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: },
1c444d34ff 2024-08-28 arcade: num::TryFromIntError,
1c444d34ff 2024-08-28 arcade: sync::{
1c444d34ff 2024-08-28 arcade: Arc,
1c444d34ff 2024-08-28 arcade: Mutex
1c444d34ff 2024-08-28 arcade: },
1c444d34ff 2024-08-28 arcade: };
1c444d34ff 2024-08-28 arcade:
e624ef9d66 2025-04-20 arcade: use anyhow::{
e624ef9d66 2025-04-20 arcade: bail,
e624ef9d66 2025-04-20 arcade: Result,
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: use async_std::task;
e624ef9d66 2025-04-20 arcade: use chrono::DateTime;
e624ef9d66 2025-04-20 arcade: use frankenstein::{
e624ef9d66 2025-04-20 arcade: client_reqwest::Bot,
e624ef9d66 2025-04-20 arcade: methods::{
e624ef9d66 2025-04-20 arcade: GetUpdatesParams,
e624ef9d66 2025-04-20 arcade: SendMessageParams
e624ef9d66 2025-04-20 arcade: },
e624ef9d66 2025-04-20 arcade: types::{
e624ef9d66 2025-04-20 arcade: AllowedUpdate,
e624ef9d66 2025-04-20 arcade: MessageEntityType,
e624ef9d66 2025-04-20 arcade: User,
e624ef9d66 2025-04-20 arcade: },
e624ef9d66 2025-04-20 arcade: updates::UpdateContent,
e624ef9d66 2025-04-20 arcade: AsyncTelegramApi,
e624ef9d66 2025-04-20 arcade: ParseMode,
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: use thiserror::Error;
e624ef9d66 2025-04-20 arcade:
1c444d34ff 2024-08-28 arcade: #[derive(Error, Debug)]
1c444d34ff 2024-08-28 arcade: pub enum RssError {
a2880e5100 2025-04-19 arcade: // #[error(transparent)]
a2880e5100 2025-04-19 arcade: // Tg(#[from] TgError),
1c444d34ff 2024-08-28 arcade: #[error(transparent)]
1c444d34ff 2024-08-28 arcade: Int(#[from] TryFromIntError),
79c91a5357 2024-08-02 arcade: }
9171c791eb 2021-11-13 arcade:
9171c791eb 2021-11-13 arcade: #[derive(Clone)]
9171c791eb 2021-11-13 arcade: pub struct Core {
e624ef9d66 2025-04-20 arcade: owner_chat: i64,
a2880e5100 2025-04-19 arcade: pub tg: Bot,
e624ef9d66 2025-04-20 arcade: pub me: User,
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,
9171c791eb 2021-11-13 arcade: }
9171c791eb 2021-11-13 arcade:
9171c791eb 2021-11-13 arcade: impl Core {
0340541002 2025-04-24 arcade: pub async fn new(settings: config::Config) -> Result<Core> {
e624ef9d66 2025-04-20 arcade: let owner_chat = settings.get_int("owner")?;
28da2e2a00 2022-03-12 arcade: let api_key = settings.get_string("api_key")?;
e624ef9d66 2025-04-20 arcade: let tg = Bot::new(&api_key);
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") {
45e34762e4 2023-05-28 arcade: let proxy = reqwest::Proxy::all(proxy)?;
45e34762e4 2023-05-28 arcade: client = client.proxy(proxy);
45e34762e4 2023-05-28 arcade: }
45e34762e4 2023-05-28 arcade: let http_client = client.build()?;
e624ef9d66 2025-04-20 arcade: let me = tg.get_me().await?;
e624ef9d66 2025-04-20 arcade: let me = me.result;
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,
0340541002 2025-04-24 arcade: db: Db::new(&settings.get_string("pg")?)?,
9171c791eb 2021-11-13 arcade: sources: Arc::new(Mutex::new(HashSet::new())),
45e34762e4 2023-05-28 arcade: http_client,
0340541002 2025-04-24 arcade: };
0340541002 2025-04-24 arcade: let mut clone = core.clone();
