ENH make generated SVMlight files self-describing in a comment

Comment is *not* used to parse the files, since that would break
compatibility with other tools; it's just there for the user.

sklearn/datasets/svmlight_format.py

@@ -25,6 +25,7 @@ import numpy as np
 import scipy.sparse as sp
 
 from ._svmlight_format import _load_svmlight_file
+from .. import __version__
 from ..utils import atleast2d_or_csr
 
 
@@ -200,6 +201,10 @@ def _dump_svmlight(X, y, f, one_based):
     else:
         line_pattern = u"%f %s\n"
 
+    f.write("# Generated by dump_svmlight_file from scikit-learn %s\n"
+            % __version__)
+    f.write("# Column indices are %s-based\n" % ["zero", "one"][one_based])
+
     for i in xrange(X.shape[0]):
         s = u" ".join([value_pattern % (j + one_based, X[i, j])
                        for j in X[i].nonzero()[is_sp]])

sklearn/datasets/tests/test_svmlight_format.py

@@ -11,8 +11,10 @@ from numpy.testing import assert_array_equal
 from numpy.testing import assert_array_almost_equal
 from nose.tools import assert_raises, raises
 
+import sklearn
 from sklearn.datasets import (load_svmlight_file, load_svmlight_files,
                               dump_svmlight_file)
+from sklearn.utils.testing import assert_in
 
 currdir = os.path.dirname(os.path.abspath(__file__))
 datafile = os.path.join(currdir, "data", "svmlight_classification.txt")
@@ -175,6 +177,12 @@ def test_dump():
                 dump_svmlight_file(X.astype(dtype), y, f,
                                    zero_based=zero_based)
                 f.seek(0)
+
+                comment = f.readline()
+                assert_in("scikit-learn %s" % sklearn.__version__, comment)
+                comment = f.readline()
+                assert_in(["one", "zero"][zero_based] + "-based", comment)
+
                 X2, y2 = load_svmlight_file(f, dtype=dtype,
                                             zero_based=zero_based)
                 assert_equal(X2.dtype, dtype)