#!/bin/bash
# Parse Tellico database to BioOSM index.htm
# $1 = tellico .tc file
# Print result to stdout
#
# <iframe width="500" height="500" frameborder="0" scrolling="no" marginheight="0" marginwidth="0" src="https://www.openstreetmap.org/export/embed.html?bbox=14.978000%2C49.990000%2C15.022000%2C50.010000&amp;layer=mapnik&amp;marker=50.000000%2C15.000000"></iframe>
# // vectorLayer.removeFeatures(atol)
# // map.removeLayer(vectorLayer)
#
#
# Creates:
#   (to stdout) HTML file with the map and static entries
#   ./det/ID.htm details of each and every entry
#   ./bioosm.json JSON of all entries
#   ./bioosm.rss updated RSS feed
#   ./plainlog.htm sorted entries for easy human/alg
#

rm -rfv "./det/*"
mkdir -p "./det"

ICON_animalia="animalia.png"
ICON_bacteria="bacteria.png"
ICON_fungi="fungi.png"
ICON_chromista="chromista.png"
ICON_plantae="plantae.svg"
ICON_plostenkoid="plostenkoid.png"
ICON_protista="protista.png"
ICON_biohazard="biohazard.svg"

json=""

rss=$'<?xml version="1.0" encoding="UTF-8" ?>\n'
rss+=$'<rss version="2.0">\n'
rss+=$'<channel>\n'
rss+=$'<title>Brmlab BioOSM RSS feed</title>\n'
rss+=$'<description>New entries on BioOSM database</description>\n'
rss+=$'<link>https://bioosm.s0c4.net</link>\n'
rss+=$'<lastBuildDate></lastBuildDate>\n'
rss+=$'<pubDate></pubDate>\n'
rss+=$'<ttl>86400</ttl>\n'

# Return icon by Regnum
function geticon () {
	case "$1" in
		"Animalia")
			echo "$ICON_animalia";;
		"Bacteria")
			echo "$ICON_bacteria";;
		"Fungi")
			echo "$ICON_fungi";;
		"Chromista")
			echo "$ICON_chromista";;
		"Plantae")
			echo "$ICON_plantae";;
		"Protista")
			echo "$ICON_protista";;
		*)
			echo "$ICON_biohazard";;
	esac
}

function safehtml () {
	echo "$1" | sed -e "s/'/\&#39;/g" -e 's/"/\&#34;/g' -e 's/</\&lt;/g' -e 's/>/\&gt;/g' -e 's/^[ ]*//g' -e 's/[ ]*$//g'
}

i=1 # GLOBAL increment must persist multiple spagety() calls; incremented to infinity
# Core function generating entries matching selection criteria (ie Kingdom)
# !!! BEZ ODSAZENI !!!
# $1 = list of entries
# $2 = layer name
# $3 = icon of the points
function spagety () {

	bylo="" # Minula pozice
	sam="" # HTML samostatneho zaznamu
	osam="" # predchozi $sam
	spoj="" # HTML had z popisu stejnych pozic
	ospoj="" # Predchozi $spoj
	ox="" # Old X
	oy="" # Old Y
	datelog="" # Entries sorted by date for plaintext log
	json="var data='{\"version\":\"$(date --rfc-3339=seconds)\",\"arma\":["