cb86e770f9 2022-03-15 arcade: task::spawn(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) => {
e624ef9d66 2025-04-20 arcade: if let Err(err) = clone.send(format!("š {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: };
cb86e770f9 2022-03-15 arcade: task::sleep(delay).await;
9171c791eb 2021-11-13 arcade: }
e624ef9d66 2025-04-20 arcade: });
9171c791eb 2021-11-13 arcade: Ok(core)
9171c791eb 2021-11-13 arcade: }
9171c791eb 2021-11-13 arcade:
0340541002 2025-04-24 arcade: pub async fn stream(&mut self) -> Result<()> {
e624ef9d66 2025-04-20 arcade: let mut offset: i64 = 0;
e624ef9d66 2025-04-20 arcade: let mut params = GetUpdatesParams {
e624ef9d66 2025-04-20 arcade: offset: None,
e624ef9d66 2025-04-20 arcade: limit: Some(100),
e624ef9d66 2025-04-20 arcade: timeout: Some(300),
e624ef9d66 2025-04-20 arcade: allowed_updates: Some(vec![AllowedUpdate::Message]),
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: loop {
e624ef9d66 2025-04-20 arcade: let updates = self.tg.get_updates(¶ms).await?.result;
e624ef9d66 2025-04-20 arcade: if updates.is_empty() {
e624ef9d66 2025-04-20 arcade: offset = 0;
e624ef9d66 2025-04-20 arcade: params.offset = None;
e624ef9d66 2025-04-20 arcade: continue;
e624ef9d66 2025-04-20 arcade: }
e624ef9d66 2025-04-20 arcade: for update in updates {
e624ef9d66 2025-04-20 arcade: if i64::from(update.update_id) >= offset {
e624ef9d66 2025-04-20 arcade: offset = i64::from(update.update_id) + 1;
e624ef9d66 2025-04-20 arcade: params.offset = Some(offset);
e624ef9d66 2025-04-20 arcade: }
e624ef9d66 2025-04-20 arcade: if let UpdateContent::Message(msg) = update.content {
e624ef9d66 2025-04-20 arcade: if let Some(text) = msg.text {
e624ef9d66 2025-04-20 arcade: if let Some(entities) = msg.entities {
e624ef9d66 2025-04-20 arcade: let chars: Vec<u16> = text.encode_utf16().collect();
e624ef9d66 2025-04-20 arcade: for entity in entities {
e624ef9d66 2025-04-20 arcade: if entity.type_field == MessageEntityType::BotCommand && entity.offset != 0 {
e624ef9d66 2025-04-20 arcade: bail!("commands should be at message start");
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: let cmd = String::from_utf16_lossy(&chars[entity.offset as usize..entity.length as usize]);
e624ef9d66 2025-04-20 arcade: let words: Vec<&str> = text.split_whitespace().collect();
e624ef9d66 2025-04-20 arcade: match cmd.as_ref() {
e624ef9d66 2025-04-20 arcade: "/check" | "/clean" | "/enable" | "/delete" | "/disable" => { command::command(self, msg.chat.id, words).await? },
e624ef9d66 2025-04-20 arcade: "/start" => { command::start(self, msg.chat.id).await?; },
e624ef9d66 2025-04-20 arcade: "/list" => { command::list(self, msg.chat.id).await?; },
e624ef9d66 2025-04-20 arcade: "/add" | "/update" => { command::update(self, msg.chat.id, words).await?; },
e624ef9d66 2025-04-20 arcade: any => {
e624ef9d66 2025-04-20 arcade: self.send(format!("\\#error\n```\nUnknown command: {any}\n```"),
e624ef9d66 2025-04-20 arcade: Some(msg.chat.id),
e624ef9d66 2025-04-20 arcade: Some(ParseMode::MarkdownV2)
e624ef9d66 2025-04-20 arcade: ).await?;
e624ef9d66 2025-04-20 arcade: },
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: };
e624ef9d66 2025-04-20 arcade: }
