Compare commits

..

No commits in common. "66e72aeb54c1aa200e15b42ffab7b4c4719a0e1d" and "c5cb4eb524300e87c1194b1ab33f01e5bf34134e" have entirely different histories.

4 changed files with 80 additions and 2353 deletions

2051
Cargo.lock generated

File diff suppressed because it is too large Load Diff

View File

@ -18,4 +18,3 @@ tracing = "0.1.40"
tracing-subscriber = { version = "0.3.18", features = ["env-filter"] } tracing-subscriber = { version = "0.3.18", features = ["env-filter"] }
serde_json = "1.0.120" serde_json = "1.0.120"
serde = { version = "1.0.204", features = ["derive"] } serde = { version = "1.0.204", features = ["derive"] }
surrealdb = "1.5.4"

View File

@ -1,26 +1,27 @@
use std::{collections::HashMap, fmt::Display, fs, hint::black_box, sync::Arc}; use std::{collections::HashMap, fmt::Display, fs, sync::Arc};
use crate::Context; use crate::Context;
use anyhow::Error; use anyhow::Error;
use poise::{ use poise::{serenity_prelude::{Cache, CacheHttp, ChannelId, ChannelType, GetMessages, GuildChannel, Http, Message}, CreateReply};
serenity_prelude::{ use serde::Serialize;
Cache, CacheHttp, ChannelId, ChannelType, GetMessages, GuildChannel, Http, Message,
},
CreateReply,
};
use serde::{Deserialize, Serialize};
use surrealdb::{engine::remote::ws::Ws, opt::auth::Root, sql::Thing, Surreal};
use tokio::time::Instant; use tokio::time::Instant;
use tracing::{debug, error, info, trace}; use tracing::{debug, error, info, trace};
#[derive(Serialize, Deserialize)] #[derive(Serialize)]
struct Server {
channels: Vec<Channel>,
#[serde(skip_serializing_if = "Vec::is_empty")]
orphanage: Vec<GuildChannel>,
#[serde(skip_serializing)]
needs_clean: bool,
}
#[derive(Serialize)]
struct Channel { struct Channel {
this: GuildChannel, this: GuildChannel,
#[serde(skip_serializing_if = "Vec::is_empty")] #[serde(skip_serializing_if = "Vec::is_empty")]
#[serde(default)]
children: Vec<Channel>, children: Vec<Channel>,
#[serde(skip_serializing_if = "Vec::is_empty")] #[serde(skip_serializing_if = "Vec::is_empty")]
#[serde(default)]
messages: Vec<Message>, messages: Vec<Message>,
} }
@ -33,34 +34,16 @@ impl Channel {
messages: Vec::new(), messages: Vec::new(),
} }
} }
fn is_category(&self) -> bool {
!self.children.is_empty() && self.messages.is_empty()
}
}
#[derive(Serialize, Deserialize)]
struct Server {
#[serde(default)]
name: String,
channels: Vec<Channel>,
#[serde(skip_serializing_if = "Vec::is_empty")]
#[serde(default)]
orphanage: Vec<GuildChannel>,
#[serde(skip_serializing)]
#[serde(default)]
needs_clean: bool,
} }
impl Display for Server { impl Display for Server {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
fn print(f: &mut std::fmt::Formatter<'_>, tab: usize, channel: &Vec<Channel>) { fn print(f: &mut std::fmt::Formatter<'_>, tab: usize, channel: &Vec<Channel>) {
for i in channel { for i in channel {
for _ in 0..tab { for _ in 0..tab { let _ = write!(f, "\t"); }
let _ = write!(f, "\t"); let _ = writeln!(f, "{} {}", prefix(i.this.kind),i.this.name);
} print(f, tab+1, &i.children);
let _ = writeln!(f, "{} {}", prefix(i.this.kind), i.this.name);
print(f, tab + 1, &i.children);
} }
} }
@ -84,10 +67,11 @@ impl Display for Server {
print(f, 0, &self.channels); print(f, 0, &self.channels);
if self.needs_clean { if self.needs_clean {
let _ = writeln!(f, "Orphans: (please clean() before displaying...)"); let _ = writeln!(f, "Orphans: (please clean() before displaying...)");
for i in &self.orphanage { for i in &self.orphanage {
