riftenlabs-indexer/src/bcmr/bcmrdownloader.rs

450 lines
14 KiB
Rust
Raw Normal View History

// Copyright (C) 2024 Riften Labs AS
//
// This software is licensed under the GNU Affero General Public License (AGPL), version 3.0 or later.
// A copy of the license can be found in the LICENSE file or at https://www.gnu.org/licenses/agpl-3.0.html
use std::{
sync::{atomic::AtomicBool, Arc},
thread::{self, JoinHandle},
time::Duration,
};
use crate::bcmr::parsedbcmr::ParsedBCMR;
use crate::db::bcmr::insert_bcmr_data;
use crate::db::bcmr::update_bcmr_failure;
use crate::{
bcmr::parse_bcmr_from_opreturn,
db::{
bcmr::{get_entries_missing_bcmr_download, AuthChainEntry},
DBPool,
},
};
use anyhow::*;
use bitcoin_hashes::{hex::ToHex, sha256, Hash};
use bitcoincash::{Script, TokenID};
use log::{info, warn};
use rand::thread_rng;
use serde_json::Value;
use ureq::Agent;
use serde_json::json;
use rand::seq::SliceRandom;
use rayon::prelude::*;
use std::result::Result::Ok;
// We are generous on timeout to allow for slow ipfs gateway
const DOWNLOAD_TIMEOUT: Duration = Duration::from_secs(60);
const IPFS_GATEWAY: &str = "https://ipfs.io/ipfs/";
const MAX_BCMR_SIZE: usize = 1024 * 1024 * 100; // 100 MB
const MAX_PARALLEL_DOWNLOADS: usize = 5;
pub struct BCMRDownloader {
db: DBPool,
keep_running: Arc<AtomicBool>,
download_thread: Option<JoinHandle<()>>,
}
const LOOP_SLEEP_TIME: Duration = Duration::from_secs(10);
fn get_download_candidates(db: &DBPool) -> Vec<AuthChainEntry> {
let conn = match db.get() {
Ok(c) => c,
Err(e) => {
warn!("bcmr: Failed to get a db connection: {}", e);
return Vec::default();
}
};
match get_entries_missing_bcmr_download(&conn) {
Ok(c) => c,
Err(e) => {
warn!("bcmr: Failed to fetch bcmr download candidates: {}", e);
Vec::default()
}
}
}
fn get_url(url: &str, max_download_size: usize) -> Result<(String, String)> {
let agent = Agent::new();
let response = agent.get(url).timeout(DOWNLOAD_TIMEOUT).call()?;
if response.status() != 200 {
bail!("Failed to fetch {}: HTTP {}", url, response.status());
}
let mut reader = response.into_reader();
let mut content = Vec::new();
let mut buffer = [0; 1024]; // Read in chunks of 1KB.
while let Ok(count) = reader.read(&mut buffer) {
if count == 0 {
break;
}
if content.len() + count > max_download_size {
bail!("Download exceeded the maximum allowed size");
}
content.extend_from_slice(&buffer[..count]);
}
let content_str = String::from_utf8(content)?;
let hash = sha256::Hash::hash(content_str.as_bytes());
Ok((content_str, hash.to_hex()))
}
fn fetch_bcmr(
urls: &[String],
expected_hash: &str,
) -> (
Option<(String, String)>,
String, /* error */
bool, /* fatal error (don't try again) */
) {
let mut candidate: Option<(String, String)> = None;
let mut errors: Vec<String> = Vec::new();
for url in urls {
let url = if let Some(stripped) = url.strip_prefix("ipfs://") {
format!("{}{}", IPFS_GATEWAY, stripped)
} else if !url.starts_with("https://") {
format!("https://{}", url)
} else {
url.to_string()
};
let (contents, actual_hash) = match get_url(&url, MAX_BCMR_SIZE) {
Ok(c) => c,
Err(e) => {
errors.push(e.to_string());
continue;
}
};
if actual_hash == expected_hash {
return (Some((contents, actual_hash)), String::default(), false);
} else {
candidate = Some((contents, actual_hash))
}
}
// No URL's with matching exepected hash
if candidate.is_some() {
(candidate, String::default(), false)
} else if errors.is_empty() {
(None, "No URLs to fetch BCMR from".to_string(), true)
} else {
(
None,
format!("Errors fetching BCMR: {}", errors.join("; ")),
false,
)
}
}
pub fn parse_bcmr_json(
bcmr: &Value,
token_id: &TokenID,
actual_hash: String,
expected_hash: String,
) -> Result<ParsedBCMR> {
let identities = bcmr
.get("identities")
.context("BCMR missing 'identities'")?;
let token_history = identities
.get(token_id.to_hex())
.context("'identities' does not contain token ID")?;
let first_entry = token_history
.as_object()
.context("token history is not an object")?
