scripts: add additional regex for dl_cleanup
[openwrt/staging/jow.git] / 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"(.+)[-_]v(\d+)\.(\d+)\.(\d+)(\w?)"), parseVer_123), # xxx-v1.2.3a
123 (re.compile(r"(.+)[-_](\d+)_(\d+)_(\d+)"), parseVer_123), # xxx-1_2_3
124 (re.compile(r"(.+)[-_](\d+)\.(\d+)(\w?)"), parseVer_12), # xxx-1.2a
125 (re.compile(r"(.+)[-_]v(\d+)\.(\d+)(\w?)"), parseVer_12), # xxx-v1.2a
126 (re.compile(r"(.+)[-_]r?(\d+)"), parseVer_r), # xxx-r1111
127 )
128
129 blacklist = [
130 ("wl_apsta", re.compile(r"wl_apsta.*")),
131 (".fw", re.compile(r".*\.fw")),
132 (".arm", re.compile(r".*\.arm")),
133 (".bin", re.compile(r".*\.bin")),
134 ("rt-firmware", re.compile(r"RT[\d\w]+_Firmware.*")),
135 ]
136
137
138 class EntryParseError(Exception):
139 pass
140
141
142 class Entry:
143 def __init__(self, directory, filename):
144 self.directory = directory
145 self.filename = filename
146 self.progname = ""
147 self.fileext = ""
148
149 for ext in extensions:
150 if filename.endswith(ext):
151 filename = filename[0 : 0 - len(ext)]
152 self.fileext = ext
153 break
154 else:
155 print(self.filename, "has an unknown file-extension")
156 raise EntryParseError("ext")
157 for (regex, parseVersion) in versionRegex:
158 match = regex.match(filename)
159 if match:
160 (self.progname, self.version) = parseVersion(
161 match, directory + "/" + filename + self.fileext
162 )
163 break
164 else:
165 print(self.filename, "has an unknown version pattern")
166 raise EntryParseError("ver")
167
168 def getPath(self):
169 return (self.directory + "/" + self.filename).replace("//", "/")
170
171 def deleteFile(self):
172 path = self.getPath()
173 print("Deleting", path)
174 if not opt_dryrun:
175 os.unlink(path)
176
177 def __ge__(self, y):
178 return self.version >= y.version
179
180
181 def usage():
182 print("OpenWrt download directory cleanup utility")
183 print("Usage: " + sys.argv[0] + " [OPTIONS] <path/to/dl>")
184 print("")
185 print(" -d|--dry-run Do a dry-run. Don't delete any files")
186 print(" -B|--show-blacklist Show the blacklist and exit")
187 print(" -w|--whitelist ITEM Remove ITEM from blacklist")
188
189
190 def main(argv):
191 global opt_dryrun
192
193 try:
194 (opts, args) = getopt.getopt(
195 argv[1:],
196 "hdBw:",
197 [
198 "help",
199 "dry-run",
200 "show-blacklist",
201 "whitelist=",
202 ],
203 )
204 if len(args) != 1:
205 usage()
206 return 1
207 except getopt.GetoptError as e:
208 usage()
209 return 1
210 directory = args[0]
211
212 if not os.path.exists(directory):
213 print("Can't find dl path", directory)
214 return 1
215
216 for (o, v) in opts:
217 if o in ("-h", "--help"):
218 usage()
219 return 0
220 if o in ("-d", "--dry-run"):
221 opt_dryrun = True
222 if o in ("-w", "--whitelist"):
223 for i in range(0, len(blacklist)):
224 (name, regex) = blacklist[i]
225 if name == v:
226 del blacklist[i]
227 break
228 else:
229 print("Whitelist error: Item", v, "is not in blacklist")
230 return 1
231 if o in ("-B", "--show-blacklist"):
232 for (name, regex) in blacklist:
233 sep = "\t\t"
234 if len(name) >= 8:
235 sep = "\t"
236 print("%s%s(%s)" % (name, sep, regex.pattern))
237 return 0
238
239 # Create a directory listing and parse the file names.
240 entries = []
241 for filename in os.listdir(directory):
242 if filename == "." or filename == "..":
243 continue
244 for (name, regex) in blacklist:
245 if regex.match(filename):
246 if opt_dryrun:
247 print(filename, "is blacklisted")
248 break
249 else:
250 try:
251 entries.append(Entry(directory, filename))
252 except EntryParseError as e:
253 pass
254
255 # Create a map of programs
256 progmap = {}
257 for entry in entries:
258 if entry.progname in progmap.keys():
259 progmap[entry.progname].append(entry)
260 else:
261 progmap[entry.progname] = [
262 entry,
263 ]
264
265 # Traverse the program map and delete everything but the last version
266 for prog in progmap:
267 lastVersion = None
268 versions = progmap[prog]
269 for version in versions:
270 if lastVersion is None or version >= lastVersion:
271 lastVersion = version
272 if lastVersion:
273 for version in versions:
274 if version is not lastVersion:
275 version.deleteFile()
276 if opt_dryrun:
277 print("Keeping", lastVersion.getPath())
278
279 return 0
280
281
282 if __name__ == "__main__":
283 sys.exit(main(sys.argv))