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