Mercurial > ec-dotfiles
annotate bin/moefetch @ 298:a872c40368dd
Why .sh.
author | Edho Arief <edho@myconan.net> |
---|---|
date | Thu, 09 Feb 2012 08:30:09 +0700 |
parents | bin/moefetch.sh@d7e5a2e70cf3 |
children | d5ac851f3225 |
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 |
3 # Copyright (c) 2009, edogawaconan <me@myconan.net> | |
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 | |
159 | 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 ${_version} |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
40 Copyright (c) 2009 edogawaconan <me@myconan.net> |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
41 " |
159 | 42 } |
43 | |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
44 # Sanitize path. Totally safe. Usage: cmd "$(safe_path "${filename}")" |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
45 safe_path() |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
46 { |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
47 # It all depends on the first character. |
232 | 48 start=$(printf "%s" "$*" | cut -c 1) |
49 path= | |
50 case "${start}" in | |
51 .|/) path="$*";; # . and / is safe. No change. | |
52 *) path="./$*";; # Anything else must be prefixed with ./ | |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
53 esac |
232 | 54 printf "%s" "${path}" # Return. |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
55 } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
56 |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
57 # Checks md5. OpenSSL should be available on anything usable. |
230 | 58 get_md5() { cat "$(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 | 73 # fatal error handler |
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 | 77 exit 1 |
78 } | |
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 | 87 # help message |
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 ${DEFAULT_SITE}. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
104 -u USERNAME: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
105 Specifies username for login. |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
106 TAGS: |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
107 Tags you want to download. Separated by spaces. Tag name follows standard Danbooru tagging scheme." |
193 | 108 exit 2 |
159 | 109 } |
110 | |
111 # generate link by transforming xml | |
112 Generate_Link() { | |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
113 echo " |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
114 Fetching XML file" |
232 | 115 tempnum=1000 |
116 iternum=1 | |
195
652d9e268cee
test migration to printf
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
194
diff
changeset
|
117 > "${TEMP_PREFIX}-list" |
232 | 118 while [ "${tempnum}" -ge 1000 ]; do |
119 url="http://${SITE}/post/index.xml?tags=$(get_cleantags "${TAGS}")&offset=0&limit=1000&page=${iternum}" | |
120 [ ${_use_login} -eq 1 ] && url="${url}&login=${LOGIN_USER}&password_hash=${LOGIN_PASS}" | |
121 wget --quiet "${url}" -O "${TEMP_PREFIX}-xml" -e continue=off || Err_Fatal "Failed download catalog file" | |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
122 printf "Processing XML file... " |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
123 # xslt evilry |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
124 xsltproc - "${TEMP_PREFIX}-xml" <<EOF | sed 's/.*\(http.*\)\(\/[a-f0-9]\{32\}\).*\.\([^\.]*\)/\1\2.\3/g' | grep ^http > "${TEMP_PREFIX}-templist" |
148 | 125 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0"> |
126 <xsl:output method="xml" indent="yes"/> | |
127 <xsl:template match="post"> | |
128 <xsl:value-of select="@file_url" /> | |
129 </xsl:template> | |
130 </xsl:stylesheet> | |
131 EOF | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
132 tempnum=$(grep -c . "${TEMP_PREFIX}-templist") |
232 | 133 iternum=$((iternum + 1)) |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
134 cat "${TEMP_PREFIX}-templist" >> "${TEMP_PREFIX}-list" |
232 | 135 echo "${tempnum} file(s) available" |
213
dd95cf01602c
working around limit
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
212
diff
changeset
|
136 done |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
137 numfiles=$(grep -c . "${TEMP_PREFIX}-list") |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
138 echo "${numfiles} file(s) available on server" |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
139 [ "${numfiles}" -gt 0 ] || Err_Fatal "Error in processing list or no files can be found with specified tag(s) or site." |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
140 } |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
141 |
148 | 142 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
143 progress_init() { |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
144 _last="-" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
145 printf "${_last}" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
146 } |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
147 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
148 progress_anim() { |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
