0340541002 2025-04-24 1: use crate::{
0340541002 2025-04-24 2: command,
0340541002 2025-04-24 3: sql::Db,
0340541002 2025-04-24 4: };
1c444d34ff 2024-08-28 5:
9171c791eb 2021-11-13 6: use std::{
c1e27b74ed 2021-11-13 7: borrow::Cow,
9171c791eb 2021-11-13 8: collections::{
9171c791eb 2021-11-13 9: BTreeMap,
9171c791eb 2021-11-13 10: HashSet,
9171c791eb 2021-11-13 11: },
c6d3e97290 2025-07-01 12: };
c6d3e97290 2025-07-01 13:
c6d3e97290 2025-07-01 14: use async_std::{
c6d3e97290 2025-07-01 15: task,
1c444d34ff 2024-08-28 16: sync::{
1c444d34ff 2024-08-28 17: Arc,
1c444d34ff 2024-08-28 18: Mutex
1c444d34ff 2024-08-28 19: },
1c444d34ff 2024-08-28 20: };
acb0a4ac54 2025-09-28 21: use chrono::{
acb0a4ac54 2025-09-28 22: DateTime,
acb0a4ac54 2025-09-28 23: Local,
acb0a4ac54 2025-09-28 24: };
5a4aab7687 2025-06-29 25: use lazy_static::lazy_static;
5a4aab7687 2025-06-29 26: use regex::Regex;
acb0a4ac54 2025-09-28 27: use reqwest::header::{
acb0a4ac54 2025-09-28 28: CACHE_CONTROL,
acb0a4ac54 2025-09-28 29: EXPIRES,
acb0a4ac54 2025-09-28 30: LAST_MODIFIED
acb0a4ac54 2025-09-28 31: };
bb89b6fab8 2025-06-15 32: use tgbot::{
bb89b6fab8 2025-06-15 33: api::Client,
bb89b6fab8 2025-06-15 34: handler::UpdateHandler,
bb89b6fab8 2025-06-15 35: types::{
bb89b6fab8 2025-06-15 36: Bot,
bb89b6fab8 2025-06-15 37: ChatPeerId,
bb89b6fab8 2025-06-15 38: Command,
bb89b6fab8 2025-06-15 39: GetBot,
bb89b6fab8 2025-06-15 40: Message,
bb89b6fab8 2025-06-15 41: ParseMode,
bb89b6fab8 2025-06-15 42: SendMessage,
bb89b6fab8 2025-06-15 43: Update,
bb89b6fab8 2025-06-15 44: UpdateType,
bb89b6fab8 2025-06-15 45: UserPeerId,
bb89b6fab8 2025-06-15 46: },
44575a91d3 2025-07-09 47: };
44575a91d3 2025-07-09 48: use stacked_errors::{
44575a91d3 2025-07-09 49: Result,
44575a91d3 2025-07-09 50: StackableErr,
44575a91d3 2025-07-09 51: anyhow,
44575a91d3 2025-07-09 52: bail,
5a4aab7687 2025-06-29 53: };
5a4aab7687 2025-06-29 54:
5a4aab7687 2025-06-29 55: lazy_static!{
5a4aab7687 2025-06-29 56: pub static ref RE_SPECIAL: Regex = Regex::new(r"([\-_*\[\]()~`>#+|{}\.!])").unwrap();
5a4aab7687 2025-06-29 57: }
5a4aab7687 2025-06-29 58:
5a4aab7687 2025-06-29 59: /// Encodes special HTML entities to prevent them interfering with Telegram HTML
5a4aab7687 2025-06-29 60: pub fn encode (text: &str) -> Cow<'_, str> {
5a4aab7687 2025-06-29 61: RE_SPECIAL.replace_all(text, "\\$1")
5a4aab7687 2025-06-29 62: }
9171c791eb 2021-11-13 63:
9171c791eb 2021-11-13 64: #[derive(Clone)]
9171c791eb 2021-11-13 65: pub struct Core {
bb89b6fab8 2025-06-15 66: owner_chat: ChatPeerId,
bb89b6fab8 2025-06-15 67: // max_delay: u16,
bb89b6fab8 2025-06-15 68: pub tg: Client,
bb89b6fab8 2025-06-15 69: pub me: Bot,
0340541002 2025-04-24 70: pub db: Db,
9171c791eb 2021-11-13 71: sources: Arc<Mutex<HashSet<Arc<i32>>>>,
45e34762e4 2023-05-28 72: http_client: reqwest::Client,
9171c791eb 2021-11-13 73: }
