0340541002 2025-04-24 arcade: use crate::{
0340541002 2025-04-24 arcade: command,
0340541002 2025-04-24 arcade: sql::Db,
0340541002 2025-04-24 arcade: };
1c444d34ff 2024-08-28 arcade:
9171c791eb 2021-11-13 arcade: use std::{
c1e27b74ed 2021-11-13 arcade: borrow::Cow,
9171c791eb 2021-11-13 arcade: collections::{
9171c791eb 2021-11-13 arcade: BTreeMap,
9171c791eb 2021-11-13 arcade: HashSet,
9171c791eb 2021-11-13 arcade: },
dc7c43b010 2026-01-01 arcade: sync::Arc,
44575a91d3 2025-07-09 arcade: };
44575a91d3 2025-07-09 arcade:
dc7c43b010 2026-01-01 arcade: use async_compat::Compat;
acb0a4ac54 2025-09-28 arcade: use chrono::{
acb0a4ac54 2025-09-28 arcade: DateTime,
acb0a4ac54 2025-09-28 arcade: Local,
44575a91d3 2025-07-09 arcade: };
5a4aab7687 2025-06-29 arcade: use lazy_static::lazy_static;
5a4aab7687 2025-06-29 arcade: use regex::Regex;
acb0a4ac54 2025-09-28 arcade: use reqwest::header::{
acb0a4ac54 2025-09-28 arcade: CACHE_CONTROL,
acb0a4ac54 2025-09-28 arcade: EXPIRES,
acb0a4ac54 2025-09-28 arcade: LAST_MODIFIED
dc7c43b010 2026-01-01 arcade: };
dc7c43b010 2026-01-01 arcade: use smol::{
dc7c43b010 2026-01-01 arcade: Timer,
dc7c43b010 2026-01-01 arcade: lock::Mutex,
acb0a4ac54 2025-09-28 arcade: };
bb89b6fab8 2025-06-15 arcade: use tgbot::{
bb89b6fab8 2025-06-15 arcade: api::Client,
bb89b6fab8 2025-06-15 arcade: handler::UpdateHandler,
bb89b6fab8 2025-06-15 arcade: types::{
bb89b6fab8 2025-06-15 arcade: Bot,
bb89b6fab8 2025-06-15 arcade: ChatPeerId,
bb89b6fab8 2025-06-15 arcade: Command,
bb89b6fab8 2025-06-15 arcade: GetBot,
bb89b6fab8 2025-06-15 arcade: Message,
bb89b6fab8 2025-06-15 arcade: ParseMode,
bb89b6fab8 2025-06-15 arcade: SendMessage,
bb89b6fab8 2025-06-15 arcade: Update,
bb89b6fab8 2025-06-15 arcade: UpdateType,
bb89b6fab8 2025-06-15 arcade: UserPeerId,
1bd041d00f 2025-05-20 arcade: },
44575a91d3 2025-07-09 arcade: };
44575a91d3 2025-07-09 arcade: use stacked_errors::{
44575a91d3 2025-07-09 arcade: Result,
44575a91d3 2025-07-09 arcade: StackableErr,
44575a91d3 2025-07-09 arcade: anyhow,
44575a91d3 2025-07-09 arcade: bail,
5a4aab7687 2025-06-29 arcade: };
5a4aab7687 2025-06-29 arcade:
5a4aab7687 2025-06-29 arcade: lazy_static!{
5a4aab7687 2025-06-29 arcade: pub static ref RE_SPECIAL: Regex = Regex::new(r"([\-_*\[\]()~`>#+|{}\.!])").unwrap();
5a4aab7687 2025-06-29 arcade: }
5a4aab7687 2025-06-29 arcade:
5a4aab7687 2025-06-29 arcade: /// Encodes special HTML entities to prevent them interfering with Telegram HTML
5a4aab7687 2025-06-29 arcade: pub fn encode (text: &str) -> Cow<'_, str> {
5a4aab7687 2025-06-29 arcade: RE_SPECIAL.replace_all(text, "\\$1")
5a4aab7687 2025-06-29 arcade: }
9171c791eb 2021-11-13 arcade:
9171c791eb 2021-11-13 arcade: #[derive(Clone)]
9171c791eb 2021-11-13 arcade: pub struct Core {
bb89b6fab8 2025-06-15 arcade: owner_chat: ChatPeerId,
bb89b6fab8 2025-06-15 arcade: // max_delay: u16,
bb89b6fab8 2025-06-15 arcade: pub tg: Client,
bb89b6fab8 2025-06-15 arcade: pub me: Bot,