	while IFS='' read -r radek || [[ -n "$radek" ]]; do
		co=$(echo "$radek" | grep -oe ' <title>.*</title> ' | sed -E 's/( )*<(\/)?title>( )*//g')
		co=$(safehtml "$co")
		obr=$(echo "$radek" | grep -oe ' <image>.*</image> ' | sed -E 's/( )*<(\/)?image>( )*//g' | grep -oPe '[^/]*$')
		obr=$(safehtml "$obr")
		kde=$(echo "$radek" | grep -oe '<locality>.*</locality>' | grep -oPe '[0-9]+\.[0-9]{1,10}[, ]*[0-9]+\.[0-9]{1,10}')
		kde=$(safehtml "$kde")
		kdex=$(echo "$kde" | grep -oPe '[0-9]+\.[0-9]{1,10}$')
		kdex=$(safehtml "$kdex")
		kdey=$(echo "$kde" | grep -oPe '^[0-9]+\.[0-9]{1,10}')
		kdey=$(safehtml "$kdey")
		kdy=$(echo "$radek" | grep -oe ' <taken-date>.*</taken-date> ' | sed -E 's/( )*<(\/)?taken-date>( )*//g' | sed -E 's/([0-9]+)\.([0-9]+)\.([0-9]{4})/\3-\2-\1/g' | sed -E 's/-([0-9])-/-0\1-/g' | sed -E 's/-([0-9])$/-0\1/g')
		kdy=$(safehtml "$kdy")
		skop_typ=$(echo "$radek" | grep -oe ' <new-field-1>.*</new-field-1> ' | sed -E 's/( )*<(\/)?new-field-1>( )*//g')
		skop_typ=$(safehtml "$skop_typ")
		skop=$(echo "$radek" | grep -oe ' <microscope>.*</microscope> ' | sed -E 's/( )*<(\/)?microscope>( )*//g')
		skop=$(safehtml "$skop")
		imp=$(echo "$radek" | grep -oe ' <new-field-3>.*</new-field-3> ' | sed -E 's/( )*<(\/)?new-field-3>( )*//g')
		imp=$(safehtml "$imp")
		reg=$(echo "$radek" | grep -oe ' <new-field-4>.*</new-field-4> ' | sed -E 's/( )*<(\/)?new-field-4>( )*//g')
		reg=$(safehtml "$reg")
		phy=$(echo "$radek" | grep -oe ' <phylum>.*</phylum> ' | sed -E 's/( )*<(\/)?phylum>( )*//g')
		phy=$(safehtml "$phy")
		cla=$(echo "$radek" | grep -oe ' <class>.*</class> ' | sed -E 's/( )*<(\/)?class>( )*//g')
		cla=$(safehtml "$cla")
		ord=$(echo "$radek" | grep -oe ' <order>.*</order> ' | sed -E 's/( )*<(\/)?order>( )*//g')
		ord=$(safehtml "$ord")
		fam=$(echo "$radek" | grep -oe ' <family>.*</family> ' | sed -E 's/( )*<(\/)?family>( )*//g')
		fam=$(safehtml "$fam")
		gen=$(echo "$radek" | grep -oe ' <genus>.*</genus> ' | sed -E 's/( )*<(\/)?genus( )*>//g')
		gen=$(safehtml "$gen")
		spe=$(echo "$radek" | grep -oe ' <species>.*</species> ' | sed -E 's/( )*<(\/)?species>( )*//g')
		spe=$(safehtml "$spe")
		loc=$(echo "$radek" | grep -oe ' <locality---plaintext>.*</locality---plaintext> ' | sed -E 's/( )*<(\/)?locality---plaintext>( )*//g')
		loc=$(safehtml "$loc")
		des=$(echo "$radek" | grep -oe ' <description>.*</description> ' | sed -E 's/( )*<(\/)?description>( )*//g')
		des=$(safehtml "$des")
		vid=$(echo "$radek" | grep -oe ' <videolink>.*</videolink> ' | sed -E 's/( )*<(\/)?videolink>( )*//g')
		vid=$(safehtml "$vid")
		bio=$(echo "$radek" | grep -oe ' <biotope>.*</biotope> ' | sed -E 's/( )*<(\/)?biotope>( )*//g')
		bio=$(safehtml "$bio")
		eng=$(echo "$radek" | grep -oe ' <enlargement>.*</enlargement> ' | sed -E 's/( )*<(\/)?enlargement>( )*//g')
		eng=$(safehtml "$eng")
		obs=$(echo "$radek" | grep -oe ' <observer>.*</observer> ' | sed -E 's/( )*<(\/)?observer>( )*//g')
		obs=$(safehtml "$obs")
		lnk=$(echo "$radek" | grep -oe ' <localitylink>.*</localitylink> ' | sed -E 's/( )*<(\/)?localitylink>( )*//g')
		lnk=$(safehtml "$lnk")
		har=$(echo "$radek" | grep -oe ' <harvesting-method>.*</harvesting-method> ' | sed -E 's/( )*<(\/)?harvesting-method>( )*//g')
		den=$(echo "$radek" | grep -oe ' <population-density>.*</population-density> ' | sed -E 's/( )*<(\/)?population-density>( )*//g')
		dur=$(echo "$radek" | grep -oe ' <min>.*</min> ' | sed -E 's/( )*<(\/)?min>( )*//g')
		lor=$(echo "$radek" | grep -oe ' <locality-reference>.*</locality-reference> ' | sed -E 's/( )*<(\/)?locality-reference>( )*//g')
		lrl=$(echo "$radek" | grep -oe ' <locality-reference---link>.*</locality-reference---link> ' | sed -E 's/( )*<(\/)?locality-reference---link>( )*//g')
		lov=$(echo "$radek" | grep -oe ' <locality---vobrazek>.*</locality---vobrazek> ' | sed -E 's/( )*<(\/)?locality---vobrazek>( )*//g' | grep -oPe '[^/]*$')

