diff options
Diffstat (limited to 'plot_exps.py')
-rwxr-xr-x | plot_exps.py | 236 |
1 files changed, 129 insertions, 107 deletions
diff --git a/plot_exps.py b/plot_exps.py index 46784bc..39529bd 100755 --- a/plot_exps.py +++ b/plot_exps.py | |||
@@ -2,76 +2,14 @@ | |||
2 | from __future__ import print_function | 2 | from __future__ import print_function |
3 | 3 | ||
4 | import os | 4 | import os |
5 | import re | ||
6 | import plot | ||
7 | import shutil as sh | 5 | import shutil as sh |
8 | 6 | import sys | |
9 | from collections import defaultdict | ||
10 | from optparse import OptionParser | 7 | from optparse import OptionParser |
11 | from gnuplot import Plot, curve | 8 | from parse.dir_map import DirMap |
12 | from random import randrange | 9 | from parse.tuple_table import ReducedTupleTable |
13 | 10 | from parse.col_map import ColMap | |
14 | class StyleMaker(object): | 11 | from collections import namedtuple,defaultdict |
15 | LINE_WIDTH = 1.5 | 12 | import matplotlib.pyplot as plot |
16 | POINT_SIZE = 0.6 | ||
17 | BEST_COLORS = [ | ||
18 | '#ff0000', # red | ||
19 | '#000001', # black | ||
20 | '#0000ff', # blue | ||
21 | '#be00c4', # purple | ||
22 | '#ffd700', # yellow | ||
23 | ] | ||
24 | |||
25 | def __init__(csvs): | ||
26 | self.main_key, self.col_map = __find_columns(csvs) | ||
27 | self.cur_style = 1 | ||
28 | |||
29 | # Use this for least-common varying attribute | ||
30 | self.main_map = {} | ||
31 | # Everything else is a color | ||
32 | self.color_map = TupleTable(self.col_map) | ||
33 | |||
34 | def __find_columns(csvs): | ||
35 | vals = defaultdict(lambda:set) | ||
36 | |||
37 | for csv in csvs: | ||
38 | to_decode = os.path.splitext(csv_file)[0] | ||
39 | params = plot.decode(to_decode) | ||
40 | for k,v in params.iteritems: | ||
41 | vals[k].add(v) | ||
42 | |||
43 | try: | ||
44 | main_key = min([(k,v) for (k,v) in thing.iteritems() if len(v) > 1], | ||
45 | key=operator.itemgetter(1))[0] | ||
46 | except ValueError: | ||
47 | main_key = None | ||
48 | |||
49 | col_map = ColMap() | ||
50 | for k,v in vals.iterkeys(): | ||
51 | if k == self.main_key: continue | ||
52 | for i in v: | ||
53 | self.col_map.try_add(k, i) | ||
54 | return (main_key, col_map) | ||
55 | |||
56 | def __rand_color(): | ||
57 | return "#%s" % "".join([hex(randrange(0, 255))[2:] for i in range(3)]) | ||
58 | |||
59 | def get_style(csv): | ||
60 | to_decode = os.path.splitext(csv_file)[0] | ||
61 | params = plot.decode(to_decode) | ||
62 | |||
63 | if kv not in self.color_map: | ||
64 | color = best.pop() if BEST_COLORS else __rand_color() | ||
65 | self.color_map.add_exp(params, color) | ||
66 | |||
67 | if self.main_key in params: | ||
68 | val = params[self.main_key] | ||
69 | if val not in self.main_map: | ||
70 | self.main_map[val] = self.cur_style | ||
71 | self.cur_style += 1 | ||
72 | style = self.main_map[val] | ||
73 | else: | ||
74 | style = 1 | ||
75 | 13 | ||
76 | def parse_args(): | 14 | def parse_args(): |
77 | parser = OptionParser("usage: %prog [options] [csv_dir]...") | 15 | parser = OptionParser("usage: %prog [options] [csv_dir]...") |
@@ -83,54 +21,138 @@ def parse_args(): | |||
83 | 21 | ||
84 | return parser.parse_args() | 22 | return parser.parse_args() |
85 | 23 | ||
86 | def get_label(kv): | ||
87 | label = [] | ||