9171c791eb 2021-11-13 74:
9171c791eb 2021-11-13 75: impl Core {
0340541002 2025-04-24 76: pub async fn new(settings: config::Config) -> Result<Core> {
44575a91d3 2025-07-09 77: let owner_chat = ChatPeerId::from(settings.get_int("owner").stack()?);
44575a91d3 2025-07-09 78: let api_key = settings.get_string("api_key").stack()?;
44575a91d3 2025-07-09 79: let tg = Client::new(&api_key).stack()?;
45e34762e4 2023-05-28 80:
45e34762e4 2023-05-28 81: let mut client = reqwest::Client::builder();
9d8a6738fd 2023-07-30 82: if let Ok(proxy) = settings.get_string("proxy") {
44575a91d3 2025-07-09 83: let proxy = reqwest::Proxy::all(proxy).stack()?;
45e34762e4 2023-05-28 84: client = client.proxy(proxy);
45e34762e4 2023-05-28 85: }
44575a91d3 2025-07-09 86: let http_client = client.build().stack()?;
44575a91d3 2025-07-09 87: let me = tg.execute(GetBot).await.stack()?;
0340541002 2025-04-24 88: let core = Core {
f988dfd28f 2022-02-13 89: tg,
e624ef9d66 2025-04-20 90: me,
e624ef9d66 2025-04-20 91: owner_chat,
44575a91d3 2025-07-09 92: db: Db::new(&settings.get_string("pg").stack()?)?,
9171c791eb 2021-11-13 93: sources: Arc::new(Mutex::new(HashSet::new())),
45e34762e4 2023-05-28 94: http_client,
bb89b6fab8 2025-06-15 95: // max_delay: 60,
0340541002 2025-04-24 96: };
bb89b6fab8 2025-06-15 97: let clone = core.clone();
cb86e770f9 2022-03-15 98: task::spawn(async move {
26339860ce 2022-02-13 99: loop {
26339860ce 2022-02-13 100: let delay = match &clone.autofetch().await {
26339860ce 2022-02-13 101: Err(err) => {
07b34bcad6 2025-08-05 102: if let Err(err) = clone.send(format!("š {err}"), None, None).await {
e624ef9d66 2025-04-20 103: eprintln!("Autofetch error: {err:?}");
26339860ce 2022-02-13 104: };
cb86e770f9 2022-03-15 105: std::time::Duration::from_secs(60)
26339860ce 2022-02-13 106: },
26339860ce 2022-02-13 107: Ok(time) => *time,
9171c791eb 2021-11-13 108: };
cb86e770f9 2022-03-15 109: task::sleep(delay).await;
9171c791eb 2021-11-13 110: }
e624ef9d66 2025-04-20 111: });
9171c791eb 2021-11-13 112: Ok(core)
9171c791eb 2021-11-13 113: }
9171c791eb 2021-11-13 114:
bb89b6fab8 2025-06-15 115: pub async fn send <S>(&self, msg: S, target: Option<ChatPeerId>, mode: Option<ParseMode>) -> Result<Message>
e624ef9d66 2025-04-20 116: where S: Into<String> {
e624ef9d66 2025-04-20 117: let msg = msg.into();
e624ef9d66 2025-04-20 118:
e624ef9d66 2025-04-20 119: let mode = mode.unwrap_or(ParseMode::Html);
e624ef9d66 2025-04-20 120: let target = target.unwrap_or(self.owner_chat);
44575a91d3 2025-07-09 121: self.tg.execute(
bb89b6fab8 2025-06-15 122: SendMessage::new(target, msg)
bb89b6fab8 2025-06-15 123: .with_parse_mode(mode)
44575a91d3 2025-07-09 124: ).await.stack()
bb89b6fab8 2025-06-15 125: }
bb89b6fab8 2025-06-15 126:
acb0a4ac54 2025-09-28 127: pub async fn check (&self, id: i32, real: bool, last_scrape: Option<DateTime<Local>>) -> Result<String> {
