Annotation For src/core.rs
Logged in as anonymous

Origin for each line in src/core.rs from check-in dc7c43b010:

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: 	},
dc7c43b010 2026-01-01   12: 	sync::Arc,
44575a91d3 2025-07-09   13: };
44575a91d3 2025-07-09   14: 
dc7c43b010 2026-01-01   15: use async_compat::Compat;
acb0a4ac54 2025-09-28   16: use chrono::{
acb0a4ac54 2025-09-28   17: 	DateTime,
acb0a4ac54 2025-09-28   18: 	Local,
44575a91d3 2025-07-09   19: };
5a4aab7687 2025-06-29   20: use lazy_static::lazy_static;
5a4aab7687 2025-06-29   21: use regex::Regex;
acb0a4ac54 2025-09-28   22: use reqwest::header::{
acb0a4ac54 2025-09-28   23: 	CACHE_CONTROL,
acb0a4ac54 2025-09-28   24: 	EXPIRES,
acb0a4ac54 2025-09-28   25: 	LAST_MODIFIED
dc7c43b010 2026-01-01   26: };
dc7c43b010 2026-01-01   27: use smol::{
dc7c43b010 2026-01-01   28: 	Timer,
dc7c43b010 2026-01-01   29: 	lock::Mutex,
acb0a4ac54 2025-09-28   30: };
bb89b6fab8 2025-06-15   31: use tgbot::{
bb89b6fab8 2025-06-15   32: 	api::Client,
bb89b6fab8 2025-06-15   33: 	handler::UpdateHandler,
bb89b6fab8 2025-06-15   34: 	types::{
bb89b6fab8 2025-06-15   35: 		Bot,
bb89b6fab8 2025-06-15   36: 		ChatPeerId,
bb89b6fab8 2025-06-15   37: 		Command,
bb89b6fab8 2025-06-15   38: 		GetBot,
bb89b6fab8 2025-06-15   39: 		Message,
bb89b6fab8 2025-06-15   40: 		ParseMode,
bb89b6fab8 2025-06-15   41: 		SendMessage,
bb89b6fab8 2025-06-15   42: 		Update,
bb89b6fab8 2025-06-15   43: 		UpdateType,
bb89b6fab8 2025-06-15   44: 		UserPeerId,
1bd041d00f 2025-05-20   45: 	},
44575a91d3 2025-07-09   46: };
44575a91d3 2025-07-09   47: use stacked_errors::{
44575a91d3 2025-07-09   48: 	Result,
44575a91d3 2025-07-09   49: 	StackableErr,
44575a91d3 2025-07-09   50: 	anyhow,
44575a91d3 2025-07-09   51: 	bail,
5a4aab7687 2025-06-29   52: };
5a4aab7687 2025-06-29   53: 
5a4aab7687 2025-06-29   54: lazy_static!{
5a4aab7687 2025-06-29   55: 	pub static ref RE_SPECIAL: Regex = Regex::new(r"([\-_*\[\]()~`>#+|{}\.!])").unwrap();
5a4aab7687 2025-06-29   56: }
5a4aab7687 2025-06-29   57: 
5a4aab7687 2025-06-29   58: /// Encodes special HTML entities to prevent them interfering with Telegram HTML
5a4aab7687 2025-06-29   59: pub fn encode (text: &str) -> Cow<'_, str> {
5a4aab7687 2025-06-29   60: 	RE_SPECIAL.replace_all(text, "\\$1")
5a4aab7687 2025-06-29   61: }
9171c791eb 2021-11-13   62: 
9171c791eb 2021-11-13   63: #[derive(Clone)]
9171c791eb 2021-11-13   64: pub struct Core {
bb89b6fab8 2025-06-15   65: 	owner_chat: ChatPeerId,
bb89b6fab8 2025-06-15   66: 	// max_delay: u16,
bb89b6fab8 2025-06-15   67: 	pub tg: Client,
bb89b6fab8 2025-06-15   68: 	pub me: Bot,
0340541002 2025-04-24   69: 	pub db: Db,
9171c791eb 2021-11-13   70: 	sources: Arc<Mutex<HashSet<Arc<i32>>>>,
45e34762e4 2023-05-28   71: 	http_client: reqwest::Client,
45e34762e4 2023-05-28   72: }
45e34762e4 2023-05-28   73: 
dc7c43b010 2026-01-01   74: pub struct Post {
dc7c43b010 2026-01-01   75: 	uri: String,
dc7c43b010 2026-01-01   76: 	title: String,
dc7c43b010 2026-01-01   77: 	authors: String,
dc7c43b010 2026-01-01   78: 	summary: String,
dc7c43b010 2026-01-01   79: }
dc7c43b010 2026-01-01   80: 
9171c791eb 2021-11-13   81: impl Core {
0340541002 2025-04-24   82: 	pub async fn new(settings: config::Config) -> Result<Core> {
44575a91d3 2025-07-09   83: 		let owner_chat = ChatPeerId::from(settings.get_int("owner").stack()?);
44575a91d3 2025-07-09   84: 		let api_key = settings.get_string("api_key").stack()?;
42b29b744b 2026-01-01   85: 		let tg = Client::new(&api_key).stack()?
