combined_list[page_index][0] + ": " \ + seconds2timestr(combined_list[page_index][1]) \ + " (" + str(combined_list[page_index][3]) + " subjects, " \ + str(combined_list[page_index][2]) + " fragments)" @@ -168,3 +212,5 @@ # show 'count' per page (in order) # clear up page_index <> page_count <> page_number confusion + +# LaTeX -> PDF print out
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/scripts/generate_report.py Tue Sep 15 10:20:50 2015 +0100 @@ -0,0 +1,508 @@ +#!/usr/bin/python +# -*- coding: utf-8 -*- + +import xml.etree.ElementTree as ET +import os # for getting files from directory +import operator # for sorting data with multiple keys +import sys # for accessing command line arguments +import subprocess # for calling pdflatex +import shlex # for calling pdflatex +import matplotlib.pyplot as plt # plots +import numpy as np # numbers + +# Command line arguments +assert len(sys.argv)<4, "evaluation_stats takes at most 2 command line argument\n"+\ + "Use: python generate_report.py [results_folder] [no_render | -nr]" + +render_figures = True + +# XML results files location +if len(sys.argv) == 1: + folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder + print "Use: python generate_report.py [results_folder] [no_render | -nr]" + print "Using default path: " + folder_name +elif len(sys.argv) == 2: + folder_name = sys.argv[1] # First command line argument is folder +elif len(sys.argv) == 3: + folder_name = sys.argv[1] # First command line argument is folder + assert sys.argv[2] in ('no_render','-nr'), "Second argument not recognised. \n" +\ + "Use: python generate_report.py [results_folder] [no_render | -nr]" + # Second command line argument is [no_render | -nr] + render_figures = False + +# Turn number of seconds (int) to '[minutes] min [seconds] s' (string) +def seconds2timestr(time_in_seconds): + time_in_minutes = int(time_in_seconds/60) + remaining_seconds = int(time_in_seconds%60) + return str(time_in_minutes) + " min " + str(remaining_seconds) + " s" + +# stats initialisation +number_of_XML_files = 0 +number_of_pages = 0 +number_of_fragments = 0 +total_empty_comments = 0 +total_not_played = 0 +total_not_moved = 0 +time_per_page_accum = 0 + +# arrays initialisation +page_names = [] +real_page_names = [] # regardless of differing numbers of fragments +subject_count = [] # subjects per audioholder name +page_count = [] +duration_page = [] # duration of experiment in function of page content +duration_order = [] # duration of experiment in function of page number +fragments_per_page = [] # number of fragments for corresponding page + +# survey stats +gender = [] +age = [] + +# get username if available +for name in ('LOGNAME', 'USER', 'LNAME', 'USERNAME'): + user = os.environ.get(name) + if user: + break + else: + user = '' + + +# begin LaTeX document +header = r'''\documentclass[11pt, oneside]{article} + \usepackage{geometry} + \geometry{a4paper} + \usepackage[parfill]{parskip} % empty line instead of indent + \usepackage{graphicx} % figures + \usepackage{hyperref} + \usepackage{tikz} % pie charts + \title{Report} + \author{'''+\ + user+\ + r'''} + \graphicspath{{'''+\ + folder_name+\ + r'''/}} + %\setcounter{section}{-1} % Summary section 0 so number of sections equals number of files + \begin{document} + \maketitle + This is an automatically generated report using the `generate\_report.py' Python script + included with the Web Audio Evaluation Tool \cite{WAET} distribution which can be found + at \texttt{code.soundsoftware.ac.uk/projects/webaudioevaluationtool}. + \tableofcontents + + ''' + +footer = r'''\begin{thebibliography}{9} + \bibitem{WAET} % reference to accompanying publication + Nicholas Jillings, Brecht De Man, David Moffat and Joshua D. Reiss, + ``Web Audio Evaluation Tool: A browser-based listening test environment,'' + presented at the 12th Sound and Music Computing Conference, July 2015. + \end{thebibliography} + \end{document}''' + +body = '' + +# generate images for later use +if render_figures: + subprocess.call("python timeline_view_movement.py", shell=True) + subprocess.call("python score_parser.py", shell=True) + subprocess.call("python score_plot.py", shell=True) + +# get every XML file in folder +files_list = os.listdir(folder_name) +for file in files_list: # iterate over all files in files_list + if file.endswith(".xml"): # check if XML file + number_of_XML_files += 1 + tree = ET.parse(folder_name + '/' + file) + root = tree.getroot() + + # PRINT name as section + body+= '\section{'+file[:-4].capitalize()+'}\n' # make section header from name without extension + + # reset for new subject + total_duration = 0 + page_number = 0 + + individual_table = '' # table with stats for this individual test file + timeline_plots = '' # plots of timeline (movements and plays) + + # DEMO survey stats + # get gender + this_subjects_gender = root.find("./posttest/radio/[@id='gender']") + if this_subjects_gender is not None: + gender.append(this_subjects_gender.get("name")) + else: + gender.append('UNAVAILABLE') + # get age + this_subjects_age = root.find("./posttest/number/[@id='age']") + if this_subjects_age is not None: + age.append(this_subjects_age.text) + + # get list of all page names + for audioholder in root.findall("./audioholder"): # iterate over pages + page_name = audioholder.get('id') # get page name + + if page_name is None: # ignore 'empty' audio_holders + print "WARNING: " + file + " contains empty audio holder. (evaluation_stats.py)" + break # move on to next + + number_of_comments = 0 # for this page + number_of_missing_comments = 0 # for this page + not_played = [] # for this page + not_moved = [] # for this page + + # 'testTime' keeps total duration: subtract time so far for duration of this audioholder + duration = float(audioholder.find("./metric/metricresult[@id='testTime']").text) - total_duration + + # total duration of test + total_duration += duration + + # number of audio elements + audioelements = audioholder.findall("./audioelement") # get audioelements + number_of_fragments += len(audioelements) # add length of this list to total + + # number of comments (interesting if comments not mandatory) + for audioelement in audioelements: + response = audioelement.find("./comment/response") + was_played = audioelement.find("./metric/metricresult/[@name='elementFlagListenedTo']") + was_moved = audioelement.find("./metric/metricresult/[@name='elementFlagMoved']") + if response.text is not None and len(response.text) > 1: + number_of_comments += 1 + else: + number_of_missing_comments += 1 + if was_played is not None and was_played.text == 'false': + not_played.append(audioelement.get('id')) + if was_moved is not None and was_moved.text == 'false': + not_moved.append(audioelement.get('id')) + + # update global counters + total_empty_comments += number_of_missing_comments + total_not_played += len(not_played) + total_not_moved += len(not_moved) + + # PRINT alerts when elements not played or markers not moved + # number of audio elements not played + if len(not_played) > 1: + body += '\\emph{\\textbf{ATTENTION: '+str(len(not_played))+\ + ' fragments were not listened to in '+page_name+'! }}'+\ + ', '.join(not_played)+'\\\\ \n' + if len(not_played) == 1: + body += '\\emph{\\textbf{ATTENTION: one fragment was not listened to in '+page_name+'! }}'+\ + not_played[0]+'\\\\ \n' + + # number of audio element markers not moved + if len(not_moved) > 1: + body += '\\emph{\\textbf{ATTENTION: '+str(len(not_moved))+\ + ' markers were not moved in '+page_name+'! }}'+\ + ', '.join(not_moved)+'\\\\ \n' + if len(not_moved) == 1: + body += '\\emph{\\textbf{ATTENTION: one marker was not moved in '+page_name+'! }}'+\ + not_moved[0]+'\\\\ \n' + + # PRINT song-specific statistic + individual_table += page_name+'&'+\ + str(number_of_comments) + '/' +\ + str(number_of_comments+number_of_missing_comments)+'&'+\ + seconds2timestr(duration)+'\\\\' + + # get timeline for this audioholder + img_path = 'timelines_movement/'+file[:-4]+'-'+page_name+'.pdf' + + # check if available + if os.path.isfile(folder_name+'/'+img_path): + # SHOW timeline image + timeline_plots += '\\includegraphics[width=\\textwidth]{'+\ + folder_name+'/'+img_path+'}\n\n' + + # keep track of duration in function of page index + if len(duration_order)>page_number: + duration_order[page_number].append(duration) + else: + duration_order.append([duration]) + + # keep list of audioholder ids and count how many times each audioholder id + # was tested, how long it took, and how many fragments there were + # (if number of fragments is different, store as different audioholder id) + if page_name in page_names: + page_index = page_names.index(page_name) # get index + # check if number of audioelements the same + if len(audioelements) == fragments_per_page[page_index]: + page_count[page_index] += 1 + duration_page[page_index].append(duration) + else: # make new entry + alt_page_name = page_name+"("+str(len(audioelements))+")" + if alt_page_name in page_names: # if already there + alt_page_index = page_names.index(alt_page_name) # get index + page_count[alt_page_index] += 1 + duration_page[alt_page_index].append(duration) + else: + page_names.append(alt_page_name) + page_count.append(1) + duration_page.append([duration]) + fragments_per_page.append(len(audioelements)) + else: + page_names.append(page_name) + page_count.append(1) + duration_page.append([duration]) + fragments_per_page.append(len(audioelements)) + + # number of subjects per audioholder regardless of differing numbers of + # fragments (for inclusion in box plots) + if page_name in real_page_names: + page_index = real_page_names.index(page_name) # get index + subject_count[page_index] += 1 + else: + real_page_names.append(page_name) + subject_count.append(1) + + # bookkeeping + page_number += 1 # increase page count for this specific test + number_of_pages += 1 # increase total number of pages + time_per_page_accum += duration # total duration (for average time spent per page) + + # PRINT table with statistics about this test + body += r'''\begin{tabular}{|p{3.5cm}|c|p{2.5cm}|} + \hline + \textbf{Song name} & \textbf{Comments} & \textbf{Duration} \\ \hline '''+\ + individual_table+\ + r'''\hline + \textbf{TOTAL} & & \textbf{'''+\ + seconds2timestr(total_duration)+\ + r'''}\\ + \hline + \end{tabular} + + ''' + # PRINT timeline plots + body += timeline_plots + +# join to footer +footer = body + footer + +# empty body again +body = '' + +# PRINT summary of everything (at start) +# unnumbered so that number of sections equals number of files +body += '\section*{Summary}\n\\addcontentsline{toc}{section}{Summary}' + +# PRINT table with statistics +body += '\\begin{tabular}{ll}' +body += r'Number of XML files: &' + str(number_of_XML_files) + r'\\' +body += r'Number of pages: &' + str(number_of_pages) + r'\\' +body += r'Number of fragments: &' + str(number_of_fragments) + r'\\' +body += r'Number of empty comments: &' + str(total_empty_comments) +\ + " (" + str(round(100.0*total_empty_comments/number_of_fragments,2)) + r"\%)\\" +body += r'Number of unplayed fragments: &' + str(total_not_played) +\ + " (" + str(round(100.0*total_not_played/number_of_fragments,2)) + r"\%)\\" +body += r'Number of unmoved markers: &' + str(total_not_moved) +\ + " (" + str(round(100.0*total_not_moved/number_of_fragments,2)) + r"\%)\\" +body += r'Average time per page: &' + seconds2timestr(time_per_page_accum/number_of_pages) + r"\\" +body += '\\end{tabular} \\vspace{1.5cm} \\\\ \n' + +# Average duration for first, second, ... page +body += " \\vspace{.5cm} Average duration per page (see also Figure \\ref{fig:avgtimeperpage}): \\\\ \n" +body += r'''\begin{tabular}{lll} + \textbf{Page} & \textbf{Duration} & \textbf{\# subjects}\\ + ''' +tpp_averages = [] # store average time per page +for page_number in range(len(duration_order)): + body += str(page_number+1) + "&" +\ + seconds2timestr(sum(duration_order[page_number])/len(duration_order[page_number])) +\ + "&"+str(len(duration_order[page_number]))+r"\\" + tpp_averages.append(sum(duration_order[page_number])/len(duration_order[page_number])) + +body += '\\end{tabular} \\vspace{1.5cm} \\\\ \n' + +# SHOW bar plot of average time per page +plt.bar(range(1,len(duration_order)+1), np.array(tpp_averages)/60) +plt.xlabel('Page order') +plt.xlim(.8, len(duration_order)+1) +plt.xticks(np.arange(1,len(duration_order)+1)+.4, range(1,len(duration_order)+1)) +plt.ylabel('Average time [minutes]') +plt.savefig(folder_name+"/time_per_page.pdf", bbox_inches='tight') +plt.close() +#TODO add error bars + + +# Sort pages by number of audioelements, then by duration + +# average duration and number of subjects per page +average_duration_page = [] +number_of_subjects_page = [] +for line in duration_page: + number_of_subjects_page.append(len(line)) + average_duration_page.append(sum(line)/len(line)) + +# combine and sort in function of number of audioelements and duration +combined_list = [page_names, average_duration_page, fragments_per_page, number_of_subjects_page] +combined_list = sorted(zip(*combined_list), key=operator.itemgetter(1, 2)) # sort + +# Show average duration for all songs +body += r'''\vspace{.5cm} Average duration per audioholder (see also Figure \ref{fig:avgtimeperaudioholder}): \\ + \begin{tabular}{llll} + \textbf{Audioholder} & \textbf{Duration} & \textbf{\# subjects} & \textbf{\# fragments} \\ + ''' +audioholder_names_ordered = [] +average_duration_audioholder_ordered = [] +number_of_subjects = [] +for page_index in range(len(page_names)): + audioholder_names_ordered.append(combined_list[page_index][0]) + average_duration_audioholder_ordered.append(combined_list[page_index][1]) + number_of_subjects.append(combined_list[page_index][3]) + body += combined_list[page_index][0] + "&" +\ + seconds2timestr(combined_list[page_index][1]) + "&" +\ + str(combined_list[page_index][3]) + "&" +\ + str(combined_list[page_index][2]) + r"\\" +body += '\\end{tabular}\n' + +# SHOW bar plot of average time per page +plt.bar(range(1,len(audioholder_names_ordered)+1), np.array(average_duration_audioholder_ordered)/60) +plt.xlabel('Audioholder') +plt.xlim(.8, len(audioholder_names_ordered)+1) +plt.xticks(np.arange(1,len(audioholder_names_ordered)+1)+.4, audioholder_names_ordered, rotation=90) +plt.ylabel('Average time [minutes]') +plt.savefig(folder_name+"/time_per_audioholder.pdf", bbox_inches='tight') +plt.close() + +# SHOW bar plot of average time per page +plt.bar(range(1,len(audioholder_names_ordered)+1), number_of_subjects) +plt.xlabel('Audioholder') +plt.xlim(.8, len(audioholder_names_ordered)+1) +plt.xticks(np.arange(1,len(audioholder_names_ordered)+1)+.4, audioholder_names_ordered, rotation=90) +plt.ylabel('Number of subjects') +ax = plt.gca() +ylims = ax.get_ylim() +yint = np.arange(int(np.floor(ylims[0])), int(np.ceil(ylims[1]))+1) +plt.yticks(yint) +plt.savefig(folder_name+"/subjects_per_audioholder.pdf", bbox_inches='tight') +plt.close() + +# SHOW both figures +body += r'''\begin{figure}[htbp] + \begin{center} + \includegraphics[width=.65\textwidth]{'''+\ + folder_name+"/time_per_page.pdf"+\ + r'''} + \caption{Average time spent per page.} + \label{fig:avgtimeperpage} + \end{center} + \end{figure} + ''' +body += r'''\begin{figure}[htbp] + \begin{center} + \includegraphics[width=.65\textwidth]{'''+\ + folder_name+"/time_per_audioholder.pdf"+\ + r'''} + \caption{Average time spent per audioholder.} + \label{fig:avgtimeperaudioholder} + \end{center} + \end{figure} + ''' +body += r'''\begin{figure}[htbp] + \begin{center} + \includegraphics[width=.65\textwidth]{'''+\ + folder_name+"/subjects_per_audioholder.pdf"+\ + r'''} + \caption{Number of subjects per audioholder.} + \label{fig:avgtimeperaudioholder} + \end{center} + \end{figure} + ''' +#TODO add error bars +#TODO layout of figures + +# SHOW boxplot per audioholder +#TODO order in decreasing order of participants +for audioholder_name in page_names: # get each name + # plot boxplot if exists (not so for the 'alt' names) + if os.path.isfile(folder_name+'/ratings/'+audioholder_name+'-ratings-box.pdf'): + body += r'''\begin{figure}[htbp] + \begin{center} + \includegraphics[width=.65\textwidth]{'''+\ + folder_name+"/ratings/"+audioholder_name+'-ratings-box.pdf'+\ + r'''} + \caption{Box plot of ratings for audioholder '''+\ + audioholder_name+' ('+str(subject_count[real_page_names.index(audioholder_name)])+\ + ''' participants).