Lines of
src/core.rs
from check-in d8c1d259a2
that are changed by the sequence of edits moving toward
check-in d3dbaf11e1:
1: use crate::{
2: Arc,
3: command,
4: Mutex,
5: sql::Db,
6: tg_bot::{
7: Callback,
8: MyMessage,
9: Tg,
10: validate,
11: },
12: };
13:
14: use std::{
15: borrow::Cow,
16: collections::{
17: BTreeMap,
18: HashSet,
19: },
20: time::Duration,
21: };
22:
23: use async_compat::Compat;
24: use chrono::{
25: DateTime,
26: Local,
27: };
28: use lazy_static::lazy_static;
29: use regex::Regex;
30: use reqwest::header::LAST_MODIFIED;
31: use smol::Timer;
32: use stacked_errors::{
33: Result,
34: StackableErr,
35: anyhow,
36: bail,
37: };
38: use tgbot::{
39: handler::UpdateHandler,
40: types::{
41: CallbackQuery,
42: ChatPeerId,
43: Command,
44: Update,
45: UpdateType,
46: UserPeerId,
47: },
48: };
49: use ttl_cache::TtlCache;
50:
51: lazy_static!{
52: pub static ref RE_SPECIAL: Regex = Regex::new(r"([\-_*\[\]()~`>#+|{}\.!])").unwrap();
53: }
54:
55: // This one does nothing except making sure only one token exists for each id
56: pub struct Token {
57: running: Arc<Mutex<HashSet<i32>>>,
58: my_id: i32,
59: }
60:
61: impl Token {
62: /// Attempts to acquire a per-id token by inserting `my_id` into the shared `running` set.
63: ///
64: /// If the id was not already present, the function inserts it and returns `Some(Token)`.
65: /// When the returned `Token` is dropped, the id will be removed from the `running` set,
66: /// allowing subsequent acquisitions for the same id.
67: ///
68: /// # Parameters
69: ///
70: /// - `running`: Shared set tracking active ids.
71: /// - `my_id`: Identifier to acquire a token for.
72: ///
73: /// # Returns
74: ///
75: /// `Ok(Token)` if the id was successfully acquired, `Error` if a token for the id is already active.
76: async fn new (running: &Arc<Mutex<HashSet<i32>>>, my_id: i32) -> Result<Token> {
77: let running = running.clone();
78: let mut set = running.lock_arc().await;
79: if set.contains(&my_id) {
80: bail!("Token already taken");
81: } else {
82: set.insert(my_id);
83: Ok(Token {
84: running,
85: my_id,
86: })
87: }
88: }
89: }
90:
91: impl Drop for Token {
92: /// Releases this token's claim on the shared running-set when the token is dropped.
93: ///
94: /// The token's identifier is removed from the shared `running` set so that future
95: /// operations for the same id may proceed.
96: ///
97: /// TODO: is using block_on inside block_on safe? Currently tested and working fine.
98: fn drop (&mut self) {
99: smol::block_on(async {
100: let mut set = self.running.lock_arc().await;
101: set.remove(&self.my_id);
102: })
103: }
104: }
105:
106: pub type FeedList = BTreeMap<i32, String>;
107: type UserCache = TtlCache<i64, Arc<Mutex<FeedList>>>;
108:
109: #[derive(Clone)]
110: pub struct Core {
111: pub tg: Tg,
112: pub db: Db,
113: pub feeds: Arc<Mutex<UserCache>>,
114: running: Arc<Mutex<HashSet<i32>>>,
115: http_client: reqwest::Client,
116: }
117:
118: // XXX Right now that part is unfinished and I guess I need to finish menu first
119: #[allow(unused)]
120: pub struct Post {
121: uri: String,
122: _title: String,
123: _authors: String,
124: _summary: String,
125: }
126:
127: impl Core {
128: /// Create a Core instance from configuration and start its background autofetch loop.
129: ///
130: /// The provided `settings` must include:
131: /// - `owner` (integer): default chat id to use as the owner/destination,
132: /// - `api_key` (string): Telegram bot API key,
133: /// - `api_gateway` (string): Telegram API gateway host,
134: /// - `pg` (string): PostgreSQL connection string,
135: /// - optional `proxy` (string): proxy URL for the HTTP client.
136: ///
137: /// On success returns an initialized `Core` with Telegram and HTTP clients, database connection,
138: /// an empty running set for per-id tokens, and a spawned background task that periodically runs
139: /// `autofetch`. If any required setting is missing or initialization fails, an error is returned.
