pkg2zero

   1 #!/usr/bin/env python
   2 # Copyright (C) 2009, Thomas Leonard
   3 # Copyright (C) 2008, Anders F Bjorklund
   4 # See the COPYING file for details, or visit http://0install.net.
   5
   6 import sys, time
   7 from optparse import OptionParser
   8 import tempfile, shutil, os
   9 from xml.dom import minidom
  10 import gzip
  11 try:
  12         import xml.etree.cElementTree as ET # Python 2.5
  13 except ImportError:
  14         try:
  15                 import xml.etree.ElementTree as ET
  16         except ImportError:
  17                 try:
  18                         import cElementTree as ET # http://effbot.org
  19                 except ImportError:
  20                         import elementtree.ElementTree as ET
  21
  22 import subprocess
  23 try:
  24         from subprocess import check_call
  25 except ImportError:
  26         def check_call(*popenargs, **kwargs):
  27                 rc = subprocess.call(*popenargs, **kwargs)
  28                 if rc != 0: raise OSError, rc
  29
  30 from zeroinstall.injector import model, qdom, distro
  31
  32 from support import read_child, add_node, Mappings
  33
  34 manifest_algorithm = 'sha1new'
  35
  36 deb_category_to_freedesktop = {
  37         'devel' : 'Development',
  38         'web' : 'Network',
  39         'graphics' : 'Graphics',
  40         'games' : 'Game',
  41 }
  42
  43 rpm_group_to_freedesktop = {
  44         'Development/Libraries' : 'Development',
  45 }
  46
  47 valid_categories = [
  48         'AudioVideo',
  49         'Audio',
  50         'Video',
  51         'Development',
  52         'Education',
  53         'Game',
  54         'Graphics',
  55         'Network',
  56         'Office',
  57         'Settings',
  58         'System',
  59         'Utility',
  60 ]
  61
  62 # Parse command-line arguments
  63
  64 parser = OptionParser('usage: %prog [options] http://.../package.deb [target-feed.xml]\n'
  65                       '       %prog [options] http://.../package.rpm [target-feed.xml]\n'
  66                       '       %prog [options] package-name [target-feed.xml]\n'
  67                       'Publish a Debian or RPM package in a Zero Install feed.\n'
  68                       "target-feed.xml is created if it doesn't already exist.")
  69 parser.add_option("-r", "--repomd-file", help="repository metadata file")
  70 parser.add_option("", "--path", help="location of packages [5/os/i386]")
  71 parser.add_option("-p", "--packages-file", help="Debian package index file")
  72 parser.add_option("-m", "--mirror", help="location of packages [http://ftp.debian.org/debian] or [http://mirror.centos.org/centos]")
  73 parser.add_option("-k", "--key", help="key to use for signing")
  74 (options, args) = parser.parse_args()
  75
  76 if len(args) < 1 or len(args) > 2:
  77         parser.print_help()
  78         sys.exit(1)
  79
  80 # Load dependency mappings
  81 mappings = Mappings()
  82
  83 class Package:
  84         name = '(unknown)'
  85         version = None
  86         arch = None
  87         category = None
  88         homepage = None
  89         buildtime = None
  90         license = None
  91
  92         def __init__(self):
  93                 self.requires = []
  94
  95 class DebRepo:
  96         def __init__(self, options):
  97                 self.packages_base_url = (options.mirror or 'http://ftp.debian.org/debian') + '/'
  98                 self.packages_file = options.packages_file or 'Packages'
  99
 100         def get_repo_metadata(self, pkg_name):
 101                 if not os.path.isfile(self.packages_file):
 102                         print >>sys.stderr, ("File '%s' not found (use -p to give its location).\n"
 103                                 "Either download one (e.g. ftp://ftp.debian.org/debian/dists/stable/main/binary-amd64/Packages.bz2),\n"
 104                                 "or specify the full URL of the .deb package to use.") % self.packages_file
 105                         sys.exit(1)
 106                 if self.packages_file.endswith('.bz2'):
 107                         import bz2
 108                         opener = bz2.BZ2File
 109                 else:
 110                         opener = file
 111                 pkg_data = "\n" + opener(self.packages_file).read()
 112                 try:
 113                         i = pkg_data.index('\nPackage: %s\n' % pkg_name)
 114                 except ValueError:
 115                         raise Exception("Package '%s' not found in Packages file '%s'." % (pkg_name, self.packages_file))
 116                 j = pkg_data.find('\n\n', i)
 117                 if j == -1:
 118                         pkg_info = pkg_data[i:]
 119                 else:
 120                         pkg_info = pkg_data[i:j]
 121                 filename = None
 122                 digest = {}
 123                 for line in pkg_info.split('\n'):
 124                         if ':' in line and not line.startswith(' '):
 125                                 key, value = line.split(':', 1)
 126                                 if key == 'Filename':
 127                                         filename = value.strip()
 128                                 elif key in ('SHA1', 'SHA256'):
 129                                         digest[key.lower()] = value.strip()
