Mercurial > ec-dotfiles
annotate bin/moefetch @ 302:b90ebadbfd5d
Forgot the other wget.
author | Edho Arief <edho@myconan.net> |
---|---|
date | Thu, 09 Feb 2012 09:12:18 +0700 |
parents | 36bc27bb32ff |
children | 21b86001b0c5 |
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 | |
302 | 37 useragent="Mozilla/5.0 (Windows NT 6.1; WOW64; rv:10.0) Gecko/20100101 Firefox/10.0" |
38 | |
159 | 39 # useless welcome message. Also version |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
40 msg_welcome() { |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
41 echo "moefetch ${_version} |
301
36bc27bb32ff
Updated version and copyright.
Edho Arief <edho@myconan.net>
parents:
300
diff
changeset
|
42 Copyright (c) 2009-2012 edogawaconan <edho@myconan.net> |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
43 " |
159 | 44 } |
45 | |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
46 # Sanitize path. Totally safe. Usage: cmd "$(safe_path "${filename}")" |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
47 safe_path() |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
48 { |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
49 # It all depends on the first character. |
232 | 50 start=$(printf "%s" "$*" | cut -c 1) |
51 path= | |
52 case "${start}" in | |
53 .|/) path="$*";; # . and / is safe. No change. | |
54 *) path="./$*";; # Anything else must be prefixed with ./ | |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
55 esac |
232 | 56 printf "%s" "${path}" # Return. |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
57 } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
58 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
59 # Checks md5. OpenSSL should be available on anything usable. |
230 | 60 get_md5() { cat "$(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
|
61 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
62 # Safely get basename. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
63 get_basename() { basename "$(safe_path "${1}")"; } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
64 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
65 # 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
|
66 get_filename() { get_basename "${1%.*}"; } |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
67 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
68 # Transformation for tag url. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
69 get_cleantags() { printf "%s " "$*" | sed -e 's/\&/%26/g;s/=/%3D/g'; } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
70 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
71 # Returns something if not an md5 value. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
72 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
|
73 |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
74 |
159 | 75 # fatal error handler |
76 Err_Fatal() { | |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
77 echo " |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
78 Fatal error: ${1}" |
159 | 79 exit 1 |
80 } | |
81 | |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
82 Err_Impossible() { |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
83 echo " |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
84 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
|
85 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
|
86 exit 1 |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
87 } |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
88 |
159 | 89 # help message |
90 Err_Help() { | |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
91 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
|
92 |
176
3d2ae9417273
even more improvement
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
175
diff
changeset
|
93 COMMAND: |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
94 (quick)fetch: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
95 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
|
96 check: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
97 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
|
98 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
99 OPTIONS: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
100 -n: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
101 Skip checking repository directory. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
102 -p PASSWORD: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
103 Specifies password for login. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
104 -s SITE_URL: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
105 Specify URL of the Danbooru powered site you want to leech from. Default is ${DEFAULT_SITE}. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
106 -u USERNAME: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
107 Specifies username for login. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
108 TAGS: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
109 Tags you want to download. Separated by spaces. Tag name follows standard Danbooru tagging scheme." |
193 | 110 exit 2 |
159 | 111 } |
112 | |
113 # generate link by transforming xml | |
114 Generate_Link() { | |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
115 echo " |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
116 Fetching XML file" |
232 | 117 tempnum=1000 |
118 iternum=1 | |
195
652d9e268cee
test migration to printf
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
194
diff
changeset
|
119 > "${TEMP_PREFIX}-list" |
232 | 120 while [ "${tempnum}" -ge 1000 ]; do |
121 url="http://${SITE}/post/index.xml?tags=$(get_cleantags "${TAGS}")&offset=0&limit=1000&page=${iternum}" | |
122 [ ${_use_login} -eq 1 ] && url="${url}&login=${LOGIN_USER}&password_hash=${LOGIN_PASS}" | |
299
d5ac851f3225
[moefetch] User agent, referer.