140: pub async fn new(settings: config::Config) -> Result<Core> {
141: let mut client = reqwest::Client::builder();
142: if let Ok(proxy) = settings.get_string("proxy") {
143: let proxy = reqwest::Proxy::all(proxy).stack()?;
144: client = client.proxy(proxy);
145: }
146:
147: let core = Core {
148: tg: Tg::new(&settings).await.stack()?,
149: db: Db::new(&settings.get_string("pg").stack()?)?,
150: feeds: Arc::new(Mutex::new(TtlCache::new(10000))),
151: running: Arc::new(Mutex::new(HashSet::new())),
152: http_client: client.build().stack()?,
153: };
154:
155: let clone = core.clone();
156: smol::spawn(Compat::new(async move {
157: loop {
158: let delay = match &clone.autofetch().await {
159: Err(err) => {
160: if let Err(err) = clone.tg.send(MyMessage::html(format!("🛑 {err}"))).await {
161: eprintln!("Autofetch error: {err:?}");
162: };
163: std::time::Duration::from_secs(60)
164: },
165: Ok(time) => *time,
166: };
167: Timer::after(delay).await;
168: }
169: })).detach();
170: Ok(core)
171: }
172:
173: /// Fetches the feed for a source, sends any newly discovered posts to the appropriate chat, and records them in the database.
174: ///
175: /// This acquires a per-source guard to prevent concurrent checks for the same `id`. If a check is already running for
176: /// the given `id`, the function returns an error. If `last_scrape` is provided, it is sent as the `If-Modified-Since`
177: /// header to the feed request. The function parses RSS or Atom feeds, sends unseen post URLs to either the source's
178: /// channel (when `real` is true) or the source owner (when `real` is false), and persists posted entries so they are
179: /// not reposted later.
180: ///
181: /// Parameters:
182: /// - `id`: Identifier of the source to check.
183: /// - `real`: When `true`, send posts to the source's channel; when `false`, send to the source owner.
184: /// - `last_scrape`: Optional timestamp used to set the `If-Modified-Since` header for the HTTP request.
185: ///
186: /// # Returns
187: ///
188: /// `Posted: N` where `N` is the number of posts processed and sent.
189: pub async fn check (&self, id: i32, real: bool, last_scrape: Option<DateTime<Local>>) -> Result<String> {
190: let mut posted: i32 = 0;
191: let mut conn = self.db.begin().await.stack()?;
192:
193: let _token = Token::new(&self.running, id).await.stack()?;
194: let source = conn.get_source(id, self.tg.owner).await.stack()?;
195: conn.set_scrape(id).await.stack()?;
196: let destination = ChatPeerId::from(match real {
197: true => source.channel_id,
198: false => source.owner,
199: });
200: let mut this_fetch: Option<DateTime<chrono::FixedOffset>> = None;
201: let mut posts: BTreeMap<DateTime<chrono::FixedOffset>, Post> = BTreeMap::new();
202:
203: let mut builder = self.http_client.get(&source.url);
204: if let Some(last_scrape) = last_scrape {
205: builder = builder.header(LAST_MODIFIED, last_scrape.to_rfc2822());
206: };
207: let response = builder.send().await.stack()?;
208: #[cfg(debug_assertions)]
209: {
210: use reqwest::header::{
211: CACHE_CONTROL,
212: EXPIRES,
213: };
214: let headers = response.headers();
215: let expires = headers.get(EXPIRES);
216: let cache = headers.get(CACHE_CONTROL);
217: if expires.is_some() || cache.is_some() {
d8c1d259a2 2026-04-23 218: println!("{} {} {:?} {:?} {:?}", Local::now().to_rfc2822(), &source.url, last_scrape, expires, cache);
219: }
220: }
221: let status = response.status();
222: let content = response.bytes().await.stack()?;
223: match rss::Channel::read_from(&content[..]) {
224: Ok(feed) => {
225: for item in feed.items() {
226: if let Some(link) = item.link() {
227: let date = match item.pub_date() {
228: Some(feed_date) => DateTime::parse_from_rfc2822(feed_date),
229: None => DateTime::parse_from_rfc3339(match item.dublin_core_ext() {
230: Some(ext) => {
231: let dates = ext.dates();
232: if dates.is_empty() {
233: bail!("Feed item has Dublin Core extension but no dates.")