} + \label{fig:avgtimeperpage} + \end{center} + \end{figure} + ''' + +# DEMO pie chart of gender distribution among subjects +genders = ['male', 'female', 'other', 'preferNotToSay', 'UNAVAILABLE'] +# TODO: get the above automatically +gender_distribution = '' +for item in genders: + number = gender.count(item) + if number>0: + gender_distribution += str("{:.2f}".format((100.0*number)/len(gender)))+\ + '/'+item.capitalize()+' ('+str(number)+'),\n' + +body += r''' + \def\angle{0} + \def\radius{3} + \def\cyclelist{{"orange","blue","red","green"}} + \newcount\cyclecount \cyclecount=-1 + \newcount\ind \ind=-1 + \begin{figure}[htbp] + \begin{center}\begin{tikzpicture}[nodes = {font=\sffamily}] + \foreach \percent/\name in {'''+\ + gender_distribution+\ + r'''} {\ifx\percent\empty\else % If \percent is empty, do nothing + \global\advance\cyclecount by 1 % Advance cyclecount + \global\advance\ind by 1 % Advance list index + \ifnum6<\cyclecount % If cyclecount is larger than list + \global\cyclecount=0 % reset cyclecount and + \global\ind=0 % reset list index + \fi + \pgfmathparse{\cyclelist[\the\ind]} % Get color from cycle list + \edef\color{\pgfmathresult} % and store as \color + % Draw angle and set labels + \draw[fill={\color!50},draw={\color}] (0,0) -- (\angle:\radius) + arc (\angle:\angle+\percent*3.6:\radius) -- cycle; + \node at (\angle+0.5*\percent*3.6:0.7*\radius) {\percent\,\%}; + \node[pin=\angle+0.5*\percent*3.6:\name] + at (\angle+0.5*\percent*3.6:\radius) {}; + \pgfmathparse{\angle+\percent*3.6} % Advance angle + \xdef\angle{\pgfmathresult} % and store in \angle + \fi + }; + \end{tikzpicture} + \caption{Representation of gender across subjects} + \label{default} + \end{center} + \end{figure} + ''' +# problem: some people entered twice? + +#TODO +# time per page in function of number of fragments (plot) +# time per participant in function of number of pages +# plot total time for each participant +# show 'count' per page (in order) + +# clear up page_index <> page_count <> page_number confusion + + +texfile = header+body+footer # add bits together + +# write TeX file +with open(folder_name + '/' + 'Report.tex','w') as f: + f.write(texfile) +proc=subprocess.Popen(shlex.split('pdflatex -output-directory='+folder_name+' '+ folder_name + '/Report.tex')) +proc.communicate() +# run again +proc=subprocess.Popen(shlex.split('pdflatex -output-directory='+folder_name+' '+ folder_name + '/Report.tex')) +proc.communicate() + +#TODO remove auxiliary LaTeX files +try: + os.remove(folder_name + '/' + 'Report.aux') + os.remove(folder_name + '/' + 'Report.log') + os.remove(folder_name + '/' + 'Report.out') + os.remove(folder_name + '/' + 'Report.toc') +except OSError: + pass + \ No newline at end of file
--- a/scripts/score_parser.py Tue Sep 15 10:16:34 2015 +0100 +++ b/scripts/score_parser.py Tue Sep 15 10:20:50 2015 +0100 @@ -2,19 +2,42 @@ import xml.etree.ElementTree as ET import os +import sys import csv -#TODO Remove DEBUG statements +# COMMAND LINE ARGUMENTS -# XML results files location (modify as needed): -folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder +assert len(sys.argv)<3, "score_parser takes at most 1 command line argument\n"+\ + "Use: python score_parser.py [rating_folder_location]" + +# XML results files location +if len(sys.argv) == 1: + folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder + print "Use: python score_parser.py [rating_folder_location]" + print "Using default path: " + folder_name +elif len(sys.argv) == 2: + folder_name = sys.argv[1] # First command line argument is folder + +# check if folder_name exists +if not os.path.exists(folder_name): + #the file is not there + print "Folder '"+folder_name+"' does not exist." + sys.exit() # terminate script execution +elif not os.access(os.path.dirname(folder_name), os.W_OK): + #the file does exist but write privileges are not given + print "No write privileges in folder '"+folder_name+"'." + + +# CODE + +# remember which files have been opened this time +file_history = [] # get every XML file in folder for file in os.listdir(folder_name): if file.endswith(".xml"): tree = ET.parse(folder_name + '/' + file) root = tree.getroot() - #print "DEBUG Reading " + file + "..." # get subject ID from XML file subject_id = file[:-4] # file name (without extension) as subject ID @@ -29,7 +52,7 @@ file_name = folder_name+'/ratings/'+page_name+'-ratings.csv' # score file name - # create folder 'ratings if not yet created + # create folder 'ratings' if not yet created if not os.path.exists(folder_name + '/ratings'): os.makedirs(folder_name + '/ratings') @@ -45,38 +68,38 @@ for audioelement in audiolist: # iterate over all audioelements fragmentnamelist.append(audioelement.get('id')) # add to list - # if file exists, get header and add 'new' fragments if os.path.isfile(file_name): - #print "DEBUG file " + file_name + " already exists - reading header" with open(file_name, 'r') as readfile: filereader = csv.reader(readfile, delimiter=',') headerrow = filereader.next() + # If file hasn't been opened yet this time, remove all rows except header + if file_name not in file_history: + with open(file_name, 'w') as writefile: + filewriter = csv.writer(writefile, delimiter=',') + headerrow = sorted(headerrow) + filewriter.writerow(headerrow) + file_history.append(file_name) + # Which of the fragmentes are in fragmentnamelist but not in headerrow? newfragments = list(set(fragmentnamelist)-set(headerrow)) newfragments = sorted(newfragments) # new fragments in alphabetical order # If not empty, read file and rewrite adding extra columns if newfragments: # if not empty - #print "DEBUG New fragments found: " + str(newfragments) with open('temp.csv', 'w') as writefile: filewriter = csv.writer(writefile, delimiter=',') filewriter.writerow(headerrow + newfragments) # write new header - #print " "+str(headerrow + newfragments) # DEBUG with open(file_name, 'r') as readfile: filereader = csv.reader(readfile, delimiter=',') filereader.next() # skip header for row in filereader: # rewrite row plus empty cells for every new fragment name - #print " Old row: " + str(row) # DEBUG filewriter.writerow(row + ['']*len(newfragments)) - #print " New row: " + str(row + ['']*len(newfragments)) # DEBUG os.rename('temp.csv', file_name) # replace old file with temp file headerrow = headerrow + newfragments - #print "DEBUG New header row: " + str(headerrow) # if not, create file and make header else: - #print ["DEBUG file " + file_name + " doesn't exist yet - making new one"] headerrow = sorted(fragmentnamelist) # sort alphabetically headerrow.insert(0,'') fragmentnamelist = fragmentnamelist[1:] #HACKY FIX inserting in firstrow also affects fragmentnamelist @@ -104,4 +127,3 @@ # write row: [subject ID, rating fragment ID 1, ..., rating fragment ID M] if any(ratingrow[1:]): # append to file if row non-empty (except subject name) filewriter.writerow(ratingrow) -
--- a/scripts/score_plot.py Tue Sep 15 10:16:34 2015 +0100 +++ b/scripts/score_plot.py Tue Sep 15 10:20:50 2015 +0100 @@ -8,20 +8,103 @@ import scipy as sp import scipy.stats -# CONFIGURATION +# COMMAND LINE ARGUMENTS -# Which type(s) of plot do you want? -enable_boxplot = True # show box plot +#TODO: Merge, implement this functionality +#TODO: Control by CLI arguments (plot types, save and/or show, ...) + +assert len(sys.argv)<4, "score_plot takes at most 2 command line arguments\n"+\ + "Use: python score_plot.py [ratings_folder_location]."+\ + "Type 'python score_plot.py -h' for more options" + +# initialise plot types (false by default) and options +enable_boxplot = False # show box plot enable_confidence = False # show confidence interval confidence = 0.90 # confidence value (for confidence interval plot) enable_individual = False # show all individual ratings -show_individual = [] # show specific individuals +show_individual = [] # show specific individuals (empty: show all individuals found) show_legend = False # show names of individuals -#TODO: Merge, implement this functionality -#TODO: Control by CLI arguments (plot types, save and/or show, ...) -# Enter folder where rating CSV files are (generated with score_parser.py or same format). -rating_folder = '../saves/ratings/' # folder with rating csv files +# DEFAULT: Looks in 'saves/ratings/' folder from 'scripts/' folder +rating_folder = "../saves/ratings/" + +# XML results files location +if len(sys.argv) == 1: # no extra arguments + enable_boxplot = True # show box plot + print "Use: python score_plot.py [rating folder] [plot_type] [-l/-legend]" + print "Type 'python score_plot.py -h' for help." + print "Using default path: " + rating_folder + " with boxplot." +else: + for arg in sys.argv: # go over all arguments + if arg == '-h': + # show help + #TODO: replace with contents of helpfile score_plot.info (or similar) + print "Use: python score_plot.py [rating_folder] [plot_type] [-l] [confidence]" + print " rating_folder:" + print " folder where output of 'score_parser' can be found, and" + print " where plots will be stored." + print " By default, '../saves/ratings/' is used." + print "" + print "PLOT TYPES" + print " Can be used in combination." + print " box | boxplot | -b" + print " Enables the boxplot" + print " conf | confidence | -c" + print " Enables the confidence interval plot" + print " ind | individual | -i" + print " Enables plot of individual ratings" + print "" + print "PLOT OPTIONS" + print " leg | legend | -l" + print " For individual plot: show legend with individual file names" + print " numeric value between 0 and 1, e.g. 0.95" + print " For confidence interval plot: confidence value" + assert False, ""# stop immediately after showing help #TODO cleaner way + + # PLOT TYPES + elif arg == 'box' or arg == 'boxplot' or arg == '-b': + enable_boxplot = True # show box plot + elif arg == 'conf' or arg == 'confidence' or arg == '-c': + enable_confidence = True # show confidence interval + #TODO add confidence value input + elif arg == 'ind' or arg == 'individual' or arg == '-i': + enable_individual = True # show all individual ratings + + # PLOT OPTIONS + elif arg == 'leg' or arg == 'legend' or arg == '-l': + if not enable_individual: + print "WARNING: The 'legend' option is only relevant to plots of "+\ + "individual ratings" + show_legend = True # show all individual ratings + elif arg.isnumeric(): + if not enable_confidence: + print "WARNING: The numeric confidence value is only relevant when "+\ + "confidence plot is enabled" + if float(arg)>0 and float(arg)<1: + confidence = float(arg) + else: + print "WARNING: The confidence value needs to be