Lines of
src/core.rs
from check-in 704bf85f8c
that are changed by the sequence of edits moving toward
check-in 5d11e7d390:
1: use crate::{
2: command,
3: sql::Db,
4: };
5:
6: use std::{
7: borrow::Cow,
8: collections::{
9: BTreeMap,
10: HashSet,
11: },
12: sync::Arc,
13: };
14:
15: use async_compat::Compat;
16: use chrono::{
17: DateTime,
18: Local,
19: };
20: use lazy_static::lazy_static;
21: use regex::Regex;
22: use reqwest::header::{
23: CACHE_CONTROL,
24: EXPIRES,
25: LAST_MODIFIED
26: };
27: use smol::{
28: Timer,
29: lock::Mutex,
30: };
31: use tgbot::{
32: api::Client,
33: handler::UpdateHandler,
34: types::{
35: Bot,
36: ChatPeerId,
37: Command,
38: GetBot,
39: Message,
40: ParseMode,
41: SendMessage,
42: Update,
43: UpdateType,
44: UserPeerId,
45: },
46: };
47: use stacked_errors::{
48: Result,
49: StackableErr,
50: anyhow,
51: bail,
52: };
53:
54: lazy_static!{
55: pub static ref RE_SPECIAL: Regex = Regex::new(r"([\-_*\[\]()~`>#+|{}\.!])").unwrap();
56: }
57:
58: /// Escape characters that are special in Telegram HTML by prefixing them with a backslash.
59: ///
60: /// This ensures the returned string can be used as HTML-formatted Telegram message content
61: /// without special characters being interpreted as HTML markup.
62: pub fn encode (text: &str) -> Cow<'_, str> {
63: RE_SPECIAL.replace_all(text, "\\$1")
64: }
65:
66: // This one does nothing except making sure only one token exists for each id
67: pub struct Token {
68: running: Arc<Mutex<HashSet<i32>>>,
69: my_id: i32,
70: }
71:
72: impl Token {
73: /// Attempts to acquire a per-id token by inserting `my_id` into the shared `running` set.
74: ///
75: /// If the id was not already present, the function inserts it and returns `Some(Token)`.
76: /// When the returned `Token` is dropped, the id will be removed from the `running` set,
77: /// allowing subsequent acquisitions for the same id.
78: ///
79: /// # Parameters
80: ///
81: /// - `running`: Shared set tracking active ids.
82: /// - `my_id`: Identifier to acquire a token for.
83: ///
84: /// # Returns
85: ///
86: /// `Ok(Token)` if the id was successfully acquired, `Error` if a token for the id is already active.
87: async fn new (running: &Arc<Mutex<HashSet<i32>>>, my_id: i32) -> Result<Token> {
88: let running = running.clone();
89: let mut set = running.lock_arc().await;
90: if set.contains(&my_id) {
91: bail!("Token already taken");
92: } else {
93: set.insert(my_id);
94: Ok(Token {
95: running,
96: my_id,
97: })
98: }
99: }
100: }
101:
102: impl Drop for Token {
103: /// Releases this token's claim on the shared running-set when the token is dropped.
104: ///
105: /// The token's identifier is removed from the shared `running` set so that future
106: /// operations for the same id may proceed.
107: fn drop (&mut self) {
108: smol::block_on(async {
109: let mut set = self.running.lock_arc().await;
110: set.remove(&self.my_id);
111: })
112: }
113: }
114:
115: #[derive(Clone)]
116: pub struct Core {
117: owner_chat: ChatPeerId,
118: // max_delay: u16,
119: pub tg: Client,
120: pub me: Bot,
121: pub db: Db,
122: running: Arc<Mutex<HashSet<i32>>>,
123: http_client: reqwest::Client,
124: }
125:
126: pub struct Post {
127: uri: String,
128: title: String,
129: authors: String,
130: summary: String,
131: }
132:
133: impl Core {
134: /// Create a Core instance from configuration and start its background autofetch loop.
135: ///
136: /// The provided `settings` must include:
137: /// - `owner` (integer): chat id to use as the default destination,
138: /// - `api_key` (string): Telegram bot API key,
139: /// - `api_gateway` (string): Telegram API gateway host,
140: /// - `pg` (string): PostgreSQL connection string,
141: /// - optional `proxy` (string): proxy URL for the HTTP client.
142: ///
143: /// On success returns an initialized `Core` with Telegram and HTTP clients, database connection,
144: /// an empty running set for per-id tokens, and a spawned background task that periodically runs
145: /// `autofetch`. If any required setting is missing or initialization fails, an error is returned.
146: pub async fn new(settings: config::Config) -> Result<Core> {
147: let owner_chat = ChatPeerId::from(settings.get_int("owner").stack()?);
148: let api_key = settings.get_string("api_key").stack()?;
149: let tg = Client::new(&api_key).stack()?