Edho Arief <edho@myconan.net>
parents:
298
diff
changeset
|
123 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
|
124 printf "Processing XML file... " |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
125 # xslt evilry |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
126 xsltproc - "${TEMP_PREFIX}-xml" <<EOF | sed 's/.*\(http.*\)\(\/[a-f0-9]\{32\}\).*\.\([^\.]*\)/\1\2.\3/g' | grep ^http > "${TEMP_PREFIX}-templist" |
148 | 127 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0"> |
128 <xsl:output method="xml" indent="yes"/> | |
129 <xsl:template match="post"> | |
130 <xsl:value-of select="@file_url" /> | |
131 </xsl:template> | |
132 </xsl:stylesheet> | |
133 EOF | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
134 tempnum=$(grep -c . "${TEMP_PREFIX}-templist") |
232 | 135 iternum=$((iternum + 1)) |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
136 cat "${TEMP_PREFIX}-templist" >> "${TEMP_PREFIX}-list" |
232 | 137 echo "${tempnum} file(s) available" |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
138 done |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
139 numfiles=$(grep -c . "${TEMP_PREFIX}-list") |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
140 echo "${numfiles} file(s) available on server" |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
141 [ "${numfiles}" -gt 0 ] || Err_Fatal "Error in processing list or no files can be found with specified tag(s) or site." |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
142 } |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
143 |
148 | 144 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
145 progress_init() { |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
146 _last="-" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
147 printf "${_last}" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
148 } |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
149 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
150 progress_anim() { |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
151 case "${_last}" in |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
152 /) _last="-";; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
153 -) _last=\\;; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
154 \\) _last=\|;; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
155 \|) _last="/";; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
156 esac |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
157 printf "\b${_last}" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
158 } |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
159 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
160 progress_done() { printf "\bdone\n"; } |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
161 |
200 | 162 # getting rid of ls (as per suggestion) |
163 Count_Files() { | |
224 | 164 numfiles=0 |
165 for dircontent in "${*}/"* "${*}/".*; do | |
251
d7e5a2e70cf3
Proper test for for loop (*, .*)
Edho Arief <edho@myconan.net>
parents:
236
diff
changeset
|
166 if [ -e "${dircontent}" ] && [ x"${dircontent}" != x"${*}/." ] && [ x"${dircontent}" != x"${*}/.." ]; then |
230 | 167 numfiles=$((numfiles + 1)) |
200 | 168 fi |
169 done | |
300 | 170 echo $((numfiles)) |
200 | 171 } |
172 | |
159 | 173 # check tools availability |
174 Check_Tools() { | |
175 # 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
|
176 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
|
177 for cmd in ${commands} |
159 | 178 do |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
179 [ "$(command -v "${cmd}")" ] || Err_Fatal "${cmd} doesn't exist in ${PATH}" |
159 | 180 done |
181 } | |
182 | |
183 # verify required folders exist and writeable | |
184 Check_Folders(){ | |
223 | 185 [ -O "${BASE_DIR}" ] || Err_Fatal "You don't own ${BASE_DIR}. Please fix ${BASE_DIR} or run this script in your own directory." |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
186 for directory in temp trash deleted "${SITE_DIR}/${TARGET_DIR}"; do |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
187 if [ ! -d "${BASE_DIR}/${directory}" ]; then |
216
a869987c4646
did I say 'mess up'?
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
215
diff
changeset
|
188 mkdir -p "${BASE_DIR}/${directory}" || Err_Impossible |
159 | 189 fi |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
190 if [ ! -O "${BASE_DIR}/${directory}" ]; then |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
191 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
|
192 chmod -R u=rwX,g=rwX,o=rwX "${BASE_DIR}/${directory}" || Err_Impossible |
159 | 193 fi |
194 done | |
223 | 195 [ "$(Count_Files "${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
|
196 for i in error ok list newlist templist; do |
196 | 197 touch "${TEMP_PREFIX}-${i}" || Fatal_Err "Error creating ${TEMP_PREFIX}-${i}. This shouldn't happen" |
159 | 198 done |
199 # | |
200 } | |
201 | |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
202 # Do some cleanup |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
203 Cleanup_Repository() { |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
204 # current dir: ${BASE_DIR}/${SITE_DIR}/${TARGET_DIR} |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
205 printf "Cleaning up repository folder... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
206 progress_init |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
207 trash_dir="${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
|
208 trashes="These files have been moved to ${trash_dir}:" |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
209 has_trash= |
223 | 210 if [ ! -d "${trash_dir}" ]; then |
216
a869987c4646
did I say 'mess up'?