.values()
.next()
.context("no entries in token history")?;
let token = first_entry
.get("token")
.context("'token' missing in token history entry")?;
let symbol = token
.get("symbol")
.context("'symbol' missing in 'token' entry")?;
let symbol = symbol.as_str().context("'symbol' field is not a string")?;
// decimals is optional and defaults to 0
let decimals_default = json!(0);
let decimals = token
.get("decimals")
.unwrap_or(&decimals_default)
.to_owned();
let decimals = {
let d_int = if let Some(d) = decimals.as_i64() {
d
} else if let Some(d_str) = decimals.as_str() {
d_str
.parse::<i64>()
.context("Invalid number in 'decimal' field for 'token")?
} else {
bail!("BCMR contains a 'decimal' field for 'token', but it's not a number")
};
if d_int < 0 {
bail!("'decimal' field for 'token' cannot be negative")
}
d_int as usize
};
let empty_string = json!("");
let name = first_entry
.get("name")
.unwrap_or(&empty_string)
.as_str()
.unwrap_or("");
let description = first_entry
.get("description")
.unwrap_or(&empty_string)
.as_str()
.unwrap_or("");
let (icon, web) = if let Some(uris) = first_entry.get("uris") {
let icon = uris
.get("icon")
.unwrap_or(&empty_string)
.as_str()
.unwrap_or("");
let web = uris
.get("web")
.unwrap_or(&empty_string)
.as_str()
.unwrap_or("");
(icon, web)
} else {
("", "")
};
Ok(ParsedBCMR::new(
&token_id.to_hex(),
symbol,
decimals,
name,
description,
icon,
web,
expected_hash,
actual_hash,
))
}
impl BCMRDownloader {
pub fn new(db: DBPool) -> Self {
Self {
db,
keep_running: Arc::new(AtomicBool::new(true)),
download_thread: None,
}
}
pub fn start(&mut self) -> Result<()> {
let db_cpy = self.db.clone();
let keep_running_cpy = self.keep_running.clone();
self.download_thread = Some(
thread::Builder::new()
.name("bcmr downloader".to_string())
.spawn(move || loop {
if !keep_running_cpy.load(std::sync::atomic::Ordering::Relaxed) {
info!("Exiting bcmr download thread");
return;
}
let mut queue = get_download_candidates(&db_cpy);
let mut rng = thread_rng();
queue.shuffle(&mut rng);
if queue.is_empty() {
thread::sleep(LOOP_SLEEP_TIME);
continue;
}
info!("bcmr: {} tokens need BCMR download", queue.len());
let pool = rayon::ThreadPoolBuilder::new()
.num_threads(MAX_PARALLEL_DOWNLOADS)
.build()
.unwrap();
pool.install(|| {
queue.par_iter().for_each(|entry| {
info!(
"bcmr: Dowloading BCMR for token {}",
entry.token_id.to_hex()
);
let bcmr = parse_bcmr_from_opreturn(&Script::from(
entry.bcmr_data.as_ref().expect("bcmr data missing").clone(),
))
.expect("invalid bcmr entry in db");
let (json_str, error, is_fatal) =
fetch_bcmr(&bcmr.uris, &bcmr.hash.to_hex());
let conn = match db_cpy.get() {
Ok(c) => c,
Err(e) => {
warn!("bcmr: Failed to get BCMR db connection: {}", e);
return;
}
};
let (json_str, actual_hash) = match json_str {
Some(j) => j,
None => {
if let Err(e) = update_bcmr_failure(
&conn,
