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                 repo_top = os.path.dirname(os.path.dirname(self.repomd_file))
 209                 for data in repomd.getElementsByTagName("data"):
 210                         if data.attributes["type"].nodeValue == "primary":
 211                                 for node in data.getElementsByTagName("location"):
 212                                         primary_file = os.path.join(repo_top, node.attributes["href"].nodeValue)
 213                 location = None
 214                 primary = ET.parse(gzip.open(primary_file))
 215                 NS = "http://linux.duke.edu/metadata/common"
 216                 metadata = primary.getroot()
 217                 pkg_data = None
 218                 for package in metadata.findall("{%s}package" % NS):
 219                         if package.find("{%s}name" % NS).text == pkg_name:
 220                                 pkg_data = package
 221                                 location = pkg_data.find("{%s}location" % NS).get("href")
 222                                 break
 223                 if pkg_data is None:
 224                         raise Exception("Package '%s' not found in repodata." % pkg_name)
 225                 checksum = pkg_data.find("{%s}checksum" % NS)
 226                 digest = {}
 227                 if checksum.get("type") == "sha":
 228                         digest["sha1"] = checksum.text
 229                 if checksum.get("type") == "sha256":
 230                         digest["sha256"] = checksum.text
 231                 if location is None:
 232                         raise Exception('location tag not found in primary metadata:\n' + primary_file)
 233                 pkg_url = self.packages_base_url + self.packages_base_dir + location
 234
 235                 return pkg_url, digest
 236
 237         def get_package_metadata(self, pkg_file):
 238                 package = Package()
 239
 240                 query_format = '%{SUMMARY}\\a%{DESCRIPTION}\\a%{NAME}\\a%{VERSION}\\a%{OS}\\a%{ARCH}\\a%{URL}\\a%{GROUP}\\a%{LICENSE}\\a%{BUILDTIME}\\a[%{REQUIRES}\\n]'
 241                 headers = read_child(['rpm', '--qf', query_format, '-qp', pkg_file]).split('\a')
 242
 243                 package.summary = headers[0].strip()
 244                 package.description = headers[1].strip()
 245
 246                 package.name = headers[2]
 247                 value = headers[3]
 248                 value = value.replace('cvs', '')
 249                 value = value.replace('svn', '')
 250                 value = distro.try_cleanup_distro_version(value)
 251                 package.version = value
 252                 value = headers[4]
 253                 package.arch = value.capitalize()
 254                 value = headers[5]
 255                 if value == 'amd64':
 256                         value = 'x86_64'
 257                 if value == 'noarch':
 258                         value = '*'
 259                 package.arch += '-' + value
 260                 value = headers[6].strip()
 261                 package.page = value
 262                 category = None
 263                 value = headers[7].strip()
 264                 package.category = rpm_group_to_freedesktop.get(value)
 265                 if not category:
 266                         print >>sys.stderr, "Warning: no mapping for RPM group '%s'" % value
 267
 268                 value = headers[8].strip()
 269                 package.license = value
 270                 value = headers[9].strip()
 271                 package.buildtime = long(value)
 272                 value = headers[10].strip()
 273                 for x in value.split('\n'):
 274                         if x.startswith('rpmlib'):
 275                                 continue
 276                         req = mappings.process(x)
 277                         if req:
 278                                 package.requires.append(req)
 279                 return package
 280
 281         def unpack(self, pkg_file, tmp):
 282                 files = read_child(['rpm', '-qlp', pkg_file])
 283                 p1 = subprocess.Popen(['rpm2cpio', pkg_file], stdout = subprocess.PIPE)
 284                 IGNORE = open('/dev/null', 'w')
 285                 p2 = subprocess.Popen(['cpio', '-dim'], stdin = p1.stdout, stderr = IGNORE, cwd = tmp)
 286                 p2.communicate()
 287                 return files
 288
 289
 290 if args[0].endswith('.deb') or options.packages_file:
 291         repo = DebRepo(options)
 292 elif args[0].endswith('.rpm') or options.repomd_file:
 293         repo = RPMRepo(options)
 294 else:
 295         print >>sys.stderr, "Use --packages-file for Debian, or --repomd-file for RPM"
 296         sys.exit(1)
 297
 298 pkg_data = None
 299
 300 scheme = args[0].split(':', 1)[0]
 301 if scheme in ('http', 'https', 'ftp'):
 302         pkg_url = args[0]
 303         digest = {}
 304 else:
 305         pkg_url, digest = repo.get_repo_metadata(args[0])
 306
 307 # Download package, if required
 308
 309 pkg_file = os.path.abspath(pkg_url.rsplit('/', 1)[1])
 310 if not os.path.exists(pkg_file):
 311         print >>sys.stderr, "File '%s' not found, so downloading from %s..." % (pkg_file, pkg_url)
 312         check_call(['wget', pkg_url])
 313
 314 # Check digest, if known
 315
 316 if "sha256" in digest:
 317         import hashlib
 318         m = hashlib.new('sha256')
 319         expected_digest = digest["sha256"]
 320 elif "sha1" in digest:
 321         try:
 322                 import hashlib
 323                 m = hashlib.new('sha1')
 324         except ImportError:
 325                 import sha
 326                 m = sha.new()
 327         expected_digest = digest["sha1"]
 328 else:
 329         m = None
 330
 331 if m:
 332         m.update(file(pkg_file).read())
 333         actual = m.hexdigest()
 334         if actual != expected_digest:
 335                 raise Exception("Incorrect digest on package file! Was " + actual + ", but expected " + expected_digest)
 336         else:
 337                 print "Package's digest matches value in reposistory metadata (" + actual + "). Good."
