feat: background downloading ⚡

Cargo.toml

@@ -51,6 +51,8 @@ paracas-instruments = { path = "crates/paracas-instruments", version = "0.1.0" }
 paracas-fetch = { path = "crates/paracas-fetch", version = "0.1.0" }
 paracas-aggregate = { path = "crates/paracas-aggregate", version = "0.1.0" }
 paracas-format = { path = "crates/paracas-format", version = "0.1.0" }
+paracas-estimate = { path = "crates/paracas-estimate", version = "0.1.0" }
+paracas-daemon = { path = "crates/paracas-daemon", version = "0.1.0" }
 
 # Async runtime
 tokio = { version = "1.42", features = ["full"] }
@@ -97,3 +99,9 @@ approx = "0.5"
 criterion = { version = "0.5", features = ["async_tokio", "html_reports"] }
 tempfile = "3.14"
 which = "7.0"
+
+# UUID
+uuid = { version = "1.0", features = ["v4", "serde"] }
+
+# Directories
+directories = "5.0"

bin/Cargo.toml

@@ -25,6 +25,8 @@ parquet = ["paracas-lib/parquet"]
 
 [dependencies]
 paracas-lib = { workspace = true }
+paracas-daemon = { workspace = true }
+paracas-estimate = { workspace = true }
 tokio = { workspace = true }
 futures = { workspace = true }
 clap = { workspace = true }

bin/README.md

@@ -1,11 +1,11 @@
-# paracas-cli
+# paracas
 
 Command-line interface for downloading Dukascopy tick data.
 
 ## Installation
 
 ```bash
-cargo install paracas-cli
+cargo install paracas
 ```
 
 ## Commands
@@ -16,10 +16,10 @@ Download tick data for an instrument:
 
 ```bash
 # Download EUR/USD ticks as CSV
-paracas download -i eurusd -s 2024-01-01 -e 2024-01-31 -o data.csv
+paracas download eurusd -s 2024-01-01 -e 2024-01-31 -o data.csv
 
 # Download as Parquet with 1-hour aggregation
-paracas download -i btcusd -s 2024-01-01 -e 2024-12-31 -o data.parquet -f parquet -t h1
+paracas download btcusd -s 2024-01-01 -e 2024-12-31 -o data.parquet -f parquet -t h1
 ```
 
