.gitignore

@@ -1,7 +0,0 @@
-*.conf
-*.txt
-log/
-*.bak
-bin/
-lib/
-*.cfg

README.md

@@ -1,7 +1,5 @@
 # DiscoRSS
 
-![DiscoRSS Logo](https://frzn.dev/~amr/images/discorss.png)
-
 ## What is it?
 
 DiscoRSS is a simple Python script to send RSS feeds to Discord webhooks. It was created because existing bots that did this set limits on the number of feeds, and self-hosting stuff is easier and better anyway. To get this working, you will require the following Python libraries:

discorss.py

@@ -12,48 +12,29 @@
 
 import requests
 import feedparser
-import hashlib
 from pathlib import Path
 import json
+import datetime
 import time
 import os
-import re
 
-config_dir = os.environ.get('XDG_CONFIG_HOME')
+config_file_path = r"/etc/discorss.conf"
-if config_dir is None:
+# config_file_path = r"discorss.conf"
-    config_file_path = r"~/.config/discorss/discorss.conf"
+# log_file_path = r"/var/log/discorss"
-    config_dir = r"~/.config/discorss"
+log_file_path = r"./log"
-else:
-    config_file_path = config_dir + r"/discorss/discorss.conf"
-log_file_path = r"/var/log/discorss"
-# log_file_path = r"./log"
 log_file_name = r"/app.log"
-# Yes, I know you "can't parse HTML with regex", but
-# just watch me.
-html_filter = re.compile(r"\<\/?([A-Za-z \"\=])*\>")
-success_codes = ["200", "201", "202", "203", "204", "205", "206"]
 
 
-# This function gets and formats the brief excerpt that goes in the embed
-# Different feeds put summaries in different fields, so we pick the best
-# one and limit it to 150 characters.
-# TODO: make the character limit smarter, as to split at a natural point
 def get_description(feed):
     try:
         temporary_string = str(feed.entries[0]["summary_detail"]["value"])
-        temporary_string = html_filter.sub("", temporary_string)
         desc = (
-            temporary_string[:150]
+            temporary_string[:150] if len(temporary_string) > 150 else temporary_string
-            if len(temporary_string) > 150
-            else temporary_string
         )
     except KeyError:
         temporary_string = str(feed.entries[0]["description"])
-        temporary_string = html_filter.sub("", temporary_string)
         desc = (
-            temporary_string[:150]
+            temporary_string[:150] if len(temporary_string) > 150 else temporary_string
-            if len(temporary_string) > 150
-            else temporary_string
         )
     return desc
 
@@ -61,59 +42,39 @@ def get_description(feed):
 def main():
     os.environ["TZ"] = "America/Toronto"
     time.tzset()
-    # Check for log and config files/paths, create empty directories if needed
     try:
         Path(log_file_path).mkdir(parents=True, exist_ok=True)
     except FileExistsError:
-        print("The logfile path {} already exists and is not a directory!".format(log_file_path))
+        print("This path already exists and is not a directory!")
+    # Load and read the config file
     if not Path(config_file_path).exists():
-        print("No config file at {}! Snarf.\n{} was created for you.".format(config_file_path, config_dir))
+        print("No config file! Snarf.  Directories were created for you.")
-        Path(config_file_path).mkdir(parents=True, exist_ok=True)
         return
     with open(config_file_path, "r") as config_file:
         app_config = json.load(config_file)
     now = time.mktime(time.localtime())
-    try:
     last_check = app_config["lastupdate"]
-    except KeyError:
+    for hook in app_config["feeds"]:
-        last_check = (
-            now - 21600
-        )  # first run, no lastupdate, check up to 6 hours ago
-    for i, hook in enumerate(app_config["feeds"]):
         # Get the feed
         feed = feedparser.parse(hook["url"])
         published_time = time.mktime(feed.entries[0]["published_parsed"])
         published_time = published_time + hook["offset"]
-        print("Parsing feed {}...".format(hook["name"]))
+        print(feed.entries[0]["published"], published_time, now)
-        # Hash the title of the latest post and use that to determine if it's been posted
-        new_hash = hashlib.sha3_512(
-            bytes(feed.entries[0]["title"], "utf-8")
-        ).hexdigest()
-        try:
-            if hook["lasthash"] != new_hash:
-                app_config["feeds"][i]["lasthash"] = new_hash
-            else:
-                continue
-        except KeyError:
-            app_config["feeds"][i]["lasthash"] = new_hash
         # Generate the webhook
         webhook = {
             "embeds": [
                 {
                     "title": str(feed.entries[0]["title"]),
                     "url": str(feed.entries[0]["link"]),
-                    "color": 216128,
+                    "color": 5814783,
-                    "footer": {
+                    "provider": {
                         "name": "DiscoRSS",
-                        # "url": "https://git.frzn.dev/amr/discorss",
+                        "url": "https://git.frzn.dev/amr/discorss",
-                    },
-                    "author": {
-                        "name": str(hook["name"]),
-                        "url": str(hook["siteurl"]),
                     },
+                    "author": {"name": str(hook["name"]), "url": str(hook["siteurl"])},
                     "fields": [
                         {
-                            "name": "Excerpt from post:",
+                            "name": str(feed.entries[0]["title"]),
                             "value": get_description(feed),
                         }
                     ],
@@ -126,17 +87,11 @@ def main():
             "content-type": "application/json",
         }
         webhook_string = json.dumps(webhook)
-
+        # print(webhook_string)
-        if published_time > last_check:
+        if published_time > last_check and published_time < now:
             r = requests.post(
                 hook["webhook"], data=webhook_string, headers=custom_header
             )
-            if r.status_code not in success_codes:
-                print(
-                    "Error {} while trying to post {}".format(
-                        r.status_code, hook["webhook"]
-                    )
-                )
         app_config["lastupdate"] = now
         with open(config_file_path, "w") as config_file:
             json.dump(app_config, config_file, indent=4)