 130                 if filename is None:
 131                         raise Exception('Filename: field not found in package data:\n' + pkg_info)
 132                 pkg_url = self.packages_base_url + filename
 133
 134                 return pkg_url, digest
 135
 136         def get_package_metadata(self, pkg_file):
 137                 package = Package()
 138
 139                 details = read_child(['dpkg-deb', '--info', pkg_file])
 140
 141                 description_and_summary = details.split('\n Description: ')[1].split('\n')
 142                 package.summary = description_and_summary[0]
 143                 description = ''
 144                 for x in description_and_summary[1:]:
 145                         if not x: continue
 146                         assert x[0] == ' '
 147                         x = x[1:]
 148                         if x[0] != ' ':
 149                                 break
 150                         if x == ' .':
 151                                 description += '\n'
 152                         else:
 153                                 description += x[1:].replace('.  ', '. ') + '\n'
 154                 package.description = description.strip()
 155
 156                 for line in details.split('\n'):
 157                         if not line: continue
 158                         assert line.startswith(' ')
 159                         line = line[1:]
 160                         if ':' in line:
 161                                 key, value = line.split(':', 1)
 162                                 value = value.strip()
 163                                 if key == 'Section':
 164                                         package.category = deb_category_to_freedesktop.get(value)
 165                                         if not package.category:
 166                                                 if value != 'libs':
 167                                                         print >>sys.stderr, "Warning: no mapping for Debian category '%s'" % value
 168                                 elif key == 'Package':
 169                                         package.name = value
 170                                 elif key == 'Version':
 171                                         value = value.replace('cvs', '')
 172                                         value = value.replace('svn', '')
 173                                         package.version = distro.try_cleanup_distro_version(value)
 174                                 elif key == 'Architecture':
 175                                         if '-' in value:
 176                                                 arch, value = value.split('-', 1)
 177                                         else:
 178                                                 arch = 'linux'
 179                                         if value == 'amd64':
 180                                                 value = 'x86_64'
 181                                         elif value == 'all':
 182                                                 value = '*'
 183                                         package.arch = arch.capitalize() + '-' + value
 184                                 elif key == 'Depends':
 185                                         for x in value.split(','):
 186                                                 req = mappings.process(x)
 187                                                 if req:
 188                                                         package.requires.append(req)
 189                 return package
 190
 191         def unpack(self, pkg_file, tmp):
 192                 return read_child(['dpkg-deb', '-X', pkg_file, tmp])
 193
 194 class RPMRepo:
 195         def __init__(self, options):
 196                 self.packages_base_url = (options.mirror or 'http://mirror.centos.org/centos') + '/'
 197                 self.packages_base_dir = (options.path or '5/os/i386') + '/'
 198                 self.repomd_file = options.repomd_file or 'repodata/repomd.xml'
 199                 if not os.path.isfile(self.repomd_file):
 200                         print >>sys.stderr, ("File '%s' not found (use -r to give its location).\n"
 201                                 "Either download one (e.g. http://mirror.centos.org/centos/5/os/i386/repodata/repomd.xml),\n"
 202                                 "or specify the full URL of the .rpm package to use.") % self.repomd_file
 203                         sys.exit(1)
 204
 205         def get_repo_metadata(self, pkg_name):
 206                 primary_file = None
 207                 repomd = minidom.parse(self.repomd_file)
 208                 for data in repomd.getElementsByTagName("data"):
 209                         if data.attributes["type"].nodeValue == "primary":
 210                                 for node in data.getElementsByTagName("location"):
 211                                         primary_file = node.attributes["href"].nodeValue
 212                 location = None
 213                 primary = ET.parse(gzip.open(primary_file))
 214                 NS = "http://linux.duke.edu/metadata/common"
 215                 metadata = primary.getroot()
 216                 pkg_data = None
 217                 for package in metadata.findall("{%s}package" % NS):