 ### List

bin/src/commands/daemon_run.rs

@@ -0,0 +1,159 @@
+//! Hidden daemon entry point for background downloads.
+//!
+//! This module provides the entry point for daemon processes spawned
+//! with `--daemon-run <job_id>`. It loads the job from disk and executes
+//! the download tasks.
+
+use crate::display::{Format, aggregate_ticks, write_ohlcv, write_ticks};
+use anyhow::{Context, Result, bail};
+use futures::StreamExt;
+use paracas_daemon::{DaemonProgress, JobId, JobStatus, StateManager};
+use paracas_lib::prelude::*;
+use std::path::PathBuf;
+
+/// Execute a background download job.
+///
+/// This is called when paracas is spawned with `--daemon-run <job_id>`.
+/// The function loads the job from disk, executes all pending tasks,
+/// and saves progress periodically.
+pub(crate) async fn daemon_run(job_id_str: &str) -> Result<()> {
+    let job_id: JobId = job_id_str.parse().context("Invalid job ID")?;
+
+    let state_manager =
+        StateManager::with_default_path().context("Failed to initialize state manager")?;
+
+    let job = state_manager.load_job(job_id).context("Job not found")?;
+
+    if !matches!(job.status, JobStatus::Pending | JobStatus::Running) {
+        bail!("Job is not in a runnable state: {:?}", job.status);
+    }
+
+    let progress = DaemonProgress::new(state_manager.clone(), job);
+
+    // Mark job as running
+    {
+        let mut job = progress.job().await;
+        job.mark_started(std::process::id());
+        state_manager.save_job(&job)?;
+    }
+
+    // Process each task
+    let job = progress.job().await;
+    for (task_idx, task) in job.tasks.iter().enumerate() {
+        if matches!(task.status, JobStatus::Completed) {
+            continue; // Skip already completed tasks
+        }
+
+        if let Err(e) = execute_task(&progress, task_idx).await {
+            progress.mark_task_failed(task_idx, &e.to_string()).await;
+        }
+
+        progress.save_checkpoint().await?;
+    }
+
+    // Mark job as completed or failed based on task results
+    if progress.all_tasks_finished().await {
+        if progress.failed_tasks().await == 0 {
+            progress.mark_job_completed().await;
+        } else {
+            let failed_count = progress.failed_tasks().await;
+            let msg = format!("{} tasks failed", failed_count);
+            progress.mark_job_failed(&msg).await;
+        }
+    }
+
+    progress.save_checkpoint().await?;
+
+    Ok(())
+}
+
+/// Execute a single download task.
+async fn execute_task(progress: &DaemonProgress, task_idx: usize) -> Result<()> {
+    progress.mark_task_running(task_idx).await;
+
+    let job = progress.job().await;
+    let task = &job.tasks[task_idx];
+
+    // Get instrument
+    let registry = InstrumentRegistry::global();
+    let instrument = registry
+        .get(&task.instrument_id)
+        .context("Unknown instrument")?;
+
+    // Parse date range
+    let start = chrono::NaiveDate::parse_from_str(&task.start_date, "%Y-%m-%d")?;
+    let end = chrono::NaiveDate::parse_from_str(&task.end_date, "%Y-%m-%d")?;
+    let range = DateRange::new(start, end)?;
+
+    // Create client
+    let config = ClientConfig {
+        concurrency: job.concurrency,
+        ..Default::default()
+    };
+    let client = DownloadClient::new(config)?;
+
+    // Download ticks
+    let mut all_ticks: Vec<Tick> = Vec::new();
+    let mut stream = paracas_lib::tick_stream_resilient(&client, instrument, range);
+    let mut hours_completed = 0u64;
+
+    while let Some(batch) = stream.next().await {
+        all_ticks.extend(batch.ticks);
+        hours_completed += 1;
+
+        // Update progress periodically (every 10 hours)
+        if hours_completed.is_multiple_of(10) {
+            progress
+                .update_task_progress(task_idx, hours_completed, all_ticks.len() as u64)
+                .await;
+        }
+    }
+
+    // Parse timeframe and aggregate if needed
+    let timeframe = task
+        .timeframe
+        .parse::<Timeframe>()
+        .map_err(|e| anyhow::anyhow!("{e}"))?;
+
+    // Parse format
+    let format = parse_format(&task.format)?;
+
+    // Write output
+    let output_path = task.output_path.clone();
+    write_output(&all_ticks, &output_path, format, timeframe)?;
+
+    let bytes_written = std::fs::metadata(&output_path)
+        .map(|m| m.len())
+        .unwrap_or(0);
+
+    progress.mark_task_completed(task_idx, bytes_written).await;
+
+    Ok(())
+}
+
+/// Parse a format string into a Format enum.
+fn parse_format(format: &str) -> Result<Format> {
+    match format.to_lowercase().as_str() {
+        "csv" => Ok(Format::Csv),
+        "json" => Ok(Format::Json),
+        "ndjson" => Ok(Format::Ndjson),
+        "parquet" => Ok(Format::Parquet),
+        _ => bail!("Unknown format: {}", format),
+    }
+}
+
+/// Write ticks or OHLCV data to the output file.
+fn write_output(
+    ticks: &[Tick],
+    output: &PathBuf,
+    format: Format,
+    timeframe: Timeframe,
+) -> Result<()> {
+    if timeframe.is_tick() {
+        write_ticks(ticks, output, format)?;
+    } else {
+        let bars = aggregate_ticks(ticks, timeframe);
+        write_ohlcv(&bars, output, format)?;
+    }
+    Ok(())
+}