234: } else {
235: &dates[0]
236: }
237: },
238: None => bail!("Feed item misses posting date."),
239: }),
240: }.stack()?;
241: posts.insert(date, Post{
242: uri: link.to_string(),
243: _title: item.title().unwrap_or("").to_string(),
244: _authors: item.author().unwrap_or("").to_string(),
245: _summary: item.content().unwrap_or("").to_string(),
246: });
247: }
248: };
249: },
250: Err(err) => match err {
251: rss::Error::InvalidStartTag => {
252: match atom_syndication::Feed::read_from(&content[..]) {
253: Ok(feed) => {
254: for item in feed.entries() {
255: let date = item.published()
256: .stack_err("Feed item missing publishing date.")?;
257: let uri = {
258: let links = item.links();
259: if links.is_empty() {
260: bail!("Feed item missing post links.");
261: } else {
262: links[0].href().to_string()
263: }
264: };
265: let _authors = item.authors().iter().map(|x| format!("{} <{:?}>", x.name(), x.email())).collect::<Vec<String>>().join(", ");
266: let _summary = if let Some(sum) = item.summary() { sum.value.clone() } else { String::new() };
267: posts.insert(*date, Post{
268: uri,
269: _title: item.title().to_string(),
270: _authors,
271: _summary,
272: });
273: };
274: },
275: Err(err) => {
276: bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
277: },
278: }
279: },
280: rss::Error::Eof => (),
281: _ => bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
282: }
283: };
284: for (date, post) in posts.iter() {
285: let post_url: Cow<str> = match source.url_re {
286: Some(ref x) => sedregex::ReplaceCommand::new(x).stack()?.execute(&post.uri),
287: None => post.uri.clone().into(),
288: };
289: if ! conn.exists(&post_url, id).await.stack()? {
290: if this_fetch.is_none() || *date > this_fetch.unwrap() {
291: this_fetch = Some(*date);
292: };
293: self.tg.send(MyMessage::html_to(match &source.iv_hash {
294: Some(hash) => format!("<a href=\"https://t.me/iv?url={post_url}&rhash={hash}\"> </a>{post_url}"),
295: None => format!("{post_url}"),
296: }, destination)).await.stack()?;
297: conn.add_post(id, date, &post_url).await.stack()?;
298: posted += 1;
299: };
300: };
301: posts.clear();
302: Ok(format!("Posted: {posted}"))
303: }
304:
305: /// Determine the delay until the next scheduled fetch and spawn background checks for any overdue sources.
306: ///
307: /// This scans the database queue, spawns background tasks to run checks for sources whose `next_fetch`
308: /// is in the past (each task uses a Core clone with the appropriate owner), and computes the shortest
309: /// duration until the next `next_fetch`.
310: async fn autofetch(&self) -> Result<std::time::Duration> {
311: let mut delay = chrono::Duration::minutes(1);
312: let now = chrono::Local::now();
313: let queue = {
314: let mut conn = self.db.begin().await.stack()?;
315: conn.get_queue().await.stack()?
316: };
317: for row in queue {
318: if let Some(next_fetch) = row.next_fetch {
319: if next_fetch < now {
320: if let (Some(owner), Some(source_id), last_scrape) = (row.owner, row.source_id, row.last_scrape) {
321: let clone = Core {
322: tg: self.tg.with_owner(owner),
323: ..self.clone()
324: };
325: let source = {
326: let mut conn = self.db.begin().await.stack()?;
327: match conn.get_one(owner, source_id).await {
328: Ok(Some(source)) => source.to_string(),
329: Ok(None) => "Source not found in database?".to_string(),
330: Err(err) => format!("Failed to fetch source data:\n{err}"),
331: }
332: };
333: smol::spawn(Compat::new(async move {
334: if let Err(err) = clone.check(source_id, true, Some(last_scrape)).await
335: && let Err(err) = clone.tg.send(MyMessage::html(format!("🛑 {source}\n<pre>{}</pre>", &err.to_string()))).await
336: {
337: eprintln!("Check error: {err}");
338: };
339: })).detach();
340: }
341: } else if next_fetch - now < delay {
342: delay = next_fetch - now;
343: }
344: }
345: };
346: delay.to_std().stack()
347: }
348:
349: /// Displays full list of managed channels for specified user
350: pub async fn list (&self, owner: UserPeerId) -> Result<String> {
351: let mut reply: Vec<String> = vec![];
352: reply.push("Channels:".into());
