From 11a330a7fc93506a44e0b14dd1ad10f7fa4a2104 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?John=20Ankarstr=C3=B6m?= <john@ankarstrom.se>
Date: Sat, 28 May 2022 00:55:08 +0200
Subject: Add "Fetch Screenwriters" menu item.

This is a separate item from "Fetch", because it takes a lot longer.
---
 pl/episode_data.pl | 74 +++++++++++++++++++++++++++++++++++++++---------------
 1 file changed, 54 insertions(+), 20 deletions(-)

(limited to 'pl/episode_data.pl')

diff --git a/pl/episode_data.pl b/pl/episode_data.pl
index af91d4e..3e0ee1b 100644
--- a/pl/episode_data.pl
+++ b/pl/episode_data.pl
@@ -1,5 +1,6 @@
 :- module(episode_data, [ensure_episode_data/0,
-			 fetch_episode_data/0,
+			 update_episode_data/0,
+			 update_screenwriters/0,
 			 retract_episode/1,
 			 episode_count/1,
 			 rate_episode/2,
@@ -26,7 +27,7 @@ detach :-
 	db_detach.
 
 ensure_episode_data :- episode_title(_, _), !.
-ensure_episode_data :- fetch_episode_data.
+ensure_episode_data :- update_episode_data.
 
 retract_episode(Ep) :-
 	(   episode_title(Ep, _)
@@ -64,12 +65,9 @@ tv_original(Ep) :- episode_datum(Ep, 'Source', 'TV Original').
 
 % Remote data retrieval.
 
-episode_number(Ep) --> integer(Ep).
-episode_number(Ep) --> integer(Ep), "WPS", integer(_).
-
-fetch_episode_data :-
+update_episode_data :-
 	findall(Ep-Title-Data,
-		(remote_episode_title_data(Ep, Title, Data)),
+		(fetch_episode_title_data(Ep, Title, Data)),
 		Set),
 	maplist(set_episode_data, Set).
 
@@ -94,20 +92,11 @@ maybe_assert_episode_datum(Ep, Key, Value) :-
 	    assert_episode_datum(Ep, Key, Value)
 	).
 
-fetch_html(H) :-
-	catch(http_load_html('https://www.detectiveconanworld.com/wiki/Anime', H),
-	      _,
-	      fail),
-	!,
-	nb_setval(html, H).
-
-remote_html(H) :-
-	(   nb_current(html, H), !
-	;   fetch_html(H)
-	).
+episode_number(Ep) --> integer(Ep).
+episode_number(Ep) --> integer(Ep), "WPS", integer(_).
 
-remote_episode_title_data(Ep, Title, ['Date'-Date, 'Source'-Source, 'Hint'-Hint]) :-
-	remote_html(H),
+fetch_episode_title_data(Ep, Title, ['Date'-Date, 'Source'-Source, 'Hint'-Hint]) :-
+	cached_html('https://www.detectiveconanworld.com/wiki/Anime', H),
 	xpath(H, //tr(td(index(3),@style='background:#f2fde9;')), R),
 	xpath(R, td(index(1),normalize_space), Ep0),
 	atom_phrase(episode_number(Ep), Ep0),
@@ -118,6 +107,51 @@ remote_episode_title_data(Ep, Title, ['Date'-Date, 'Source'-Source, 'Hint'-Hint]
 	atom_string(Source, Source1),
 	xpath(R, td(index(8),normalize_space), Hint).
 
+update_screenwriters :-
+	findall(Ep-Name,
+		(maybe_fetch_screenwriter_episode(Name, Ep)),
+		Set),
+	maplist(set_episode_screenwriter, Set).
+
+set_episode_screenwriter(Ep-Name) :-
+	maybe_assert_episode_datum(Ep, 'Screenwriter', Name).
+
+maybe_fetch_screenwriter_episode(Name, Ep) :-
+	\+ episode_datum(Ep, 'Screenwriter', Ep),
+	fetch_screenwriter_episode(Name, Ep).
+
+fetch_screenwriter(Name) :-
+	cached_html('https://www.detectiveconanworld.com/wiki/Category:Screenplay_writers', H),
+	xpath(H, //'div'(@id='mw-pages')//a, A),
+	xpath(A, /self(normalize_space), Name).
+
+fetch_screenwriter_url(Name, U) :-
+	cached_html('https://www.detectiveconanworld.com/wiki/Category:Screenplay_writers', H),
+	xpath(H, //'div'(@id='mw-pages')//a, A),
+	xpath(A, /self(normalize_space), Name),
+	xpath(A, /self(@href), U).
+
+absolute_url(R) --> "https://www.detectiveconanworld.com", R.
+screenwriter_episode(Ep) --> string(_), "(Episode ", integer(Ep), ")", string(_).
+screenwriter_episode(Ep) --> string(_), "(Episodes ", integer(Ep1), "-", integer(Ep2), ")", string(_),
+			     { between(Ep1, Ep2, Ep) }.
+
+fetch_screenwriter_episode(Name, Ep) :-
+	fetch_screenwriter_url(Name, U0),
+	atom_phrase(absolute_url(U0), U),
+	cached_html(U, H),
+	xpath(H, //div(@id='mw-content-text')//li, L),
+	xpath(L, /self(normalize_space), T),
+	atom_phrase(screenwriter_episode(Ep), T).
+
+fetch_html(U, H) :-
+	catch(http_load_html(U, H), _, fail),
+	!,
+	nb_setval(U, H).
+
+cached_html(U, H) :- nb_current(U, H), !.
+cached_html(U, H) :- fetch_html(U, H).
+
 http_load_html(URL, DOM) :-
         setup_call_cleanup(http_open(URL, In,
                            [ timeout(60)
-- 
cgit v1.2.3