daklib/binary.py

   1 #!/usr/bin/python
   2
   3 """
   4 Functions related debian binary packages
   5
   6 @contact: Debian FTPMaster <ftpmaster@debian.org>
   7 @copyright: 2009  Mike O'Connor <stew@debian.org>
   8 @license: GNU General Public License version 2 or later
   9 """
  10
  11 # This program is free software; you can redistribute it and/or modify
  12 # it under the terms of the GNU General Public License as published by
  13 # the Free Software Foundation; either version 2 of the License, or
  14 # (at your option) any later version.
  15
  16 # This program is distributed in the hope that it will be useful,
  17 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 # GNU General Public License for more details.
  20
  21 # You should have received a copy of the GNU General Public License
  22 # along with this program; if not, write to the Free Software
  23 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  24
  25 ################################################################################
  26
  27 # <Ganneff> are we going the xorg way?
  28 # <Ganneff> a dak without a dak.conf?
  29 # <stew> automatically detect the wrong settings at runtime?
  30 # <Ganneff> yes!
  31 # <mhy> well, we'll probably always need dak.conf (how do you get the database setting
  32 # <mhy> but removing most of the config into the database seems sane
  33 # <Ganneff> mhy: dont spoil the fun
  34 # <Ganneff> mhy: and i know how. we nmap localhost and check all open ports
  35 # <Ganneff> maybe one answers to sql
  36 # <stew> we will discover projectb via avahi
  37 # <mhy> you're both sick
  38 # <mhy> really fucking sick
  39
  40 ################################################################################
  41
  42 import os
  43 import sys
  44 import shutil
  45 import tarfile
  46 import commands
  47 import traceback
  48 import atexit
  49
  50 from debian_bundle import deb822
  51
  52 from dbconn import *
  53 from config import Config
  54 import utils
  55
  56 ################################################################################
  57
  58 __all__ = []
  59
  60 ################################################################################
  61
  62 class Binary(object):
  63     def __init__(self, filename, reject=None):
  64         """
  65         @type filename: string
  66         @param filename: path of a .deb
  67
  68         @type reject: function
  69         @param reject: a function to log reject messages to
  70         """
  71         self.filename = filename
  72         self.tmpdir = None
  73         self.chunks = None
  74         self.wrapped_reject = reject
  75         # Store rejects for later use
  76         self.rejects = []
  77
  78     def reject(self, message):
  79         """
  80         if we were given a reject function, send the reject message,
  81         otherwise send it to stderr.
  82         """
  83         print >> sys.stderr, message
  84         self.rejects.append(message)
  85         if self.wrapped_reject:
  86             self.wrapped_reject(message)
  87
  88     def __del__(self):
  89         """
  90         make sure we cleanup when we are garbage collected.
  91         """
  92         self._cleanup()
  93
  94     def _cleanup(self):
  95         """
  96         we need to remove the temporary directory, if we created one
  97         """
  98         if self.tmpdir and os.path.exists(self.tmpdir):
  99             shutil.rmtree(self.tmpdir)
 100             self.tmpdir = None
 101
 102     def __scan_ar(self):
 103         # get a list of the ar contents
 104         if not self.chunks:
 105
 106             cmd = "ar t %s" % (self.filename)
 107             (result, output) = commands.getstatusoutput(cmd)
 108             if result != 0:
 109                 rejected = True
 110                 print("%s: 'ar t' invocation failed." % (self.filename))
 111                 self.reject("%s: 'ar t' invocation failed." % (self.filename))
 112                 self.reject(utils.prefix_multi_line_string(output, " [ar output:] "))
 113             self.chunks = output.split('\n')
 114
 115
 116
 117     def __unpack(self):
 118         # Internal function which extracts the contents of the .ar to
 119         # a temporary directory
 120
 121         if not self.tmpdir:
 122             tmpdir = utils.temp_dirname()
 123             cwd = os.getcwd()
 124             try:
 125                 os.chdir( tmpdir )
 126                 cmd = "ar x %s %s %s" % (os.path.join(cwd,self.filename), self.chunks[1], self.chunks[2])
 127                 (result, output) = commands.getstatusoutput(cmd)
 128                 if result != 0:
 129                     print("%s: '%s' invocation failed." % (self.filename, cmd))
 130                     self.reject("%s: '%s' invocation failed." % (self.filename, cmd))
 131                     self.reject(utils.prefix_multi_line_string(output, " [ar output:] "))
 132                 else:
 133                     self.tmpdir = tmpdir
 134                     atexit.register( self._cleanup )
 135
 136             finally:
 137                 os.chdir( cwd )
 138
 139     def valid_deb(self, relaxed=False):
 140         """
 141         Check deb contents making sure the .deb contains:
 142           1. debian-binary
 143           2. control.tar.gz
 144           3. data.tar.gz or data.tar.bz2
 145         in that order, and nothing else.
