scripts/dl_cleanup.py

   1 #!/usr/bin/env python3
   2 """
   3 # OpenWrt download directory cleanup utility.
   4 # Delete all but the very last version of the program tarballs.
   5 #
   6 # Copyright (C) 2010-2015 Michael Buesch <m@bues.ch>
   7 # Copyright (C) 2013-2015 OpenWrt.org
   8 """
   9
  10 from __future__ import print_function
  11
  12 import sys
  13 import os
  14 import re
  15 import getopt
  16 import shutil
  17
  18 # Commandline options
  19 opt_dryrun = False
  20
  21
  22 def parseVer_1234(match, filepath):
  23     progname = match.group(1)
  24     progversion = (
  25         (int(match.group(2)) << 64)
  26         | (int(match.group(3)) << 48)
  27         | (int(match.group(4)) << 32)
  28         | (int(match.group(5)) << 16)
  29     )
  30     return (progname, progversion)
  31
  32
  33 def parseVer_123(match, filepath):
  34     progname = match.group(1)
  35     try:
  36         patchlevel = match.group(5)
  37     except IndexError as e:
  38         patchlevel = None
  39     if patchlevel:
  40         patchlevel = ord(patchlevel[0])
  41     else:
  42         patchlevel = 0
  43     progversion = (
  44         (int(match.group(2)) << 64)
  45         | (int(match.group(3)) << 48)
  46         | (int(match.group(4)) << 32)
  47         | patchlevel
  48     )
  49     return (progname, progversion)
  50
  51
  52 def parseVer_12(match, filepath):
  53     progname = match.group(1)
  54     try:
  55         patchlevel = match.group(4)
  56     except IndexError as e:
  57         patchlevel = None
  58     if patchlevel:
  59         patchlevel = ord(patchlevel[0])
  60     else:
  61         patchlevel = 0
  62     progversion = (int(match.group(2)) << 64) | (int(match.group(3)) << 48) | patchlevel
  63     return (progname, progversion)
  64
  65
  66 def parseVer_r(match, filepath):
  67     progname = match.group(1)
  68     progversion = int(match.group(2)) << 64
  69     return (progname, progversion)
  70
  71
  72 def parseVer_ymd_GIT_SHASUM(match, filepath):
  73     progname = match.group(1)
  74     progversion = (
  75         (int(match.group(2)) << 64)
  76         | (int(match.group(3)) << 48)
  77         | (int(match.group(4)) << 32)
  78     )
  79     return (progname, progversion)
  80
  81
  82 def parseVer_ymd(match, filepath):
  83     progname = match.group(1)
  84     progversion = (
  85         (int(match.group(2)) << 64)
  86         | (int(match.group(3)) << 48)
  87         | (int(match.group(4)) << 32)
  88     )
  89     return (progname, progversion)
  90
  91
  92 def parseVer_GIT(match, filepath):
  93     progname = match.group(1)
  94     st = os.stat(filepath)
  95     progversion = int(st.st_mtime) << 64
  96     return (progname, progversion)
  97
  98
  99 extensions = (
 100     ".tar.gz",
 101     ".tar.bz2",
 102     ".tar.xz",
 103     ".orig.tar.gz",
 104     ".orig.tar.bz2",
 105     ".orig.tar.xz",
 106     ".zip",
 107     ".tgz",
 108     ".tbz",
 109     ".txz",
 110 )
 111
 112 versionRegex = (
 113     (re.compile(r"(gcc[-_]\d+)\.(\d+)\.(\d+)"), parseVer_12),  # gcc.1.2
 114     (re.compile(r"(linux[-_]\d+\.\d+)\.(\d+)"), parseVer_r),  # linux.1
 115     (re.compile(r"(.+)[-_](\d+)\.(\d+)\.(\d+)\.(\d+)"), parseVer_1234),  # xxx-1.2.3.4
 116     (
 117         re.compile(r"(.+)[-_](\d\d\d\d)-?(\d\d)-?(\d\d)-"),
 118         parseVer_ymd_GIT_SHASUM,
 119     ),  # xxx-YYYY-MM-DD-GIT_SHASUM
 120     (re.compile(r"(.+)[-_](\d\d\d\d)-?(\d\d)-?(\d\d)"), parseVer_ymd),  # xxx-YYYY-MM-DD
 121     (re.compile(r"(.+)[-_]([0-9a-fA-F]{40,40})"), parseVer_GIT),  # xxx-GIT_SHASUM
 122     (re.compile(r"(.+)[-_](\d+)\.(\d+)\.(\d+)(\w?)"), parseVer_123),  # xxx-1.2.3a
