feat(tvldb): Import the tvldb/paroxysm source, add a Nix derivation r/993

- This imports the tvldb (actually a thing called 'paroxysm') code
  from https://git.theta.eu.org/eta/paroxysm into the monorepo.
- Additionally, I did a nix thing, yay! \o/
  (well, with tazjin's help)
- 3p/default.nix needed modifying to whitelist pgsql.

Change-Id: Icdf13ca221650dde376f632bd2dd8a087af451bf
Reviewed-on: https://cl.tvl.fyi/c/depot/+/389
Reviewed-by: tazjin <mail@tazj.in>

5 files changed, 597 insertions, 0 deletions

diff --git a/fun/tvldb/src/cfg.rs b/fun/tvldb/src/cfg.rs
new file mode 100644
index 000000000000..038795a6f114
--- /dev/null
+++ b/fun/tvldb/src/cfg.rs
@@ -0,0 +1,11 @@
+use std::collections::HashSet;
+
+#[derive(Deserialize)]
+pub struct Config {
+    pub database_url: String,
+    pub irc_config_path: String,
+    #[serde(default)]
+    pub admins: HashSet<String>,
+    #[serde(default)]
+    pub log_filter: Option<String>,
+}
diff --git a/fun/tvldb/src/keyword.rs b/fun/tvldb/src/keyword.rs
new file mode 100644
index 000000000000..78dc36f35f12
--- /dev/null
+++ b/fun/tvldb/src/keyword.rs
@@ -0,0 +1,182 @@
+use crate::models::{Entry, Keyword, NewEntry, NewKeyword};
+use diesel::pg::PgConnection;
+use diesel::prelude::*;
+use failure::Error;
+use std::borrow::Cow;
+
+pub struct KeywordDetails {
+    pub keyword: Keyword,
+    pub entries: Vec<Entry>,
+}
+impl KeywordDetails {
+    pub fn learn(&mut self, nick: &str, text: &str, dbc: &PgConnection) -> Result<usize, Error> {
+        let now = ::chrono::Utc::now().naive_utc();
+        let ins = NewEntry {
+            keyword_id: self.keyword.id,
+            idx: (self.entries.len() + 1) as _,
+            text,
+            creation_ts: now,
+            created_by: nick,
+        };
+        let new = {
+            use crate::schema::entries;
+            ::diesel::insert_into(entries::table)
+                .values(ins)
+                .get_result(dbc)?
+        };
+        self.entries.push(new);
+        Ok(self.entries.len())
+    }
+    pub fn process_moves(&mut self, moves: &[(i32, i32)], dbc: &PgConnection) -> Result<(), Error> {
+        for (oid, new_idx) in moves {
+            {
+                use crate::schema::entries::dsl::*;
+                ::diesel::update(entries.filter(id.eq(oid)))
+                    .set(idx.eq(new_idx))
+                    .execute(dbc)?;
+            }
+        }
+        self.entries = Self::get_entries(self.keyword.id, dbc)?;
+        Ok(())
+    }
+    pub fn swap(&mut self, idx_a: usize, idx_b: usize, dbc: &PgConnection) -> Result<(), Error> {
+        let mut moves = vec![];
+        for ent in self.entries.iter() {
+            if ent.idx == idx_a as i32 {
+                moves.push((ent.id, idx_b as i32));
+            }
+            if ent.idx == idx_b as i32 {
+                moves.push((ent.id, idx_a as i32));
+            }
+        }
+        if moves.len() != 2 {
+            Err(format_err!("Invalid swap operation."))?;
+        }
+        self.process_moves(&moves, dbc)?;
+        Ok(())
+    }
+    pub fn update(&mut self, idx: usize, val: &str, dbc: &PgConnection) -> Result<(), Error> {
+        let ent = self
+            .entries
+            .get_mut(idx.saturating_sub(1))
+            .ok_or(format_err!("No such element to update."))?;
+        {
+            use crate::schema::entries::dsl::*;
+            ::diesel::update(entries.filter(id.eq(ent.id)))