88 | for key, value in kv.iteritems(): | ||
89 | label += ["%s=%s" % (key.capitalize(), value)] | ||
90 | return ", ".join(label) | ||
91 | |||
92 | def add_line(plot, csv_file): | ||
93 | to_decode = os.path.splitext(csv_file)[0] | ||
94 | params = plot.decode(to_decode) | ||
95 | |||
96 | def get_stat(path, name): | ||
97 | full = os.path.abspath(path) | ||
98 | rstr = r"(?P<STAT>[^/]+)/((max|min|var|avg)/)*(%s/?)?$" % name | ||
99 | regex = re.compile(rstr, re.I | re.M) | ||
100 | match = regex.search(full) | ||
101 | return match.group("STAT") | ||
102 | |||
103 | def plot_exp(name, data_dir, out_dir): | ||
104 | p = Plot() | ||
105 | p.format = 'pdf' | ||
106 | p.output = "%s/%s.pdf" % (out_dir, name) | ||
107 | p.xlabel = name.replace("vary-", "") | ||
108 | p.ylabel = get_stat(data_dir, name) | ||
109 | p.font = 'Helvetica' | ||
110 | p.dashed_lines = True | ||
111 | p.enhanced_text = True | ||
112 | p.size = ('5.0cm', '5.0cm') | ||
113 | p.font_size = '6pt' | ||
114 | p.key = 'on bmargin center horizontal' | ||
115 | |||
116 | csvs = [f for f in os.listdir(data_dir) if re.match("*.csv", f)] | ||
117 | col_map = get_col_map(csvs) | ||
118 | 24 | ||
25 | ExpDetails = namedtuple('ExpDetails', ['variable', 'value', 'title', 'out']) | ||
26 | OUT_FORMAT = 'pdf' | ||
27 | |||
28 | def get_details(path): | ||
29 | out = "_".join(path) if path else "plot" | ||
30 | |||
31 | value = path.pop() if path else None | ||
32 | variable = path.pop() if path else None | ||
33 | |||
34 | title = value.capitalize() if value else "" | ||
35 | title += " by %s" % variable if variable else "" | ||
36 | title += " (%s)" % (", ".join(path)) if path else "" | ||
37 | |||
38 | return ExpDetails(variable, value, title, out) | ||
39 | |||
40 | |||
41 | |||
42 | class StyleMap(object): | ||
43 | COLORS = list('bgrcmyk') | ||
44 | LINES = ['-', ':', '--'] | ||
45 | MARKERS = list('.,ov^<>1234sp*hH+xDd|_') | ||
46 | ORDER = [MARKERS, COLORS, LINES] | ||
47 | DEFAULT = ["k", "-", "k"] | ||
48 | |||
49 | def __init__(self, col_list, col_values): | ||
50 | self.prop_map = dict(zip(col_list, StyleMap.ORDER)) | ||
51 | |||
52 | # Store 1 style per value | ||
53 | self.value_map = defaultdict(dict) | ||
54 | for column, styles in self.prop_map.iteritems(): | ||
55 | value_styles = self.value_map[column] | ||
56 | for value in sorted(col_values[column]): | ||
57 | value_styles[value] = styles.pop(0) | ||
58 | styles += [value_styles[value]] | ||
59 | |||
60 | def get_style(self, kv): | ||
61 | style = '' | ||
62 | for k,v in kv.iteritems(): | ||
63 | if k in self.value_map: | ||
64 | style += self.value_map[k][v] | ||
65 | return style | ||
66 | |||
67 | def get_key(self): | ||
68 | key = [] | ||
69 | for column, properties in self.prop_map.iteritems(): | ||
70 | idx = StyleMap.ORDER.index(properties) | ||
71 | prop_string = StyleMap.DEFAULT[idx] + "%s" | ||
72 | for value, prop in self.value_map[column].iteritems(): | ||
73 | style = plot.plot([],[], prop_string%prop)[0] | ||
74 | key += [(style, "%s:%s" % (column, value))] | ||
75 | return sorted(key, key=lambda x:x[1]) | ||
76 | |||
77 | def plot_by_variable(dir_map, col_map, out_dir, force): | ||
78 | num_plots = 0 | ||
79 | id = 0 | ||
80 | for _,_ in dir_map.leafs(1): | ||
81 | num_plots += 1 | ||
