#!/bin/sh

rm -rf /tmp/apgte
mkdir /tmp/apgte

curl -s 'https://practicalguidetoevil.wordpress.com/table-of-contents/' | hq 'sel [class=entry-content]' | grep -v sharedaddy | grep -v jp-relatedposts | while read line; do
	printf '%s' "$line" | grep -q "</h2>$" && {
		printf 'book\t%s\n' "$line" | sed 's/<[^<>]*>//g'
	} || {
		printf '%s' "$line" | hq 'sel a | totext' | sed -e 's/ ( /\t/' -e 's/..$//'
	}
done | while read -r line; do
	echo $line
	test "$(printf '%s' "$line" | cut -f 1)" = "book" && {
		echo $book.pdf
		test -n "$book" && weasyprint /tmp/"$book.html" /tmp/apgte/"$book.pdf"
		echo Commentated $book.pdf
		test -n "$book" && weasyprint /tmp/"Commentated $book.html" /tmp/apgte/"Commentated $book.pdf"
		book="$(printf '%s' "$line" | cut -f 2)"
		printf "<h1>%s</h1><hr>" "$book" >/tmp/"$book.html"
		printf "<h1>%s</h1><hr>" "$book" >/tmp/"Commentated $book.html"
	} || {
		curl -s "$(printf '%s' "$line" | cut -f 2)" >/tmp/apgte.html
		{
			printf '<h2>'
			printf '%s' "$line" | cut -f 1
			printf '</h2>'
			hq 'sel [class=entry-content]' </tmp/apgte.html | grep -v sharedaddy | grep -v jp-relatedposts
		} >> /tmp/"$book.html"
		{
			printf '<h2>'
			printf '%s' "$line" | cut -f 1
			printf '</h2>'
			hq 'sel [class=entry-content]' </tmp/apgte.html | grep -v sharedaddy | grep -v jp-relatedposts
			printf '<br><br><hr><br><br>'
			hq 'sel .comment-content p:not(:last-child), .comment-author cite' </tmp/apgte.html >/tmp/comments.html
			hq 'sel li.comment | . class' </tmp/apgte.html | while read -r line; do
				line="$(printf %s "$line" | tr ' ' '\n' | grep '^depth-' | sed 's/depth-//')"
				printf '<div style="padding: 0 0 0 %sem;">' "$((line - 1))"
				awk 'BEGIN { a = 0 } /^<cite / { a++ } { if (a > 1) { print >>"/tmp/comments2.html" } else { print } }' </tmp/comments.html
				[ -f /tmp/comments2.html ] && mv /tmp/comments2.html /tmp/comments.html
				printf '</div>'
			done
			printf '<br><br><hr><br><br>'
		} >> /tmp/"Commentated $book.html"
	}
done

echo $book.pdf
test -n "$book" && weasyprint /tmp/"$book.html" /tmp/apgte/"$book.pdf"
echo Commentated $book.pdf
test -n "$book" && weasyprint /tmp/"Commentated $book.html" /tmp/apgte/"Commentated $book.pdf"


printf "<h1>Extra Chapters</h1><hr>" >/tmp/"Extra Chapters.html"
printf "<h1>Extra Chapters</h1><hr>" >/tmp/"Commentated Extra Chapters.html"

curl -s 'https://practicalguidetoevil.wordpress.com/extra-chapters/' | hq 'sel [class=entry-content]' | grep -v sharedaddy | grep -v jp-relatedposts | while read line; do
	printf '%s' "$line" | hq 'sel a | totext' | sed -e 's/ ( /\t/' -e 's/..$//'
done | while read -r line; do
	echo $line
	curl -s "$(printf '%s' "$line" | cut -f 2)" >/tmp/apgte.html
	{
		printf '<h2>'
		printf '%s' "$line" | cut -f 1
		printf '</h2>'
		hq 'sel [class=entry-content]' </tmp/apgte.html | grep -v sharedaddy | grep -v jp-relatedposts
	} >> /tmp/"Extra Chapters.html"
	{
		printf '<h2>'
		printf '%s' "$line" | cut -f 1
		printf '</h2>'
		hq 'sel [class=entry-content]' </tmp/apgte.html | grep -v sharedaddy | grep -v jp-relatedposts
		printf '<br><br><hr><br><br>'
		hq 'sel .comment-content p:not(:last-child), .comment-author cite' </tmp/apgte.html >/tmp/comments.html
		hq 'sel li.comment | . class' </tmp/apgte.html | while read -r line; do
			line="$(printf %s "$line" | tr ' ' '\n' | grep '^depth-' | sed 's/depth-//')"
			printf '<div style="padding: 0 0 0 %sem;">' "$((line - 1))"
			awk 'BEGIN { a = 0 } /^<cite / { a++ } { if (a > 1) { print >>"/tmp/comments2.html" } else { print } }' </tmp/comments.html
			[ -f /tmp/comments2.html ] && mv /tmp/comments2.html /tmp/comments.html
			printf '</div>'
		done
		printf '<br><br><hr><br><br>'
	} >> /tmp/"Commentated Extra Chapters.html"
done

echo Extra Chapters
weasyprint "/tmp/Extra Chapters.html" "/tmp/apgte/Extra Chapters.pdf"
echo Commentated Extra Chapters
weasyprint "/tmp/Commentated Extra Chapters.html" "/tmp/apgte/Commentated Extra Chapters.pdf"