0340541002 2025-04-24 arcade: pub db: Db,
9171c791eb 2021-11-13 arcade: sources: Arc<Mutex<HashSet<Arc<i32>>>>,
45e34762e4 2023-05-28 arcade: http_client: reqwest::Client,
45e34762e4 2023-05-28 arcade: }
45e34762e4 2023-05-28 arcade:
dc7c43b010 2026-01-01 arcade: pub struct Post {
dc7c43b010 2026-01-01 arcade: uri: String,
dc7c43b010 2026-01-01 arcade: title: String,
dc7c43b010 2026-01-01 arcade: authors: String,
dc7c43b010 2026-01-01 arcade: summary: String,
dc7c43b010 2026-01-01 arcade: }
dc7c43b010 2026-01-01 arcade:
9171c791eb 2021-11-13 arcade: impl Core {
0340541002 2025-04-24 arcade: pub async fn new(settings: config::Config) -> Result<Core> {
44575a91d3 2025-07-09 arcade: let owner_chat = ChatPeerId::from(settings.get_int("owner").stack()?);
44575a91d3 2025-07-09 arcade: let api_key = settings.get_string("api_key").stack()?;
42b29b744b 2026-01-01 arcade: let tg = Client::new(&api_key).stack()?
42b29b744b 2026-01-01 arcade: .with_host(settings.get_string("api_gateway").stack()?);
45e34762e4 2023-05-28 arcade:
45e34762e4 2023-05-28 arcade: let mut client = reqwest::Client::builder();
9d8a6738fd 2023-07-30 arcade: if let Ok(proxy) = settings.get_string("proxy") {
44575a91d3 2025-07-09 arcade: let proxy = reqwest::Proxy::all(proxy).stack()?;
45e34762e4 2023-05-28 arcade: client = client.proxy(proxy);
45e34762e4 2023-05-28 arcade: }
44575a91d3 2025-07-09 arcade: let http_client = client.build().stack()?;
44575a91d3 2025-07-09 arcade: let me = tg.execute(GetBot).await.stack()?;
0340541002 2025-04-24 arcade: let core = Core {
f988dfd28f 2022-02-13 arcade: tg,
e624ef9d66 2025-04-20 arcade: me,
e624ef9d66 2025-04-20 arcade: owner_chat,
44575a91d3 2025-07-09 arcade: db: Db::new(&settings.get_string("pg").stack()?)?,
9171c791eb 2021-11-13 arcade: sources: Arc::new(Mutex::new(HashSet::new())),
45e34762e4 2023-05-28 arcade: http_client,
bb89b6fab8 2025-06-15 arcade: // max_delay: 60,
0340541002 2025-04-24 arcade: };
bb89b6fab8 2025-06-15 arcade: let clone = core.clone();
dc7c43b010 2026-01-01 arcade: smol::spawn(Compat::new(async move {
26339860ce 2022-02-13 arcade: loop {
26339860ce 2022-02-13 arcade: let delay = match &clone.autofetch().await {
26339860ce 2022-02-13 arcade: Err(err) => {
07b34bcad6 2025-08-05 arcade: if let Err(err) = clone.send(format!("🛑 {err}"), None, None).await {
e624ef9d66 2025-04-20 arcade: eprintln!("Autofetch error: {err:?}");
26339860ce 2022-02-13 arcade: };
cb86e770f9 2022-03-15 arcade: std::time::Duration::from_secs(60)
26339860ce 2022-02-13 arcade: },
26339860ce 2022-02-13 arcade: Ok(time) => *time,
9171c791eb 2021-11-13 arcade: };
dc7c43b010 2026-01-01 arcade: Timer::after(delay).await;
9171c791eb 2021-11-13 arcade: }
dc7c43b010 2026-01-01 arcade: })).detach();
9171c791eb 2021-11-13 arcade: Ok(core)
9171c791eb 2021-11-13 arcade: }
9171c791eb 2021-11-13 arcade:
bb89b6fab8 2025-06-15 arcade: pub async fn send <S>(&self, msg: S, target: Option<ChatPeerId>, mode: Option<ParseMode>) -> Result<Message>