		# >&2 echo "$co - $har - $den - $dur - $lor - $lrl - $lov"

		detid=$(echo "brmlab$co$obr$kde$kdy" | md5sum | grep -oe '[a-zA-Z0-9]*')

		if [[ "$kdex" = "" ]]; then
			>&2 echo -e "\033[31m" $(echo "$radek" | grep -oe '<locality>.*</locality>') " $co\033[0m"
			continue;
		fi
		zkurvenetellico=$(echo "$obr" | grep -oe '\.[a-zA-Z]*$')
		if [[ "$zkurvenetellico" = ".jpeg" ]]; then
			>&2 echo -e "\033[31m$obr $co\033[0m"
			continue;
		fi

		if [[ "$spe" = "sp." ]]; then
			wikisp="https://en.wikipedia.org/wiki/$gen"
		else
			wikisp="https://en.wikipedia.org/wiki/$gen $spe"
		fi

		json+="{"
		json+=" \"qualis\":\"$co\","
		json+=" \"imperium\":\"$imp\","
		json+=" \"regnum\":\"$reg\","
		json+=" \"phylum\":\"$phy\","
		json+=" \"classis\":\"$cla\","
		json+=" \"ordo\":\"$ord\","
		json+=" \"familia\":\"$fam\","
		json+=" \"genus\":\"$gen\","
		json+=" \"specie\":\"$spe\","
		json+=" \"locus\":\"$kde\","
		json+=" \"x\":\"$kdex\","
		json+=" \"y\":\"$kdey\","
		json+=" \"locplain\":\"$loc\","
		json+=" \"depictio\":\"$des\","
		json+=" \"quando\":\"$kdy\","
		json+=" \"obr\":\"$obr\","
		json+=" \"micro\":\"$skop\","
		json+=" \"microtyp\":\"$skop_typ\","
		json+=" \"vid\":\"$vid\","
		json+=" \"bio\":\"$bio\","
		json+=" \"amp\":\"$eng\","
		json+=" \"spec\":\"$obs\","
		json+=" \"detid\":\"$detid\","
		json+=" \"har\":\"$har\","
		json+=" \"den\":\"$den\","
		json+=" \"dur\":\"$dur\","
		json+=" \"lor\":\"$lor\","
		json+=" \"lrl\":\"$lrl\","
		json+=" \"lov\":\"$lov\","
		json+=" \"lnk\":\"$lnk\"" # Lastone without ','
		json+="},"


		rss+=$'<item>\n'
		rss+=$'  <title>'"$co"$'</title>\n'
		rss+=$'  <description>'"$des"$'</description>\n'
		rss+=$'  <link>http://brmlab.s0c4.net/bioosm/det/'"$detid"$'.htm</link>\n'
		rss+=$'  <guid isPermaLink="false">'"$detid"$'</guid>\n'
		rss+=$'  <pubDate>'"$kdy"$'</pubDate>\n'
		rss+=$'</item>\n'