 123     (re.compile(r"(.+)[-_]v(\d+)\.(\d+)\.(\d+)(\w?)"), parseVer_123),  # xxx-v1.2.3a
 124     (re.compile(r"(.+)[-_](\d+)_(\d+)_(\d+)"), parseVer_123),  # xxx-1_2_3
 125     (re.compile(r"(.+)[-_](\d+)\.(\d+)(\w?)"), parseVer_12),  # xxx-1.2a
 126     (re.compile(r"(.+)[-_]v(\d+)\.(\d+)(\w?)"), parseVer_12),  # xxx-v1.2a
 127     (re.compile(r"(.+)[-_]r?(\d+)"), parseVer_r),  # xxx-r1111
 128 )
 129
 130 blacklist = [
 131     ("wl_apsta", re.compile(r"wl_apsta.*")),
 132     (".fw", re.compile(r".*\.fw")),
 133     (".arm", re.compile(r".*\.arm")),
 134     (".bin", re.compile(r".*\.bin")),
 135     ("rt-firmware", re.compile(r"RT[\d\w]+_Firmware.*")),
 136 ]
 137
 138
 139 class EntryParseError(Exception):
 140     pass
 141
 142
 143 class Entry:
 144     def __init__(self, directory, builddir, filename):
 145         self.directory = directory
 146         self.filename = filename
 147         self.builddir = builddir
 148         self.progname = ""
 149         self.fileext = ""
 150         self.filenoext = ""
 151
 152         if os.path.isdir(self.getPath()):
 153             self.filenoext = filename
 154         else:
 155             for ext in extensions:
 156                 if filename.endswith(ext):
 157                     filename = filename[0 : 0 - len(ext)]
 158                     self.filenoext = filename
 159                     self.fileext = ext
 160                     break
 161             else:
 162                 print(self.filename, "has an unknown file-extension")
 163                 raise EntryParseError("ext")
 164         for (regex, parseVersion) in versionRegex:
 165             match = regex.match(filename)
 166             if match:
 167                 (self.progname, self.version) = parseVersion(
 168                     match, directory + "/" + filename + self.fileext
 169                 )
 170                 break
 171         else:
 172             print(self.filename, "has an unknown version pattern")
 173             raise EntryParseError("ver")
 174
 175     def getPath(self):
 176         return (self.directory + "/" + self.filename).replace("//", "/")
 177
 178     def getBuildPaths(self):
 179         paths = []
 180         for subdir in os.scandir(self.builddir):
 181             package_build_dir = os.path.join(subdir.path, self.filenoext)
 182             if os.path.exists(package_build_dir):
 183                 paths.append(package_build_dir)
 184         return paths
 185
 186     def deleteFile(self):
 187         path = self.getPath()
 188         print("Deleting", path)
 189         if not opt_dryrun:
 190             if os.path.isdir(path):
 191                 shutil.rmtree(path)
 192             else:
 193                 os.unlink(path)
 194
 195     def deleteBuildDir(self):
 196         paths = self.getBuildPaths()
 197         for path in paths:
 198             print("Deleting BuildDir", path)
 199             if not opt_dryrun:
 200                     shutil.rmtree(path)
 201
 202     def __ge__(self, y):
 203         return self.version >= y.version
 204
 205
 206 def usage():
 207     print("OpenWrt download directory cleanup utility")
 208     print("Usage: " + sys.argv[0] + " [OPTIONS] <path/to/dl>")
 209     print("")
 210     print(" -d|--dry-run            Do a dry-run. Don't delete any files")
 211     print(" -B|--show-blacklist     Show the blacklist and exit")
 212     print(" -w|--whitelist ITEM     Remove ITEM from blacklist")
 213     print(
 214         " -D|--download-dir       Provide path to dl dir to clean also the build directory"
 215     )
 216     print(
 217         " -b|--build-dir          Provide path to build dir to clean also the build directory"
 218     )
 219
 220
 221 def main(argv):
 222     global opt_dryrun
 223