bin/src/commands/download.rs

@@ -0,0 +1,221 @@
+//! Download command implementation.
+//!
+//! This module handles downloading tick data from Dukascopy and writing it to various output formats.
+
+use crate::display::{Format, aggregate_ticks, write_ohlcv, write_ticks};
+use anyhow::{Context, Result};
+use chrono::NaiveDate;
+use futures::StreamExt;
+use indicatif::{ProgressBar, ProgressStyle};
+use paracas_daemon::{DaemonSpawner, DownloadJob, InstrumentTask, StateManager};
+use paracas_lib::prelude::*;
+use std::path::PathBuf;
+
+/// Download tick data for an instrument.
+#[allow(clippy::too_many_arguments)]
+pub(crate) async fn download(
+    instrument_id: &str,
+    start_str: Option<&str>,
+    end_str: Option<&str>,
+    output: Option<PathBuf>,
+    format: Format,
+    timeframe_str: Option<&str>,
+    concurrency: usize,
+    background: bool,
+    _yes: bool,
+    quiet: bool,
+) -> Result<()> {
+    // Handle background mode
+    if background {
+        return spawn_background_download(
+            instrument_id,
+            start_str,
+            end_str,
+            output,
+            format,
+            timeframe_str,
+            concurrency,
+        );
+    }
+
+    // Lookup instrument
+    let registry = InstrumentRegistry::global();
+    let instrument = registry
+        .get(instrument_id)
+        .with_context(|| format!("Unknown instrument: {instrument_id}"))?;
+
+    // Parse start date (default to instrument's earliest available data)
+    let start = match start_str {
+        Some(s) => NaiveDate::parse_from_str(s, "%Y-%m-%d")
+            .with_context(|| format!("Invalid start date: {s}"))?,
+        None => instrument
+            .start_tick_date()
+            .map(|dt| dt.date_naive())
+            .unwrap_or_else(|| NaiveDate::from_ymd_opt(2003, 5, 5).expect("valid date")),
+    };
+
+    // Parse end date (default to today)
+    let end = match end_str {
+        Some(s) => NaiveDate::parse_from_str(s, "%Y-%m-%d")
+            .with_context(|| format!("Invalid end date: {s}"))?,
+        None => chrono::Utc::now().date_naive(),
+    };
+
+    let range = DateRange::new(start, end)?;
+
+    // Determine output path (default to <instrument>.<format>)
+    let output = output
+        .unwrap_or_else(|| PathBuf::from(format!("{}.{}", instrument_id, format.extension())));
+
+    // Parse timeframe
+    let timeframe = match timeframe_str {
+        Some(tf) => tf
+            .parse::<Timeframe>()
+            .map_err(|e| anyhow::anyhow!("{e}"))?,
+        None => Timeframe::Tick,
+    };
+
+    // Create client
+    let config = ClientConfig {
+        concurrency,
+        ..Default::default()
+    };
+    let client = DownloadClient::new(config)?;
+
+    // Setup progress bar
+    let total_hours = range.total_hours() as u64;
+    let progress = if quiet {
+        ProgressBar::hidden()
+    } else {
+        let pb = ProgressBar::new(total_hours);
+        pb.set_style(
+            ProgressStyle::default_bar()
+                .template("{spinner:.green} [{elapsed_precise}] [{bar:40.cyan/blue}] {pos}/{len} hours ({percent}%) {msg}")
+                .expect("Invalid progress template")
+                .progress_chars("=>-"),
+        );
+        pb.set_message(format!("{} {} -> {}", instrument.id(), start, end));
+        pb
+    };
+
+    // Download and collect ticks using the resilient stream
+    // This will retry on transient errors and skip hours that fail after retries
+    let mut all_ticks: Vec<Tick> = Vec::new();
+    let mut skipped_hours = 0u64;
