Mercurial > hg > cc > cirrus_home
changeset 97:2b880f2ce894
basic help format hacking works
author | Henry S. Thompson <ht@inf.ed.ac.uk> |
---|---|
date | Fri, 16 Apr 2021 12:55:05 +0000 |
parents | a7e72a254790 |
children | 1a4c5fdc2923 |
files | bin/ix.py |
diffstat | 1 files changed, 28 insertions(+), 15 deletions(-) [+] |
line wrap: on
line diff
--- a/bin/ix.py Fri Apr 16 09:01:16 2021 +0000 +++ b/bin/ix.py Fri Apr 16 12:55:05 2021 +0000 @@ -7,24 +7,31 @@ Note that if no output flag(s) is/are given, the whole WARC record will be output, more efficiently than would be the case if all three flags were given.''' -import argparse +import sys, argparse, regex + +HACK=regex.compile('\[-x\]\n\s*\[length\] \[offset\] \[filename\]') +class HackFormat(argparse.RawDescriptionHelpFormatter): + def format_help(self): + global FOO + FOO=argparse.RawDescriptionHelpFormatter.format_help(self) + return HACK.sub('\n [ ( -x | length offset filename ) ]', + FOO) + parser = argparse.ArgumentParser( description='''Extract records from warc files given length, offset and file triples. Input one triple on command line, or triples from stdin as tab-delimited lines or complete cdx index lines.''', - epilog='''[ -x | length option file] - -Note that if no output flag(s) is/are given, + epilog='''Note that if no output flag(s) is/are given, the whole WARC record will be output, more efficiently than would be the case if all three flags were given.''', add_help=False, conflict_handler='resolve', - formatter_class=argparse.RawDescriptionHelpFormatter + formatter_class=HackFormat ) -parser.add_argument('--help',help='Show help',action='store_true') +parser.add_argument('--help',help='Show help',action='help') parser.add_argument('-d','--debug',help='Debug output',action='store_true') parser.add_argument('-w','--warc',help='output WARC headers', action='store_true') @@ -32,17 +39,23 @@ action='store_true') parser.add_argument('-b','--body',help='output HTTP body', action='store_true') -parser.add_argument('-e','--exec',help='pipes each result thru EXEC') +parser.add_argument('-c','--cmd',help='pipes each result thru CMD') sg=parser.add_mutually_exclusive_group() sg.add_argument('-x','--index', help='take lines of triples from a cdx index file as input', action='store_true') -tg=sg.add_argument_group('triple','explicit triple') -tg.add_argument('length',type=int, - help='length in bytes of gzipped record') -tg.add_argument('offset',type=int, - help='start position in bytes of gzipped record') -tg.add_argument('filename', - help='name of gzipped Common Crawl WARC-format file') +sg.add_argument('length',type=int, + help='length in bytes of gzipped record', + nargs='?') +parser.add_argument('offset',type=int, + help='start position in bytes of gzipped record', + nargs='?') +parser.add_argument('filename', + help='name of gzipped Common Crawl WARC-format file', + nargs='?') -parser.print_help() +#parser.print_help() +pa=parser.parse_args(sys.argv[1:]) +# We have to enforce our own check.. +print(pa) +