		# Generate detail page
		detid=$(echo "brmlab$co$obr$kde$kdy" | md5sum | grep -oe '[a-zA-Z0-9]*')
		detpg="<html><head>"
		detpg+="<meta http-equiv='Content-Type' content='text/html; charset=utf-8'>"
		detpg+="<link rel='stylesheet' href='../style.css' type='text/css'>"
		detpg+="<title>Brmlab - BioOSM $co $kde</title>"
		detpg+="</head><body>"
		detpg+="<table class='dettbl'><tr>"
		detpg+="<h1>$co (<a href='$wikisp' target='_blank'>wiki</a>)</h1>"
		detpg+="<td style='width:33%;vertical-align:top'>"
		if [[ "$imp" != "" ]]; then
			detpg+="<b>Imperium:</b> <a href='https://en.wikipedia.org/wiki/$imp' target='_blank'>$imp</a><br>"
		fi
		if [[ "$reg" != "" ]]; then
			detpg+="<b>Regnum:</b> <a href='https://en.wikipedia.org/wiki/$reg' target='_blank'>$reg</a><br>"
		fi
		if [[ "$phy" != "" ]]; then
			detpg+="<b>Phylum:</b> <a href='https://en.wikipedia.org/wiki/$phy' target='_blank'>$phy</a><br>"
		fi
		if [[ "$cla" != "" ]]; then
			detpg+="<b>Classis:</b> <a href='https://en.wikipedia.org/wiki/$cla' target='_blank'>$cla</a><br>"
		fi
		if [[ "$ord" != "" ]]; then
			detpg+="<b>Ordo:</b> <a href='https://en.wikipedia.org/wiki/$ord' target='_blank'>$ord</a><br>"
		fi
		if [[ "$fam" != "" ]]; then
			detpg+="<b>Familia:</b> <a href='https://en.wikipedia.org/wiki/$fam' target='_blank'>$fam</a><br>"
		fi
		if [[ "$gen" != "" ]]; then
			detpg+="<b>Genus:</b> <a href='https://en.wikipedia.org/wiki/$gen' target='_blank'>$gen</a><br>"
		fi
		if [[ "$spe" != "" ]]; then
			detpg+="<b>Specie:</b> <a href='$wikisp' target='_blank'>$spe</a><br>"
		fi
		detpg+="<br>"
		if [[ "$kdy" != "" ]]; then
			detpg+="<b>Quando:</b> $kdy<br>"
		fi
		if [[ "$lnk" != "" ]]; then
			loc="<a href=\"$lnk\" target=\"_blank\">$loc</a>" # Yes, escaped "
		fi
		detpg+="<b>Locus:</b> $kde ($loc)<br>"
		if [[ "$obs" != "" ]]; then
			detpg+="<b>Spectator:</b> $obs<br>"
		fi
		if [[ "$bio" != "" ]]; then
			detpg+="<b>Biotope:</b> $bio<br>"
		fi
		if [[ "$skop" != "" ]]; then
			detpg+="<b>Microscopium:</b> $skop ($skop_typ)<br>"
		fi
		if [[ "$eng" != "" ]]; then
			detpg+="<b>Amplificatio:</b> $eng<br>"
		fi
		if [[ "$vid" != "" ]]; then
			detpg+="<b>Video:</b> <a href=\"$vid\" target=\"_blank\">$vid</a><br>"
		fi
		if [[ "$har" != "" ]]; then
			detpg+="<b>Collecting method:</b> $har<br>"
		fi
		if [[ "$den" != "" ]]; then
			detpg+="<b>Individuals found:</b> $den<br>"
		fi
		if [[ "$dur" != "" ]]; then
			detpg+="<b>Duration on stage:</b> $dur<br>"
		fi
		if [[ "$lrl" != "" ]]; then
			detpg+="<b>Locality reference:</b> <a href=\"$lrl\" target=\"_blank\">$lor</a><br>"
		elif [[ "$lor" != "" ]]; then
			detpg+="<b>Locality reference:</b> $lor<br>"
		fi
		if [[ "$des" != "" ]]; then
			detpg+="<b>Depictio:</b> $des<br>"
		fi
		detpg+="<br><b>More on <a href=\"../\">BioOSM map</a></b><br>"
		detpg+="<br>"

