refactor: move cli-only utils out of staged-sync (#2984)

bin/reth/src/args/mod.rs

@@ -24,4 +24,6 @@ mod stage_args;
 pub use stage_args::StageEnum;
 
 mod gas_price_oracle_args;
+pub mod utils;
+
 pub use gas_price_oracle_args::GasPriceOracleArgs;

bin/reth/src/args/payload_builder_args.rs

@@ -1,4 +1,4 @@
-use crate::{utils::parse_duration_from_secs, version::default_extradata};
+use crate::{args::utils::parse_duration_from_secs, version::default_extradata};
 use clap::{
     builder::{RangedU64ValueParser, TypedValueParser},
     Arg, Args, Command,

bin/reth/src/args/utils.rs

@@ -0,0 +1,138 @@
+//! Clap parser utilities
+
+use reth_primitives::{AllGenesisFormats, BlockHashOrNumber, ChainSpec, GOERLI, MAINNET, SEPOLIA};
+use reth_revm::primitives::B256 as H256;
+use std::{
+    net::{IpAddr, Ipv4Addr, SocketAddr, ToSocketAddrs},
+    path::PathBuf,
+    str::FromStr,
+    sync::Arc,
+    time::Duration,
+};
+
+/// Helper to parse a [Duration] from seconds
+pub fn parse_duration_from_secs(arg: &str) -> eyre::Result<Duration, std::num::ParseIntError> {
+    let seconds = arg.parse()?;
+    Ok(Duration::from_secs(seconds))
+}
+
+/// Clap value parser for [ChainSpec]s that takes either a built-in chainspec or the path
+/// to a custom one.
+pub fn chain_spec_value_parser(s: &str) -> eyre::Result<Arc<ChainSpec>, eyre::Error> {
+    Ok(Arc::new(match s {
+        "mainnet" => MAINNET.clone(),
+        "goerli" => GOERLI.clone(),
+        "sepolia" => SEPOLIA.clone(),
+        _ => {
+            let raw = std::fs::read_to_string(PathBuf::from(shellexpand::full(s)?.into_owned()))?;
+            serde_json::from_str(&raw)?
+        }
+    }))
+}
+
+/// Clap value parser for [ChainSpec]s that takes either a built-in genesis format or the path
+/// to a custom one.
+pub fn genesis_value_parser(s: &str) -> eyre::Result<Arc<ChainSpec>, eyre::Error> {
+    Ok(Arc::new(match s {
+        "mainnet" => MAINNET.clone(),
+        "goerli" => GOERLI.clone(),
+        "sepolia" => SEPOLIA.clone(),
+        _ => {
+            let raw = std::fs::read_to_string(PathBuf::from(shellexpand::full(s)?.into_owned()))?;
+            let genesis: AllGenesisFormats = serde_json::from_str(&raw)?;
+            genesis.into()
+        }
+    }))
+}
+
+/// Parse [BlockHashOrNumber]
+pub fn hash_or_num_value_parser(value: &str) -> eyre::Result<BlockHashOrNumber, eyre::Error> {
+    match H256::from_str(value) {
+        Ok(hash) => Ok(BlockHashOrNumber::Hash(hash)),
+        Err(_) => Ok(BlockHashOrNumber::Number(value.parse()?)),
+    }
+}
+
+/// Error thrown while parsing a socket address.
+#[derive(thiserror::Error, Debug)]
+pub enum SocketAddressParsingError {
+    /// Failed to convert the string into a socket addr
+    #[error("Cannot parse socket address: {0}")]
+    Io(#[from] std::io::Error),
+    /// Input must not be empty
+    #[error("Cannot parse socket address from empty string")]
+    Empty,
+    /// Failed to parse the address
+    #[error("Could not parse socket address from {0}")]
+    Parse(String),
+    /// Failed to parse port
+    #[error("Could not parse port: {0}")]
+    Port(#[from] std::num::ParseIntError),
+}
+
+/// Parse a [SocketAddr] from a `str`.
+///
+/// The following formats are checked:
+///
+/// - If the value can be parsed as a `u16` or starts with `:` it is considered a port, and the
+/// hostname is set to `localhost`.
+/// - If the value contains `:` it is assumed to be the format `<host>:<port>`
+/// - Otherwise it is assumed to be a hostname
+///
+/// An error is returned if the value is empty.
+pub fn parse_socket_address(value: &str) -> eyre::Result<SocketAddr, SocketAddressParsingError> {
+    if value.is_empty() {
+        return Err(SocketAddressParsingError::Empty)
+    }
+
+    if let Some(port) = value.strip_prefix(':').or_else(|| value.strip_prefix("localhost:")) {
+        let port: u16 = port.parse()?;
+        return Ok(SocketAddr::new(IpAddr::V4(Ipv4Addr::LOCALHOST), port))
+    }
+    if let Ok(port) = value.parse::<u16>() {
+        return Ok(SocketAddr::new(IpAddr::V4(Ipv4Addr::LOCALHOST), port))
+    }
+    value
+        .to_socket_addrs()?
+        .next()
+        .ok_or_else(|| SocketAddressParsingError::Parse(value.to_string()))
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use proptest::prelude::Rng;
+    use secp256k1::rand::thread_rng;
+
+    #[test]
+    fn parse_chain_spec() {
+        for chain in ["mainnet", "sepolia", "goerli"] {
+            chain_spec_value_parser(chain).unwrap();
+            genesis_value_parser(chain).unwrap();
+        }
+    }
+
+    #[test]
+    fn parse_socket_addresses() {
+        for value in ["localhost:9000", ":9000", "9000"] {
+            let socket_addr = parse_socket_address(value)
+                .unwrap_or_else(|_| panic!("could not parse socket address: {value}"));
+
+            assert!(socket_addr.ip().is_loopback());
+            assert_eq!(socket_addr.port(), 9000);
+        }
+    }
+
+    #[test]
+    fn parse_socket_address_random() {
+        let port: u16 = thread_rng().gen();
+
+        for value in [format!("localhost:{port}"), format!(":{port}"), port.to_string()] {
+            let socket_addr = parse_socket_address(&value)
+                .unwrap_or_else(|_| panic!("could not parse socket address: {value}"));
+
+            assert!(socket_addr.ip().is_loopback());
+            assert_eq!(socket_addr.port(), port);
+        }
+    }
+}

