upstream/0.5.3/estwolefind

#! /bin/sh

#================================================================
# estwolels
# List the path of cache files of wwwoffle
#================================================================


# set variables
LANG=C ; export LANG
LC_ALL=C ; export LC_ALL
progname="estwolels"
spooldir=/var/spool/wwwoffle
denysufs="css|js|dtd|rdf|rss|swf|class|md5"
denysufs="$denysufs|png|gif|jpg|jpeg|jpe|bmp|tif|tiff"
denysufs="$denysufs|pnm|pbm|pgm|ppm|xbm|xpm|ps|eps|au|svg|dvi|ico"
denysufs="$denysufs|mid|midi|kar|mp3|mp2|au|snd|wav|aif|aiff"
denysufs="$denysufs|mpg|mpeg|mpe|qt|mov|avi"
denysufs="$denysufs|pdf|rtf|rtx|doc|xls|ppt|xdw|csv|tsv"
denysufs="$denysufs|gz|zip|bz2|lzh|lha|tar|bin|cpio|shar|jar|war"
sizemax="1048576"

# show help message
if [ "$1" = "--help" ]
then
  printf 'List the path and the URL of cache files of wwwoffle.\n'
  printf '\n'
  printf 'Usage:\n'
  printf '  %s [spooldir]\n' "$progname"
  printf '\n'
  exit 0
fi


# check the spool directory
if [ -n "$1" ]
then
  spooldir="$1"
fi
if [ -d "$spooldir/http" ] && [ -r "$spooldir/http" ] && [ -x "$spooldir/http" ]
then
  true
else
  printf '%s: cannot scan %s\n' "$progname" "$spooldir/http" 1>&2
  exit 1
fi


# list the path and the URL of cache files
ls "$spooldir/http" |
while read domain
do
  wwwoffle-ls "http://$domain"
done |
fgrep " http://" |
egrep -i -v "\.($denysufs)$" |
awk -F ' ' -v sdir="$spooldir/http" -v sizemax="$sizemax" '
{
  if($2 < 1 || $2 > sizemax) next
  file = $0
  sub(/ .*/, "", file)
  domain = $0
  sub(/.* http:\/\//, "", domain)
  sub(/\/.*/, "", domain)
  url = $0
  sub(/.* http:\/\//, "", url)
  printf("%s/%s/%s\thttp://%s\n", sdir, domain, file, url)
}
'


# exit normally
exit 0


# END OF FILE
1	#! /bin/sh
2
3	#================================================================
4	# estwolels
5	# List the path of cache files of wwwoffle
6	#================================================================
7
8
9	# set variables
10	LANG=C ; export LANG
11	LC_ALL=C ; export LC_ALL
12	progname="estwolels"
13	spooldir=/var/spool/wwwoffle
14	denysufs="css\|js\|dtd\|rdf\|rss\|swf\|class\|md5"
15	denysufs="$denysufs\|png\|gif\|jpg\|jpeg\|jpe\|bmp\|tif\|tiff"
16	denysufs="$denysufs\|pnm\|pbm\|pgm\|ppm\|xbm\|xpm\|ps\|eps\|au\|svg\|dvi\|ico"
17	denysufs="$denysufs\|mid\|midi\|kar\|mp3\|mp2\|au\|snd\|wav\|aif\|aiff"
18	denysufs="$denysufs\|mpg\|mpeg\|mpe\|qt\|mov\|avi"
19	denysufs="$denysufs\|pdf\|rtf\|rtx\|doc\|xls\|ppt\|xdw\|csv\|tsv"
20	denysufs="$denysufs\|gz\|zip\|bz2\|lzh\|lha\|tar\|bin\|cpio\|shar\|jar\|war"
21	sizemax="1048576"
22
23	# show help message
24	if [ "$1" = "--help" ]
25	then
26	printf 'List the path and the URL of cache files of wwwoffle.\n'
27	printf '\n'
28	printf 'Usage:\n'
29	printf ' %s [spooldir]\n' "$progname"
30	printf '\n'
31	exit 0
32	fi
33
34
35	# check the spool directory
36	if [ -n "$1" ]
37	then
38	spooldir="$1"
39	fi
40	if [ -d "$spooldir/http" ] && [ -r "$spooldir/http" ] && [ -x "$spooldir/http" ]
41	then
42	true
43	else
44	printf '%s: cannot scan %s\n' "$progname" "$spooldir/http" 1>&2
45	exit 1
46	fi
47
48
49	# list the path and the URL of cache files
50	ls "$spooldir/http" \|
51	while read domain
52	do
53	wwwoffle-ls "http://$domain"
54	done \|
55	fgrep " http://" \|
56	egrep -i -v "\.($denysufs)$" \|
57	awk -F ' ' -v sdir="$spooldir/http" -v sizemax="$sizemax" '
58	{
59	if($2 < 1 \|\| $2 > sizemax) next
60	file = $0
61	sub(/ .*/, "", file)
62	domain = $0
63	sub(/.* http:\/\//, "", domain)
64	sub(/\/.*/, "", domain)
65	url = $0
66	sub(/.* http:\/\//, "", url)
67	printf("%s/%s/%s\thttp://%s\n", sdir, domain, file, url)
68	}
69	'
70
71
72	# exit normally
73	exit 0
74
75
76
77	# END OF FILE