e624ef9d66 2025-04-20 arcade: where S: Into<String> {
e624ef9d66 2025-04-20 arcade: let msg = msg.into();
e624ef9d66 2025-04-20 arcade:
e624ef9d66 2025-04-20 arcade: let mode = mode.unwrap_or(ParseMode::Html);
e624ef9d66 2025-04-20 arcade: let target = target.unwrap_or(self.owner_chat);
44575a91d3 2025-07-09 arcade: self.tg.execute(
bb89b6fab8 2025-06-15 arcade: SendMessage::new(target, msg)
bb89b6fab8 2025-06-15 arcade: .with_parse_mode(mode)
44575a91d3 2025-07-09 arcade: ).await.stack()
bb89b6fab8 2025-06-15 arcade: }
bb89b6fab8 2025-06-15 arcade:
acb0a4ac54 2025-09-28 arcade: pub async fn check (&self, id: i32, real: bool, last_scrape: Option<DateTime<Local>>) -> Result<String> {
bb89b6fab8 2025-06-15 arcade: let mut posted: i32 = 0;
44575a91d3 2025-07-09 arcade: let mut conn = self.db.begin().await.stack()?;
bb89b6fab8 2025-06-15 arcade:
bb89b6fab8 2025-06-15 arcade: let id = {
c6d3e97290 2025-07-01 arcade: let mut set = self.sources.lock_arc().await;
bb89b6fab8 2025-06-15 arcade: match set.get(&id) {
bb89b6fab8 2025-06-15 arcade: Some(id) => id.clone(),
bb89b6fab8 2025-06-15 arcade: None => {
bb89b6fab8 2025-06-15 arcade: let id = Arc::new(id);
9171c791eb 2021-11-13 arcade: set.insert(id.clone());
9171c791eb 2021-11-13 arcade: id.clone()
9171c791eb 2021-11-13 arcade: },
9171c791eb 2021-11-13 arcade: }
9171c791eb 2021-11-13 arcade: };
9171c791eb 2021-11-13 arcade: let count = Arc::strong_count(&id);
9171c791eb 2021-11-13 arcade: if count == 2 {
44575a91d3 2025-07-09 arcade: let source = conn.get_source(*id, self.owner_chat).await.stack()?;
44575a91d3 2025-07-09 arcade: conn.set_scrape(*id).await.stack()?;
bb89b6fab8 2025-06-15 arcade: let destination = ChatPeerId::from(match real {
e624ef9d66 2025-04-20 arcade: true => source.channel_id,
e624ef9d66 2025-04-20 arcade: false => source.owner,
bb89b6fab8 2025-06-15 arcade: });
9910c2209c 2023-08-04 arcade: let mut this_fetch: Option<DateTime<chrono::FixedOffset>> = None;
dc7c43b010 2026-01-01 arcade: let mut posts: BTreeMap<DateTime<chrono::FixedOffset>, Post> = BTreeMap::new();
9910c2209c 2023-08-04 arcade:
acb0a4ac54 2025-09-28 arcade: let mut builder = self.http_client.get(&source.url);
acb0a4ac54 2025-09-28 arcade: if let Some(last_scrape) = last_scrape {
acb0a4ac54 2025-09-28 arcade: builder = builder.header(LAST_MODIFIED, last_scrape.to_rfc2822());
acb0a4ac54 2025-09-28 arcade: };
acb0a4ac54 2025-09-28 arcade: let response = builder.send().await.stack()?;
acb0a4ac54 2025-09-28 arcade: {
acb0a4ac54 2025-09-28 arcade: let headers = response.headers();
acb0a4ac54 2025-09-28 arcade: let expires = headers.get(EXPIRES);
acb0a4ac54 2025-09-28 arcade: let cache = headers.get(CACHE_CONTROL);
acb0a4ac54 2025-09-28 arcade: if expires.is_some() || cache.is_some() {
acb0a4ac54 2025-09-28 arcade: println!("{} {} {:?} {:?} {:?}", Local::now().to_rfc2822(), &source.url, last_scrape, expires, cache);
acb0a4ac54 2025-09-28 arcade: }
acb0a4ac54 2025-09-28 arcade: }