150: .with_host(settings.get_string("api_gateway").stack()?);
151:
152: let mut client = reqwest::Client::builder();
153: if let Ok(proxy) = settings.get_string("proxy") {
154: let proxy = reqwest::Proxy::all(proxy).stack()?;
155: client = client.proxy(proxy);
156: }
157: let http_client = client.build().stack()?;
158: let me = tg.execute(GetBot).await.stack()?;
159: let core = Core {
160: tg,
161: me,
162: owner_chat,
163: db: Db::new(&settings.get_string("pg").stack()?)?,
164: running: Arc::new(Mutex::new(HashSet::new())),
165: http_client,
166: // max_delay: 60,
167: };
168: let clone = core.clone();
169: smol::spawn(Compat::new(async move {
170: loop {
171: let delay = match &clone.autofetch().await {
172: Err(err) => {
173: if let Err(err) = clone.send(format!("🛑 {err}"), None, None).await {
174: eprintln!("Autofetch error: {err:?}");
175: };
176: std::time::Duration::from_secs(60)
177: },
178: Ok(time) => *time,
179: };
180: Timer::after(delay).await;
181: }
182: })).detach();
183: Ok(core)
184: }
185:
186: pub async fn send <S>(&self, msg: S, target: Option<ChatPeerId>, mode: Option<ParseMode>) -> Result<Message>
187: where S: Into<String> {
188: let msg = msg.into();
189:
190: let mode = mode.unwrap_or(ParseMode::Html);
191: let target = target.unwrap_or(self.owner_chat);
192: self.tg.execute(
193: SendMessage::new(target, msg)
194: .with_parse_mode(mode)
195: ).await.stack()
196: }
197:
198: /// Fetches the feed for a source, sends any newly discovered posts to the appropriate chat, and records them in the database.
199: ///
200: /// This acquires a per-source guard to prevent concurrent checks for the same `id`. If a check is already running for
201: /// the given `id`, the function returns an error. If `last_scrape` is provided, it is sent as the `If-Modified-Since`
202: /// header to the feed request. The function parses RSS or Atom feeds, sends unseen post URLs to either the source's
203: /// channel (when `real` is true) or the source owner (when `real` is false), and persists posted entries so they are
204: /// not reposted later.
205: ///
206: /// Parameters:
207: /// - `id`: Identifier of the source to check.
208: /// - `real`: When `true`, send posts to the source's channel; when `false`, send to the source owner.
209: /// - `last_scrape`: Optional timestamp used to set the `If-Modified-Since` header for the HTTP request.
210: ///
211: /// # Returns
212: ///
213: /// `Posted: N` where `N` is the number of posts processed and sent.
214: pub async fn check (&self, id: i32, real: bool, last_scrape: Option<DateTime<Local>>) -> Result<String> {
215: let mut posted: i32 = 0;
216: let mut conn = self.db.begin().await.stack()?;
217:
218: let _token = Token::new(&self.running, id).await.stack()?;
219: let source = conn.get_source(id, self.owner_chat).await.stack()?;
220: conn.set_scrape(id).await.stack()?;
221: let destination = ChatPeerId::from(match real {
222: true => source.channel_id,
223: false => source.owner,
224: });
225: let mut this_fetch: Option<DateTime<chrono::FixedOffset>> = None;
226: let mut posts: BTreeMap<DateTime<chrono::FixedOffset>, Post> = BTreeMap::new();
227:
228: let mut builder = self.http_client.get(&source.url);
229: if let Some(last_scrape) = last_scrape {
230: builder = builder.header(LAST_MODIFIED, last_scrape.to_rfc2822());
231: };
232: let response = builder.send().await.stack()?;
233: {
234: let headers = response.headers();
235: let expires = headers.get(EXPIRES);
236: let cache = headers.get(CACHE_CONTROL);
237: if expires.is_some() || cache.is_some() {
238: println!("{} {} {:?} {:?} {:?}", Local::now().to_rfc2822(), &source.url, last_scrape, expires, cache);
239: }
240: }
241: let status = response.status();
242: let content = response.bytes().await.stack()?;
243: match rss::Channel::read_from(&content[..]) {
244: Ok(feed) => {
245: for item in feed.items() {
246: if let Some(link) = item.link() {
247: let date = match item.pub_date() {
248: Some(feed_date) => DateTime::parse_from_rfc2822(feed_date),
249: None => DateTime::parse_from_rfc3339(match item.dublin_core_ext() {
250: Some(ext) => {
251: let dates = ext.dates();
252: if dates.is_empty() {
253: bail!("Feed item has Dublin Core extension but no dates.")