bb89b6fab8 2025-06-15 128: let mut posted: i32 = 0;
44575a91d3 2025-07-09 129: let mut conn = self.db.begin().await.stack()?;
bb89b6fab8 2025-06-15 130:
bb89b6fab8 2025-06-15 131: let id = {
c6d3e97290 2025-07-01 132: let mut set = self.sources.lock_arc().await;
bb89b6fab8 2025-06-15 133: match set.get(&id) {
bb89b6fab8 2025-06-15 134: Some(id) => id.clone(),
bb89b6fab8 2025-06-15 135: None => {
bb89b6fab8 2025-06-15 136: let id = Arc::new(id);
9171c791eb 2021-11-13 137: set.insert(id.clone());
9171c791eb 2021-11-13 138: id.clone()
9171c791eb 2021-11-13 139: },
9171c791eb 2021-11-13 140: }
9171c791eb 2021-11-13 141: };
9171c791eb 2021-11-13 142: let count = Arc::strong_count(&id);
9171c791eb 2021-11-13 143: if count == 2 {
44575a91d3 2025-07-09 144: let source = conn.get_source(*id, self.owner_chat).await.stack()?;
44575a91d3 2025-07-09 145: conn.set_scrape(*id).await.stack()?;
bb89b6fab8 2025-06-15 146: let destination = ChatPeerId::from(match real {
e624ef9d66 2025-04-20 147: true => source.channel_id,
e624ef9d66 2025-04-20 148: false => source.owner,
bb89b6fab8 2025-06-15 149: });
9171c791eb 2021-11-13 150: let mut this_fetch: Option<DateTime<chrono::FixedOffset>> = None;
9171c791eb 2021-11-13 151: let mut posts: BTreeMap<DateTime<chrono::FixedOffset>, String> = BTreeMap::new();
45e34762e4 2023-05-28 152:
acb0a4ac54 2025-09-28 153: let mut builder = self.http_client.get(&source.url);
acb0a4ac54 2025-09-28 154: if let Some(last_scrape) = last_scrape {
acb0a4ac54 2025-09-28 155: builder = builder.header(LAST_MODIFIED, last_scrape.to_rfc2822());
acb0a4ac54 2025-09-28 156: };
acb0a4ac54 2025-09-28 157: let response = builder.send().await.stack()?;
acb0a4ac54 2025-09-28 158: {
acb0a4ac54 2025-09-28 159: let headers = response.headers();
acb0a4ac54 2025-09-28 160: let expires = headers.get(EXPIRES);
acb0a4ac54 2025-09-28 161: let cache = headers.get(CACHE_CONTROL);
acb0a4ac54 2025-09-28 162: if expires.is_some() || cache.is_some() {
acb0a4ac54 2025-09-28 163: println!("{} {} {:?} {:?} {:?}", Local::now().to_rfc2822(), &source.url, last_scrape, expires, cache);
acb0a4ac54 2025-09-28 164: }
acb0a4ac54 2025-09-28 165: }
613a665847 2021-11-15 166: let status = response.status();
44575a91d3 2025-07-09 167: let content = response.bytes().await.stack()?;
9171c791eb 2021-11-13 168: match rss::Channel::read_from(&content[..]) {
9171c791eb 2021-11-13 169: Ok(feed) => {
9171c791eb 2021-11-13 170: for item in feed.items() {
f988dfd28f 2022-02-13 171: if let Some(link) = item.link() {
f988dfd28f 2022-02-13 172: let date = match item.pub_date() {
f988dfd28f 2022-02-13 173: Some(feed_date) => DateTime::parse_from_rfc2822(feed_date),
f988dfd28f 2022-02-13 174: None => DateTime::parse_from_rfc3339(&item.dublin_core_ext().unwrap().dates()[0]),
44575a91d3 2025-07-09 175: }.stack()?;