+                .set(text.eq(val))
+                .execute(dbc)?;
+        }
+        ent.text = val.to_string();
+        Ok(())
+    }
+    pub fn delete(&mut self, idx: usize, dbc: &PgConnection) -> Result<(), Error> {
+        // step 1: delete the element
+        {
+            let ent = self
+                .entries
+                .get(idx.saturating_sub(1))
+                .ok_or(format_err!("No such element to delete."))?;
+            {
+                use crate::schema::entries::dsl::*;
+                ::diesel::delete(entries.filter(id.eq(ent.id))).execute(dbc)?;
+            }
+        }
+        // step 2: move all the elements in front of it back one
+        let mut moves = vec![];
+        for ent in self.entries.iter() {
+            if idx > ent.idx as _ {
+                moves.push((ent.id, ent.idx.saturating_sub(1)));
+            }
+        }
+        self.process_moves(&moves, dbc)?;
+        Ok(())
+    }
+    pub fn add_zwsp_to_name(name: &str) -> Option<String> {
+        let second_index = name.char_indices().nth(1).map(|(i, _)| i)?;
+        let (start, end) = name.split_at(second_index);
+        Some(format!("{}​{}", start, end))
+    }
+    pub fn format_entry(&self, idx: usize) -> Option<String> {
+        if let Some(ent) = self.entries.get(idx.saturating_sub(1)) {
+            let gen_clr = if self.keyword.chan == "*" {
+                "\x0307"
+            } else {
+                ""
+            };
+            let zwsp_name = Self::add_zwsp_to_name(&self.keyword.name)
+                .unwrap_or_else(|| self.keyword.name.clone());
+            Some(format!(
+                "\x02{}{}\x0f\x0315[{}/{}]\x0f: {} \x0f\x0314[{}]\x0f",
+                gen_clr,
+                zwsp_name,
+                idx,
+                self.entries.len(),
+                ent.text,
+                ent.creation_ts.date()
+            ))
+        } else {
+            None
+        }
+    }
+    pub fn get_or_create(word: &str, c: &str, dbc: &PgConnection) -> Result<Self, Error> {
+        if let Some(ret) = Self::get(word, c, dbc)? {
+            Ok(ret)
+        } else {
+            Ok(Self::create(word, c, dbc)?)
+        }
+    }
+    pub fn create(word: &str, c: &str, dbc: &PgConnection) -> Result<Self, Error> {
+        let val = NewKeyword {
+            name: word,
+            chan: c,
+        };
+        let ret: Keyword = {
+            use crate::schema::keywords;
+            ::diesel::insert_into(keywords::table)
+                .values(val)
+                .get_result(dbc)?
+        };
+        Ok(KeywordDetails {
+            keyword: ret,
+            entries: vec![],
+        })
+    }
+    fn get_entries(kid: i32, dbc: &PgConnection) -> Result<Vec<Entry>, Error> {
+        let entries: Vec<Entry> = {
+            use crate::schema::entries::dsl::*;
+            entries
+                .filter(keyword_id.eq(kid))
+                .order_by(idx.asc())
+                .load(dbc)?
+        };
+        Ok(entries)
+    }
+    pub fn get<'a, T: Into<Cow<'a, str>>>(
+        word: T,
+        c: &str,
+        dbc: &PgConnection,
+    ) -> Result<Option<Self>, Error> {
+        let word = word.into();
+        let keyword: Option<Keyword> = {
+            use crate::schema::keywords::dsl::*;
+            keywords
+                .filter(name.ilike(word).and(chan.eq(c).or(chan.eq("*"))))
+                .first(dbc)
+                .optional()?