		lhx=$(echo "$kdex-0.022" | bc)
		lhy=$(echo "$kdey-0.010" | bc)
		pdx=$(echo "$kdex+0.022" | bc)
		pdy=$(echo "$kdey+0.010" | bc)
		detpg+="<iframe id='minimapa' src='https://www.openstreetmap.org/export/embed.html?bbox=$lhx%2C$lhy%2C$pdx%2C$pdy&amp;layer=mapnik&amp;marker=$kdey%2C$kdex'></iframe>"
		detpg+="</td><td style='min-width:66%;vertical-align:top'>"

		if [[ "$obr" != "" ]]; then
			detpg+="<a href=\"../img/$obr\"><img src=\"../img/$obr\" alt=\"$co\" class='detimg'></a>"
		else
			if [[ "$vid" != "" ]]; then
				detyt=$(echo "$vid" | sed -e 's/youtube.com\/watch?v=/youtube.com\/embed\//g')
				detpg+="<iframe class='detvid' src='$detyt' frameborder='0' allow='encrypted-media' allowfullscreen></iframe>"
			else
				detpg+="<img src=\"../$3\" alt=\"video\" class='detimg'>"
			fi
		fi
		if [[ "$lov" != "" ]]; then
			detpg+="<img class=\"detimg\" src=\"../lokality/$lov\" alt=\"Locality photo\"><br>"
		fi

		detpg+="</td></tr></table>"
		detpg+="</body></html>"
		echo "$detpg" > "./det/$detid.htm"

		# Plain log of entries
		datelog+="$kdy $kde <a href='./det/$detid.htm' target='_blank'>$co</a><br>\n"

		>&2 echo -e "\033[32m$co $obs\033[0m"

		ox="$kdex"
		oy="$kdey"

		i=$(($i+1))

	done <<< $(echo "$1") # Zere parametr spaget, nikoli $1 scriptu !!!
	echo "$json{}]}" | sed -e "s/^var data='//g" > "./bioosm.json"
	echo "$json{}]}'"

	rss+=$'</channel>\n'
	rss+=$'</rss>\n'
	echo "$rss" > "./bioosm.rss"

} # KONEC FCE SPAGETY



# Parse Tellico file to list of entries
praseTC=$(unzip -p "$1" tellico.xml | tr -d '\n' | sed -e $'s/<entry id="/\\\n<entry id="/g' -e $'s/<\/entry>/<\/entry>\\\n/g' | grep -Pe '^<entry id="' | grep "<strcit-do-mapy>yep</strcit-do-mapy>" | sed -E 's/(^.*)(<locality>.*<\/locality>)(.*$)/\2\1\3/' | sort)

# Sorted unique Kingdoms (to be passed as selection criteria)
kingdoms=$(echo "$praseTC" | grep -oe ' <new-field-4>.*</new-field-4> ' | sed -E 's/( )*<(\/)?new-field-4>( )*//g' | sort | uniq)

