Mercurial > ec-dotfiles

annotate bin/moefetch @ 299:d5ac851f3225

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
[moefetch] User agent, referer.
author Edho Arief <edho@myconan.net>
date Thu, 09 Feb 2012 08:47:47 +0700
parents a872c40368dd
children 4879900244f7
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
221
e891b563b797 wrong rule caused mass headache
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 220
diff changeset
1 #!/bin/sh
148
edhoprima
parents:
diff changeset
2
edhoprima
parents:
diff changeset
3 # Copyright (c) 2009, edogawaconan <me@myconan.net>
edhoprima
parents:
diff changeset
4 #
edhoprima
parents:
diff changeset
5 # Permission to use, copy, modify, and/or distribute this software for any
edhoprima
parents:
diff changeset
6 # purpose with or without fee is hereby granted, provided that the above
edhoprima
parents:
diff changeset
7 # copyright notice and this permission notice appear in all copies.
edhoprima
parents:
diff changeset
8 #
edhoprima
parents:
diff changeset
9 # THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
edhoprima
parents:
diff changeset
10 # WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
edhoprima
parents:
diff changeset
11 # MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
edhoprima
parents:
diff changeset
12 # ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
edhoprima
parents:
diff changeset
13 # WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
edhoprima
parents:
diff changeset
14 # ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
edhoprima
parents:
diff changeset
15 # OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
16 #
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
17 # Lots of bugs here. Use with care
148
edhoprima
parents:
diff changeset
18 # USE WITH CARE
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
19 #
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
20 # what it does: fetch every picture that has the specified TAGS.
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
21 # requirement: wget, libxslt, openssl
148
edhoprima
parents:
diff changeset
22
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
23 # program additional paths for: cut, sed, wc, openssl, wget, xsltproc, grep
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
24 ADDITIONAL_PATH=
148
edhoprima
parents:
diff changeset
25
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
26 # default server address. Danbooru only! I do not take responsibility of stupidity.
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
27 DEFAULT_SITE="moe.imouto.org"
148
edhoprima
parents:
diff changeset
28
edhoprima
parents:
diff changeset
29 # base directory. make sure it's writeable. I do not take responsibility if you don't own the folder and files as no check is done for this one.
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
30 # Structure is ${BASE_DIR}/<TAGS>
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
31 # Absolute path only.
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
32 # Leave empty to use whatever folder you're running this at
193
ac6533a8fb51 - Documentation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 192
diff changeset
33 BASE_DIR=
148
edhoprima
parents:
diff changeset
34
edhoprima
parents:
diff changeset
35 # not user modifiable from here
edhoprima
parents:
diff changeset
36
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
37 # useless welcome message. Also version
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
38 msg_welcome() {
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
39 echo "moefetch ${_version}
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
40 Copyright (c) 2009 edogawaconan <me@myconan.net>
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
41 "
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
42 }
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
43
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
44 # Sanitize path. Totally safe. Usage: cmd "$(safe_path "${filename}")"
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
45 safe_path()
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
46 {
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
47 # It all depends on the first character.
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
48 start=$(printf "%s" "$*" | cut -c 1)
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
49 path=
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
50 case "${start}" in
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
51 .|/) path="$*";; # . and / is safe. No change.
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
52 *) path="./$*";; # Anything else must be prefixed with ./
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
53 esac
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
54 printf "%s" "${path}" # Return.
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
55 }
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
56
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
57 # Checks md5. OpenSSL should be available on anything usable.
230
e922fb1e858f - fixes on openssl output
edhoprima
parents: 229
diff changeset
58 get_md5() { cat "$(safe_path "${1}")" | openssl dgst -md5 | tail -n 1 | sed -e 's/.*\([[:xdigit:]]\{32\}\).*/\1/'; }
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
59
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
60 # Safely get basename.
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
61 get_basename() { basename "$(safe_path "${1}")"; }
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
62
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
63 # Safely get filename (basename without the extension).
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
64 get_filename() { get_basename "${1%.*}"; }
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
65
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
66 # Transformation for tag url.
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
67 get_cleantags() { printf "%s " "$*" | sed -e 's/\&/%26/g;s/=/%3D/g'; }
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
68
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
69 # Returns something if not an md5 value.
