Mercurial > ec-dotfiles
annotate bin/moefetch @ 327:c9d8fbccc1d4
[ren] More rules.
author | Edho Arief <edho@myconan.net> |
---|---|
date | Wed, 07 Mar 2012 13:16:10 +0700 |
parents | 21b86001b0c5 |
children | 110d50856dde |
rev | line source |
---|---|
221
e891b563b797
wrong rule caused mass headache
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
220
diff
changeset
|
1 #!/bin/sh |
148 | 2 |
301
36bc27bb32ff
Updated version and copyright.
Edho Arief <edho@myconan.net>
parents:
300
diff
changeset
|
3 # Copyright (c) 2009-2012, edogawaconan <edho@myconan.net> |
148 | 4 # |
5 # Permission to use, copy, modify, and/or distribute this software for any | |
6 # purpose with or without fee is hereby granted, provided that the above | |
7 # copyright notice and this permission notice appear in all copies. | |
8 # | |
9 # THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES | |
10 # WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF | |
11 # MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR | |
12 # ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES | |
13 # WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN | |
14 # ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF | |
15 # OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. | |
159 | 16 # |
17 # Lots of bugs here. Use with care | |
148 | 18 # USE WITH CARE |
159 | 19 # |
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 | 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 | 24 ADDITIONAL_PATH= |
148 | 25 |
159 | 26 # default server address. Danbooru only! I do not take responsibility of stupidity. |
27 DEFAULT_SITE="moe.imouto.org" | |
148 | 28 |
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 | 30 # Structure is ${BASE_DIR}/<TAGS> |
31 # Absolute path only. | |
32 # Leave empty to use whatever folder you're running this at | |
193 | 33 BASE_DIR= |
148 | 34 |
35 # not user modifiable from here | |
36 | |
305
21b86001b0c5
[moefetch] Added basic safeguard
Edho Arief <edho@myconan.net>
parents:
302
diff
changeset
|
37 # stop on any error |
21b86001b0c5
[moefetch] Added basic safeguard
Edho Arief <edho@myconan.net>
parents:
302
diff
changeset
|
38 set -e |
21b86001b0c5
[moefetch] Added basic safeguard
Edho Arief <edho@myconan.net>
parents:
302
diff
changeset
|
39 # ensures all variables initialized |
21b86001b0c5
[moefetch] Added basic safeguard
Edho Arief <edho@myconan.net>
parents:
302
diff
changeset
|
40 set -u |
302 | 41 useragent="Mozilla/5.0 (Windows NT 6.1; WOW64; rv:10.0) Gecko/20100101 Firefox/10.0" |
42 | |
159 | 43 # useless welcome message. Also version |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
44 msg_welcome() { |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
45 echo "moefetch ${_version} |
301
36bc27bb32ff
Updated version and copyright.
Edho Arief <edho@myconan.net>
parents:
300
diff
changeset
|
46 Copyright (c) 2009-2012 edogawaconan <edho@myconan.net> |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
47 " |
159 | 48 } |
49 | |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
50 # Sanitize path. Totally safe. Usage: cmd "$(safe_path "${filename}")" |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
51 safe_path() |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
52 { |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
53 # It all depends on the first character. |
232 | 54 start=$(printf "%s" "$*" | cut -c 1) |
55 path= | |
56 case "${start}" in | |
57 .|/) path="$*";; # . and / is safe. No change. | |
58 *) path="./$*";; # Anything else must be prefixed with ./ | |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
59 esac |
232 | 60 printf "%s" "${path}" # Return. |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
61 } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
62 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
63 # Checks md5. OpenSSL should be available on anything usable. |
230 | 64 get_md5() { cat "$(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
|
65 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
66 # Safely get basename. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
67 get_basename() { basename "$(safe_path "${1}")"; } |
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 # 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
|
70 get_filename() { get_basename "${1%.*}"; } |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
71 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
72 # Transformation for tag url. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
73 get_cleantags() { printf "%s " "$*" | sed -e 's/\&/%26/g;s/=/%3D/g'; } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
74 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
75 # Returns something if not an md5 value. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
76 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
|
77 |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
78 |
159 | 79 # fatal error handler |
80 Err_Fatal() { | |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
81 echo " |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
82 Fatal error: ${1}" |
159 | 83 exit 1 |
84 } | |
85 | |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
86 Err_Impossible() { |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
87 echo " |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
88 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
|
89 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
|
90 exit 1 |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
91 } |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
92 |
159 | 93 # help message |
94 Err_Help() { | |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
95 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
|
96 |
176
3d2ae9417273
even more improvement
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
175
diff
changeset
|
97 COMMAND: |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
98 (quick)fetch: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