353: let mut conn = self.db.begin().await.stack()?;
354: for row in conn.get_list(owner).await.stack()? {
355: reply.push(row.to_string());
356: };
357: Ok(reply.join("\n\n"))
358: }
359:
360: /// Returns current cached list of feed for requested user, or loads data from database
361: pub async fn get_feeds (&self, owner: i64) -> Result<Arc<Mutex<FeedList>>> {
362: let mut feeds = self.feeds.lock_arc().await;
363: Ok(match feeds.get(&owner) {
364: None => {
365: let mut conn = self.db.begin().await.stack()?;
366: let feed_list = conn.get_feeds(owner).await.stack()?;
367: let mut map = BTreeMap::new();
368: for feed in feed_list {
369: map.insert(feed.source_id, feed.channel);
370: };
371: let res = Arc::new(Mutex::new(map));
372: feeds.insert(owner, res.clone(), Duration::from_secs(60 * 60 * 3));
373: res
374: },
375: Some(res) => res.clone(),
376: })
377: }
378:
379: /// Adds feed to cached list
380: pub async fn add_feed (&self, owner: i64, source_id: i32, channel: String) -> Result<()> {
381: let mut inserted = true;
382: {
383: let mut feeds = self.feeds.lock_arc().await;
384: if let Some(feed) = feeds.get_mut(&owner) {
385: let mut feed = feed.lock_arc().await;
386: feed.insert(source_id, channel);
387: } else {
388: inserted = false;
389: }
390: }
391: // in case insert failed - we miss the entry we needed to expand, reload everything from
392: // database
393: if !inserted {
394: self.get_feeds(owner).await.stack()?;
395: }
396: Ok(())
397: }
398:
399: /// Removes feed from cached list
400: pub async fn rm_feed (&self, owner: i64, source_id: &i32) -> Result<()> {
401: let mut dropped = false;
402: {
403: let mut feeds = self.feeds.lock_arc().await;
404: if let Some(feed) = feeds.get_mut(&owner) {
405: let mut feed = feed.lock_arc().await;
406: feed.remove(source_id);
407: dropped = true;
408: }
409: }
410: // in case we failed to found feed we need to remove - just reload everything from database
411: if !dropped {
412: self.get_feeds(owner).await.stack()?;
413: }
414: Ok(())
415: }
416:
417: pub async fn cb (&self, query: &CallbackQuery, cb: &str) -> Result<()> {
418: let cb: Callback = toml::from_str(cb).stack()?;
419: todo!();
420: Ok(())
421: }
422: }
423:
424: impl UpdateHandler for Core {
425: /// Dispatches an incoming Telegram update to a matching command handler and reports handler errors to the originating chat.
426: ///
427: /// This method inspects the update; if it contains a message that can be parsed as a bot command,
428: /// it executes the corresponding command handler. If the handler returns an error, the error text
429: /// is sent back to the message's chat. Unknown commands produce an error which is also reported to the chat.
430: async fn handle (&self, update: Update) -> () {
431: match update.update_type {
432: UpdateType::Message(msg) => {
433: if let Ok(cmd) = Command::try_from(*msg) {
434: let msg = cmd.get_message();
435: let words = cmd.get_args();
436: let command = cmd.get_name();
437: let res = match command {
438: "/check" | "/clean" | "/enable" | "/delete" | "/disable" => command::command(self, command, msg, words).await,
439: "/start" => command::start(self, msg).await,
440: "/list" => command::list(self, msg).await,
441: "/test" => command::test(self, msg).await,
442: "/add" | "/update" => command::update(self, command, msg, words).await,
443: any => Err(anyhow!("Unknown command: {any}")),
444: };
445: if let Err(err) = res {
446: match validate(&err.to_string()) {
447: Ok(text) => {
448: if let Err(err2) = self.tg.send(MyMessage::html_to(
449: format!("#error<pre>{}</pre>", text),
450: msg.chat.get_id(),
451: )).await {
452: dbg!(err2);
453: }
454: },
455: Err(err2) => {
456: dbg!(err2);
457: },
458: }
459: }
460: } else {
461: // not a command
462: }
463: },
464: UpdateType::CallbackQuery(query) => {
465: if let Some(ref cb) = query.data
466: && let Err(err) = self.cb(&query, cb).await
467: && let Err(err) = self.tg.answer_cb(query.id, err.to_string()).await
468: {
469: println!("{err:?}");
470: }
471: },
472: _ => {
473: println!("Unhandled UpdateKind:\n{update:?}")
474: },
475: }
476: }
477: }