Show More
@@ -1,868 +1,873 b'' | |||
|
1 | 1 | # RhodeCode VCSServer provides access to different vcs backends via network. |
|
2 | 2 | # Copyright (C) 2014-2019 RhodeCode GmbH |
|
3 | 3 | # |
|
4 | 4 | # This program is free software; you can redistribute it and/or modify |
|
5 | 5 | # it under the terms of the GNU General Public License as published by |
|
6 | 6 | # the Free Software Foundation; either version 3 of the License, or |
|
7 | 7 | # (at your option) any later version. |
|
8 | 8 | # |
|
9 | 9 | # This program is distributed in the hope that it will be useful, |
|
10 | 10 | # but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
11 | 11 | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
|
12 | 12 | # GNU General Public License for more details. |
|
13 | 13 | # |
|
14 | 14 | # You should have received a copy of the GNU General Public License |
|
15 | 15 | # along with this program; if not, write to the Free Software Foundation, |
|
16 | 16 | # Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
|
17 | 17 | import collections |
|
18 | 18 | import logging |
|
19 | 19 | import os |
|
20 | 20 | import posixpath as vcspath |
|
21 | 21 | import re |
|
22 | 22 | import stat |
|
23 | 23 | import traceback |
|
24 | 24 | import urllib |
|
25 | 25 | import urllib2 |
|
26 | 26 | from functools import wraps |
|
27 | 27 | |
|
28 | 28 | import more_itertools |
|
29 | 29 | import pygit2 |
|
30 | 30 | from pygit2 import Repository as LibGit2Repo |
|
31 | 31 | from dulwich import index, objects |
|
32 | 32 | from dulwich.client import HttpGitClient, LocalGitClient |
|
33 | 33 | from dulwich.errors import ( |
|
34 | 34 | NotGitRepository, ChecksumMismatch, WrongObjectException, |
|
35 | 35 | MissingCommitError, ObjectMissing, HangupException, |
|
36 | 36 | UnexpectedCommandError) |
|
37 | 37 | from dulwich.repo import Repo as DulwichRepo |
|
38 | 38 | from dulwich.server import update_server_info |
|
39 | 39 | |
|
40 | 40 | from vcsserver import exceptions, settings, subprocessio |
|
41 | 41 | from vcsserver.utils import safe_str |
|
42 | 42 | from vcsserver.base import RepoFactory, obfuscate_qs, raise_from_original |
|
43 | 43 | from vcsserver.hgcompat import ( |
|
44 | 44 | hg_url as url_parser, httpbasicauthhandler, httpdigestauthhandler) |
|
45 | 45 | from vcsserver.git_lfs.lib import LFSOidStore |
|
46 | 46 | |
|
47 | 47 | DIR_STAT = stat.S_IFDIR |
|
48 | 48 | FILE_MODE = stat.S_IFMT |
|
49 | 49 | GIT_LINK = objects.S_IFGITLINK |
|
50 | 50 | |
|
51 | 51 | log = logging.getLogger(__name__) |
|
52 | 52 | |
|
53 | 53 | |
|
54 | 54 | def reraise_safe_exceptions(func): |
|
55 | 55 | """Converts Dulwich exceptions to something neutral.""" |
|
56 | 56 | |
|
57 | 57 | @wraps(func) |
|
58 | 58 | def wrapper(*args, **kwargs): |
|
59 | 59 | try: |
|
60 | 60 | return func(*args, **kwargs) |
|
61 | 61 | except (ChecksumMismatch, WrongObjectException, MissingCommitError, ObjectMissing,) as e: |
|
62 | 62 | exc = exceptions.LookupException(org_exc=e) |
|
63 | 63 | raise exc(safe_str(e)) |
|
64 | 64 | except (HangupException, UnexpectedCommandError) as e: |
|
65 | 65 | exc = exceptions.VcsException(org_exc=e) |
|
66 | 66 | raise exc(safe_str(e)) |
|
67 | 67 | except Exception as e: |
|
68 | 68 | # NOTE(marcink): becuase of how dulwich handles some exceptions |
|
69 | 69 | # (KeyError on empty repos), we cannot track this and catch all |
|
70 | 70 | # exceptions, it's an exceptions from other handlers |
|
71 | 71 | #if not hasattr(e, '_vcs_kind'): |
|
72 | 72 | #log.exception("Unhandled exception in git remote call") |
|
73 | 73 | #raise_from_original(exceptions.UnhandledException) |
|
74 | 74 | raise |
|
75 | 75 | return wrapper |
|
76 | 76 | |
|
77 | 77 | |
|
78 | 78 | class Repo(DulwichRepo): |
|
79 | 79 | """ |
|
80 | 80 | A wrapper for dulwich Repo class. |
|
81 | 81 | |
|
82 | 82 | Since dulwich is sometimes keeping .idx file descriptors open, it leads to |
|
83 | 83 | "Too many open files" error. We need to close all opened file descriptors |
|
84 | 84 | once the repo object is destroyed. |
|
85 | 85 | """ |
|
86 | 86 | def __del__(self): |
|
87 | 87 | if hasattr(self, 'object_store'): |
|
88 | 88 | self.close() |
|
89 | 89 | |
|
90 | 90 | |
|
91 | 91 | class Repository(LibGit2Repo): |
|
92 | 92 | |
|
93 | 93 | def __enter__(self): |
|
94 | 94 | return self |
|
95 | 95 | |
|
96 | 96 | def __exit__(self, exc_type, exc_val, exc_tb): |
|
97 | 97 | self.free() |
|
98 | 98 | |
|
99 | 99 | |
|
100 | 100 | class GitFactory(RepoFactory): |
|
101 | 101 | repo_type = 'git' |
|
102 | 102 | |
|
103 | 103 | def _create_repo(self, wire, create, use_libgit2=False): |
|
104 | 104 | if use_libgit2: |
