fedi2html/fedi2html.sh

#!/bin/sh
#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
# Name: fedi2html.sh
# Desc: 
# Reqs: 
# Date:
#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――

POST_TEMPLATE='
<article class="comment">
<a class="user" href="$ACCOUNT_URL">
    <img class="avatar" src="$ACCOUNT_AVATAR">
    <strong class="username">$ACCOUNT_NAME</strong>
    <em class="useraddress">$ACCOUNT_ID</em>
</a>
<a class="address" href="$POST_URL">
    <time title="$POST_DATE">$(date --date=$POST_DATE)</time>
</a>
<section>
    $POST_CONTENT
</section>
<div class="attachments">
    $POST_ATTACHMENTS
</div>
<div class="responses">
    $POST_RESPONSES
</div>
</article>
'


ATTACH_TEMPLATE='
<a href="$ATTACH_URL">
    <div class="attachment">
        <strong>$ATTACH_NAME</strong>
        $ATTACH_DESC
    </div>
</a>'

ATTACH_IMAGE_TEMPLATE='
<figure>
    <a href="$ATTACH_URL">
       <img class="attachment" src="$ATTACH_URL" alt="$ATTACH_DESC" title="$ATTACH_DESC">
    </a>
    <figcaption>
        <a href="$ATTACH_URL"><b>$ATTACH_NAME</b></a><br>
        $(if test $(echo "$ATTACH_DESC" | wc -c) -gt 52; then
              echo $ATTACH_DESC | head -c 53 | sed 's%[[:space:]]*$%%'
              echo …
          else
              echo $ATTACH_DESC
          fi)
    </figcaption>
</figure>
'

EMOJI_TEMPLATE='<img class="emoji" src="$EMOJI_URL" alt="$EMOJI_SHORTCODE" title="$EMOJI_SHORTCODE">'


# Given a note’s JSON, render it as HTML.
# The most important part of the script!
#     render_post $post_data $context_data $tree_level
render_post() {
	local post_data="$1"
	local responses_data="$2"
	local POST_TREE_LEVEL="$3"

	local ACCOUNT_URL="$(echo "$post_data"    | jq -r .account.url)"
	local ACCOUNT_ID="$(echo "$post_data"     | jq -r .account.fqn)"
	local ACCOUNT_NAME="$(echo "$post_data"   | jq -r .account.display_name | replace_emojis "$(echo "$post_data" | jq -r '.account')")"
	local ACCOUNT_AVATAR="$(echo "$post_data" | jq -r .account.avatar)"
	local POST_URL="$(echo "$post_data"       | jq -r .url)"
	local POST_DATE="$(echo "$post_data"      | jq -r .created_at)"
	local POST_CONTENT="$(echo "$post_data"   | jq -r .content | replace_emojis "$post_data")"
	local POST_ATTACHMENTS="$(media_attachments "$post_data")"
	local POST_RESPONSES="$(render_responses "$post_data" "$responses_data" "$POST_TREE_LEVEL")"
	env_subst "$POST_TEMPLATE"
}


# Render a post’s responses one-by-one and recursively.
# Each branch of the response tree will be rendered completely before proceeding
# to the next.
#     render_responses $post_data $context_data $tree_level
render_responses() {
	local post_data="$1"
	local responses_data="$2"
	local level="$3"
	if test -z "$level"; then level=0; fi

    local id="$(echo "$post_data" | jq -r '.id')"
	local responses="$(echo "$responses_data" | grep "in_reply_to_id.*$id")"
	local IFS="
"

	for response in $responses; do
		render_post "$response" "$responses_data" "$(expr "$level" + 1)"
	done
}


# Accepts a string over stdin; it will replace all emoji shortcodes along stdin
# input with appropriate <img> HTML, based on $EMOJI_TEMPLATE, and based on
# a post’s JSON.
#     echo ":blobcat:" | replace_emojis $post_data
replace_emojis() {
	local post_data="$1"
	local emojis="$(echo "$post_data" | jq -r '.emojis[]|(.url + "\t" + .shortcode)')"
	local temp="$(mktemp)"
	local IFS="
"
	cat > "$temp"

	for line in $emojis; do
		local EMOJI_URL="$(echo "$line" | awk -F'\t' '{print $1}')"
		local EMOJI_SHORTCODE="$(echo "$line" | awk -F'\t' '{print $2}')"
		local value="$(env_subst "$EMOJI_TEMPLATE")"
        sed -i "s%:${EMOJI_SHORTCODE}:%${value}%g" "$temp"
	done
	cat "$temp"
	rm "$temp"
}


