Annotation For src/core.rs
Logged in as anonymous

Origin for each line in src/core.rs from check-in c6d3e97290:

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: 	},
1bd041d00f 2025-05-20   12: };
1bd041d00f 2025-05-20   13: 
1bd041d00f 2025-05-20   14: use anyhow::{
1bd041d00f 2025-05-20   15: 	anyhow,
1bd041d00f 2025-05-20   16: 	bail,
1bd041d00f 2025-05-20   17: 	Result,
1bd041d00f 2025-05-20   18: };
c6d3e97290 2025-07-01   19: use async_std::{
c6d3e97290 2025-07-01   20: 	task,
c6d3e97290 2025-07-01   21: 	sync::{
c6d3e97290 2025-07-01   22: 		Arc,
c6d3e97290 2025-07-01   23: 		Mutex
c6d3e97290 2025-07-01   24: 	},
c6d3e97290 2025-07-01   25: };
1bd041d00f 2025-05-20   26: use chrono::DateTime;
5a4aab7687 2025-06-29   27: use lazy_static::lazy_static;
5a4aab7687 2025-06-29   28: use regex::Regex;
bb89b6fab8 2025-06-15   29: use tgbot::{
bb89b6fab8 2025-06-15   30: 	api::Client,
bb89b6fab8 2025-06-15   31: 	handler::UpdateHandler,
1bd041d00f 2025-05-20   32: 	types::{
bb89b6fab8 2025-06-15   33: 		Bot,
bb89b6fab8 2025-06-15   34: 		ChatPeerId,
bb89b6fab8 2025-06-15   35: 		Command,
bb89b6fab8 2025-06-15   36: 		GetBot,
bb89b6fab8 2025-06-15   37: 		Message,
bb89b6fab8 2025-06-15   38: 		ParseMode,
bb89b6fab8 2025-06-15   39: 		SendMessage,
bb89b6fab8 2025-06-15   40: 		Update,
bb89b6fab8 2025-06-15   41: 		UpdateType,
bb89b6fab8 2025-06-15   42: 		UserPeerId,
1bd041d00f 2025-05-20   43: 	},
1bd041d00f 2025-05-20   44: };
5a4aab7687 2025-06-29   45: 
5a4aab7687 2025-06-29   46: lazy_static!{
5a4aab7687 2025-06-29   47: 	pub static ref RE_SPECIAL: Regex = Regex::new(r"([\-_*\[\]()~`>#+|{}\.!])").unwrap();
5a4aab7687 2025-06-29   48: }
5a4aab7687 2025-06-29   49: 
5a4aab7687 2025-06-29   50: /// Encodes special HTML entities to prevent them interfering with Telegram HTML
5a4aab7687 2025-06-29   51: pub fn encode (text: &str) -> Cow<'_, str> {
5a4aab7687 2025-06-29   52: 	RE_SPECIAL.replace_all(text, "\\$1")
5a4aab7687 2025-06-29   53: }
9171c791eb 2021-11-13   54: 
9171c791eb 2021-11-13   55: #[derive(Clone)]
9171c791eb 2021-11-13   56: pub struct Core {
bb89b6fab8 2025-06-15   57: 	owner_chat: ChatPeerId,
bb89b6fab8 2025-06-15   58: 	// max_delay: u16,
bb89b6fab8 2025-06-15   59: 	pub tg: Client,
bb89b6fab8 2025-06-15   60: 	pub me: Bot,
0340541002 2025-04-24   61: 	pub db: Db,
9171c791eb 2021-11-13   62: 	sources: Arc<Mutex<HashSet<Arc<i32>>>>,
45e34762e4 2023-05-28   63: 	http_client: reqwest::Client,
9171c791eb 2021-11-13   64: }
9171c791eb 2021-11-13   65: 
9171c791eb 2021-11-13   66: impl Core {
0340541002 2025-04-24   67: 	pub async fn new(settings: config::Config) -> Result<Core> {