42b29b744b 2026-01-01   86: 			.with_host(settings.get_string("api_gateway").stack()?);
45e34762e4 2023-05-28   87: 
45e34762e4 2023-05-28   88: 		let mut client = reqwest::Client::builder();
9d8a6738fd 2023-07-30   89: 		if let Ok(proxy) = settings.get_string("proxy") {
44575a91d3 2025-07-09   90: 			let proxy = reqwest::Proxy::all(proxy).stack()?;
45e34762e4 2023-05-28   91: 			client = client.proxy(proxy);
45e34762e4 2023-05-28   92: 		}
44575a91d3 2025-07-09   93: 		let http_client = client.build().stack()?;
44575a91d3 2025-07-09   94: 		let me = tg.execute(GetBot).await.stack()?;
0340541002 2025-04-24   95: 		let core = Core {
f988dfd28f 2022-02-13   96: 			tg,
e624ef9d66 2025-04-20   97: 			me,
e624ef9d66 2025-04-20   98: 			owner_chat,
44575a91d3 2025-07-09   99: 			db: Db::new(&settings.get_string("pg").stack()?)?,
9171c791eb 2021-11-13  100: 			sources: Arc::new(Mutex::new(HashSet::new())),
45e34762e4 2023-05-28  101: 			http_client,
bb89b6fab8 2025-06-15  102: 			// max_delay: 60,
0340541002 2025-04-24  103: 		};
bb89b6fab8 2025-06-15  104: 		let clone = core.clone();
dc7c43b010 2026-01-01  105: 		smol::spawn(Compat::new(async move {
26339860ce 2022-02-13  106: 			loop {
26339860ce 2022-02-13  107: 				let delay = match &clone.autofetch().await {
26339860ce 2022-02-13  108: 					Err(err) => {
07b34bcad6 2025-08-05  109: 						if let Err(err) = clone.send(format!("šŸ›‘ {err}"), None, None).await {
e624ef9d66 2025-04-20  110: 							eprintln!("Autofetch error: {err:?}");
26339860ce 2022-02-13  111: 						};
cb86e770f9 2022-03-15  112: 						std::time::Duration::from_secs(60)
26339860ce 2022-02-13  113: 					},
26339860ce 2022-02-13  114: 					Ok(time) => *time,
9171c791eb 2021-11-13  115: 				};
dc7c43b010 2026-01-01  116: 				Timer::after(delay).await;
9171c791eb 2021-11-13  117: 			}
dc7c43b010 2026-01-01  118: 		})).detach();
9171c791eb 2021-11-13  119: 		Ok(core)
9171c791eb 2021-11-13  120: 	}
9171c791eb 2021-11-13  121: 
bb89b6fab8 2025-06-15  122: 	pub async fn send <S>(&self, msg: S, target: Option<ChatPeerId>, mode: Option<ParseMode>) -> Result<Message>
e624ef9d66 2025-04-20  123: 	where S: Into<String> {
e624ef9d66 2025-04-20  124: 		let msg = msg.into();
e624ef9d66 2025-04-20  125: 
e624ef9d66 2025-04-20  126: 		let mode = mode.unwrap_or(ParseMode::Html);
e624ef9d66 2025-04-20  127: 		let target = target.unwrap_or(self.owner_chat);
44575a91d3 2025-07-09  128: 		self.tg.execute(
bb89b6fab8 2025-06-15  129: 			SendMessage::new(target, msg)
bb89b6fab8 2025-06-15  130: 				.with_parse_mode(mode)
44575a91d3 2025-07-09  131: 		).await.stack()
bb89b6fab8 2025-06-15  132: 	}
bb89b6fab8 2025-06-15  133: 
acb0a4ac54 2025-09-28  134: 	pub async fn check (&self, id: i32, real: bool, last_scrape: Option<DateTime<Local>>) -> Result<String> {
bb89b6fab8 2025-06-15  135: 		let mut posted: i32 = 0;
44575a91d3 2025-07-09  136: 		let mut conn = self.db.begin().await.stack()?;