let _ = write!(f, "{} {},", prefix(i.kind), i.name); let _ = write!(f, "{} {},", prefix(i.kind),i.name);
} }
} }
@ -96,12 +80,12 @@ impl Display for Server {
} }
impl Server { impl Server {
fn new(name: impl Into<String>) -> Self {
fn new() -> Self {
Self { Self {
name: name.into(),
channels: Vec::new(), channels: Vec::new(),
orphanage: Vec::new(), orphanage: Vec::new(),
needs_clean: false, needs_clean: false
} }
} }
@ -113,9 +97,9 @@ impl Server {
if let Some(x) = Self::search_by_id(&mut child.children, find) { if let Some(x) = Self::search_by_id(&mut child.children, find) {
return Some(x); return Some(x);
} }
} }
None None
} }
fn add(&mut self, insert: GuildChannel) { fn add(&mut self, insert: GuildChannel) {
// make sure the new item wants a parent // make sure the new item wants a parent
@ -125,12 +109,12 @@ impl Server {
match Self::search_by_id(&mut self.channels, parent_id) { match Self::search_by_id(&mut self.channels, parent_id) {
Some(parent_node) => { Some(parent_node) => {
parent_node.children.push(Channel::new(insert)); parent_node.children.push(Channel::new(insert));
} },
None => { None => {
// couldn't find parent, store somewhere else until it's parent is added... // couldn't find parent, store somewhere else until it's parent is added...
self.orphanage.push(insert); self.orphanage.push(insert);
self.needs_clean = true; self.needs_clean = true;
} },
} }
} else { } else {
self.channels.push(Channel::new(insert)); self.channels.push(Channel::new(insert));
@ -139,9 +123,7 @@ impl Server {
/// Cleans out the orphan channels, finding them parents. You'll want to use this before displaying anything. /// Cleans out the orphan channels, finding them parents. You'll want to use this before displaying anything.
fn clean(&mut self) { fn clean(&mut self) {
if !self.needs_clean { if !self.needs_clean {return;}
return;
}
// Look thru the orphanage and try to find parents // Look thru the orphanage and try to find parents
for orphan in &self.orphanage { for orphan in &self.orphanage {
@ -178,20 +160,12 @@ impl Server {
} }
/// Get all messages for 1 channel and children /// Get all messages for 1 channel and children
async fn get_messages( async fn get_messages(channel: &mut Channel, cache: impl CacheHttp + Clone, settings: GetMessages) {
channel: &mut Channel,
cache: impl CacheHttp + Clone,
settings: GetMessages,
) {
// Loop thru all the messages in the channel in batches. // Loop thru all the messages in the channel in batches.
// Adding each batch to the current channel's messages the whole time. // Adding each batch to the current channel's messages the whole time.
let mut last_id = channel.this.last_message_id; let mut last_id = channel.this.last_message_id;
while let Some(last) = last_id { while let Some(last) = last_id {
match channel match channel.this.messages(cache.clone(), settings.before(last)).await {
.this
.messages(cache.clone(), settings.before(last))
.await
{
Ok(mut ok) => { Ok(mut ok) => {
if ok.is_empty() { if ok.is_empty() {
debug!("Reached the beginning of {}", channel.this.name); debug!("Reached the beginning of {}", channel.this.name);
@ -205,15 +179,12 @@ impl Server {
last_id = Some(l.id); last_id = Some(l.id);
} }
} }
} },
Err(e) => { Err(e) => {
error!( error!("While reading messages in \"{}\" before `{}` - {e}", channel.this.name, last);
"While reading messages in \"{}\" before `{}` - {e}",
channel.this.name, last
);
// Stop reading this channel on an error. // Stop reading this channel on an error.