f988dfd28f 2022-02-13 176: let url = link;
26339860ce 2022-02-13 177: posts.insert(date, url.to_string());
9171c791eb 2021-11-13 178: }
9171c791eb 2021-11-13 179: };
9171c791eb 2021-11-13 180: },
9171c791eb 2021-11-13 181: Err(err) => match err {
9171c791eb 2021-11-13 182: rss::Error::InvalidStartTag => {
2c36f015d8 2024-11-23 183: match atom_syndication::Feed::read_from(&content[..]) {
2c36f015d8 2024-11-23 184: Ok(feed) => {
2c36f015d8 2024-11-23 185: for item in feed.entries() {
2c36f015d8 2024-11-23 186: let date = item.published().unwrap();
2c36f015d8 2024-11-23 187: let url = item.links()[0].href();
2c36f015d8 2024-11-23 188: posts.insert(*date, url.to_string());
2c36f015d8 2024-11-23 189: };
2c36f015d8 2024-11-23 190: },
2c36f015d8 2024-11-23 191: Err(err) => {
07b34bcad6 2025-08-05 192: bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
2c36f015d8 2024-11-23 193: },
2c36f015d8 2024-11-23 194: }
9171c791eb 2021-11-13 195: },
9171c791eb 2021-11-13 196: rss::Error::Eof => (),
07b34bcad6 2025-08-05 197: _ => bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
9171c791eb 2021-11-13 198: }
9171c791eb 2021-11-13 199: };
9171c791eb 2021-11-13 200: for (date, url) in posts.iter() {
9910c2209c 2023-08-04 201: let post_url: Cow<str> = match source.url_re {
44575a91d3 2025-07-09 202: Some(ref x) => sedregex::ReplaceCommand::new(x).stack()?.execute(url),
659724c658 2021-12-08 203: None => url.into(),
659724c658 2021-12-08 204: };
44575a91d3 2025-07-09 205: if let Some(exists) = conn.exists(&post_url, *id).await.stack()? {
9910c2209c 2023-08-04 206: if ! exists {
9910c2209c 2023-08-04 207: if this_fetch.is_none() || *date > this_fetch.unwrap() {
9910c2209c 2023-08-04 208: this_fetch = Some(*date);
9910c2209c 2023-08-04 209: };
e624ef9d66 2025-04-20 210: self.send( match &source.iv_hash {
e624ef9d66 2025-04-20 211: Some(hash) => format!("<a href=\"https://t.me/iv?url={post_url}&rhash={hash}\"> </a>{post_url}"),
e624ef9d66 2025-04-20 212: None => format!("{post_url}"),
44575a91d3 2025-07-09 213: }, Some(destination), Some(ParseMode::Html)).await.stack()?;
44575a91d3 2025-07-09 214: conn.add_post(*id, date, &post_url).await.stack()?;
659724c658 2021-12-08 215: };
9171c791eb 2021-11-13 216: };
9171c791eb 2021-11-13 217: posted += 1;
9171c791eb 2021-11-13 218: };
9171c791eb 2021-11-13 219: posts.clear();
9171c791eb 2021-11-13 220: };
0340541002 2025-04-24 221: Ok(format!("Posted: {posted}"))
26339860ce 2022-02-13 222: }
26339860ce 2022-02-13 223:
bb89b6fab8 2025-06-15 224: async fn autofetch(&self) -> Result<std::time::Duration> {
26339860ce 2022-02-13 225: let mut delay = chrono::Duration::minutes(1);
9910c2209c 2023-08-04 226: let now = chrono::Local::now();
bb89b6fab8 2025-06-15 227: let queue = {
44575a91d3 2025-07-09 228: let mut conn = self.db.begin().await.stack()?;
44575a91d3 2025-07-09 229: conn.get_queue().await.stack()?
