]>
Commit | Line | Data |
---|---|---|
7ebf9245 | 1 | #!/usr/bin/perl -w |
b396edfa | 2 | # $Id$ |
7ebf9245 | 3 | |
56ea0f1e | 4 | use IPC::Run qw(run); |
5 | ||
7a0595ea | 6 | $commits_list = "pld-cvs-commit\@lists.pld-linux.org"; |
7ebf9245 | 7 | |
b5c674c5 | 8 | $spool_dir = "./spool"; |
01806192 MM |
9 | $copy_dir = "src"; # relative to ftp root |
10 | $no_url_dir = "./upload"; | |
11 | $df_server = "distfiles.pld-linux.org"; | |
51fe5d4e | 12 | $df_scp = "plddist\@ep09.pld-linux.org:ftp"; |
62f7358e | 13 | $user_agent = "PLD/distfiles"; |
7ebf9245 MM |
14 | |
15 | @md5 = (); | |
ccbaa374 | 16 | @url = (); |
7ebf9245 MM |
17 | $problems = ""; |
18 | $normal_out = ""; | |
19 | $requester = ""; | |
20 | $file = ""; | |
47f8fda7 MM |
21 | $fetched_count = 0; |
22 | $force_reply = 0; | |
bca91c1b | 23 | $req_login = "nobody"; |
44cd280f | 24 | @files = (); |
7ebf9245 MM |
25 | |
26 | # try lookup some file in spool, exit if it cannot be done | |
27 | sub find_file_in_spool() | |
28 | { | |
29 | opendir(DIR, $spool_dir) || die "can't opendir $spool_dir: $!"; | |
30 | while (1) { | |
31 | my $f = readdir(DIR); | |
32 | defined $f or last; | |
33 | -f "$spool_dir/$f" or next; | |
34 | $file = "$spool_dir/$f"; | |
35 | last; | |
36 | } | |
37 | closedir(DIR); | |
38 | ||
39 | exit 0 if ($file eq ""); | |
40 | } | |
41 | ||
42 | # read file from spool, and try unlink it. if cannot unlink -- exit | |
576dff03 | 43 | # sets $requester (email), $problems, @md5 (arrays of md5's) |
ccbaa374 | 44 | # and @url (map from md5 to urls) |
7ebf9245 MM |
45 | sub read_spool_file() |
46 | { | |
47 | open(F, "< $file") || exit 0; | |
48 | $requester = <F>; | |
49 | chomp $requester; | |
576dff03 | 50 | $requester =~ /^[a-zA-Z_0-9@.-]+$/ |
7ebf9245 | 51 | or die "$file: evil requester: $requester"; |
bca91c1b MM |
52 | $req_login = ""; |
53 | $requester =~ /^([^@]+)\@/ and $req_login = $1; | |
576dff03 | 54 | $req_login =~ /^[a-z0-9A-Z_.]+$/ or die "$file: evil requester $requester"; |
47f8fda7 MM |
55 | my $flags = <F>; |
56 | $force_reply++ if ($flags =~ /force-reply/); | |
576dff03 | 57 | |
7ebf9245 MM |
58 | while (<F>) { |
59 | if (/^ERROR/) { | |
60 | s/^ERROR: //; | |
61 | $problems .= $_; | |
f2a640bf | 62 | next; |
7ebf9245 | 63 | } |
e4708e3e | 64 | if (/^([a-f0-9]{32})\s+((ftp|http|https|no-url|no-url-copy):\/\/([=\@\?a-z0-9A-Z:\+\~\.,\-\/_]|\%[0-9])+)\s*$/) { |
65 | if (/\/$/) { | |
66 | $problems .= "$file: cannot fetch dir"; | |
67 | } else { | |
68 | push @md5, $1; | |
69 | push @url, $2; | |
70 | } | |
71 | } else { | |
72 | $problems .= "FILE: $file: corrupted"; | |
73 | } | |
7ebf9245 MM |
74 | } |
75 | close(F); | |
76 | ||
77 | unlink($file) || exit 0; | |
78 | } | |
79 | ||
44cd280f MM |
80 | sub basename($) |
81 | { | |
82 | my $f = shift; | |
83 | $f =~ s|.*/||; | |
84 | return $f; | |
85 | } | |
86 | ||
01806192 | 87 | sub by_md5($$) |
7ebf9245 MM |
88 | { |
89 | my ($md5, $url) = @_; | |
7ebf9245 | 90 | $md5 =~ /^(.)(.)/; |
01806192 | 91 | return "/by-md5/$1/$2/$md5/" . basename($url); |