 218                         if package.find("{%s}name" % NS).text == pkg_name:
 219                                 pkg_data = package
 220                                 location = pkg_data.find("{%s}location" % NS).get("href")
 221                                 break
 222                 if pkg_data is None:
 223                         raise Exception("Package '%s' not found in repodata." % pkg_name)
 224                 checksum = pkg_data.find("{%s}checksum" % NS)
 225                 digest = {}
 226                 if checksum.get("type") == "sha":
 227                         digest["sha1"] = checksum.text
 228                 if checksum.get("type") == "sha256":
 229                         digest["sha256"] = checksum.text
 230                 if location is None:
 231                         raise Exception('location tag not found in primary metadata:\n' + primary_file)
 232                 pkg_url = self.packages_base_url + self.packages_base_dir + location
 233
 234                 return pkg_url, digest
 235
 236         def get_package_metadata(self, pkg_file):
 237                 package = Package()
 238
 239                 query_format = '%{SUMMARY}\\a%{DESCRIPTION}\\a%{NAME}\\a%{VERSION}\\a%{OS}\\a%{ARCH}\\a%{URL}\\a%{GROUP}\\a%{LICENSE}\\a%{BUILDTIME}\\a[%{REQUIRES}\\n]'
 240                 headers = read_child(['rpm', '--qf', query_format, '-qp', pkg_file]).split('\a')
 241
 242                 package.summary = headers[0].strip()
 243                 package.description = headers[1].strip()
 244
 245                 package.name = headers[2]
 246                 value = headers[3]
 247                 value = value.replace('cvs', '')
 248                 value = value.replace('svn', '')
 249                 value = distro.try_cleanup_distro_version(value)
 250                 package.version = value
 251                 value = headers[4]
 252                 package.arch = value.capitalize()
 253                 value = headers[5]
 254                 if value == 'amd64':
 255                         value = 'x86_64'
 256                 if value == 'noarch':
 257                         value = '*'
 258                 package.arch += '-' + value
 259                 value = headers[6].strip()
 260                 package.page = value
 261                 category = None
 262                 value = headers[7].strip()
 263                 package.category = rpm_group_to_freedesktop.get(value)
 264                 if not category:
 265                         print >>sys.stderr, "Warning: no mapping for RPM group '%s'" % value
 266
 267                 value = headers[8].strip()
 268                 package.license = value
 269                 value = headers[9].strip()
 270                 package.buildtime = long(value)
 271                 value = headers[10].strip()
 272                 for x in value.split('\n'):
 273                         if x.startswith('rpmlib'):
 274                                 continue
 275                         req = mappings.process(x)
 276                         if req:
 277                                 package.requires.append(req)
 278                 return package
 279
 280         def unpack(self, pkg_file, tmp):
 281                 files = read_child(['rpm', '-qlp', pkg_file])
 282                 p1 = subprocess.Popen(['rpm2cpio', pkg_file], stdout = subprocess.PIPE)
 283                 IGNORE = open('/dev/null', 'w')
 284                 p2 = subprocess.Popen(['cpio', '-dim'], stdin = p1.stdout, stderr = IGNORE, cwd = tmp)
 285                 p2.communicate()
 286                 return files
 287
 288
 289 if args[0].endswith('.deb') or options.packages_file:
 290         repo = DebRepo(options)
 291 elif args[0].endswith('.rpm') or options.repomd_file:
 292         repo = RPMRepo(options)
 293 else:
 294         print >>sys.stderr, "Use --packages-file for Debian, or --repomd-file for RPM"
 295         sys.exit(1)
 296
 297 pkg_data = None
 298
 299 scheme = args[0].split(':', 1)[0]
 300 if scheme in ('http', 'https', 'ftp'):
 301         pkg_url = args[0]
 302         digest = {}
 303 else:
 304         pkg_url, digest = repo.get_repo_metadata(args[0])
 305
 306 # Download package, if required
 307
 308 pkg_file = os.path.abspath(pkg_url.rsplit('/', 1)[1])
 309 if not os.path.exists(pkg_file):
 310         print >>sys.stderr, "File '%s' not found, so downloading from %s..." % (pkg_file, pkg_url)
 311         check_call(['wget', pkg_url])
 312
 313 # Check digest, if known
 314
 315 if "sha256" in digest:
 316         import hashlib
 317         m = hashlib.new('sha256')
 318         expected_digest = digest["sha256"]
 319 elif "sha1" in digest:
 320         try:
 321                 import hashlib
 322                 m = hashlib.new('sha1')
 323         except ImportError:
 324                 import sha
 325                 m = sha.new()
 326         expected_digest = digest["sha1"]
 327 else:
 328         m = None
 329
 330 if m:
 331         m.update(file(pkg_file).read())
 332         actual = m.hexdigest()
 333         if actual != expected_digest:
 334                 raise Exception("Incorrect digest on package file! Was " + actual + ", but expected " + expected_digest)
 335         else:
 336                 print "Package's digest matches value in reposistory metadata (" + actual + "). Good."