|
105 | 105 | return Repository(wire['path']) |
|
106 | 106 | else: |
|
107 | 107 | repo_path = str_to_dulwich(wire['path']) |
|
108 | 108 | return Repo(repo_path) |
|
109 | 109 | |
|
110 | 110 | def repo(self, wire, create=False, use_libgit2=False): |
|
111 | 111 | """ |
|
112 | 112 | Get a repository instance for the given path. |
|
113 | 113 | """ |
|
114 | 114 | region = self._cache_region |
|
115 | 115 | context = wire.get('context', None) |
|
116 | 116 | repo_path = wire.get('path', '') |
|
117 | 117 | context_uid = '{}'.format(context) |
|
118 | 118 | cache = wire.get('cache', True) |
|
119 | 119 | cache_on = context and cache |
|
120 | 120 | |
|
121 | 121 | @region.conditional_cache_on_arguments(condition=cache_on) |
|
122 | 122 | def create_new_repo(_repo_type, _repo_path, _context_uid, _use_libgit2): |
|
123 | 123 | return self._create_repo(wire, create, use_libgit2) |
|
124 | 124 | |
|
125 | 125 | repo = create_new_repo(self.repo_type, repo_path, context_uid, use_libgit2) |
|
126 | 126 | return repo |
|
127 | 127 | |
|
128 | 128 | def repo_libgit2(self, wire): |
|
129 | 129 | return self.repo(wire, use_libgit2=True) |
|
130 | 130 | |
|
131 | 131 | |
|
132 | 132 | class GitRemote(object): |
|
133 | 133 | |
|
134 | 134 | def __init__(self, factory): |
|
135 | 135 | self._factory = factory |
|
136 | 136 | self.peeled_ref_marker = '^{}' |
|
137 | 137 | self._bulk_methods = { |
|
138 | 138 | "date": self.date, |
|
139 | 139 | "author": self.author, |
|
140 | 140 | "message": self.message, |
|
141 | 141 | "parents": self.parents, |
|
142 | 142 | "_commit": self.revision, |
|
143 | 143 | } |
|
144 | 144 | |
|
145 | 145 | def _wire_to_config(self, wire): |
|
146 | 146 | if 'config' in wire: |
|
147 | 147 | return dict([(x[0] + '_' + x[1], x[2]) for x in wire['config']]) |
|
148 | 148 | return {} |
|
149 | 149 | |
|
150 | 150 | def _remote_conf(self, config): |
|
151 | 151 | params = [ |
|
152 | 152 | '-c', 'core.askpass=""', |
|
153 | 153 | ] |
|
154 | 154 | ssl_cert_dir = config.get('vcs_ssl_dir') |
|
155 | 155 | if ssl_cert_dir: |
|
156 | 156 | params.extend(['-c', 'http.sslCAinfo={}'.format(ssl_cert_dir)]) |
|
157 | 157 | return params |
|
158 | 158 | |
|
159 | 159 | @reraise_safe_exceptions |
|
160 | 160 | def is_empty(self, wire): |
|
161 | 161 | repo = self._factory.repo_libgit2(wire) |
|
162 | 162 | |
|
163 | 163 | try: |
|
164 |
|
|
|
164 | has_head = repo.head.name | |
|
165 | if has_head: | |
|
166 | return False | |
|
167 | ||
|
168 | # NOTE(marcink): check again using more expensive method | |
|
169 | return repo.is_empty | |
|
165 | 170 | except Exception: |
|
166 |
|
|
|
171 | pass | |
|
172 | ||
|
173 | return True | |
|
167 | 174 | |
|
168 | 175 | @reraise_safe_exceptions |
|
169 | 176 | def add_object(self, wire, content): |
|
170 | 177 | repo = self._factory.repo(wire) |
|
171 | 178 | blob = objects.Blob() |
|
172 | 179 | blob.set_raw_string(content) |
|
173 | 180 | repo.object_store.add_object(blob) |
|
174 | 181 | return blob.id |
|
175 | 182 | |
|
176 | 183 | @reraise_safe_exceptions |
|
177 | 184 | def assert_correct_path(self, wire): |
|
178 | 185 | try: |
|
179 | 186 | self._factory.repo_libgit2(wire) |
|
180 | 187 | except pygit2.GitError: |
|
181 | 188 | path = wire.get('path') |
|
182 | 189 | tb = traceback.format_exc() |
|
183 | 190 | log.debug("Invalid Git path `%s`, tb: %s", path, tb) |
|
184 | 191 | return False |
|
185 | 192 | |
|
186 | 193 | return True |
|
187 | 194 | |
|
188 | 195 | @reraise_safe_exceptions |
|
189 | 196 | def bare(self, wire): |
|
190 | 197 | repo = self._factory.repo_libgit2(wire) |
|
191 | 198 | return repo.is_bare |
|
192 | 199 | |
|
193 | 200 | @reraise_safe_exceptions |
|
194 | 201 | def blob_as_pretty_string(self, wire, sha): |
|
195 | 202 | repo_init = self._factory.repo_libgit2(wire) |
|
196 | 203 | with repo_init as repo: |
|
197 | 204 | blob_obj = repo[sha] |
|
198 | 205 | blob = blob_obj.data |
|
199 | 206 | return blob |
|
200 | 207 | |
|
201 | 208 | @reraise_safe_exceptions |
|
202 | 209 | def blob_raw_length(self, wire, sha): |
|
203 | 210 | repo_init = self._factory.repo_libgit2(wire) |
|
204 | 211 | with repo_init as repo: |
|
205 | 212 | blob = repo[sha] |
|
206 | 213 | return blob.size |
|
207 | 214 | |
|
208 | 215 | def _parse_lfs_pointer(self, raw_content): |
|
209 | 216 | |
|
210 | 217 | spec_string = 'version https://git-lfs.github.com/spec' |
|
211 | 218 | if raw_content and raw_content.startswith(spec_string): |
|
212 | 219 | pattern = re.compile(r""" |
|
213 | 220 | (?:\n)? |
|
214 | 221 | ^version[ ]https://git-lfs\.github\.com/spec/(?P<spec_ver>v\d+)\n |
|
215 | 222 | ^oid[ ] sha256:(?P<oid_hash>[0-9a-f]{64})\n |
