#!/bin/sh

# SPDX-License-Identifier: GPL-3.0-or-later
# SPDX-FileCopyrightText: 2024 Jonas Smedegaard <dr@jones.dk>

# iterate through host list load web page for each, pause, and gather data
#
# Requires sway, firefox-esr, jq
# Recommends (for postprocessing): heif-examples, poppler-utils

# break on error
set -eu

# top 1000 domains
#  * go to <https://dataforseo.com/free-seo-stats/top-1000-websites>
#  * select "Denmark"
#  * download
json_file="ranked_domains.json"

# count Firefox windows
firefox_windows() {
	swaymsg -t get_tree | jq '[recurse(.nodes[]? | .nodes[]?)] | map(select(.app_id == "firefox-esr")) | length'
}

# count already opened Firefox windows
baseline_windows=$(firefox_windows)

# iterate through top domains and collect data about each
jq -c '.[]' "$json_file" | while read -r item; do
	domain=$(echo "$item" | jq -r '.domain')
	pos=$(echo "$item" | jq -r '.position')

	# skip if screenshot already exist for this domain
	[ ! -e "$pos.png" ] || continue

	# load front page of www host at domain into Firefox
	firefox --new-window "https://www.$domain"

	# wait until firefox window is closed
	while true; do
		sleep 1
		if [ "$(firefox_windows)" -eq "$baseline_windows" ]; then
			break
		fi
	done

	# collect PNG screenshot
	#  * use tool grimshot
	find ~/ -mindepth 1 -maxdepth 1 -name '*.png' -exec mv '{}' "$pos.png" ';'

	# collect HAR network timing data
	#  * Open debugger window: F12
	#  * Select pane "Network"
	#  * From rightmost pane, select "Save all as HAR"
	find ~/data -mindepth 1 -maxdepth 1 -name '*.har' -exec mv '{}' "$pos.har" ';'

	# collect data: PNG screenshot, HAR network timing and PDF screenshot
	#  * use plugin <https://addons.mozilla.org/da/firefox/addon/save-pdf/>
	find ~/data -mindepth 1 -maxdepth 1 -name '*' -exec mv '{}' "$pos.pdf" ';'
done

# Post-processing to extract text strings and compress images:
#find -name '*.pdf' -type f -exec pdftotext -raw '{}' \;
#find -name '*.png' -type f -exec heif-enc -A '{}' \;