149 case "${_last}" in |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
150 /) _last="-";; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
151 -) _last=\\;; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
152 \\) _last=\|;; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
153 \|) _last="/";; |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
154 esac |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
155 printf "\b${_last}" |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
156 } |
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
157 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
158 progress_done() { printf "\bdone\n"; } |
205
2e866999c042
now with useless animation
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
204
diff
changeset
|
159 |
200 | 160 # getting rid of ls (as per suggestion) |
161 Count_Files() { | |
224 | 162 numfiles=0 |
163 for dircontent in "${*}/"* "${*}/".*; do | |
251
d7e5a2e70cf3
Proper test for for loop (*, .*)
Edho Arief <edho@myconan.net>
parents:
236
diff
changeset
|
164 if [ -e "${dircontent}" ] && [ x"${dircontent}" != x"${*}/." ] && [ x"${dircontent}" != x"${*}/.." ]; then |
230 | 165 numfiles=$((numfiles + 1)) |
200 | 166 fi |
167 done | |
230 | 168 echo $((numfiles - 2)) |
200 | 169 } |
170 | |
159 | 171 # check tools availability |
172 Check_Tools() { | |
173 # 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
|
174 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
|
175 for cmd in ${commands} |
159 | 176 do |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
177 [ "$(command -v "${cmd}")" ] || Err_Fatal "${cmd} doesn't exist in ${PATH}" |
159 | 178 done |
179 } | |
180 | |
181 # verify required folders exist and writeable | |
182 Check_Folders(){ | |
223 | 183 [ -O "${BASE_DIR}" ] || Err_Fatal "You don't own ${BASE_DIR}. Please fix ${BASE_DIR} or run this script in your own directory." |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
184 for directory in temp trash deleted "${SITE_DIR}/${TARGET_DIR}"; do |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
185 if [ ! -d "${BASE_DIR}/${directory}" ]; then |
216
a869987c4646
did I say 'mess up'?
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
215
diff
changeset
|
186 mkdir -p "${BASE_DIR}/${directory}" || Err_Impossible |
159 | 187 fi |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
188 if [ ! -O "${BASE_DIR}/${directory}" ]; then |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
189 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
|
190 chmod -R u=rwX,g=rwX,o=rwX "${BASE_DIR}/${directory}" || Err_Impossible |
159 | 191 fi |
192 done | |
223 | 193 [ "$(Count_Files "${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
|
194 for i in error ok list newlist templist; do |
196 | 195 touch "${TEMP_PREFIX}-${i}" || Fatal_Err "Error creating ${TEMP_PREFIX}-${i}. This shouldn't happen" |
159 | 196 done |
197 # | |
198 } | |
199 | |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
200 # Do some cleanup |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
201 Cleanup_Repository() { |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
202 # current dir: ${BASE_DIR}/${SITE_DIR}/${TARGET_DIR} |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
203 printf "Cleaning up repository folder... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
204 progress_init |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
205 trash_dir="${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
|
206 trashes="These files have been moved to ${trash_dir}:" |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
207 has_trash= |
223 | 208 if [ ! -d "${trash_dir}" ]; then |
216
a869987c4646
did I say 'mess up'?
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
215
diff
changeset
|
209 mkdir -p "${trash_dir}" || Err_Impossible |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
210 else |
223 | 211 if [ ! -O "${trash_dir}" ]; then |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
212 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
|
213 fi |
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 for trash in "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}/"* |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
216 do |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
217 is_trash= |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
218 if [ -d "${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
|
219 is_trash=1 |
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
220 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
|
221 mv -f -- "${trash}" "${trash_dir}" || Err_Impossible |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
222 trashes="${trashes} |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
223 $(get_basename "${trash}")" |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
224 fi |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
225 progress_anim |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
226 done |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
227 rmdir "${trash_dir}" 2>/dev/null |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
228 progress_done |
