Mercurial > hg > fetch
annotate fetch.py @ 24:b1f65f3bd1bc
pretend to flesh out git fetcher
| author | Jeff Hammel <jhammel@mozilla.com> |
|---|---|
| date | Thu, 10 Nov 2011 08:16:48 -0800 |
| parents | 12ad9ab11860 |
| children | e628ce3ae49f |
| rev | line source |
|---|---|
| 0 | 1 #!/usr/bin/env python |
| 2 | |
| 3 """ | |
| 4 fetch stuff from the interwebs | |
| 5 """ | |
| 6 | |
| 7 import os | |
| 23 | 8 import shutil |
| 0 | 9 import sys |
| 10 import optparse | |
| 11 | |
|
8
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
12 __all__ = ['Fetcher', 'Fetch', 'main'] |
|
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
13 |
|
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
14 def which(executable, path=os.environ['PATH']): |
| 15 | 15 """python equivalent of which; should really be in the stdlib""" |
| 16 dirs = path.split(os.pathsep) | |
| 17 for dir in dirs: | |
| 18 if os.path.isfile(os.path.join(dir, executable)): | |
| 19 return os.path.join(dir, executable) | |
| 7 | 20 |
| 0 | 21 class Fetcher(object): |
| 15 | 22 """abstract base class for resource fetchers""" |
| 0 | 23 |
| 15 | 24 @classmethod |
| 25 def match(cls, _type): | |
| 26 return _type == cls.type | |
| 0 | 27 |
| 23 | 28 def __init__(self, url, clobber=True): |
| 17 | 29 self.subpath = None |
| 30 if '#' in url: | |
| 31 url, self.subpath = url.rsplit('#') | |
| 15 | 32 self.url = url |
| 23 | 33 self.clobber = clobber |
| 0 | 34 |
| 15 | 35 def __call__(self, dest): |
| 17 | 36 raise NotImplementedError("Should be called by implementing class") |
| 37 | |
| 38 @classmethod | |
| 39 def doc(cls): | |
| 40 """return docstring for the instance""" | |
| 41 retval = getattr(cls, '__doc__', '').strip() | |
| 42 return ' '.join(retval.split()) | |
| 0 | 43 |
| 7 | 44 ### standard dispatchers - always available |
| 0 | 45 |
| 7 | 46 import tarfile |
| 0 | 47 import urllib2 |
| 7 | 48 from StringIO import StringIO |
| 0 | 49 |
| 5 | 50 class FileFetcher(Fetcher): |
| 15 | 51 """fetch a single file""" |
| 0 | 52 |
| 15 | 53 type = 'file' |
| 0 | 54 |
| 15 | 55 @classmethod |
| 56 def download(cls, url): | |
| 57 return urllib2.urlopen(url).read() | |
| 0 | 58 |
| 15 | 59 def __call__(self, dest): |
| 60 if os.path.isdir(dest): | |
| 61 filename = self.url.rsplit('/', 1)[-1] | |
| 62 dest = os.path.join(dest, filename) | |
| 63 f = file(dest, 'w') | |
| 64 f.write(self.download(self.url)) | |
| 65 f.close() | |
| 0 | 66 |
|
6
86f6f99e421b
add types for unimplemented dispatchers
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
67 |
| 5 | 68 class TarballFetcher(FileFetcher): |
| 15 | 69 """fetch and extract a tarball""" |
| 0 | 70 |
| 15 | 71 type = 'tar' |
| 0 | 72 |
| 15 | 73 def __call__(self, dest): |
|
24
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
74 if os.path.exists(dest): |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
75 assert os.path.isdir(dest) |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
76 else: |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
77 os.mkdirs(dest) |
| 17 | 78 if self.subpath: |
| 79 raise NotImplementedError("should extract only a subpath of a tarball but I haven't finished it yet") | |
| 15 | 80 buffer = StringIO() |
| 81 buffer.write(self.download(self.url)) | |
| 82 buffer.seek(0) | |
| 83 tf = tarfile.open(mode='r', fileobj=buffer) | |
| 84 tf.extract(dest) | |
| 7 | 85 |
|
8
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
86 fetchers = [FileFetcher, TarballFetcher] |
|
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
87 |
|
24
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
88 ### VCS fetchers |
|
8
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
89 |
|
11
726c3d288733
* add convenience import in __init__
Jeff Hammel <jhammel@mozilla.com>
parents:
10
diff
changeset
|
90 import subprocess |
| 19 | 91 try: |
| 92 from subprocess import check_call as call | |
| 93 except ImportErorr: | |
| 94 raise # we need check_call, kinda | |
|
11
726c3d288733
* add convenience import in __init__
Jeff Hammel <jhammel@mozilla.com>
parents:
10
diff
changeset
|
95 |
| 17 | 96 class VCSFetcher(Fetcher): |
|
24
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
