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