 337 else:
 338         print >>sys.stderr, "Note: no SHA-1 or SHA-256 digest known for this package, so not checking..."
 339
 340 # Extract meta-data from package
 341
 342 pkg_metadata = repo.get_package_metadata(pkg_file)
 343
 344 # Unpack package, find binaries and .desktop files, and add to cache
 345
 346 possible_mains = []
 347 icondata = None
 348 tmp = tempfile.mkdtemp(prefix = 'pkg2zero-')
 349 try:
 350         files = repo.unpack(pkg_file, tmp)
 351
 352         icon = None
 353         images = {}
 354         for f in files.split('\n'):
 355                 if f.startswith('./'):
 356                         f = f[2:]
 357                 elif f.startswith('/'):
 358                         f = f[1:]
 359                 assert not f.startswith('/')
 360                 full = os.path.join(tmp, f)
 361                 if f.endswith('.desktop'):
 362                         for line in file(full):
 363                                 if line.startswith('Categories'):
 364                                         for cat in line.split('=', 1)[1].split(';'):
 365                                                 cat = cat.strip()
 366                                                 if cat in valid_categories:
 367                                                         category = cat
 368                                                         break
 369                                 elif line.startswith('Icon'):
 370                                         icon = line.split('=', 1)[1].strip()
 371                 elif f.startswith('bin/') or f.startswith('usr/bin/') or f.startswith('usr/games/'):
 372                         if os.path.isfile(full):
 373                                 possible_mains.append(f)
 374                 elif f.endswith('.png'):
 375                         images[f] = full
 376                         images[os.path.basename(f)] = full
 377                         # make sure to also map basename without the extension
 378                         images[os.path.splitext(os.path.basename(f))[0]] = full
 379
 380         icondata = None
 381         if icon in images:
 382                 print "Using %s for icon" % os.path.basename(images[icon])
 383                 icondata = file(images[icon]).read()
 384
 385         manifest = read_child(['0store', 'manifest', tmp, manifest_algorithm])
 386         digest = manifest.rsplit('\n', 2)[1]
 387         check_call(['0store', 'add', digest, tmp])
 388 finally:
 389         shutil.rmtree(tmp)
 390
 391 if possible_mains:
 392         possible_mains = sorted(possible_mains, key = len)
 393         pkg_main = possible_mains[0]
 394         if len(possible_mains) > 1:
 395                 print "Warning: several possible main binaries found:"
 396                 print "- " + pkg_main + " (I chose this one)"
 397                 for x in possible_mains[1:]:
 398                         print "- " + x
 399 else:
 400         pkg_main = None
 401
 402 # Make sure we haven't added this version already...
 403
 404 if len(args) > 1:
 405         target_feed_file = args[1]
 406         target_icon_file = args[1].replace('.xml', '.png')
 407 else:
 408         target_feed_file = pkg_metadata.name + '.xml'
 409         target_icon_file = pkg_metadata.name + '.png'
 410
 411 feed_uri = None
 412 icon_uri = None
 413 if os.path.isfile(target_feed_file):
 414         dummy_dist = distro.Distribution()
 415         dom = qdom.parse(file(target_feed_file))
 416         old_target_feed = model.ZeroInstallFeed(dom, local_path = target_feed_file, distro = dummy_dist)
 417         existing_impl = old_target_feed.implementations.get(digest)
 418         if existing_impl:
 419                 print >>sys.stderr, ("Feed '%s' already contains an implementation with this digest!\n%s" % (target_feed_file, existing_impl))
 420                 sys.exit(1)
 421 else:
 422         # No target, so need to pick a URI
 423         feed_uri = mappings.lookup(pkg_metadata.name)
 424         if feed_uri is None:
 425                 suggestion = mappings.get_suggestion(pkg_metadata.name)
 426                 uri = raw_input('Enter the URI for this feed [%s]: ' % suggestion).strip()
 427                 if not uri:
 428                         uri = suggestion
 429                 assert uri.startswith('http://') or uri.startswith('https://') or uri.startswith('ftp://'), uri
 430                 feed_uri = uri
 431                 mappings.add_mapping(pkg_metadata.name, uri)
 432
 433         if icondata and not os.path.isfile(target_icon_file):
 434                 file = open(target_icon_file, 'wb')
 435                 file.write(icondata)
 436                 file.close()
 437         if icon_uri is None:
 438                 suggestion = 'http://0install.net/feed_icons/' + target_icon_file
 439                 uri = raw_input('Enter the URI for this icon [%s]: ' % suggestion).strip()
 440                 if not uri:
 441                         uri = suggestion
 442                 assert uri.startswith('http://') or uri.startswith('https://') or uri.startswith('ftp://'), uri
 443                 icon_uri = uri
 444
 445 # Create a local feed with just the new version...
