Annotation For src/core.rs
Logged in as anonymous

Origin for each line in src/core.rs from check-in 0340541002:

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: 	},
1c444d34ff 2024-08-28   12: 	num::TryFromIntError,
1c444d34ff 2024-08-28   13: 	sync::{
1c444d34ff 2024-08-28   14: 		Arc,
1c444d34ff 2024-08-28   15: 		Mutex
1c444d34ff 2024-08-28   16: 	},
1c444d34ff 2024-08-28   17: };
1c444d34ff 2024-08-28   18: 
e624ef9d66 2025-04-20   19: use anyhow::{
e624ef9d66 2025-04-20   20: 	bail,
e624ef9d66 2025-04-20   21: 	Result,
e624ef9d66 2025-04-20   22: };
e624ef9d66 2025-04-20   23: use async_std::task;
e624ef9d66 2025-04-20   24: use chrono::DateTime;
e624ef9d66 2025-04-20   25: use frankenstein::{
e624ef9d66 2025-04-20   26: 	client_reqwest::Bot,
e624ef9d66 2025-04-20   27: 	methods::{
e624ef9d66 2025-04-20   28: 		GetUpdatesParams,
e624ef9d66 2025-04-20   29: 		SendMessageParams
e624ef9d66 2025-04-20   30: 	},
e624ef9d66 2025-04-20   31: 	types::{
e624ef9d66 2025-04-20   32: 		AllowedUpdate,
e624ef9d66 2025-04-20   33: 		MessageEntityType,
e624ef9d66 2025-04-20   34: 		User,
e624ef9d66 2025-04-20   35: 	},
e624ef9d66 2025-04-20   36: 	updates::UpdateContent,
e624ef9d66 2025-04-20   37: 	AsyncTelegramApi,
e624ef9d66 2025-04-20   38: 	ParseMode,
e624ef9d66 2025-04-20   39: };
e624ef9d66 2025-04-20   40: use thiserror::Error;
e624ef9d66 2025-04-20   41: 
1c444d34ff 2024-08-28   42: #[derive(Error, Debug)]
1c444d34ff 2024-08-28   43: pub enum RssError {
a2880e5100 2025-04-19   44: 	// #[error(transparent)]
a2880e5100 2025-04-19   45: 	// Tg(#[from] TgError),
1c444d34ff 2024-08-28   46: 	#[error(transparent)]
1c444d34ff 2024-08-28   47: 	Int(#[from] TryFromIntError),
79c91a5357 2024-08-02   48: }
9171c791eb 2021-11-13   49: 
9171c791eb 2021-11-13   50: #[derive(Clone)]
9171c791eb 2021-11-13   51: pub struct Core {
e624ef9d66 2025-04-20   52: 	owner_chat: i64,
a2880e5100 2025-04-19   53: 	pub tg: Bot,
e624ef9d66 2025-04-20   54: 	pub me: User,
0340541002 2025-04-24   55: 	pub db: Db,
9171c791eb 2021-11-13   56: 	sources: Arc<Mutex<HashSet<Arc<i32>>>>,
45e34762e4 2023-05-28   57: 	http_client: reqwest::Client,
9171c791eb 2021-11-13   58: }
9171c791eb 2021-11-13   59: 
9171c791eb 2021-11-13   60: impl Core {
0340541002 2025-04-24   61: 	pub async fn new(settings: config::Config) -> Result<Core> {
e624ef9d66 2025-04-20   62: 		let owner_chat = settings.get_int("owner")?;
28da2e2a00 2022-03-12   63: 		let api_key = settings.get_string("api_key")?;
e624ef9d66 2025-04-20   64: 		let tg = Bot::new(&api_key);
45e34762e4 2023-05-28   65: 
45e34762e4 2023-05-28   66: 		let mut client = reqwest::Client::builder();
9d8a6738fd 2023-07-30   67: 		if let Ok(proxy) = settings.get_string("proxy") {
45e34762e4 2023-05-28   68: 			let proxy = reqwest::Proxy::all(proxy)?;
45e34762e4 2023-05-28   69: 			client = client.proxy(proxy);
45e34762e4 2023-05-28   70: 		}
45e34762e4 2023-05-28   71: 		let http_client = client.build()?;
e624ef9d66 2025-04-20   72: 		let me = tg.get_me().await?;
e624ef9d66 2025-04-20   73: 		let me = me.result;
0340541002 2025-04-24   74: 		let core = Core {
f988dfd28f 2022-02-13   75: 			tg,
e624ef9d66 2025-04-20   76: 			me,
e624ef9d66 2025-04-20   77: 			owner_chat,