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
70 is_not_md5() { get_filename "$1" | sed -e 's/\([0-9a-f]\{32\}\)//g'; }
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
71
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
72
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
73 # fatal error handler
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
74 Err_Fatal() {
214
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
75 echo "
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
76 Fatal error: ${1}"
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
77 exit 1
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
78 }
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
79
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
80 Err_Impossible() {
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
81 echo "
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
82 Impossible error. Or you modified content of the working directories when the script is running.
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
83 Please report to moefetch.googlecode.com if you see this message (complete with entire run log)"
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
84 exit 1
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
85 }
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
86
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
87 # help message
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
88 Err_Help() {
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
89 echo "moefetch.sh COMMAND [-n] [-p PASSWORD] [-s SITE_URL] [-u USERNAME] TAGS
174
0948e76a57a1 added help. Bump to 0.1-beta2
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 173
diff changeset
90
176
3d2ae9417273 even more improvement
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 175
diff changeset
91 COMMAND:
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
92 (quick)fetch:
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
93 Do a complete update. Add prefix quick to skip file checking
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
94 check:
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
95 Get list of new files, clean up local folder and print total new files
175
5b7a154dbd21 cosmetics fix for help message
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 174
diff changeset
96
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
97 OPTIONS:
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
98 -n:
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
99 Skip checking repository directory.
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
100 -p PASSWORD:
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
101 Specifies password for login.
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
102 -s SITE_URL:
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
103 Specify URL of the Danbooru powered site you want to leech from. Default is ${DEFAULT_SITE}.
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
104 -u USERNAME:
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
105 Specifies username for login.
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
106 TAGS:
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
107 Tags you want to download. Separated by spaces. Tag name follows standard Danbooru tagging scheme."
193
ac6533a8fb51 - Documentation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 192
diff changeset
108 exit 2
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
109 }
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
110
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
111 # generate link by transforming xml
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
112 Generate_Link() {
214
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
113 echo "
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
114 Fetching XML file"
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
115 tempnum=1000
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
116 iternum=1
299
d5ac851f3225 [moefetch] User agent, referer.
Edho Arief <edho@myconan.net>
parents: 298
diff changeset
117 useragent="Mozilla/5.0 (Windows NT 6.1; WOW64; rv:10.0) Gecko/20100101 Firefox/10.0"
195
652d9e268cee test migration to printf
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 194
diff changeset
118 > "${TEMP_PREFIX}-list"
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
119 while [ "${tempnum}" -ge 1000 ]; do
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
120 url="http://${SITE}/post/index.xml?tags=$(get_cleantags "${TAGS}")&offset=0&limit=1000&page=${iternum}"
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
121 [ ${_use_login} -eq 1 ] && url="${url}&login=${LOGIN_USER}&password_hash=${LOGIN_PASS}"
299
d5ac851f3225 [moefetch] User agent, referer.
Edho Arief <edho@myconan.net>
parents: 298
diff changeset
122 wget --quiet "${url}" -O "${TEMP_PREFIX}-xml" --referer="http://${SITE}/post" --user-agent="${useragent}" -e continue=off || Err_Fatal "Failed download catalog file"
214
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
123 printf "Processing XML file... "
213
dd95cf01602c working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 212
diff changeset
124 # xslt evilry
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
125 xsltproc - "${TEMP_PREFIX}-xml" <<EOF | sed 's/.*\(http.*\)\(\/[a-f0-9]\{32\}\).*\.\([^\.]*\)/\1\2.\3/g' | grep ^http > "${TEMP_PREFIX}-templist"
148
edhoprima
parents:
diff changeset
126 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
edhoprima
parents:
diff changeset
127 <xsl:output method="xml" indent="yes"/>
edhoprima
parents:
diff changeset
128 <xsl:template match="post">
edhoprima
parents:
diff changeset
129 <xsl:value-of select="@file_url" />
edhoprima
parents:
diff changeset
130 </xsl:template>
edhoprima
parents:
diff changeset
131 </xsl:stylesheet>
edhoprima
parents:
diff changeset
132 EOF
235
649b7d4b056a Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents: 234
diff changeset
133 tempnum=$(grep -c . "${TEMP_PREFIX}-templist")
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
134 iternum=$((iternum + 1))
213
dd95cf01602c working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 212
diff changeset
135 cat "${TEMP_PREFIX}-templist" >> "${TEMP_PREFIX}-list"
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
136 echo "${tempnum} file(s) available"
213
dd95cf01602c working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 212
diff changeset
137 done
235
649b7d4b056a Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents: 234
diff changeset
138 numfiles=$(grep -c . "${TEMP_PREFIX}-list")
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
139 echo "${numfiles} file(s) available on server"
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
140 [ "${numfiles}" -gt 0 ] || Err_Fatal "Error in processing list or no files can be found with specified tag(s) or site."