+    let mut stream = paracas_lib::tick_stream_resilient(&client, instrument, range);
+
+    while let Some(batch) = stream.next().await {
+        if batch.had_error() {
+            skipped_hours += 1;
+        }
+        all_ticks.extend(batch.ticks);
+        progress.inc(1);
+    }
+
+    let finish_msg = if skipped_hours > 0 {
+        format!(
+            "Downloaded {} ticks ({} hours skipped due to errors)",
+            all_ticks.len(),
+            skipped_hours
+        )
+    } else {
+        format!("Downloaded {} ticks", all_ticks.len())
+    };
+    progress.finish_with_message(finish_msg);
+
+    // Aggregate if needed
+    if timeframe.is_tick() {
+        // Write raw ticks
+        write_ticks(&all_ticks, &output, format)?;
+    } else {
+        // Aggregate to OHLCV
+        let bars = aggregate_ticks(&all_ticks, timeframe);
+        write_ohlcv(&bars, &output, format)?;
+    }
+
+    if !quiet {
+        println!("Output written to: {}", output.display());
+    }
+
+    Ok(())
+}
+
+/// Spawn a background download job for a single instrument.
+#[allow(clippy::too_many_arguments)]
+fn spawn_background_download(
+    instrument_id: &str,
+    start_str: Option<&str>,
+    end_str: Option<&str>,
+    output: Option<PathBuf>,
+    format: Format,
+    timeframe_str: Option<&str>,
+    concurrency: usize,
+) -> Result<()> {
+    let registry = InstrumentRegistry::global();
+    let instrument = registry
+        .get(instrument_id)
+        .with_context(|| format!("Unknown instrument: {instrument_id}"))?;
+
+    // Determine start date
+    let start = start_str
+        .map(|s| s.to_string())
+        .or_else(|| {
+            instrument
+                .start_tick_date()
+                .map(|d| d.format("%Y-%m-%d").to_string())
+        })
+        .unwrap_or_else(|| "2003-05-05".to_string());
+
+    // Determine end date
+    let end = end_str
+        .map(|s| s.to_string())
+        .unwrap_or_else(|| chrono::Utc::now().format("%Y-%m-%d").to_string());
+
+    // Determine output path
+    let output_path = output
+        .unwrap_or_else(|| PathBuf::from(format!("{}.{}", instrument_id, format.extension())));
+
+    // Make output path absolute
+    let output_path = if output_path.is_absolute() {
+        output_path
+    } else {
+        std::env::current_dir()
+            .unwrap_or_else(|_| PathBuf::from("."))
+            .join(output_path)
+    };
+
+    // Calculate total hours for progress tracking
+    let start_date = NaiveDate::parse_from_str(&start, "%Y-%m-%d")?;
+    let end_date = NaiveDate::parse_from_str(&end, "%Y-%m-%d")?;
+    let range = DateRange::new(start_date, end_date)?;
+
+    // Determine timeframe string (default to "tick")
+    let timeframe = timeframe_str
+        .map(|s| s.to_string())
+        .unwrap_or_else(|| "tick".to_string());
+
+    let task = InstrumentTask::new(
+        instrument_id.to_string(),
+        start,
+        end,
+        output_path,
+        format.to_string(),
+        timeframe,
+        range.total_hours() as u32,
+    );
+
+    let mut job = DownloadJob::new(vec![task], concurrency);
+
+    let state_manager =
+        StateManager::with_default_path().context("Failed to initialize state manager")?;
+    let spawner = DaemonSpawner::new(state_manager).context("Failed to create daemon spawner")?;
+    let job_id = spawner
+        .spawn(&mut job)
+        .context("Failed to spawn background job")?;
+
+    println!("Background download started.");
+    println!("Job ID: {}", job_id);
+    println!("Check status with: paracas status {}", job_id);
+
+    Ok(())
+}

