Adding Py3k support + Travis tester

Author: perimosocordiae

.travis.yml

@@ -2,6 +2,7 @@ language: python
 sudo: false
 python:
   - "2.7"
+  - "3.4"
 install:
   - if [[ "$TRAVIS_PYTHON_VERSION" == 2.* ]]; then
       wget http://repo.continuum.io/miniconda/Miniconda-3.4.2-Linux-x86_64.sh -O miniconda.sh;

metric_learn/__init__.py

@@ -1,7 +1,9 @@
-from itml import ITML
-from lmnn import LMNN
-from lsml import LSML
-from sdml import SDML
-from nca import NCA
-from lfda import LFDA
-from rca import RCA
+from __future__ import absolute_import
+
+from .itml import ITML
+from .lmnn import LMNN
+from .lsml import LSML
+from .sdml import SDML
+from .nca import NCA
+from .lfda import LFDA
+from .rca import RCA

metric_learn/itml.py

@@ -11,9 +11,11 @@
 or semi-definite programming.
 """
 
+from __future__ import print_function, absolute_import
 import numpy as np
+from six.moves import xrange
 from sklearn.metrics import pairwise_distances
-from base_metric import BaseMetricLearner
+from .base_metric import BaseMetricLearner
 
 
 class ITML(BaseMetricLearner):
@@ -108,9 +110,9 @@ def fit(self, X, constraints, bounds=None, A0=None, verbose=False):
         break
       lambdaold = _lambda.copy()
       if verbose:
-        print 'itml iter: %d, conv = %f' % (it, conv)
+        print('itml iter: %d, conv = %f' % (it, conv))
     if verbose:
-      print 'itml converged at iter: %d, conv = %f' % (it, conv)
+      print('itml converged at iter: %d, conv = %f' % (it, conv))
     return self
 
   def metric(self):

metric_learn/lfda.py

@@ -10,11 +10,12 @@
 The core optimization problem of LFDA is solved as a generalized
 eigenvalue problem.
 """
-from __future__ import division
+from __future__ import division, absolute_import
 import numpy as np
 import scipy
+from six.moves import xrange
 from sklearn.metrics import pairwise_distances
-from base_metric import BaseMetricLearner
+from .base_metric import BaseMetricLearner
 
 
 class LFDA(BaseMetricLearner):

metric_learn/lmnn.py

@@ -9,10 +9,12 @@
 """
 #TODO: periodic recalculation of impostors, PCA initialization
 
+from __future__ import print_function, absolute_import
 import numpy as np
 from collections import Counter
+from six.moves import xrange
 from sklearn.metrics import pairwise_distances
-from base_metric import BaseMetricLearner
+from .base_metric import BaseMetricLearner
 
 
 # commonality between LMNN implementations
@@ -134,7 +136,7 @@ def fit(self, X, labels, verbose=False):
       delta_obj = objective - objective_old
 
       if verbose:
-        print it, objective, delta_obj, total_active, learn_rate
+        print(it, objective, delta_obj, total_active, learn_rate)
 
       # update step size
       if delta_obj > 0:
@@ -152,11 +154,11 @@ def fit(self, X, labels, verbose=False):
       # check for convergence
       if it > min_iter and abs(delta_obj) < convergence_tol:
         if verbose:
-          print "LMNN converged with objective", objective
+          print("LMNN converged with objective", objective)
         break
     else:
       if verbose:
-        print "LMNN didn't converge in %(max_iter)d steps." % self.params
+        print("LMNN didn't converge in %(max_iter)d steps." % self.params)
 
     # store the last L
     self.L = L

metric_learn/lsml.py

@@ -7,10 +7,12 @@
 Paper: http://www.cs.ucla.edu/~weiwang/paper/ICDM12.pdf
 """
 
-from random import choice
+from __future__ import print_function, absolute_import
 import numpy as np
 import scipy.linalg
-from base_metric import BaseMetricLearner
+from random import choice
+from six.moves import xrange
+from .base_metric import BaseMetricLearner
 
 
 class LSML(BaseMetricLearner):
@@ -63,14 +65,14 @@ def fit(self, X, constraints, weights=None, prior=None, verbose=False):
     s_best = self._total_loss(self.M, prior_inv)
     step_sizes = np.logspace(-10, 0, 10)
     if verbose:
-      print 'initial loss', s_best
+      print('initial loss', s_best)
     for it in xrange(1, self.max_iter+1):
       grad = self._gradient(self.M, prior_inv)
       grad_norm = scipy.linalg.norm(grad)
       if grad_norm < self.tol:
         break
       if verbose:
-        print 'gradient norm', grad_norm
+        print('gradient norm', grad_norm)
       M_best = None
       for step_size in step_sizes:
         step_size /= grad_norm
@@ -83,12 +85,12 @@ def fit(self, X, constraints, weights=None, prior=None, verbose=False):
           s_best = cur_s
           M_best = new_metric
       if verbose:
-        print 'iter', it, 'cost', s_best, 'best step', l_best * grad_norm
+        print('iter', it, 'cost', s_best, 'best step', l_best * grad_norm)
       if M_best is None:
         break
       self.M = M_best
     else:
-      print "Didn't converge after %d iterations. Final loss: %f" % (it, s_best)
+      print("Didn't converge after", it, "iterations. Final loss:", s_best)
     return self
 
   def _comparison_loss(self, metric):

metric_learn/nca.py

@@ -3,8 +3,10 @@
 Ported to Python from https://github.com/vomjom/nca
 """
 
+from __future__ import absolute_import
 import numpy as np
-from base_metric import BaseMetricLearner
+from six.moves import xrange
+from .base_metric import BaseMetricLearner
 
 
 class NCA(BaseMetricLearner):

metric_learn/rca.py

@@ -9,9 +9,12 @@
 
 'Learning distance functions using equivalence relations', ICML 2003
 """
+
+from __future__ import absolute_import
 import numpy as np
 import random
-from base_metric import BaseMetricLearner
+from six.moves import xrange
+from .base_metric import BaseMetricLearner
 
 
 class RCA(BaseMetricLearner):

metric_learn/sdml.py

@@ -8,12 +8,13 @@
 Paper: http://lms.comp.nus.edu.sg/sites/default/files/publication-attachments/icml09-guojun.pdf
 """
 
-from random import choice
+from __future__ import absolute_import
 import numpy as np
+from random import choice
+from scipy.sparse.csgraph import laplacian
 from sklearn.covariance import graph_lasso
 from sklearn.utils.extmath import pinvh
-from scipy.sparse.csgraph import laplacian
-from base_metric import BaseMetricLearner
+from .base_metric import BaseMetricLearner
 
 
 class SDML(BaseMetricLearner):

setup.py

@@ -22,7 +22,8 @@
       install_requires=[
           'numpy',
           'scipy',
-          'scikit-learn'
+          'scikit-learn',
+          'six'
       ],
       extras_require=dict(
           docs=['sphinx', 'numpydoc'],