From 85708f5b91f3b033e4adf1f0cd81b9d39dbd1b19 Mon Sep 17 00:00:00 2001
From: CyberLeo <cyberleo@cyberleo.net>
Date: Sat, 19 May 2012 08:05:43 -0500
Subject: [PATCH] What is this I don't even...

---
 .gitignore     |  3 ++
 e621           |  1 +
 fetch          | 69 ++++++++++++++++++++++++++++++++++++
 gelbooru       |  1 +
 lib/kvs.sh     | 56 +++++++++++++++++++++++++++++
 lib/taggery.sh | 95 ++++++++++++++++++++++++++++++++++++++++++++++++++
 6 files changed, 225 insertions(+)
 create mode 120000 e621
 create mode 100755 fetch
 create mode 120000 gelbooru
 create mode 100644 lib/kvs.sh
 create mode 100644 lib/taggery.sh

diff --git a/.gitignore b/.gitignore
index e69de29..3e7eb75 100644
--- a/.gitignore
+++ b/.gitignore
@@ -0,0 +1,3 @@
+.*.kvdb
+md5/*
+tag/*
diff --git a/e621 b/e621
new file mode 120000
index 0000000..ef77e7f
--- /dev/null
+++ b/e621
@@ -0,0 +1 @@
+fetch
\ No newline at end of file
diff --git a/fetch b/fetch
new file mode 100755
index 0000000..16c5dbf
--- /dev/null
+++ b/fetch
@@ -0,0 +1,69 @@
+#!/bin/sh -e
+
+pebkac() {
+  [ "${*}" ] && printf "%s\n\n" "${*}"
+  cat <<EOF
+Usage: $(basename "${0}") <id>
+
+Given an ID number from the Gelbooru site, downloads and stashes the full-size
+image in the md5 directory and adds whatever tags are found into the tag area.
+EOF
+  kill $$
+  exit 1
+}
+
+[ "${*}" ] || pebkac
+id="${1}"
+[ -z "$(echo "${id}" | tr -d '[0-9]')" ] || pebkac "ID should be an integer"
+
+taggery_name="$(basename "${0}")"
+case "${taggery_name}" in
+e621)
+  e621_image_url_filter() {
+    sed -e '/Size: /!d; s/^.*<a //; s/>.*$//; s/^.*href="\([^"]*\)".*$/\1/; /:\/\//!s/^/http:\/\/e621.net/'
+  }
+  e621_image_tags_filter() {
+    sed -e '/id="tag-sidebar"/,/<\/ul>/!d; /class="tag-type-/!d; s/\([^?]\)<\/a>.*$/\1/; s/^.*>//; s/ /_/g'
+  }
+  taggery_fmturl="http://e621.net/post/show/%s"
+  taggery_image_url_filter=e621_image_url_filter
+  taggery_image_tags_filter=e621_image_tags_filter
+  ;;
+gelbooru)
+  gelbooru_image_url_filter() {
+    sed -e '/>Original image</!d; s/>Original image.*//i; s/^.*<a //i; s/^.*href="\([^"]*\)".*$/\1/i'
+  }
+  gelbooru_image_tags_filter() {
+    sed -e '/id="tag-sidebar"/!d; s/<\/li>/&\n/g' | sed -e '/class="tag-type-/!d; s/\([^?]\)<\/a>.*$/\1/; s/^.*>//; s/ /_/g'
+  }
+  taggery_fmturl="http://gelbooru.com/index.php?page=post&s=view&id=%s"
+  taggery_image_url_filter=gelbooru_image_url_filter
+  taggery_image_tags_filter=gelbooru_image_tags_filter
+  ;;
+*) pebkac "Unsupported" ;;
+esac
+
+# Load KVS
+kvs=".${taggery_name}.kvdb"
+. "$(dirname "${0}")/lib/kvs.sh"
+
+# Load Taggery
+. "$(dirname "${0}")/lib/taggery.sh"
+image_url() { taggery_image_url "${@}"; }
+image_tags() { taggery_image_tags "${@}"; }
+
+cd "$(dirname "${0}")"
+
+file="$(image_url "${id}")"
+tags="$(image_tags "${id}")"
+
+[ "${file}" -a "${tags}" ] || { echo "Nothing found!"; kill $$; exit 1; }
+
+name="$(basename "${file}")"
+wget -O "md5/${name}" "${file}"
+for tag in ${tags}
+do
+  echo ${tag}
+  mkdir -p "tag/${tag}"
+  ln -sf "../../md5/${name}" "tag/${tag}/"
+done
diff --git a/gelbooru b/gelbooru
new file mode 120000
index 0000000..ef77e7f
--- /dev/null
+++ b/gelbooru
@@ -0,0 +1 @@
+fetch
\ No newline at end of file
diff --git a/lib/kvs.sh b/lib/kvs.sh
new file mode 100644
index 0000000..0c2059d
--- /dev/null
+++ b/lib/kvs.sh
@@ -0,0 +1,56 @@
+t="$(printf '\t')"
+
+[ "${kvs}" ] || {
+  cat <<EOF
+KVS v0.1
+Copyright - http://wiki.cyberleo.net/wiki/CyberLeo/COPYRIGHT?version=4
+
+Set the following variables before sourcing this library:
+
+ kvs  - Path and filename of the kvdb database file
+EOF
+  kill $$
+  exit 1
+}
+
+
+[ -s "${kvs}" ] || echo "${t}${t}# KVSv0 id-key-value database" > "${kvs}"
+
+# Add or replace a value stored for a given key and id in the kvs
+kvs_set() {
+  [ "${1}" -a "${2}" ] || return 255
+  local id="${1}"
+  local var="${2}"
+  local val="${3}"
+  printf "%s\t%s\t%s\n" "${id}" "${var}" "${val}" >> "${kvs}"
+}
+
+# Fetch the most recent value stored for a given key and id from the kvs
+kvs_get() {
+  [ "${1}" -a "${2}" ] || return 255
+  local id="${1}"
+  local var="${2}"
+  grep "^${id}${t}${var}${t}" "${kvs}" | tail -n 1 | cut -d"${t}" -f3-
+}
+
+# Does the kvs have any entries for a given ID?