bb89b6fab8 2025-06-15   68: 		let owner_chat = ChatPeerId::from(settings.get_int("owner")?);
28da2e2a00 2022-03-12   69: 		let api_key = settings.get_string("api_key")?;
bb89b6fab8 2025-06-15   70: 		let tg = Client::new(&api_key)?;
45e34762e4 2023-05-28   71: 
45e34762e4 2023-05-28   72: 		let mut client = reqwest::Client::builder();
9d8a6738fd 2023-07-30   73: 		if let Ok(proxy) = settings.get_string("proxy") {
45e34762e4 2023-05-28   74: 			let proxy = reqwest::Proxy::all(proxy)?;
45e34762e4 2023-05-28   75: 			client = client.proxy(proxy);
45e34762e4 2023-05-28   76: 		}
45e34762e4 2023-05-28   77: 		let http_client = client.build()?;
bb89b6fab8 2025-06-15   78: 		let me = tg.execute(GetBot).await?;
0340541002 2025-04-24   79: 		let core = Core {
f988dfd28f 2022-02-13   80: 			tg,
e624ef9d66 2025-04-20   81: 			me,
e624ef9d66 2025-04-20   82: 			owner_chat,
0340541002 2025-04-24   83: 			db: Db::new(&settings.get_string("pg")?)?,
9171c791eb 2021-11-13   84: 			sources: Arc::new(Mutex::new(HashSet::new())),
45e34762e4 2023-05-28   85: 			http_client,
bb89b6fab8 2025-06-15   86: 			// max_delay: 60,
0340541002 2025-04-24   87: 		};
bb89b6fab8 2025-06-15   88: 		let clone = core.clone();
cb86e770f9 2022-03-15   89: 		task::spawn(async move {
26339860ce 2022-02-13   90: 			loop {
26339860ce 2022-02-13   91: 				let delay = match &clone.autofetch().await {
26339860ce 2022-02-13   92: 					Err(err) => {
e624ef9d66 2025-04-20   93: 						if let Err(err) = clone.send(format!("šŸ›‘ {err:?}"), None, None).await {
e624ef9d66 2025-04-20   94: 							eprintln!("Autofetch error: {err:?}");
26339860ce 2022-02-13   95: 						};
cb86e770f9 2022-03-15   96: 						std::time::Duration::from_secs(60)
26339860ce 2022-02-13   97: 					},
26339860ce 2022-02-13   98: 					Ok(time) => *time,
9171c791eb 2021-11-13   99: 				};
cb86e770f9 2022-03-15  100: 				task::sleep(delay).await;
9171c791eb 2021-11-13  101: 			}
e624ef9d66 2025-04-20  102: 		});
9171c791eb 2021-11-13  103: 		Ok(core)
9171c791eb 2021-11-13  104: 	}
9171c791eb 2021-11-13  105: 
bb89b6fab8 2025-06-15  106: 	pub async fn send <S>(&self, msg: S, target: Option<ChatPeerId>, mode: Option<ParseMode>) -> Result<Message>
e624ef9d66 2025-04-20  107: 	where S: Into<String> {
e624ef9d66 2025-04-20  108: 		let msg = msg.into();
e624ef9d66 2025-04-20  109: 
e624ef9d66 2025-04-20  110: 		let mode = mode.unwrap_or(ParseMode::Html);
e624ef9d66 2025-04-20  111: 		let target = target.unwrap_or(self.owner_chat);
bb89b6fab8 2025-06-15  112: 		Ok(self.tg.execute(
bb89b6fab8 2025-06-15  113: 			SendMessage::new(target, msg)
bb89b6fab8 2025-06-15  114: 				.with_parse_mode(mode)
bb89b6fab8 2025-06-15  115: 		).await?)