0340541002 2025-04-24   78: 			db: Db::new(&settings.get_string("pg")?)?,
9171c791eb 2021-11-13   79: 			sources: Arc::new(Mutex::new(HashSet::new())),
45e34762e4 2023-05-28   80: 			http_client,
0340541002 2025-04-24   81: 		};
0340541002 2025-04-24   82: 		let mut clone = core.clone();
cb86e770f9 2022-03-15   83: 		task::spawn(async move {
26339860ce 2022-02-13   84: 			loop {
26339860ce 2022-02-13   85: 				let delay = match &clone.autofetch().await {
26339860ce 2022-02-13   86: 					Err(err) => {
e624ef9d66 2025-04-20   87: 						if let Err(err) = clone.send(format!("šŸ›‘ {err:?}"), None, None).await {
e624ef9d66 2025-04-20   88: 							eprintln!("Autofetch error: {err:?}");
26339860ce 2022-02-13   89: 						};
cb86e770f9 2022-03-15   90: 						std::time::Duration::from_secs(60)
26339860ce 2022-02-13   91: 					},
26339860ce 2022-02-13   92: 					Ok(time) => *time,
9171c791eb 2021-11-13   93: 				};
cb86e770f9 2022-03-15   94: 				task::sleep(delay).await;
9171c791eb 2021-11-13   95: 			}
e624ef9d66 2025-04-20   96: 		});
9171c791eb 2021-11-13   97: 		Ok(core)
9171c791eb 2021-11-13   98: 	}
9171c791eb 2021-11-13   99: 
0340541002 2025-04-24  100: 	pub async fn stream(&mut self) -> Result<()> {
e624ef9d66 2025-04-20  101: 		let mut offset: i64 = 0;
e624ef9d66 2025-04-20  102: 		let mut params = GetUpdatesParams {
e624ef9d66 2025-04-20  103: 			offset: None,
e624ef9d66 2025-04-20  104: 			limit: Some(100),
e624ef9d66 2025-04-20  105: 			timeout: Some(300),
e624ef9d66 2025-04-20  106: 			allowed_updates: Some(vec![AllowedUpdate::Message]),
e624ef9d66 2025-04-20  107: 		};
e624ef9d66 2025-04-20  108: 		loop {
e624ef9d66 2025-04-20  109: 			let updates = self.tg.get_updates(&params).await?.result;
e624ef9d66 2025-04-20  110: 			if updates.is_empty() {
e624ef9d66 2025-04-20  111: 				offset = 0;
e624ef9d66 2025-04-20  112: 				params.offset = None;
e624ef9d66 2025-04-20  113: 				continue;
e624ef9d66 2025-04-20  114: 			}
e624ef9d66 2025-04-20  115: 			for update in updates {
e624ef9d66 2025-04-20  116: 				if i64::from(update.update_id) >= offset {
e624ef9d66 2025-04-20  117: 					offset = i64::from(update.update_id) + 1;
e624ef9d66 2025-04-20  118: 					params.offset = Some(offset);
e624ef9d66 2025-04-20  119: 				}
e624ef9d66 2025-04-20  120: 				if let UpdateContent::Message(msg) = update.content {
e624ef9d66 2025-04-20  121: 					if let Some(text) = msg.text {
e624ef9d66 2025-04-20  122: 						if let Some(entities) = msg.entities {
e624ef9d66 2025-04-20  123: 							let chars: Vec<u16> = text.encode_utf16().collect();
e624ef9d66 2025-04-20  124: 							for entity in entities {
e624ef9d66 2025-04-20  125: 								if entity.type_field == MessageEntityType::BotCommand && entity.offset != 0 {
e624ef9d66 2025-04-20  126: 									bail!("commands should be at message start");
e624ef9d66 2025-04-20  127: 								};
e624ef9d66 2025-04-20  128: 								let cmd = String::from_utf16_lossy(&chars[entity.offset as usize..entity.length as usize]);
e624ef9d66 2025-04-20  129: 								let words: Vec<&str> = text.split_whitespace().collect();
e624ef9d66 2025-04-20  130: 								match cmd.as_ref() {
e624ef9d66 2025-04-20  131: 									"/check" | "/clean" | "/enable" | "/delete" | "/disable" => { command::command(self, msg.chat.id, words).await? },
