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 import glob
  40
  41 from daklib import utils
  42 from daklib.dbconn import Archive, Component, DBConn, Suite, get_suite, get_suite_architectures
  43 #from daklib.regexes import re_includeinpdiff
  44 import re
  45 re_includeinpdiff = re.compile(r"(Translation-[a-zA-Z_]+\.(?:bz2|xz))")
  46
  47 ################################################################################
  48
  49 Cnf = None
  50 Logger = None
  51 Options = None
  52
  53 ################################################################################
  54
  55 def usage (exit_code=0):
  56     print """Usage: dak generate-index-diffs [OPTIONS] [suites]
  57 Write out ed-style diffs to Packages/Source lists
  58
  59   -h, --help            show this help and exit
  60   -a <archive>          generate diffs for suites in <archive>
  61   -c                    give the canonical path of the file
  62   -p                    name for the patch (defaults to current time)
  63   -d                    name for the hardlink farm for status
  64   -m                    how many diffs to generate
  65   -n                    take no action
  66     """
  67     sys.exit(exit_code)
  68
  69 def tryunlink(file):
  70     try:
  71         os.unlink(file)
  72     except OSError:
  73         print "warning: removing of %s denied" % (file)
  74
  75 def smartstat(file):
  76     for ext in ["", ".gz", ".bz2", ".xz"]:
  77         if os.path.isfile(file + ext):
  78             return (ext, os.stat(file + ext))
  79     return (None, None)
  80
  81 def smartlink(f, t):
  82     if os.path.isfile(f):
  83         os.link(f,t)
  84     elif os.path.isfile("%s.gz" % (f)):
  85         os.system("gzip -d < %s.gz > %s" % (f, t))
  86     elif os.path.isfile("%s.bz2" % (f)):
  87         os.system("bzip2 -d < %s.bz2 > %s" % (f, t))
  88     elif os.path.isfile("%s.xz" % (f)):
  89         os.system("xz -d < %s.xz > %s" % (f, t))
  90     else:
  91         print "missing: %s" % (f)
  92         raise IOError(f)
  93
  94 def smartopen(file):
  95     if os.path.isfile(file):
  96         f = open(file, "r")
  97     elif os.path.isfile("%s.gz" % file):
  98         f = create_temp_file(os.popen("zcat %s.gz" % file, "r"))
  99     elif os.path.isfile("%s.bz2" % file):
 100         f = create_temp_file(os.popen("bzcat %s.bz2" % file, "r"))
 101     elif os.path.isfile("%s.xz" % file):
 102         f = create_temp_file(os.popen("xzcat %s.xz" % file, "r"))
 103     else:
 104         f = None
 105     return f
 106
 107 def pipe_file(f, t):
 108     f.seek(0)
 109     while 1:
 110         l = f.read()
 111         if not l: break
 112         t.write(l)
 113     t.close()
 114
 115 class Updates:
 116     def __init__(self, readpath = None, max = 56):
 117         self.can_path = None
 118         self.history = {}
 119         self.history_order = []
 120         self.max = max
 121         self.readpath = readpath
 122         self.filesizehashes = None
 123
 124         if readpath:
 125             try:
 126                 f = open(readpath + "/Index")
 127                 x = f.readline()
 128
 129                 def read_hashs(ind, hashind, f, self, x=x):
 130                     while 1:
 131                         x = f.readline()
 132                         if not x or x[0] != " ": break
 133                         l = x.split()
 134                         fname = l[2]
 135                         if fname.endswith('.gz'):
 136                             fname = fname[:-3]
 137                         if not self.history.has_key(fname):
 138                             self.history[fname] = [None,None,None]
 139                             self.history_order.append(fname)
 140                         if not self.history[fname][ind]:
 141                             self.history[fname][ind] = (int(l[1]), None, None)
 142                         if hashind == 1:
 143                             self.history[fname][ind] = (int(self.history[fname][ind][0]), l[0], self.history[fname][ind][2])
 144                         else:
 145                             self.history[fname][ind] = (int(self.history[fname][ind][0]), self.history[fname][ind][1], l[0])