between 0 and 1" + + # FOLDER NAME + else: + # assume it's the folder name + rating_folder = arg + +# at least one plot type should be selected: box plot by default +if not enable_boxplot and not enable_confidence and not enable_individual: + enable_boxplot = True + +# check if folder_name exists +if not os.path.exists(rating_folder): + #the file is not there + print "Folder '"+rating_folder+"' does not exist." + sys.exit() # terminate script execution +elif not os.access(os.path.dirname(rating_folder), os.W_OK): + #the file does exist but write rating_folder are not given + print "No write privileges in folder '"+rating_folder+"'." + + +# CONFIGURATION # Font settings font = {'weight' : 'bold', @@ -131,7 +214,7 @@ plt.title(page_name) plt.xlabel('Fragment') plt.xlim(0, len(headerrow)+1) # only show relevant region, leave space left & right) - plt.xticks(range(1, len(headerrow)+1), headerrow) # show fragment names + plt.xticks(range(1, len(headerrow)+1), headerrow, rotation=90) # show fragment names plt.ylabel('Rating') plt.ylim(0,1) @@ -146,5 +229,5 @@ plot_type = ("-box" if enable_boxplot else "") + \ ("-conf" if enable_confidence else "") + \ ("-ind" if enable_individual else "") - plt.savefig(rating_folder+page_name+plot_type+".png") + plt.savefig(rating_folder+page_name+plot_type+".pdf", bbox_inches='tight') plt.close()
--- a/scripts/timeline_view.py Tue Sep 15 10:16:34 2015 +0100 +++ b/scripts/timeline_view.py Tue Sep 15 10:20:50 2015 +0100 @@ -1,14 +1,36 @@ #!/usr/bin/python import xml.etree.ElementTree as ET -import os -import matplotlib.pyplot as plt +import os # list files in directory +import sys # command line arguments +import matplotlib.pyplot as plt # plots +import matplotlib.patches as patches # rectangles + +# COMMAND LINE ARGUMENTS + +assert len(sys.argv)<3, "timeline_view takes at most 1 command line argument\n"+\ + "Use: python timeline_view.py [XML_files_location]" + +# XML results files location +if len(sys.argv) == 1: + folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder + print "Use: python timeline_view.py [XML_files_location]" + print "Using default path: " + folder_name +elif len(sys.argv) == 2: + folder_name = sys.argv[1] # First command line argument is folder + +# check if folder_name exists +if not os.path.exists(folder_name): + #the file is not there + print "Folder '"+folder_name+"' does not exist." + sys.exit() # terminate script execution +elif not os.access(os.path.dirname(folder_name), os.W_OK): + #the file does exist but write privileges are not given + print "No write privileges in folder '"+folder_name+"'." + # CONFIGURATION -# XML results files location (modify as needed): -folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder - # Folder where to store timelines timeline_folder = folder_name + '/timelines/' # Stores in 'saves/timelines/' @@ -20,6 +42,16 @@ # Colormap for to cycle through colormap = ['b', 'r', 'g', 'c', 'm', 'y', 'k'] +# if enabled, x-axis shows time per audioholder, not total test time +show_audioholder_time = True + +# bar height (<1 to avoid overlapping) +bar_height = 0.6 + +# figure size +fig_width = 25 +fig_height = 5 + # CODE @@ -34,11 +66,14 @@ root = tree.getroot() subject_id = file[:-4] # drop '.xml' + time_offset = 0 # test starts at zero + # ONE TIMELINE PER PAGE - make new plot per page # get list of all page names for audioholder in root.findall("./audioholder"): # iterate over pages page_name = audioholder.get('id') # get page name + plot_empty = True # check if any data is plotted if page_name is None: # ignore 'empty' audio_holders break @@ -56,9 +91,13 @@ increment = 0 # increased for every new audioelement audioelements_names = [] # store names of audioelements + # get axes handle + fig = plt.figure(figsize=(fig_width, fig_height)) + ax = fig.add_subplot(111) #, aspect='equal' + # for page [page_name], print comments related to fragment [id] for tuple in data: - audioelement = tuple[1] + audioelement = tuple[1] if audioelement is not None: # Check it exists audio_id = str(audioelement.get('id')) audioelements_names.append(audio_id) @@ -66,41 +105,55 @@ # for this audioelement, loop over all listen events listen_events = audioelement.findall("./metric/metricresult/[@name='elementListenTracker']/event") for event in listen_events: + # mark this plot as not empty + plot_empty = False + # get testtime: start and stop - start_time = event.find('testtime').get('start') - stop_time = event.find('testtime').get('stop') + start_time = float(event.find('testtime').get('start'))-time_offset + stop_time = float(event.find('testtime').get('stop'))-time_offset # event lines: - plt.plot([start_time, start_time], # x-values + ax.plot([start_time, start_time], # x-values [0, N_audioelements+1], # y-values color='k' ) - plt.plot([stop_time, stop_time], # x-values + ax.plot([stop_time, stop_time], # x-values [0, N_audioelements+1], # y-values color='k' ) # plot time: - plt.plot([start_time, stop_time], # x-values - [N_audioelements-increment, N_audioelements-increment], # y-values - color=colormap[increment%len(colormap)], - linewidth=6 + ax.add_patch( + patches.Rectangle( + (start_time, N_audioelements-increment-bar_height/2), # (x, y) + stop_time - start_time, # width + bar_height, # height + color=colormap[increment%len(colormap)] # colour ) + ) - increment+=1 - + increment+=1 # to next audioelement + + # subtract total audioholder length