# Given a post’s JSON data, return the appropriate HTML corresponding to its
# media attachments, if any. Will return an empty string if none.
#     media_attachments $post_data
media_attachments() {
	local post_data="$1"
	local attachments="$(echo "$post_data" | jq -r '.media_attachments[]|(.type + "\t" + .url + "\t" + .description + "\t" + .preview_url)')"
	local IFS="
"
	for line in $attachments; do
		local ATTACH_TYPE="$(echo "$line" | awk -F'\t' '{print $1}')"
		local ATTACH_URL="$(echo "$line" | awk -F'\t' '{print $2}')"
		local ATTACH_DESC="$(echo "$line" | awk -F'\t' '{print $3}')"
		local ATTACH_PREVIEW="$(echo "$line" | awk -F'\t' '{print $4}')"
		local ATTACH_NAME="$(basename "$ATTACH_URL")"
		if test "$ATTACH_TYPE" = "image"; then
			env_subst "$ATTACH_IMAGE_TEMPLATE"
		else
			env_subst "$ATTACH_TEMPLATE"
		fi
	done
}


# Pass a post’s context JSON along stdin; out comes the response_data in JSON.
#     fetch_post_context $url | context_to_responses
context_to_responses() {
	jq '.descendants' \
		| jq 'sort_by(.created_at)' \
		| jq 'reverse' \
		| jq -cr '.[]'
}


# Make a request to the /api/v1/statuses/:id/$request API endpoint.
#     statuses_api_request $post_url $request
statuses_api_request() {
	local post_url="$1"
	local api_request="$2"
	if test -n "$api_request"; then
		api_request="/$api_request"
	fi

	local id="$(url_id "$url")"
	local server="$(url_server "$url")"
	curl --location --header 'Accept: application/json,application/activity+json' \
		 "$server/api/v1/statuses/${id}${api_request}"
}


# Require the context-JSON of a post, by URL.
#     fetch_post_context $url
fetch_post_context() {
	local url="$1"
	statuses_api_request "$url" "context"
}


# Given a post URL, request its JSON.
#     fetch_post $url
fetch_post() {
	local url="$1"
	statuses_api_request "$url"
}


# Return the ID of a post, based on its URL.
#     url_id $url
url_id() {
	local url="$1"
	# Pleroma-style URLs:  https://jam.xwx.moe/notice/Ac6PIZAP0ZzkMTYBBg
	# Mastodon-style URLs: https://esperanto.masto.host/@minjo/111461250815264185
	if echo "$url" | grep "/notice/" > /dev/null; then
		echo "$url" \
			| sed 's%.*/notice/%%'
	else
		echo "$url" \
			| sed 's%.*/@[[:alnum:]]*/%%'
	fi
}


# Return the server (including protocol) of a post, based on its URL.
#     url_server $url
url_server() {
	local url="$1"
	local protocol="$(echo "$url" | grep --only-matching '[[:alnum:]]*://')"
	printf "$protocol"
	echo "$url" \
		| sed 's%^'"$protocol"'%%' \
		| sed 's%/.*%%'
}


# Sanitize a template-string.
# AKA, escape quotation-marks.
#     prep_template $template
prep_template() {
	local template="$1"
	echo "$template" \
		| sed 's%\"%\\\"%g'
}


# Rough replacement for gettext’s envsubst. Safe!
# This will evaluate a string’s shell variables (somewhat) safely
# Probably not good for general use — but for our purposes, it only subsitutes
# along the “first level.” So environment variables are replaced, but those
# variables’ contents (AKA, post contents) aren’t evaluated.
#     env_subst $template
#     env_subst "$SHELL"  →  "/bin/sh"
env_subst() {
	local template="$1"
	eval "echo \"$(prep_template "$template")\""
}


usage() {
	echo "usage: $(basename "$0") POST_URL"
	echo
	echo "$(basename "$0") does exactly what it says on the tin: It formats"
	echo "a fediverse post (and its replies) into simple-and-embeddable HTML."
	echo
	echo "It works with posts from any server that supports Mastodon’s API,"
	echo "including Pleroma, Akkoma, Glitch, etc."
	echo
	echo "Notably, it supports post-atachments and custom-emoji. Keep in mind"
	echo "that images are all fetched from remote sources. It is recommended,"
	echo "if privacy or total archival, is a concern, to use wget(1)’s --mirror"
	echo "(or something like it) to fetch even these foreign files."
}


URL="$1"

if test -z "$URL"; then
	usage 1>&2
	exit 2
elif test "$URL" = "-h" -o "$URL" = "--help"; then
	usage
	exit 0
fi


POST="$(fetch_post "$URL")"
RESPONSES="$(fetch_post_context "$URL" | context_to_responses)"
render_post "$POST" "$RESPONSES" 0