97 |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
98 command = None |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
99 |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
100 def call(*args, **kwargs): |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
101 assert command is not None, "Abstract base class" |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
102 call([self.command] + list(args), **kwargs) |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
103 |
| 17 | 104 def __init__(self, url, export=True): |
| 105 """ | |
| 106 - export : whether to strip the versioning information | |
| 107 """ | |
| 108 Fetcher.__init__(self, url) | |
| 109 self.export = export | |
| 110 | |
|
24
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
111 def __call__(self, dest): |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
112 raise NotImplementedError("Abstract base class") |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
113 |
|
8
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
114 if which('hg'): |
|
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
115 |
| 17 | 116 class HgFetcher(VCSFetcher): |
| 15 | 117 """checkout a mercurial repository""" |
| 118 type = 'hg' | |
| 0 | 119 |
| 19 | 120 def __init__(self, url, export=True): |
| 121 VCSFetcher.__init__(self, url, export=True) | |
| 122 self.hg = which('hg') | |
| 123 | |
| 15 | 124 def __call__(self, dest): |
| 23 | 125 if os.path.exists(dest): |
| 15 | 126 assert os.path.isdir(dest) and os.path.exists(os.path.join(dest, '.hg')) |
| 19 | 127 call([self.hg, 'pull', self.url], cwd=dest) |
| 128 call([self.hg, 'update'], cwd=dest) | |
| 129 else: | |
| 23 | 130 if not os.path.exists(dest): |
| 131 os.mkdirs(dest) | |
| 132 call([self.hg, 'clone', self.url, dest]) | |
|
11
726c3d288733
* add convenience import in __init__
Jeff Hammel <jhammel@mozilla.com>
parents:
10
diff
changeset
|
133 |
| 15 | 134 fetchers.append(HgFetcher) |
|
6
86f6f99e421b
add types for unimplemented dispatchers
Jeff Hammel <jhammel@mozilla.com>
parents:
5
diff
changeset
|
135 |
|
24
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
136 |
| 15 | 137 if which('git'): |
| 17 | 138 |
| 15 | 139 class GitFetcher(Fetcher): |
| 140 """checkout a git repository""" | |
| 141 type = 'git' | |
|
8
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
142 |
| 20 | 143 def __init__(self, url, export=True): |
| 144 VCSFetcher.__init__(self, url, export=True) | |
| 23 | 145 self.git = which('git') |
| 146 | |
|
24
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
147 def __call__(self, dest): |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
148 if os.path.exists(dest): |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
149 assert os.path.isdir(dest) and os.path.exists(os.path.join(dest, '.git')) |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
150 call([self.git, 'pull', self.url], cwd=dest) |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
151 call([self.hg, 'update'], cwd=dest) |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
152 else: |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
153 if not os.path.exists(dest): |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
154 os.mkdirs(dest) |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
155 call([self.hg, 'clone', self.url, dest]) |
|
b1f65f3bd1bc
pretend to flesh out git fetcher
Jeff Hammel <jhammel@mozilla.com>
parents:
23
diff
changeset
|
156 |
| 20 | 157 |
| 158 fetchers.append(GitFetcher) | |
| 17 | 159 |
| 16 | 160 __all__ += [i.__name__ for i in fetchers] |
|
8
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
161 |
| 0 | 162 class Fetch(object): |
| 163 | |
| 21 | 164 def __init__(self, fetchers=fetchers[:], relative_to=None, strict=True): |
| 15 | 165 self.fetchers = fetchers |
| 166 self.relative_to = relative_to | |
| 167 self.strict = strict | |
| 0 | 168 |
| 15 | 169 def fetcher(self, _type): |
| 170 """find the fetcher for the appropriate type""" | |
| 171 for fetcher in fetchers: | |
| 172 if fetcher.match(_type): | |
| 173 return fetcher | |
| 0 | 174 |
| 15 | 175 def __call__(self, url, destination, type, **options): |
| 176 fetcher = self.fetcher(type) | |
