graphex

#!/usr/bin/env python

# Simple way to make a scatter plot of numbers in a text file

# example usage: graphex "F1 score for threshold (\S+) is (\S+)\(P (\S+), R (\S+)\)"
#                                                X        Y1       Y2       Y3
# will make a scatter plot of X vs Y1, X vs Y2, and X vs Y3
# If only one field is given, it will automatically number starting from 0.

# You can name fields now too:
# graphex "Z: (?P<Z>\d+) X: (?P<X>\d+) Y: (?P<Y>\d+)"
# will graph X vs Z and X vs Y.

# graphex "(?P<cats>\S+) (?P<dogs>\S+) (?P<rats>\S+)"
# Since there's no field named 'x' or 'X', it will use automatic numbering.

# graphex "\s+[0-9.]+\s+[0-9.]+\s+[0-9.]+\s+[0-9.]+\s+[0-9.]+\s+(?P<Y>0.\d+)\s+[0-9.]+\s+(?P<Z>0.\d+)\s+\((?P<X>[0-9.]+)\)"
# Extracts f-scores and regularization from cvlm

# Dependencies:
#   matplotlib: http://matplotlib.sourceforge.net/

# Author: David McClosky
# Homepage: http://zorglish.org
# Code: http://github.com/dmcc

import sys, re, pylab
regex = re.compile(sys.argv[1])
points = []
match_index = 0
names = None
for line in sys.stdin:
    match = regex.search(line)
    if match:
        groupdict = match.groupdict()
        if groupdict:
            names = sorted(groupdict.keys())
            items = sorted(groupdict.items())
            if not ('x' in groupdict or 'X' in groupdict):
                items.insert(0, ('X', match_index))
            all_values = [float(value) for key, value in items]
        else:
            all_values = map(float, match.groups())
            if len(all_values) == 1:
                all_values = [match_index] + all_values

        points.append(all_values)
        match_index += 1

colors = iter('rgb')

if not points:
    print("Error: didn't match any lines")
    raise SystemExit

transformed = list(zip(*points))
x = transformed[0]
for y in transformed[1:]:
    pylab.scatter(x, y, c=next(colors))
    for item in sorted(zip(x, y)):
        print(item)

if names:
    pylab.legend(tuple(names[1:]))

pylab.show()