|
216 | 223 | ^size[ ](?P<oid_size>[0-9]+)\n |
|
217 | 224 | (?:\n)? |
|
218 | 225 | """, re.VERBOSE | re.MULTILINE) |
|
219 | 226 | match = pattern.match(raw_content) |
|
220 | 227 | if match: |
|
221 | 228 | return match.groupdict() |
|
222 | 229 | |
|
223 | 230 | return {} |
|
224 | 231 | |
|
225 | 232 | @reraise_safe_exceptions |
|
226 | 233 | def is_large_file(self, wire, sha): |
|
227 | 234 | repo_init = self._factory.repo_libgit2(wire) |
|
228 | 235 | |
|
229 | 236 | with repo_init as repo: |
|
230 | 237 | blob = repo[sha] |
|
231 | 238 | if blob.is_binary: |
|
232 | 239 | return {} |
|
233 | 240 | |
|
234 | 241 | return self._parse_lfs_pointer(blob.data) |
|
235 | 242 | |
|
236 | 243 | @reraise_safe_exceptions |
|
237 | 244 | def in_largefiles_store(self, wire, oid): |
|
238 | 245 | repo = self._factory.repo_libgit2(wire) |
|
239 | 246 | conf = self._wire_to_config(wire) |
|
240 | 247 | |
|
241 | 248 | store_location = conf.get('vcs_git_lfs_store_location') |
|
242 | 249 | if store_location: |
|
243 | 250 | repo_name = repo.path |
|
244 | 251 | store = LFSOidStore( |
|
245 | 252 | oid=oid, repo=repo_name, store_location=store_location) |
|
246 | 253 | return store.has_oid() |
|
247 | 254 | |
|
248 | 255 | return False |
|
249 | 256 | |
|
250 | 257 | @reraise_safe_exceptions |
|
251 | 258 | def store_path(self, wire, oid): |
|
252 | 259 | repo = self._factory.repo_libgit2(wire) |
|
253 | 260 | conf = self._wire_to_config(wire) |
|
254 | 261 | |
|
255 | 262 | store_location = conf.get('vcs_git_lfs_store_location') |
|
256 | 263 | if store_location: |
|
257 | 264 | repo_name = repo.path |
|
258 | 265 | store = LFSOidStore( |
|
259 | 266 | oid=oid, repo=repo_name, store_location=store_location) |
|
260 | 267 | return store.oid_path |
|
261 | 268 | raise ValueError('Unable to fetch oid with path {}'.format(oid)) |
|
262 | 269 | |
|
263 | 270 | @reraise_safe_exceptions |
|
264 | 271 | def bulk_request(self, wire, rev, pre_load): |
|
265 | 272 | result = {} |
|
266 | 273 | for attr in pre_load: |
|
267 | 274 | try: |
|
268 | 275 | method = self._bulk_methods[attr] |
|
269 | 276 | args = [wire, rev] |
|
270 | 277 | result[attr] = method(*args) |
|
271 | 278 | except KeyError as e: |
|
272 | 279 | raise exceptions.VcsException(e)("Unknown bulk attribute: %s" % attr) |
|
273 | 280 | return result |
|
274 | 281 | |
|
275 | 282 | def _build_opener(self, url): |
|
276 | 283 | handlers = [] |
|
277 | 284 | url_obj = url_parser(url) |
|
278 | 285 | _, authinfo = url_obj.authinfo() |
|
279 | 286 | |
|
280 | 287 | if authinfo: |
|
281 | 288 | # create a password manager |
|
282 | 289 | passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm() |
|
283 | 290 | passmgr.add_password(*authinfo) |
|
284 | 291 | |
|
285 | 292 | handlers.extend((httpbasicauthhandler(passmgr), |
|
286 | 293 | httpdigestauthhandler(passmgr))) |
|
287 | 294 | |
|
288 | 295 | return urllib2.build_opener(*handlers) |
|
289 | 296 | |
|
290 | 297 | def _type_id_to_name(self, type_id): |
|
291 | 298 | return { |
|
292 | 299 | 1: b'commit', |
|
293 | 300 | 2: b'tree', |
|
294 | 301 | 3: b'blob', |
|
295 | 302 | 4: b'tag' |
|
296 | 303 | }[type_id] |
|
297 | 304 | |
|
298 | 305 | @reraise_safe_exceptions |
|
299 | 306 | def check_url(self, url, config): |
|
300 | 307 | url_obj = url_parser(url) |
|
301 | 308 | test_uri, _ = url_obj.authinfo() |
|
302 | 309 | url_obj.passwd = '*****' if url_obj.passwd else url_obj.passwd |
|
303 | 310 | url_obj.query = obfuscate_qs(url_obj.query) |
|
304 | 311 | cleaned_uri = str(url_obj) |
|
305 | 312 | log.info("Checking URL for remote cloning/import: %s", cleaned_uri) |
|
306 | 313 | |
|
307 | 314 | if not test_uri.endswith('info/refs'): |
|
308 | 315 | test_uri = test_uri.rstrip('/') + '/info/refs' |
|
309 | 316 | |
|
310 | 317 | o = self._build_opener(url) |
|
311 | 318 | o.addheaders = [('User-Agent', 'git/1.7.8.0')] # fake some git |
|
312 | 319 | |
|
313 | 320 | q = {"service": 'git-upload-pack'} |
|
314 | 321 | qs = '?%s' % urllib.urlencode(q) |
|
315 | 322 | cu = "%s%s" % (test_uri, qs) |
|
316 | 323 | req = urllib2.Request(cu, None, {}) |
|
317 | 324 | |
|
318 | 325 | try: |
|
319 | 326 | log.debug("Trying to open URL %s", cleaned_uri) |
|
320 | 327 | resp = o.open(req) |
|
321 | 328 | if resp.code != 200: |
|
322 | 329 | raise exceptions.URLError()('Return Code is not 200') |
|
323 | 330 | except Exception as e: |
|
324 | 331 | log.warning("URL cannot be opened: %s", cleaned_uri, exc_info=True) |
|
325 | 332 | # means it cannot be cloned |
|
326 | 333 | raise exceptions.URLError(e)("[%s] org_exc: %s" % (cleaned_uri, e)) |
|
327 | 334 | |
|