bb89b6fab8 2025-06-15  116: 	}
bb89b6fab8 2025-06-15  117: 
bb89b6fab8 2025-06-15  118: 	pub async fn check (&self, id: i32, real: bool) -> Result<String> {
e624ef9d66 2025-04-20  119: 		let mut posted: i32 = 0;
0340541002 2025-04-24  120: 		let mut conn = self.db.begin().await?;
285ce2bc31 2024-09-17  121: 
285ce2bc31 2024-09-17  122: 		let id = {
c6d3e97290 2025-07-01  123: 			let mut set = self.sources.lock_arc().await;
bb89b6fab8 2025-06-15  124: 			match set.get(&id) {
9171c791eb 2021-11-13  125: 				Some(id) => id.clone(),
9171c791eb 2021-11-13  126: 				None => {
bb89b6fab8 2025-06-15  127: 					let id = Arc::new(id);
9171c791eb 2021-11-13  128: 					set.insert(id.clone());
9171c791eb 2021-11-13  129: 					id.clone()
9171c791eb 2021-11-13  130: 				},
9171c791eb 2021-11-13  131: 			}
9171c791eb 2021-11-13  132: 		};
9171c791eb 2021-11-13  133: 		let count = Arc::strong_count(&id);
9171c791eb 2021-11-13  134: 		if count == 2 {
bb89b6fab8 2025-06-15  135: 			let source = conn.get_source(*id, self.owner_chat).await?;
61899b5618 2025-04-26  136: 			conn.set_scrape(*id).await?;
bb89b6fab8 2025-06-15  137: 			let destination = ChatPeerId::from(match real {
e624ef9d66 2025-04-20  138: 				true => source.channel_id,
e624ef9d66 2025-04-20  139: 				false => source.owner,
bb89b6fab8 2025-06-15  140: 			});
9171c791eb 2021-11-13  141: 			let mut this_fetch: Option<DateTime<chrono::FixedOffset>> = None;
9171c791eb 2021-11-13  142: 			let mut posts: BTreeMap<DateTime<chrono::FixedOffset>, String> = BTreeMap::new();
45e34762e4 2023-05-28  143: 
9910c2209c 2023-08-04  144: 			let response = self.http_client.get(&source.url).send().await?;
613a665847 2021-11-15  145: 			let status = response.status();
613a665847 2021-11-15  146: 			let content = response.bytes().await?;
9171c791eb 2021-11-13  147: 			match rss::Channel::read_from(&content[..]) {
9171c791eb 2021-11-13  148: 				Ok(feed) => {
9171c791eb 2021-11-13  149: 					for item in feed.items() {
f988dfd28f 2022-02-13  150: 						if let Some(link) = item.link() {
f988dfd28f 2022-02-13  151: 							let date = match item.pub_date() {
f988dfd28f 2022-02-13  152: 								Some(feed_date) => DateTime::parse_from_rfc2822(feed_date),
f988dfd28f 2022-02-13  153: 								None => DateTime::parse_from_rfc3339(&item.dublin_core_ext().unwrap().dates()[0]),
f988dfd28f 2022-02-13  154: 							}?;
f988dfd28f 2022-02-13  155: 							let url = link;
26339860ce 2022-02-13  156: 							posts.insert(date, url.to_string());
9171c791eb 2021-11-13  157: 						}
9171c791eb 2021-11-13  158: 					};
9171c791eb 2021-11-13  159: 				},
9171c791eb 2021-11-13  160: 				Err(err) => match err {
9171c791eb 2021-11-13  161: 					rss::Error::InvalidStartTag => {
2c36f015d8 2024-11-23  162: 						match atom_syndication::Feed::read_from(&content[..]) {
2c36f015d8 2024-11-23  163: 							Ok(feed) => {
2c36f015d8 2024-11-23  164: 								for item in feed.entries() {
2c36f015d8 2024-11-23  165: 									let date = item.published().unwrap();
2c36f015d8 2024-11-23  166: 									let url = item.links()[0].href();
2c36f015d8 2024-11-23  167: 									posts.insert(*date, url.to_string());
2c36f015d8 2024-11-23  168: 								};
2c36f015d8 2024-11-23  169: 							},
2c36f015d8 2024-11-23  170: 							Err(err) => {
e624ef9d66 2025-04-20  171: 								bail!("Unsupported or mangled content:\n{:?}\n{err:#?}\n{status:#?}\n", &source.url)
2c36f015d8 2024-11-23  172: 							},
2c36f015d8 2024-11-23  173: 						}
9171c791eb 2021-11-13  174: 					},
9171c791eb 2021-11-13  175: 					rss::Error::Eof => (),
e624ef9d66 2025-04-20  176: 					_ => bail!("Unsupported or mangled content:\n{:?}\n{err:#?}\n{status:#?}\n", &source.url)
9171c791eb 2021-11-13  177: 				}
9171c791eb 2021-11-13  178: 			};