e624ef9d66 2025-04-20  132: 									"/start" => { command::start(self, msg.chat.id).await?; },
e624ef9d66 2025-04-20  133: 									"/list" => { command::list(self, msg.chat.id).await?; },
e624ef9d66 2025-04-20  134: 									"/add" | "/update" => { command::update(self, msg.chat.id, words).await?; },
e624ef9d66 2025-04-20  135: 									any => {
e624ef9d66 2025-04-20  136: 										self.send(format!("\\#error\n```\nUnknown command: {any}\n```"),
e624ef9d66 2025-04-20  137: 											Some(msg.chat.id),
e624ef9d66 2025-04-20  138: 											Some(ParseMode::MarkdownV2)
e624ef9d66 2025-04-20  139: 										).await?;
e624ef9d66 2025-04-20  140: 									},
e624ef9d66 2025-04-20  141: 								};
e624ef9d66 2025-04-20  142: 							};
e624ef9d66 2025-04-20  143: 						};
e624ef9d66 2025-04-20  144: 					};
e624ef9d66 2025-04-20  145: 				};
e624ef9d66 2025-04-20  146: 			}
e624ef9d66 2025-04-20  147: 		}
e624ef9d66 2025-04-20  148: 	}
e624ef9d66 2025-04-20  149: 
e624ef9d66 2025-04-20  150: 	pub async fn send <S>(&self, msg: S, target: Option<i64>, mode: Option<ParseMode>) -> Result<()>
e624ef9d66 2025-04-20  151: 	where S: Into<String> {
e624ef9d66 2025-04-20  152: 		let msg = msg.into();
e624ef9d66 2025-04-20  153: 
e624ef9d66 2025-04-20  154: 		let mode = mode.unwrap_or(ParseMode::Html);
e624ef9d66 2025-04-20  155: 		let target = target.unwrap_or(self.owner_chat);
e624ef9d66 2025-04-20  156: 		let send = SendMessageParams::builder()
e624ef9d66 2025-04-20  157: 			.chat_id(target)
e624ef9d66 2025-04-20  158: 			.text(msg)
e624ef9d66 2025-04-20  159: 			.parse_mode(mode)
e624ef9d66 2025-04-20  160: 			.build();
e624ef9d66 2025-04-20  161: 		self.tg.send_message(&send).await?;
e624ef9d66 2025-04-20  162: 		Ok(())
e624ef9d66 2025-04-20  163: 	}
e624ef9d66 2025-04-20  164: 
0340541002 2025-04-24  165: 	pub async fn check (&mut self, id: &i32, owner: i64, real: bool) -> Result<String> {
e624ef9d66 2025-04-20  166: 		let mut posted: i32 = 0;
0340541002 2025-04-24  167: 		let mut conn = self.db.begin().await?;
285ce2bc31 2024-09-17  168: 
9171c791eb 2021-11-13  169: 		let id = {
9171c791eb 2021-11-13  170: 			let mut set = self.sources.lock().unwrap();
9171c791eb 2021-11-13  171: 			match set.get(id) {
9171c791eb 2021-11-13  172: 				Some(id) => id.clone(),
9171c791eb 2021-11-13  173: 				None => {
9171c791eb 2021-11-13  174: 					let id = Arc::new(*id);
9171c791eb 2021-11-13  175: 					set.insert(id.clone());
9171c791eb 2021-11-13  176: 					id.clone()
9171c791eb 2021-11-13  177: 				},
9171c791eb 2021-11-13  178: 			}
9171c791eb 2021-11-13  179: 		};
9171c791eb 2021-11-13  180: 		let count = Arc::strong_count(&id);
9171c791eb 2021-11-13  181: 		if count == 2 {
0340541002 2025-04-24  182: 			let source = conn.get_source(*id, owner).await?;
9171c791eb 2021-11-13  183: 			let destination = match real {
e624ef9d66 2025-04-20  184: 				true => source.channel_id,
e624ef9d66 2025-04-20  185: 				false => source.owner,
9171c791eb 2021-11-13  186: 			};
9171c791eb 2021-11-13  187: 			let mut this_fetch: Option<DateTime<chrono::FixedOffset>> = None;
9171c791eb 2021-11-13  188: 			let mut posts: BTreeMap<DateTime<chrono::FixedOffset>, String> = BTreeMap::new();
45e34762e4 2023-05-28  189: 
9910c2209c 2023-08-04  190: 			let response = self.http_client.get(&source.url).send().await?;
