#!/bin/sh
set -u

# root directory of the cryptobox development environment
ROOT_DIR="$(dirname $(dirname $0))"

# retrieve these pages from the wiki
PAGES="CryptoBox CryptoBoxDev CryptoBoxKonzept CryptoBoxEn"

# base URL
WIKI_HOST="https://systemausfall.org"
# the trailing slash is important
WIKI_URL="/wikis/howto/"
# header and footer
HEADER_FILE="$ROOT_DIR/tools/doc_header.inc"
FOOTER_FILE="$ROOT_DIR/tools/doc_footer.inc"

CBOX_CGI="/cryptobox?action=doc\&page="

DEST_DIR="$ROOT_DIR/cbox-tree.d/usr/share/doc/cryptobox/html"
IMAGE_DIR="$ROOT_DIR/cbox-tree.d/var/www/cryptobox-img"
TMP_DIR=/tmp/$(basename $0)-$$.d

[ ! -e "$DEST_DIR" ] && echo "$DEST_DIR does not exist" && exit 1

for PAGE in $PAGES; do
	PAGE_SRC="$WIKI_HOST$WIKI_URL$PAGE"
	echo "Importing $PAGE:"

	TMP_FILE=$TMP_DIR/${PAGE}.html
	mkdir -p "$TMP_DIR"

	echo "	downloading the page ..."
	wget --quiet --output-document="$TMP_FILE" "$PAGE_SRC" || { echo "Downloading ($PAGE_SRC) failed!"; exit 1; }

	echo "	removing header and footer ..."
	# break lines before start of content
	sed -i 's#<div [^>]* id="content" [^>]*>#_END_OF_HEADER_\n#' "$TMP_FILE"
	sed -i 's#<div id="footer">#\n_START_OF_FOOTER_#' "$TMP_FILE"
	# remove all lines before and after "body"
	sed -i '1,/_END_OF_HEADER_/d; /_START_OF_FOOTER_/,$d' "$TMP_FILE"

	#echo "	adding header and footer ..."
	#[ -e "${TMP_FILE}.0" ] && rm "${TMP_FILE}.0"
	#mv "$TMP_FILE" "${TMP_FILE}.0"
	#cat "$HEADER_FILE" "${TMP_FILE}.0" "$FOOTER_FILE" >"$TMP_FILE"
	#rm "${TMP_FILE}.0"

	echo "	downloading requisites ..."
	wget --quiet --ignore-tags=a --no-clobber --page-requisites --convert-links --no-parent --no-directories --base="$WIKI_HOST$WIKI_URL" --directory-prefix="$TMP_DIR" --html-extension --force-html --input-file="$TMP_FILE" || { echo "Downloading requisites for ($PAGE_SRC) failed!"; exit 1; }

	echo "	adjusting links for requisites ..."
	find "$TMP_DIR" -type f | grep -v '\.html$' | while read a
	  do	fname=$(basename "$a")
	  	sed -i "s#=\"[^\"]*/$fname\"#=\"/cryptobox-img/$fname\"#g" "$TMP_FILE"
	 done

	echo "	adjusting wiki links ..."
	# redirect wiki links to cryptobox cgi
	sed -i "s#=\"$WIKI_URL\([^\.]*\)\"#=\"$CBOX_CGI\1\"#g" "$TMP_FILE"

	echo "	finished!"
  done

mv "$TMP_DIR"/*.html "$DEST_DIR"
mv "$TMP_DIR"/* "$IMAGE_DIR"
rmdir "$TMP_DIR"