Mercurial > repos > bcclaywell > argo_navis
annotate bin/parse_pact_tree.py @ 2:7eaf6f9abd28 draft default tip
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b-dirty
author | bcclaywell |
---|---|
date | Mon, 12 Oct 2015 17:57:38 -0400 |
parents | d67268158946 |
children |
rev | line source |
---|---|
0
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
1 #!/usr/bin/env python |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
2 import argparse |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
3 import csv |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
4 import re |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
5 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
6 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
7 coord_re = re.compile("\{([-\d\.]+),([-\d\.]+)\}") |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
8 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
9 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
10 def parse_rules(handle): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
11 # Returns a very raw and literal translation of the out.rules results from PACT |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
12 def get_nodes(line): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
13 return [int(x) for x in line.split()] |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
14 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
15 def int_if_intable(a): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
16 try: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
17 return int(a) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
18 except: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
19 return a |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
20 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
21 def get_map(line, imgfn=int_if_intable): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
22 coll = (x.split('->') for x in line.split()) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
23 return dict([(int_if_intable(a), imgfn(b)) for a, b in coll]) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
24 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
25 def parse_coordinate(text): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
26 m = coord_re.match(text) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
27 return (float(m.group(1)), float(m.group(2))) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
28 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
29 tips = get_nodes(handle.next()) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
30 trunk_nodes = get_nodes(handle.next()) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
31 connections = get_map(handle.next()) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
32 labels = get_map(handle.next()) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
33 coordinates = get_map(handle.next(), parse_coordinate) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
34 tip_names = get_map(handle.next(), str) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
35 return dict(tips=tips, trunk_nodes=trunk_nodes, connections=connections, labels=labels, |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
36 coordinates=coordinates, tip_names=tip_names) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
37 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
38 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
39 def get_row(parsed_tree, n_id): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
40 # This gives us the row data (as seen in final table) for the given n_id value |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
41 if n_id in parsed_tree['tips']: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
42 klass = "tip" |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
43 name = parsed_tree['tip_names'][n_id] |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
44 else: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
45 klass = "trunk" |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
46 name = "" |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
47 try: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
48 parent_id = parsed_tree['connections'][n_id] |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
49 except KeyError: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
50 parent_id = n_id |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
51 klass = "root" |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
52 label = parsed_tree['labels'][n_id] |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
53 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
54 x, y = parsed_tree['coordinates'][n_id] |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
55 parent_x, parent_y = parsed_tree['coordinates'][parent_id] |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
56 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
57 return dict(id=n_id, klass=klass, name=name, parent_id=parent_id, x=x, y=y, parent_x=parent_x, |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
58 parent_y=parent_y, label=label) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
59 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
60 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
61 def parsed_to_table(parsed_tree): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
62 # Cols are going to be: |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
63 # id, parent_id, label, klass, name, x, y, parent_x, parent_y, |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
64 for n_id in parsed_tree['coordinates'].keys(): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
65 yield get_row(parsed_tree, n_id) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
66 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
67 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
68 def get_args(): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
69 parser = argparse.ArgumentParser(prog="parse_pact_tree.py", |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
70 description="""Utility for parsing the output of PACT into a form renderable by ggplot""") |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
71 parser.add_argument('input', type=argparse.FileType('r')) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
72 parser.add_argument('output', type=argparse.FileType('w')) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
73 return parser.parse_args() |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
74 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
75 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
76 def main(): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
77 # Get args, run the parser, spit out the results into a file |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
78 args = get_args() |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
79 data = parse_rules(args.input) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
80 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
81 writer = csv.DictWriter(args.output, |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
82 fieldnames=["id", "parent_id", "label", "klass", "name", "x", "y", "parent_x", "parent_y"]) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
83 writer.writeheader() |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
84 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
85 for row in parsed_to_table(data): |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
86 writer.writerow(row) |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
87 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
88 args.input.close() |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
89 args.output.close() |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
90 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
91 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
92 if __name__ == '__main__': |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
93 main() |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
94 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
95 |
d67268158946
planemo upload commit a3f181f5f126803c654b3a66dd4e83a48f7e203b
bcclaywell
parents:
diff
changeset
|
96 |