9171c791eb 2021-11-13  179: 			for (date, url) in posts.iter() {
9910c2209c 2023-08-04  180: 				let post_url: Cow<str> = match source.url_re {
3dc9cddd4d 2023-08-04  181: 					Some(ref x) => sedregex::ReplaceCommand::new(x)?.execute(url),
659724c658 2021-12-08  182: 					None => url.into(),
659724c658 2021-12-08  183: 				};
0340541002 2025-04-24  184: 				if let Some(exists) = conn.exists(&post_url, *id).await? {
9910c2209c 2023-08-04  185: 					if ! exists {
9910c2209c 2023-08-04  186: 						if this_fetch.is_none() || *date > this_fetch.unwrap() {
9910c2209c 2023-08-04  187: 							this_fetch = Some(*date);
9910c2209c 2023-08-04  188: 						};
e624ef9d66 2025-04-20  189: 						self.send( match &source.iv_hash {
e624ef9d66 2025-04-20  190: 							Some(hash) => format!("<a href=\"https://t.me/iv?url={post_url}&rhash={hash}\"> </a>{post_url}"),
e624ef9d66 2025-04-20  191: 							None => format!("{post_url}"),
e624ef9d66 2025-04-20  192: 						}, Some(destination), Some(ParseMode::Html)).await?;
0340541002 2025-04-24  193: 						conn.add_post(*id, date, &post_url).await?;
659724c658 2021-12-08  194: 					};
9171c791eb 2021-11-13  195: 				};
9171c791eb 2021-11-13  196: 				posted += 1;
9171c791eb 2021-11-13  197: 			};
9171c791eb 2021-11-13  198: 			posts.clear();
9171c791eb 2021-11-13  199: 		};
0340541002 2025-04-24  200: 		Ok(format!("Posted: {posted}"))
26339860ce 2022-02-13  201: 	}
26339860ce 2022-02-13  202: 
bb89b6fab8 2025-06-15  203: 	async fn autofetch(&self) -> Result<std::time::Duration> {
26339860ce 2022-02-13  204: 		let mut delay = chrono::Duration::minutes(1);
9910c2209c 2023-08-04  205: 		let now = chrono::Local::now();
bb89b6fab8 2025-06-15  206: 		let queue = {
bb89b6fab8 2025-06-15  207: 			let mut conn = self.db.begin().await?;
bb89b6fab8 2025-06-15  208: 			conn.get_queue().await?
bb89b6fab8 2025-06-15  209: 		};
bb89b6fab8 2025-06-15  210: 		for row in queue {
9910c2209c 2023-08-04  211: 			if let Some(next_fetch) = row.next_fetch {
9910c2209c 2023-08-04  212: 				if next_fetch < now {
9910c2209c 2023-08-04  213: 					if let (Some(owner), Some(source_id)) = (row.owner, row.source_id) {
bb89b6fab8 2025-06-15  214: 						let clone = Core {
bb89b6fab8 2025-06-15  215: 							owner_chat: ChatPeerId::from(owner),
9910c2209c 2023-08-04  216: 							..self.clone()
9910c2209c 2023-08-04  217: 						};
b4af85e31b 2025-06-29  218: 						let source = {
b4af85e31b 2025-06-29  219: 							let mut conn = self.db.begin().await?;
b4af85e31b 2025-06-29  220: 							match conn.get_one(owner, source_id).await {
b4af85e31b 2025-06-29  221: 								Ok(Some(source)) => source.to_string(),
b4af85e31b 2025-06-29  222: 								Ok(None) => "Source not found in database?".to_string(),
b4af85e31b 2025-06-29  223: 								Err(err) => format!("Failed to fetch source data:\n{err}"),
b4af85e31b 2025-06-29  224: 							}
b4af85e31b 2025-06-29  225: 						};
9910c2209c 2023-08-04  226: 						task::spawn(async move {
bb89b6fab8 2025-06-15  227: 							if let Err(err) = clone.check(source_id, true).await {
c6d3e97290 2025-07-01  228: 								if let Err(err) = clone.send(&format!("{source}\n\nšŸ›‘ {}", encode(&err.to_string())), None, Some(ParseMode::MarkdownV2)).await {
e624ef9d66 2025-04-20  229: 									eprintln!("Check error: {err:?}");
79c91a5357 2024-08-02  230: 									// clone.disable(&source_id, owner).await.unwrap();
9910c2209c 2023-08-04  231: 								};
9910c2209c 2023-08-04  232: 							};
9910c2209c 2023-08-04  233: 						});
9910c2209c 2023-08-04  234: 					}
9910c2209c 2023-08-04  235: 				} else if next_fetch - now < delay {
9910c2209c 2023-08-04  236: 					delay = next_fetch - now;
9910c2209c 2023-08-04  237: 				}
9910c2209c 2023-08-04  238: 			}
9910c2209c 2023-08-04  239: 		};
26339860ce 2022-02-13  240: 		Ok(delay.to_std()?)
