dak/generate_index_diffs.py

   1 #!/usr/bin/env python
   2
   3 """ generates partial package updates list"""
   4
   5 ###########################################################
   6
   7 # idea and basic implementation by Anthony, some changes by Andreas
   8 # parts are stolen from 'dak generate-releases'
   9 #
  10 # Copyright (C) 2004, 2005, 2006  Anthony Towns <aj@azure.humbug.org.au>
  11 # Copyright (C) 2004, 2005  Andreas Barth <aba@not.so.argh.org>
  12
  13 # This program is free software; you can redistribute it and/or modify
  14 # it under the terms of the GNU General Public License as published by
  15 # the Free Software Foundation; either version 2 of the License, or
  16 # (at your option) any later version.
  17
  18 # This program is distributed in the hope that it will be useful,
  19 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  20 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21 # GNU General Public License for more details.
  22
  23 # You should have received a copy of the GNU General Public License
  24 # along with this program; if not, write to the Free Software
  25 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  26
  27
  28 # < elmo> bah, don't bother me with annoying facts
  29 # < elmo> I was on a roll
  30
  31
  32 ################################################################################
  33
  34 import sys
  35 import os
  36 import tempfile
  37 import subprocess
  38 import time
  39 import apt_pkg
  40
  41 from daklib import utils
  42 from daklib.dbconn import get_suite, get_suite_architectures
  43
  44 ################################################################################
  45
  46 Cnf = None
  47 Logger = None
  48 Options = None
  49
  50 ################################################################################
  51
  52 def usage (exit_code=0):
  53     print """Usage: dak generate-index-diffs [OPTIONS] [suites]
  54 Write out ed-style diffs to Packages/Source lists
  55
  56   -h, --help            show this help and exit
  57   -c                    give the canonical path of the file
  58   -p                    name for the patch (defaults to current time)
  59   -n                    take no action
  60     """
  61     sys.exit(exit_code)
  62
  63
  64 def tryunlink(file):
  65     try:
  66         os.unlink(file)
  67     except OSError:
  68         print "warning: removing of %s denied" % (file)
  69
  70 def smartstat(file):
  71     for ext in ["", ".gz", ".bz2"]:
  72         if os.path.isfile(file + ext):
  73             return (ext, os.stat(file + ext))
  74     return (None, None)
  75
  76 def smartlink(f, t):
  77     if os.path.isfile(f):
  78         os.link(f,t)
  79     elif os.path.isfile("%s.gz" % (f)):
  80         os.system("gzip -d < %s.gz > %s" % (f, t))
  81     elif os.path.isfile("%s.bz2" % (f)):
  82         os.system("bzip2 -d < %s.bz2 > %s" % (f, t))
  83     else:
  84         print "missing: %s" % (f)
  85         raise IOError, f
  86
  87 def smartopen(file):
  88     if os.path.isfile(file):
  89         f = open(file, "r")
  90     elif os.path.isfile("%s.gz" % file):
  91         f = create_temp_file(os.popen("zcat %s.gz" % file, "r"))
  92     elif os.path.isfile("%s.bz2" % file):
  93         f = create_temp_file(os.popen("bzcat %s.bz2" % file, "r"))
  94     else:
  95         f = None
  96     return f
  97
  98 def pipe_file(f, t):
  99     f.seek(0)
 100     while 1:
 101         l = f.read()
 102         if not l: break
 103         t.write(l)
 104     t.close()
 105
 106 class Updates:
 107     def __init__(self, readpath = None, max = 14):
 108         self.can_path = None
 109         self.history = {}
 110         self.history_order = []
 111         self.max = max
 112         self.readpath = readpath
 113         self.filesizesha1 = None
 114
 115         if readpath:
 116             try:
 117                 f = open(readpath + "/Index")
 118                 x = f.readline()
 119
 120                 def read_hashs(ind, f, self, x=x):
 121                     while 1:
 122                         x = f.readline()
 123                         if not x or x[0] != " ": break
 124                         l = x.split()
 125                         if not self.history.has_key(l[2]):
 126                             self.history[l[2]] = [None,None]
 127                             self.history_order.append(l[2])
 128                         self.history[l[2]][ind] = (l[0], int(l[1]))
 129                     return x
 130
 131                 while x:
 132                     l = x.split()
 133
 134                     if len(l) == 0:
 135                         x = f.readline()
 136                         continue
 137
 138                     if l[0] == "SHA1-History:":