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
215
diff
changeset
|
211 mkdir -p "${trash_dir}" || Err_Impossible |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
212 else |
223 | 213 if [ ! -O "${trash_dir}" ]; then |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
214 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
|
215 fi |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
216 fi |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
217 for trash in "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}/"* |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
218 do |
300 | 219 if [ -e "${trash}" ]; then |
220 is_trash= | |
221 if [ -d "${trash}" ] || [ -n "$(is_not_md5 "${trash}")" ] || [ -z "$(grep "$(get_basename "${trash}")" "${TEMP_PREFIX}-list")" ]; then | |
222 is_trash=1 | |
223 has_trash=1 | |
224 mv -f -- "${trash}" "${trash_dir}" || Err_Impossible | |
225 trashes="${trashes} | |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
226 $(get_basename "${trash}")" |
300 | 227 fi |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
228 fi |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
229 progress_anim |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
230 done |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
231 rmdir "${trash_dir}" 2>/dev/null |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
232 progress_done |
223 | 233 [ -n "${has_trash}" ] && echo "${trashes}" |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
234 } |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
235 |
159 | 236 # check files correctness |
237 Check_Files() { | |
223 | 238 if [ ! -n "${ISNEW}" ]; then |
239 [ -z "${NOCLEAN}" ] && Cleanup_Repository | |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
240 printf "Checking for errors... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
241 progress_init |
232 | 242 files_error="These files do not match its md5:" |
243 files_notdanbooru="These files are not checked:" | |
244 has_err_filename= | |
245 has_err_md5= | |
196 | 246 > "${TEMP_PREFIX}-error" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
247 > "${TEMP_PREFIX}-ok" |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
248 for file in "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}/"* |
159 | 249 do |
251
d7e5a2e70cf3
Proper test for for loop (*, .*)
Edho Arief <edho@myconan.net>
parents:
236
diff
changeset
|
250 if [ -e "${file}" ]; then |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
251 if [ -n "$(is_not_md5 "${file}")" ] || [ -d "${file}" ]; then |
232 | 252 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
|
253 $(get_basename "${file}")" |
232 | 254 has_err_filename=1 |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
255 else |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
256 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
|
257 echo "$(get_basename "${file}")" >> "${TEMP_PREFIX}-ok" |
217 | 258 else |
259 rm "${file}" || Err_Fatal "Error removing ${file}" | |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
260 echo "$(get_basename "${file}")" >> "${TEMP_PREFIX}-error" |
232 | 261 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
|
262 $(get_basename "${file}")" |
232 | 263 has_err_md5=1 |
217 | 264 fi |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
265 fi |
148 | 266 fi |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
267 progress_anim |
148 | 268 done |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
269 progress_done |
232 | 270 if [ ! -n "${has_err_md5}" ] && [ ! -n "${has_err_filename}" ]; then |
203 | 271 echo "All files OK" |
170 | 272 else |
232 | 273 if [ -n "${has_err_md5}" ]; then |
274 echo "${files_error}" | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
275 echo "$(grep -c . "${TEMP_PREFIX}-error") file(s) removed" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
276 fi |
232 | 277 [ -n "${has_err_filename}" ] && echo "${files_notdanbooru}" |
170 | 278 fi |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
279 echo "$(grep -c . "${TEMP_PREFIX}-ok") file(s) available locally" |
148 | 280 |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
281 printf "Generating list of new files... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
282 progress_init |
218 | 283 cp -f "${TEMP_PREFIX}-list" "${TEMP_PREFIX}-templist" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
284 while read -r is_ok; do |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
285 grep -v "${is_ok}" "${TEMP_PREFIX}-templist" > "${TEMP_PREFIX}-newlist" |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
286 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
|
287 progress_anim |
203 | 288 done < "${TEMP_PREFIX}-ok" |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
289 progress_done |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
290 echo "$(grep -c . "${TEMP_PREFIX}-newlist") file(s) to be downloaded" |
148 | 291 else |
223 | 292 if [ -n "${ISQUICK}" ]; then |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
293 echo "Quick mode selected. Skipping check" |
152 | 294 else |
295 echo "Empty local repository" | |
296 fi | |
200 | 297 cat "${TEMP_PREFIX}-list" > "${TEMP_PREFIX}-newlist" |
148 | 298 fi |
299 } | |
300 | |
159 | 301 # start downloading the images |
302 Fetch_Images() { | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
303 if [ "$(grep -c . "${TEMP_PREFIX}-newlist")" -eq 0 ]; then |
148 | 304 echo "No new file" |
305 else | |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