 446
 447 template = '''<interface xmlns="http://zero-install.sourceforge.net/2004/injector/interface">
 448 </interface>'''
 449 doc = minidom.parseString(template)
 450 root = doc.documentElement
 451
 452 add_node(root, 'name', pkg_metadata.name)
 453 add_node(root, 'summary', pkg_metadata.summary)
 454 add_node(root, 'description', pkg_metadata.description)
 455 feed_for = add_node(root, 'feed-for', '')
 456 if feed_uri:
 457         feed_for.setAttribute('interface', feed_uri)
 458 if icon_uri:
 459         icon = add_node(root, 'icon')
 460         icon.setAttribute('href', icon_uri)
 461         icon.setAttribute('type', 'image/png')
 462 if pkg_metadata.homepage:
 463         add_node(root, 'homepage', pkg_metadata.homepage)
 464 if pkg_metadata.category:
 465         add_node(root, 'category', pkg_metadata.category)
 466
 467 package = add_node(root, 'package-implementation', '')
 468 package.setAttribute('package', pkg_metadata.name)
 469
 470 group = add_node(root, 'group', '')
 471 if pkg_metadata.arch:
 472         group.setAttribute('arch', pkg_metadata.arch)
 473 else:
 474         print >>sys.stderr, "No Architecture: field in package"
 475 if pkg_metadata.license:
 476         group.setAttribute('license', pkg_metadata.license)
 477
 478 for req in pkg_metadata.requires:
 479         req_element = add_node(group, 'requires', before = '\n    ', after = '')
 480         req_element.setAttribute('interface', req.interface)
 481         binding = add_node(req_element, 'environment', before = '\n      ', after = '\n    ')
 482         binding.setAttribute('name', 'LD_LIBRARY_PATH')
 483         binding.setAttribute('insert', 'usr/lib')
 484
 485 if pkg_main:
 486         group.setAttribute('main', pkg_main)
 487         package.setAttribute('main', '/' + pkg_main)
 488
 489 impl = add_node(group, 'implementation', before = '\n    ', after = '\n  ')
 490 impl.setAttribute('id', digest)
 491 assert pkg_metadata.version
 492 impl.setAttribute('version', pkg_metadata.version)
 493
 494 if pkg_metadata.buildtime:
 495         impl.setAttribute('released', time.strftime('%Y-%m-%d', time.localtime(pkg_metadata.buildtime)))
 496 else:
 497         impl.setAttribute('released', time.strftime('%Y-%m-%d'))
 498
 499 archive = add_node(impl, 'archive', before = '\n      ', after = '\n    ')
 500 archive.setAttribute('href', pkg_url)
 501 archive.setAttribute('size', str(os.path.getsize(pkg_file)))
 502
 503 # Add our new version to the main feed...
 504
 505 output_stream = tempfile.NamedTemporaryFile(prefix = 'pkg2zero-')
 506 try:
 507         output_stream.write("<?xml version='1.0'?>\n")
 508         root.writexml(output_stream)
 509         output_stream.write('\n')
 510         output_stream.flush()
 511
 512         publishing_options = []
 513         if options.key:
 514                 # Note: 0publish < 0.16 requires the --xmlsign option too
 515                 publishing_options += ['--xmlsign', '--key', options.key]
 516         check_call([os.environ['PUBLISH_COMMAND']] + publishing_options + ['--local', output_stream.name, target_feed_file])
 517         print "Added version %s to %s" % (pkg_metadata.version, target_feed_file)
 518 finally:
 519         output_stream.close()