613a665847 2021-11-15  191: 			let status = response.status();
613a665847 2021-11-15  192: 			let content = response.bytes().await?;
9171c791eb 2021-11-13  193: 			match rss::Channel::read_from(&content[..]) {
9171c791eb 2021-11-13  194: 				Ok(feed) => {
9171c791eb 2021-11-13  195: 					for item in feed.items() {
f988dfd28f 2022-02-13  196: 						if let Some(link) = item.link() {
f988dfd28f 2022-02-13  197: 							let date = match item.pub_date() {
f988dfd28f 2022-02-13  198: 								Some(feed_date) => DateTime::parse_from_rfc2822(feed_date),
f988dfd28f 2022-02-13  199: 								None => DateTime::parse_from_rfc3339(&item.dublin_core_ext().unwrap().dates()[0]),
f988dfd28f 2022-02-13  200: 							}?;
f988dfd28f 2022-02-13  201: 							let url = link;
26339860ce 2022-02-13  202: 							posts.insert(date, url.to_string());
9171c791eb 2021-11-13  203: 						}
9171c791eb 2021-11-13  204: 					};
9171c791eb 2021-11-13  205: 				},
9171c791eb 2021-11-13  206: 				Err(err) => match err {
9171c791eb 2021-11-13  207: 					rss::Error::InvalidStartTag => {
2c36f015d8 2024-11-23  208: 						match atom_syndication::Feed::read_from(&content[..]) {
2c36f015d8 2024-11-23  209: 							Ok(feed) => {
2c36f015d8 2024-11-23  210: 								for item in feed.entries() {
2c36f015d8 2024-11-23  211: 									let date = item.published().unwrap();
2c36f015d8 2024-11-23  212: 									let url = item.links()[0].href();
2c36f015d8 2024-11-23  213: 									posts.insert(*date, url.to_string());
2c36f015d8 2024-11-23  214: 								};
2c36f015d8 2024-11-23  215: 							},
2c36f015d8 2024-11-23  216: 							Err(err) => {
e624ef9d66 2025-04-20  217: 								bail!("Unsupported or mangled content:\n{:?}\n{err:#?}\n{status:#?}\n", &source.url)
2c36f015d8 2024-11-23  218: 							},
2c36f015d8 2024-11-23  219: 						}
9171c791eb 2021-11-13  220: 					},
9171c791eb 2021-11-13  221: 					rss::Error::Eof => (),
e624ef9d66 2025-04-20  222: 					_ => bail!("Unsupported or mangled content:\n{:?}\n{err:#?}\n{status:#?}\n", &source.url)
9171c791eb 2021-11-13  223: 				}
9171c791eb 2021-11-13  224: 			};
9171c791eb 2021-11-13  225: 			for (date, url) in posts.iter() {
9910c2209c 2023-08-04  226: 				let post_url: Cow<str> = match source.url_re {
3dc9cddd4d 2023-08-04  227: 					Some(ref x) => sedregex::ReplaceCommand::new(x)?.execute(url),
659724c658 2021-12-08  228: 					None => url.into(),
659724c658 2021-12-08  229: 				};
0340541002 2025-04-24  230: 				if let Some(exists) = conn.exists(&post_url, *id).await? {
9910c2209c 2023-08-04  231: 					if ! exists {
9910c2209c 2023-08-04  232: 						if this_fetch.is_none() || *date > this_fetch.unwrap() {
9910c2209c 2023-08-04  233: 							this_fetch = Some(*date);
9910c2209c 2023-08-04  234: 						};
e624ef9d66 2025-04-20  235: 						self.send( match &source.iv_hash {
e624ef9d66 2025-04-20  236: 							Some(hash) => format!("<a href=\"https://t.me/iv?url={post_url}&rhash={hash}\"> </a>{post_url}"),
e624ef9d66 2025-04-20  237: 							None => format!("{post_url}"),
e624ef9d66 2025-04-20  238: 						}, Some(destination), Some(ParseMode::Html)).await?;
0340541002 2025-04-24  239: 						conn.add_post(*id, date, &post_url).await?;
659724c658 2021-12-08  240: 					};
