Mercurial > hg > fetch
view fetch/main.py @ 0:3497a30190d2
initial commit of fetch, WIP
author | Jeff Hammel <jhammel@mozilla.com> |
---|---|
date | Sun, 18 Sep 2011 10:33:04 -0700 |
parents | |
children | 0c344ed749ac |
line wrap: on
line source
#!/usr/bin/env python """ fetch stuff from the interwebs """ import os import sys import optparse class Fetcher(object): """abstract base class for resource fetchers""" @classmethod def match(cls, _type): return _type == cls.type def __init__(self, url): self.url = url def __call__(self, dest): raise NotImplementedError import urllib2 class FileFetcher(object): """fetch a single file""" type = 'file' @classmethod def download(cls, url): return urllib2.urlopen(url).read() def __call__(self, dest): if os.path.isdir(dest): filename = url.rsplit('/', 1)[-1] dest = os.path.join(dest, filename) f = file(dest, 'w') f.write(self.download(self.url)) f.close() class TarballFetcher(object): """fetch and extract a tarball""" type = 'tar' class HgFetcher(object): """checkout a mercurial repository""" class GitFetcher(object): """checkout a git repository""" fetchers = [FileFetcher] class Fetch(object): def __init__(self, fetchers, relative_to=None, strict=True): self.fetchers = fetchers self.relative_to = relative_to self.strict = strict def fetcher(self, _type): """find the fetcher for the appropriate type""" for fetcher in fetchers: if fetcher.match(_type): return fetcher def __call__(self, url, destination, type, **options): fetcher = self.fetcher(type) assert fetcher is not None fetcher = fetcher(url, **options) fetcher(destination) def fetch(self, *items): for item in items: self(item['url']) format_string = "[URL] [destination] [type] <options>" def read_manifests(*manifests): """ read some manifests and return the items Format: %s """ % format_string # sanity check assert not [i for i in manifest if not.os.path.exists(i)] retval = [] for manifest in manifests: for line in file(i).readlines(): line = line.strip() if line.startswith('#'): pass line = line.split() if len(line) not in (3,4): raise Exception("Format should be: %s; line %s" % (format_string, line)) options = {} if len(line) == 4: option_string = line.pop().rstrip(',') try: options = dict([[j.strip() for j in i.split('=', 1)] for i in option_string.split(',')]) except: raise Exception("Options format should be: key=value,key2=value2,...; got %s" % option_string) url, dest, _type = line retval.append(dict(url=url, dest=dest, type=_type, options=options, manifest=manifest)) return retval def main(args=sys.argv[:]): # parse command line options usage = '%prog [options] manifest [manifest] [...]' # description formatter class PlainDescriptionFormatter(optparse.IndentedHelpFormatter): def format_description(self, description): if description: return description + '\n' else: return '' parser = optparse.OptionParser(usage=usage, description=__doc__, formatter=PlainDescriptionFormatter()) parser.add_option('-o', '--output', help="output relative to this location vs. the manifest location") parser.add_option('-d', '--dest', action='append', help="output only these destinations") parser.add_option('-s', '--strict', action='store_true', default=False, help="fail on error") options, args = parser.parse_args(args) if not args: parser.print_help() parser.exit() items = read_manifests(*args) fetch = Fetch(fetchers, strict=options.strict) # ensure the all the required fetchers are available # TODO: to Fetch class types = set([i['type'] for i in items]) assert not [i for i in types if [True for fetcher in fetchers if fetcher.match(i)]] # download the files fetch.fetch(*items) if __name__ == '__main__': main()