shi/shi2.sh

#!/bin/sh
# TODO: gif support and thumbnails if Imagemagick is installed
# TODO: More comments
# TODO: discard reply if parent doesnt exist, right now shi hangs

backend="./http_backend.sh"
db_backend="./file_db_backend.sh"

. ./globals.sh
. ./util.sh
. "$backend"
. "$db_backend"


sanitize_text() {
	# Sanitizes text, the first argument is the text to be
	# sanitized, the rest represent what sanitations should be applied
	# (html, sed, etc.)

	# sed sanitation will prepend a leading backslash to sed characters
	# html sanitation will escape '&', '<', and '>' and replace newlines 
	# with '<br>' tags

	# NOTE: 'sed' must be placed before 'html" in the options if they
	# are both specified (This should be fixed later)

	local text="$1" # text to be sanitized
	
	for option in "$@" # parse options to find specified sanitations
	do
		case "$option" in "$text")
			;; # if it is the first option, ignore it
		"sed")
			echo "sanitizing text $text for $option" >> $log_file
			text="$(echo "$text" \
				| sed -e 's/&/&amp;/g' \
				      -e 's/</\&lt;/g' \
				      -e 's/>/\&gt;/g' \
				      -e 's/\([-$^*()+{\[\.?\/]\)/\\\1/g')"
			;;
		"html")
			echo "sanitizing text $text for $option" >> $log_file
			text="$(echo "$text" \
				| sed -e ':a;N;$!ba;s/\n/<br>/g')"
			;;
		esac
	done
	
	echo "$text"
}

get_op_post_html() {
	# Get the html of an op post if it has already been generated
	local temp_file="$(mktemp)"
	local post_db_path="$1"
	local post_html_path="$(get_post_data "$post_db_path" "html_path")" 
	local post_id="$(get_post_data "$post_db_path" "post_id")"
	local trim="$2" # how many replies to get
	local post_replies="$(get_post_data "$post_db_path" "replies")"
	local reply_num="$(echo "$post_replies" | wc -l)" # number of replies to the post


	cat "$post_html_path" \
	| sed -e ':a' -e 'N' -e '$!ba' \
	      -e "s/.*\(<!--BEGIN OP POST $post_id-->.*<!--END OP POST $post_id-->\).*/\1/g" \
	> "$temp_file"
	
	if [ $reply_num -gt $trim ]
	then
		# id of the most recent reply to be included on the board page
		local reply_lim="$(tac "$post_html_path" \
			| grep -E "<!--END REPLY [0-9]+-->" \
			| head -n "$trim" \
			| tail -n1 | grep -Eo '[0-9]+')"


		sed -i -e ':a' -e 'N' -e '$!ba' \
			-e "s/<!--BEGIN REPLY [0-9]\+-->.*<!--BEGIN REPLY $reply_lim-->/<!--BEGIN REPLY $reply_lim-->/g" \
			"$temp_file"

		cat "$temp_file"
	else 
		cat "$temp_file"
	fi


	rm "$temp_file"
}

gen_post_html() {
	# Takes a path to a database entry and generates html
	# based on it's content
	
	# TODO: implement thumbnails	
	# TODO: separate html frontend to enable creation of other frontends
	# eval

	local post_db_path="$1" # db path to post
	local template_path="" # filesystem path to the post template to be used

	echo "[Genereate post html]" >> "$log_file"
	
	echo "Generating post html for $post_db_path" >> "$log_file"
	local date="$(date -u)"
	local user="$(get_post_data "$post_db_path" "user")"
	local title="$(get_post_data "$post_db_path" "title")"
	local content="$(get_post_data "$post_db_path" "content")"
	local image_path="$(get_post_data "$post_db_path" "image_path")"
	local image_size=""
	local image_name=""
	local post_id="$(get_post_data "$post_db_path" "post_id")"
	local post_type="$(get_post_data "$post_db_path" "post_type")" 


	# http path to post directory (generated from html path)
	local post_url_path="$(get_post_data "$post_db_path" "html_path")" 
	post_url_path="$(dirname "$post_url_path")"

