Merge pull request #14 from lfuelling/master

add config option to filter for language

config.json

@@ -1,4 +1,5 @@
 {
+    "lang": "en",
 	"site": "https://botsin.space",
 	"cw": null,
 	"learn_from_cw": false

main.py

@@ -106,6 +106,18 @@ patterns = {
 	"pid": re.compile(r"[^\/]+$"),
 }
 
+
+def insert_toot(oii, acc, post, cursor):  # extracted to prevent duplication
+	pid = patterns["pid"].search(oii['object']['id']).group(0)
+	cursor.execute("REPLACE INTO toots (id, cw, userid, uri, content) VALUES (?, ?, ?, ?, ?)", (
+		pid,
+		1 if (oii['object']['summary'] != None and oii['object']['summary'] != "") else 0,
+		acc.id,
+		oii['object']['id'],
+		post
+	))
+
+
 for f in following:
 	last_toot = c.execute("SELECT id FROM `toots` WHERE userid LIKE ? ORDER BY id DESC LIMIT 1", (f.id,)).fetchone()
 	if last_toot != None:
@@ -172,15 +184,11 @@ for f in following:
 							#we've caught up to the notices we've already downloaded, so we can stop now
 							#you might be wondering, "lynne, what if the instance ratelimits you after 40 posts, and they've made 60 since main.py was last run? wouldn't the bot miss 20 posts and never be able to see them?" to which i reply, "it's called mstdn-ebooks not fediverse-ebooks. pleroma support is an afterthought"
 							done = True
-					pid = patterns["pid"].search(oi['object']['id']).group(0)
+					if cfg['lang']:
-					c.execute("REPLACE INTO toots (id, cw, userid, uri, content) VALUES (?, ?, ?, ?, ?)", (
+						if oi['object']['contentMap'][cfg['lang']]:  # filter for language
-						pid,
+							insert_toot(oi, f, toot, c)
-						1 if (oi['object']['summary'] != None and oi['object']['summary'] != "") else 0,
+					else:
-						f.id,
+						insert_toot(oi, f, toot, c)
-						oi['object']['id'],
-						toot
-						)
-					)
 					pass
 				except:
 					pass #ignore any toots that don't successfully go into the DB