From 58b7c785473a48d681f9722ebf74708bf29f4bc4 Mon Sep 17 00:00:00 2001
From: Michael Walker <mike@barrucadu.co.uk>
Date: Fri, 6 Sep 2024 14:06:31 +0100
Subject: [PATCH] [nyarlathotep] Use summary for kjp-to-hacksrus, not title

tumblr truncates the title: https://hacksrus.xyz/notice/Alj49cnMAXJLmAUYKm

I don't think I need to impose my own length limit here, as KJP posts
are pretty short.
---
 hosts/nyarlathotep/configuration.nix       | 3 ++-
 hosts/nyarlathotep/jobs/rss-to-mastodon.py | 9 +++++++--
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/hosts/nyarlathotep/configuration.nix b/hosts/nyarlathotep/configuration.nix
index dc79d324..a2c5d037 100644
--- a/hosts/nyarlathotep/configuration.nix
+++ b/hosts/nyarlathotep/configuration.nix
@@ -605,10 +605,11 @@ in
     startAt = "hourly";
     serviceConfig = {
       ExecStart =
-        let python = pkgs.python3.withPackages (ps: [ ps.docopt ps.feedparser ps.requests ]);
+        let python = pkgs.python3.withPackages (ps: [ ps.beautifulsoup4 ps.docopt ps.feedparser ps.requests ]);
         in concatStringsSep " " [
           "${python}/bin/python3"
           (pkgs.writeText "rss-to-mastodon.py" (fileContents ./jobs/rss-to-mastodon.py))
+          "--use-summary"
           "-d https://hacksrus.xyz/"
           "-f https://kingjamesprogramming.tumblr.com/rss"
           "-l /persist/var/lib/rss-to-mastodon/kjp-hacksrus.txt"
diff --git a/hosts/nyarlathotep/jobs/rss-to-mastodon.py b/hosts/nyarlathotep/jobs/rss-to-mastodon.py
index 1b851714..9d4abcd1 100644
--- a/hosts/nyarlathotep/jobs/rss-to-mastodon.py
+++ b/hosts/nyarlathotep/jobs/rss-to-mastodon.py
@@ -5,10 +5,11 @@
 Requires the API_KEY environment variable to be set.
 
 Usage:
-  rss-to-mastodon [--dry-run] -d <domain> -f <feed-url> -l <history-file> [-e <entries>] [-v <visibility>]
+  rss-to-mastodon [--dry-run] [--use-summary] -d <domain> -f <feed-url> -l <history-file> [-e <entries>] [-v <visibility>]
 
 Options:
   --dry-run          just print what would be published
+  --use-summary      use the (de-HTMLised) sumamry field, rather than the title
   -d <domain>        api domain
   -f <feed-url>      rss feed URL
   -l <history-file>  file to log feed item IDs to (to prevent double-posting)
@@ -16,6 +17,7 @@
   -v <visibility>    visibility of entries [default: public]
 """
 
+import bs4
 import docopt
 import feedparser
 import html.parser
@@ -28,6 +30,7 @@
 
 args = docopt.docopt(__doc__)
 dry_run = args["--dry-run"]
+use_summary = args["--use-summary"]
 api_domain = args["-d"]
 feed_url = args["-f"]
 history_file = pathlib.Path(args["-l"])
@@ -64,9 +67,11 @@
 
 # if there are multiple items, post the older ones first
 for item in reversed(items):
-    # handle entities
     title = html.parser.unescape(item["title"])
 
+    if use_summary:
+        title = bs4.BeautifulSoup(item["summary"], "html.parser").get_text().strip()
+
     print(item["id"])
     print(title)
     print()