	# http path to image
	local image_url_path=""

	# http path to thumbail
	local thumb_url_path=""

	# truncate the url path to start at $board_dir
	case "$post_type" in "oppost")
		post_url_path="$(echo "$post_url_path" \
				      | rev \
				      | cut -f -3 -d '/' \
				      | rev)" 
		;;
	"reply")
		post_url_path="$(echo "$post_url_path" \
				      | rev \
				      | cut -f -4 -d '/' \
				      | rev)" 
		;;
	esac

	post_url_path="$shi_sub_url/$post_url_path"
	
	if [ -z "$image_path" ] # if no image is found
	then
		post_type="$post_type"-noimage # specify the post type as imageless
		echo "Image path is null, -noimage post" >> "$log_file"
	else
		image_size="$(du -h "$image_path" | cut -f 1)"
		image_name="$(basename "$image_path")"
		image_url_path="$post_url_path/$image_name"
		thumb_url_path="$post_url_path/thumb_$image_name.jpg"
	fi

	template_path="$template_dir/$post_type".template

	content="$(sanitize_text "$content" sed html)" # sanitize content
	title="$(sanitize_text "$title" sed)" # sanitize title for sed
	post_url_path="$(sanitize_text "$post_url_path" sed)" # sanitize post url for sed
	image_url_path="$(sanitize_text "$image_url_path" sed)" # sanitize image path for sed
	thumb_url_path="$(sanitize_text "$thumb_url_path" sed)" # sanitize image path for sed

	# read through the appropriate template and replace 
	# placeholders with their actual values
	while read -r line 	
	do
		echo "$line" \
		| sed -e "s/{{{ POSTNUM }}}/$post_id/g" \
		      -e "s/{{{ POSTURL }}}/$post_url_path/g" \
		      -e "s/{{{ DATE }}}/$date/g" \
		      -e "s/{{{ USER }}}/$user/g" \
		      -e "s/{{{ POSTTITLE }}}/$title/g" \
		      -e "s/{{{ POSTTEXT }}}/$content/g" \
		      -e "s/{{{ IMAGEPATH }}}/$image_url_path/g" \
		      -e "s/{{{ THUMBPATH }}}/$thumb_url_path/g" \
		      -e "s/{{{ IMAGENAME }}}/$image_name/g" \
		      -e "s/{{{ IMAGESIZE }}}/$image_size/g"
	done < "$template_path"
}

insert_post() {
	# TODO: cache (?) the position of <!--BEGIN POSTS--> in the html
	# currently a large and exponential slowdown

	# read from a cache file, if it doesn't exist, make it

	local post_db_path="$1" # path to the post data
	local target_html="$2" # the html to be modified
	local trim="3" # how many replies to display on the main board page

	echo "[Insert post into html]" >> "$log_file"
	echo "Inserting post $post_db_path into $target_html" >> "$log_file"

	local post_id="$(get_post_data "$post_db_path" "post_id")"
	local post_type="$(get_post_data "$post_db_path" "post_type")"
	local post_replies=""
	local parent=""
	local temp_file="$(mktemp)"

	local replace="" # regex to find the exitsing post's html
	local insert_at="" # html will be inserted after the line containing this string
	
	case "$post_type" in "oppost")
		replace="<!--BEGIN OP POST $post_id-->.*<!--END OP POST $post_id-->"	
		insert_at="<!--BEGIN POSTS-->"
		;;
	"reply")
		parent="$(get_post_data "$post_db_path" "parent")"
		replace="<!--BEGIN REPLY $post_id-->.*<!--END REPLY $post_id-->\n"
		# find existing reply end tags in the parent post html
		insert_at="$(cat "$target_html" | grep -E "<!--END REPLY [0-9]+-->")" 
		echo "inserting at $insert_at" >> "$log_file"
		if [ -n "$insert_at" ] # if reply end tags are found
		then
			# set `insert_at` to the end tag of the lowermost reply
			insert_at="$(echo "$insert_at" | tail -n1)" 
		else
			# else set `insert_at` to beginning of replies
			insert_at="<!--BEGIN REPLIES $parent-->"
		fi
		;;
	esac
		echo "inserting at $insert_at" >> "$log_file"