9910c2209c 2023-08-04 arcade: let status = response.status();
44575a91d3 2025-07-09 arcade: let content = response.bytes().await.stack()?;
9171c791eb 2021-11-13 arcade: match rss::Channel::read_from(&content[..]) {
9171c791eb 2021-11-13 arcade: Ok(feed) => {
9171c791eb 2021-11-13 arcade: for item in feed.items() {
f988dfd28f 2022-02-13 arcade: if let Some(link) = item.link() {
f988dfd28f 2022-02-13 arcade: let date = match item.pub_date() {
f988dfd28f 2022-02-13 arcade: Some(feed_date) => DateTime::parse_from_rfc2822(feed_date),
f988dfd28f 2022-02-13 arcade: None => DateTime::parse_from_rfc3339(&item.dublin_core_ext().unwrap().dates()[0]),
44575a91d3 2025-07-09 arcade: }.stack()?;
dc7c43b010 2026-01-01 arcade: let uri = link.to_string();
dc7c43b010 2026-01-01 arcade: let title = item.title().unwrap_or("").to_string();
dc7c43b010 2026-01-01 arcade: let authors = item.author().unwrap_or("").to_string();
dc7c43b010 2026-01-01 arcade: let summary = item.content().unwrap_or("").to_string();
dc7c43b010 2026-01-01 arcade: posts.insert(date, Post{
dc7c43b010 2026-01-01 arcade: uri,
dc7c43b010 2026-01-01 arcade: title,
dc7c43b010 2026-01-01 arcade: authors,
dc7c43b010 2026-01-01 arcade: summary,
dc7c43b010 2026-01-01 arcade: });
9171c791eb 2021-11-13 arcade: }
9171c791eb 2021-11-13 arcade: };
9171c791eb 2021-11-13 arcade: },
9171c791eb 2021-11-13 arcade: Err(err) => match err {
9171c791eb 2021-11-13 arcade: rss::Error::InvalidStartTag => {
2c36f015d8 2024-11-23 arcade: match atom_syndication::Feed::read_from(&content[..]) {
2c36f015d8 2024-11-23 arcade: Ok(feed) => {
2c36f015d8 2024-11-23 arcade: for item in feed.entries() {
2c36f015d8 2024-11-23 arcade: let date = item.published().unwrap();
dc7c43b010 2026-01-01 arcade: let uri = item.links()[0].href().to_string();
dc7c43b010 2026-01-01 arcade: let title = item.title().to_string();
dc7c43b010 2026-01-01 arcade: let authors = item.authors().iter().map(|x| format!("{} <{:?}>", x.name(), x.email())).collect::<Vec<String>>().join(", ");
dc7c43b010 2026-01-01 arcade: let summary = if let Some(sum) = item.summary() { sum.value.clone() } else { String::new() };
dc7c43b010 2026-01-01 arcade: posts.insert(*date, Post{
dc7c43b010 2026-01-01 arcade: uri,
dc7c43b010 2026-01-01 arcade: title,
dc7c43b010 2026-01-01 arcade: authors,
dc7c43b010 2026-01-01 arcade: summary,
dc7c43b010 2026-01-01 arcade: });
2c36f015d8 2024-11-23 arcade: };
2c36f015d8 2024-11-23 arcade: },
2c36f015d8 2024-11-23 arcade: Err(err) => {
07b34bcad6 2025-08-05 arcade: bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
2c36f015d8 2024-11-23 arcade: },
2c36f015d8 2024-11-23 arcade: }
9910c2209c 2023-08-04 arcade: },
9910c2209c 2023-08-04 arcade: rss::Error::Eof => (),
07b34bcad6 2025-08-05 arcade: _ => bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
9910c2209c 2023-08-04 arcade: }
9910c2209c 2023-08-04 arcade: };
dc7c43b010 2026-01-01 arcade: for (date, post) in posts.iter() {