bin/src/commands/download_all.rs

@@ -0,0 +1,351 @@
+//! Download all instruments command.
+//!
+//! This module handles batch downloading of multiple instruments, with support for
+//! category filtering, parallel downloads, and download estimation.
+
+use crate::display::{Format, aggregate_ticks, parse_category, write_ohlcv, write_ticks};
+use anyhow::{Context, Result};
+use chrono::NaiveDate;
+use futures::stream::{self, StreamExt};
+use indicatif::{MultiProgress, ProgressBar, ProgressStyle};
+use paracas_daemon::{DaemonSpawner, DownloadJob, InstrumentTask, StateManager};
+use paracas_estimate::Estimator;
+use paracas_lib::prelude::*;
+use std::io::Write as _;
+use std::path::PathBuf;
+
+/// Execute the download-all command.
+#[allow(clippy::too_many_arguments)]
+pub(crate) async fn download_all(
+    category: Option<&str>,
+    start_str: Option<&str>,
+    end_str: Option<&str>,
+    output_dir: PathBuf,
+    format: Format,
+    timeframe_str: Option<&str>,
+    parallel_instruments: usize,
+    concurrency: usize,
+    background: bool,
+    yes: bool,
+    quiet: bool,
+) -> Result<()> {
+    // 1. Get instruments based on category filter (or all)
+    let registry = InstrumentRegistry::global();
+    let instruments: Vec<_> = match category {
+        Some(cat) => {
+            let category = parse_category(cat)?;
+            registry.by_category(category).collect()
+        }
+        None => registry.all().collect(),
+    };
+
+    if instruments.is_empty() {
+        anyhow::bail!("No instruments found matching criteria");
+    }
+
+    // Parse end date (default to today)
+    let today = chrono::Utc::now().date_naive();
+    let end = match end_str {
+        Some(s) => NaiveDate::parse_from_str(s, "%Y-%m-%d")
+            .with_context(|| format!("Invalid end date: {s}"))?,
+        None => today,
+    };
+
+    // Parse start date or use earliest instrument date
+    let start = match start_str {
+        Some(s) => NaiveDate::parse_from_str(s, "%Y-%m-%d")
+            .with_context(|| format!("Invalid start date: {s}"))?,
+        None => {
+            // Use the earliest start date among all selected instruments
+            instruments
+                .iter()
+                .filter_map(|i| i.start_tick_date())
+                .map(|dt| dt.date_naive())
+                .min()
+                .unwrap_or_else(|| NaiveDate::from_ymd_opt(2003, 5, 5).expect("valid date"))
+        }
+    };
+
+    let range = DateRange::new(start, end)?;
+
+    // 2. Show estimate and get confirmation
+    let estimator = Estimator::global();
+    let estimate = estimator.estimate_batch(&instruments, &range);
+
+    if !yes && !quiet {
+        println!("Download plan:");
+        println!("  Instruments: {}", instruments.len());
+        println!("  Date range: {} to {}", start, end);
+        println!(
+            "  Estimated download size: {}",
+            Estimator::format_bytes(estimate.estimated_compressed_bytes)
+        );
+        println!(
+            "  Estimated output size: {}",
+            Estimator::format_bytes(estimate.estimated_output_bytes)
+        );
+        println!(
+            "  Estimated time: {}",
+            Estimator::format_duration(estimate.estimated_duration)
+        );
+        println!();
+
+        // Simple y/n confirmation
+        print!("Proceed with download? [y/N] ");
+        std::io::stdout().flush()?;
+        let mut input = String::new();
+        std::io::stdin().read_line(&mut input)?;
+        if !input.trim().eq_ignore_ascii_case("y") {
+            println!("Cancelled.");
+            return Ok(());
+        }
+    }
+
+    // 3. If background mode, spawn daemon
+    if background {
+        return spawn_background_download_all(
+            &instruments,
+            start,
+            end,
+            &output_dir,
+            format,
+            timeframe_str,
+            concurrency,
+        );
+    }
+
+    // 4. Create output directory if needed
+    std::fs::create_dir_all(&output_dir)?;
+
+    // 5. Parse timeframe
+    let timeframe = match timeframe_str {
+        Some(tf) => tf
+            .parse::<Timeframe>()
+            .map_err(|e| anyhow::anyhow!("{e}"))?,
+        None => Timeframe::Tick,
+    };
+
+    // 6. Download instruments in parallel
+    let multi_progress = MultiProgress::new();
+
+    let results: Vec<_> = stream::iter(instruments.into_iter())
+        .map(|instrument| {
+            let pb = multi_progress.add(ProgressBar::new(100));
+            pb.set_style(
+                ProgressStyle::default_bar()
+                    .template("{prefix:.bold} [{bar:30.cyan/blue}] {percent}% {msg}")
+                    .unwrap()
+                    .progress_chars("=>-"),
+            );
+            pb.set_prefix(format!("{:>12}", instrument.id()));
+
+            download_single_instrument(
+                instrument,
+                start,
+                end,
+                output_dir.clone(),
+                format,
+                timeframe,
+                concurrency,
+                pb,
+                quiet,
+            )
+        })
+        .buffer_unordered(parallel_instruments)
+        .collect()
+        .await;
+
+    // 7. Report summary
+    let (successes, failures): (Vec<_>, Vec<_>) = results.iter().partition(|r| r.is_ok());
+
+    if !quiet {
+        println!("\nDownload complete:");
+        println!("  Successful: {}", successes.len());
+        if !failures.is_empty() {
+            println!("  Failed: {}", failures.len());
+            for (i, err) in failures.iter().enumerate() {
+                if let Err(e) = err {
