Converted the codes to be Python 3.x-compatible

biosppy/biometrics.py

@@ -247,7 +247,7 @@ def list_subjects(self):
 
         """
 
-        subjects = self._subject2label.keys()
+        subjects = list(self._subject2label.keys())
 
         return subjects
 
@@ -356,7 +356,7 @@ def batch_train(self, data=None):
         if data is None:
             raise TypeError("Please specify the data to train.")
 
-        for sub, val in data.iteritems():
+        for sub, val in data.items():
             if val is None:
                 try:
                     self.dismiss(sub, deferred=True)
@@ -399,10 +399,10 @@ def update_thresholds(self, fraction=1.):
 
         # gather data to test
         data = {}
-        for subject, label in self._subject2label.iteritems():
+        for subject, label in self._subject2label.items():
             # select a random fraction of the training data
             aux = self.io_load(label)
-            indx = range(len(aux))
+            indx = list(range(len(aux)))
             use, _ = utils.random_fraction(indx, fraction, sort=True)
 
             data[subject] = aux[use]
@@ -411,7 +411,7 @@ def update_thresholds(self, fraction=1.):
         _, res = self.evaluate(data, ths)
 
         # choose thresholds at EER
-        for subject, label in self._subject2label.iteritems():
+        for subject, label in self._subject2label.items():
             EER_auth = res['subject'][subject]['authentication']['rates']['EER']
             self.set_auth_thr(label, EER_auth[self.EER_IDX, 0], ready=True)
 
@@ -639,7 +639,7 @@ def evaluate(self, data, thresholds=None, show=False):
             thresholds = self.get_thresholds()
 
         # get subjects
-        subjects = filter(lambda item: self.check_subject(item), data.keys())
+        subjects = [item for item in list(data.keys()) if self.check_subject(item)]
         if len(subjects) == 0:
             raise ValueError("No enrolled subjects in test set.")
 
@@ -722,15 +722,15 @@ def cross_validation(cls, data, labels, cv, thresholds=None, **kwargs):
                 lbl = labels[item]
                 train_idx[lbl].append(item)
 
-            train_data = {sub: data[idx] for sub, idx in train_idx.iteritems()}
+            train_data = {sub: data[idx] for sub, idx in train_idx.items()}
 
             # test data set
             test_idx = collections.defaultdict(list)
             for item in test:
                 lbl = labels[item]
                 test_idx[lbl].append(item)
 
-            test_data = {sub: data[idx] for sub, idx in test_idx.iteritems()}
+            test_data = {sub: data[idx] for sub, idx in test_idx.items()}
 
             # instantiate classifier
             clf = cls(**kwargs)
@@ -832,8 +832,8 @@ def _prepare(self, data, targets=None):
 
         # target class labels
         if targets is None:
-            targets = self._subject2label.values()
-        elif isinstance(targets, basestring):
+            targets = list(self._subject2label.values())
+        elif isinstance(targets, str):
             targets = [targets]
 
         return data
@@ -975,7 +975,7 @@ def _authenticate(self, data, label, threshold):
         # select based on subject label
         aux = []
         ns = len(dists)
-        for i in xrange(ns):
+        for i in range(ns):
             aux.append(dists[i, train_labels[i, :] == label])
 
         dists = np.array(aux)
@@ -984,7 +984,7 @@ def _authenticate(self, data, label, threshold):
         dists = dists[:, :self.k]
 
         decision = np.zeros(ns, dtype='bool')
-        for i in xrange(ns):
+        for i in range(ns):
             # compare distances to threshold
             count = np.sum(dists[i, :] <= threshold)
 
@@ -1014,8 +1014,8 @@ def _get_thresholds(self):
             return np.linspace(self.min_thr, 1., 100)
 
         maxD = []
-        for _ in xrange(3):
-            for label in self._subject2label.values():
+        for _ in range(3):
+            for label in list(self._subject2label.values()):
                 # randomly select samples
                 aux = self.io_load(label)
                 ind = np.random.randint(0, aux.shape[0], 3)
@@ -1064,7 +1064,7 @@ def _identify(self, data, threshold=None):
         ns = len(dists)
 
         labels = []
-        for i in xrange(ns):
+        for i in range(ns):
             lbl, _ = majority_rule(train_labels[i, :], random=True)
 
             # compare distances to threshold
@@ -1102,8 +1102,8 @@ def _prepare(self, data, targets=None):
 
         # target class labels
         if targets is None:
-            targets = self._subject2label.values()
-        elif isinstance(targets, basestring):
+            targets = list(self._subject2label.values())
+        elif isinstance(targets, str):
             targets = [targets]
 
         dists = []
@@ -1415,7 +1415,7 @@ def _authenticate(self, data, label, threshold):
         aux = aux[sel, :]
 
         decision = []
-        for i in xrange(ns):
+        for i in range(ns):
             # determine majority
             predMax, count = majority_rule(aux[:, i], random=True)
             rate = float(count) / norm
@@ -1483,7 +1483,7 @@ def _identify(self, data, threshold=None):
             norm = 1.0
 
         labels = []
-        for i in xrange(ns):
+        for i in range(ns):
             # determine majority
             predMax, count = majority_rule(aux[:, i], random=True)
             rate = float(count) / norm
@@ -1524,11 +1524,11 @@ def _prepare(self, data, targets=None):
 
