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"]:
  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     else:
  89         print "missing: %s" % (f)
  90         raise IOError(f)
  91
  92 def smartopen(file):
  93     if os.path.isfile(file):
  94         f = open(file, "r")
  95     elif os.path.isfile("%s.gz" % file):
  96         f = create_temp_file(os.popen("zcat %s.gz" % file, "r"))
  97     elif os.path.isfile("%s.bz2" % file):
  98         f = create_temp_file(os.popen("bzcat %s.bz2" % file, "r"))
  99     else:
 100         f = None
 101     return f
 102
 103 def pipe_file(f, t):
 104     f.seek(0)
 105     while 1:
 106         l = f.read()
 107         if not l: break
 108         t.write(l)
 109     t.close()
 110
 111 class Updates:
 112     def __init__(self, readpath = None, max = 56):
 113         self.can_path = None
 114         self.history = {}
 115         self.history_order = []
 116         self.max = max
 117         self.readpath = readpath
 118         self.filesizesha1 = None
 119
 120         if readpath:
 121             try:
 122                 f = open(readpath + "/Index")
 123                 x = f.readline()
 124
 125                 def read_hashs(ind, f, self, x=x):
 126                     while 1:
 127                         x = f.readline()
 128                         if not x or x[0] != " ": break
 129                         l = x.split()
 130                         fname = l[2]
 131                         if fname.endswith('.gz'):
 132                             fname = fname[:-3]
 133                         if not self.history.has_key(fname):
 134                             self.history[fname] = [None,None,None]
 135                             self.history_order.append(fname)
 136                         self.history[fname][ind] = (l[0], int(l[1]))
 137                     return x
 138
 139                 while x:
 140                     l = x.split()
 141
 142                     if len(l) == 0:
 143                         x = f.readline()
 144                         continue
 145
 146                     if l[0] == "SHA1-History:":
 147                         x = read_hashs(0,f,self)
 148                         continue
 149
 150                     if l[0] == "SHA1-Patches:":
 151                         x = read_hashs(1,f,self)
 152                         continue
 153
 154                     if l[0] == "SHA1-Download:":
 155                         x = read_hashs(2,f,self)
 156                         continue
 157
 158                     if l[0] == "Canonical-Name:" or l[0]=="Canonical-Path:":
 159                         self.can_path = l[1]
 160
 161                     if l[0] == "SHA1-Current:" and len(l) == 3:
 162                         self.filesizesha1 = (l[1], int(l[2]))
 163
 164                     x = f.readline()
 165
 166             except IOError:
 167                 0
 168
 169     def dump(self, out=sys.stdout):
 170         if self.can_path:
 171             out.write("Canonical-Path: %s\n" % (self.can_path))
 172
 173         if self.filesizesha1:
 174             out.write("SHA1-Current: %s %7d\n" % (self.filesizesha1))
 175
 176         hs = self.history
 177         l = self.history_order[:]
 178
 179         cnt = len(l)
 180         if cnt > self.max:
 181             for h in l[:cnt-self.max]:
 182                 tryunlink("%s/%s.gz" % (self.readpath, h))
 183                 del hs[h]
 184             l = l[cnt-self.max:]
 185             self.history_order = l[:]
 186
 187         out.write("SHA1-History:\n")
 188         for h in l:
 189             out.write(" %s %7d %s\n" % (hs[h][0][0], hs[h][0][1], h))
 190         out.write("SHA1-Patches:\n")
 191         for h in l:
 192             out.write(" %s %7d %s\n" % (hs[h][1][0], hs[h][1][1], h))
 193         out.write("SHA1-Download:\n")
 194         for h in l:
 195             if hs[h][2]:
 196                 out.write(" %s %7d %s.gz\n" % (hs[h][2][0], hs[h][2][1], h))
 197
 198 def create_temp_file(r):
 199     f = tempfile.TemporaryFile()
 200     while 1:
 201         x = r.readline()
 202         if not x: break