bb89b6fab8 2025-06-15  137: 
bb89b6fab8 2025-06-15  138: 		let id = {
c6d3e97290 2025-07-01  139: 			let mut set = self.sources.lock_arc().await;
bb89b6fab8 2025-06-15  140: 			match set.get(&id) {
bb89b6fab8 2025-06-15  141: 				Some(id) => id.clone(),
bb89b6fab8 2025-06-15  142: 				None => {
bb89b6fab8 2025-06-15  143: 					let id = Arc::new(id);
9171c791eb 2021-11-13  144: 					set.insert(id.clone());
9171c791eb 2021-11-13  145: 					id.clone()
9171c791eb 2021-11-13  146: 				},
9171c791eb 2021-11-13  147: 			}
9171c791eb 2021-11-13  148: 		};
9171c791eb 2021-11-13  149: 		let count = Arc::strong_count(&id);
9171c791eb 2021-11-13  150: 		if count == 2 {
44575a91d3 2025-07-09  151: 			let source = conn.get_source(*id, self.owner_chat).await.stack()?;
44575a91d3 2025-07-09  152: 			conn.set_scrape(*id).await.stack()?;
bb89b6fab8 2025-06-15  153: 			let destination = ChatPeerId::from(match real {
e624ef9d66 2025-04-20  154: 				true => source.channel_id,
e624ef9d66 2025-04-20  155: 				false => source.owner,
bb89b6fab8 2025-06-15  156: 			});
9910c2209c 2023-08-04  157: 			let mut this_fetch: Option<DateTime<chrono::FixedOffset>> = None;
dc7c43b010 2026-01-01  158: 			let mut posts: BTreeMap<DateTime<chrono::FixedOffset>, Post> = BTreeMap::new();
9910c2209c 2023-08-04  159: 
acb0a4ac54 2025-09-28  160: 			let mut builder = self.http_client.get(&source.url);
acb0a4ac54 2025-09-28  161: 			if let Some(last_scrape) = last_scrape {
acb0a4ac54 2025-09-28  162: 				builder = builder.header(LAST_MODIFIED, last_scrape.to_rfc2822());
acb0a4ac54 2025-09-28  163: 			};
acb0a4ac54 2025-09-28  164: 			let response = builder.send().await.stack()?;
acb0a4ac54 2025-09-28  165: 			{
acb0a4ac54 2025-09-28  166: 				let headers = response.headers();
acb0a4ac54 2025-09-28  167: 				let expires = headers.get(EXPIRES);
acb0a4ac54 2025-09-28  168: 				let cache = headers.get(CACHE_CONTROL);
acb0a4ac54 2025-09-28  169: 				if expires.is_some() || cache.is_some() {
acb0a4ac54 2025-09-28  170: 					println!("{} {} {:?} {:?} {:?}", Local::now().to_rfc2822(), &source.url, last_scrape, expires, cache);
acb0a4ac54 2025-09-28  171: 				}
acb0a4ac54 2025-09-28  172: 			}
9910c2209c 2023-08-04  173: 			let status = response.status();
44575a91d3 2025-07-09  174: 			let content = response.bytes().await.stack()?;
9171c791eb 2021-11-13  175: 			match rss::Channel::read_from(&content[..]) {
9171c791eb 2021-11-13  176: 				Ok(feed) => {
9171c791eb 2021-11-13  177: 					for item in feed.items() {
f988dfd28f 2022-02-13  178: 						if let Some(link) = item.link() {
f988dfd28f 2022-02-13  179: 							let date = match item.pub_date() {
f988dfd28f 2022-02-13  180: 								Some(feed_date) => DateTime::parse_from_rfc2822(feed_date),
f988dfd28f 2022-02-13  181: 								None => DateTime::parse_from_rfc3339(&item.dublin_core_ext().unwrap().dates()[0]),