 146         """
 147         self.__scan_ar()
 148         rejected = not self.chunks
 149         if relaxed:
 150             if len(self.chunks) < 3:
 151                 rejected = True
 152                 self.reject("%s: found %d chunks, expected at least 3." % (self.filename, len(self.chunks)))
 153         else:
 154             if len(self.chunks) != 3:
 155                 rejected = True
 156                 self.reject("%s: found %d chunks, expected 3." % (self.filename, len(self.chunks)))
 157         if self.chunks[0] != "debian-binary":
 158             rejected = True
 159             self.reject("%s: first chunk is '%s', expected 'debian-binary'." % (self.filename, self.chunks[0]))
 160         if not rejected and self.chunks[1] != "control.tar.gz":
 161             rejected = True
 162             self.reject("%s: second chunk is '%s', expected 'control.tar.gz'." % (self.filename, self.chunks[1]))
 163         if not rejected and self.chunks[2] not in [ "data.tar.bz2", "data.tar.gz" ]:
 164             rejected = True
 165             self.reject("%s: third chunk is '%s', expected 'data.tar.gz' or 'data.tar.bz2'." % (self.filename, self.chunks[2]))
 166
 167         return not rejected
 168
 169     def scan_package(self, bootstrap_id=0, relaxed=False, session=None):
 170         """
 171         Unpack the .deb, do sanity checking, and gather info from it.
 172
 173         Currently information gathering consists of getting the contents list. In
 174         the hopefully near future, it should also include gathering info from the
 175         control file.
 176
 177         @type bootstrap_id: int
 178         @param bootstrap_id: the id of the binary these packages
 179           should be associated or zero meaning we are not bootstrapping
 180           so insert into a temporary table
 181
 182         @return: True if the deb is valid and contents were imported
 183         """
 184         result = False
 185         rejected = not self.valid_deb(relaxed)
 186         if not rejected:
 187             self.__unpack()
 188
 189
 190             cwd = os.getcwd()
 191             if not rejected and self.tmpdir:
 192                 try:
 193                     os.chdir(self.tmpdir)
 194                     if self.chunks[1] == "control.tar.gz":
 195                         control = tarfile.open(os.path.join(self.tmpdir, "control.tar.gz" ), "r:gz")
 196                         control.extract('./control', self.tmpdir )
 197                     if self.chunks[2] == "data.tar.gz":
 198                         data = tarfile.open(os.path.join(self.tmpdir, "data.tar.gz"), "r:gz")
 199                     elif self.chunks[2] == "data.tar.bz2":
 200                         data = tarfile.open(os.path.join(self.tmpdir, "data.tar.bz2" ), "r:bz2")
 201
 202                     if bootstrap_id:
 203                         result = insert_content_paths(bootstrap_id, [tarinfo.name for tarinfo in data if not tarinfo.isdir()], session)
 204                     else:
 205                         pkgs = deb822.Packages.iter_paragraphs(file(os.path.join(self.tmpdir,'control')))
 206                         pkg = pkgs.next()
 207                         result = insert_pending_content_paths(pkg, [tarinfo.name for tarinfo in data if not tarinfo.isdir()], session)
 208
 209                 except:
 210                     traceback.print_exc()
 211
 212             os.chdir(cwd)
 213         self._cleanup()
 214         return result
 215
 216     def check_utf8_package(self, package):
 217         """
 218         Unpack the .deb, do sanity checking, and gather info from it.
 219
 220         Currently information gathering consists of getting the contents list. In
 221         the hopefully near future, it should also include gathering info from the
 222         control file.