9171c791eb 2021-11-13  241: 				};
9171c791eb 2021-11-13  242: 				posted += 1;
9171c791eb 2021-11-13  243: 			};
9171c791eb 2021-11-13  244: 			posts.clear();
9171c791eb 2021-11-13  245: 		};
0340541002 2025-04-24  246: 		conn.set_scrape(*id).await?;
0340541002 2025-04-24  247: 		Ok(format!("Posted: {posted}"))
0340541002 2025-04-24  248: 	}
0340541002 2025-04-24  249: 
0340541002 2025-04-24  250: 	async fn autofetch(&mut self) -> Result<std::time::Duration> {
0340541002 2025-04-24  251: 		let mut delay = chrono::Duration::minutes(1);
0340541002 2025-04-24  252: 		let now = chrono::Local::now();
0340541002 2025-04-24  253: 		let mut conn = self.db.begin().await?;
0340541002 2025-04-24  254: 		for row in conn.get_queue().await? {
9910c2209c 2023-08-04  255: 			if let Some(next_fetch) = row.next_fetch {
9910c2209c 2023-08-04  256: 				if next_fetch < now {
9910c2209c 2023-08-04  257: 					if let (Some(owner), Some(source_id)) = (row.owner, row.source_id) {
0340541002 2025-04-24  258: 						let mut clone = Core {
e624ef9d66 2025-04-20  259: 							owner_chat: owner,
9910c2209c 2023-08-04  260: 							..self.clone()
9910c2209c 2023-08-04  261: 						};
9910c2209c 2023-08-04  262: 						task::spawn(async move {
9910c2209c 2023-08-04  263: 							if let Err(err) = clone.check(&source_id, owner, true).await {
e624ef9d66 2025-04-20  264: 								if let Err(err) = clone.send(&format!("šŸ›‘ {err:?}"), None, None).await {
e624ef9d66 2025-04-20  265: 									eprintln!("Check error: {err:?}");
79c91a5357 2024-08-02  266: 									// clone.disable(&source_id, owner).await.unwrap();
9910c2209c 2023-08-04  267: 								};
9910c2209c 2023-08-04  268: 							};
9910c2209c 2023-08-04  269: 						});
9910c2209c 2023-08-04  270: 					}
9910c2209c 2023-08-04  271: 				} else if next_fetch - now < delay {
9910c2209c 2023-08-04  272: 					delay = next_fetch - now;
9910c2209c 2023-08-04  273: 				}
9910c2209c 2023-08-04  274: 			}
9910c2209c 2023-08-04  275: 		};
26339860ce 2022-02-13  276: 		Ok(delay.to_std()?)
26339860ce 2022-02-13  277: 	}
26339860ce 2022-02-13  278: 
0340541002 2025-04-24  279: 	pub async fn list (&mut self, owner: i64) -> Result<String> {
26339860ce 2022-02-13  280: 		let mut reply: Vec<Cow<str>> = vec![];
26339860ce 2022-02-13  281: 		reply.push("Channels:".into());
0340541002 2025-04-24  282: 		let mut conn = self.db.begin().await?;
0340541002 2025-04-24  283: 		for row in conn.get_list(owner).await? {
79c91a5357 2024-08-02  284: 			reply.push(format!("\n\\#ļøāƒ£ {} \\*ļøāƒ£ `{}` {}\nšŸ”— `{}`", row.source_id, row.channel,
9910c2209c 2023-08-04  285: 				match row.enabled {
26339860ce 2022-02-13  286: 					true  => "šŸ”„ enabled",
26339860ce 2022-02-13  287: 					false => "ā›” disabled",
9910c2209c 2023-08-04  288: 				}, row.url).into());
9910c2209c 2023-08-04  289: 			if let Some(hash) = &row.iv_hash {
e624ef9d66 2025-04-20  290: 				reply.push(format!("IV: `{hash}`").into());
26339860ce 2022-02-13  291: 			}
9910c2209c 2023-08-04  292: 			if let Some(re) = &row.url_re {
e624ef9d66 2025-04-20  293: 				reply.push(format!("RE: `{re}`").into());
a7f91033c0 2021-11-13  294: 			}
a7f91033c0 2021-11-13  295: 		};
a7f91033c0 2021-11-13  296: 		Ok(reply.join("\n"))
e624ef9d66 2025-04-20  297: 	}
9171c791eb 2021-11-13  298: }