 146                     return x
 147
 148                 while x:
 149                     l = x.split()
 150
 151                     if len(l) == 0:
 152                         x = f.readline()
 153                         continue
 154
 155                     if l[0] == "SHA1-History:":
 156                         x = read_hashs(0,1,f,self)
 157                         continue
 158
 159                     if l[0] == "SHA256-History:":
 160                         x = read_hashs(0,2,f,self)
 161                         continue
 162
 163                     if l[0] == "SHA1-Patches:":
 164                         x = read_hashs(1,1,f,self)
 165                         continue
 166
 167                     if l[0] == "SHA256-Patches:":
 168                         x = read_hashs(1,2,f,self)
 169                         continue
 170
 171                     if l[0] == "SHA1-Download:":
 172                         x = read_hashs(2,1,f,self)
 173                         continue
 174
 175                     if l[0] == "SHA256-Download:":
 176                         x = read_hashs(2,2,f,self)
 177                         continue
 178
 179                     if l[0] == "Canonical-Name:" or l[0]=="Canonical-Path:":
 180                         self.can_path = l[1]
 181
 182                     if l[0] == "SHA1-Current:" and len(l) == 3:
 183                         if not self.filesizehashes:
 184                             self.filesizehashes = (int(l[2]), None, None)
 185                         self.filesizehashes = (int(self.filesizehashes[0]), l[1], self.filesizehashes[2])
 186
 187                     if l[0] == "SHA256-Current:" and len(l) == 3:
 188                         if not self.filesizehashes:
 189                             self.filesizehashes = (int(l[2]), None, None)
 190                         self.filesizehashes = (int(self.filesizehashes[0]), self.filesizehashes[2], l[1])
 191
 192                     x = f.readline()
 193
 194             except IOError:
 195                 0
 196
 197     def dump(self, out=sys.stdout):
 198         if self.can_path:
 199             out.write("Canonical-Path: %s\n" % (self.can_path))
 200
 201         if self.filesizehashes:
 202             if self.filesizehashes[1]:
 203                 out.write("SHA1-Current: %s %7d\n" % (self.filesizehashes[1], self.filesizehashes[0]))
 204             if self.filesizehashes[2]:
 205                 out.write("SHA256-Current: %s %7d\n" % (self.filesizehashes[2], self.filesizehashes[0]))
 206
 207         hs = self.history
 208         l = self.history_order[:]
 209
 210         cnt = len(l)
 211         if cnt > self.max:
 212             for h in l[:cnt-self.max]:
 213                 tryunlink("%s/%s.gz" % (self.readpath, h))
 214                 del hs[h]
 215             l = l[cnt-self.max:]
 216             self.history_order = l[:]
 217
 218         out.write("SHA1-History:\n")
 219         for h in l:
 220             if hs[h][0] and hs[h][0][1]:
 221                 out.write(" %s %7d %s\n" % (hs[h][0][1], hs[h][0][0], h))
 222         out.write("SHA256-History:\n")
 223         for h in l:
 224             if hs[h][0] and hs[h][0][2]:
 225                 out.write(" %s %7d %s\n" % (hs[h][0][2], hs[h][0][0], h))
 226         out.write("SHA1-Patches:\n")
 227         for h in l:
 228             if hs[h][1] and hs[h][1][1]:
 229                 out.write(" %s %7d %s\n" % (hs[h][1][1], hs[h][1][0], h))
 230         out.write("SHA256-Patches:\n")
 231         for h in l:
 232             if hs[h][1] and hs[h][1][2]:
 233                 out.write(" %s %7d %s\n" % (hs[h][1][2], hs[h][1][0], h))
 234         out.write("SHA1-Download:\n")
 235         for h in l:
 236             if hs[h][2] and hs[h][2][1]:
 237                 out.write(" %s %7d %s.gz\n" % (hs[h][2][1], hs[h][2][0], h))
 238         out.write("SHA256-Download:\n")
 239         for h in l:
 240             if hs[h][2] and hs[h][2][2]:
 241                 out.write(" %s %7d %s.gz\n" % (hs[h][2][2], hs[h][2][0], h))
 242
 243 def create_temp_file(r):
 244     f = tempfile.TemporaryFile()
 245     while 1:
 246         x = r.readline()
 247         if not x: break