6ce39ee7 MM |
92 | } |
93 | ||
01806192 | 94 | sub got_on_distfiles($$) |
6ce39ee7 | 95 | { |
01806192 MM |
96 | my ($md5, $url) = @_; |
97 | my $p = by_md5($md5, $url); | |
98 | my $l = `lftp -c 'debug 0; open $df_server; quote size $p'`; | |
99 | return $l =~ /^213 /; | |
100 | } | |
101 | ||
102 | sub copy_to_df($$) | |
103 | { | |
104 | my ($from, $to) = @_; | |
105 | my $cmd = "scp -r -B -q $from $df_scp/$to"; | |
106 | open(E, "$cmd 2>&1 |") or die; | |
107 | my $oops = ""; | |
108 | while (<E>) { | |
109 | $oops .= $_; | |
110 | } | |
219df45c MM |
111 | $oops .= "\nThe command has exited with a non-zero status." |
112 | unless (close (E)); | |
01806192 MM |
113 | $problems .= "scp problems: $cmd:\n$oops\n" |
114 | if ($oops ne ""); | |
115 | return ($oops ne ""); | |
6ce39ee7 MM |
116 | } |
117 | ||
b5c674c5 | 118 | sub move_file($$$) |
6ce39ee7 | 119 | { |
b5c674c5 | 120 | my ($md5, $url, $local_copy) = @_; |
6ce39ee7 | 121 | |
01806192 | 122 | my $bn = basename($url); |
576dff03 | 123 | |
01806192 MM |
124 | if ($local_copy ne "$tmp_dir/$md5/$bn") { |
125 | if (system("mv -f \"$local_copy\" \"$tmp_dir/$md5/$bn\"")) { | |
126 | $problems .= "FATAL: cannot move $local_copy to $tmp_dir\n"; | |
127 | return; | |
128 | } | |
129 | $local_copy = "$tmp_dir/$md5/$bn"; | |
130 | } | |
576dff03 | 131 | |
01806192 MM |
132 | if (open(D, "> $tmp_dir/$md5/$bn.desc")) { |
133 | print D "URL: $url\n"; | |
d286a317 | 134 | print D "Login: $req_login\n"; |
01806192 MM |
135 | print D "MD5: $md5\n"; |
136 | print D 'Fetched-by: $Id$'."\n"; | |
137 | print D "Time: " . time . "\n"; | |
138 | close(D); | |
7ebf9245 | 139 | } else { |
01806192 MM |
140 | $problems .= "ERROR: cannot write $bn.desc\n"; |
141 | } | |
142 | ||
143 | my $dir = by_md5($md5, $url); | |
144 | $dir =~ s|/[^/]+/[^/]+$||; | |
145 | if (copy_to_df("$tmp_dir/$md5/", $dir) == 0) { | |
576dff03 | 146 | $normal_out .= |
44cd280f MM |
147 | "STORED: $url\n" . |
148 | "\t$md5 " . basename($url) . "\n" . | |
01806192 | 149 | "\tSize: " . (-s $local_copy) . " bytes\n"; |
47f8fda7 | 150 | $fetched_count++; |
7ebf9245 MM |
151 | } |
152 | } | |
153 | ||
b5c674c5 MM |
154 | sub make_src_symlink($$) |
155 | { | |
156 | my ($md5, $url) = @_; | |
576dff03 | 157 | |
01806192 | 158 | return unless ($url =~ /^no-url/); |
576dff03 | 159 | |
01806192 MM |
160 | my $b = basename($url); |
161 | if (open(S, "> $tmp_dir/$b.link")) { | |
b5c674c5 | 162 | if ($url =~ /^no-url-copy/) { |
01806192 | 163 | print S (".." . by_md5($md5, $url)); |
b5c674c5 | 164 | } else { |
01806192 | 165 | print S "REMOVE"; |
b5c674c5 | 166 | } |
01806192 MM |
167 | close(S); |
168 | copy_to_df("$tmp_dir/$b.link", "$copy_dir/$b.link"); | |
b5c674c5 | 169 | } else { |
01806192 | 170 | $problems .= "ERROR: cannot write $tmp_dir/$b.link\n"; |
b5c674c5 MM |
171 | } |
172 | } | |
173 | ||
174 | sub md5($) | |
175 | { | |
176 | my $file = shift; | |
56ea0f1e | 177 | my $in = ""; |
178 | my $md5 = ""; | |
179 | my $err = ""; | |
180 | my @cmd = ("md5sum", $file); | |
181 | ||