+                    println!("    {}: {}", i + 1, e);
+                }
+            }
+        }
+    }
+
+    // Return error if any downloads failed
+    if !failures.is_empty() {
+        anyhow::bail!(
+            "{} out of {} downloads failed",
+            failures.len(),
+            successes.len() + failures.len()
+        );
+    }
+
+    Ok(())
+}
+
+/// Download a single instrument with progress tracking.
+#[allow(clippy::too_many_arguments)]
+async fn download_single_instrument(
+    instrument: &Instrument,
+    start: NaiveDate,
+    end: NaiveDate,
+    output_dir: PathBuf,
+    format: Format,
+    timeframe: Timeframe,
+    concurrency: usize,
+    progress: ProgressBar,
+    quiet: bool,
+) -> Result<()> {
+    // Adjust start date based on instrument's available data
+    let effective_start = instrument
+        .start_tick_date()
+        .map_or(start, |instrument_start| {
+            let instrument_start_date = instrument_start.date_naive();
+            if start < instrument_start_date {
+                instrument_start_date
+            } else {
+                start
+            }
+        });
+
+    // Skip if the instrument has no data in the requested range
+    if effective_start > end {
+        progress.finish_with_message("skipped (no data)");
+        return Ok(());
+    }
+
+    let range = DateRange::new(effective_start, end)?;
+    let total_hours = range.total_hours() as u64;
+    progress.set_length(total_hours);
+
+    // Create client
+    let config = ClientConfig {
+        concurrency,
+        ..Default::default()
+    };
+    let client = DownloadClient::new(config)?;
+
+    // Download and collect ticks
+    let mut all_ticks: Vec<Tick> = Vec::new();
+    let mut skipped_hours = 0u64;
+    let mut stream = paracas_lib::tick_stream_resilient(&client, instrument, range);
+
+    while let Some(batch) = stream.next().await {
+        if batch.had_error() {
+            skipped_hours += 1;
+        }
+        all_ticks.extend(batch.ticks);
+        progress.inc(1);
+    }
+
+    let tick_count = all_ticks.len();
+    let finish_msg = if skipped_hours > 0 {
+        format!("{} ticks ({} hrs skipped)", tick_count, skipped_hours)
+    } else {
+        format!("{} ticks", tick_count)
+    };
+    progress.finish_with_message(finish_msg);
+
+    // Determine output path
+    let output_path = output_dir.join(format!("{}.{}", instrument.id(), format.extension()));
+
+    // Aggregate if needed
+    if timeframe.is_tick() {
+        write_ticks(&all_ticks, &output_path, format)?;
+    } else {
+        let bars = aggregate_ticks(&all_ticks, timeframe);
+        write_ohlcv(&bars, &output_path, format)?;
+    }
+
+    if !quiet {
+        progress.println(format!("  Written: {}", output_path.display()));
+    }
+
+    Ok(())
+}
+
+/// Spawn a background download job for multiple instruments.
+#[allow(clippy::too_many_arguments)]
+fn spawn_background_download_all(
+    instruments: &[&Instrument],
+    start: NaiveDate,
+    end: NaiveDate,
+    output_dir: &PathBuf,
+    format: Format,
+    timeframe_str: Option<&str>,
+    concurrency: usize,
+) -> Result<()> {
+    // Make output directory absolute
+    let output_dir = if output_dir.is_absolute() {
+        output_dir.clone()
+    } else {
+        std::env::current_dir()
+            .unwrap_or_else(|_| PathBuf::from("."))
+            .join(output_dir)
+    };
+
+    // Create output directory if needed
+    std::fs::create_dir_all(&output_dir)?;
+
+    // Determine timeframe string (default to "tick")
+    let timeframe = timeframe_str
+        .map(|s| s.to_string())
+        .unwrap_or_else(|| "tick".to_string());
+
+    // Create tasks for each instrument
+    let mut tasks = Vec::with_capacity(instruments.len());
+
+    for instrument in instruments {
+        // Adjust start date based on instrument's available data
+        let effective_start = instrument
+            .start_tick_date()
+            .map_or(start, |instrument_start| {
+                let instrument_start_date = instrument_start.date_naive();
+                if start < instrument_start_date {
+                    instrument_start_date
+                } else {
+                    start
+                }
+            });
+
+        // Skip if the instrument has no data in the requested range
+        if effective_start > end {
+            continue;
+        }
+
+        let range = DateRange::new(effective_start, end)?;
+        let output_path = output_dir.join(format!("{}.{}", instrument.id(), format.extension()));
+
+        let task = InstrumentTask::new(
+            instrument.id().to_string(),
+            effective_start.format("%Y-%m-%d").to_string(),
+            end.format("%Y-%m-%d").to_string(),
+            output_path,
+            format.to_string(),
+            timeframe.clone(),
+            range.total_hours() as u32,
+        );
+
+        tasks.push(task);
+    }
+
+    if tasks.is_empty() {
+        anyhow::bail!("No instruments with data in the specified date range");
+    }
+
+    let mut job = DownloadJob::new(tasks, concurrency);
+
+    let state_manager =
+        StateManager::with_default_path().context("Failed to initialize state manager")?;
+    let spawner = DaemonSpawner::new(state_manager).context("Failed to create daemon spawner")?;
+    let job_id = spawner
+        .spawn(&mut job)
+        .context("Failed to spawn background job")?;
+
+    println!("Background download started.");
+    println!("Job ID: {}", job_id);
+    println!("Instruments: {}", job.tasks.len());
+    println!("Check status with: paracas status {}", job_id);
+
+    Ok(())
+}