214
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
141 }
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
142
148
edhoprima
parents:
diff changeset
143
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
144 progress_init() {
205
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
145 _last="-"
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
146 printf "${_last}"
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
147 }
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
148
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
149 progress_anim() {
205
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
150 case "${_last}" in
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
151 /) _last="-";;
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
152 -) _last=\\;;
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
153 \\) _last=\|;;
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
154 \|) _last="/";;
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
155 esac
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
156 printf "\b${_last}"
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
157 }
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
158
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
159 progress_done() { printf "\bdone\n"; }
205
2e866999c042 now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 204
diff changeset
160
200
8efa600ebfdb purge ls
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 199
diff changeset
161 # getting rid of ls (as per suggestion)
8efa600ebfdb purge ls
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 199
diff changeset
162 Count_Files() {
224
0ac1805621d4 fix for FreeBSD
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 223
diff changeset
163 numfiles=0
0ac1805621d4 fix for FreeBSD
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 223
diff changeset
164 for dircontent in "${*}/"* "${*}/".*; do
251
d7e5a2e70cf3 Proper test for for loop (*, .*)
Edho Arief <edho@myconan.net>
parents: 236
diff changeset
165 if [ -e "${dircontent}" ] && [ x"${dircontent}" != x"${*}/." ] && [ x"${dircontent}" != x"${*}/.." ]; then
230
e922fb1e858f - fixes on openssl output
edhoprima
parents: 229
diff changeset
166 numfiles=$((numfiles + 1))
200
8efa600ebfdb purge ls
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 199
diff changeset
167 fi
8efa600ebfdb purge ls
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 199
diff changeset
168 done
230
e922fb1e858f - fixes on openssl output
edhoprima
parents: 229
diff changeset
169 echo $((numfiles - 2))
200
8efa600ebfdb purge ls
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 199
diff changeset
170 }
8efa600ebfdb purge ls
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 199
diff changeset
171
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
172 # check tools availability
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
173 Check_Tools() {
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
174 # verify all programs required do indeed exist
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
175 commands="cut sed wc wget xsltproc xargs rm mkdir chown comm grep date openssl"
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
176 for cmd in ${commands}
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
177 do
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
178 [ "$(command -v "${cmd}")" ] || Err_Fatal "${cmd} doesn't exist in ${PATH}"
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
179 done
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
180 }
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
181
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
182 # verify required folders exist and writeable
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
183 Check_Folders(){
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
184 [ -O "${BASE_DIR}" ] || Err_Fatal "You don't own ${BASE_DIR}. Please fix ${BASE_DIR} or run this script in your own directory."
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
185 for directory in temp trash deleted "${SITE_DIR}/${TARGET_DIR}"; do
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
186 if [ ! -d "${BASE_DIR}/${directory}" ]; then
216
a869987c4646 did I say 'mess up'?
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 215
diff changeset
187 mkdir -p "${BASE_DIR}/${directory}" || Err_Impossible
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
188 fi
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
189 if [ ! -O "${BASE_DIR}/${directory}" ]; then
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
190 echo "You don't own the ${BASE_DIR}/${directory}, applying globally writeable permission on it"
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
191 chmod -R u=rwX,g=rwX,o=rwX "${BASE_DIR}/${directory}" || Err_Impossible
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
192 fi
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
193 done
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
194 [ "$(Count_Files "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}")" -eq 0 ] && ISNEW=1
201
30d2fb656029 scrapping grep -vf
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 200
diff changeset
195 for i in error ok list newlist templist; do
196
4d28f3a957ee cleanup~
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 195
diff changeset
196 touch "${TEMP_PREFIX}-${i}" || Fatal_Err "Error creating ${TEMP_PREFIX}-${i}. This shouldn't happen"
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
197 done
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
198 #
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
199 }
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
200
187
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
201 # Do some cleanup
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
202 Cleanup_Repository() {
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
203 # current dir: ${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}
207
17d816a63b4c final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 206
diff changeset
204 printf "Cleaning up repository folder... "
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
205 progress_init
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
206 trash_dir="${BASE_DIR}/trash/${trash_dir}/$(date -u "+${SITE_DIR}-${TARGET_DIR}-%Y%m%d-%H.%M")"
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
207 trashes="These files have been moved to ${trash_dir}:"
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
208 has_trash=
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
209 if [ ! -d "${trash_dir}" ]; then
216
a869987c4646 did I say 'mess up'?