         # target class labels
         if self._nbSubjects == 1:
-            pairs = self._models.keys()
+            pairs = list(self._models.keys())
         else:
             if targets is None:
-                pairs = self._models.keys()
-            elif isinstance(targets, basestring):
+                pairs = list(self._models.keys())
+            elif isinstance(targets, str):
                 labels = list(
                     set(self._subject2label.values()) - set([targets]))
                 pairs = [[targets, lbl] for lbl in labels]
@@ -1563,9 +1563,9 @@ def _train(self, enroll=None, dismiss=None):
             dismiss = []
 
         # process dismiss
-        pairs = self._models.keys()
+        pairs = list(self._models.keys())
         for t in dismiss:
-            pairs = filter(lambda p: t in p, pairs)
+            pairs = [p for p in pairs if t in p]
 
         for p in pairs:
             self._del_clf(p)
@@ -1590,11 +1590,11 @@ def _train(self, enroll=None, dismiss=None):
 
         # check singles
         if self._nbSubjects == 1:
-            label = self._subject2label.values()[0]
+            label = list(self._subject2label.values())[0]
             X = self.io_load(label)
             self._get_single_clf(X, label)
         elif self._nbSubjects > 1:
-            aux = filter(lambda p: '' in p, self._models.keys())
+            aux = [p for p in list(self._models.keys()) if '' in p]
             if len(aux) != 0:
                 for p in aux:
                     self._del_clf(p)
@@ -1850,7 +1850,7 @@ def get_subject_results(results=None,
     R = np.zeros(nth, dtype='float')
     CM = []
 
-    for i in xrange(nth):  # for each threshold
+    for i in range(nth):  # for each threshold
         # authentication
         for k, lbl in enumerate(subject_idx):  # for each subject
             subject_tst = subjects[k]
@@ -2163,7 +2163,7 @@ def combination(results=None, weights=None):
         weights = {}
 
     # compile results to find all classes
-    vec = results.values()
+    vec = list(results.values())
     if len(vec) == 0:
         raise CombinationError("No keys found.")
 
@@ -2182,13 +2182,13 @@ def combination(results=None, weights=None):
         # multi-class
         counts = np.zeros(nb, dtype='float')
 
-        for n in results.iterkeys():
+        for n in results.keys():
             # ensure array
             res = np.array(results[n])
             ns = float(len(res))
 
             # get count for each unique class
-            for i in xrange(nb):
+            for i in range(nb):
                 aux = float(np.sum(res == unq[i]))
                 w = weights.get(n, 1.)
                 counts[i] += ((aux / ns) * w)

biosppy/clustering.py

@@ -134,7 +134,7 @@ def hierarchical(data=None,
                        'ward', 'weighted']:
         raise ValueError("Unknown linkage criterion '%r'." % linkage)
 
-    if not isinstance(metric, basestring):
+    if not isinstance(metric, str):
         raise TypeError("Please specify the distance metric as a string.")
 
     N = len(data)
@@ -405,19 +405,19 @@ def create_coassoc(ensemble=None, N=None):
     nparts = len(ensemble)
     assoc = 0
     for part in ensemble:
-        nsamples = np.array([len(part[key]) for key in part.iterkeys()])
+        nsamples = np.array([len(part[key]) for key in part.keys()])
         dim = np.sum(nsamples * (nsamples - 1)) / 2
 
         I = np.zeros(dim)
         J = np.zeros(dim)
         X = np.ones(dim)
         ntriplets = 0
 
-        for v in part.itervalues():
+        for v in part.values():
             nb = len(v)
             if nb > 0:
-                for h in xrange(nb):
-                    for f in xrange(h + 1, nb):
+                for h in range(nb):
+                    for f in range(h + 1, nb):
                         I[ntriplets] = v[h]
                         J[ntriplets] = v[f]
                         ntriplets += 1
@@ -539,7 +539,7 @@ def mdist_templates(data=None,
         clusters = {0: np.arange(len(data), dtype='int')}
 
     # cluster labels
-    ks = clusters.keys()
+    ks = list(clusters.keys())
 
     # remove the outliers' cluster, if present
     if '-1' in ks:
@@ -631,7 +631,7 @@ def centroid_templates(data=None, clusters=None, ntemplates=1):
         raise TypeError("Please specify a data clustering.")
 