254: } else {
255: &dates[0]
256: }
257: },
258: None => bail!("Feed item misses posting date."),
259: }),
260: }.stack()?;
261: let uri = link.to_string();
262: let title = item.title().unwrap_or("").to_string();
263: let authors = item.author().unwrap_or("").to_string();
264: let summary = item.content().unwrap_or("").to_string();
265: posts.insert(date, Post{
266: uri,
267: title,
268: authors,
269: summary,
270: });
271: }
272: };
273: },
274: Err(err) => match err {
275: rss::Error::InvalidStartTag => {
276: match atom_syndication::Feed::read_from(&content[..]) {
277: Ok(feed) => {
278: for item in feed.entries() {
704bf85f8c 2026-01-06 279: let date = item.published().unwrap();
704bf85f8c 2026-01-06 280: let uri = item.links()[0].href().to_string();
281: let title = item.title().to_string();
282: let authors = item.authors().iter().map(|x| format!("{} <{:?}>", x.name(), x.email())).collect::<Vec<String>>().join(", ");
283: let summary = if let Some(sum) = item.summary() { sum.value.clone() } else { String::new() };
284: posts.insert(*date, Post{
285: uri,
286: title,
287: authors,
288: summary,
289: });
290: };
291: },
292: Err(err) => {
293: bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
294: },
295: }
296: },
297: rss::Error::Eof => (),
298: _ => bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
299: }
300: };
301: for (date, post) in posts.iter() {
302: let post_url: Cow<str> = match source.url_re {
303: Some(ref x) => sedregex::ReplaceCommand::new(x).stack()?.execute(&post.uri),
304: None => post.uri.clone().into(),
305: };
306: if let Some(exists) = conn.exists(&post_url, id).await.stack()? {
307: if ! exists {
308: if this_fetch.is_none() || *date > this_fetch.unwrap() {
309: this_fetch = Some(*date);
310: };
311: self.send( match &source.iv_hash {
312: Some(hash) => format!("<a href=\"https://t.me/iv?url={post_url}&rhash={hash}\"> </a>{post_url}"),
313: None => format!("{post_url}"),
314: }, Some(destination), Some(ParseMode::Html)).await.stack()?;
315: conn.add_post(id, date, &post_url).await.stack()?;
316: };
317: };
318: posted += 1;
319: };
320: posts.clear();
321: Ok(format!("Posted: {posted}"))
322: }
323:
324: async fn autofetch(&self) -> Result<std::time::Duration> {
325: let mut delay = chrono::Duration::minutes(1);
326: let now = chrono::Local::now();
327: let queue = {
328: let mut conn = self.db.begin().await.stack()?;
329: conn.get_queue().await.stack()?
330: };
331: for row in queue {
332: if let Some(next_fetch) = row.next_fetch {
333: if next_fetch < now {
334: if let (Some(owner), Some(source_id), last_scrape) = (row.owner, row.source_id, row.last_scrape) {
335: let clone = Core {
336: owner_chat: ChatPeerId::from(owner),
337: ..self.clone()
338: };
339: let source = {
340: let mut conn = self.db.begin().await.stack()?;
341: match conn.get_one(owner, source_id).await {
342: Ok(Some(source)) => source.to_string(),
343: Ok(None) => "Source not found in database?".to_string(),
344: Err(err) => format!("Failed to fetch source data:\n{err}"),
345: }
346: };
347: smol::spawn(Compat::new(async move {
348: if let Err(err) = clone.check(source_id, true, Some(last_scrape)).await {
349: if let Err(err) = clone.send(&format!("🛑 {source}\n{}", encode(&err.to_string())), None, Some(ParseMode::MarkdownV2)).await {
350: eprintln!("Check error: {err}");
351: // clone.disable(&source_id, owner).await.unwrap();
352: };
353: };
354: })).detach();
355: }
356: } else if next_fetch - now < delay {
357: delay = next_fetch - now;
358: }
359: }
360: };
361: delay.to_std().stack()
362: }
363:
364: pub async fn list (&self, owner: UserPeerId) -> Result<String> {
365: let mut reply: Vec<String> = vec![];
366: reply.push("Channels:".into());
367: let mut conn = self.db.begin().await.stack()?;
368: for row in conn.get_list(owner).await.stack()? {
369: reply.push(row.to_string());
370: };
371: Ok(reply.join("\n\n"))
372: }
373: }
374:
375: impl UpdateHandler for Core {
376: async fn handle (&self, update: Update) {
377: if let UpdateType::Message(msg) = update.update_type {
378: if let Ok(cmd) = Command::try_from(msg) {
379: let msg = cmd.get_message();
380: let words = cmd.get_args();
381: let command = cmd.get_name();
382: let res = match command {
383: "/check" | "/clean" | "/enable" | "/delete" | "/disable" => command::command(self, command, msg, words).await,
384: "/start" => command::start(self, msg).await,
385: "/list" => command::list(self, msg).await,
386: "/add" | "/update" => command::update(self, command, msg, words).await,
387: any => Err(anyhow!("Unknown command: {any}")),
388: };
389: if let Err(err) = res {
390: if let Err(err2) = self.send(format!("\\#error\n```\n{err}\n```"),
391: Some(msg.chat.get_id()),
392: Some(ParseMode::MarkdownV2)
393: ).await{
394: dbg!(err2);
395: };
396: }
397: };
398: };
399: }
400: }