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 215
diff changeset
210 mkdir -p "${trash_dir}" || Err_Impossible
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
211 else
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
212 if [ ! -O "${trash_dir}" ]; then
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
213 chmod -R u=rwX,g=rwX,o=rwX "${BASE_DIR}/${directory}" || Err_Impossible
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
214 fi
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
215 fi
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
216 for trash in "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}/"*
187
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
217 do
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
218 is_trash=
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
219 if [ -d "${trash}" ] || [ -n "$(is_not_md5 "${trash}")" ] || [ -z "$(grep "$(get_basename "${trash}")" "${TEMP_PREFIX}-list")" ]; then
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
220 is_trash=1
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
221 has_trash=1
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
222 mv -f -- "${trash}" "${trash_dir}" || Err_Impossible
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
223 trashes="${trashes}
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
224 $(get_basename "${trash}")"
187
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
225 fi
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
226 progress_anim
187
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
227 done
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
228 rmdir "${trash_dir}" 2>/dev/null
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
229 progress_done
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
230 [ -n "${has_trash}" ] && echo "${trashes}"
187
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
231 }
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
232
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
233 # check files correctness
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
234 Check_Files() {
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
235 if [ ! -n "${ISNEW}" ]; then
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
236 [ -z "${NOCLEAN}" ] && Cleanup_Repository
207
17d816a63b4c final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 206
diff changeset
237 printf "Checking for errors... "
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
238 progress_init
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
239 files_error="These files do not match its md5:"
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
240 files_notdanbooru="These files are not checked:"
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
241 has_err_filename=
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
242 has_err_md5=
196
4d28f3a957ee cleanup~
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 195
diff changeset
243 > "${TEMP_PREFIX}-error"
214
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
244 > "${TEMP_PREFIX}-ok"
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
245 for file in "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}/"*
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
246 do
251
d7e5a2e70cf3 Proper test for for loop (*, .*)
Edho Arief <edho@myconan.net>
parents: 236
diff changeset
247 if [ -e "${file}" ]; then
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
248 if [ -n "$(is_not_md5 "${file}")" ] || [ -d "${file}" ]; then
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
249 files_notdanbooru="${files_notdanbooru}
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
250 $(get_basename "${file}")"
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
251 has_err_filename=1
214
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
252 else
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
253 if [ "$(get_md5 "${file}")" = "$(get_filename "${file}")" ]; then
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
254 echo "$(get_basename "${file}")" >> "${TEMP_PREFIX}-ok"
217
77cd21d714f6 screwup fix
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 216
diff changeset
255 else
77cd21d714f6 screwup fix
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 216
diff changeset
256 rm "${file}" || Err_Fatal "Error removing ${file}"
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
257 echo "$(get_basename "${file}")" >> "${TEMP_PREFIX}-error"
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
258 files_error="${files_error}
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
259 $(get_basename "${file}")"
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
260 has_err_md5=1
217
77cd21d714f6 screwup fix
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 216
diff changeset
261 fi
187
efd957294c8c refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 186
diff changeset
262 fi
148
edhoprima
parents:
diff changeset
263 fi
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
264 progress_anim
148
edhoprima
parents:
diff changeset
265 done
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
266 progress_done
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
267 if [ ! -n "${has_err_md5}" ] && [ ! -n "${has_err_filename}" ]; then
203
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 202
diff changeset
268 echo "All files OK"
170
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 169
diff changeset
269 else
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
270 if [ -n "${has_err_md5}" ]; then
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
271 echo "${files_error}"
235
649b7d4b056a Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents: 234
diff changeset
272 echo "$(grep -c . "${TEMP_PREFIX}-error") file(s) removed"
214
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
273 fi
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
274 [ -n "${has_err_filename}" ] && echo "${files_notdanbooru}"
170
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 169
diff changeset
275 fi
235
649b7d4b056a Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents: 234
diff changeset
276 echo "$(grep -c . "${TEMP_PREFIX}-ok") file(s) available locally"
148
edhoprima
parents:
diff changeset
277
207
17d816a63b4c final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 206
diff changeset
278 printf "Generating list of new files... "
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
279 progress_init
218
aeca29670e26 broken!