e624ef9d66 2025-04-20 arcade: }
e624ef9d66 2025-04-20 arcade: }
e624ef9d66 2025-04-20 arcade:
e624ef9d66 2025-04-20 arcade: pub async fn send <S>(&self, msg: S, target: Option<i64>, mode: Option<ParseMode>) -> Result<()>
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);
e624ef9d66 2025-04-20 arcade: let send = SendMessageParams::builder()
e624ef9d66 2025-04-20 arcade: .chat_id(target)
e624ef9d66 2025-04-20 arcade: .text(msg)
e624ef9d66 2025-04-20 arcade: .parse_mode(mode)
e624ef9d66 2025-04-20 arcade: .build();
e624ef9d66 2025-04-20 arcade: self.tg.send_message(&send).await?;
e624ef9d66 2025-04-20 arcade: Ok(())
e624ef9d66 2025-04-20 arcade: }
e624ef9d66 2025-04-20 arcade:
0340541002 2025-04-24 arcade: pub async fn check (&mut self, id: &i32, owner: i64, real: bool) -> Result<String> {
e624ef9d66 2025-04-20 arcade: let mut posted: i32 = 0;
0340541002 2025-04-24 arcade: let mut conn = self.db.begin().await?;
285ce2bc31 2024-09-17 arcade:
9171c791eb 2021-11-13 arcade: let id = {
9171c791eb 2021-11-13 arcade: let mut set = self.sources.lock().unwrap();
9171c791eb 2021-11-13 arcade: match set.get(id) {
9171c791eb 2021-11-13 arcade: Some(id) => id.clone(),
9171c791eb 2021-11-13 arcade: None => {
9171c791eb 2021-11-13 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 {
0340541002 2025-04-24 arcade: let source = conn.get_source(*id, owner).await?;
61899b5618 2025-04-26 arcade: conn.set_scrape(*id).await?;
9171c791eb 2021-11-13 arcade: let destination = match real {
e624ef9d66 2025-04-20 arcade: true => source.channel_id,
e624ef9d66 2025-04-20 arcade: false => source.owner,
9171c791eb 2021-11-13 arcade: };
9171c791eb 2021-11-13 arcade: let mut this_fetch: Option<DateTime<chrono::FixedOffset>> = None;
9171c791eb 2021-11-13 arcade: let mut posts: BTreeMap<DateTime<chrono::FixedOffset>, String> = BTreeMap::new();
45e34762e4 2023-05-28 arcade:
9910c2209c 2023-08-04 arcade: let response = self.http_client.get(&source.url).send().await?;
613a665847 2021-11-15 arcade: let status = response.status();
613a665847 2021-11-15 arcade: let content = response.bytes().await?;
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]),
f988dfd28f 2022-02-13 arcade: }?;
f988dfd28f 2022-02-13 arcade: let url = link;
26339860ce 2022-02-13 arcade: posts.insert(date, url.to_string());
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();
2c36f015d8 2024-11-23 arcade: let url = item.links()[0].href();
2c36f015d8 2024-11-23 arcade: posts.insert(*date, url.to_string());
2c36f015d8 2024-11-23 arcade: };
2c36f015d8 2024-11-23 arcade: },
2c36f015d8 2024-11-23 arcade: Err(err) => {
e624ef9d66 2025-04-20 arcade: bail!("Unsupported or mangled content:\n{:?}\n{err:#?}\n{status:#?}\n", &source.url)
2c36f015d8 2024-11-23 arcade: },
2c36f015d8 2024-11-23 arcade: }
9171c791eb 2021-11-13 arcade: },
9171c791eb 2021-11-13 arcade: rss::Error::Eof => (),
e624ef9d66 2025-04-20 arcade: _ => bail!("Unsupported or mangled content:\n{:?}\n{err:#?}\n{status:#?}\n", &source.url)
9171c791eb 2021-11-13 arcade: }
9171c791eb 2021-11-13 arcade: };
9171c791eb 2021-11-13 arcade: for (date, url) in posts.iter() {
