usage: conll.py [-h] [--file FILE] [--directory DIRECTORY]
                [--extension EXTENSION] [--section SECTION]
                [--tagmode {nochange,tagfile,onetagperword,remove,random}]
                [--tagfile TAGFILE] [--ambigious]
                [--formmode {nochange,formfile,remove}] [--formfile FORMFILE]
                [--subsmode {nochange,best}] [--subsfile SUBSFILE]

CoNLL file transformer

optional arguments:
  -h, --help            show this help message and exit
  --file FILE           CoNLL files to read
  --directory DIRECTORY
                        CoNLL directory to read
  --extension EXTENSION
                        CoNLL file extension to read
  --section SECTION     WSJ sections to be filtered out
  --tagmode {nochange,tagfile,onetagperword,remove,random}
                        Tag manipulation operation on corpus
  --tagfile TAGFILE     Tag file to be used. Only valid when used with
                        --tagmode tagfile|random
  --ambigious           Threat tags as ambigious by not using a dictionary
  --formmode {nochange,formfile,remove}
                        Form manipulation operation on corpus
  --formfile FORMFILE   Form file to be used. Only valid when used with
                        --formmode formfile
  --subsmode {nochange,best}
                        Form manipulation operation on corpus
  --subsfile SUBSFILE   Substitution file to be used. Only valid when used
                        with --subsmode formfile

$ python2.7 filter.py -h
usage: filter.py [-h] [--wsj10 | --wsj20 | --wsj40 | --wsj]
                 inputwildcard output

Cleanup & Filter CoNLL corpus

positional arguments:
  inputwildcard  File wildcard showing CoNLL corpus file(s) including one or
                 more dependency graphs
  output         Target CoNLL corpus file used to store clean and filtered
                 CoNLL corpus

optional arguments:
  -h, --help     show this help message and exit
  --wsj10        Cleaned sentences of maximum length 10
  --wsj20        Cleaned sentences of maximum length 20
  --wsj40        Cleaned sentences of maximum length 40
  --wsj          Cleaned sentences of full corpus

$ python2.7 eval.py -h
usage: eval.py [-h] [--ignoreroot] [--minlength MINLENGTH] goldfile modelfile

Evaluate two parsings

positional arguments:
  goldfile              Source CoNLL corpus file including gold dependency
                        graphs
  modelfile             Model CoNLL corpus file including model dependency
                        grapgs

optional arguments:
  -h, --help            show this help message and exit
  --ignoreroot
  --minlength MINLENGTH
                        Minimum sentence length to be considered in evaluation

$ python2.7 parser.py -h
usage: parser.py [-h] [--rhead | --lhead] input output

General some popular baseline parsings for given corpus files

positional arguments:
  input       Source CoNLL corpus file including gold dependency graphs
  output      Model CoNLL corpus file including model dependency grapgs

optional arguments:
  -h, --help  show this help message and exit
  --rhead     Right head parsing
  --lhead     Left head parsing

$ python2.7 genconll.py -h
usage: genconll.py [-h] [--parallel parallel]
                   input output [sections [sections ...]]

Generate CoNLL format by reading penntreebank trees

positional arguments:
  input                Treebank directory containing sections of penntree
                       corpus
  output               Target CoNLL directory
  sections             Section filter for generation

optional arguments:
  -h, --help           show this help message and exit
  --parallel parallel  Number of parallel slaves to perform conversion