 223
 224         @type package: string
 225         @param package: the name of the package to be checked
 226
 227         @rtype: boolean
 228         @return: True if the deb is valid and contents were imported
 229         """
 230         rejected = not self.valid_deb(True)
 231         self.__unpack()
 232
 233         if not rejected and self.tmpdir:
 234             cwd = os.getcwd()
 235             try:
 236                 os.chdir(self.tmpdir)
 237                 if self.chunks[1] == "control.tar.gz":
 238                     control = tarfile.open(os.path.join(self.tmpdir, "control.tar.gz" ), "r:gz")
 239                     control.extract('control', self.tmpdir )
 240                 if self.chunks[2] == "data.tar.gz":
 241                     data = tarfile.open(os.path.join(self.tmpdir, "data.tar.gz"), "r:gz")
 242                 elif self.chunks[2] == "data.tar.bz2":
 243                     data = tarfile.open(os.path.join(self.tmpdir, "data.tar.bz2" ), "r:bz2")
 244
 245                 for tarinfo in data:
 246                     try:
 247                         unicode( tarinfo.name )
 248                     except:
 249                         print >> sys.stderr, "E: %s has non-unicode filename: %s" % (package,tarinfo.name)
 250
 251             except:
 252                 traceback.print_exc()
 253                 result = False
 254
 255             os.chdir(cwd)
 256
 257 __all__.append('Binary')
 258
 259 def copy_temporary_contents(package, version, archname, deb, reject, session=None):
 260     """
 261     copy the previously stored contents from the temp table to the permanant one
 262
 263     during process-unchecked, the deb should have been scanned and the
 264     contents stored in pending_content_associations
 265     """
 266
 267     cnf = Config()
 268
 269     privatetrans = False
 270     if session is None:
 271         session = DBConn().session()
 272         privatetrans = True
 273
 274     arch = get_architecture(archname, session=session)
 275
 276     # first see if contents exist:
 277     in_pcaq = """SELECT 1 FROM pending_content_associations
 278                                WHERE package=:package
 279                                AND version=:version
 280                                AND architecture=:archid LIMIT 1"""
 281
 282     vals = {'package': package,
 283             'version': version,
 284             'archid': arch.arch_id}
 285
 286     exists = None
 287     check = session.execute(in_pcaq, vals)
 288
 289     if check.rowcount > 0:
 290         # This should NOT happen.  We should have added contents
 291         # during process-unchecked.  if it did, log an error, and send
 292         # an email.
 293         subst = {
 294             "__PACKAGE__": package,
 295             "__VERSION__": version,
 296             "__ARCH__": arch,
 297             "__TO_ADDRESS__": cnf["Dinstall::MyAdminAddress"],
 298             "__DAK_ADDRESS__": cnf["Dinstall::MyEmailAddress"] }
 299
 300         message = utils.TemplateSubst(subst, cnf["Dir::Templates"]+"/missing-contents")
 301         utils.send_mail(message)
 302
 303         # Temporarily disable contents storage until we re-do the table layout
 304         #exists = Binary(deb, reject).scan_package()
 305
 306     if exists:
 307         sql = """INSERT INTO content_associations(binary_pkg,filepath,filename)
 308                  SELECT currval('binaries_id_seq'), filepath, filename FROM pending_content_associations
 309                  WHERE package=:package AND version=:version AND architecture=:archid"""
 310         session.execute(sql, vals)
 311
 312         sql = """DELETE from pending_content_associations
 313                  WHERE package=:package AND version=:version AND architecture=:archid"""
 314         session.execute(sql, vals)
 315         session.commit()
 316
 317     if privatetrans:
 318         session.close()
 319
 320     return exists
 321
 322 __all__.append('copy_temporary_contents')