+        };
+        if let Some(k) = keyword {
+            let entries = Self::get_entries(k.id, dbc)?;
+            if let Some(e0) = entries.get(0) {
+                if e0.text.starts_with("see: ") {
+                    return Self::get(e0.text.replace("see: ", ""), c, dbc);
+                }
+            }
+            Ok(Some(KeywordDetails {
+                keyword: k,
+                entries,
+            }))
+        } else {
+            Ok(None)
+        }
+    }
+}
diff --git a/fun/tvldb/src/main.rs b/fun/tvldb/src/main.rs
new file mode 100644
index 000000000000..9fe7584b61b7
--- /dev/null
+++ b/fun/tvldb/src/main.rs
@@ -0,0 +1,353 @@
+extern crate irc;
+extern crate serde;
+#[macro_use]
+extern crate serde_derive;
+#[macro_use]
+extern crate diesel;
+extern crate chrono;
+extern crate config;
+extern crate env_logger;
+#[macro_use]
+extern crate log;
+#[macro_use]
+extern crate failure;
+extern crate regex;
+#[macro_use]
+extern crate lazy_static;
+extern crate rand;
+
+use crate::cfg::Config;
+use crate::keyword::KeywordDetails;
+use diesel::pg::PgConnection;
+use diesel::r2d2::{ConnectionManager, Pool};
+use failure::Error;
+use irc::client::prelude::*;
+use rand::rngs::ThreadRng;
+use rand::{thread_rng, Rng};
+use regex::{Captures, Regex};
+use std::collections::HashMap;
+use std::fmt::Display;
+
+mod cfg;
+mod keyword;
+mod models;
+mod schema;
+
+pub struct App {
+    client: IrcClient,
+    pg: Pool<ConnectionManager<PgConnection>>,
+    rng: ThreadRng,
+    cfg: Config,
+    last_msgs: HashMap<String, HashMap<String, String>>,
+}
+
+impl App {
+    pub fn report_error<T: Display>(
+        &mut self,
+        nick: &str,
+        chan: &str,
+        msg: T,
+    ) -> Result<(), Error> {
+        self.client
+            .send_notice(nick, format!("[{}] \x0304Error:\x0f {}", chan, msg))?;
+        Ok(())
+    }
+    pub fn keyword_from_captures(
+        &mut self,
+        learn: &::regex::Captures,
+        nick: &str,
+        chan: &str,
+    ) -> Result<KeywordDetails, Error> {
+        let db = self.pg.get()?;
+        debug!("Fetching keyword for captures: {:?}", learn);
+        let subj = &learn["subj"];
+        let learn_chan = if learn.name("gen").is_some() {
+            "*"
+        } else {
+            chan
+        };
+        if !chan.starts_with("#") && learn_chan != "*" {
+            Err(format_err!("Only general entries may be taught via PM."))?;
+        }
+        debug!("Fetching keyword '{}' for chan {}", subj, learn_chan);
+        let kwd = KeywordDetails::get_or_create(subj, learn_chan, &db)?;
+        if kwd.keyword.chan == "*" && !self.cfg.admins.contains(nick) {
+            Err(format_err!(
+                "Only administrators can create or modify general entries."