bin/reth/src/chain/import.rs

@@ -8,6 +8,7 @@ use eyre::Context;
 use futures::{Stream, StreamExt};
 use reth_beacon_consensus::BeaconConsensus;
 
+use crate::args::utils::genesis_value_parser;
 use reth_config::Config;
 use reth_db::database::Database;
 use reth_downloaders::{
@@ -16,10 +17,7 @@ use reth_downloaders::{
 };
 use reth_interfaces::consensus::Consensus;
 use reth_primitives::{ChainSpec, H256};
-use reth_staged_sync::utils::{
-    chainspec::genesis_value_parser,
-    init::{init_db, init_genesis},
-};
+use reth_staged_sync::utils::init::{init_db, init_genesis};
 use reth_stages::{
     prelude::*,
     stages::{

bin/reth/src/chain/init.rs

@@ -1,10 +1,10 @@
-use crate::dirs::{DataDirPath, MaybePlatformPath};
+use crate::{
+    args::utils::genesis_value_parser,
+    dirs::{DataDirPath, MaybePlatformPath},
+};
 use clap::Parser;
 use reth_primitives::ChainSpec;
-use reth_staged_sync::utils::{
-    chainspec::genesis_value_parser,
-    init::{init_db, init_genesis},
-};
+use reth_staged_sync::utils::init::{init_db, init_genesis};
 use std::sync::Arc;
 use tracing::info;
 

bin/reth/src/db/mod.rs

@@ -1,5 +1,6 @@
 //! Database debugging tool
 use crate::{
+    args::utils::genesis_value_parser,
     dirs::{DataDirPath, MaybePlatformPath},
     utils::DbTool,
 };
@@ -9,7 +10,6 @@ use eyre::WrapErr;
 use human_bytes::human_bytes;
 use reth_db::{database::Database, tables};
 use reth_primitives::ChainSpec;
-use reth_staged_sync::utils::chainspec::genesis_value_parser;
 use std::sync::Arc;
 use tracing::error;
 