echo '<!DOCTYPE html>'
echo '<html>'
echo '<head>'
echo '	<meta http-equiv="Content-Type" content="text/html; charset=utf-8">'
echo '	<title>brmlab - BioOSM</title>'
echo '	<link rel="stylesheet" href="./style.css" type="text/css">'
echo '	<link rel="favicon" href="./'$ICON_biohazard'">'
echo '	<script src="./OpenLayers.js"></script>'
echo '</head>'
echo '<body>'
# Generate select
echo "<div id='plainlog'><b><a href='./bioosm.rss'>RSS feed</a><br><a href='./plainlog.htm'>Quick news log</a></b></div>"
echo "<h1 id='title'>BioOSM - arma biologica in tuis vicinia</h1>"
ksel="<select id='reg' onChange='genphy()'>"
ksel="$ksel<option value='ALL' selected>ALL</option>"
while IFS='' read -r king || [[ -n "$king" ]]; do
	ksel="$ksel<option value='"$king"'>$king</option>"
done <<< $(echo "$kingdoms")
ksel="$ksel</select>"
echo "Regnum: $ksel "

echo "Phylum: <select id='phy' onChange='gencla()'><option>ALL</option></select> "
echo "Classis: <select id='cla' onChange='genord()'><option>ALL</option></select> "
echo "Ordo: <select id='ord' onChange='genfam()'><option>ALL</option></select> "
echo "Familia: <select id='fam' onChange='gengen()'><option>ALL</option></select> "
echo "Genus: <select id='gen' onChange='genspe()'><option>ALL</option></select>"
echo "Specie: <select id='spe' onChange='armageddon()'><option>ALL</option></select>"
echo "<br><input type='button' value='Platyhelminthes' onClick='plostenkoidy()'>"
echo "<input type='checkbox' id='studchecker' onChange='studdrv()'>&nbsp;Springs"
#echo "<input type='button' value='Bugs' onClick='buggy()'>"
echo "<br><br><div id='map' class='smallmap'></div>"
echo "<br><p>Welcome! You can download the <a href='./bioosm.json' target='_blank'>source data as JSON file</a><br>"
echo "This site is owned by members of the <a href='https://brmlab.cz' target='_blank'>hackerspace brmlab</a>. See our wiki page for <a href='https://brmlab.cz/project/bioosm/start' target='_blank'>details and contact info</a><br>License: CC-BY-SA</p>"
echo "</body>"

echo "<script>"
echo 'map = new OpenLayers.Map("map");'
echo 'map.addLayer(new OpenLayers.Layer.OSM());'
echo 'epsg4326 =  new OpenLayers.Projection("EPSG:4326"); //WGS 1984 projection'
echo 'projectTo = map.getProjectionObject(); //The map projection (Spherical Mercator)'
echo 'var lonLat = new OpenLayers.LonLat(15.478, 49.817).transform(epsg4326, projectTo);'
echo 'var zoom=8;'
echo 'map.setCenter (lonLat, zoom);'

echo 'var vectorLayer = new OpenLayers.Layer.Vector("Overlay",{attribution:"<a href=\"http://brmlab.cz\">Brmlab</a>"});'

echo '// Define markers as "features" of the vector layer:'

echo 'var atol = new OpenLayers.Feature.Vector('
echo '	new OpenLayers.Geometry.Point(14.41876, 50.10554).transform(epsg4326, projectTo),'
echo '		{description:"Atomovy atol Brmkini"} ,'
echo '		{externalGraphic: "./biohazard.svg", graphicHeight:8, graphicWidth:8, graphicXOffset:-4, graphicYOffset:-4},'
echo ');'
echo 'vectorLayer.addFeatures(atol);'
# Sem prijdou dalsi staticke ficury
echo "map.addLayer(vectorLayer);"

spagety "$praseTC" "ALL" "$ICON_biohazard"

echo '</script>'
echo '<script src="./bioosm.js"></script>' # Must be after JSON
echo "</html>"

# Create plaintext log
dateloghtm="<html><head>"
dateloghtm+="<meta http-equiv='Content-Type' content='text/html; charset=utf-8'>"
dateloghtm+="<link rel='stylesheet' href='./style.css' type='text/css'>"
dateloghtm+="<title>Brmlab - BioOSM plaintext log</title>"
dateloghtm+="</head><body>"
datelogsort=$(echo -e "$datelog" | sort -r | uniq)
dateloghtm+="$datelogsort"
dateloghtm+="</body></html>"
echo "$dateloghtm" > "./plainlog.htm"