26339860ce 2022-02-13  241: 	}
26339860ce 2022-02-13  242: 
bb89b6fab8 2025-06-15  243: 	pub async fn list (&self, owner: UserPeerId) -> Result<String> {
b4af85e31b 2025-06-29  244: 		let mut reply: Vec<String> = vec![];
26339860ce 2022-02-13  245: 		reply.push("Channels:".into());
0340541002 2025-04-24  246: 		let mut conn = self.db.begin().await?;
0340541002 2025-04-24  247: 		for row in conn.get_list(owner).await? {
b4af85e31b 2025-06-29  248: 			reply.push(row.to_string());
e624ef9d66 2025-04-20  249: 		};
b4af85e31b 2025-06-29  250: 		Ok(reply.join("\n\n"))
bb89b6fab8 2025-06-15  251: 	}
bb89b6fab8 2025-06-15  252: }
bb89b6fab8 2025-06-15  253: 
bb89b6fab8 2025-06-15  254: impl UpdateHandler for Core {
bb89b6fab8 2025-06-15  255: 	async fn handle (&self, update: Update) {
bb89b6fab8 2025-06-15  256: 		if let UpdateType::Message(msg) = update.update_type {
bb89b6fab8 2025-06-15  257: 			if let Ok(cmd) = Command::try_from(msg) {
bb89b6fab8 2025-06-15  258: 				let msg = cmd.get_message();
bb89b6fab8 2025-06-15  259: 				let words = cmd.get_args();
fae13a0e55 2025-06-28  260: 				let command = cmd.get_name();
fae13a0e55 2025-06-28  261: 				let res = match command {
fae13a0e55 2025-06-28  262: 					"/check" | "/clean" | "/enable" | "/delete" | "/disable" => command::command(self, command, msg, words).await,
bb89b6fab8 2025-06-15  263: 					"/start" => command::start(self, msg).await,
bb89b6fab8 2025-06-15  264: 					"/list" => command::list(self, msg).await,
fae13a0e55 2025-06-28  265: 					"/add" | "/update" => command::update(self, command, msg, words).await,
bb89b6fab8 2025-06-15  266: 					any => Err(anyhow!("Unknown command: {any}")),
bb89b6fab8 2025-06-15  267: 				};
bb89b6fab8 2025-06-15  268: 				if let Err(err) = res {
bb89b6fab8 2025-06-15  269: 					if let Err(err2) = self.send(format!("\\#error\n```\n{err:?}\n```"),
bb89b6fab8 2025-06-15  270: 						Some(msg.chat.get_id()),
bb89b6fab8 2025-06-15  271: 						Some(ParseMode::MarkdownV2)
bb89b6fab8 2025-06-15  272: 					).await{
bb89b6fab8 2025-06-15  273: 						dbg!(err2);
bb89b6fab8 2025-06-15  274: 					};
bb89b6fab8 2025-06-15  275: 				}
bb89b6fab8 2025-06-15  276: 			};
bb89b6fab8 2025-06-15  277: 		};
e624ef9d66 2025-04-20  278: 	}
9171c791eb 2021-11-13  279: }