&entry.utxo,
&format!("Failed to fetch BCMR: {}", error),
is_fatal,
) {
warn!("bcmr: Failed to set BCMR error: {}", e)
}
return;
}
};
let json: Value = match serde_json::from_str(&json_str) {
Ok(b) => b,
Err(e) => {
if let Err(e) = update_bcmr_failure(
&conn,
&entry.utxo,
&format!("BCMR invalid JSON error: {}", e),
true,
) {
warn!("bcmr: Failed to set BCMR error: {}", e)
}
return;
}
};
let bcmr_parsed = match parse_bcmr_json(
&json,
&entry.token_id,
actual_hash,
bcmr.hash.to_hex(),
) {
Ok(b) => b,
Err(err) => {
if let Err(e) = update_bcmr_failure(
&conn,
&entry.utxo,
&format!("BCMR contents error: {}", err),
true,
) {
warn!("bcmr: Failed to set BCMR error: {}", e)
}
return;
}
};
if let Err(err) = insert_bcmr_data(&conn, &entry.utxo, &bcmr_parsed) {
warn!("bcmr: Failed to insert BCMR data {}", err)
}
});
});
thread::sleep(LOOP_SLEEP_TIME)
})
.expect("failed to start bcmr download thread"),
);
Ok(())
}
}
impl Drop for BCMRDownloader {
fn drop(&mut self) {
self.keep_running
.store(false, std::sync::atomic::Ordering::SeqCst);
if let Some(thread) = self.download_thread.take() {
// Wake the thread in case it is sleeping
thread.thread().unpark();
match thread.join() {
Ok(_) => info!("bcmr download thread done"),
Err(e) => warn!("Failed to join bcmr download thread: {:?}", e),
}
}
}
}
#[cfg(test)]
mod tests {
use super::*;
use bitcoin_hashes::hex::FromHex;
#[test]
fn test_bcmr_parse() {
let bcmr = r#"
{
"$schema": "https://cashtokens.org/bcmr-v2.schema.json",
"version": {
"major": 0,
"minor": 1,
"patch": 0
},
"latestRevision": "2023-05-15T12:18:32.912Z",
"registryIdentity": {
"name": "Fallout Coin",
"description": "One of the earliest coins ever created on BCH",
"uris": {
"icon": "https://c3-soft.com/tokens/icon.ico",
"web": "https://c3-soft.com/tokens/",
"registry": "https://c3-soft.com/tokens/registry.json"
}
},
"identities": {
"83e12eea20b19a9a0906bb0521ff18520db69a4a8136293bafbfca0acb2c2313": {
"2023-05-12T12:00:00.000Z": {
"name": "Fallout Coin",
"description": "Commemorative Coin for the computer game Fallout",
"token": {
"category": "83e12eea20b19a9a0906bb0521ff18520db69a4a8136293bafbfca0acb2c2313",
"symbol": "FC",
"decimals": 2
},
"uris": {
"icon": "https://c3-soft.com/tokens/fallout.ico"
}
}
}
},
"license": "CC0-1.0"
}
"#;
let dummy_hash = TokenID::all_zeros().to_hex();
let parsed = parse_bcmr_json(
&serde_json::from_str(bcmr).unwrap(),
&TokenID::from_hex("83e12eea20b19a9a0906bb0521ff18520db69a4a8136293bafbfca0acb2c2313")
.unwrap(),
dummy_hash.clone(),
dummy_hash,
)
.unwrap();
assert_eq!("FC", parsed.token.symbol);
assert_eq!(2, parsed.token.decimals);
assert_eq!("Fallout Coin", parsed.name);
assert_eq!(
"Commemorative Coin for the computer game Fallout",
parsed.description
);
assert_eq!(
"https://c3-soft.com/tokens/fallout.ico",
parsed.uris.icon.unwrap()
);
assert_eq!(None, parsed.uris.web);
}
}