182 | run \@cmd, \$in, \$md5, \$err; | |
183 | if ($err ne "") { | |
184 | chomp($err); | |
a6b09732 | 185 | $problems .= "FATAL: " . $err . "\n"; |
186 | return "error"; | |
56ea0f1e | 187 | } |
188 | chomp $md5; | |
b5c674c5 MM |
189 | $md5 =~ /^([a-f0-9]{32})/ and $md5 = $1; |
190 | return $md5; | |
191 | } | |
192 | ||
193 | sub handle_no_url($$) | |
194 | { | |
195 | my ($md5, $url) = @_; | |
576dff03 | 196 | |
e4708e3e | 197 | unless ($url =~ m#://([^/]+)#) { |
198 | $problems .= "$url: corrupted! (no-url)"; | |
199 | return; | |
200 | } | |
b5c674c5 | 201 | my $basename = $1; |
bca91c1b | 202 | my $file = "$no_url_dir/$req_login/$basename"; |
b5c674c5 MM |
203 | |
204 | if (-f $file) { | |
205 | my $computed_md5 = md5($file); | |
206 | if ($computed_md5 ne $md5) { | |
207 | $problems .= "FATAL: $file md5 mismatch, needed $md5, got $computed_md5\n"; | |
208 | } else { | |
209 | move_file($md5, $url, $file); | |
210 | make_src_symlink($md5, $url); | |
211 | } | |
212 | } else { | |
bca91c1b | 213 | $problems .= "FATAL: $file was not uploaded\n"; |
b5c674c5 MM |
214 | } |
215 | } | |
216 | ||
7ebf9245 MM |
217 | sub fetch_file($$) |
218 | { | |
219 | my ($md5, $url) = @_; | |
220 | my $out = ""; | |
08ea99b9 | 221 | my $all_out = ""; |
01806192 MM |
222 | my $bn = basename($url); |
223 | my $local = "$tmp_dir/$md5/$bn"; | |
8f58f574 JB |
224 | my $cmd = "wget -nv --no-check-certificate --user-agent=$user_agent -O $local \"$url\""; |
225 | my $cmd2 = "wget -nv --no-check-certificate --user-agent=$user_agent --passive-ftp -O $local \"$url\""; | |
b5c674c5 | 226 | |
01806192 | 227 | push @files, $bn; |
44cd280f | 228 | |
01806192 | 229 | if (got_on_distfiles($md5, $url)) { |
576dff03 | 230 | $normal_out .= |
00eefa39 | 231 | "ALREADY GOT: $url\n" . |
01806192 | 232 | "\t$md5 " . basename($url) . "\n"; |
b5c674c5 | 233 | make_src_symlink($md5, $url); |
6ce39ee7 MM |
234 | return; |
235 | } | |
01806192 | 236 | |
2ae8fdec | 237 | mkdir("$tmp_dir/$md5"); |
576dff03 | 238 | |
b5c674c5 MM |
239 | if ($url =~ /^no-url/) { |
240 | handle_no_url($md5, $url); | |
241 | return; | |
242 | } | |
576dff03 | 243 | |
7ebf9245 MM |
244 | open(W, "$cmd 2>&1 |"); |
245 | while (<W>) { | |
08ea99b9 | 246 | $all_out .= $_; |
7ebf9245 MM |
247 | /URL:.*\s+\-\>\s+.*/ and next; |
248 | $out .= $_; | |
249 | } | |
250 | close(W); | |
251 | if ($out ne "") { | |
252 | $problems .= "$cmd:\n$out\n\n"; | |
253 | } | |
e23ae36a | 254 | if (-f $local && -s $local == 0 && $url =~ /^ftp:/) { |
f4f705aa | 255 | $out = ""; |
c0467b40 | 256 | open(W, "$cmd2 2>&1 |"); |
257 | while (<W>) { | |
f4f705aa | 258 | $all_out .= "\n\t\t$_"; |
c0467b40 | 259 | /URL:.*\s+\-\>\s+.*/ and next; |
260 | $out .= $_; | |
261 | } | |
262 | close(W); | |
263 | if ($out ne "") { | |
264 | $problems .= "$cmd:\n$out\n\n"; | |
265 | } | |
266 | } | |
4db63fc1 | 267 | if (-r $local && -s $local > 0) { |
01806192 | 268 | my $computed_md5 = md5($local); |
7ebf9245 MM |
269 | if ($computed_md5 ne $md5) { |
270 | $problems .= "FATAL: $url md5 mismatch, needed $md5, got $computed_md5\n"; | |