82 | sys.stderr.write("Plotting by variable...") | ||
83 | |||
84 | for plot_path, plot_node in dir_map.leafs(1): | ||
85 | id += 1 | ||
86 | details = get_details(plot_path) | ||
87 | out_fname = "%s/%s.%s" % (out_dir, details.out, OUT_FORMAT) | ||
88 | if os.path.exists(out_fname) and not force: | ||
89 | continue | ||
90 | |||
91 | # Kinda bad... | ||
92 | first_csv = plot_node.children.keys()[0] | ||
93 | first_config = ColMap.decode(first_csv[:first_csv.index('.csv')]) | ||
94 | columns = filter(lambda c: c in first_config, col_map.columns()) | ||
95 | |||
96 | style_map = StyleMap(columns, col_map.get_values()) | ||
97 | |||
98 | figure = plot.figure() | ||
99 | axes = figure.add_subplot(111) | ||
100 | |||
101 | for line_path, line_node in plot_node.children.iteritems(): | ||
102 | encoded = line_path[:line_path.index(".csv")] | ||
103 | config = ColMap.decode(encoded) | ||
104 | style = style_map.get_style(config) | ||
105 | |||
106 | values = sorted(line_node.values, key=lambda tup: tup[0]) | ||
107 | xvalues, yvalues = zip(*values) | ||
108 | |||
109 | plot.plot(xvalues, yvalues, style) | ||
110 | |||
111 | lines, labels = zip(*style_map.get_key()) | ||
112 | |||
113 | axes.legend(tuple(lines), tuple(labels), prop={'size':10}) | ||
114 | axes.set_ylabel(details.value) | ||
115 | axes.set_xlabel(details.variable) | ||
116 | axes.set_xlim(0, axes.get_xlim()[1] + 1) | ||
117 | axes.set_ylim(0, axes.get_ylim()[1] + 1) | ||
118 | |||
119 | axes.set_title(details.title) | ||
120 | |||
121 | plot.savefig(out_fname, format=OUT_FORMAT) | ||
122 | |||
123 | sys.stderr.write('\r {0:.2%}'.format(float(id)/num_plots)) | ||
124 | sys.stderr.write('\n') | ||
125 | |||
126 | def plot_exp(data_dir, out_dir, force): | ||
127 | print("Reading data...") | ||
128 | dir_map = DirMap.read(data_dir) | ||
129 | print("Sorting configs...") | ||
130 | tuple_table = ReducedTupleTable.from_dir_map(dir_map) | ||
131 | col_map = tuple_table.get_col_map() | ||
132 | |||
133 | if not os.path.exists(out_dir): | ||
134 | os.mkdir(out_dir) | ||
135 | |||
136 | print("Plotting data...") | ||
137 | plot_by_variable(dir_map, col_map, out_dir, force) | ||
138 | # plot_by_config(tuple_table, out_dir) | ||
119 | 139 | ||
120 | def main(): | 140 | def main(): |
121 | opts, args = parse_args() | 141 | opts, args = parse_args() |
122 | args = args or [os.getcwd()] | 142 | args = args or [os.getcwd()] |
123 | 143 | ||
124 | # if opts.force and os.path.exists(opts.out_dir): | 144 | if opts.force and os.path.exists(opts.out_dir): |
125 | # sh.rmtree(opts.out_dir) | 145 | sh.rmtree(opts.out_dir) |
126 | # if not os.path.exists(opts.out_dir): | 146 | if not os.path.exists(opts.out_dir): |
127 | # os.mkdir(opts.out_dir) | 147 | os.mkdir(opts.out_dir) |
128 | 148 | ||
129 | for exp in args: | 149 | for exp in args: |
130 | name = os.path.split(exp)[1] | 150 | name = os.path.split(exp)[1] |
131 | out_dir = "%s/%s" % (opts.out_dir, exp) | 151 | if exp != os.getcwd(): |
132 | 152 | out_dir = "%s/%s" % (opts.out_dir, name) | |
133 | plot_exp(name, exp, out_dir) | 153 | else: |
154 | out_dir = os.getcwd() | ||
155 | plot_exp(exp, out_dir, opts.force) | ||
134 | 156 | ||
135 | if __name__ == '__main__': | 157 | if __name__ == '__main__': |
136 | main() | 158 | main() |