 203         f.write(x)
 204     r.close()
 205     del x,r
 206     f.flush()
 207     f.seek(0)
 208     return f
 209
 210 def sizesha1(f):
 211     size = os.fstat(f.fileno())[6]
 212     f.seek(0)
 213     sha1sum = apt_pkg.sha1sum(f)
 214     return (sha1sum, size)
 215
 216 def genchanges(Options, outdir, oldfile, origfile, maxdiffs = 56):
 217     if Options.has_key("NoAct"):
 218         print "Not acting on: od: %s, oldf: %s, origf: %s, md: %s" % (outdir, oldfile, origfile, maxdiffs)
 219         return
 220
 221     patchname = Options["PatchName"]
 222
 223     # origfile = /path/to/Packages
 224     # oldfile  = ./Packages
 225     # newfile  = ./Packages.tmp
 226     # difffile = outdir/patchname
 227     # index   => outdir/Index
 228
 229     # (outdir, oldfile, origfile) = argv
 230
 231     newfile = oldfile + ".new"
 232     difffile = "%s/%s" % (outdir, patchname)
 233
 234     upd = Updates(outdir, int(maxdiffs))
 235     (oldext, oldstat) = smartstat(oldfile)
 236     (origext, origstat) = smartstat(origfile)
 237     if not origstat:
 238         print "%s: doesn't exist" % (origfile)
 239         return
 240     if not oldstat:
 241         print "%s: initial run" % (origfile)
 242         os.link(origfile + origext, oldfile + origext)
 243         return
 244
 245     if oldstat[1:3] == origstat[1:3]:
 246         #print "%s: hardlink unbroken, assuming unchanged" % (origfile)
 247         return
 248
 249     oldf = smartopen(oldfile)
 250     oldsizesha1 = sizesha1(oldf)
 251
 252     # should probably early exit if either of these checks fail
 253     # alternatively (optionally?) could just trim the patch history
 254
 255     #if upd.filesizesha1:
 256     #    if upd.filesizesha1 != oldsizesha1:
 257     #        print "info: old file " + oldfile + " changed! %s %s => %s %s" % (upd.filesizesha1 + oldsizesha1)
 258
 259     if Options.has_key("CanonicalPath"): upd.can_path=Options["CanonicalPath"]
 260
 261     if os.path.exists(newfile): os.unlink(newfile)
 262     smartlink(origfile, newfile)
 263     newf = open(newfile, "r")
 264     newsizesha1 = sizesha1(newf)
 265     newf.close()
 266
 267     if newsizesha1 == oldsizesha1:
 268         os.unlink(newfile)
 269         oldf.close()
 270         #print "%s: unchanged" % (origfile)
 271     else:
 272         if not os.path.isdir(outdir):
 273             os.mkdir(outdir)
 274
 275         w = os.popen("diff --ed - %s | gzip --rsyncable -c -9 > %s.gz" %
 276                      (newfile, difffile), "w")
 277         pipe_file(oldf, w)
 278         oldf.close()
 279
 280         difff = smartopen(difffile)
 281         difsizesha1 = sizesha1(difff)
 282         difff.close()
 283
 284         difffgz = open(difffile + ".gz", "r")
 285         difgzsizesha1 = sizesha1(difffgz)
 286         difffgz.close()
 287
 288         upd.history[patchname] = (oldsizesha1, difsizesha1, difgzsizesha1)
 289         upd.history_order.append(patchname)
 290
 291         upd.filesizesha1 = newsizesha1
 292
 293         os.unlink(oldfile + oldext)
 294         os.link(origfile + origext, oldfile + origext)
 295         os.unlink(newfile)
 296
 297         f = open(outdir + "/Index", "w")
 298         upd.dump(f)
 299         f.close()
 300
 301
 302 def main():
 303     global Cnf, Options, Logger
 304
 305     os.umask(0o002)
 306
 307     Cnf = utils.get_conf()
 308     Arguments = [ ('h', "help", "Generate-Index-Diffs::Options::Help"),
 309                   ('a', 'archive', 'Generate-Index-Diffs::Options::Archive', 'hasArg'),
 310                   ('c', None, "Generate-Index-Diffs::Options::CanonicalPath", "hasArg"),
 311                   ('p', "patchname", "Generate-Index-Diffs::Options::PatchName", "hasArg"),
 312                   ('d', "tmpdir", "Generate-Index-Diffs::Options::TempDir", "hasArg"),
 313                   ('m', "maxdiffs", "Generate-Index-Diffs::Options::MaxDiffs", "hasArg"),
 314                   ('n', "n-act", "Generate-Index-Diffs::Options::NoAct"),
 315                 ]
 316     suites = apt_pkg.parse_commandline(Cnf,Arguments,sys.argv)
 317     Options = Cnf.subtree("Generate-Index-Diffs::Options")