271 | } else { | |
d642cf7e | 272 | my $testcmd = "file \"$local\" |"; |
273 | my $testres = ""; | |
a14ed9b7 | 274 | if ($url =~ /^(http|https):/ && $local =~ /\.(tar\.(bz2|gz)|tgz|zip|jar|xpi)$/) { |
d642cf7e | 275 | open(T, $testcmd) or die; |
276 | $testres = <T>; | |
277 | close(T); | |
278 | } | |
279 | if ($testres =~ /empty|(ASCII|HTML|SGML).*text/) { | |
280 | $testres =~ s/.*://; | |
281 | $problems .= "FATAL: data returned from $url: $testres"; | |
282 | } else { | |
283 | move_file($md5, $url, $local); | |
284 | } | |
7ebf9245 | 285 | } |
4db63fc1 | 286 | } elsif (-f $local && -s $local > 0) { |
287 | $problems .= "FATAL: $url ($md5) was not fetched ($cmd: $all_out): file is not readable\n"; | |
7ebf9245 | 288 | } else { |
08ea99b9 | 289 | $problems .= "FATAL: $url ($md5) was not fetched ($cmd: $all_out)\n"; |
7ebf9245 | 290 | } |
01806192 MM |
291 | # save space |
292 | unlink($local); | |
7ebf9245 MM |
293 | } |
294 | ||
295 | sub fetch_files() | |
296 | { | |
297 | $problems .= "\n\n" if ($problems ne ""); | |
ccbaa374 | 298 | foreach $i (0..$#md5) { |
299 | fetch_file($md5[$i], $url[$i]); | |
7ebf9245 MM |
300 | } |
301 | } | |
302 | ||
303 | sub send_email() | |
304 | { | |
9a1e36f3 MM |
305 | open(EMAIL, "| /usr/sbin/sendmail -t"); |
306 | #open(EMAIL, "| cat"); | |
44cd280f | 307 | my $marker = ""; |
7ebf9245 | 308 | if ($problems ne "") { |
44cd280f | 309 | $marker = "ERRORS: "; |
7ebf9245 | 310 | } |
04550b88 MM |
311 | my $req_login; |
312 | $requester =~ /^(.*)\@/ and $req_login = $1; | |
44cd280f MM |
313 | |
314 | splice(@files, 10, @files - 10, "...") | |
315 | if (@files > 10); | |
576dff03 ER |
316 | |
317 | print EMAIL | |
04550b88 | 318 | "From: $req_login <$requester> |
482655f9 MM |
319 | To: $commits_list |
320 | Cc: $requester | |
44cd280f | 321 | Subject: DISTFILES: ${marker}@{files} |
e157986d MM |
322 | Message-ID: <$$." . time . "\@distfiles.pld-linux.org> |
323 | X-distfiles-program: file-fetcher.pl | |
324 | X-distfiles-version: " . '$Id$' . " | |
7ebf9245 MM |
325 | |
326 | $problems | |
47f8fda7 MM |
327 | Files fetched: $fetched_count |
328 | ||
7ebf9245 | 329 | $normal_out |
e157986d MM |
330 | |
331 | -- | |
d8cd573c | 332 | Virtually Yours: distfiles. |
7ebf9245 MM |
333 | "; |
334 | close(EMAIL) or die; | |
335 | } | |
336 | ||
01806192 MM |
337 | sub make_tmp_dir() |
338 | { | |
339 | my $id = `uuidgen 2>/dev/null`; | |
340 | chomp $id; | |
341 | $id = rand if (!defined $id or $id eq ""); | |
342 | $tmp_dir = "./tmp/$id"; | |
343 | mkdir($tmp_dir) or die; | |
344 | } | |
345 | ||
346 | sub clean_tmp_dir() | |
347 | { | |
348 | system("rm -rf $tmp_dir") | |
349 | if ($tmp_dir ne "" && -d $tmp_dir); | |
350 | } | |
351 | ||
2e33bbbc MM |
352 | umask(002); |
353 | ||
7ebf9245 MM |
354 | find_file_in_spool(); |
355 | read_spool_file(); | |
01806192 | 356 | make_tmp_dir(); |
7ebf9245 | 357 | fetch_files(); |
47f8fda7 | 358 | send_email() unless (!$force_reply and $problems eq "" and $fetched_count == 0); |
01806192 | 359 | clean_tmp_dir(); |