328 | 335 | # now detect if it's proper git repo |
|
329 | 336 | gitdata = resp.read() |
|
330 | 337 | if 'service=git-upload-pack' in gitdata: |
|
331 | 338 | pass |
|
332 | 339 | elif re.findall(r'[0-9a-fA-F]{40}\s+refs', gitdata): |
|
333 | 340 | # old style git can return some other format ! |
|
334 | 341 | pass |
|
335 | 342 | else: |
|
336 | 343 | raise exceptions.URLError()( |
|
337 | 344 | "url [%s] does not look like an git" % (cleaned_uri,)) |
|
338 | 345 | |
|
339 | 346 | return True |
|
340 | 347 | |
|
341 | 348 | @reraise_safe_exceptions |
|
342 | 349 | def clone(self, wire, url, deferred, valid_refs, update_after_clone): |
|
343 | 350 | # TODO(marcink): deprecate this method. Last i checked we don't use it anymore |
|
344 | 351 | remote_refs = self.pull(wire, url, apply_refs=False) |
|
345 | 352 | repo = self._factory.repo(wire) |
|
346 | 353 | if isinstance(valid_refs, list): |
|
347 | 354 | valid_refs = tuple(valid_refs) |
|
348 | 355 | |
|
349 | 356 | for k in remote_refs: |
|
350 | 357 | # only parse heads/tags and skip so called deferred tags |
|
351 | 358 | if k.startswith(valid_refs) and not k.endswith(deferred): |
|
352 | 359 | repo[k] = remote_refs[k] |
|
353 | 360 | |
|
354 | 361 | if update_after_clone: |
|
355 | 362 | # we want to checkout HEAD |
|
356 | 363 | repo["HEAD"] = remote_refs["HEAD"] |
|
357 | 364 | index.build_index_from_tree(repo.path, repo.index_path(), |
|
358 | 365 | repo.object_store, repo["HEAD"].tree) |
|
359 | 366 | |
|
360 | 367 | # TODO: this is quite complex, check if that can be simplified |
|
361 | 368 | @reraise_safe_exceptions |
|
362 | 369 | def commit(self, wire, commit_data, branch, commit_tree, updated, removed): |
|
363 | 370 | repo = self._factory.repo(wire) |
|
364 | 371 | object_store = repo.object_store |
|
365 | 372 | |
|
366 | 373 | # Create tree and populates it with blobs |
|
367 | 374 | commit_tree = commit_tree and repo[commit_tree] or objects.Tree() |
|
368 | 375 | |
|
369 | 376 | for node in updated: |
|
370 | 377 | # Compute subdirs if needed |
|
371 | 378 | dirpath, nodename = vcspath.split(node['path']) |
|
372 | 379 | dirnames = map(safe_str, dirpath and dirpath.split('/') or []) |
|
373 | 380 | parent = commit_tree |
|
374 | 381 | ancestors = [('', parent)] |
|
375 | 382 | |
|
376 | 383 | # Tries to dig for the deepest existing tree |
|
377 | 384 | while dirnames: |
|
378 | 385 | curdir = dirnames.pop(0) |
|
379 | 386 | try: |
|
380 | 387 | dir_id = parent[curdir][1] |
|
381 | 388 | except KeyError: |
|
382 | 389 | # put curdir back into dirnames and stops |
|
383 | 390 | dirnames.insert(0, curdir) |
|
384 | 391 | break |
|
385 | 392 | else: |
|
386 | 393 | # If found, updates parent |
|
387 | 394 | parent = repo[dir_id] |
|
388 | 395 | ancestors.append((curdir, parent)) |
|
389 | 396 | # Now parent is deepest existing tree and we need to create |
|
390 | 397 | # subtrees for dirnames (in reverse order) |
|
391 | 398 | # [this only applies for nodes from added] |
|
392 | 399 | new_trees = [] |
|
393 | 400 | |
|
394 | 401 | blob = objects.Blob.from_string(node['content']) |
|
395 | 402 | |
|
396 | 403 | if dirnames: |
|
397 | 404 | # If there are trees which should be created we need to build |
|
398 | 405 | # them now (in reverse order) |
|
399 | 406 | reversed_dirnames = list(reversed(dirnames)) |
|
400 | 407 | curtree = objects.Tree() |
|
401 | 408 | curtree[node['node_path']] = node['mode'], blob.id |
|
402 | 409 | new_trees.append(curtree) |
|
403 | 410 | for dirname in reversed_dirnames[:-1]: |
|
404 | 411 | newtree = objects.Tree() |
|
405 | 412 | newtree[dirname] = (DIR_STAT, curtree.id) |
|
406 | 413 | new_trees.append(newtree) |
|
407 | 414 | curtree = newtree |
|
408 | 415 | parent[reversed_dirnames[-1]] = (DIR_STAT, curtree.id) |
|
409 | 416 | else: |
|
410 | 417 | parent.add(name=node['node_path'], mode=node['mode'], hexsha=blob.id) |
|
411 | 418 | |
|
412 | 419 | new_trees.append(parent) |
|
413 | 420 | # Update ancestors |
|
414 | 421 | reversed_ancestors = reversed( |
|
415 | 422 | [(a[1], b[1], b[0]) for a, b in zip(ancestors, ancestors[1:])]) |
|
416 | 423 | for parent, tree, path in reversed_ancestors: |
|
417 | 424 | parent[path] = (DIR_STAT, tree.id) |
|
418 | 425 | object_store.add_object(tree) |
|
419 | 426 | |
|
420 | 427 | object_store.add_object(blob) |
|
421 | 428 | for tree in new_trees: |
|
422 | 429 | object_store.add_object(tree) |
|
423 | 430 | |
|
424 | 431 | for node_path in removed: |
|
425 | 432 | paths = node_path.split('/') |
|
426 | 433 | tree = commit_tree |
|
427 | 434 | trees = [tree] |
|
428 | 435 | # Traverse deep into the forest... |
|
429 | 436 | for path in paths: |