from subsequent audioholder event times + audioholder_time = audioholder.find("./metric/metricresult/[@id='testTime']") + if audioholder_time is not None and show_audioholder_time: + time_offset = float(audioholder_time.text) + + if not plot_empty: + # set plot parameters + plt.title('Timeline ' + file + ": "+page_name) + plt.xlabel('Time [seconds]') + plt.ylabel('Fragment') + plt.ylim(0, N_audioelements+1) + + #y-ticks: fragment IDs, top to bottom + plt.yticks(range(N_audioelements, 0, -1), audioelements_names) # show fragment names + + + #plt.show() # uncomment to show plot; comment when just saving + #exit() + + plt.savefig(timeline_folder+subject_id+"-"+page_name+".pdf", bbox_inches='tight') + plt.close() + #TODO: if 'nonsensical' or unknown: dashed line until next event #TODO: Vertical lines for fragment looping point - - plt.title('Timeline ' + file) #TODO add song too - plt.xlabel('Time [seconds]') - plt.ylabel('Fragment') - plt.ylim(0, N_audioelements+1) - - #y-ticks: fragment IDs, top to bottom - plt.yticks(range(N_audioelements, 0, -1), audioelements_names) # show fragment names - - - #plt.show() # uncomment to show plot; comment when just saving - #exit() - - plt.savefig(timeline_folder+subject_id+"-"+page_name+".png") - plt.close() \ No newline at end of file + \ No newline at end of file
--- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/scripts/timeline_view_movement.py Tue Sep 15 10:20:50 2015 +0100 @@ -0,0 +1,257 @@ +#!/usr/bin/python + +import xml.etree.ElementTree as ET +import os # list files in directory +import sys # command line arguments +import matplotlib.pyplot as plt # plots +import matplotlib.patches as patches # rectangles + + +# COMMAND LINE ARGUMENTS + +assert len(sys.argv)<3, "timeline_view_movement takes at most 1 command line argument\n"+\ + "Use: python timeline_view_movement.py [XML_files_location]" + +# XML results files location +if len(sys.argv) == 1: + folder_name = "../saves" # Looks in 'saves/' folder from 'scripts/' folder + print "Use: python timeline_view_movement.py [XML_files_location]" + print "Using default path: " + folder_name +elif len(sys.argv) == 2: + folder_name = sys.argv[1] # First command line argument is folder + +# check if folder_name exists +if not os.path.exists(folder_name): + #the file is not there + print "Folder '"+folder_name+"' does not exist." + sys.exit() # terminate script execution +elif not os.access(os.path.dirname(folder_name), os.W_OK): + #the file does exist but write privileges are not given + print "No write privileges in folder '"+folder_name+"'." + + +# CONFIGURATION + +# Folder where to store timelines +timeline_folder = folder_name + '/timelines_movement/' # Stores in 'saves/timelines_movement/' by default + +# Font settings +font = {'weight' : 'bold', + 'size' : 16} +plt.rc('font', **font) + +# Colormap for to cycle through +colormap = ['b', 'g', 'c', 'm', 'y', 'k'] + +# figure size +fig_width = 25 +fig_height = 10 + + +# CODE + +# create timeline_folder if not yet created +if not os.path.exists(timeline_folder): + os.makedirs(timeline_folder) + +# get every XML file in folder +for file in os.listdir(folder_name): + if file.endswith(".xml"): + tree = ET.parse(folder_name + '/' + file) + root = tree.getroot() + subject_id = file[:-4] # drop '.xml' + + previous_audioholder_time = 0 # time spent before current audioholder + time_offset = 0 # test starts at zero + + # ONE TIMELINE PER PAGE - make new plot per page + + # get list of all page names + for audioholder in root.findall("./audioholder"): # iterate over pages + page_name = audioholder.get('id') # get page name + plot_empty = True # check if any data is plotted + + if page_name is None: # ignore 'empty' audio_holders + print "Skipping empty audioholder name from "+subject_id+"." + break + + # subtract total audioholder length from subsequent audioholder event times + audioholder_time_temp = audioholder.find("./metric/metricresult/[@id='testTime']") + if audioholder_time_temp is not None: + audioholder_time = float(audioholder_time_temp.text) + else: + print "Skipping audioholder without total time specified from "+subject_id+"." + break + + # get audioelements + audioelements = audioholder.findall("./audioelement") + + # sort alphabetically + data = [] + for elem in audioelements: # from http://effbot.org/zone/element-sort.htm + key = elem.get("id") + data.append((key, elem)) + data.sort() + + N_audioelements = len(audioelements) # number of audio elements for this page + increment = 0 # increased for every new audioelement + + # get axes handle + fig = plt.figure(figsize=(fig_width, fig_height)) + ax = fig.add_subplot(111) + + # for page [page_name], print comments related to fragment [id] + #for tuple in data: + # audioelement = tuple[1] + for tuple in data: + audioelement = tuple[1] + if audioelement is not None: # Check it exists + audio_id = str(audioelement.get('id')) + + # break if no initial position or move events registered + initial_position_temp = audioelement.find("./metric/metricresult/[@name='elementInitialPosition']") + if initial_position_temp is None: + print "Skipping "+page_name+" from "+subject_id+": does not have initial positions specified." + break + + # get move events, initial and eventual position + initial_position = float(initial_position_temp.text) + move_events = audioelement.findall("./metric/metricresult/[@name='elementTrackerFull']/timepos") + final_position = float(audioelement.find("./value").text) + + # get listen events + start_times_global = [] + stop_times_global = [] + listen_events = audioelement.findall("./metric/metricresult/[@name='elementListenTracker']/event") + for event in listen_events: + # get testtime: start and stop + start_times_global.append(float(event.find('testtime').get('start'))-time_offset) + stop_times_global.append(float(event.find('testtime').get('stop'))-time_offset) + + # display fragment name at start + plt.text(0,initial_position+0.02,audio_id,color=colormap[increment%len(colormap)]) #,rotation=45 + + # previous position and time + previous_position = initial_position + previous_time = 0 + + # assume not playing at start + currently_playing = False # keep track of whether fragment is playing during move event + + # draw all segments except final one + for event in move_events: + # mark this plot as not empty + plot_empty = False + + # get time and final position of move event + new_time = float(event.find("./time").text)-time_offset + new_position = float(event.find("./position").text) + + # get play/stop events since last move until current move event + stop_times = [] + start_times = [] + # is there a play and/or stop event between previous_time and new_time? + for time in start_times_global: + if time>previous_time and time<new_time: + start_times.append(time) + for time in stop_times_global: + if time>previous_time and time<new_time: + stop_times.append(time) + # if no play/stop events between move events, find out whether playing + + segment_start = previous_time # first segment starts at previous move event + + # draw segments (horizontal line) + while len(start_times)+len(stop_times)>0: # while still play/stop events left + if len(stop_times)<1: # upcoming event is 'play' + # draw non-playing segment from segment_start to 'play' + currently_playing = False + segment_stop = start_times.pop(0) # remove and return first item + elif len(start_times)<1: # upcoming event is 'stop' + # draw playing segment (red) from segment_start to 'stop' + currently_playing = True + segment_stop = stop_times.pop(0) # remove and return first item + elif start_times[0]<stop_times[0]: # upcoming event is 'play' + # draw non-playing segment from segment_start to 'play' + currently_playing = False + segment_stop = start_times.pop(0) # remove and return first item + else: # stop_times[0]<start_times[0]: upcoming event is 'stop' + # draw playing segment (red) from segment_start to 'stop' + currently_playing = True + segment_stop = stop_times.pop(0) # remove and return first item + + # draw segment + plt.plot([segment_start, segment_stop], # x-values + [previous_position, previous_position], # y-values + color='r' if currently_playing else colormap[increment%len(colormap)], + linewidth=3 + ) + segment_start = segment_stop # move on to next segment + currently_playing = not currently_playing # toggle to draw final segment correctly + + # draw final segment (horizontal line) from last 'segment_start' to current move event time + plt.plot([segment_start, new_time], # x-values + [previous_position, previous_position], # y-values + # color depends on playing during move event or not: + color='r' if currently_playing else colormap[increment%len(colormap)], + linewidth=3 + ) + + # vertical line from previous to current position + #TODO red if currently playing, orig color if not + plt.plot([new_time, new_time], # x-values + [previous_position, new_position], # y-values + # color depends on playing during move event or not: + color='r' if currently_playing else colormap[increment%len(colormap)], + linewidth=3 + ) + + # update previous_position value + previous_position = new_position + previous_time = new_time + + # draw final segment + # horizontal line from previous time to end of audioholder + plt.plot([previous_time, audioholder_time-time_offset], # x-values + [previous_position, previous_position], # y-values + color=colormap[increment%len(colormap)], + linewidth=3 + ) + + # display fragment name at end + plt.text(audioholder_time-time_offset,previous_position,\ + audio_id,color=colormap[increment%len(colormap)]) #,rotation=45 + + increment+=1 # to next audioelement + + last_audioholder_duration = audioholder_time-time_offset + time_offset = audioholder_time + + if not plot_empty: # if plot is not empty, show or store + # set plot parameters + plt.title('Timeline ' + file + ": "+page_name) + plt.xlabel('Time [seconds]') + plt.xlim(0, last_audioholder_duration) + plt.ylabel('Rating') # default + plt.ylim(0, 1) # rating between 0 and 1 + + #y-ticks: labels on rating axis + label_positions = [] + label_text = [] + scale_tags = root.findall("./BrowserEvalProjectDocument/audioHolder/interface/scale") + scale_title = root.find("./BrowserEvalProjectDocument/audioHolder/interface/title") + for tag in scale_tags: + label_positions.append(float(tag.get('position'))/100) # on a scale from 0 to 100 + label_text.append(tag.text) + if len(label_positions) > 0: # if any labels available + plt.yticks(label_positions, label_text) # show rating axis labels + # set label Y-axis + if scale_title is not None: + plt.ylabel(scale_title.text) + + #plt.show() # uncomment to show plot; comment when just saving + #exit() + + plt.savefig(timeline_folder+subject_id+"-"+page_name+".pdf", bbox_inches='tight') + plt.close() + \ No newline at end of file