99 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
|
100 check: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
101 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
|
102 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
103 OPTIONS: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
104 -n: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
105 Skip checking repository directory. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
106 -p PASSWORD: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
107 Specifies password for login. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
108 -s SITE_URL: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
109 Specify URL of the Danbooru powered site you want to leech from. Default is ${DEFAULT_SITE}. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
110 -u USERNAME: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
111 Specifies username for login. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
112 TAGS: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
113 Tags you want to download. Separated by spaces. Tag name follows standard Danbooru tagging scheme." |
193 | 114 exit 2 |
159 | 115 } |
116 | |
117 # generate link by transforming xml | |
118 Generate_Link() { | |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
119 echo " |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
120 Fetching XML file" |
232 | 121 tempnum=1000 |
122 iternum=1 | |
195
652d9e268cee
test migration to printf
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
194
diff
changeset
|
123 > "${TEMP_PREFIX}-list" |
232 | 124 while [ "${tempnum}" -ge 1000 ]; do |
125 url="http://${SITE}/post/index.xml?tags=$(get_cleantags "${TAGS}")&offset=0&limit=1000&page=${iternum}" | |
126 [ ${_use_login} -eq 1 ] && url="${url}&login=${LOGIN_USER}&password_hash=${LOGIN_PASS}" | |
299
d5ac851f3225
[moefetch] User agent, referer.
Edho Arief <edho@myconan.net>
parents:
298
diff
changeset
|
127 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
|
128 printf "Processing XML file... " |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
129 # xslt evilry |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
130 xsltproc - "${TEMP_PREFIX}-xml" <<EOF | sed 's/.*\(http.*\)\(\/[a-f0-9]\{32\}\).*\.\([^\.]*\)/\1\2.\3/g' | grep ^http > "${TEMP_PREFIX}-templist" |
148 | 131 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0"> |
132 <xsl:output method="xml" indent="yes"/> | |
133 <xsl:template match="post"> | |
134 <xsl:value-of select="@file_url" /> | |
135 </xsl:template> | |
136 </xsl:stylesheet> | |
137 EOF | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
138 tempnum=$(grep -c . "${TEMP_PREFIX}-templist") |
232 | 139 iternum=$((iternum + 1)) |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
140 cat "${TEMP_PREFIX}-templist" >> "${TEMP_PREFIX}-list" |
232 | 141 echo "${tempnum} file(s) available" |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
142 done |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
143 numfiles=$(grep -c . "${TEMP_PREFIX}-list") |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
144 echo "${numfiles} file(s) available on server" |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
145 [ "${numfiles}" -gt 0 ] || Err_Fatal "Error in processing list or no files can be found with specified tag(s) or site." |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
146 } |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
147 |
148 | 148 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
149 progress_init() { |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
150 _last="-" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
151 printf "${_last}" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
152 } |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
153 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
154 progress_anim() { |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
155 case "${_last}" in |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
156 /) _last="-";; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
157 -) _last=\\;; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
158 \\) _last=\|;; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
159 \|) _last="/";; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
160 esac |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
161 printf "\b${_last}" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
162 } |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
163 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
164 progress_done() { printf "\bdone\n"; } |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
165 |
200 | 166 # getting rid of ls (as per suggestion) |
167 Count_Files() { | |
224 | 168 numfiles=0 |
169 for dircontent in "${*}/"* "${*}/".*; do | |
251
d7e5a2e70cf3
Proper test for for loop (*, .*)
Edho Arief <edho@myconan.net>
parents:
236
diff
changeset
|
170 if [ -e "${dircontent}" ] && [ x"${dircontent}" != x"${*}/." ] && [ x"${dircontent}" != x"${*}/.." ]; then |
230 | 171 numfiles=$((numfiles + 1)) |
200 | 172 fi |
173 done | |
300 | 174 echo $((numfiles)) |
200 | 175 } |
176 | |
159 | 177 # check tools availability |
178 Check_Tools() { | |
179 # 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
|
180 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
|
181 for cmd in ${commands} |
159 | 182 do |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
183 [ "$(command -v "${cmd}")" ] || Err_Fatal "${cmd} doesn't exist in ${PATH}" |
159 | 184 done |
185 } | |
186 | |
187 # verify required folders exist and writeable | |
188 Check_Folders(){ | |
223 | 189 [ -O "${BASE_DIR}" ] || Err_Fatal "You don't own ${BASE_DIR}. Please fix ${BASE_DIR} or run this script in your own directory." |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
190 for directory in temp trash deleted "${SITE_DIR}/${TARGET_DIR}"; do |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
191 if [ ! -d "${BASE_DIR}/${directory}" ]; then |
216
a869987c4646
did I say 'mess up'?