 139                         x = read_hashs(0,f,self)
 140                         continue
 141
 142                     if l[0] == "SHA1-Patches:":
 143                         x = read_hashs(1,f,self)
 144                         continue
 145
 146                     if l[0] == "Canonical-Name:" or l[0]=="Canonical-Path:":
 147                         self.can_path = l[1]
 148
 149                     if l[0] == "SHA1-Current:" and len(l) == 3:
 150                         self.filesizesha1 = (l[1], int(l[2]))
 151
 152                     x = f.readline()
 153
 154             except IOError:
 155                 0
 156
 157     def dump(self, out=sys.stdout):
 158         if self.can_path:
 159             out.write("Canonical-Path: %s\n" % (self.can_path))
 160
 161         if self.filesizesha1:
 162             out.write("SHA1-Current: %s %7d\n" % (self.filesizesha1))
 163
 164         hs = self.history
 165         l = self.history_order[:]
 166
 167         cnt = len(l)
 168         if cnt > self.max:
 169             for h in l[:cnt-self.max]:
 170                 tryunlink("%s/%s.gz" % (self.readpath, h))
 171                 del hs[h]
 172             l = l[cnt-self.max:]
 173             self.history_order = l[:]
 174
 175         out.write("SHA1-History:\n")
 176         for h in l:
 177             out.write(" %s %7d %s\n" % (hs[h][0][0], hs[h][0][1], h))
 178         out.write("SHA1-Patches:\n")
 179         for h in l:
 180             out.write(" %s %7d %s\n" % (hs[h][1][0], hs[h][1][1], h))
 181
 182 def create_temp_file(r):
 183     f = tempfile.TemporaryFile()
 184     while 1:
 185         x = r.readline()
 186         if not x: break
 187         f.write(x)
 188     r.close()
 189     del x,r
 190     f.flush()
 191     f.seek(0)
 192     return f
 193
 194 def sizesha1(f):
 195     size = os.fstat(f.fileno())[6]
 196     f.seek(0)
 197     sha1sum = apt_pkg.sha1sum(f)
 198     return (sha1sum, size)
 199
 200 def genchanges(Options, outdir, oldfile, origfile, maxdiffs = 14):
 201     if Options.has_key("NoAct"):
 202         return
 203
 204     patchname = Options["PatchName"]
 205
 206     # origfile = /path/to/Packages
 207     # oldfile  = ./Packages
 208     # newfile  = ./Packages.tmp
 209     # difffile = outdir/patchname
 210     # index   => outdir/Index
 211
 212     # (outdir, oldfile, origfile) = argv
 213
 214     newfile = oldfile + ".new"
 215     difffile = "%s/%s" % (outdir, patchname)
 216
 217     upd = Updates(outdir, int(maxdiffs))
 218     (oldext, oldstat) = smartstat(oldfile)
 219     (origext, origstat) = smartstat(origfile)
 220     if not origstat:
 221         print "%s: doesn't exist" % (origfile)
 222         return
 223     if not oldstat:
 224         print "%s: initial run" % (origfile)
 225         os.link(origfile + origext, oldfile + origext)
 226         return
 227
 228     if oldstat[1:3] == origstat[1:3]:
 229         print "%s: hardlink unbroken, assuming unchanged" % (origfile)
 230         return
 231
 232     oldf = smartopen(oldfile)
 233     oldsizesha1 = sizesha1(oldf)
 234
 235     # should probably early exit if either of these checks fail
 236     # alternatively (optionally?) could just trim the patch history
 237
 238     if upd.filesizesha1:
 239         if upd.filesizesha1 != oldsizesha1:
 240             print "info: old file " + oldfile + " changed! %s %s => %s %s" % (upd.filesizesha1 + oldsizesha1)
 241
 242     if Options.has_key("CanonicalPath"): upd.can_path=Options["CanonicalPath"]
 243
 244     if os.path.exists(newfile): os.unlink(newfile)
 245     smartlink(origfile, newfile)
 246     newf = open(newfile, "r")
 247     newsizesha1 = sizesha1(newf)
 248     newf.close()
 249
 250     if newsizesha1 == oldsizesha1:
 251         os.unlink(newfile)
 252         oldf.close()
 253         print "%s: unchanged" % (origfile)
 254     else:
 255         if not os.path.isdir(outdir):
 256             os.mkdir(outdir)
 257
 258         w = os.popen("diff --ed - %s | gzip -c -9 > %s.gz" %
 259                      (newfile, difffile), "w")
 260         pipe_file(oldf, w)
 261         oldf.close()
 262
 263         difff = smartopen(difffile)
 264         difsizesha1 = sizesha1(difff)
 265         difff.close()
 266
 267         upd.history[patchname] = (oldsizesha1, difsizesha1)
 268         upd.history_order.append(patchname)
 269
 270         upd.filesizesha1 = newsizesha1
 271
 272         os.unlink(oldfile + oldext)
 273         os.link(origfile + origext, oldfile + origext)
 274         os.unlink(newfile)
 275
 276         f = open(outdir + "/Index", "w")
 277         upd.dump(f)
 278         f.close()
 279
 280
 281 def main():
 282     global Cnf, Options, Logger
 283
 284     os.umask(0002)
 285
 286     Cnf = utils.get_conf()