     # cluster labels
-    ks = clusters.keys()
+    ks = list(clusters.keys())
 
     # remove the outliers' cluster, if present
     if '-1' in ks:
@@ -839,7 +839,7 @@ def outliers_dmean(data=None,
             outliers.append(i)
 
     outliers = np.unique(outliers)
-    normal = np.setdiff1d(range(len(data)), outliers, assume_unique=True)
+    normal = np.setdiff1d(list(range(len(data))), outliers, assume_unique=True)
 
     # output
     clusters = {-1: outliers, 0: normal}
@@ -987,7 +987,7 @@ def _merge_clusters(clusters):
 
     """
 
-    keys = clusters.keys()
+    keys = list(clusters.keys())
 
     # outliers
     if -1 in keys:

biosppy/metrics.py

@@ -83,7 +83,7 @@ def pdist(X, metric='euclidean', p=2, w=None, V=None, VI=None):
 
     """
 
-    if isinstance(metric, basestring):
+    if isinstance(metric, str):
         if metric == 'pcosine':
             metric = pcosine
 
@@ -127,7 +127,7 @@ def cdist(XA, XB, metric='euclidean', p=2, V=None, VI=None, w=None):
 
     """
 
-    if isinstance(metric, basestring):
+    if isinstance(metric, str):
         if metric == 'pcosine':
             metric = pcosine
 

biosppy/plotting.py

@@ -770,7 +770,7 @@ def _plot_multichannel(ts=None,
 
     # check labels
     if labels is None:
-        labels = ['Ch. %d' % i for i in xrange(nch)]
+        labels = ['Ch. %d' % i for i in range(nch)]
 
     if nch < nrows:
         nrows = nch
@@ -794,7 +794,7 @@ def _plot_multichannel(ts=None,
     ax0.grid()
     axs = {(0, 0): ax0}
 
-    for i in xrange(1, nch - 1):
+    for i in range(1, nch - 1):
         a = i % nrows
         b = int(np.floor(i / float(nrows)))
         ax = fig.add_subplot(gs[a, b], sharex=ax0)
@@ -822,7 +822,7 @@ def _plot_multichannel(ts=None,
     if xlabel is not None:
         ax.set_xlabel(xlabel)
 
-        for b in xrange(0, ncols - 1):
+        for b in range(0, ncols - 1):
             a = nrows - 1
             ax = axs[(a, b)]
             ax.set_xlabel(xlabel)
@@ -1042,7 +1042,7 @@ def plot_biometrics(assessment=None, eer_idx=None, path=None, show=False):
     id_ax = fig.add_subplot(122)
 
     # subject results
-    for sub in assessment['subject'].iterkeys():
+    for sub in assessment['subject'].keys():
         auth_rates = assessment['subject'][sub]['authentication']['rates']
         _ = _plot_rates(ths, auth_rates, ['FAR', 'FRR'],
                         lw=MINOR_LW,
@@ -1134,7 +1134,7 @@ def plot_clustering(data=None, clusters=None, path=None, show=False):
     ymin, ymax = _yscaling(data, alpha=1.2)
 
     # determine number of clusters
-    keys = clusters.keys()
+    keys = list(clusters.keys())
     nc = len(keys)
 
     if nc <= 4:

biosppy/signals/bvp.py

@@ -181,7 +181,7 @@ def find_onsets(signal=None, sampling_rate=1000., sm_size=None, size=None,
 
         # analyze between maxima of 2nd derivative of ss
         detected = False
-        for i in xrange(1, len(dpidx) + 1):
+        for i in range(1, len(dpidx) + 1):
             try:
                 v, u = dpidx[i - 1], dpidx[i]
             except IndexError:

biosppy/signals/ecg.py

@@ -664,11 +664,11 @@ def engzee_segmenter(signal=None, sampling_rate=1000., threshold=0.48):
     mmp = 0.2
 
     # Differentiator (1)
-    y1 = map(lambda i: signal[i] - signal[i - 4], xrange(4, len(signal)))
+    y1 = [signal[i] - signal[i - 4] for i in range(4, len(signal))]
 
     # Low pass filter (2)
     c = [1, 4, 6, 4, 1, -1, -4, -6, -4, -1]
-    y2 = np.array(map(lambda n: np.dot(c, y1[n - 9:n + 1]), xrange(9, len(y1))))
+    y2 = np.array([np.dot(c, y1[n - 9:n + 1]) for n in range(9, len(y1))])
     y2_len = len(y2)
 
     # vars
@@ -1087,7 +1087,7 @@ def hamilton_segmenter(signal=None, sampling_rate=1000.):
             pass
 
         # getting positive peaks
-        for i in xrange(len(pospeaks) - 1):
+        for i in range(len(pospeaks) - 1):
             if abs(pospeaks[0][1] - pospeaks[i + 1][1]) > adjacency:
                 twopeaks.append(pospeaks[i + 1])
                 break
@@ -1097,7 +1097,7 @@ def hamilton_segmenter(signal=None, sampling_rate=1000.):
             error[0] = True
 
         # getting negative peaks
-        for i in xrange(len(negpeaks) - 1):
+        for i in range(len(negpeaks) - 1):
             if abs(negpeaks[0][1] - negpeaks[i + 1][1]) > adjacency:
                 twonegpeaks.append(negpeaks[i + 1])
                 break