dak/generate_index_diffs.py

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