|
430 | 437 | try: |
|
431 | 438 | obj = repo[tree[path][1]] |
|
432 | 439 | if isinstance(obj, objects.Tree): |
|
433 | 440 | trees.append(obj) |
|
434 | 441 | tree = obj |
|
435 | 442 | except KeyError: |
|
436 | 443 | break |
|
437 | 444 | # Cut down the blob and all rotten trees on the way back... |
|
438 | 445 | for path, tree in reversed(zip(paths, trees)): |
|
439 | 446 | del tree[path] |
|
440 | 447 | if tree: |
|
441 | 448 | # This tree still has elements - don't remove it or any |
|
442 | 449 | # of it's parents |
|
443 | 450 | break |
|
444 | 451 | |
|
445 | 452 | object_store.add_object(commit_tree) |
|
446 | 453 | |
|
447 | 454 | # Create commit |
|
448 | 455 | commit = objects.Commit() |
|
449 | 456 | commit.tree = commit_tree.id |
|
450 | 457 | for k, v in commit_data.iteritems(): |
|
451 | 458 | setattr(commit, k, v) |
|
452 | 459 | object_store.add_object(commit) |
|
453 | 460 | |
|
454 | 461 | self.create_branch(wire, branch, commit.id) |
|
455 | 462 | |
|
456 | 463 | # dulwich set-ref |
|
457 | 464 | ref = 'refs/heads/%s' % branch |
|
458 | 465 | repo.refs[ref] = commit.id |
|
459 | 466 | |
|
460 | 467 | return commit.id |
|
461 | 468 | |
|
462 | 469 | @reraise_safe_exceptions |
|
463 | 470 | def pull(self, wire, url, apply_refs=True, refs=None, update_after=False): |
|
464 | 471 | if url != 'default' and '://' not in url: |
|
465 | 472 | client = LocalGitClient(url) |
|
466 | 473 | else: |
|
467 | 474 | url_obj = url_parser(url) |
|
468 | 475 | o = self._build_opener(url) |
|
469 | 476 | url, _ = url_obj.authinfo() |
|
470 | 477 | client = HttpGitClient(base_url=url, opener=o) |
|
471 | 478 | repo = self._factory.repo(wire) |
|
472 | 479 | |
|
473 | 480 | determine_wants = repo.object_store.determine_wants_all |
|
474 | 481 | if refs: |
|
475 | 482 | def determine_wants_requested(references): |
|
476 | 483 | return [references[r] for r in references if r in refs] |
|
477 | 484 | determine_wants = determine_wants_requested |
|
478 | 485 | |
|
479 | 486 | try: |
|
480 | 487 | remote_refs = client.fetch( |
|
481 | 488 | path=url, target=repo, determine_wants=determine_wants) |
|
482 | 489 | except NotGitRepository as e: |
|
483 | 490 | log.warning( |
|
484 | 491 | 'Trying to fetch from "%s" failed, not a Git repository.', url) |
|
485 | 492 | # Exception can contain unicode which we convert |
|
486 | 493 | raise exceptions.AbortException(e)(repr(e)) |
|
487 | 494 | |
|
488 | 495 | # mikhail: client.fetch() returns all the remote refs, but fetches only |
|
489 | 496 | # refs filtered by `determine_wants` function. We need to filter result |
|
490 | 497 | # as well |
|
491 | 498 | if refs: |
|
492 | 499 | remote_refs = {k: remote_refs[k] for k in remote_refs if k in refs} |
|
493 | 500 | |
|
494 | 501 | if apply_refs: |
|
495 | 502 | # TODO: johbo: Needs proper test coverage with a git repository |
|
496 | 503 | # that contains a tag object, so that we would end up with |
|
497 | 504 | # a peeled ref at this point. |
|
498 | 505 | for k in remote_refs: |
|
499 | 506 | if k.endswith(self.peeled_ref_marker): |
|
500 | 507 | log.debug("Skipping peeled reference %s", k) |
|
501 | 508 | continue |
|
502 | 509 | repo[k] = remote_refs[k] |
|
503 | 510 | |
|
504 | 511 | if refs and not update_after: |
|
505 | 512 | # mikhail: explicitly set the head to the last ref. |
|
506 | 513 | repo['HEAD'] = remote_refs[refs[-1]] |
|
507 | 514 | |
|
508 | 515 | if update_after: |
|
509 | 516 | # we want to checkout HEAD |
|
510 | 517 | repo["HEAD"] = remote_refs["HEAD"] |
|
511 | 518 | index.build_index_from_tree(repo.path, repo.index_path(), |
|
512 | 519 | repo.object_store, repo["HEAD"].tree) |
|
513 | 520 | return remote_refs |
|
514 | 521 | |
|
515 | 522 | @reraise_safe_exceptions |
|
516 | 523 | def sync_fetch(self, wire, url, refs=None): |
|
517 | 524 | repo = self._factory.repo(wire) |
|
518 | 525 | if refs and not isinstance(refs, (list, tuple)): |
|
519 | 526 | refs = [refs] |
|
520 | 527 | config = self._wire_to_config(wire) |
|
521 | 528 | # get all remote refs we'll use to fetch later |
|
522 | 529 | output, __ = self.run_git_command( |
|
523 | 530 | wire, ['ls-remote', url], fail_on_stderr=False, |
|
524 | 531 | _copts=self._remote_conf(config), |
|
525 | 532 | extra_env={'GIT_TERMINAL_PROMPT': '0'}) |
|
526 | 533 | |
|
527 | 534 | remote_refs = collections.OrderedDict() |
|
528 | 535 | fetch_refs = [] |
|
529 | 536 | |
|
530 | 537 | for ref_line in output.splitlines(): |
|
531 | 538 | sha, ref = ref_line.split('\t') |
|
532 | 539 | sha = sha.strip() |
|
533 | 540 | if ref in remote_refs: |
|
534 | 541 | # duplicate, skip |
|
535 | 542 | continue |
|