last_id = None; last_id = None;
} },
} }
} }
// Then recurse into children channels // Then recurse into children channels
@ -230,228 +201,17 @@ impl Server {
for channel in this { for channel in this {
total += channel.messages.len(); total += channel.messages.len();
total += walk(&channel.children); total += walk(&channel.children);
} };
total total
} }
walk(&self.channels) walk(&self.channels)
} }
async fn to_surreal(&self) -> surrealdb::Result<()> {
trace!("Connecting to surrealdb...");
// Connect to the server
let db = Surreal::new::<Ws>("127.0.0.1:8000").await?;
db.signin(Root {
username: "root",
password: "root",
})
.await?;
db.use_ns("egress").use_db(self.name.clone()).await?;
// =========================================================
// Ingress data
// Data will only be in three layers
// Layer 1: Categories (no parent)
// Layer 2: Channels (might have parent)
// Layer 3: Messages (has parent)
trace!("Starting ingress...");
for cat in self.channels.iter() {
match cat.this.kind {
ChannelType::Text => {
// This is a text channel
#[derive(Serialize)]
struct ChannelWrapper {
name: String,
nsfw: bool,
discord_id: u64,
discord_parent_id: Option<u64>,
topic: String,
}
let chan = &cat.this;
let dpi = if let Some(val) = chan.parent_id {Some(val.get())} else { None };
let new_channel: Vec<Thing> = db
.create("channel")
.content(ChannelWrapper {
name: chan.name.to_owned(),
nsfw: chan.nsfw,
discord_id: chan.id.get(),
discord_parent_id: dpi,
topic: chan.topic.to_owned().unwrap_or(String::new()),
})
.await?;
for msg in cat.messages.iter() {
#[derive(Serialize)]
struct Author {
nickname: String,
username: String,
/// B64 encoded string of image (for now)
avatar: String,
id: u64,
}
#[derive(Serialize)]
struct Attachment {
content_type: String,
filename: String,
url: String,
}
#[derive(Serialize)]
struct Reaction {
count: u64,
emoji: String,
}
#[derive(Serialize)]
struct MessageWrapper {
// FIXME learn how to do references
parent: Thing,
author: Author,
content: String,
utc_timestamp: String,
mentions: Vec<u64>,
attachments: Vec<Attachment>,
reactions: Vec<Reaction>,
pinned: bool,
}
let _: Vec<Thing> = db
.create("message")
.content(MessageWrapper {
parent: new_channel[0].clone(),
author: Author {
id: msg.author.id.get(),
nickname: msg.author.name.to_owned(),
username: msg.author.global_name.clone().unwrap_or(String::new()),
avatar: {
match msg.author.avatar {
Some(hash) => {
format!(
"https://cdn.discordapp.com/avatars/{}/{}.webp",
msg.author.id,
hash,
)
},
None => String::new(),
}
}
},
content: msg.content.clone(),
utc_timestamp: msg.timestamp.to_utc().to_string(),
mentions: msg.mentions.iter().map(|f| f.id.get()).collect(),
attachments: msg.attachments.iter().map(|f| Attachment {
content_type: f.content_type.clone().unwrap_or(String::new()),
filename: f.filename.to_owned(),
url: f.url.to_owned(),
}).collect(),
reactions: msg.reactions.iter().map(|f| Reaction {
count: f.count,
emoji: f.reaction_type.as_data(),
}).collect(),
pinned: msg.pinned,
})
.await?;
}
}
ChannelType::Private => todo!(),
ChannelType::Voice => todo!(),
ChannelType::GroupDm => todo!(),
ChannelType::Category => todo!(),
ChannelType::News => todo!(),
ChannelType::NewsThread => todo!(),
ChannelType::PublicThread => todo!(),
ChannelType::PrivateThread => todo!(),
ChannelType::Stage => todo!(),
ChannelType::Directory => todo!(),
ChannelType::Forum => todo!(),
ChannelType::Unknown(_) => todo!(),
_ => todo!(),
}
// TODO learn why this is a vec
