Class FeatureDictReader

object --+    
         |    
      dict --+
             |
            FeatureDictReader

Decodes data feature vectors (represented as index:value lines in a text file) into string:value feature dictionary objects.

These were probably encoded in the first place with the matching "FeatureDictWriter" class. The basic strategy is to just back-translate the index numbers into the actual string representation features. This class (which extends the dict class) stores the index:feature mappings for subsequent access (note that this is the inverse of the FeatureDictWriter's feature:index mappings).

Note that the default implementation assumes the features are represented as simple text strings. If you want a more sophisticated object representation, you'll have to extend this class and override the parseFeature method to translate the string into the feature object.

>>> from cStringIO import StringIO
>>> infile = StringIO();        # doctest can't handle multi-line strings well
>>> print >> infile, ""         # Test blank line robustness
>>> print >> infile, "# 0 a"    # So write it out as a StringIO first
>>> print >> infile, "# 1 s"
>>> print >> infile, "# 2 d"
>>> print >> infile, "# 3 g"
>>> print >> infile, "# 4 f"
>>> print >> infile, "# 5 A"
>>> print >> infile, "# 6 F"
>>> print >> infile, "# 7 S"
>>> print >> infile, "# 8 D"
>>> print >> infile, "# 9 h"
>>> print >> infile, "# 10 G"
>>> print >> infile, "# 11 H"
>>> print >> infile, "asdfsdfg UNKNOWN_ID 0:1 1:2 2:2 3:1 4:2 "
>>> print >> infile, "asdfasdfASDF UNKNOWN_ID 0:2 1:2 2:2 4:2 5:1 6:1 7:1 8:1 "
>>> print >> infile, "dfghDFGH UNKNOWN_ID 2:1 3:1 4:1 6:1 8:1 9:1 10:1 11:1 "
>>> infile = StringIO(infile.getvalue());
>>>
>>> featureReader = FeatureDictReader(infile);
>>> # Read out and print the contents of each feature dictionary
>>> for featureDict in featureReader:
...     print str(featureDict)
{'a': 1.0, 's': 2.0, 'd': 2.0, 'g': 1.0, 'f': 2.0}
{'a': 2.0, 'A': 1.0, 'd': 2.0, 'F': 1.0, 'f': 2.0, 'S': 1.0, 's': 2.0, 'D': 1.0}
{'D': 1.0, 'g': 1.0, 'F': 1.0, 'h': 1.0, 'f': 1.0, 'G': 1.0, 'H': 1.0, 'd': 1.0}
>>> for description in featureReader.objDescriptions:
...     print description
asdfsdfg
asdfasdfASDF
dfghDFGH

>>> from cStringIO import StringIO
>>> reader = FeatureDictReader(StringIO(""))
>>> for (featureDict, objDescr) in reader:
...     print objDescr, featureDict