 338 else:
 339         print >>sys.stderr, "Note: no SHA-1 or SHA-256 digest known for this package, so not checking..."
 340
 341 # Extract meta-data from package
 342
 343 pkg_metadata = repo.get_package_metadata(pkg_file)
 344
 345 # Unpack package, find binaries and .desktop files, and add to cache
 346
 347 possible_mains = []
 348 icondata = None
 349 tmp = tempfile.mkdtemp(prefix = 'pkg2zero-')
 350 try:
 351         files = repo.unpack(pkg_file, tmp)
 352
 353         icon = None
 354         images = {}
 355         for f in files.split('\n'):
 356                 if f.startswith('./'):
 357                         f = f[2:]
 358                 elif f.startswith('/'):
 359                         f = f[1:]
 360                 assert not f.startswith('/')
 361                 full = os.path.join(tmp, f)
 362                 if f.endswith('.desktop'):
 363                         for line in file(full):
 364                                 if line.startswith('Categories'):
 365                                         for cat in line.split('=', 1)[1].split(';'):
 366                                                 cat = cat.strip()
 367                                                 if cat in valid_categories:
 368                                                         category = cat
 369                                                         break
 370                                 elif line.startswith('Icon'):
 371                                         icon = line.split('=', 1)[1].strip()
 372                 elif f.startswith('bin/') or f.startswith('usr/bin/') or f.startswith('usr/games/'):
 373                         if os.path.isfile(full):
 374                                 possible_mains.append(f)
 375                 elif f.endswith('.png'):
 376                         images[f] = full
 377                         images[os.path.basename(f)] = full
 378                         # make sure to also map basename without the extension
 379                         images[os.path.splitext(os.path.basename(f))[0]] = full
 380
 381         icondata = None
 382         if icon in images:
 383                 print "Using %s for icon" % os.path.basename(images[icon])
 384                 icondata = file(images[icon]).read()
 385
 386         manifest = read_child(['0store', 'manifest', tmp, manifest_algorithm])
 387         digest = manifest.rsplit('\n', 2)[1]
 388         check_call(['0store', 'add', digest, tmp])
 389 finally:
 390         shutil.rmtree(tmp)
 391
 392 if possible_mains:
 393         possible_mains = sorted(possible_mains, key = len)
 394         pkg_main = possible_mains[0]
 395         if len(possible_mains) > 1:
 396                 print "Warning: several possible main binaries found:"
 397                 print "- " + pkg_main + " (I chose this one)"
 398                 for x in possible_mains[1:]:
 399                         print "- " + x
 400 else:
 401         pkg_main = None
 402
 403 # Make sure we haven't added this version already...