// Do the first iteration of channels a bit different, so as to name it "category".
let new_category: Vec<Thing> = db.create("category").content(&cat.this).await?;
import_messages(&cat.messages, &new_category[0], &db).await?;
// Ok, now automatically recurse the rest of the structure and auto import as channels
// and messages.
import_channel(&cat.children, &new_category[0], &db).await?;
}
async fn import_channel(
channels: &Vec<Channel>,
parent: &Thing,
db: &Surreal<surrealdb::engine::remote::ws::Client>,
) -> surrealdb::Result<()> {
for channel in channels.iter() {
trace!("Importing channel \"{}\"", channel.this.name);
#[derive(Serialize)]
struct ChannelWrapper<'a, 'b> {
channel: &'a GuildChannel,
surreal_parent: &'b Thing,
}
let new_channel: Vec<Thing> = db
.create("channel")
.content(ChannelWrapper {
channel: &channel.this,
surreal_parent: &parent,
})
.await?;
import_messages(&channel.messages, &new_channel[0], &db).await?;
// async recursion - thus box
Box::pin(import_channel(&channel.children, &new_channel[0], &db)).await?;
}
Ok(())
}
async fn import_messages(
msgs: &Vec<Message>,
parent: &Thing,
db: &Surreal<surrealdb::engine::remote::ws::Client>,
) -> surrealdb::Result<()> {
trace!("Importing {} messages...", msgs.len());
for msg in msgs.iter() {
#[derive(Serialize)]
struct MessageWrapper<'a, 'b> {
message: &'a Message,
surreal_parent: &'b Thing,
}
let created: Vec<Thing> = db
.create("message")
.content(MessageWrapper {
message: &msg,
surreal_parent: &parent,
})
.await?;
trace!("Imported message {:?}", created);
}
Ok(())
}
// Data is all in
// =========================================================
Ok(())
}
} }
#[poise::command(slash_command, rename = "scrape_all", guild_only, owners_only)] #[poise::command(slash_command, rename = "scrape_all", guild_only, owners_only)]
pub async fn scrape_all(ctx: Context<'_>, pretty_print: bool) -> Result<(), Error> { pub async fn scrape_all(ctx: Context<'_>, pretty_print: bool) -> Result<(), Error> {
let guild = ctx let guild = ctx.guild_id().unwrap().to_partial_guild(ctx.serenity_context()).await.unwrap();
.guild_id()
.unwrap()
.to_partial_guild(ctx.serenity_context())
.await
.unwrap();
let invoker = ctx.author().name.clone(); let invoker = ctx.author().name.clone();
if let Some(nickname) = ctx.author().nick_in(ctx.http(), guild.id).await { if let Some(nickname) = ctx.author().nick_in(ctx.http(), guild.id).await {
@ -461,7 +221,7 @@ pub async fn scrape_all(ctx: Context<'_>, pretty_print: bool) -> Result<(), Erro
} }
if let Ok(map) = guild.channels(ctx.http()).await { if let Ok(map) = guild.channels(ctx.http()).await {
let mut server = index(map, guild.name).await; let mut server = index(map).await;
match ctx.reply("Starting scrape...").await { match ctx.reply("Starting scrape...").await {
Ok(ok) => { Ok(ok) => {
let start = Instant::now(); let start = Instant::now();
@ -469,48 +229,38 @@ pub async fn scrape_all(ctx: Context<'_>, pretty_print: bool) -> Result<(), Erro
let end = start.elapsed().as_millis(); let end = start.elapsed().as_millis();
let msg_count = server.message_count(); let msg_count = server.message_count();
if let Err(e) = server.to_surreal().await { let print = if pretty_print {
error!("{e}"); serde_json::to_string_pretty(&server)
} else {
serde_json::to_string(&server)
}; };
// let print = if pretty_print { match print {
// serde_json::to_string_pretty(&server) Ok(ok) => {
// } else { if let Err(e) = fs::write("server.json", ok) {