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
215
diff
changeset
|
192 mkdir -p "${BASE_DIR}/${directory}" || Err_Impossible |
159 | 193 fi |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
194 if [ ! -O "${BASE_DIR}/${directory}" ]; then |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
195 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
|
196 chmod -R u=rwX,g=rwX,o=rwX "${BASE_DIR}/${directory}" || Err_Impossible |
159 | 197 fi |
198 done | |
223 | 199 [ "$(Count_Files "${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
|
200 for i in error ok list newlist templist; do |
196 | 201 touch "${TEMP_PREFIX}-${i}" || Fatal_Err "Error creating ${TEMP_PREFIX}-${i}. This shouldn't happen" |
159 | 202 done |
203 # | |
204 } | |
205 | |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
206 # Do some cleanup |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
207 Cleanup_Repository() { |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
208 # current dir: ${BASE_DIR}/${SITE_DIR}/${TARGET_DIR} |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
209 printf "Cleaning up repository folder... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
210 progress_init |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
211 trash_dir="${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
|
212 trashes="These files have been moved to ${trash_dir}:" |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
213 has_trash= |
223 | 214 if [ ! -d "${trash_dir}" ]; then |
216
a869987c4646
did I say 'mess up'?
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
215
diff
changeset
|
215 mkdir -p "${trash_dir}" || Err_Impossible |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
216 else |
223 | 217 if [ ! -O "${trash_dir}" ]; then |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
218 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
|
219 fi |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
220 fi |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
221 for trash in "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}/"* |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
222 do |
300 | 223 if [ -e "${trash}" ]; then |
224 is_trash= | |
225 if [ -d "${trash}" ] || [ -n "$(is_not_md5 "${trash}")" ] || [ -z "$(grep "$(get_basename "${trash}")" "${TEMP_PREFIX}-list")" ]; then | |
226 is_trash=1 | |
227 has_trash=1 | |
228 mv -f -- "${trash}" "${trash_dir}" || Err_Impossible | |
229 trashes="${trashes} | |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
230 $(get_basename "${trash}")" |
300 | 231 fi |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
232 fi |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
233 progress_anim |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
234 done |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
235 rmdir "${trash_dir}" 2>/dev/null |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
236 progress_done |
223 | 237 [ -n "${has_trash}" ] && echo "${trashes}" |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
238 } |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
239 |
159 | 240 # check files correctness |
241 Check_Files() { | |
223 | 242 if [ ! -n "${ISNEW}" ]; then |
243 [ -z "${NOCLEAN}" ] && Cleanup_Repository | |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
244 printf "Checking for errors... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
245 progress_init |
232 | 246 files_error="These files do not match its md5:" |
247 files_notdanbooru="These files are not checked:" | |
248 has_err_filename= | |
249 has_err_md5= | |
196 | 250 > "${TEMP_PREFIX}-error" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
251 > "${TEMP_PREFIX}-ok" |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
252 for file in "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}/"* |
159 | 253 do |
251
d7e5a2e70cf3
Proper test for for loop (*, .*)
Edho Arief <edho@myconan.net>
parents:
236
diff
changeset
|
254 if [ -e "${file}" ]; then |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
255 if [ -n "$(is_not_md5 "${file}")" ] || [ -d "${file}" ]; then |
232 | 256 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
|
257 $(get_basename "${file}")" |