+kvs_has_id() {
+  [ "${1}" ] || return 255
+  local id="${1}"
+  grep -q "^${id}${t}" "${kvs}"
+}
+
+# Remove a given key with a given ID from the kvs
+kvs_unset() {
+  [ "${1}" -a "${2}" ] || return 255
+  local id="${1}"
+  local var="${2}"
+  grep -v "^${id}${t}${var}${t}" "${kvs}" > "${kvs}.tmp" && mv -f "${kvs}.tmp" "${kvs}" || rm -f "${kvs}.tmp"
+}
+
+# Remove all keys with a given ID from the kvs
+kvs_unset_all() {
+  [ "${1}" ] || return 255
+  local id="${1}"
+  grep -v "^${id}${t}" "${kvs}" > "${kvs}.tmp" && mv -f "${kvs}.tmp" "${kvs}" || rm -f "${kvs}.tmp"
+}
diff --git a/lib/taggery.sh b/lib/taggery.sh
new file mode 100644
index 0000000..1801b53
--- /dev/null
+++ b/lib/taggery.sh
@@ -0,0 +1,95 @@
+[ "${taggery_name}" -a "${taggery_fmturl}" -a "${taggery_image_url_filter}" -a "${taggery_image_tags_filter}" ] || {
+  cat <<EOF
+Taggery v0.1
+Copyright - http://wiki.cyberleo.net/wiki/CyberLeo/COPYRIGHT?version=4
+
+Set the following variables before sourcing this library:
+
+ taggery_name              - A string token to use when formatting things
+ taggery_fmturl            - A printf-compatible format string that will
+                             transform an ID into the URL of a 'view' page
+ taggery_image_url_filter  - A program or function that takes a webpage on
+                             stdin and provides on stdout the URL to the raw
+                             image to be downloaded.
+ taggery_image_tags_filter - A program or function that takes a webpage on
+                             stdin and provides on stdout a list of newline-
+                             delimited tags.
+
+Examples:
+ Format URL:
+  taggery_fmturl="http://e621.net/post/show/%s"
+  taggery_fmturl="http://gelbooru.com/index.php?page=post&s=view&id=%s"
+
+ Filters:
+  e621_image_url_filter() { sed -e '...'; }
+  taggery_image_url_filter="e621_image_url_filter"
+EOF
+  kill $$
+  exit 1
+}
+
+taggery_cleanup() {
+  local cleanup="$(kvs_get clean up)"
+  if [ "${cleanup}" ]
+  then
+    local id
+    for id in ${cleanup}
+    do
+      taggery_free_page "${id}"
+    done
+  fi
+  kvs_unset clean up
+}
+trap "taggery_cleanup" EXIT HUP INT TERM KILL
+
+taggery_register_cleanup() {
+  [ "${1}" ] || return 1
+  local id="${1}"
+  kvs_set clean up "$(kvs_get clean up) ${id}"
+}
+
+taggery_init_page() {
+  [ "${1}" ] || return 1
+  local id="${1}"
+  kvs_set "${id}" tmp "$(mktemp ".taggery_${id}.XXXXXXXX")"
+  kvs_set "${id}" url "$(printf "${taggery_fmturl}" "${id}")"
+  taggery_register_cleanup "${id}"
+}
+
+taggery_free_page() {
+  [ "${1}" ] || return 1
+  local id="${1}"
+  local tmp="$(kvs_get "${id}" tmp)"
+  [ -f "${tmp}" -a "${tmp%%_*}" = ".taggery" ] && rm -f "${tmp}"
+  kvs_unset_all "${id}"
+}
+
+taggery_fetch() {
+  [ "${1}" ] || return 1
+  local id="${1}"
+  kvs_has_id "${id}" || taggery_init_page "${id}"
+  local url="$(kvs_get "${id}" url)"
+  local tmp="$(kvs_get "${id}" tmp)"
+  wget -qO "${tmp}" "${url}"
+}
+
+taggery_page() {
+  [ "${1}" ] || return 1
+  local id="${1}"
+  kvs_has_id "${id}" || taggery_init_page "${id}"
+  local tmp="$(kvs_get "${id}" tmp)"
+  [ -s "${tmp}" ] || taggery_fetch "${id}"
+  cat "${tmp}"
+}
+
+taggery_image_url() {
+  [ "${1}" ] || return 1
+  local id="${1}"
+  taggery_page "${id}" | "${taggery_image_url_filter}"
+}
+
+taggery_image_tags() {
+  [ "${1}" ] || return 1
+  local id="${1}"
+  taggery_page "${id}" | "${taggery_image_tags_filter}"
+}
-- 
2.45.0