223 | 229 [ -n "${has_trash}" ] && echo "${trashes}" |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
230 } |
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
231 |
159 | 232 # check files correctness |
233 Check_Files() { | |
223 | 234 if [ ! -n "${ISNEW}" ]; then |
235 [ -z "${NOCLEAN}" ] && Cleanup_Repository | |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
236 printf "Checking for errors... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
237 progress_init |
232 | 238 files_error="These files do not match its md5:" |
239 files_notdanbooru="These files are not checked:" | |
240 has_err_filename= | |
241 has_err_md5= | |
196 | 242 > "${TEMP_PREFIX}-error" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
243 > "${TEMP_PREFIX}-ok" |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
244 for file in "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}/"* |
159 | 245 do |
251
d7e5a2e70cf3
Proper test for for loop (*, .*)
Edho Arief <edho@myconan.net>
parents:
236
diff
changeset
|
246 if [ -e "${file}" ]; then |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
247 if [ -n "$(is_not_md5 "${file}")" ] || [ -d "${file}" ]; then |
232 | 248 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
|
249 $(get_basename "${file}")" |
232 | 250 has_err_filename=1 |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
251 else |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
252 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
|
253 echo "$(get_basename "${file}")" >> "${TEMP_PREFIX}-ok" |
217 | 254 else |
255 rm "${file}" || Err_Fatal "Error removing ${file}" | |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
256 echo "$(get_basename "${file}")" >> "${TEMP_PREFIX}-error" |
232 | 257 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
|
258 $(get_basename "${file}")" |
232 | 259 has_err_md5=1 |
217 | 260 fi |
187
efd957294c8c
refactoring. cleanup. etc.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
186
diff
changeset
|
261 fi |
148 | 262 fi |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
263 progress_anim |
148 | 264 done |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
265 progress_done |
232 | 266 if [ ! -n "${has_err_md5}" ] && [ ! -n "${has_err_filename}" ]; then |
203 | 267 echo "All files OK" |
170 | 268 else |
232 | 269 if [ -n "${has_err_md5}" ]; then |
270 echo "${files_error}" | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
271 echo "$(grep -c . "${TEMP_PREFIX}-error") file(s) removed" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
272 fi |
232 | 273 [ -n "${has_err_filename}" ] && echo "${files_notdanbooru}" |
170 | 274 fi |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
275 echo "$(grep -c . "${TEMP_PREFIX}-ok") file(s) available locally" |
148 | 276 |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
277 printf "Generating list of new files... " |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
278 progress_init |
218 | 279 cp -f "${TEMP_PREFIX}-list" "${TEMP_PREFIX}-templist" |
214
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
280 while read -r is_ok; do |
a6624fb9b317
major cleanup. tweaking.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
213
diff
changeset
|
281 grep -v "${is_ok}" "${TEMP_PREFIX}-templist" > "${TEMP_PREFIX}-newlist" |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
282 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
|
283 progress_anim |
203 | 284 done < "${TEMP_PREFIX}-ok" |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
285 progress_done |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
286 echo "$(grep -c . "${TEMP_PREFIX}-newlist") file(s) to be downloaded" |
148 | 287 else |
223 | 288 if [ -n "${ISQUICK}" ]; then |
207
17d816a63b4c
final progress version
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
206
diff
changeset
|
289 echo "Quick mode selected. Skipping check" |
152 | 290 else |
291 echo "Empty local repository" | |
292 fi | |
200 | 293 cat "${TEMP_PREFIX}-list" > "${TEMP_PREFIX}-newlist" |
148 | 294 fi |
295 } | |
296 | |
159 | 297 # start downloading the images |
298 Fetch_Images() { | |
235
649b7d4b056a
Use "grep -c ." instead of "echo $(wc -l <" evilry. I should stop trying to fix this script.
Edho Prima Arief <me@myconan.net>
parents:
234
diff
changeset
|
299 if [ "$(grep -c . "${TEMP_PREFIX}-newlist")" -eq 0 ]; then |
148 | 300 echo "No new file" |
301 else | |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
302 printf "Downloading files... " |
160
68227a30d0b3
forgot to fix Fetch_Images to reflect new folder naming scheme
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
159
diff
changeset
|
303 cd "${BASE_DIR}/${SITE_DIR}/${TARGET_DIR}" |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
304 wget -e continue=on -i "${TEMP_PREFIX}-newlist" -o "${TEMP_PREFIX}.log" |
148 | 305 fi |
306 } | |
307 | |
159 | 308 # initialize base variables and initial command check |
227 | 309 init() |
310 { | |
159 | 311 # path initialization |
227 | 312 # check if additional path is specified |
313 if [ -n "${ADDITIONAL_PATH}" ] | |
314 then | |
315 # insert the additional path | |
316 PATH="${ADDITIONAL_PATH}:${PATH}" | |
317 export PATH | |
318 fi | |