9910c2209c 2023-08-04 arcade: let post_url: Cow<str> = match source.url_re {
dc7c43b010 2026-01-01 arcade: Some(ref x) => sedregex::ReplaceCommand::new(x).stack()?.execute(&post.uri),
dc7c43b010 2026-01-01 arcade: None => post.uri.clone().into(),
9910c2209c 2023-08-04 arcade: };
44575a91d3 2025-07-09 arcade: if let Some(exists) = conn.exists(&post_url, *id).await.stack()? {
9910c2209c 2023-08-04 arcade: if ! exists {
9910c2209c 2023-08-04 arcade: if this_fetch.is_none() || *date > this_fetch.unwrap() {
9910c2209c 2023-08-04 arcade: this_fetch = Some(*date);
9910c2209c 2023-08-04 arcade: };
e624ef9d66 2025-04-20 arcade: self.send( match &source.iv_hash {
e624ef9d66 2025-04-20 arcade: Some(hash) => format!("<a href=\"https://t.me/iv?url={post_url}&rhash={hash}\"> </a>{post_url}"),
e624ef9d66 2025-04-20 arcade: None => format!("{post_url}"),
44575a91d3 2025-07-09 arcade: }, Some(destination), Some(ParseMode::Html)).await.stack()?;
44575a91d3 2025-07-09 arcade: conn.add_post(*id, date, &post_url).await.stack()?;
9910c2209c 2023-08-04 arcade: };
9171c791eb 2021-11-13 arcade: };
9171c791eb 2021-11-13 arcade: posted += 1;
9171c791eb 2021-11-13 arcade: };
9171c791eb 2021-11-13 arcade: posts.clear();
9171c791eb 2021-11-13 arcade: };
0340541002 2025-04-24 arcade: Ok(format!("Posted: {posted}"))
26339860ce 2022-02-13 arcade: }
26339860ce 2022-02-13 arcade:
bb89b6fab8 2025-06-15 arcade: async fn autofetch(&self) -> Result<std::time::Duration> {
26339860ce 2022-02-13 arcade: let mut delay = chrono::Duration::minutes(1);
9910c2209c 2023-08-04 arcade: let now = chrono::Local::now();
bb89b6fab8 2025-06-15 arcade: let queue = {
44575a91d3 2025-07-09 arcade: let mut conn = self.db.begin().await.stack()?;
44575a91d3 2025-07-09 arcade: conn.get_queue().await.stack()?
bb89b6fab8 2025-06-15 arcade: };
bb89b6fab8 2025-06-15 arcade: for row in queue {
9910c2209c 2023-08-04 arcade: if let Some(next_fetch) = row.next_fetch {
9910c2209c 2023-08-04 arcade: if next_fetch < now {
acb0a4ac54 2025-09-28 arcade: if let (Some(owner), Some(source_id), last_scrape) = (row.owner, row.source_id, row.last_scrape) {
bb89b6fab8 2025-06-15 arcade: let clone = Core {
bb89b6fab8 2025-06-15 arcade: owner_chat: ChatPeerId::from(owner),
9910c2209c 2023-08-04 arcade: ..self.clone()
9910c2209c 2023-08-04 arcade: };
b4af85e31b 2025-06-29 arcade: let source = {
44575a91d3 2025-07-09 arcade: let mut conn = self.db.begin().await.stack()?;
b4af85e31b 2025-06-29 arcade: match conn.get_one(owner, source_id).await {
b4af85e31b 2025-06-29 arcade: Ok(Some(source)) => source.to_string(),
4c144972c0 2026-01-02 arcade: Ok(None) => "Source not found in database?".to_string(),
b4af85e31b 2025-06-29 arcade: Err(err) => format!("Failed to fetch source data:\n{err}"),
b4af85e31b 2025-06-29 arcade: }
b4af85e31b 2025-06-29 arcade: };
dc7c43b010 2026-01-01 arcade: smol::spawn(Compat::new(async move {
acb0a4ac54 2025-09-28 arcade: if let Err(err) = clone.check(source_id, true, Some(last_scrape)).await {