 404
 405 if len(args) > 1:
 406         target_feed_file = args[1]
 407         target_icon_file = args[1].replace('.xml', '.png')
 408 else:
 409         target_feed_file = pkg_metadata.name + '.xml'
 410         target_icon_file = pkg_metadata.name + '.png'
 411
 412 feed_uri = None
 413 icon_uri = None
 414 if os.path.isfile(target_feed_file):
 415         dummy_dist = distro.Distribution()
 416         dom = qdom.parse(file(target_feed_file))
 417         old_target_feed = model.ZeroInstallFeed(dom, local_path = target_feed_file, distro = dummy_dist)
 418         existing_impl = old_target_feed.implementations.get(digest)
 419         if existing_impl:
 420                 print >>sys.stderr, ("Feed '%s' already contains an implementation with this digest!\n%s" % (target_feed_file, existing_impl))
 421                 sys.exit(1)
 422 else:
 423         # No target, so need to pick a URI
 424         feed_uri = mappings.lookup(pkg_metadata.name)
 425         if feed_uri is None:
 426                 suggestion = mappings.get_suggestion(pkg_metadata.name)
 427                 uri = raw_input('Enter the URI for this feed [%s]: ' % suggestion).strip()
 428                 if not uri:
 429                         uri = suggestion
 430                 assert uri.startswith('http://') or uri.startswith('https://') or uri.startswith('ftp://'), uri
 431                 feed_uri = uri
 432                 mappings.add_mapping(pkg_metadata.name, uri)
 433
 434         if icondata and not os.path.isfile(target_icon_file):
 435                 file = open(target_icon_file, 'wb')
 436                 file.write(icondata)
 437                 file.close()
 438         if icon_uri is None:
 439                 suggestion = 'http://0install.net/feed_icons/' + target_icon_file
 440                 uri = raw_input('Enter the URI for this icon [%s]: ' % suggestion).strip()
 441                 if not uri:
 442                         uri = suggestion
 443                 assert uri.startswith('http://') or uri.startswith('https://') or uri.startswith('ftp://'), uri
 444                 icon_uri = uri
 445
 446 # Create a local feed with just the new version...
 447
 448 template = '''<interface xmlns="http://zero-install.sourceforge.net/2004/injector/interface">
 449 </interface>'''
 450 doc = minidom.parseString(template)
 451 root = doc.documentElement
 452
 453 add_node(root, 'name', pkg_metadata.name)
 454 add_node(root, 'summary', pkg_metadata.summary)
 455 add_node(root, 'description', pkg_metadata.description)
 456 feed_for = add_node(root, 'feed-for', '')
 457 if feed_uri:
 458         feed_for.setAttribute('interface', feed_uri)
 459 if icon_uri:
 460         icon = add_node(root, 'icon')
 461         icon.setAttribute('href', icon_uri)
 462         icon.setAttribute('type', 'image/png')
 463 if pkg_metadata.homepage:
 464         add_node(root, 'homepage', pkg_metadata.homepage)
 465 if pkg_metadata.category:
 466         add_node(root, 'category', pkg_metadata.category)
 467
 468 package = add_node(root, 'package-implementation', '')
 469 package.setAttribute('package', pkg_metadata.name)
 470
 471 group = add_node(root, 'group', '')
 472 if pkg_metadata.arch:
 473         group.setAttribute('arch', pkg_metadata.arch)
 474 else:
 475         print >>sys.stderr, "No Architecture: field in package"
 476 if pkg_metadata.license:
 477         group.setAttribute('license', pkg_metadata.license)
 478
 479 for req in pkg_metadata.requires:
 480         req_element = add_node(group, 'requires', before = '\n    ', after = '')
 481         req_element.setAttribute('interface', req.interface)
 482         binding = add_node(req_element, 'environment', before = '\n      ', after = '\n    ')
 483         binding.setAttribute('name', 'LD_LIBRARY_PATH')
 484         binding.setAttribute('insert', 'usr/lib')
 485
 486 if pkg_main:
 487         group.setAttribute('main', pkg_main)
 488         package.setAttribute('main', '/' + pkg_main)
 489
 490 impl = add_node(group, 'implementation', before = '\n    ', after = '\n  ')
 491 impl.setAttribute('id', digest)
 492 assert pkg_metadata.version
 493 impl.setAttribute('version', pkg_metadata.version)
 494
 495 if pkg_metadata.buildtime:
 496         impl.setAttribute('released', time.strftime('%Y-%m-%d', time.localtime(pkg_metadata.buildtime)))
 497 else:
 498         impl.setAttribute('released', time.strftime('%Y-%m-%d'))
 499
 500 archive = add_node(impl, 'archive', before = '\n      ', after = '\n    ')
 501 archive.setAttribute('href', pkg_url)
 502 archive.setAttribute('size', str(os.path.getsize(pkg_file)))
 503
 504 # Add our new version to the main feed...
 505
 506 output_stream = tempfile.NamedTemporaryFile(prefix = 'pkg2zero-')
 507 try:
 508         output_stream.write("<?xml version='1.0'?>\n")
 509         root.writexml(output_stream)
 510         output_stream.write('\n')
 511         output_stream.flush()
 512
 513         publishing_options = []
 514         if options.key:
 515                 # Note: 0publish < 0.16 requires the --xmlsign option too
 516                 publishing_options += ['--xmlsign', '--key', options.key]
 517         check_call([os.environ['PUBLISH_COMMAND']] + publishing_options + ['--local', output_stream.name, target_feed_file])
 518         print "Added version %s to %s" % (pkg_metadata.version, target_feed_file)
 519 finally:
 520         output_stream.close()