Mercurial > repos > stevecassidy > nltktools
comparison g_chart_parser.py @ 2:a47980ef2b96 draft
planemo upload for repository https://github.com/Alveo/alveo-galaxy-tools commit b5b26e9118f2ad8af109d606746b39a5588f0511-dirty
author | stevecassidy |
---|---|
date | Wed, 01 Nov 2017 01:19:55 -0400 |
parents | fb617586f4b2 |
children |
comparison
equal
deleted
inserted
replaced
1:fb617586f4b2 | 2:a47980ef2b96 |
---|---|
1 import sys | 1 import sys |
2 import nltk | 2 import nltk |
3 import argparse | 3 import argparse |
4 from nltk.corpus import PlaintextCorpusReader | 4 |
5 | 5 |
6 def arguments(): | 6 def arguments(): |
7 parser = argparse.ArgumentParser(description="run NER on a text") | 7 parser = argparse.ArgumentParser(description="run NER on a text") |
8 parser.add_argument('--input', required=True, action="store", type=str, help="input text file") | 8 parser.add_argument('--input', required=True, action="store", type=str, help="input text file") |
9 parser.add_argument('--grammar', required=True, action="store", type=str, help="grammar file") | 9 parser.add_argument('--grammar', required=True, action="store", type=str, help="grammar file") |
10 parser.add_argument('--output', required=True, action="store", type=str, help="output file path") | 10 parser.add_argument('--output', required=True, action="store", type=str, help="output file path") |
11 args = parser.parse_args() | 11 return parser.parse_args() |
12 return args | |
13 | 12 |
14 | 13 |
15 def chart_parse(in_file, grammar_file, out_file): | 14 def chart_parse(in_file, grammar_file, out_file): |
16 with open(in_file, 'r') as fd: | 15 with open(in_file, 'r') as fd: |
17 text = fd.read() | 16 text = fd.read() |
30 for t in trees: | 29 for t in trees: |
31 output.write(t.pformat()) | 30 output.write(t.pformat()) |
32 output.write('\n') | 31 output.write('\n') |
33 | 32 |
34 except Exception as e: | 33 except Exception as e: |
35 message = "Error with parsing. Check the input files are correct and the grammar contains every word in the input sequence. \n----\n" + str(e) + "\n" | 34 message = """Error with parsing. Check the input files are correct |
35 and the grammar contains every word in the input sequence. \n----\n""" + str(e) + "\n" | |
36 sys.stderr.write(message) | 36 sys.stderr.write(message) |
37 sys.exit() | 37 sys.exit() |
38 output.close() | 38 output.close() |
39 | 39 |
40 | |
40 if __name__ == '__main__': | 41 if __name__ == '__main__': |
41 args = arguments() | 42 args = arguments() |
42 chart_parse(args.input, args.grammar, args.output) | 43 chart_parse(args.input, args.grammar, args.output) |