536 | 543 | if ref.endswith(self.peeled_ref_marker): |
|
537 | 544 | log.debug("Skipping peeled reference %s", ref) |
|
538 | 545 | continue |
|
539 | 546 | # don't sync HEAD |
|
540 | 547 | if ref in ['HEAD']: |
|
541 | 548 | continue |
|
542 | 549 | |
|
543 | 550 | remote_refs[ref] = sha |
|
544 | 551 | |
|
545 | 552 | if refs and sha in refs: |
|
546 | 553 | # we filter fetch using our specified refs |
|
547 | 554 | fetch_refs.append('{}:{}'.format(ref, ref)) |
|
548 | 555 | elif not refs: |
|
549 | 556 | fetch_refs.append('{}:{}'.format(ref, ref)) |
|
550 | 557 | log.debug('Finished obtaining fetch refs, total: %s', len(fetch_refs)) |
|
551 | 558 | if fetch_refs: |
|
552 | 559 | for chunk in more_itertools.chunked(fetch_refs, 1024 * 4): |
|
553 | 560 | fetch_refs_chunks = list(chunk) |
|
554 | 561 | log.debug('Fetching %s refs from import url', len(fetch_refs_chunks)) |
|
555 | 562 | _out, _err = self.run_git_command( |
|
556 | 563 | wire, ['fetch', url, '--force', '--prune', '--'] + fetch_refs_chunks, |
|
557 | 564 | fail_on_stderr=False, |
|
558 | 565 | _copts=self._remote_conf(config), |
|
559 | 566 | extra_env={'GIT_TERMINAL_PROMPT': '0'}) |
|
560 | 567 | |
|
561 | 568 | return remote_refs |
|
562 | 569 | |
|
563 | 570 | @reraise_safe_exceptions |
|
564 | 571 | def sync_push(self, wire, url, refs=None): |
|
565 | 572 | if not self.check_url(url, wire): |
|
566 | 573 | return |
|
567 | 574 | config = self._wire_to_config(wire) |
|
568 | 575 | repo = self._factory.repo(wire) |
|
569 | 576 | self.run_git_command( |
|
570 | 577 | wire, ['push', url, '--mirror'], fail_on_stderr=False, |
|
571 | 578 | _copts=self._remote_conf(config), |
|
572 | 579 | extra_env={'GIT_TERMINAL_PROMPT': '0'}) |
|
573 | 580 | |
|
574 | 581 | @reraise_safe_exceptions |
|
575 | 582 | def get_remote_refs(self, wire, url): |
|
576 | 583 | repo = Repo(url) |
|
577 | 584 | return repo.get_refs() |
|
578 | 585 | |
|
579 | 586 | @reraise_safe_exceptions |
|
580 | 587 | def get_description(self, wire): |
|
581 | 588 | repo = self._factory.repo(wire) |
|
582 | 589 | return repo.get_description() |
|
583 | 590 | |
|
584 | 591 | @reraise_safe_exceptions |
|
585 | 592 | def get_missing_revs(self, wire, rev1, rev2, path2): |
|
586 | 593 | repo = self._factory.repo(wire) |
|
587 | 594 | LocalGitClient(thin_packs=False).fetch(path2, repo) |
|
588 | 595 | |
|
589 | 596 | wire_remote = wire.copy() |
|
590 | 597 | wire_remote['path'] = path2 |
|
591 | 598 | repo_remote = self._factory.repo(wire_remote) |
|
592 | 599 | LocalGitClient(thin_packs=False).fetch(wire["path"], repo_remote) |
|
593 | 600 | |
|
594 | 601 | revs = [ |
|
595 | 602 | x.commit.id |
|
596 | 603 | for x in repo_remote.get_walker(include=[rev2], exclude=[rev1])] |
|
597 | 604 | return revs |
|
598 | 605 | |
|
599 | 606 | @reraise_safe_exceptions |
|
600 | 607 | def get_object(self, wire, sha): |
|
601 | 608 | repo = self._factory.repo_libgit2(wire) |
|
602 | 609 | |
|
603 | 610 | missing_commit_err = 'Commit {} does not exist for `{}`'.format(sha, wire['path']) |
|
604 | 611 | try: |
|
605 | 612 | commit = repo.revparse_single(sha) |
|
606 | 613 | except (KeyError, ValueError) as e: |
|
607 | 614 | raise exceptions.LookupException(e)(missing_commit_err) |
|
608 | 615 | |
|
609 | 616 | if isinstance(commit, pygit2.Tag): |
|
610 | 617 | commit = repo.get(commit.target) |
|
611 | 618 | |
|
612 | 619 | # check for dangling commit |
|
613 | 620 | branches = [x for x in repo.branches.with_commit(commit.hex)] |
|
614 | 621 | if not branches: |
|
615 | 622 | raise exceptions.LookupException(None)(missing_commit_err) |
|
616 | 623 | |
|
617 | 624 | commit_id = commit.hex |
|
618 | 625 | type_id = commit.type |
|
619 | 626 | |
|
620 | 627 | return { |
|
621 | 628 | 'id': commit_id, |
|
622 | 629 | 'type': self._type_id_to_name(type_id), |
|
623 | 630 | 'commit_id': commit_id, |
|
624 | 631 | 'idx': 0 |
|
625 | 632 | } |
|
626 | 633 | |
|
627 | 634 | @reraise_safe_exceptions |
|
628 | 635 | def get_refs(self, wire): |
|
629 | 636 | repo = self._factory.repo_libgit2(wire) |
|
630 | 637 | |
|
631 | 638 | result = {} |
|
632 | 639 | for ref in repo.references: |
|
633 | 640 | peeled_sha = repo.lookup_reference(ref).peel() |
|
634 | 641 | result[ref] = peeled_sha.hex |
|
635 | 642 | |
|
636 | 643 | return result |
|
637 | 644 | |
|
638 | 645 | @reraise_safe_exceptions |
|
639 | 646 | def head(self, wire, show_exc=True): |
|
640 | 647 | repo = self._factory.repo_libgit2(wire) |
|
641 | 648 | try: |
|
642 | 649 | return repo.head.peel().hex |
|
643 | 650 | except Exception: |
|
644 | 651 | if show_exc: |
|
645 | 652 | raise |
|
646 | 653 | |
|
647 | 654 | @reraise_safe_exceptions |
