def collectUserFeatures(data, user, featureFunction):
directory = data + '/' + user
features = []
for site in os.listdir(directory):
path = directory + '/' + site
stream = util.filterKeystrokes(util.openStream(path))
features.append(featureFunction(stream))
return chain(*features)
def collectUserFeatures(data, user, featureFunction):
directory = data + '/' + user
features = []
for site in os.listdir(directory):
path = directory + '/' + site
stream = util.filterKeystrokes(util.openStream(path))
features.append(featureFunction(stream))
return chain(*features)
def readNewick(self, filename, readData=None):
# use simple parsing if pyparsing is not available
if pyparsing == None:
return self.readBigNewick(filename)
try:
# default data reader
if readData == None:
readData = self.readData
# get parse tree
text = util.readUntil(util.openStream(filename), ";")[0] + ";"
expr = newickParser(text)[0]
# walk the parse tree and build the tree
self.clear()
def walk(expr):
if isinstance(expr, pyparsing.ParseResults):
# parse name
if "name" in expr:
node = TreeNode(expr["name"])
else:
node = TreeNode(self.newName())
if "data" in expr:
readData(node, expr["data"])
# recurse
for child in expr:
ret = walk(child)
if ret:
self.addChild(node, ret)
return node
self.root = walk(expr)
# test for boot strap presence
for node in self.nodes.itervalues():
if "boot" in node.data:
self.defaultData["boot"] = 0
break
self.setDefaultData()
except RuntimeError:
# try simpler parser
return self.readBigNewick(filename)
def readNewick(self, filename, readData=None):
# use simple parsing if pyparsing is not available
if pyparsing == None:
return self.readBigNewick(filename)
try:
# default data reader
if readData == None:
readData = self.readData
# get parse tree
text = util.readUntil(util.openStream(filename), ";")[0] + ";"
expr = newickParser(text)[0]
# walk the parse tree and build the tree
self.clear()
def walk(expr):
if isinstance(expr, pyparsing.ParseResults):
# parse name
if "name" in expr:
node = TreeNode(expr["name"])
else:
node = TreeNode(self.newName())
if "data" in expr:
readData(node, expr["data"])
# recurse
for child in expr:
ret = walk(child)
if ret:
self.addChild(node, ret)
return node
self.root = walk(expr)
# test for boot strap presence
for node in self.nodes.itervalues():
if "boot" in node.data:
self.defaultData["boot"] = 0
break
self.setDefaultData()
except RuntimeError:
# try simpler parser
return self.readBigNewick(filename)
def readTreeColorMap(filename):
"""Reads a tree colormap from a file"""
infile = util.openStream(filename)
maps = []
for line in infile:
expr, red, green, blue = line.rstrip().split("\t")
maps.append([expr, map(float, (red, green, blue))])
name2color = makeExprMapping(maps)
def leafmap(node):
return name2color(node.name)
return treeColorMap(leafmap)
def readTreeColorMap(filename):
"""Reads a tree colormap from a file"""
infile = util.openStream(filename)
maps = []
for line in infile:
expr, red, green, blue = line.rstrip().split("\t")
maps.append([expr, map(float, (red, green, blue))])
name2color = makeExprMapping(maps)
def leafmap(node):
return name2color(node.name)
return treeColorMap(leafmap)
def extractFeatureVectors(data, statisticFn):
userList = os.listdir(data)
count = 0
for user in userList: # collect all segmented data
userdir = data + '/' + user
for site in os.listdir(userdir):
path = userdir + '/' + site
stream = util.filterKeystrokes(util.openStream(path))
sessions = util.segmentStream(stream, 600)
for session in sessions:
# Generate each feature
anyEmptyFeatures = False
vector = [count]