4c144972c0 2026-01-02 arcade: if let Err(err) = clone.send(&format!("🛑 {source}\n{}", encode(&err.to_string())), None, Some(ParseMode::MarkdownV2)).await {
ec71df21e1 2025-11-07 arcade: eprintln!("Check error: {err}");
79c91a5357 2024-08-02 arcade: // clone.disable(&source_id, owner).await.unwrap();
9910c2209c 2023-08-04 arcade: };
9910c2209c 2023-08-04 arcade: };
dc7c43b010 2026-01-01 arcade: })).detach();
9910c2209c 2023-08-04 arcade: }
9910c2209c 2023-08-04 arcade: } else if next_fetch - now < delay {
9910c2209c 2023-08-04 arcade: delay = next_fetch - now;
9910c2209c 2023-08-04 arcade: }
9910c2209c 2023-08-04 arcade: }
9910c2209c 2023-08-04 arcade: };
44575a91d3 2025-07-09 arcade: delay.to_std().stack()
9910c2209c 2023-08-04 arcade: }
9910c2209c 2023-08-04 arcade:
bb89b6fab8 2025-06-15 arcade: pub async fn list (&self, owner: UserPeerId) -> Result<String> {
b4af85e31b 2025-06-29 arcade: let mut reply: Vec<String> = vec![];
9910c2209c 2023-08-04 arcade: reply.push("Channels:".into());
44575a91d3 2025-07-09 arcade: let mut conn = self.db.begin().await.stack()?;
44575a91d3 2025-07-09 arcade: for row in conn.get_list(owner).await.stack()? {
b4af85e31b 2025-06-29 arcade: reply.push(row.to_string());
e624ef9d66 2025-04-20 arcade: };
b4af85e31b 2025-06-29 arcade: Ok(reply.join("\n\n"))
bb89b6fab8 2025-06-15 arcade: }
bb89b6fab8 2025-06-15 arcade: }
bb89b6fab8 2025-06-15 arcade:
bb89b6fab8 2025-06-15 arcade: impl UpdateHandler for Core {
bb89b6fab8 2025-06-15 arcade: async fn handle (&self, update: Update) {
bb89b6fab8 2025-06-15 arcade: if let UpdateType::Message(msg) = update.update_type {
bb89b6fab8 2025-06-15 arcade: if let Ok(cmd) = Command::try_from(msg) {
bb89b6fab8 2025-06-15 arcade: let msg = cmd.get_message();
bb89b6fab8 2025-06-15 arcade: let words = cmd.get_args();
fae13a0e55 2025-06-28 arcade: let command = cmd.get_name();
fae13a0e55 2025-06-28 arcade: let res = match command {
fae13a0e55 2025-06-28 arcade: "/check" | "/clean" | "/enable" | "/delete" | "/disable" => command::command(self, command, msg, words).await,
bb89b6fab8 2025-06-15 arcade: "/start" => command::start(self, msg).await,
bb89b6fab8 2025-06-15 arcade: "/list" => command::list(self, msg).await,
fae13a0e55 2025-06-28 arcade: "/add" | "/update" => command::update(self, command, msg, words).await,
bb89b6fab8 2025-06-15 arcade: any => Err(anyhow!("Unknown command: {any}")),
bb89b6fab8 2025-06-15 arcade: };
bb89b6fab8 2025-06-15 arcade: if let Err(err) = res {
07b34bcad6 2025-08-05 arcade: if let Err(err2) = self.send(format!("\\#error\n```\n{err}\n```"),
bb89b6fab8 2025-06-15 arcade: Some(msg.chat.get_id()),
bb89b6fab8 2025-06-15 arcade: Some(ParseMode::MarkdownV2)
bb89b6fab8 2025-06-15 arcade: ).await{
bb89b6fab8 2025-06-15 arcade: dbg!(err2);
bb89b6fab8 2025-06-15 arcade: };
bb89b6fab8 2025-06-15 arcade: }
bb89b6fab8 2025-06-15 arcade: };
bb89b6fab8 2025-06-15 arcade: };
e624ef9d66 2025-04-20 arcade: }
9171c791eb 2021-11-13 arcade: }