#!/bin/env python2.6
"""
This is our first BME 205 program illustrating some basic file I/O,
printing, conditionals, and data structures.

This program reads 6-column BED data and outputs a list of genes
ranked by score (from lowest to highest).

Input data can either be piped to the program:

    cat sample_data.bed | ./BME_205_Python_lecture_firstProg.py

or name of file containing input data can be given as an argument:

    ./BME_205_Python_lecture_firstProg.py sample_data.bed

Blame Andrew Uzilov (auzilov@ucsc.edu), Fall 2009.
"""

import sys

# Write to STDERR using usual print syntax.
print >> sys.stderr, 'We are running this Python version:', sys.version_info

# Write to STDERR using string formatting operator (%).
print >> sys.stderr, 'Arguments are: %s' % sys.argv

# Read data from file or STDIN?
if len (sys.argv) == 2:
    # Read from file; open file for reading.
    infile = open (sys.argv[1], 'r')
elif len (sys.argv) == 1:
    # Read from STDIN, which is always "open for reading".
    infile = sys.stdin
else:
    # Oops, incorrect command-line invocation.
    # Print usage and scare the user with an error message.
    print >> sys.stderr, \
        __doc__, \
        '\nERROR: You gave the program %d args; only 0 or 1 args allowed.' \
        % (len (sys.argv) - 1)
    sys.exit (1)

# Read in BED data.
genesScores = []  # list of 2-tuples, each tuple is: (gene, score)
for line in infile:
    line = line.rstrip ('\n\r')  # strip newline and MS-DOS ^M 

    if line[:5] == 'track':
        continue  # skip 'track' lines

    fields = line.split()
    assert len (fields) == 6, 'Not a 6-column BED file'

    gene  = fields[3]
    score = int (fields[4])
    genesScores.append ((gene, score))  # append 2-tuple

def compareGenes (a, b):
    """Comparator function for gene/score tuples.

    Each input argument is a 2-tuple: (gene, score).
    """
    if a[1] < b[1]:
        return -1
    elif a[1] > b[1]:
        return 1
    else:
        return 0
   
# Sort genes by their scores.
sortedGenes = sorted (genesScores, compareGenes)

# Output sorted gene list.
for gene, score in sortedGenes:
    print "gene %s score %d" % (gene, score)