bin/reth/src/debug_cmd/execution.rs

@@ -27,10 +27,9 @@ use reth_network::NetworkHandle;
 use reth_network_api::NetworkInfo;
 use reth_primitives::{stage::StageId, BlockHashOrNumber, BlockNumber, ChainSpec, H256};
 use reth_provider::{providers::get_stage_checkpoint, ShareableDatabase, Transaction};
-use reth_staged_sync::utils::{
-    chainspec::genesis_value_parser,
-    init::{init_db, init_genesis},
-};
+use reth_staged_sync::utils::init::{init_db, init_genesis};
+
+use crate::args::utils::genesis_value_parser;
 use reth_stages::{
     sets::DefaultStages,
     stages::{

bin/reth/src/debug_cmd/merkle.rs

@@ -1,5 +1,8 @@
 //! Command for debugging merkle trie calculation.
-use crate::dirs::{DataDirPath, MaybePlatformPath};
+use crate::{
+    args::utils::genesis_value_parser,
+    dirs::{DataDirPath, MaybePlatformPath},
+};
 use clap::Parser;
 use reth_db::{cursor::DbCursorRO, tables, transaction::DbTx};
 use reth_primitives::{
@@ -7,7 +10,7 @@ use reth_primitives::{
     ChainSpec,
 };
 use reth_provider::Transaction;
-use reth_staged_sync::utils::{chainspec::genesis_value_parser, init::init_db};
+use reth_staged_sync::utils::init::init_db;
 use reth_stages::{
     stages::{
         AccountHashingStage, ExecutionStage, ExecutionStageThresholds, MerkleStage,

bin/reth/src/dirs.rs

@@ -1,6 +1,6 @@
 //! reth data directories.
+use crate::utils::parse_path;
 use reth_primitives::Chain;
-use reth_staged_sync::utils::parse_path;
 use std::{
     env::VarError,
     fmt::{Debug, Display, Formatter},

bin/reth/src/node/mod.rs

@@ -51,11 +51,7 @@ use reth_provider::{
 use reth_revm::Factory;
 use reth_revm_inspectors::stack::Hook;
 use reth_rpc_engine_api::EngineApi;
-use reth_staged_sync::utils::{
-    chainspec::genesis_value_parser,
-    init::{init_db, init_genesis},
-    parse_socket_address,
-};
+use reth_staged_sync::utils::init::{init_db, init_genesis};
 use reth_stages::{
     prelude::*,
     stages::{
@@ -74,7 +70,13 @@ use std::{
 use tokio::sync::{mpsc::unbounded_channel, oneshot, watch};
 use tracing::*;
 
-use crate::{args::PayloadBuilderArgs, dirs::MaybePlatformPath};
+use crate::{
+    args::{
+        utils::{genesis_value_parser, parse_socket_address},
+        PayloadBuilderArgs,
+    },
+    dirs::MaybePlatformPath,
+};
 use reth_interfaces::p2p::headers::client::HeadersClient;
 use reth_payload_builder::PayloadBuilderService;
 use reth_primitives::bytes::BytesMut;

bin/reth/src/p2p/mod.rs

@@ -1,6 +1,10 @@
 //! P2P Debugging tool
 use crate::{
-    args::{get_secret_key, DiscoveryArgs},
+    args::{
+        get_secret_key,
+        utils::{chain_spec_value_parser, hash_or_num_value_parser},
+        DiscoveryArgs,
+    },
     dirs::{DataDirPath, MaybePlatformPath},
     utils::get_single_header,
 };
@@ -12,7 +16,6 @@ use reth_discv4::NatResolver;
 use reth_interfaces::p2p::bodies::client::BodiesClient;
 use reth_primitives::{BlockHashOrNumber, ChainSpec, NodeRecord};
 use reth_provider::ShareableDatabase;
-use reth_staged_sync::utils::{chainspec::chain_spec_value_parser, hash_or_num_value_parser};
 use std::{path::PathBuf, sync::Arc};
 