44575a91d3 2025-07-09  182: 							}.stack()?;
dc7c43b010 2026-01-01  183: 							let uri = link.to_string();
dc7c43b010 2026-01-01  184: 							let title = item.title().unwrap_or("").to_string();
dc7c43b010 2026-01-01  185: 							let authors = item.author().unwrap_or("").to_string();
dc7c43b010 2026-01-01  186: 							let summary = item.content().unwrap_or("").to_string();
dc7c43b010 2026-01-01  187: 							posts.insert(date, Post{
dc7c43b010 2026-01-01  188: 								uri,
dc7c43b010 2026-01-01  189: 								title,
dc7c43b010 2026-01-01  190: 								authors,
dc7c43b010 2026-01-01  191: 								summary,
dc7c43b010 2026-01-01  192: 							});
9171c791eb 2021-11-13  193: 						}
9171c791eb 2021-11-13  194: 					};
9171c791eb 2021-11-13  195: 				},
9171c791eb 2021-11-13  196: 				Err(err) => match err {
9171c791eb 2021-11-13  197: 					rss::Error::InvalidStartTag => {
2c36f015d8 2024-11-23  198: 						match atom_syndication::Feed::read_from(&content[..]) {
2c36f015d8 2024-11-23  199: 							Ok(feed) => {
2c36f015d8 2024-11-23  200: 								for item in feed.entries() {
2c36f015d8 2024-11-23  201: 									let date = item.published().unwrap();
dc7c43b010 2026-01-01  202: 									let uri = item.links()[0].href().to_string();
dc7c43b010 2026-01-01  203: 									let title = item.title().to_string();
dc7c43b010 2026-01-01  204: 									let authors = item.authors().iter().map(|x| format!("{} <{:?}>", x.name(), x.email())).collect::<Vec<String>>().join(", ");
dc7c43b010 2026-01-01  205: 									let summary = if let Some(sum) = item.summary() { sum.value.clone() } else { String::new() };
dc7c43b010 2026-01-01  206: 									posts.insert(*date, Post{
dc7c43b010 2026-01-01  207: 										uri,
dc7c43b010 2026-01-01  208: 										title,
dc7c43b010 2026-01-01  209: 										authors,
dc7c43b010 2026-01-01  210: 										summary,
dc7c43b010 2026-01-01  211: 									});
2c36f015d8 2024-11-23  212: 								};
2c36f015d8 2024-11-23  213: 							},
2c36f015d8 2024-11-23  214: 							Err(err) => {
07b34bcad6 2025-08-05  215: 								bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
2c36f015d8 2024-11-23  216: 							},
2c36f015d8 2024-11-23  217: 						}
9910c2209c 2023-08-04  218: 					},
9910c2209c 2023-08-04  219: 					rss::Error::Eof => (),
07b34bcad6 2025-08-05  220: 					_ => bail!("Unsupported or mangled content:\n{:?}\n{err}\n{status:#?}\n", &source.url)
9910c2209c 2023-08-04  221: 				}
9910c2209c 2023-08-04  222: 			};
dc7c43b010 2026-01-01  223: 			for (date, post) in posts.iter() {
9910c2209c 2023-08-04  224: 				let post_url: Cow<str> = match source.url_re {
dc7c43b010 2026-01-01  225: 					Some(ref x) => sedregex::ReplaceCommand::new(x).stack()?.execute(&post.uri),
dc7c43b010 2026-01-01  226: 					None => post.uri.clone().into(),
9910c2209c 2023-08-04  227: 				};
44575a91d3 2025-07-09  228: 				if let Some(exists) = conn.exists(&post_url, *id).await.stack()? {