+            ))?;
+        }
+        Ok(kwd)
+    }
+    pub fn handle_move(
+        &mut self,
+        target: &str,
+        nick: &str,
+        chan: &str,
+        mv: Captures,
+    ) -> Result<(), Error> {
+        let db = self.pg.get()?;
+        let idx = &mv["idx"];
+        let idx = match idx[1..(idx.len() - 1)].parse::<usize>() {
+            Ok(i) => i,
+            Err(e) => Err(format_err!("Could not parse index: {}", e))?,
+        };
+        let new_idx = match mv["new_idx"].parse::<i32>() {
+            Ok(i) => i,
+            Err(e) => Err(format_err!("Could not parse target index: {}", e))?,
+        };
+        let mut kwd = self.keyword_from_captures(&mv, nick, chan)?;
+        if new_idx < 0 {
+            kwd.delete(idx, &db)?;
+            self.client.send_notice(
+                target,
+                format!("\x02{}\x0f: Deleted entry {}.", kwd.keyword.name, idx),
+            )?;
+        } else {
+            kwd.swap(idx, new_idx as _, &db)?;
+            self.client.send_notice(
+                target,
+                format!(
+                    "\x02{}\x0f: Swapped entries {} and {}.",
+                    kwd.keyword.name, idx, new_idx
+                ),
+            )?;
+        }
+        Ok(())
+    }
+    pub fn handle_learn(
+        &mut self,
+        target: &str,
+        nick: &str,
+        chan: &str,
+        learn: Captures,
+    ) -> Result<(), Error> {
+        let db = self.pg.get()?;
+        let val = &learn["val"];
+        let mut kwd = self.keyword_from_captures(&learn, nick, chan)?;
+        let idx = kwd.learn(nick, val, &db)?;
+        self.client
+            .send_notice(target, kwd.format_entry(idx).unwrap())?;
+        Ok(())
+    }
+    pub fn handle_insert_last_quote(
+        &mut self,
+        target: &str,
+        nick: &str,
+        chan: &str,
+        qlast: Captures,
+    ) -> Result<(), Error> {
+        let db = self.pg.get()?;
+        let mut kwd = self.keyword_from_captures(&qlast, nick, chan)?;
+        let chan_lastmsgs = self
+            .last_msgs
+            .entry(chan.to_string())
+            .or_insert(HashMap::new());
+        let val = if let Some(last) = chan_lastmsgs.get(&kwd.keyword.name.to_ascii_lowercase()) {
+            format!("<{}> {}", &kwd.keyword.name, last)
+        } else {
+            Err(format_err!("I dunno what {} said...", kwd.keyword.name))?
+        };
+        let idx = kwd.learn(nick, &val, &db)?;
+        self.client
+            .send_notice(target, kwd.format_entry(idx).unwrap())?;
+        Ok(())
+    }
+    pub fn handle_increment(
+        &mut self,
+        target: &str,
+        nick: &str,
+        chan: &str,
+        icr: Captures,
+    ) -> Result<(), Error> {
+        let db = self.pg.get()?;
+        let mut kwd = self.keyword_from_captures(&icr, nick, chan)?;
+        let is_incr = &icr["incrdecr"] == "++";
+        let now = chrono::Utc::now().naive_utc().date();
+        let mut idx = None;
+        for (i, ent) in kwd.entries.iter().enumerate() {
+            if ent.creation_ts.date() == now {
+                if let Ok(val) = ent.text.parse::<i32>() {
+                    let val = if is_incr { val + 1 } else { val - 1 };
+                    idx = Some((i + 1, val));
+                }
+            }
+        }
+        if let Some((i, val)) = idx {
+            kwd.update(i, &val.to_string(), &db)?;
+            self.client.send_notice(target, kwd.format_entry(i).unwrap())?;
+        } else {
+            let val = if is_incr { 1 } else { -1 };
+            let idx = kwd.learn(nick, &val.to_string(), &db)?;
+            self.client
+                .send_notice(target, kwd.format_entry(idx).unwrap())?;
+        }
+        Ok(())
+    }
+    pub fn handle_query(
+        &mut self,
+        target: &str,
+        nick: &str,
+        chan: &str,
+        query: Captures,
+    ) -> Result<(), Error> {
+        let db = self.pg.get()?;
+        let subj = &query["subj"];
+        let idx = match query.name("idx") {
+            Some(i) => {
+                let i = i.as_str();