9910c2209c 2023-08-04 arcade: let post_url: Cow<str> = match source.url_re {
3dc9cddd4d 2023-08-04 arcade: Some(ref x) => sedregex::ReplaceCommand::new(x)?.execute(url),
659724c658 2021-12-08 arcade: None => url.into(),
659724c658 2021-12-08 arcade: };
0340541002 2025-04-24 arcade: if let Some(exists) = conn.exists(&post_url, *id).await? {
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}"),
e624ef9d66 2025-04-20 arcade: }, Some(destination), Some(ParseMode::Html)).await?;
0340541002 2025-04-24 arcade: conn.add_post(*id, date, &post_url).await?;
659724c658 2021-12-08 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}"))
0340541002 2025-04-24 arcade: }
0340541002 2025-04-24 arcade:
0340541002 2025-04-24 arcade: async fn autofetch(&mut self) -> Result<std::time::Duration> {
0340541002 2025-04-24 arcade: let mut delay = chrono::Duration::minutes(1);
0340541002 2025-04-24 arcade: let now = chrono::Local::now();
0340541002 2025-04-24 arcade: let mut conn = self.db.begin().await?;
0340541002 2025-04-24 arcade: for row in conn.get_queue().await? {
9910c2209c 2023-08-04 arcade: if let Some(next_fetch) = row.next_fetch {
9910c2209c 2023-08-04 arcade: if next_fetch < now {
9910c2209c 2023-08-04 arcade: if let (Some(owner), Some(source_id)) = (row.owner, row.source_id) {
0340541002 2025-04-24 arcade: let mut clone = Core {
e624ef9d66 2025-04-20 arcade: owner_chat: owner,
9910c2209c 2023-08-04 arcade: ..self.clone()
9910c2209c 2023-08-04 arcade: };
9910c2209c 2023-08-04 arcade: task::spawn(async move {
9910c2209c 2023-08-04 arcade: if let Err(err) = clone.check(&source_id, owner, true).await {
e624ef9d66 2025-04-20 arcade: if let Err(err) = clone.send(&format!("š {err:?}"), None, None).await {
e624ef9d66 2025-04-20 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: };
9910c2209c 2023-08-04 arcade: });
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: };
26339860ce 2022-02-13 arcade: Ok(delay.to_std()?)
26339860ce 2022-02-13 arcade: }
26339860ce 2022-02-13 arcade:
0340541002 2025-04-24 arcade: pub async fn list (&mut self, owner: i64) -> Result<String> {
26339860ce 2022-02-13 arcade: let mut reply: Vec<Cow<str>> = vec![];
26339860ce 2022-02-13 arcade: reply.push("Channels:".into());
0340541002 2025-04-24 arcade: let mut conn = self.db.begin().await?;
0340541002 2025-04-24 arcade: for row in conn.get_list(owner).await? {
79c91a5357 2024-08-02 arcade: reply.push(format!("\n\\#ļøā£ {} \\*ļøā£ `{}` {}\nš `{}`", row.source_id, row.channel,
9910c2209c 2023-08-04 arcade: match row.enabled {
26339860ce 2022-02-13 arcade: true => "š enabled",
26339860ce 2022-02-13 arcade: false => "ā disabled",
9910c2209c 2023-08-04 arcade: }, row.url).into());
9910c2209c 2023-08-04 arcade: if let Some(hash) = &row.iv_hash {
e624ef9d66 2025-04-20 arcade: reply.push(format!("IV: `{hash}`").into());
26339860ce 2022-02-13 arcade: }
9910c2209c 2023-08-04 arcade: if let Some(re) = &row.url_re {
e624ef9d66 2025-04-20 arcade: reply.push(format!("RE: `{re}`").into());
a7f91033c0 2021-11-13 arcade: }
a7f91033c0 2021-11-13 arcade: };
a7f91033c0 2021-11-13 arcade: Ok(reply.join("\n"))
e624ef9d66 2025-04-20 arcade: }
9171c791eb 2021-11-13 arcade: }