for ind in FEATURE_INDICES:
features = viz.collectFeatures(session, viz.FEATURES[ind])
#if len(features) < 2:
# anyEmptyFeatures = True
# break
vector.extend(statisticFn(features))
if not anyEmptyFeatures: yield vector
count += 1
def extractFeatureVectors(data, statisticFn):
userList = os.listdir(data)
count = 0
for user in userList: # collect all segmented data
userdir = data + '/' + user
for site in os.listdir(userdir):
path = userdir + '/' + site
stream = util.filterKeystrokes(util.openStream(path))
sessions = util.segmentStream(stream, 600)
for session in sessions:
# Generate each feature
anyEmptyFeatures = False
vector = [count]
for ind in FEATURE_INDICES:
features = viz.collectFeatures(session, viz.FEATURES[ind])
#if len(features) < 2:
# anyEmptyFeatures = True
# break
vector.extend(statisticFn(features))
if not anyEmptyFeatures: yield vector
count += 1
def writeNewick(self, out=sys.stdout, writeData=None, oneline=False):
"""Write the tree in newick notation"""
self.writeNewickNode(self.root,
util.openStream(out, "w"),
writeData=writeData,
oneline=oneline)
def writeNewick(self, out = sys.stdout, writeData=None, oneline=False):
"""Write the tree in newick notation"""
self.writeNewickNode(self.root, util.openStream(out, "w"),
writeData=writeData, oneline=oneline)
def main(args):
createDirectoryIfNotExist(args.odir)
userList = os.listdir(args.dir)
for user in userList:
createDirectoryIfNotExist(args.odir + '/' + user)
sites = os.listdir(args.dir + '/' + user)
for site in sites:
relName = user + '/' + site
createDirectoryIfNotExist(args.odir + '/' + relName)
stream = util.filterKeystrokes(util.openStream(args.dir + '/' + relName))
sessions = util.segmentStream(stream)
# We want histograms of keystroke usage per user
allLengths = list(util.getAllKeystrokeLengths(stream))
if len(allLengths) > 0:
plt.clf()
plt.hist(list(allLengths), 200)
plt.savefig(args.odir + '/' + user + '/all-kl-' + site + '.png')
for i in range(len(keyCombos)):
lengths = list(util.getKeystrokeLengths(stream, keyCombos[i]))
if len(lengths) > 0:
plt.clf()
plt.hist(list(lengths), 200)
plt.savefig(args.odir + '/' + user + '/' + names[i] + '-kl-' + site + '.png')
# Also want histograms of word data, per user
wordData = zip(*chain(*[util.getWordData(s) for s in sessions]))
if len(wordData) > 0 and len(wordData[0]) > 0:
plt.clf()
plt.hist(list(d for d in wordData[0] if abs(d) < 10), 200)
plt.savefig(args.odir + '/' + user + '/word-dur-' + site + '.png')
plt.clf()
plt.hist(list(wordData[1]), 200)
plt.savefig(args.odir + '/' + user + '/word-len-' + site + '.png')
# Key overlaps
#keyOverlaps = list(chain.from_iterable(util.getKeyOverlaps(s) for s in sessions))
keyOverlaps = list(x for x in util.getKeyOverlaps(stream) if abs(x) < 5)
if len(keyOverlaps) > 0:
plt.clf()
plt.hist(keyOverlaps, 100)
plt.savefig(args.odir + '/' + user + '/overlap-' + site + '.png')
# Word pauses
wordPauses = list(x for x in util.getWordPauses(stream) if abs(x) < 60)
if len(wordPauses) > 0:
plt.clf()
plt.hist(wordPauses, 100)
plt.savefig(args.odir + '/' + user + '/word-pause-' + site + '.png')
# Time between shift key and modified key
shiftTime = list(x for x in util.getModifierDelays(stream, util._SHIFT) if abs(x) < 5)
if len(shiftTime) > 0:
plt.clf()
plt.hist(shiftTime, 100)
plt.savefig(args.odir + '/' + user + '/shift-delay-' + site + '.png')
# Time between shift-to-shift
shiftShift= list(x for x in util.getModifierDelays(stream, util._SHIFT) if abs(x) < 1200)
if len(shiftShift) > 0:
plt.clf()
plt.hist(shiftShift, 100)
plt.savefig(args.odir + '/' + user + '/shift-shift-' + site + '.png')