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 217
diff changeset
280 cp -f "${TEMP_PREFIX}-list" "${TEMP_PREFIX}-templist"
214
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
281 while read -r is_ok; do
a6624fb9b317 major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 213
diff changeset
282 grep -v "${is_ok}" "${TEMP_PREFIX}-templist" > "${TEMP_PREFIX}-newlist"
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
283 cp -f "${TEMP_PREFIX}-newlist" "${TEMP_PREFIX}-templist" || Err_Impossible
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
284 progress_anim
203
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 202
diff changeset
285 done < "${TEMP_PREFIX}-ok"
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
286 progress_done
235
649b7d4b056a Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents: 234
diff changeset
287 echo "$(grep -c . "${TEMP_PREFIX}-newlist") file(s) to be downloaded"
148
edhoprima
parents:
diff changeset
288 else
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
289 if [ -n "${ISQUICK}" ]; then
207
17d816a63b4c final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 206
diff changeset
290 echo "Quick mode selected. Skipping check"
152
67df02877319 added quickfetch for skipping file checking
edhoprima
parents: 151
diff changeset
291 else
67df02877319 added quickfetch for skipping file checking
edhoprima
parents: 151
diff changeset
292 echo "Empty local repository"
67df02877319 added quickfetch for skipping file checking
edhoprima
parents: 151
diff changeset
293 fi
200
8efa600ebfdb purge ls
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 199
diff changeset
294 cat "${TEMP_PREFIX}-list" > "${TEMP_PREFIX}-newlist"
148
edhoprima
parents:
diff changeset
295 fi
edhoprima
parents:
diff changeset
296 }
edhoprima
parents:
diff changeset
297
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
298 # start downloading the images
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
299 Fetch_Images() {
235
649b7d4b056a Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents: 234
diff changeset
300 if [ "$(grep -c . "${TEMP_PREFIX}-newlist")" -eq 0 ]; then
148
edhoprima
parents:
diff changeset
301 echo "No new file"
edhoprima
parents:
diff changeset
302 else
231
4c0fd276665e - for some reason I broke the getopts logic again. Fixed
edhoprima
parents: 230
diff changeset
303 printf "Downloading files... "
160
68227a30d0b3 forgot to fix Fetch_Images to reflect new folder naming scheme
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 159
diff changeset
304 cd "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}"
231
4c0fd276665e - for some reason I broke the getopts logic again. Fixed
edhoprima
parents: 230
diff changeset
305 wget -e continue=on -i "${TEMP_PREFIX}-newlist" -o "${TEMP_PREFIX}.log"
148
edhoprima
parents:
diff changeset
306 fi
edhoprima
parents:
diff changeset
307 }
edhoprima
parents:
diff changeset
308
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
309 # initialize base variables and initial command check
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
310 init()
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
311 {
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
312 # path initialization
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
313 # check if additional path is specified
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
314 if [ -n "${ADDITIONAL_PATH}" ]
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
315 then
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
316 # insert the additional path
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
317 PATH="${ADDITIONAL_PATH}:${PATH}"
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
318 export PATH
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
319 fi
158
cba73f6a96bb grep check. OpenSolaris' default grep doesn't support -f
edhoprima
parents: 157
diff changeset
320
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
321 # misc variables
166
cc60e8cf7793 dunno :<
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 165
diff changeset
322 ISQUICK=
cc60e8cf7793 dunno :<
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 165
diff changeset
323 ISNEW=
215
710082ce6788 major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 214
diff changeset
324
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
325 # minimum number of arguments: 2 (command and tag). If less than two, exit and print help message
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
326 [ $# -lt 2 ] && Err_Help
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
327 case "$1" in
174
0948e76a57a1 added help. Bump to 0.1-beta2
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 173
diff changeset
328 check|fetch|quickfetch)
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
329 echo "Starting..."