 248         f.write(x)
 249     r.close()
 250     del x,r
 251     f.flush()
 252     f.seek(0)
 253     return f
 254
 255 def sizehashes(f):
 256     size = os.fstat(f.fileno())[6]
 257     f.seek(0)
 258     sha1sum = apt_pkg.sha1sum(f)
 259     f.seek(0)
 260     sha256sum = apt_pkg.sha256sum(f)
 261     return (size, sha1sum, sha256sum)
 262
 263 def genchanges(Options, outdir, oldfile, origfile, maxdiffs = 56):
 264     if Options.has_key("NoAct"):
 265         print "Not acting on: od: %s, oldf: %s, origf: %s, md: %s" % (outdir, oldfile, origfile, maxdiffs)
 266         return
 267
 268     patchname = Options["PatchName"]
 269
 270     # origfile = /path/to/Packages
 271     # oldfile  = ./Packages
 272     # newfile  = ./Packages.tmp
 273     # difffile = outdir/patchname
 274     # index   => outdir/Index
 275
 276     # (outdir, oldfile, origfile) = argv
 277
 278     newfile = oldfile + ".new"
 279     difffile = "%s/%s" % (outdir, patchname)
 280
 281     upd = Updates(outdir, int(maxdiffs))
 282     (oldext, oldstat) = smartstat(oldfile)
 283     (origext, origstat) = smartstat(origfile)
 284     if not origstat:
 285         print "%s: doesn't exist" % (origfile)
 286         return
 287     if not oldstat:
 288         print "%s: initial run" % (origfile)
 289         os.link(origfile + origext, oldfile + origext)
 290         return
 291
 292     if oldstat[1:3] == origstat[1:3]:
 293         #print "%s: hardlink unbroken, assuming unchanged" % (origfile)
 294         return
 295
 296     oldf = smartopen(oldfile)
 297     oldsizehashes = sizehashes(oldf)
 298
 299     # should probably early exit if either of these checks fail
 300     # alternatively (optionally?) could just trim the patch history
 301
 302     #if upd.filesizesha1:
 303     #    if upd.filesizesha1 != oldsizesha1:
 304     #        print "info: old file " + oldfile + " changed! %s %s => %s %s" % (upd.filesizesha1 + oldsizesha1)
 305
 306     if Options.has_key("CanonicalPath"): upd.can_path=Options["CanonicalPath"]
 307
 308     if os.path.exists(newfile): os.unlink(newfile)
 309     smartlink(origfile, newfile)
 310     newf = open(newfile, "r")
 311     newsizehashes = sizehashes(newf)
 312     newf.close()
 313
 314     if newsizehashes == oldsizehashes:
 315         os.unlink(newfile)
 316         oldf.close()
 317         #print "%s: unchanged" % (origfile)
 318     else:
 319         if not os.path.isdir(outdir):
 320             os.mkdir(outdir)
 321
 322         w = os.popen("diff --ed - %s | gzip --rsyncable -c -9 > %s.gz" %
 323                      (newfile, difffile), "w")
 324         pipe_file(oldf, w)
 325         oldf.close()
 326
 327         difff = smartopen(difffile)
 328         difsizehashes = sizehashes(difff)
 329         difff.close()
 330
 331         difffgz = open(difffile + ".gz", "r")
 332         difgzsizehashes = sizehashes(difffgz)
 333         difffgz.close()
 334
 335         upd.history[patchname] = (oldsizehashes, difsizehashes, difgzsizehashes)
 336         upd.history_order.append(patchname)
 337
 338         upd.filesizehashes = newsizehashes
 339
 340         os.unlink(oldfile + oldext)
 341         os.link(origfile + origext, oldfile + origext)
 342         os.unlink(newfile)
 343
 344         with open(outdir + "/Index.new", "w") as f:
 345             upd.dump(f)
 346         os.rename(outdir + "/Index.new", outdir + "/Index")
 347
 348
 349 def main():
 350     global Cnf, Options, Logger
 351
 352     os.umask(0o002)
 353
 354     Cnf = utils.get_conf()
 355     Arguments = [ ('h', "help", "Generate-Index-Diffs::Options::Help"),
 356                   ('a', 'archive', 'Generate-Index-Diffs::Options::Archive', 'hasArg'),
 357                   ('c', None, "Generate-Index-Diffs::Options::CanonicalPath", "hasArg"),
 358                   ('p', "patchname", "Generate-Index-Diffs::Options::PatchName", "hasArg"),
 359                   ('d', "tmpdir", "Generate-Index-Diffs::Options::TempDir", "hasArg"),
 360                   ('m', "maxdiffs", "Generate-Index-Diffs::Options::MaxDiffs", "hasArg"),
 361                   ('n', "n-act", "Generate-Index-Diffs::Options::NoAct"),
 362                 ]
 363     suites = apt_pkg.parse_commandline(Cnf,Arguments,sys.argv)