306 printf "Downloading files... " |
160
68227a30d0b3
forgot to fix Fetch_Images to reflect new folder naming scheme
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
159
diff
changeset
|
307 cd "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}" |
302 | 308 wget -e continue=on -i "${TEMP_PREFIX}-newlist" -o "${TEMP_PREFIX}.log" --referer="http://${SITE}/post" --user-agent="${useragent}" |
148 | 309 fi |
310 } | |
311 | |
159 | 312 # initialize base variables and initial command check |
227 | 313 init() |
314 { | |
159 | 315 # path initialization |
227 | 316 # check if additional path is specified |
317 if [ -n "${ADDITIONAL_PATH}" ] | |
318 then | |
319 # insert the additional path | |
320 PATH="${ADDITIONAL_PATH}:${PATH}" | |
321 export PATH | |
322 fi | |
158
cba73f6a96bb
grep check. OpenSolaris' default grep doesn't support -f
edhoprima
parents:
157
diff
changeset
|
323 |
159 | 324 # misc variables |
166 | 325 ISQUICK= |
326 ISNEW= | |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
327 |
227 | 328 # minimum number of arguments: 2 (command and tag). If less than two, exit and print help message |
159 | 329 [ $# -lt 2 ] && Err_Help |
330 case "$1" in | |
174
0948e76a57a1
added help. Bump to 0.1-beta2
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
173
diff
changeset
|
331 check|fetch|quickfetch) |
159 | 332 echo "Starting..." |
333 JOB="$1" | |
334 ;; | |
335 *) | |
336 Err_Help | |
337 ;; | |
338 esac | |
339 shift | |
230 | 340 SITE= |
341 TAGS= | |
232 | 342 has_pass=0 |
343 has_user=0 | |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
344 x=1 |
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
345 while getopts "s:nu:p:" opt |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
346 do |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
347 case "$opt" in |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
348 s) SITE="$OPTARG";; |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
349 n) NOCLEAN=1;; |
227 | 350 p) |
234
58ad057cd2ec
- Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents:
232
diff
changeset
|
351 LOGIN_PASS=$(printf "%s" "$OPTARG" | openssl dgst -sha1 | sed -e 's/.*\([[:xdigit:]]\{40\}\).*/\1/') |
232 | 352 has_pass=1 |
227 | 353 ;; |
354 u) | |
355 LOGIN_USER="$OPTARG" | |
232 | 356 has_user=1 |
227 | 357 ;; |
185
6d926d4b3c5a
initial clean system support
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
184
diff
changeset
|
358 esac |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
359 x=$OPTIND |
185
6d926d4b3c5a
initial clean system support
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
184
diff
changeset
|
360 done |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
361 shift $(($x-1)) |
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
362 if [ "$1" = -- ]; then shift; fi |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
363 TAGS="$@" |
223 | 364 [ -n "${SITE}" ] || SITE=${DEFAULT_SITE} |
365 [ -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
|
366 # Get base folder - default, current folder or fallback to ${HOME} |
223 | 367 [ -n "${BASE_DIR}" ] || BASE_DIR=${PWD} |
368 [ -n "${BASE_DIR}" ] || BASE_DIR=${HOME} | |
227 | 369 [ -n "$(echo "${BASE_DIR}" | cut -c1 | grep \/)" ] || BASE_DIR="/${BASE_DIR}" |
370 # see if both pass and use are set. If they're set, switch _use_login variable content to 1. | |
232 | 371 [ ${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
|
372 |
159 | 373 echo "Tags: ${TAGS}" |
374 # slash is not wanted for folder name | |
193 | 375 TARGET_DIR=$(echo "${TAGS}" | sed -e 's/\//_/g') |
376 SITE_DIR=$(echo "${SITE}" | sed -e 's/\/$//g;s/\//_/g') | |
195
652d9e268cee
test migration to printf
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
194
diff
changeset
|
377 TEMP_PREFIX="${BASE_DIR}/temp/${SITE_DIR}-${TARGET_DIR}" |
159 | 378 } |
148 | 379 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
380 # global variables goes here |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
381 init_globals() |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
382 { |
301
36bc27bb32ff
Updated version and copyright.
Edho Arief <edho@myconan.net>
parents:
300
diff
changeset
|
383 _version="1.0-rc3" # version of this script |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
384 _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
|
385 } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
386 |
227 | 387 main() |
388 { | |
234
58ad057cd2ec
- Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents:
232
diff
changeset
|
389 # 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
|
390 POSIXLY_CORRECT=1 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
391 #initialize global variables |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
392 init_globals |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
393 #print welcome message |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
394 msg_welcome |
227 | 395 # initialization |
396 init "$@" | |
397 Check_Tools | |
398 Check_Folders | |
158
cba73f6a96bb
grep check. OpenSolaris' default grep doesn't support -f
edhoprima
parents:
157
diff
changeset
|
399 |
148 | 400 |
227 | 401 # let's do the job! |
402 case "${JOB}" in | |
403 check) | |
404 Generate_Link | |
405 Check_Files | |
406 ;; | |
407 fetch) | |
408 Generate_Link | |
409 Check_Files | |
410 Fetch_Images | |
411 ;; | |
412 quickfetch) | |
413 ISNEW=1 | |
414 ISQUICK=1 | |
415 Generate_Link | |
416 Check_Files | |
417 Fetch_Images | |
418 ;; | |
419 esac | |
420 } | |
421 | |
422 # call the main routine! | |
423 main "$@" | |
424 |