232 | 258 has_err_filename=1 |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
259 else |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
260 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
|
261 echo "$(get_basename "${file}")" >> "${TEMP_PREFIX}-ok" |
217 | 262 else |
263 rm "${file}" || Err_Fatal "Error removing ${file}" | |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
264 echo "$(get_basename "${file}")" >> "${TEMP_PREFIX}-error" |
232 | 265 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
|
266 $(get_basename "${file}")" |
232 | 267 has_err_md5=1 |
217 | 268 fi |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
269 fi |
148 | 270 fi |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
271 progress_anim |
148 | 272 done |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
273 progress_done |
232 | 274 if [ ! -n "${has_err_md5}" ] && [ ! -n "${has_err_filename}" ]; then |
203 | 275 echo "All files OK" |
170 | 276 else |
232 | 277 if [ -n "${has_err_md5}" ]; then |
278 echo "${files_error}" | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
279 echo "$(grep -c . "${TEMP_PREFIX}-error") file(s) removed" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
280 fi |
232 | 281 [ -n "${has_err_filename}" ] && echo "${files_notdanbooru}" |
170 | 282 fi |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
283 echo "$(grep -c . "${TEMP_PREFIX}-ok") file(s) available locally" |
148 | 284 |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
285 printf "Generating list of new files... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
286 progress_init |
218 | 287 cp -f "${TEMP_PREFIX}-list" "${TEMP_PREFIX}-templist" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
288 while read -r is_ok; do |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
289 grep -v "${is_ok}" "${TEMP_PREFIX}-templist" > "${TEMP_PREFIX}-newlist" |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
290 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
|
291 progress_anim |
203 | 292 done < "${TEMP_PREFIX}-ok" |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
293 progress_done |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
294 echo "$(grep -c . "${TEMP_PREFIX}-newlist") file(s) to be downloaded" |
148 | 295 else |
223 | 296 if [ -n "${ISQUICK}" ]; then |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
297 echo "Quick mode selected. Skipping check" |
152 | 298 else |
299 echo "Empty local repository" | |
300 fi | |
200 | 301 cat "${TEMP_PREFIX}-list" > "${TEMP_PREFIX}-newlist" |
148 | 302 fi |
303 } | |
304 | |
159 | 305 # start downloading the images |
306 Fetch_Images() { | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
307 if [ "$(grep -c . "${TEMP_PREFIX}-newlist")" -eq 0 ]; then |
148 | 308 echo "No new file" |
309 else | |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
310 printf "Downloading files... " |
160
68227a30d0b3
forgot to fix Fetch_Images to reflect new folder naming scheme
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
159
diff
changeset
|
311 cd "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}" |
302 | 312 wget -e continue=on -i "${TEMP_PREFIX}-newlist" -o "${TEMP_PREFIX}.log" --referer="http://${SITE}/post" --user-agent="${useragent}" |
148 | 313 fi |
314 } | |
315 | |
159 | 316 # initialize base variables and initial command check |
227 | 317 init() |
318 { | |
159 | 319 # path initialization |
227 | 320 # check if additional path is specified |
321 if [ -n "${ADDITIONAL_PATH}" ] | |
322 then | |
323 # insert the additional path | |
324 PATH="${ADDITIONAL_PATH}:${PATH}" | |
325 export PATH | |
326 fi | |
158
cba73f6a96bb
grep check. OpenSolaris' default grep doesn't support -f
edhoprima
parents:
157
diff
changeset
|
327 |
159 | 328 # misc variables |
166 | 329 ISQUICK= |
330 ISNEW= | |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
331 |
227 | 332 # minimum number of arguments: 2 (command and tag). If less than two, exit and print help message |
159 | 333 [ $# -lt 2 ] && Err_Help |
334 case "$1" in | |
174
0948e76a57a1
added help. Bump to 0.1-beta2
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
173
diff
changeset
|
335 check|fetch|quickfetch) |