 287     Arguments = [ ('h', "help", "Generate-Index-Diffs::Options::Help"),
 288                   ('c', None, "Generate-Index-Diffs::Options::CanonicalPath", "hasArg"),
 289                   ('p', "patchname", "Generate-Index-Diffs::Options::PatchName", "hasArg"),
 290                   ('r', "rootdir", "Generate-Index-Diffs::Options::RootDir", "hasArg"),
 291                   ('d', "tmpdir", "Generate-Index-Diffs::Options::TempDir", "hasArg"),
 292                   ('m', "maxdiffs", "Generate-Index-Diffs::Options::MaxDiffs", "hasArg"),
 293                   ('n', "n-act", "Generate-Index-Diffs::Options::NoAct"),
 294                 ]
 295     suites = apt_pkg.ParseCommandLine(Cnf,Arguments,sys.argv)
 296     Options = Cnf.SubTree("Generate-Index-Diffs::Options")
 297     if Options.has_key("Help"): usage()
 298
 299     maxdiffs = Options.get("MaxDiffs::Default", "14")
 300     maxpackages = Options.get("MaxDiffs::Packages", maxdiffs)
 301     maxcontents = Options.get("MaxDiffs::Contents", maxdiffs)
 302     maxsources = Options.get("MaxDiffs::Sources", maxdiffs)
 303
 304     if not Options.has_key("PatchName"):
 305         format = "%Y-%m-%d-%H%M.%S"
 306         Options["PatchName"] = time.strftime( format )
 307
 308     AptCnf = apt_pkg.newConfiguration()
 309     apt_pkg.ReadConfigFileISC(AptCnf,utils.which_apt_conf_file())
 310
 311     if Options.has_key("RootDir"): Cnf["Dir::Root"] = Options["RootDir"]
 312
 313     if not suites:
 314         suites = Cnf.SubTree("Suite").List()
 315
 316     for suitename in suites:
 317         print "Processing: " + suitename
 318         SuiteBlock = Cnf.SubTree("Suite::" + suitename)
 319
 320         suiteobj = get_suite(suitename.lower())
 321
 322         # Use the canonical version of the suite name
 323         suite = suiteobj.suite_name
 324
 325         if suiteobj.untouchable:
 326             print "Skipping: " + suite + " (untouchable)"
 327             continue
 328
 329         architectures = get_suite_architectures(suite, skipall=True)
 330
 331         if SuiteBlock.has_key("Components"):
 332             components = SuiteBlock.ValueList("Components")
 333         else:
 334             components = []
 335
 336         suite_suffix = Cnf.Find("Dinstall::SuiteSuffix")
 337         if components and suite_suffix:
 338             longsuite = suite + "/" + suite_suffix
 339         else:
 340             longsuite = suite
 341
 342         tree = SuiteBlock.get("Tree", "dists/%s" % (longsuite))
 343
 344         if AptCnf.has_key("tree::%s" % (tree)):
 345             sections = AptCnf["tree::%s::Sections" % (tree)].split()
 346         elif AptCnf.has_key("bindirectory::%s" % (tree)):
 347             sections = AptCnf["bindirectory::%s::Sections" % (tree)].split()
 348         else:
 349             aptcnf_filename = os.path.basename(utils.which_apt_conf_file())
 350             print "ALERT: suite %s not in %s, nor untouchable!" % (suite, aptcnf_filename)
 351             continue
 352
 353         for archobj in architectures:
 354             architecture = archobj.arch_string
 355
 356             if architecture != "source":
 357                 # Process Contents
 358                 file = "%s/Contents-%s" % (Cnf["Dir::Root"] + tree,
 359                         architecture)
 360                 storename = "%s/%s_contents_%s" % (Options["TempDir"], suite, architecture)
 361                 genchanges(Options, file + ".diff", storename, file, \
 362                   Cnf.get("Suite::%s::Generate-Index-Diffs::MaxDiffs::Contents" % (suite), maxcontents))
 363
 364             # use sections instead of components since dak.conf
 365             # treats "foo/bar main" as suite "foo", suitesuffix "bar" and
 366             # component "bar/main". suck.
 367
 368             for component in sections:
 369                 if architecture == "source":
 370                     longarch = architecture
 371                     packages = "Sources"
 372                     maxsuite = maxsources
 373                 else:
 374                     longarch = "binary-%s"% (architecture)
 375                     packages = "Packages"
 376                     maxsuite = maxpackages
 377
 378                 file = "%s/%s/%s/%s" % (Cnf["Dir::Root"] + tree,
 379                            component, longarch, packages)
 380                 storename = "%s/%s_%s_%s" % (Options["TempDir"], suite, component, architecture)
 381                 genchanges(Options, file + ".diff", storename, file, \
 382                   Cnf.get("Suite::%s::Generate-Index-Diffs::MaxDiffs::%s" % (suite, packages), maxsuite))
 383
 384 ################################################################################
 385
 386 if __name__ == '__main__':
 387     main()