+                match &i[1..(i.len() - 1)] {
+                    "*" => Some(-1),
+                    x => x.parse::<usize>().map(|x| x as i32).ok(),
+                }
+            }
+            None => None,
+        };
+        debug!("Querying {} with idx {:?}", subj, idx);
+        match KeywordDetails::get(subj, chan, &db)? {
+            Some(kwd) => {
+                if let Some(mut idx) = idx {
+                    if idx == -1 {
+                        // 'get all entries' ('*' parses into this)
+                        for i in 0..kwd.entries.len() {
+                            self.client.send_notice(
+                                nick,
+                                format!("[{}] {}", chan, kwd.format_entry(i + 1).unwrap()),
+                            )?;
+                        }
+                    } else {
+                        if idx == 0 {
+                            idx = 1;
+                        }
+                        if let Some(ent) = kwd.format_entry(idx as _) {
+                            self.client.send_notice(target, ent)?;
+                        } else {
+                            let pluralised = if kwd.entries.len() == 1 {
+                                "entry"
+                            } else {
+                                "entries"
+                            };
+                            self.client.send_notice(
+                                target,
+                                format!(
+                                    "\x02{}\x0f: only has \x02\x0304{}\x0f {}",
+                                    subj,
+                                    kwd.entries.len(),
+                                    pluralised
+                                ),
+                            )?;
+                        }
+                    }
+                } else {
+                    let entry = if kwd.entries.len() < 2 {
+                        1 // because [1, 1) does not a range make
+                    } else {
+                        self.rng.gen_range(1, kwd.entries.len())
+                    };
+                    if let Some(ent) = kwd.format_entry(entry) {
+                        self.client.send_notice(target, ent)?;
+                    } else {
+                        self.client
+                            .send_notice(target, format!("\x02{}\x0f: no entries yet", subj))?;
+                    }
+                }
+            }
+            None => {
+                self.client
+                    .send_notice(target, format!("\x02{}\x0f: never heard of it", subj))?;
+            }
+        }
+        Ok(())
+    }
+    pub fn handle_privmsg(&mut self, from: &str, chan: &str, msg: &str) -> Result<(), Error> {
+        lazy_static! {
+            static ref LEARN_RE: Regex =
+                Regex::new(r#"^\?\?(?P<gen>!)?\s*(?P<subj>[^\[:]*):\s*(?P<val>.*)"#).unwrap();
+            static ref QUERY_RE: Regex =
+                Regex::new(r#"^\?\?\s*(?P<subj>[^\[:]*)(?P<idx>\[[^\]]+\])?"#).unwrap();
+            static ref QLAST_RE: Regex = Regex::new(r#"^\?\?\s*(?P<subj>[^\[:]*)!"#).unwrap();
+            static ref INCREMENT_RE: Regex =
+                Regex::new(r#"^\?\?(?P<gen>!)?\s*(?P<subj>[^\[:]*)(?P<incrdecr>\+\+|\-\-)"#)
+                    .unwrap();
+            static ref MOVE_RE: Regex = Regex::new(
+                r#"^\?\?(?P<gen>!)?\s*(?P<subj>[^\[:]*)(?P<idx>\[[^\]]+\])->(?P<new_idx>.*)"#
+            )
+            .unwrap();
+        }
+        let nick = from.split("!").next().ok_or(format_err!(
+            "Received PRIVMSG from a source without nickname (failed to split n!u@h)"
+        ))?;
+        let target = if chan.starts_with("#") { chan } else { nick };
+        debug!("[{}] <{}> {}", chan, nick, msg);
+        if let Some(learn) = LEARN_RE.captures(msg) {
+            self.handle_learn(target, nick, chan, learn)?;
+        } else if let Some(qlast) = QLAST_RE.captures(msg) {
+            self.handle_insert_last_quote(target, nick, chan, qlast)?;
