bin/comics: add script for updating xkcd, add $state_file config

xkcd is special (read: other than all others scripts) because it does not
contain a timestamp information in the comics. we have to track the last
script that was added to our mirror. so the xkcd script needs a state-file
containing the last xkcd that was mirrored. the date of the xkcd is the
day of the first discovery of a new comic.

2 files changed, 54 insertions, 1 deletions

diff --git a/bin/comics/xkcd b/bin/comics/xkcd
new file mode 100755
index 0000000..5af680b
--- /dev/null
+++ b/bin/comics/xkcd
@@ -0,0 +1,45 @@
+#!/bin/bash
+
+source $(pwd)/$(dirname $0)/../../etc/settings.sh
+tmp=$(mktemp -d)
+cd "${tmp}"
+
+browser="Mozilla/4.76 [de] (X11; U; Linux 2.2.18 i586)"
+newn="${day}.png"
+
+wget ${wget_args} -U "$browser" http://www.xkcd.com -O index.html
+
+url=$(grep -A 1 "Image URL" index.html | sed 'N;s/.*\(http.*png\)<\?.*/\1/')
+imgname=$(echo "$url" | tr '/' '\n' | tail -1)
+
+# get additional information
+img_tag=$(grep -A2 "src=\"${url}" index.html | sed 'N;N;s/.*\(<img [^>]\+>\).*/\1/')
+title=$(sed 's/.*title="\([^"]*\)".*/\1/'<<< $img_tag)
+alt=$(sed 's/.*alt="\([^"]*\)".*/\1/'<<< $img_tag)
+
+if [ ! -s "${image_dir}/$newn" ]; then
+    # no image availalable for current date
+
+    if ! grep -q "${url}" "${state_file}" >/dev/null; then
+        # new image
+        echo $url > "${state_file}"
+
+        wget ${wget_args} -U "$browser" -O "${tmp}/${newn}" \
+            --header="Referer: http://www.xkcd.com/" "$url"
+
+        if [ -s "${tmp}/${newn}" ]; then
+            # save
+            mv "${tmp}/${newn}" "${image_dir}/$newn"
+            echo "$title" > "${image_dir}/${day}.title"
+            echo "$alt" > "${image_dir}/${day}.alt"
+
+            # update symlinks
+            rm -f "${comic_dir}/latest.png" "${comic_dir}/latest.title" "${comic_dir}/latest.alt"
+            ln -s "${image_offset}/$newn" "${comic_dir}/latest.png"
+            ln -s "${image_offset}/${day}.title" "${comic_dir}/latest.title"
+            ln -s "${image_offset}/${day}.alt" "${comic_dir}/latest.alt"
+        fi
+    fi
+fi
+
+rm -rf ${tmp}
diff --git a/etc/settings.sh b/etc/settings.sh
index d1f5486..d837e7d 100644
--- a/etc/settings.sh
+++ b/etc/settings.sh
@@ -11,7 +11,11 @@ comic=$(basename $0)
 year=$(date -d"$daysago days ago" +%Y)
 month=$(date -d"$daysago days ago" +%Y-%m)
 day=$(date -d"$daysago days ago" +%d)
-comic_dir=$(pwd)/$(dirname $0)/../../htdocs/imgs/${comic}
+
+top="$(pwd)/$(dirname $0)/../../"
+state_dir=${top}/var/state/
+state_file=${state_dir}/${comic}.state
+comic_dir=${top}/htdocs/imgs/${comic}
 image_offset=${month}/
 image_dir=${comic_dir}/${image_offset}
 
@@ -19,4 +23,8 @@ if [ ! -d "${image_dir}" ]; then
     mkdir -p "${image_dir}"
 fi
 
+if [ ! -d "${state_dir}" ]; then
+    mkdir -p "${state_dir}"
+fi
+
 wget_args="--timeout 100 --wait=1000 --no-cache"