 224     try:
 225         (opts, args) = getopt.getopt(
 226             argv[1:],
 227             "hdBw:D:b:",
 228             [
 229                 "help",
 230                 "dry-run",
 231                 "show-blacklist",
 232                 "whitelist=",
 233                 "download-dir=",
 234                 "build-dir=",
 235             ],
 236         )
 237     except getopt.GetoptError as e:
 238         usage()
 239         return 1
 240
 241     directory = "dl/"
 242     builddir = "build_dir/"
 243
 244     for (o, v) in opts:
 245         if o in ("-h", "--help"):
 246             usage()
 247             return 0
 248         if o in ("-d", "--dry-run"):
 249             opt_dryrun = True
 250         if o in ("-w", "--whitelist"):
 251             for i in range(0, len(blacklist)):
 252                 (name, regex) = blacklist[i]
 253                 if name == v:
 254                     del blacklist[i]
 255                     break
 256             else:
 257                 print("Whitelist error: Item", v, "is not in blacklist")
 258                 return 1
 259         if o in ("-B", "--show-blacklist"):
 260             for (name, regex) in blacklist:
 261                 sep = "\t\t"
 262                 if len(name) >= 8:
 263                     sep = "\t"
 264                 print("%s%s(%s)" % (name, sep, regex.pattern))
 265             return 0
 266         if o in ("-D", "--download-dir"):
 267             directory = v
 268         if o in ("-b", "--build-dir"):
 269             builddir = v
 270
 271     if args:
 272         directory = args[0]
 273
 274     if not os.path.exists(directory):
 275         print("Can't find download directory", directory)
 276         return 1
 277
 278     if not os.path.exists(builddir):
 279         print("Can't find build directory", builddir)
 280         return 1
 281
 282     # Create a directory listing and parse the file names.
 283     entries = []
 284     for filename in os.listdir(directory):
 285         if filename == "." or filename == "..":
 286             continue
 287         for (name, regex) in blacklist:
 288             if regex.match(filename):
 289                 if opt_dryrun:
 290                     print(filename, "is blacklisted")
 291                 break
 292         else:
 293             try:
 294                 entries.append(Entry(directory, builddir, filename))
 295             except EntryParseError as e:
 296                 pass
 297
 298     # Create a map of programs
 299     progmap = {}
 300     for entry in entries:
 301         if entry.progname in progmap.keys():
 302             progmap[entry.progname].append(entry)
 303         else:
 304             progmap[entry.progname] = [
 305                 entry,
 306             ]
 307
 308     # Traverse the program map and delete everything but the last version
 309     for prog in progmap:
 310         lastVersion = None
 311         versions = progmap[prog]
 312         for version in versions:
 313             if lastVersion:
 314                 if os.path.isdir(lastVersion.getPath()) and not os.path.isdir(version.getPath()):
 315                     continue
 316             if lastVersion is None or version >= lastVersion:
 317                 lastVersion = version
 318         if lastVersion:
 319             for version in versions:
 320                 if version is not lastVersion:
 321                     version.deleteFile()
 322                     if builddir:
 323                         version.deleteBuildDir()
 324             if opt_dryrun:
 325                 print("Keeping", lastVersion.getPath())
 326
 327     return 0
 328
 329
 330 if __name__ == "__main__":
 331     sys.exit(main(sys.argv))