9910c2209c 2023-08-04  229: 					if ! exists {
9910c2209c 2023-08-04  230: 						if this_fetch.is_none() || *date > this_fetch.unwrap() {
9910c2209c 2023-08-04  231: 							this_fetch = Some(*date);
9910c2209c 2023-08-04  232: 						};
e624ef9d66 2025-04-20  233: 						self.send( match &source.iv_hash {
e624ef9d66 2025-04-20  234: 							Some(hash) => format!("<a href=\"https://t.me/iv?url={post_url}&rhash={hash}\"> </a>{post_url}"),
e624ef9d66 2025-04-20  235: 							None => format!("{post_url}"),
44575a91d3 2025-07-09  236: 						}, Some(destination), Some(ParseMode::Html)).await.stack()?;
44575a91d3 2025-07-09  237: 						conn.add_post(*id, date, &post_url).await.stack()?;
9910c2209c 2023-08-04  238: 					};
9171c791eb 2021-11-13  239: 				};
9171c791eb 2021-11-13  240: 				posted += 1;
9171c791eb 2021-11-13  241: 			};
9171c791eb 2021-11-13  242: 			posts.clear();
9171c791eb 2021-11-13  243: 		};
0340541002 2025-04-24  244: 		Ok(format!("Posted: {posted}"))
26339860ce 2022-02-13  245: 	}
26339860ce 2022-02-13  246: 
bb89b6fab8 2025-06-15  247: 	async fn autofetch(&self) -> Result<std::time::Duration> {
26339860ce 2022-02-13  248: 		let mut delay = chrono::Duration::minutes(1);
9910c2209c 2023-08-04  249: 		let now = chrono::Local::now();
bb89b6fab8 2025-06-15  250: 		let queue = {
44575a91d3 2025-07-09  251: 			let mut conn = self.db.begin().await.stack()?;
44575a91d3 2025-07-09  252: 			conn.get_queue().await.stack()?
bb89b6fab8 2025-06-15  253: 		};
bb89b6fab8 2025-06-15  254: 		for row in queue {
9910c2209c 2023-08-04  255: 			if let Some(next_fetch) = row.next_fetch {
9910c2209c 2023-08-04  256: 				if next_fetch < now {
acb0a4ac54 2025-09-28  257: 					if let (Some(owner), Some(source_id), last_scrape) = (row.owner, row.source_id, row.last_scrape) {
bb89b6fab8 2025-06-15  258: 						let clone = Core {
bb89b6fab8 2025-06-15  259: 							owner_chat: ChatPeerId::from(owner),
9910c2209c 2023-08-04  260: 							..self.clone()
9910c2209c 2023-08-04  261: 						};
b4af85e31b 2025-06-29  262: 						let source = {
44575a91d3 2025-07-09  263: 							let mut conn = self.db.begin().await.stack()?;
b4af85e31b 2025-06-29  264: 							match conn.get_one(owner, source_id).await {
b4af85e31b 2025-06-29  265: 								Ok(Some(source)) => source.to_string(),
44575a91d3 2025-07-09  266: 								Ok(None) => "Source not found in database.stack()?".to_string(),
b4af85e31b 2025-06-29  267: 								Err(err) => format!("Failed to fetch source data:\n{err}"),
b4af85e31b 2025-06-29  268: 							}
b4af85e31b 2025-06-29  269: 						};
dc7c43b010 2026-01-01  270: 						smol::spawn(Compat::new(async move {
acb0a4ac54 2025-09-28  271: 							if let Err(err) = clone.check(source_id, true, Some(last_scrape)).await {
c6d3e97290 2025-07-01  272: 								if let Err(err) = clone.send(&format!("{source}\n\nšŸ›‘ {}", encode(&err.to_string())), None, Some(ParseMode::MarkdownV2)).await {