159 | 336 echo "Starting..." |
337 JOB="$1" | |
338 ;; | |
339 *) | |
340 Err_Help | |
341 ;; | |
342 esac | |
343 shift | |
230 | 344 SITE= |
345 TAGS= | |
232 | 346 has_pass=0 |
347 has_user=0 | |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
348 x=1 |
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
349 while getopts "s:nu:p:" opt |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
350 do |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
351 case "$opt" in |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
352 s) SITE="$OPTARG";; |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
353 n) NOCLEAN=1;; |
227 | 354 p) |
234
58ad057cd2ec
- Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents:
232
diff
changeset
|
355 LOGIN_PASS=$(printf "%s" "$OPTARG" | openssl dgst -sha1 | sed -e 's/.*\([[:xdigit:]]\{40\}\).*/\1/') |
232 | 356 has_pass=1 |
227 | 357 ;; |
358 u) | |
359 LOGIN_USER="$OPTARG" | |
232 | 360 has_user=1 |
227 | 361 ;; |
185
6d926d4b3c5a
initial clean system support
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
184
diff
changeset
|
362 esac |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
363 x=$OPTIND |
185
6d926d4b3c5a
initial clean system support
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
184
diff
changeset
|
364 done |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
365 shift $(($x-1)) |
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
366 if [ "$1" = -- ]; then shift; fi |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
367 TAGS="$@" |
223 | 368 [ -n "${SITE}" ] || SITE=${DEFAULT_SITE} |
369 [ -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
|
370 # Get base folder - default, current folder or fallback to ${HOME} |
223 | 371 [ -n "${BASE_DIR}" ] || BASE_DIR=${PWD} |
372 [ -n "${BASE_DIR}" ] || BASE_DIR=${HOME} | |
227 | 373 [ -n "$(echo "${BASE_DIR}" | cut -c1 | grep \/)" ] || BASE_DIR="/${BASE_DIR}" |
374 # see if both pass and use are set. If they're set, switch _use_login variable content to 1. | |
232 | 375 [ ${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
|
376 |
159 | 377 echo "Tags: ${TAGS}" |
378 # slash is not wanted for folder name | |
193 | 379 TARGET_DIR=$(echo "${TAGS}" | sed -e 's/\//_/g') |
380 SITE_DIR=$(echo "${SITE}" | sed -e 's/\/$//g;s/\//_/g') | |
195
652d9e268cee
test migration to printf
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
194
diff
changeset
|
381 TEMP_PREFIX="${BASE_DIR}/temp/${SITE_DIR}-${TARGET_DIR}" |
159 | 382 } |
148 | 383 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
384 # global variables goes here |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
385 init_globals() |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
386 { |
301
36bc27bb32ff
Updated version and copyright.
Edho Arief <edho@myconan.net>
parents:
300
diff
changeset
|
387 _version="1.0-rc3" # version of this script |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
388 _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
|
389 } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
390 |
227 | 391 main() |
392 { | |
234
58ad057cd2ec
- Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents:
232
diff
changeset
|
393 # 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
|
394 POSIXLY_CORRECT=1 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
395 #initialize global variables |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
396 init_globals |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
397 #print welcome message |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
398 msg_welcome |
227 | 399 # initialization |
400 init "$@" | |
401 Check_Tools | |
402 Check_Folders | |
158
cba73f6a96bb
grep check. OpenSolaris' default grep doesn't support -f
edhoprima
parents:
157
diff
changeset
|
403 |
148 | 404 |
227 | 405 # let's do the job! |
406 case "${JOB}" in | |
407 check) | |
408 Generate_Link | |
409 Check_Files | |
410 ;; | |
411 fetch) | |
412 Generate_Link | |
413 Check_Files | |
414 Fetch_Images | |
415 ;; | |
416 quickfetch) | |
417 ISNEW=1 | |
418 ISQUICK=1 | |
419 Generate_Link | |
420 Check_Files | |
421 Fetch_Images | |
422 ;; | |
423 esac | |
424 } | |
425 | |
426 # call the main routine! | |
427 main "$@" | |
428 |