From 146cbbb39eb3affee603c6d54082a7ea3359eb3b Mon Sep 17 00:00:00 2001 From: tv Date: Sun, 4 Sep 2011 20:02:20 +0200 Subject: //meinsack: initial commit --- meinsack/index | 77 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 77 insertions(+) create mode 100755 meinsack/index (limited to 'meinsack/index') diff --git a/meinsack/index b/meinsack/index new file mode 100755 index 00000000..2b3e455a --- /dev/null +++ b/meinsack/index @@ -0,0 +1,77 @@ +#! /bin/sh +set -euf +cd $(dirname $(readlink -f $0))/.. # cd // + +plz=${plz+"&plz=$plz"} +str=${str+"&str=$str"} +anr=${anr+"&anr=$anr"} + +url="http://www.sita-deutschland.de/sita/ywbase_Abfallkalender_Stuttgart.nsf/frmSelect?ReadForm$plz$str$anr" + +# echo $url + +html="$(curl -fsS "$url" | iconv -f latin1 -t utf8)" + +if echo "$html" | fgrep -q 'Ihre Eingabe war nicht eindeutig'; then + { + echo "Error: multiple locations found" + echo "$html" | + query/jquery/index 'table.tab:nth-child(2n) .bordertop > .text' | + sed ' + s/[[:space:]]\+/ /g + s/'"`printf '\xc2\xa0'`"'//g;# kill unicode non-breaking space + ' | + while read plz && + read str && + read ort && + read bes1 && + read bes2 && + read anr; do + str=$(echo "$str"|sed 's/<[^>]*>//g') + echo "export anr=$anr if you meant \"$ort\"" + done | sort | uniq + } >&2 + exit 2 +elif test $(echo "$html" | grep -c 'Abfallkalender für') -gt 1 && + test -z "${ort-}"; then + { + echo "Error: multiple locations found" + echo "$html" | + query/jquery/index '.tab span.text,.subheadline' | + sed -n ' + s/[[:space:]]\+/ /g + s/'"`printf '\xc2\xa0'`"'//g;# kill unicode non-breaking space + s/^Abfallkalender.*[^0-9]\([0-9]\+\)[^(]*(\(.*\))/export ort=\2/p + ' | sort | uniq + } >&2 + exit 3 +else + result="$(echo "$html" | + query/jquery/index '.tab span.text,.subheadline' | { + if test -n "${ort-}"; then + sed -n " + /($ort)/,/^Abfallkalender/{p;b} + /($ort)/,\${p;b} + " + else + cat + fi + } | { + grep -v Abfallkalender | + sed ' + s/'"`printf '\xc2\xa0'`"'//g;# kill unicode non-breaking space + s/^[A-Z][a-z]\.\([0-9]*\)\.\([0-9]*\)\.\([0-9]*\)$/\3-\2-\1/ + s/^[0-9].*/ "&T00:00:00.000Z"/ + $!s/$/,/ + 1s/^/[\n/ + $s/$/\n]/ + ' + })" + if ! echo "$result" | grep .; then + { + echo "Error: your query had no effect" + echo "url=\"$url\"" + } >&2 + exit 4 + fi +fi -- cgit v1.2.3