Change the word filter configuration to allow providing reasons, fix emotes not working in filters, and implement separate OOC/IC/PDA filters (#61606)

code/controllers/configuration/configuration.dm

@@ -24,8 +24,24 @@
 	/// If the configuration is loaded
 	var/loaded = FALSE
 
+	/// A regex that matches words blocked IC
 	var/static/regex/ic_filter_regex
 
+	/// A regex that matches words blocked OOC
+	var/static/regex/ooc_filter_regex
+
+	/// A regex that matches words blocked IC, but not in PDAs
+	var/static/regex/ic_outside_pda_filter_regex
+
+	/// An assoc list of blocked IC words to their reasons
+	var/static/list/ic_filter_reasons
+
+	/// An assoc list of words that are blocked IC, but not in PDAs, to their reasons
+	var/static/list/ic_outside_pda_filter_reasons
+
+	/// An assoc list of words that are blocked both IC and OOC to their reasons
+	var/static/list/shared_filter_reasons
+
 /datum/controller/configuration/proc/admin_reload()
 	if(IsAdminAdvancedProcCall())
 		return
@@ -338,17 +354,84 @@ Example config:
 				log_config("Unknown command in map vote config: '[command]'")
 
 /datum/controller/configuration/proc/LoadChatFilter()
-	var/list/in_character_filter = list()
-	if(!fexists("[directory]/in_character_filter.txt"))
+	if(!fexists("[directory]/word_filter.toml"))
+		load_legacy_chat_filter()
 		return
+
+	log_config("Loading config file word_filter.toml...")
+
+	var/list/word_filter = rustg_read_toml_file("[directory]/word_filter.toml")
+	if (!islist(word_filter))
+		var/message = "The word filter configuration did not output a list, contact someone with configuration access to make sure it's setup properly."
+		log_config(message)
+		DelayedMessageAdmins(message)
+		return
+
+	ic_filter_reasons = try_extract_from_word_filter(word_filter, "ic")
+	ic_outside_pda_filter_reasons = try_extract_from_word_filter(word_filter, "ic_outside_pda")
+	shared_filter_reasons = try_extract_from_word_filter(word_filter, "shared")
+
+	update_chat_filter_regexes()
+
+/datum/controller/configuration/proc/load_legacy_chat_filter()
+	if (!fexists("[directory]/in_character_filter.txt"))
+		return
+
 	log_config("Loading config file in_character_filter.txt...")
-	for(var/line in world.file2list("[directory]/in_character_filter.txt"))
-		if(!line)
+
+	ic_filter_reasons = list()
+	ic_outside_pda_filter_reasons = list()
+	shared_filter_reasons = list()
+
+	for (var/line in world.file2list("[directory]/in_character_filter.txt"))
+		if (!line)
 			continue
-		if(findtextEx(line,"#",1,2))
+		if (findtextEx(line, "#", 1, 2))
 			continue
-		in_character_filter += REGEX_QUOTE(line)
-	ic_filter_regex = in_character_filter.len ? regex("\\b([jointext(in_character_filter, "|")])\\b", "i") : null
+		// The older filter didn't apply to PDA
+		ic_outside_pda_filter_reasons[line] = "No reason available"
+
+	update_chat_filter_regexes()
+
+/// Will update the internal regexes of the chat filter based on the filter reasons
+/datum/controller/configuration/proc/update_chat_filter_regexes()
+	ic_filter_regex = compile_filter_regex(ic_filter_reasons + ic_outside_pda_filter_reasons + shared_filter_reasons)
+	ic_outside_pda_filter_regex = compile_filter_regex(ic_filter_reasons + shared_filter_reasons)
+	ooc_filter_regex = compile_filter_regex(shared_filter_reasons)
+
+/datum/controller/configuration/proc/try_extract_from_word_filter(list/word_filter, key)
+	var/list/banned_words = word_filter[key]
+
+	if (isnull(banned_words))
+		return list()
+	else if (!islist(banned_words))
+		var/message = "The word filter configuration's '[key]' key was invalid, contact someone with configuration access to make sure it's setup properly."
+		log_config(message)
+		DelayedMessageAdmins(message)
+		return list()
+
+	return banned_words
+
+/datum/controller/configuration/proc/compile_filter_regex(list/banned_words)
+	if (isnull(banned_words) || banned_words.len == 0)
+		return null
+
+	var/static/regex/should_join_on_word_bounds = regex(@"^\w+$")
+
+	// Stuff like emoticons needs another split, since there's no way to get ":)" on a word bound.
+	// Furthermore, normal words need to be on word bounds, so "(adminhelp)" gets filtered.
+	var/list/to_join_on_whitespace_splits = list()
+	var/list/to_join_on_word_bounds = list()
+
+	for (var/banned_word in banned_words)
+		if (findtext(banned_word, should_join_on_word_bounds))
+			to_join_on_word_bounds += REGEX_QUOTE(banned_word)
+		else
+			to_join_on_whitespace_splits += REGEX_QUOTE(banned_word)
+
+	var/whitespace_split = @"(?:(?:^|\s+)(" + jointext(to_join_on_whitespace_splits, "|") + @")(?:$|\s+))"
+	var/word_bounds = @"(\b(" + jointext(to_join_on_word_bounds, "|") + "))"
+	return regex("([whitespace_split]|[word_bounds])", "i")
 
 //Message admins when you can.
 /datum/controller/configuration/proc/DelayedMessageAdmins(text)