 /// `reth p2p` command

bin/reth/src/stage/drop.rs

@@ -1,6 +1,6 @@
 //! Database debugging tool
 use crate::{
-    args::StageEnum,
+    args::{utils::genesis_value_parser, StageEnum},
     dirs::{DataDirPath, MaybePlatformPath},
     utils::DbTool,
 };
@@ -12,7 +12,7 @@ use reth_db::{
     transaction::DbTxMut,
 };
 use reth_primitives::{stage::StageId, ChainSpec};
-use reth_staged_sync::utils::{chainspec::genesis_value_parser, init::insert_genesis_state};
+use reth_staged_sync::utils::init::insert_genesis_state;
 use std::sync::Arc;
 use tracing::info;
 

bin/reth/src/stage/dump/mod.rs

@@ -8,7 +8,7 @@ use reth_db::{
     cursor::DbCursorRO, database::Database, table::TableImporter, tables, transaction::DbTx,
 };
 use reth_primitives::ChainSpec;
-use reth_staged_sync::utils::{chainspec::genesis_value_parser, init::init_db};
+use reth_staged_sync::utils::init::init_db;
 use std::{path::PathBuf, sync::Arc};
 use tracing::info;
 
@@ -22,6 +22,7 @@ mod execution;
 use execution::dump_execution_stage;
 
 mod merkle;
+use crate::args::utils::genesis_value_parser;
 use merkle::dump_merkle_stage;
 
 /// `reth dump-stage` command

bin/reth/src/stage/run.rs

@@ -2,7 +2,7 @@
 //!
 //! Stage debugging tool
 use crate::{
-    args::{get_secret_key, NetworkArgs, StageEnum},
+    args::{get_secret_key, utils::chain_spec_value_parser, NetworkArgs, StageEnum},
     dirs::{DataDirPath, MaybePlatformPath},
     prometheus_exporter,
     version::SHORT_VERSION,
@@ -16,7 +16,7 @@ use reth_primitives::{
     ChainSpec,
 };
 use reth_provider::{ShareableDatabase, Transaction};
-use reth_staged_sync::utils::{chainspec::chain_spec_value_parser, init::init_db};
+use reth_staged_sync::utils::init::init_db;
 use reth_stages::{
     stages::{
         BodyStage, ExecutionStage, ExecutionStageThresholds, IndexAccountHistoryStage,

bin/reth/src/stage/unwind.rs

@@ -10,9 +10,9 @@ use reth_db::{
 };
 use reth_primitives::{BlockHashOrNumber, ChainSpec};
 use reth_provider::Transaction;
-use reth_staged_sync::utils::chainspec::genesis_value_parser;
 use std::{ops::RangeInclusive, sync::Arc};
 
+use crate::args::utils::genesis_value_parser;
 use reth_db::cursor::DbCursorRO;
 
 /// `reth stage unwind` command

bin/reth/src/utils.rs

@@ -12,14 +12,17 @@ use reth_interfaces::p2p::{
     priority::Priority,
 };
 use reth_primitives::{BlockHashOrNumber, HeadersDirection, SealedHeader};
-use std::{path::Path, time::Duration};
+use std::{
+    env::VarError,
+    path::{Path, PathBuf},
+};
 use tracing::info;
 
 /// Get a single header from network
 pub async fn get_single_header<Client>(
     client: Client,
     id: BlockHashOrNumber,
-) -> eyre::Result<SealedHeader>
+) -> Result<SealedHeader>
 where
     Client: HeadersClient,
 {
@@ -104,8 +107,8 @@ impl<'a, DB: Database> DbTool<'a, DB> {
     }
 }
 
-/// Helper to parse a [Duration] from seconds
-pub fn parse_duration_from_secs(arg: &str) -> Result<Duration, std::num::ParseIntError> {
-    let seconds = arg.parse()?;
-    Ok(Duration::from_secs(seconds))
+/// Parses a user-specified path with support for environment variables and common shorthands (e.g.
+/// ~ for the user's home directory).
+pub fn parse_path(value: &str) -> Result<PathBuf, shellexpand::LookupError<VarError>> {
+    shellexpand::full(value).map(|path| PathBuf::from(path.into_owned()))
 }