// serde_json::to_string(&server) error!("Problem writing server to disk: {e}");
// }; }
},
// match print { Err(err) => {
// Ok(ok) => { error!("Trying to serialize server: {err}");
// if let Err(e) = fs::write("server.json", ok) { },
// error!("Problem writing server to disk: {e}"); }
// }
// },
// Err(err) => {
// error!("Trying to serialize server: {err}");
// },
// }
// Done. Print stats. // Done. Print stats.
let _ = ok.edit(ctx, CreateReply::default().content( let _ = ok.edit(ctx, CreateReply::default().content(
&format!("Done. Stats: \n```toml\nMessages saved: {msg_count}\nElapsed time: {end}ms\n```") &format!("Done. Stats: \n```toml\nMessages saved: {msg_count}\nElapsed time: {end}ms\n```")
)).await; )).await;
debug!("Scraped server in {}ms", end); debug!("Scraped server in {}ms", end);
} },
Err(e) => error!("{e} - While trying to reply to scrape command"), Err(e) => error!("{e} - While trying to reply to scrape command"),
} }
} }
Ok(()) Ok(())
} }
pub async fn from_json() {
let data = fs::read_to_string("server.json").unwrap();
let server: Server = serde_json::from_str(&data).unwrap();
server.to_surreal().await.unwrap();
}
/// Get server's topology (and runs clean) /// Get server's topology (and runs clean)
async fn index(map: HashMap<ChannelId, GuildChannel>, name: impl Into<String>) -> Server { async fn index(map: HashMap<ChannelId, GuildChannel>) -> Server {
let mut server = Server::new(name); let mut server = Server::new();
// iterate thru all channels // iterate thru all channels
map.into_iter().for_each(|(_id, current)| { map.into_iter().for_each(|(_id, current)| {
// println!("{} {} {:?}", current.name, current.id, current.parent_id); // println!("{} {} {:?}", current.name, current.id, current.parent_id);
@ -525,12 +275,7 @@ async fn index(map: HashMap<ChannelId, GuildChannel>, name: impl Into<String>) -
// NOTE!!! Make sure these names in quotes are lowercase! // NOTE!!! Make sure these names in quotes are lowercase!
#[poise::command(slash_command, rename = "index", guild_only)] #[poise::command(slash_command, rename = "index", guild_only)]
pub async fn index_cmd(ctx: Context<'_>) -> Result<(), Error> { pub async fn index_cmd(ctx: Context<'_>) -> Result<(), Error> {
let guild = ctx let guild = ctx.guild_id().unwrap().to_partial_guild(ctx.serenity_context()).await.unwrap();
.guild_id()
.unwrap()
.to_partial_guild(ctx.serenity_context())
.await
.unwrap();
let invoker = ctx.author().name.clone(); let invoker = ctx.author().name.clone();
if let Some(nickname) = ctx.author().nick_in(ctx.http(), guild.id).await { if let Some(nickname) = ctx.author().nick_in(ctx.http(), guild.id).await {
info!("{invoker} ({nickname}) is indexing {}", guild.name); info!("{invoker} ({nickname}) is indexing {}", guild.name);
@ -540,9 +285,9 @@ pub async fn index_cmd(ctx: Context<'_>) -> Result<(), Error> {
match guild.channels(ctx.http()).await { match guild.channels(ctx.http()).await {
Ok(ok) => { Ok(ok) => {
let server = index(ok, guild.name).await; let server = index(ok).await;
let _ = ctx.reply(server.to_string()).await; let _ = ctx.reply(server.to_string()).await;
} },
Err(_) => todo!(), Err(_) => todo!(),
} }
Ok(()) Ok(())

View File

@ -15,8 +15,6 @@ static ENV: Lazy<BotEnv> = Lazy::new(|| {
#[tokio::main] #[tokio::main]
async fn main() { async fn main() {
command::from_json().await;
return;
// Start the tracing subscriber // Start the tracing subscriber
let filter = EnvFilter::builder() let filter = EnvFilter::builder()