lib/taggery.sh

   1 [ "${taggery_name}" -a "${taggery_fmturl}" -a "${taggery_image_url_filter}" -a "${taggery_image_tags_filter}" ] || {
   2   cat <<EOF
   3 Taggery v0.1
   4 Copyright - http://wiki.cyberleo.net/wiki/CyberLeo/COPYRIGHT?version=4
   5
   6 Set the following variables before sourcing this library:
   7
   8  taggery_base              - The base directory where the taggery database
   9                              resides. This should contain the directories
  10                              md5/ tag/ pool/
  11  taggery_libs              - The base directory where the code resides.
  12
  13 The following should be defined by the profile:
  14
  15  taggery_name              - A string token to use when formatting things
  16  taggery_fmturl            - A printf-compatible format string that will
  17                              transform an ID into the URL of a 'view' page.
  18  taggery_image_url_filter  - A program or function that takes a webpage on
  19                              stdin and provides on stdout the URL to the raw
  20                              image to be downloaded.
  21  taggery_image_tags_filter - A program or function that takes a webpage on
  22                              stdin and provides on stdout a list of newline-
  23                              delimited tags.
  24  taggery_image_referer     - A program or function that takes an id and
  25                              returns the referer URL for that id.
  26
  27 Examples:
  28  Format URL:
  29   taggery_fmturl="http://e621.net/post/show/%s"
  30   taggery_fmturl="http://gelbooru.com/index.php?page=post&s=view&id=%s"
  31
  32  Filters:
  33   e621_image_url_filter() { sed -e '...'; }
  34   taggery_image_url_filter="e621_image_url_filter"
  35 EOF
  36   kill $$
  37   exit 1
  38 }
  39
  40 taggery_cleanup() {
  41   local cleanup="$(kvs_get "${taggery_name}" cleanup)"
  42   if [ "${cleanup}" ]
  43   then
  44     local id
  45     for id in ${cleanup}
  46     do
  47       taggery_free_page "${id}"
  48     done
  49   fi
  50   kvs_unset "${taggery_name}" cleanup
  51 }
  52 [ "${DEBUG}" ] || trap "taggery_cleanup" EXIT HUP INT TERM KILL
  53
  54 taggery_register_cleanup() {
  55   [ "${1}" ] || return 1
  56   local id="${1}"
  57   kvs_set "${taggery_name}" cleanup "$(kvs_get "${taggery_name}" cleanup) ${id}"
  58 }
  59
  60 taggery_init_page() {
  61   [ "${1}" ] || return 1
  62   local id="${1}"
  63   kvs_set "${taggery_name}/${id}" tmp "$(mktemp "${taggery_temp}/.taggery_${id}.XXXXXXXX")"
  64   kvs_set "${taggery_name}/${id}" url "$(printf "${taggery_fmturl}" "${id}")"
  65   taggery_register_cleanup "${id}"
  66 }
  67
  68 taggery_free_page() {
  69   [ "${1}" ] || return 1
  70   local id="${1}"
  71   local tmp="$(kvs_get "${taggery_name}/${id}" tmp)"
  72   [ -f "${tmp}" -a "${tmp%%_*}" = "${taggery_temp}/.taggery" ] && rm -f "${tmp}"
  73   kvs_unset_all "${taggery_name}/${id}"
  74 }
  75
  76 taggery_fetch_page() {
  77   [ "${1}" ] || return 1
  78   local id="${1}"
  79   kvs_has_id "${taggery_name}/${id}" || taggery_init_page "${id}"
  80   local url="$(kvs_get "${taggery_name}/${id}" url)"
  81   local tmp="$(kvs_get "${taggery_name}/${id}" tmp)"
  82   wget ${taggery_user_agent:+--user-agent="${taggery_user_agent}"} -qO "${tmp}" "${url}"
  83 }
  84
  85 taggery_page() {
  86   [ "${1}" ] || return 1
  87   local id="${1}"
  88   kvs_has_id "${taggery_name}/${id}" || taggery_init_page "${id}"
  89   local tmp="$(kvs_get "${taggery_name}/${id}" tmp)"
  90   [ -s "${tmp}" ] || taggery_fetch_page "${id}"
  91   cat "${tmp}"
  92 }
  93
  94 taggery_id() {
  95   [ "${1}" ] || return 1
  96   local url="${1}"
  97   echo "${url}" | "${taggery_id_filter}"
  98 }
  99
 100 taggery_image_name() {
 101   [ "${1}" ] || return 1
 102   local id="${1}"
 103   local url="$(taggery_image_url "${id}")"
 104   if [ "${taggery_image_name_filter}" ]
 105   then
 106     "${taggery_image_name_filter}" "${id}" "${url}"
 107   else
 108     name="$(basename "${url}")"
 109     name="${name%%\?*}"
 110     echo "${name}"
 111   fi
 112 }
 113
 114 taggery_image_url() {
 115   [ "${1}" ] || return 1
 116   local id="${1}"
 117   taggery_page "${id}" | "${taggery_image_url_filter}"
 118 }
 119
 120 taggery_image_referer() {
 121   [ "${1}" ] || return 1
 122   local id="${1}"
 123   [ "${taggery_image_referer}" ] || return 0
 124   "${taggery_image_referer}" "${id}"
 125 }
 126
 127 taggery_fetch_image() {
 128   [ "${1}" ] || return 1
 129   local id="${1}"
 130   name="$(taggery_image_name "${id}")"
 131   file="$(taggery_image_url "${id}")"
 132   referer="$(taggery_image_referer "${id}")"
 133   wget ${taggery_user_agent:+--user-agent="${taggery_user_agent}"} ${referer:+--referer="${referer}"} -O "${taggery_base}/md5/${name}" "${file}"
 134 }
 135
 136 taggery_image_tags() {
 137   [ "${1}" ] || return 1
 138   local id="${1}"
 139   echo "${taggery_name}"
 140   taggery_page "${id}" | "${taggery_image_tags_filter}"
 141 }
 142
 143 taggery_sanitize_tag() {
 144   [ "${1}" ] || return 1
 145   local tag="${1}"
 146   echo "${tag}" | sed -e 's/\//%47/g; s/[\]/\\&/g'
 147 }
 148
 149 taggery_link_tags() {
 150   [ "${1}" ] || return 1
 151   local id="${1}"
 152   taggery_image_tags "${id}" | while read tag
 153   do
 154     [ "${tag}" ] || continue
 155     echo "${tag}"
 156     tag="$(taggery_sanitize_tag "${tag}")"
 157     mkdir -p "${taggery_base}/tag/${tag}"
 158     ln -sf "../../md5/${name}" "${taggery_base}/tag/${tag}/"
 159   done
 160 }
 161
 162 : ${taggery_temp:=${taggery_base}/tmp}
 163
 164 mkdir -p "${taggery_temp}" "${taggery_base}/md5" "${taggery_base}/tag" "${taggery_base}/pool"
 165 TMPDIR="${taggery_temp}"
 166 kvs="${taggery_temp}/taggery.kvs"
 167 . "${taggery_libs}/kvs.sh"