+        } else if let Some(mv) = MOVE_RE.captures(msg) {
+            self.handle_move(target, nick, chan, mv)?;
+        } else if let Some(icr) = INCREMENT_RE.captures(msg) {
+            self.handle_increment(target, nick, chan, icr)?;
+        } else if let Some(query) = QUERY_RE.captures(msg) {
+            self.handle_query(target, nick, chan, query)?;
+        } else {
+            let chan_lastmsgs = self
+                .last_msgs
+                .entry(chan.to_string())
+                .or_insert(HashMap::new());
+            chan_lastmsgs.insert(nick.to_string().to_ascii_lowercase(), msg.to_string());
+        }
+        Ok(())
+    }
+    pub fn handle_msg(&mut self, m: Message) -> Result<(), Error> {
+        match m.command {
+            Command::PRIVMSG(channel, message) => {
+                if let Some(src) = m.prefix {
+                    if let Err(e) = self.handle_privmsg(&src, &channel, &message) {
+                        warn!("error handling command in {} (src {}): {}", channel, src, e);
+                        if let Some(nick) = src.split("!").next() {
+                            self.report_error(nick, &channel, e)?;
+                        }
+                    }
+                }
+            }
+            Command::INVITE(nick, channel) => {
+                if self.cfg.admins.contains(&nick) {
+                    info!("Joining {} after admin invite", channel);
+                    self.client.send_join(channel)?;
+                }
+            }
+            _ => {}
+        }
+        Ok(())
+    }
+}
+fn main() -> Result<(), Error> {
+    println!("[+] loading configuration");
+    let default_log_filter = "paroxysm=info".to_string();
+    let mut settings = config::Config::default();
+    settings.merge(config::Environment::with_prefix("PARX"))?;
+    let cfg: Config = settings.try_into()?;
+    let env = env_logger::Env::new()
+        .default_filter_or(cfg.log_filter.clone().unwrap_or(default_log_filter));
+    env_logger::init_from_env(env);
+    info!("paroxysm starting up");
+    info!("connecting to database at {}", cfg.database_url);
+    let pg = Pool::new(ConnectionManager::new(&cfg.database_url))?;
+    info!("connecting to IRC using config {}", cfg.irc_config_path);
+    let client = IrcClient::new(&cfg.irc_config_path)?;
+    client.identify()?;
+    let st = client.stream();
+    let mut app = App {
+        client,
+        pg,
+        cfg,
+        rng: thread_rng(),
+        last_msgs: HashMap::new(),
+    };
+    info!("running!");
+    st.for_each_incoming(|m| {
+        if let Err(e) = app.handle_msg(m) {
+            warn!("Error processing message: {}", e);
+        }
+    })?;
+    Ok(())
+}
diff --git a/fun/tvldb/src/models.rs b/fun/tvldb/src/models.rs
new file mode 100644
index 000000000000..78ddce0af96c
--- /dev/null
+++ b/fun/tvldb/src/models.rs
@@ -0,0 +1,33 @@
+use crate::schema::{entries, keywords};
+use chrono::NaiveDateTime;
+
+#[derive(Queryable)]
+pub struct Keyword {
+    pub id: i32,
+    pub name: String,
+    pub chan: String,
+}
+#[derive(Queryable)]
+pub struct Entry {
+    pub id: i32,
+    pub keyword_id: i32,
+    pub idx: i32,
+    pub text: String,
+    pub creation_ts: NaiveDateTime,
+    pub created_by: String,
+}
+#[derive(Insertable)]
+#[table_name = "keywords"]
+pub struct NewKeyword<'a> {
+    pub name: &'a str,
+    pub chan: &'a str,
+}
+#[derive(Insertable)]
+#[table_name = "entries"]
+pub struct NewEntry<'a> {
+    pub keyword_id: i32,
+    pub idx: i32,
+    pub text: &'a str,
+    pub creation_ts: NaiveDateTime,
+    pub created_by: &'a str,
+}
diff --git a/fun/tvldb/src/schema.rs b/fun/tvldb/src/schema.rs
new file mode 100644
index 000000000000..ef4044531ee7
--- /dev/null
+++ b/fun/tvldb/src/schema.rs
@@ -0,0 +1,18 @@
+table! {
+    entries (id) {
+        id -> Int4,
+        keyword_id -> Int4,
+        idx -> Int4,
+        text -> Varchar,
+        creation_ts -> Timestamp,
+        created_by -> Varchar,
+    }
+}
+
+table! {
+    keywords (id) {
+        id -> Int4,
+        name -> Varchar,
+        chan -> Varchar,
+    }
+}