| 177 assert fetcher is not None, "No fetcher found for type '%s'" % type | |
| 178 fetcher = fetcher(url, **options) | |
| 179 fetcher(destination) | |
| 2 | 180 |
| 15 | 181 def fetch(self, *items): |
| 2 | 182 |
| 15 | 183 if self.strict: |
| 184 # ensure all the required fetchers are available | |
| 185 types = set([i['type'] for i in items]) | |
| 186 assert not [i for i in types | |
| 187 if [True for fetcher in fetchers if fetcher.match(i)]] | |
| 4 | 188 |
| 15 | 189 for item in items: |
| 4 | 190 |
| 15 | 191 # fix up relative paths |
| 192 dest = item['dest'] | |
| 193 if not os.path.isabs(dest): | |
| 194 relative_to = self.relative_to or os.path.dirname(os.path.abspath(item['manifest'])) | |
| 195 dest = os.path.join(relative_to, dest) | |
| 4 | 196 |
| 15 | 197 # fetch the items |
| 198 self(item['url'], destination=dest, type=item['type'], **item['options']) | |
| 0 | 199 |
| 21 | 200 |
| 0 | 201 format_string = "[URL] [destination] [type] <options>" |
| 202 def read_manifests(*manifests): | |
| 15 | 203 """ |
| 204 read some manifests and return the items | |
| 205 | |
| 206 Format: | |
| 207 %s | |
| 208 """ % format_string | |
| 0 | 209 |
| 15 | 210 # sanity check |
| 211 assert not [i for i in manifests if not os.path.exists(i)] | |
| 0 | 212 |
| 15 | 213 retval = [] |
| 0 | 214 |
| 15 | 215 for manifest in manifests: |
| 216 for line in file(i).readlines(): | |
| 217 line = line.strip() | |
| 218 if line.startswith('#') or not line: | |
| 219 continue | |
| 220 line = line.split() | |
| 221 if len(line) not in (3,4): | |
| 222 raise Exception("Format should be: %s; line %s" % (format_string, line)) | |
| 223 options = {} | |
| 224 if len(line) == 4: | |
| 225 option_string = line.pop().rstrip(',') | |
| 226 try: | |
| 227 options = dict([[j.strip() for j in i.split('=', 1)] | |
| 228 for i in option_string.split(',')]) | |
| 229 except: | |
| 230 raise Exception("Options format should be: key=value,key2=value2,...; got %s" % option_string) | |
| 0 | 231 |
| 15 | 232 url, dest, _type = line |
| 233 retval.append(dict(url=url, dest=dest, type=_type, options=options, manifest=manifest)) | |
| 234 return retval | |
| 0 | 235 |
| 2 | 236 def main(args=sys.argv[1:]): |
| 0 | 237 |
| 15 | 238 # parse command line options |
| 239 usage = '%prog [options] manifest [manifest] [...]' | |
| 0 | 240 |
| 15 | 241 class PlainDescriptionFormatter(optparse.IndentedHelpFormatter): |
| 242 def format_description(self, description): | |
| 243 if description: | |
| 244 return description + '\n' | |
| 245 else: | |
| 246 return '' | |
| 0 | 247 |
| 15 | 248 parser = optparse.OptionParser(usage=usage, description=__doc__, formatter=PlainDescriptionFormatter()) |
| 249 parser.add_option('-o', '--output', | |
| 250 help="output relative to this location vs. the manifest location") | |
| 17 | 251 parser.add_option('-d', '--dest', # XXX unused |
| 15 | 252 action='append', |
| 253 help="output only these destinations") | |
| 254 parser.add_option('-s', '--strict', | |
| 255 action='store_true', default=False, | |
| 256 help="fail on error") | |
| 257 parser.add_option('--list-fetchers', dest='list_fetchers', | |
| 258 action='store_true', default=False, | |
| 259 help='list available fetchers and exit') | |
| 260 options, args = parser.parse_args(args) | |
| 0 | 261 |
| 15 | 262 if options.list_fetchers: |
| 17 | 263 types = set() |
| 264 for fetcher in fetchers: | |
| 265 if fetcher.type in types: | |
| 266 continue # occluded, should probably display separately | |
| 267 print '%s : %s' % (fetcher.type, fetcher.doc()) | |
| 268 types.add(fetcher.type) | |
| 15 | 269 parser.exit() |
|
8
cf00d46b1bfb
pretend like we have a pluggable system to start debugging it
Jeff Hammel <jhammel@mozilla.com>
parents:
7
diff
changeset
|
270 |
| 15 | 271 if not args: |
| 17 | 272 # TODO: could read from stdin |
| 15 | 273 parser.print_help() |
| 274 parser.exit() | |
| 0 | 275 |
| 15 | 276 items = read_manifests(*args) |
| 16 | 277 fetch = Fetch(fetchers, strict=options.strict) |
| 0 | 278 |
| 15 | 279 # download the files |
| 280 fetch.fetch(*items) | |
| 0 | 281 |
| 282 if __name__ == '__main__': | |
| 15 | 283 main() |
| 0 | 284 |