|
648 | 655 | def init(self, wire): |
|
649 | 656 | repo_path = str_to_dulwich(wire['path']) |
|
650 | 657 | self.repo = Repo.init(repo_path) |
|
651 | 658 | |
|
652 | 659 | @reraise_safe_exceptions |
|
653 | 660 | def init_bare(self, wire): |
|
654 | 661 | repo_path = str_to_dulwich(wire['path']) |
|
655 | 662 | self.repo = Repo.init_bare(repo_path) |
|
656 | 663 | |
|
657 | 664 | @reraise_safe_exceptions |
|
658 | 665 | def revision(self, wire, rev): |
|
659 | 666 | repo = self._factory.repo_libgit2(wire) |
|
660 | 667 | commit = repo[rev] |
|
661 | 668 | obj_data = { |
|
662 | 669 | 'id': commit.id.hex, |
|
663 | 670 | } |
|
664 | 671 | # tree objects itself don't have tree_id attribute |
|
665 | 672 | if hasattr(commit, 'tree_id'): |
|
666 | 673 | obj_data['tree'] = commit.tree_id.hex |
|
667 | 674 | |
|
668 | 675 | return obj_data |
|
669 | 676 | |
|
670 | 677 | @reraise_safe_exceptions |
|
671 | 678 | def date(self, wire, rev): |
|
672 | 679 | repo = self._factory.repo_libgit2(wire) |
|
673 | 680 | commit = repo[rev] |
|
674 | 681 | # TODO(marcink): check dulwich difference of offset vs timezone |
|
675 | 682 | return [commit.commit_time, commit.commit_time_offset] |
|
676 | 683 | |
|
677 | 684 | @reraise_safe_exceptions |
|
678 | 685 | def author(self, wire, rev): |
|
679 | 686 | repo = self._factory.repo_libgit2(wire) |
|
680 | 687 | commit = repo[rev] |
|
681 | 688 | if commit.author.email: |
|
682 | 689 | return u"{} <{}>".format(commit.author.name, commit.author.email) |
|
683 | 690 | |
|
684 | 691 | return u"{}".format(commit.author.raw_name) |
|
685 | 692 | |
|
686 | 693 | @reraise_safe_exceptions |
|
687 | 694 | def message(self, wire, rev): |
|
688 | 695 | repo = self._factory.repo_libgit2(wire) |
|
689 | 696 | commit = repo[rev] |
|
690 | 697 | return commit.message |
|
691 | 698 | |
|
692 | 699 | @reraise_safe_exceptions |
|
693 | 700 | def parents(self, wire, rev): |
|
694 | 701 | repo = self._factory.repo_libgit2(wire) |
|
695 | 702 | commit = repo[rev] |
|
696 | 703 | return [x.hex for x in commit.parent_ids] |
|
697 | 704 | |
|
698 | 705 | @reraise_safe_exceptions |
|
699 | 706 | def set_refs(self, wire, key, value): |
|
700 | 707 | repo = self._factory.repo_libgit2(wire) |
|
701 | 708 | repo.references.create(key, value, force=True) |
|
702 | 709 | |
|
703 | 710 | @reraise_safe_exceptions |
|
704 | 711 | def create_branch(self, wire, branch_name, commit_id, force=False): |
|
705 | 712 | repo = self._factory.repo_libgit2(wire) |
|
706 | 713 | commit = repo[commit_id] |
|
707 | 714 | |
|
708 | 715 | if force: |
|
709 | 716 | repo.branches.local.create(branch_name, commit, force=force) |
|
710 | 717 | elif not repo.branches.get(branch_name): |
|
711 | 718 | # create only if that branch isn't existing |
|
712 | 719 | repo.branches.local.create(branch_name, commit, force=force) |
|
713 | 720 | |
|
714 | 721 | @reraise_safe_exceptions |
|
715 | 722 | def remove_ref(self, wire, key): |
|
716 | 723 | repo = self._factory.repo_libgit2(wire) |
|
717 | 724 | repo.references.delete(key) |
|
718 | 725 | |
|
719 | 726 | @reraise_safe_exceptions |
|
720 | 727 | def tag_remove(self, wire, tag_name): |
|
721 | 728 | repo = self._factory.repo_libgit2(wire) |
|
722 | 729 | key = 'refs/tags/{}'.format(tag_name) |
|
723 | 730 | repo.references.delete(key) |
|
724 | 731 | |
|
725 | 732 | @reraise_safe_exceptions |
|
726 | 733 | def tree_changes(self, wire, source_id, target_id): |
|
727 | 734 | # TODO(marcink): remove this seems it's only used by tests |
|
728 | 735 | repo = self._factory.repo(wire) |
|
729 | 736 | source = repo[source_id].tree if source_id else None |
|
730 | 737 | target = repo[target_id].tree |
|
731 | 738 | result = repo.object_store.tree_changes(source, target) |
|
732 | 739 | return list(result) |
|
733 | 740 | |
|
734 | 741 | @reraise_safe_exceptions |
|
735 | 742 | def tree_and_type_for_path(self, wire, commit_id, path): |
|
736 | 743 | repo_init = self._factory.repo_libgit2(wire) |
|
737 | 744 | |
|
738 | 745 | with repo_init as repo: |
|
739 | 746 | commit = repo[commit_id] |
|
740 | 747 | try: |
|
741 | 748 | tree = commit.tree[path] |
|
742 | 749 | except KeyError: |
|
743 | 750 | return None, None, None |
|
744 | 751 | |
|
745 | 752 | return tree.id.hex, tree.type, tree.filemode |
|
746 | 753 | |
|
747 | 754 | @reraise_safe_exceptions |
|
748 | 755 | def tree_items(self, wire, tree_id): |
|
749 | 756 | repo_init = self._factory.repo_libgit2(wire) |
|
750 | 757 | |
|
751 | 758 | with repo_init as repo: |
|
752 | 759 | try: |
|
753 | 760 | tree = repo[tree_id] |
|
754 | 761 | except KeyError: |
|
755 | 762 | raise ObjectMissing('No tree with id: {}'.format(tree_id)) |