	# remove any existing instances of the post in the html
	sed -i \
	-e ':a' \
	-e 'N' \
	-e '$!ba' \
	-e "s/$replace//g" "$target_html"


	while read -r line
	do
		echo "$line"
		if echo "$line" | grep -qE "$insert_at"
		then
			echo "Inserting html at line $line matching $insert_at" >> "$log_file"
			if [ "$post_type" = "oppost" ]
			then
				post_replies="$(get_post_data "$post_db_path" "replies")"
				if [ -n "$post_replies" ]
				then
					get_op_post_html "$post_db_path" "$trim"
				else
					gen_post_html "$post_db_path"
				fi
			else
				gen_post_html "$post_db_path"
			fi
		fi
			
	done < "$target_html" > "$temp_file"
	cat "$temp_file" > "$target_html"
	
	
#	NOTE: This section only applies when regenerating an entire board
#	If the post is an oppost, we will need to also generate the replies
#	if [ "$post_type" = "oppost" ]
#	then
#		# NOTE: Very redundant, find a way to process both files at once
#		post_replies=$(get_post_data "$post_db_path" "replies" | tac)
#		for reply_db_path in $post_replies
#		do
#			echo "Inserting "$reply_db_path" into post html" >> "$log_file"
#			insert_post "$reply_db_path" "$(get_post_data "$post_db_path" "html_path")"
#		done
#		
#		# NOTE: This will break if the board directory has spaces
#		for reply_db_path in $(echo $post_replies | cut -f "$trim"- -d ' ')
#		do
#			echo "Inserting "$reply_db_path" into board html" >> "$log_file"
#			insert_post "$reply_db_path" "$target_html"
#		done
#	fi
}

start_listener() {
	# This function reads a continuous stream of data
	# (provided by the `listen` function from a user
	# specified backend) and generates posts from the
	# data it receives.

	local status=0 # 0 is normal, 1 is receiving post data

	local board=""
	local parent=""
	local user=""
	local title=""
	local content=""
	local image_name=""
	local image_content=""
	local post_id=""
	local post_type=""
	local post_html_path=""

	local latest=""
	
	listen | while read -r line
	do
	
		if [ "$line" = "End new post" ]
		then
			echo [New post] >> "$log_file"
			status=0
			latest="$(cat "$latest_file")"
			post_id="$(( latest + 1 ))"
			post_db_path="$(create_post \
				"$board" \
				"$parent" \
				"$user" \
				"$title" \
				"$content" \
				"$image_name" \
				"$image_content" \
				"$post_id" \
				"$post_type")"

			post_html_path="$(get_post_data "$post_db_path" "html_path")"
			parent_html_path=""
			parent_db_path=""
			
			
			if [ "$post_type" == "oppost" ]
			then
				cp "$board_dir/$board/index.html.template" "$post_html_path"
				cp "$template_dir/main.css.template" "$(dirname "$post_html_path")/main.css"
				insert_post "$post_db_path" "$post_html_path"
				insert_post "$post_db_path" "$board_dir/$board/index.html"
			else
				parent_html_path="$board_dir/$board/$parent/index.html"
				parent_db_path="$board_dir/$board/$parent"
				insert_post "$post_db_path" "$parent_html_path"
				insert_post "$parent_db_path" "$board_dir/$board/index.html"
			fi

			
			echo "$post_id" > "$latest_file" # update the latest post file
			echo "Finished adding post $post_id at $(date)"
		fi
	
		if [ "$status" -eq 1 ]
		then
			key="$(echo "$line" | cut -f 1 -d :)" value="$(echo "$line" | cut -f 2- -d :)"
			
			echo "Reading key $key from backend" >> "$log_file"
			case "$key" in "board")
				board="$value"
				;;
			"parent")
				parent="$value"
				if [ -z "$parent" ]
				then
					post_type="oppost"	
				else
					post_type="reply"
				fi
				;;
			"user")
				user="$value"
				;;
			"title")
				title="$value"
				;;
			"content")
				content="$value"
				;;
			"image_name")
				image_name="$value"
				;;
			"image_content")
				image_content="$value"
				;;
			esac
		fi

		if [ "$line" = "New post" ]
		then
			echo "Processing new post" >> "$log_file"
			status=1
		fi
	done
}

trap break EXIT INT TERM