From: Stefan Metzmacher <metze@samba.org>
Date: Wed, 5 Aug 2015 11:06:22 +0000 (+0200)
Subject: add import_oldnews.sh helper script
X-Git-Url: http://git.samba.org/?a=commitdiff_plain;h=c9f87ffacfb0daf011d8808a557c21f2c89a3e91;p=metze%2Ftest%2Fsamba-web.git

add import_oldnews.sh helper script

This can be used like this:

./import_oldnews.sh generated_news/latest_10_headlines.html generated_news/latest_10_bodies.html

Signed-off-by: Stefan Metzmacher <metze@samba.org>
---

diff --git a/import_oldnews.sh b/import_oldnews.sh
new file mode 100755
index 00000000..f6ab6bc6
--- /dev/null
+++ b/import_oldnews.sh
@@ -0,0 +1,61 @@
+#!/bin/sh
+#
+
+LC_ALL=C
+export LC_ALL
+LANG=C
+export LANG
+LANGUAGE=C
+export LANGUAGE
+
+set -u
+set -e
+umask 0022
+
+HEADLINES=${1}
+BODIES=${2}
+
+LINES="$(cat ${HEADLINES} | grep '<li>')"
+NLINES=$(echo "${LINES}" | wc -l)
+
+BODYH5="$(grep --line-number '<h5>' ${BODIES})"
+
+for i in $(seq 1 $NLINES); do
+	l=$(echo "${LINES}" | head -${i} | tail -1 | sed -e 's!^[\t ]*!!')
+	d=$(echo "${l}" | sed -e 's!^<li>\([^<]*\)<a href=.*!\1!')
+	n=$(echo "${l}" | sed -e 's!.*a href="#\([^"]*\)".*!\1!')
+	utctime=$(date --date="${d}" --utc +"%Y%m%d-%H%M%S")
+	headlinefile="posted_news/${utctime}.${n}.headline.html"
+	bodyfile="posted_news/${utctime}.${n}.body.html"
+	echo "${i}: ${headlinefile}"
+	{
+		echo "<!-- BEGIN: ${headlinefile} -->"
+		echo "${l}"
+		echo "<!-- END: ${headlinefile} -->"
+	} > ${headlinefile}
+
+	boundary=$(echo "${BODYH5}" | grep -A1 "<h5><a name=\"${n}\">" | cut -d ':' -f1)
+	tmp=$(echo "${boundary}" | wc -l)
+	case "${tmp}" in
+	2)
+		begin=$(echo "${boundary}" | head -1)
+		next=$(echo "${boundary}" | tail -1)
+		end=$(expr ${next} - 1)
+		len=$(expr ${end} - ${begin})
+		;;
+	1)
+		len="10000"
+		;;
+	*)
+		echo "invalid boundary:"
+		echo "${boundary}"
+		exit 1
+		;;
+	esac
+	echo "${i}: ${bodyfile}"
+	{
+		echo "<!-- BEGIN: ${bodyfile} -->"
+		grep -A${len} "<h5><a name=\"${n}\">" ${BODIES}
+		echo "<!-- END: ${bodyfile} -->"
+	} > ${bodyfile}
+done