ec71df21e1 2025-11-07  273: 									eprintln!("Check error: {err}");
79c91a5357 2024-08-02  274: 									// clone.disable(&source_id, owner).await.unwrap();
9910c2209c 2023-08-04  275: 								};
9910c2209c 2023-08-04  276: 							};
dc7c43b010 2026-01-01  277: 						})).detach();
9910c2209c 2023-08-04  278: 					}
9910c2209c 2023-08-04  279: 				} else if next_fetch - now < delay {
9910c2209c 2023-08-04  280: 					delay = next_fetch - now;
9910c2209c 2023-08-04  281: 				}
9910c2209c 2023-08-04  282: 			}
9910c2209c 2023-08-04  283: 		};
44575a91d3 2025-07-09  284: 		delay.to_std().stack()
9910c2209c 2023-08-04  285: 	}
9910c2209c 2023-08-04  286: 
bb89b6fab8 2025-06-15  287: 	pub async fn list (&self, owner: UserPeerId) -> Result<String> {
b4af85e31b 2025-06-29  288: 		let mut reply: Vec<String> = vec![];
9910c2209c 2023-08-04  289: 		reply.push("Channels:".into());
44575a91d3 2025-07-09  290: 		let mut conn = self.db.begin().await.stack()?;
44575a91d3 2025-07-09  291: 		for row in conn.get_list(owner).await.stack()? {
b4af85e31b 2025-06-29  292: 			reply.push(row.to_string());
e624ef9d66 2025-04-20  293: 		};
b4af85e31b 2025-06-29  294: 		Ok(reply.join("\n\n"))
bb89b6fab8 2025-06-15  295: 	}
bb89b6fab8 2025-06-15  296: }
bb89b6fab8 2025-06-15  297: 
bb89b6fab8 2025-06-15  298: impl UpdateHandler for Core {
bb89b6fab8 2025-06-15  299: 	async fn handle (&self, update: Update) {
bb89b6fab8 2025-06-15  300: 		if let UpdateType::Message(msg) = update.update_type {
bb89b6fab8 2025-06-15  301: 			if let Ok(cmd) = Command::try_from(msg) {
bb89b6fab8 2025-06-15  302: 				let msg = cmd.get_message();
bb89b6fab8 2025-06-15  303: 				let words = cmd.get_args();
fae13a0e55 2025-06-28  304: 				let command = cmd.get_name();
fae13a0e55 2025-06-28  305: 				let res = match command {
fae13a0e55 2025-06-28  306: 					"/check" | "/clean" | "/enable" | "/delete" | "/disable" => command::command(self, command, msg, words).await,
bb89b6fab8 2025-06-15  307: 					"/start" => command::start(self, msg).await,
bb89b6fab8 2025-06-15  308: 					"/list" => command::list(self, msg).await,
fae13a0e55 2025-06-28  309: 					"/add" | "/update" => command::update(self, command, msg, words).await,
bb89b6fab8 2025-06-15  310: 					any => Err(anyhow!("Unknown command: {any}")),
bb89b6fab8 2025-06-15  311: 				};
bb89b6fab8 2025-06-15  312: 				if let Err(err) = res {
07b34bcad6 2025-08-05  313: 					if let Err(err2) = self.send(format!("\\#error\n```\n{err}\n```"),
bb89b6fab8 2025-06-15  314: 						Some(msg.chat.get_id()),
bb89b6fab8 2025-06-15  315: 						Some(ParseMode::MarkdownV2)
bb89b6fab8 2025-06-15  316: 					).await{
bb89b6fab8 2025-06-15  317: 						dbg!(err2);
bb89b6fab8 2025-06-15  318: 					};
bb89b6fab8 2025-06-15  319: 				}
bb89b6fab8 2025-06-15  320: 			};
bb89b6fab8 2025-06-15  321: 		};
e624ef9d66 2025-04-20  322: 	}
9171c791eb 2021-11-13  323: }