 364     Options = Cnf.subtree("Generate-Index-Diffs::Options")
 365     if Options.has_key("Help"): usage()
 366
 367     maxdiffs = Options.get("MaxDiffs::Default", "56")
 368     maxpackages = Options.get("MaxDiffs::Packages", maxdiffs)
 369     maxcontents = Options.get("MaxDiffs::Contents", maxdiffs)
 370     maxsources = Options.get("MaxDiffs::Sources", maxdiffs)
 371
 372     if not Options.has_key("PatchName"):
 373         format = "%Y-%m-%d-%H%M.%S"
 374         Options["PatchName"] = time.strftime( format )
 375
 376     session = DBConn().session()
 377
 378     if not suites:
 379         query = session.query(Suite.suite_name)
 380         if Options.get('Archive'):
 381             query = query.join(Suite.archive).filter(Archive.archive_name == Options['Archive'])
 382         suites = [ s.suite_name for s in query ]
 383
 384     for suitename in suites:
 385         print "Processing: " + suitename
 386
 387         suiteobj = get_suite(suitename.lower(), session=session)
 388
 389         # Use the canonical version of the suite name
 390         suite = suiteobj.suite_name
 391
 392         if suiteobj.untouchable:
 393             print "Skipping: " + suite + " (untouchable)"
 394             continue
 395
 396         architectures = get_suite_architectures(suite, skipall=True, session=session)
 397         components = [ c.component_name for c in session.query(Component.component_name) ]
 398
 399         suite_suffix = Cnf.find("Dinstall::SuiteSuffix")
 400         if components and suite_suffix:
 401             longsuite = suite + "/" + suite_suffix
 402         else:
 403             longsuite = suite
 404
 405         tree = os.path.join(suiteobj.archive.path, 'dists', longsuite)
 406
 407         # See if there are Translations which might need a new pdiff
 408         cwd = os.getcwd()
 409         for component in components:
 410             #print "DEBUG: Working on %s" % (component)
 411             workpath=os.path.join(tree, component, "i18n")
 412             if os.path.isdir(workpath):
 413                 os.chdir(workpath)
 414                 for dirpath, dirnames, filenames in os.walk(".", followlinks=True, topdown=True):
 415                     for entry in filenames:
 416                         if not re_includeinpdiff.match(entry):
 417                             #print "EXCLUDING %s" % (entry)
 418                             continue
 419                         (fname, fext) = os.path.splitext(entry)
 420                         processfile=os.path.join(workpath, fname)
 421                         #print "Working: %s" % (processfile)
 422                         storename="%s/%s_%s_%s" % (Options["TempDir"], suite, component, fname)
 423                         #print "Storefile: %s" % (storename)
 424                         genchanges(Options, processfile + ".diff", storename, processfile, maxdiffs)
 425         os.chdir(cwd)
 426
 427         for archobj in architectures:
 428             architecture = archobj.arch_string
 429
 430             for component in components:
 431                 if architecture == "source":
 432                     longarch = architecture
 433                     packages = "Sources"
 434                     maxsuite = maxsources
 435                 else:
 436                     longarch = "binary-%s" % (architecture)
 437                     packages = "Packages"
 438                     maxsuite = maxpackages
 439
 440                 # Process Contents
 441                 file = "%s/%s/Contents-%s" % (tree, component, architecture)
 442                 storename = "%s/%s_%s_contents_%s" % (Options["TempDir"], suite, component, architecture)
 443                 genchanges(Options, file + ".diff", storename, file, maxcontents)
 444
 445                 file = "%s/%s/%s/%s" % (tree, component, longarch, packages)
 446                 storename = "%s/%s_%s_%s" % (Options["TempDir"], suite, component, architecture)
 447                 genchanges(Options, file + ".diff", storename, file, maxsuite)
 448
 449 ################################################################################
 450
 451 if __name__ == '__main__':
 452     main()