bb89b6fab8 2025-06-15 230: };
bb89b6fab8 2025-06-15 231: for row in queue {
9910c2209c 2023-08-04 232: if let Some(next_fetch) = row.next_fetch {
9910c2209c 2023-08-04 233: if next_fetch < now {
acb0a4ac54 2025-09-28 234: if let (Some(owner), Some(source_id), last_scrape) = (row.owner, row.source_id, row.last_scrape) {
bb89b6fab8 2025-06-15 235: let clone = Core {
bb89b6fab8 2025-06-15 236: owner_chat: ChatPeerId::from(owner),
9910c2209c 2023-08-04 237: ..self.clone()
9910c2209c 2023-08-04 238: };
b4af85e31b 2025-06-29 239: let source = {
44575a91d3 2025-07-09 240: let mut conn = self.db.begin().await.stack()?;
b4af85e31b 2025-06-29 241: match conn.get_one(owner, source_id).await {
b4af85e31b 2025-06-29 242: Ok(Some(source)) => source.to_string(),
44575a91d3 2025-07-09 243: Ok(None) => "Source not found in database.stack()?".to_string(),
b4af85e31b 2025-06-29 244: Err(err) => format!("Failed to fetch source data:\n{err}"),
b4af85e31b 2025-06-29 245: }
b4af85e31b 2025-06-29 246: };
9910c2209c 2023-08-04 247: task::spawn(async move {
acb0a4ac54 2025-09-28 248: if let Err(err) = clone.check(source_id, true, Some(last_scrape)).await {
c6d3e97290 2025-07-01 249: if let Err(err) = clone.send(&format!("{source}\n\nš {}", encode(&err.to_string())), None, Some(ParseMode::MarkdownV2)).await {
e624ef9d66 2025-04-20 250: eprintln!("Check error: {err:?}");
79c91a5357 2024-08-02 251: // clone.disable(&source_id, owner).await.unwrap();
9910c2209c 2023-08-04 252: };
9910c2209c 2023-08-04 253: };
9910c2209c 2023-08-04 254: });
9910c2209c 2023-08-04 255: }
9910c2209c 2023-08-04 256: } else if next_fetch - now < delay {
9910c2209c 2023-08-04 257: delay = next_fetch - now;
9910c2209c 2023-08-04 258: }
9910c2209c 2023-08-04 259: }
9910c2209c 2023-08-04 260: };
44575a91d3 2025-07-09 261: delay.to_std().stack()
9910c2209c 2023-08-04 262: }
9910c2209c 2023-08-04 263:
bb89b6fab8 2025-06-15 264: pub async fn list (&self, owner: UserPeerId) -> Result<String> {
b4af85e31b 2025-06-29 265: let mut reply: Vec<String> = vec![];
9910c2209c 2023-08-04 266: reply.push("Channels:".into());
44575a91d3 2025-07-09 267: let mut conn = self.db.begin().await.stack()?;
44575a91d3 2025-07-09 268: for row in conn.get_list(owner).await.stack()? {
b4af85e31b 2025-06-29 269: reply.push(row.to_string());
e624ef9d66 2025-04-20 270: };
b4af85e31b 2025-06-29 271: Ok(reply.join("\n\n"))
bb89b6fab8 2025-06-15 272: }
bb89b6fab8 2025-06-15 273: }
bb89b6fab8 2025-06-15 274:
bb89b6fab8 2025-06-15 275: impl UpdateHandler for Core {
bb89b6fab8 2025-06-15 276: async fn handle (&self, update: Update) {
bb89b6fab8 2025-06-15 277: if let UpdateType::Message(msg) = update.update_type {
bb89b6fab8 2025-06-15 278: if let Ok(cmd) = Command::try_from(msg) {
bb89b6fab8 2025-06-15 279: let msg = cmd.get_message();
bb89b6fab8 2025-06-15 280: let words = cmd.get_args();
fae13a0e55 2025-06-28 281: let command = cmd.get_name();
fae13a0e55 2025-06-28 282: let res = match command {
fae13a0e55 2025-06-28 283: "/check" | "/clean" | "/enable" | "/delete" | "/disable" => command::command(self, command, msg, words).await,
bb89b6fab8 2025-06-15 284: "/start" => command::start(self, msg).await,
bb89b6fab8 2025-06-15 285: "/list" => command::list(self, msg).await,
fae13a0e55 2025-06-28 286: "/add" | "/update" => command::update(self, command, msg, words).await,
bb89b6fab8 2025-06-15 287: any => Err(anyhow!("Unknown command: {any}")),
bb89b6fab8 2025-06-15 288: };
bb89b6fab8 2025-06-15 289: if let Err(err) = res {
07b34bcad6 2025-08-05 290: if let Err(err2) = self.send(format!("\\#error\n```\n{err}\n```"),
bb89b6fab8 2025-06-15 291: Some(msg.chat.get_id()),
bb89b6fab8 2025-06-15 292: Some(ParseMode::MarkdownV2)
bb89b6fab8 2025-06-15 293: ).await{
bb89b6fab8 2025-06-15 294: dbg!(err2);
bb89b6fab8 2025-06-15 295: };
bb89b6fab8 2025-06-15 296: }
bb89b6fab8 2025-06-15 297: };
bb89b6fab8 2025-06-15 298: };
e624ef9d66 2025-04-20 299: }
9171c791eb 2021-11-13 300: }