158
cba73f6a96bb
grep check. OpenSolaris' default grep doesn't support -f
edhoprima
parents:
157
diff
changeset
|
319 |
159 | 320 # misc variables |
166 | 321 ISQUICK= |
322 ISNEW= | |
215
710082ce6788
major cleanup part2.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
214
diff
changeset
|
323 |
227 | 324 # minimum number of arguments: 2 (command and tag). If less than two, exit and print help message |
159 | 325 [ $# -lt 2 ] && Err_Help |
326 case "$1" in | |
174
0948e76a57a1
added help. Bump to 0.1-beta2
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
173
diff
changeset
|
327 check|fetch|quickfetch) |
159 | 328 echo "Starting..." |
329 JOB="$1" | |
330 ;; | |
331 *) | |
332 Err_Help | |
333 ;; | |
334 esac | |
335 shift | |
230 | 336 SITE= |
337 TAGS= | |
232 | 338 has_pass=0 |
339 has_user=0 | |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
340 x=1 |
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
341 while getopts "s:nu:p:" opt |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
342 do |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
343 case "$opt" in |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
344 s) SITE="$OPTARG";; |
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
345 n) NOCLEAN=1;; |
227 | 346 p) |
234
58ad057cd2ec
- Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents:
232
diff
changeset
|
347 LOGIN_PASS=$(printf "%s" "$OPTARG" | openssl dgst -sha1 | sed -e 's/.*\([[:xdigit:]]\{40\}\).*/\1/') |
232 | 348 has_pass=1 |
227 | 349 ;; |
350 u) | |
351 LOGIN_USER="$OPTARG" | |
232 | 352 has_user=1 |
227 | 353 ;; |
185
6d926d4b3c5a
initial clean system support
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
184
diff
changeset
|
354 esac |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
355 x=$OPTIND |
185
6d926d4b3c5a
initial clean system support
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
184
diff
changeset
|
356 done |
231
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
357 shift $(($x-1)) |
4c0fd276665e
- for some reason I broke the getopts logic again. Fixed
edhoprima
parents:
230
diff
changeset
|
358 if [ "$1" = -- ]; then shift; fi |
225
265a9ca47a19
- Replaced md5(sum) with openssl. Less platform dependent because the tool is same across platforms
edhoprima
parents:
224
diff
changeset
|
359 TAGS="$@" |
223 | 360 [ -n "${SITE}" ] || SITE=${DEFAULT_SITE} |
361 [ -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
|
362 # Get base folder - default, current folder or fallback to ${HOME} |
223 | 363 [ -n "${BASE_DIR}" ] || BASE_DIR=${PWD} |
364 [ -n "${BASE_DIR}" ] || BASE_DIR=${HOME} | |
227 | 365 [ -n "$(echo "${BASE_DIR}" | cut -c1 | grep \/)" ] || BASE_DIR="/${BASE_DIR}" |
366 # see if both pass and use are set. If they're set, switch _use_login variable content to 1. | |
232 | 367 [ ${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
|
368 |
159 | 369 echo "Tags: ${TAGS}" |
370 # slash is not wanted for folder name | |
193 | 371 TARGET_DIR=$(echo "${TAGS}" | sed -e 's/\//_/g') |
372 SITE_DIR=$(echo "${SITE}" | sed -e 's/\/$//g;s/\//_/g') | |
195
652d9e268cee
test migration to printf
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
194
diff
changeset
|
373 TEMP_PREFIX="${BASE_DIR}/temp/${SITE_DIR}-${TARGET_DIR}" |
159 | 374 } |
148 | 375 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
376 # global variables goes here |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
377 init_globals() |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
378 { |
232 | 379 _version="1.0-rc2" # version of this script |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
380 _use_login=0 # variable to check whether a login is used or not |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
381 } |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
382 |
227 | 383 main() |
384 { | |
234
58ad057cd2ec
- Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents:
232
diff
changeset
|
385 # removing GNU-ism as much as possible |
58ad057cd2ec
- Fix for openssl output parser for generating hashed password
Edho P. Arief <me@myconan.net>
parents:
232
diff
changeset
|
386 POSIXLY_CORRECT=1 |
228
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
387 #initialize global variables |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
388 init_globals |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
389 #print welcome message |
5d3a0645b504
- Restructured some things.
edhoprima@gmail.com <edhoprima@gmail.com>
parents:
227
diff
changeset
|
390 msg_welcome |
227 | 391 # initialization |
392 init "$@" | |
393 Check_Tools | |
394 Check_Folders | |
158
cba73f6a96bb
grep check. OpenSolaris' default grep doesn't support -f
edhoprima
parents:
157
diff
changeset
|
395 |
148 | 396 |
227 | 397 # let's do the job! |
398 case "${JOB}" in | |
399 check) | |
400 Generate_Link | |
401 Check_Files | |
402 ;; | |
403 fetch) | |
404 Generate_Link | |
405 Check_Files | |
406 Fetch_Images | |
407 ;; | |
408 quickfetch) | |
409 ISNEW=1 | |
410 ISQUICK=1 | |
411 Generate_Link | |
412 Check_Files | |
413 Fetch_Images | |
414 ;; | |
415 esac | |
416 } | |
417 | |
418 # call the main routine! | |
419 main "$@" | |
420 |