 318     if Options.has_key("Help"): usage()
 319
 320     maxdiffs = Options.get("MaxDiffs::Default", "56")
 321     maxpackages = Options.get("MaxDiffs::Packages", maxdiffs)
 322     maxcontents = Options.get("MaxDiffs::Contents", maxdiffs)
 323     maxsources = Options.get("MaxDiffs::Sources", maxdiffs)
 324
 325     if not Options.has_key("PatchName"):
 326         format = "%Y-%m-%d-%H%M.%S"
 327         Options["PatchName"] = time.strftime( format )
 328
 329     session = DBConn().session()
 330
 331     if not suites:
 332         query = session.query(Suite.suite_name)
 333         if Options.get('Archive'):
 334             query = query.join(Suite.archive).filter(Archive.archive_name == Options['Archive'])
 335         suites = [ s.suite_name for s in query ]
 336
 337     for suitename in suites:
 338         print "Processing: " + suitename
 339
 340         suiteobj = get_suite(suitename.lower(), session=session)
 341
 342         # Use the canonical version of the suite name
 343         suite = suiteobj.suite_name
 344
 345         if suiteobj.untouchable:
 346             print "Skipping: " + suite + " (untouchable)"
 347             continue
 348
 349         architectures = get_suite_architectures(suite, skipall=True, session=session)
 350         components = [ c.component_name for c in session.query(Component.component_name) ]
 351
 352         suite_suffix = Cnf.find("Dinstall::SuiteSuffix")
 353         if components and suite_suffix:
 354             longsuite = suite + "/" + suite_suffix
 355         else:
 356             longsuite = suite
 357
 358         tree = os.path.join(suiteobj.archive.path, 'dists', longsuite)
 359
 360         # See if there are Translations which might need a new pdiff
 361         cwd = os.getcwd()
 362         for component in components:
 363             #print "DEBUG: Working on %s" % (component)
 364             workpath=os.path.join(tree, component, "i18n")
 365             if os.path.isdir(workpath):
 366                 os.chdir(workpath)
 367                 for dirpath, dirnames, filenames in os.walk(".", followlinks=True, topdown=True):
 368                     for entry in filenames:
 369                         if not re_includeinpdiff.match(entry):
 370                             #print "EXCLUDING %s" % (entry)
 371                             continue
 372                         (fname, fext) = os.path.splitext(entry)
 373                         processfile=os.path.join(workpath, fname)
 374                         #print "Working: %s" % (processfile)
 375                         storename="%s/%s_%s_%s" % (Options["TempDir"], suite, component, fname)
 376                         #print "Storefile: %s" % (storename)
 377                         genchanges(Options, processfile + ".diff", storename, processfile, maxdiffs)
 378         os.chdir(cwd)
 379
 380         for archobj in architectures:
 381             architecture = archobj.arch_string
 382
 383             for component in components:
 384                 if architecture == "source":
 385                     longarch = architecture
 386                     packages = "Sources"
 387                     maxsuite = maxsources
 388                 else:
 389                     longarch = "binary-%s"% (architecture)
 390                     packages = "Packages"
 391                     maxsuite = maxpackages
 392                     # Process Contents
 393                     file = "%s/%s/Contents-%s" % (tree, component, architecture)
 394                     storename = "%s/%s_%s_contents_%s" % (Options["TempDir"], suite, component, architecture)
 395                     genchanges(Options, file + ".diff", storename, file, maxcontents)
 396
 397                 file = "%s/%s/%s/%s" % (tree, component, longarch, packages)
 398                 storename = "%s/%s_%s_%s" % (Options["TempDir"], suite, component, architecture)
 399                 genchanges(Options, file + ".diff", storename, file, maxsuite)
 400
 401 ################################################################################
 402
 403 if __name__ == '__main__':
 404     main()