Cog409: Sarja 5

# -*- coding: utf-8 -*-
# Sakari Vaelma
"""Assignment set 5, assignments 1 and 2 by Sakari Vaelma, 013328180"""


# A helper class for assignment 2
class Frequency:
    """Calculate frequencies for any immutable object in a sequencei"""
    frequencies = {}

    def add(self, item):
        """Add a immutable to frequencies, return None."""
        self.frequencies[item] = self.frequencies.get(item, 0) + 1

    def sorted(self):
        """
        Return all (item, frequency) in a list, sorted by frequency.

        The resulting frequencies are in descending order.
        """
        return sorted(self.frequencies.items(),
                      key=lambda a: a[1], # Sorting by the 2nd item (frequency)
                      reverse=True)

    def __str__(self):
        """Prints a nicely formatted list of frequencies, descending order"""
        return "\n".join(["\t%s: %s" % items
                          for items
                          in self.sorted()])


# Assignment 1
def read_file(filename):
    """Read data from file and return it in a list of dictionaries."""
    data = []

    with open(filename) as datafile:
        for line in datafile:
            line = line.strip()

            # Some concepts may not have attributes
            if " " in line:
                meta, attributes = line.split(" ", 1)
            else:
                meta, attributes = line, ""

            meta, attributes = meta.split("_"), attributes.split()
            data.append({'concept': meta[0][1:], # Leading hash removed
                         'group': meta[1],
                         'participant': int(meta[2]),
                         'education': meta[3],
                         'attributes': attributes})

    return data

# Assignment 2
def attributes_per_concept(data):
    """Calculate frequencies of attributes per concept"""
    concepts = {}

    for line in data:
        concept, attributes = line['concept'], line['attributes']
        if not concepts.has_key(concept):
            concepts[concept] = Frequency()
        map(concepts[concept].add, attributes)

    # Printing a table of concepts, attributes and frequencies:
    for concept, frequencies in concepts.items():
        print "Concept: %s\n%s\n" % (concept, frequencies)


if __name__ == "__main__":
    # Testing
    attributes_per_concept(read_file("attribuutiot.txt"))