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
330 JOB="$1"
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
331 ;;
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
332 *)
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
333 Err_Help
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
334 ;;
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
335 esac
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
336 shift
230
e922fb1e858f - fixes on openssl output
edhoprima
parents: 229
diff changeset
337 SITE=
e922fb1e858f - fixes on openssl output
edhoprima
parents: 229
diff changeset
338 TAGS=
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
339 has_pass=0
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
340 has_user=0
231
4c0fd276665e - for some reason I broke the getopts logic again. Fixed
edhoprima
parents: 230
diff changeset
341 x=1
4c0fd276665e - for some reason I broke the getopts logic again. Fixed
edhoprima
parents: 230
diff changeset
342 while getopts "s:nu:p:" opt
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
343 do
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
344 case "$opt" in
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
345 s) SITE="$OPTARG";;
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
346 n) NOCLEAN=1;;
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
347 p)
234
58ad057cd2ec - Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents: 232
diff changeset
348 LOGIN_PASS=$(printf "%s" "$OPTARG" | openssl dgst -sha1 | sed -e 's/.*\([[:xdigit:]]\{40\}\).*/\1/')
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
349 has_pass=1
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
350 ;;
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
351 u)
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
352 LOGIN_USER="$OPTARG"
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
353 has_user=1
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
354 ;;
185
6d926d4b3c5a initial clean system support
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 184
diff changeset
355 esac
231
4c0fd276665e - for some reason I broke the getopts logic again. Fixed
edhoprima
parents: 230
diff changeset
356 x=$OPTIND
185
6d926d4b3c5a initial clean system support
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 184
diff changeset
357 done
231
4c0fd276665e - for some reason I broke the getopts logic again. Fixed
edhoprima
parents: 230
diff changeset
358 shift $(($x-1))
4c0fd276665e - for some reason I broke the getopts logic again. Fixed
edhoprima
parents: 230
diff changeset
359 if [ "$1" = -- ]; then shift; fi
225
265a9ca47a19 - Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents: 224
diff changeset
360 TAGS="$@"
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
361 [ -n "${SITE}" ] || SITE=${DEFAULT_SITE}
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
362 [ -n "${TAGS}" ] || Err_Fatal "No tag specified"
181
d3b7927bdb2b restructuring and add check if the xml is processed properly
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 180
diff changeset
363 # Get base folder - default, current folder or fallback to ${HOME}
223
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
364 [ -n "${BASE_DIR}" ] || BASE_DIR=${PWD}
04ad0b0a3c63 revert back to [
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 222
diff changeset
365 [ -n "${BASE_DIR}" ] || BASE_DIR=${HOME}
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
366 [ -n "$(echo "${BASE_DIR}" | cut -c1 | grep \/)" ] || BASE_DIR="/${BASE_DIR}"
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
367 # see if both pass and use are set. If they're set, switch _use_login variable content to 1.
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
368 [ ${has_pass} -eq 1 -a ${has_user} -eq 1 ] && _use_login=1
181
d3b7927bdb2b restructuring and add check if the xml is processed properly
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 180
diff changeset
369
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
370 echo "Tags: ${TAGS}"
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
371 # slash is not wanted for folder name
193
ac6533a8fb51 - Documentation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 192
diff changeset
372 TARGET_DIR=$(echo "${TAGS}" | sed -e 's/\//_/g')
ac6533a8fb51 - Documentation
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 192
diff changeset
373 SITE_DIR=$(echo "${SITE}" | sed -e 's/\/$//g;s/\//_/g')
195
652d9e268cee test migration to printf
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 194
diff changeset
374 TEMP_PREFIX="${BASE_DIR}/temp/${SITE_DIR}-${TARGET_DIR}"
159
75fe19903b74 Major cleanup
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 158
diff changeset
375 }
148
edhoprima
parents:
diff changeset
376
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
377 # global variables goes here
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
378 init_globals()
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
379 {
232
5438d80244a3 - version bump
edhoprima
parents: 231
diff changeset
380 _version="1.0-rc2" # version of this script
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
381 _use_login=0 # variable to check whether a login is used or not
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
382 }
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
383
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
384 main()
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
385 {
234
58ad057cd2ec - Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents: 232
diff changeset
386 # removing GNU-ism as much as possible
58ad057cd2ec - Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents: 232
diff changeset
387 POSIXLY_CORRECT=1
228
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
388 #initialize global variables
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
389 init_globals
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
390 #print welcome message
5d3a0645b504 - Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 227
diff changeset
391 msg_welcome
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
392 # initialization
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
393 init "$@"
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
394 Check_Tools
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
395 Check_Folders
158
cba73f6a96bb grep check. OpenSolaris' default grep doesn't support -f
edhoprima
parents: 157
diff changeset
396
148
edhoprima
parents:
diff changeset
397
227
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
398 # let's do the job!
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
399 case "${JOB}" in
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
400 check)
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
401 Generate_Link
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
402 Check_Files
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
403 ;;
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
404 fetch)
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
405 Generate_Link
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
406 Check_Files
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
407 Fetch_Images
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
408 ;;
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
409 quickfetch)
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
410 ISNEW=1
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
411 ISQUICK=1
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
412 Generate_Link
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
413 Check_Files
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
414 Fetch_Images
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
415 ;;
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
416 esac
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
417 }
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
418
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
419 # call the main routine!
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
420 main "$@"
8b1f6f6b6a3b Bugfixes:
edhoprima@gmail.com <edhoprima@gmail.com>
parents: 226
diff changeset
421