From 54796470b1ba51cb9e29195815d6d53b8d2cc7a5 Mon Sep 17 00:00:00 2001
From: jesopo <github@lolnerd.net>
Date: Sun, 23 Jun 2019 16:03:15 +0100
Subject: [PATCH] First version of rss.py

---
 README.md        |   2 +-
 modules/rss.py   | 118 +++++++++++++++++++++++++++++++++++++++++++++++
 requirements.txt |   1 +
 3 files changed, 120 insertions(+), 1 deletion(-)
 create mode 100644 modules/rss.py

diff --git a/README.md b/README.md
index d8200419..a30d7a91 100644
--- a/README.md
+++ b/README.md
@@ -2,7 +2,7 @@
 Python3 event-driven modular IRC bot!
 
 ## Requirements
-Python3.6+ and [BeautifulSoup4](https://pypi.python.org/pypi/beautifulsoup4), [lxml](https://pypi.org/project/lxml/), [netifaces](https://pypi.org/project/netifaces/), [requests](https://pypi.org/project/requests/), [scrypt](https://pypi.python.org/pypi/scrypt), [suds](https://pypi.python.org/pypi/suds-jurko) and [tweepy](https://pypi.org/project/tweepy/). Use `pip3 install -r requirements.txt` to install them all at once.
+Python3.6+ and [BeautifulSoup4](https://pypi.python.org/pypi/beautifulsoup4), [feedparser](https://pypi.org/project/feedparser/), [lxml](https://pypi.org/project/lxml/), [netifaces](https://pypi.org/project/netifaces/), [requests](https://pypi.org/project/requests/), [scrypt](https://pypi.python.org/pypi/scrypt), [suds](https://pypi.python.org/pypi/suds-jurko) and [tweepy](https://pypi.org/project/tweepy/). Use `pip3 install -r requirements.txt` to install them all at once.
 
 ## Setup
 See [docs/help/setup.md](docs/help/setup.md).
diff --git a/modules/rss.py b/modules/rss.py
new file mode 100644
index 00000000..5cef8a72
--- /dev/null
+++ b/modules/rss.py
@@ -0,0 +1,118 @@
+from src import ModuleManager, utils
+import feedparser
+
+RSS_INTERVAL = 60 # 1 minute
+
+def _format_entry(feed_title, entry):
+    title = entry["title"]
+
+    author = entry.get("author", None)
+    author = " by %s" % author if author else ""
+
+    link = entry.get("link", None)
+    link = " - %s" % link if link else ""
+
+    return "%s: %s%s%s" % (feed_title, title, author, link)
+
+class Module(ModuleManager.BaseModule):
+    _title = "RSS"
+    def on_load(self):
+        self.timers.add("rss", RSS_INTERVAL)
+
+    @utils.hook("timer.rss")
+    def timer(self, event):
+        event["timer"].redo()
+        hook_settings = self.bot.database.channel_settings.find_by_setting(
+            "rss-hooks")
+        hooks = {}
+        for server_id, channel_name, urls in hook_settings:
+            server = self.bot.get_server_by_id(server_id)
+            if server and channel_name in server.channels:
+                channel = server.channels.get(channel_name)
+                for url in urls:
+                    if not url in hooks:
+                        hooks[url] = []
+                    hooks[url].append(channel)
+
+        for url, channels in hooks.items():
+            try:
+                feed = feedparser.parse(url)
+            except:
+                feed = None
+
+            if not feed or not feed["feed"]:
+                self.log.warn("Failed to parse rss for %s", [url],
+                    exc_info=True)
+                continue
+
+            feed_title = feed["feed"]["title"]
+            entry_formatted = {}
+
+            for channel in channels:
+                seen_ids = channel.get_setting("rss-seen-ids", [])
+                new_ids = []
+                valid = 1
+                for entry in feed["entries"]:
+                    if entry["id"] in seen_ids:
+                        new_ids.append(entry["id"])
+                        continue
+
+                    if valid == 4:
+                        continue
+                    valid += 1
+
+                    if not entry["id"] in entry_formatted:
+                        output = _format_entry(feed_title, entry)
+                        entry_formatted[entry["id"]] = output
+                    else:
+                        output = entry_formatted[entry["id"]]
+
+                    self.events.on("send.stdout").call(target=channel,
+                        module_name="RSS", server=server, message=output)
+                    new_ids.append(entry["id"])
+
+                channel.set_setting("rss-seen-ids", new_ids)
+
+    @utils.hook("received.command.rss", min_args=1, channel_only=True)
+    def rss(self, event):
+        """
+        :help: Modify RSS/Atom configuration for the current channel
+        :usage: list
+        :usage: add <url>
+        :usage: remove <url>
+        :permission: rss
+        """
+        changed = False
+        message = None
+
+        rss_hooks = event["target"].get_setting("rss-hooks", [])
+
+        subcommand = event["args_split"][0].lower()
+        if subcommand == "list":
+            event["stdout"].write("RSS hooks: %s" % ", ".join(rss_hooks))
+        elif subcommand == "add":
+            if not len(event["args_split"]) > 1:
+                raise utils.EventError("Please provide a URL")
+
+            url = event["args_split"][1]
+            if url in rss_hooks:
+                raise utils.EventError("That URL is already being watched")
+            rss_hooks.append(url)
+            changed = True
+            message = "Added RSS feed"
+        elif subcommand == "remove":
+            if not len(event["args_split"]) > 1:
+                raise utils.EventError("Please provide a URL")
+
+            url = event["args_split"][1]
+            if not url in rss_hooks:
+                raise utils.EventError("I'm not watching that URL")
+            rss_hooks.remove(url)
+            changed = True
+            message = "Removed RSS feed"
+        else:
+            raise utils.EventError("Unknown subcommand '%s'" % subcommand)
+
+        if changed:
+            event["target"].set_setting("rss-hooks", rss_hooks)
+            event["stdout"].write(message)
diff --git a/requirements.txt b/requirements.txt
index 904f66ed..35153a7e 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,5 @@
 beautifulsoup4
+feedparser
 lxml
 netifaces
 requests