|
756 | 763 | |
|
757 | 764 | result = [] |
|
758 | 765 | for item in tree: |
|
759 | 766 | item_sha = item.hex |
|
760 | 767 | item_mode = item.filemode |
|
761 | 768 | item_type = item.type |
|
762 | 769 | |
|
763 | 770 | if item_type == 'commit': |
|
764 | 771 | # NOTE(marcink): submodules we translate to 'link' for backward compat |
|
765 | 772 | item_type = 'link' |
|
766 | 773 | |
|
767 | 774 | result.append((item.name, item_mode, item_sha, item_type)) |
|
768 | 775 | return result |
|
769 | 776 | |
|
770 | 777 | @reraise_safe_exceptions |
|
771 | 778 | def update_server_info(self, wire): |
|
772 | 779 | repo = self._factory.repo(wire) |
|
773 | 780 | update_server_info(repo) |
|
774 | 781 | |
|
775 | 782 | @reraise_safe_exceptions |
|
776 | 783 | def discover_git_version(self): |
|
777 | 784 | stdout, _ = self.run_git_command( |
|
778 | 785 | {}, ['--version'], _bare=True, _safe=True) |
|
779 | 786 | prefix = 'git version' |
|
780 | 787 | if stdout.startswith(prefix): |
|
781 | 788 | stdout = stdout[len(prefix):] |
|
782 | 789 | return stdout.strip() |
|
783 | 790 | |
|
784 | 791 | @reraise_safe_exceptions |
|
785 | 792 | def get_all_commit_ids(self, wire): |
|
786 | if self.is_empty(wire): | |
|
787 | return [] | |
|
788 | 793 | |
|
789 | 794 | cmd = ['rev-list', '--reverse', '--date-order', '--branches', '--tags'] |
|
790 | 795 | try: |
|
791 | 796 | output, __ = self.run_git_command(wire, cmd) |
|
792 | 797 | return output.splitlines() |
|
793 | 798 | except Exception: |
|
794 | 799 | # Can be raised for empty repositories |
|
795 | 800 | return [] |
|
796 | 801 | |
|
797 | 802 | @reraise_safe_exceptions |
|
798 | 803 | def run_git_command(self, wire, cmd, **opts): |
|
799 | 804 | path = wire.get('path', None) |
|
800 | 805 | |
|
801 | 806 | if path and os.path.isdir(path): |
|
802 | 807 | opts['cwd'] = path |
|
803 | 808 | |
|
804 | 809 | if '_bare' in opts: |
|
805 | 810 | _copts = [] |
|
806 | 811 | del opts['_bare'] |
|
807 | 812 | else: |
|
808 | 813 | _copts = ['-c', 'core.quotepath=false', ] |
|
809 | 814 | safe_call = False |
|
810 | 815 | if '_safe' in opts: |
|
811 | 816 | # no exc on failure |
|
812 | 817 | del opts['_safe'] |
|
813 | 818 | safe_call = True |
|
814 | 819 | |
|
815 | 820 | if '_copts' in opts: |
|
816 | 821 | _copts.extend(opts['_copts'] or []) |
|
817 | 822 | del opts['_copts'] |
|
818 | 823 | |
|
819 | 824 | gitenv = os.environ.copy() |
|
820 | 825 | gitenv.update(opts.pop('extra_env', {})) |
|
821 | 826 | # need to clean fix GIT_DIR ! |
|
822 | 827 | if 'GIT_DIR' in gitenv: |
|
823 | 828 | del gitenv['GIT_DIR'] |
|
824 | 829 | gitenv['GIT_CONFIG_NOGLOBAL'] = '1' |
|
825 | 830 | gitenv['GIT_DISCOVERY_ACROSS_FILESYSTEM'] = '1' |
|
826 | 831 | |
|
827 | 832 | cmd = [settings.GIT_EXECUTABLE] + _copts + cmd |
|
828 | 833 | _opts = {'env': gitenv, 'shell': False} |
|
829 | 834 | |
|
830 | 835 | try: |
|
831 | 836 | _opts.update(opts) |
|
832 | 837 | p = subprocessio.SubprocessIOChunker(cmd, **_opts) |
|
833 | 838 | |
|
834 | 839 | return ''.join(p), ''.join(p.error) |
|
835 | 840 | except (EnvironmentError, OSError) as err: |
|
836 | 841 | cmd = ' '.join(cmd) # human friendly CMD |
|
837 | 842 | tb_err = ("Couldn't run git command (%s).\n" |
|
838 | 843 | "Original error was:%s\n" |
|
839 | 844 | "Call options:%s\n" |
|
840 | 845 | % (cmd, err, _opts)) |
|
841 | 846 | log.exception(tb_err) |
|
842 | 847 | if safe_call: |
|
843 | 848 | return '', err |
|
844 | 849 | else: |
|
845 | 850 | raise exceptions.VcsException()(tb_err) |
|
846 | 851 | |
|
847 | 852 | @reraise_safe_exceptions |
|
848 | 853 | def install_hooks(self, wire, force=False): |
|
849 | 854 | from vcsserver.hook_utils import install_git_hooks |
|
850 | 855 | repo = self._factory.repo(wire) |
|
851 | 856 | return install_git_hooks(repo.path, repo.bare, force_create=force) |
|
852 | 857 | |
|
853 | 858 | @reraise_safe_exceptions |
|
854 | 859 | def get_hooks_info(self, wire): |
|
855 | 860 | from vcsserver.hook_utils import ( |
|
856 | 861 | get_git_pre_hook_version, get_git_post_hook_version) |
|
857 | 862 | repo = self._factory.repo(wire) |
|
858 | 863 | return { |
|
859 | 864 | 'pre_version': get_git_pre_hook_version(repo.path, repo.bare), |
|
860 | 865 | 'post_version': get_git_post_hook_version(repo.path, repo.bare), |
|
861 | 866 | } |
|
862 | 867 | |
|
863 | 868 | |
|
864 | 869 | def str_to_dulwich(value): |
|
865 | 870 | """ |
|
866 | 871 | Dulwich 0.10.1a requires `unicode` objects to be passed in. |
|
867 | 872 | """ |
|
868 | 873 | return value.decode(settings.WIRE_ENCODING) |
General Comments 0
You need to be logged in to leave comments.
Login now