bin/src/commands/info.rs

@@ -0,0 +1,104 @@
+//! Info command implementation.
+//!
+//! This module handles displaying detailed information about a specific instrument,
+//! including size estimates for different time periods.
+
+use anyhow::{Context, Result};
+use paracas_estimate::Estimator;
+use paracas_lib::prelude::*;
+
+/// Show detailed information about an instrument, including size estimates.
+pub(crate) fn show_info(instrument_id: &str) -> Result<()> {
+    let registry = InstrumentRegistry::global();
+    let instrument = registry
+        .get(instrument_id)
+        .with_context(|| format!("Unknown instrument: {instrument_id}"))?;
+
+    // Basic info
+    println!("Instrument: {}", instrument.name());
+    println!("ID:         {}", instrument.id());
+    println!("Category:   {}", instrument.category());
+    println!("Description: {}", instrument.description());
+    println!("Decimal Factor: {}", instrument.decimal_factor());
+
+    if let Some(start) = instrument.start_tick_date() {
+        println!("Data Available From: {}", start.format("%Y-%m-%d"));
+
+        // Calculate estimates for different time periods
+        let today = chrono::Utc::now().date_naive();
+        let estimator = Estimator::global();
+
+        println!("\nDownload Estimates:");
+        println!(
+            "{:<20} {:>12} {:>12} {:>12}",
+            "PERIOD", "DOWNLOAD", "OUTPUT (CSV)", "EST. TIME"
+        );
+        println!("{}", "-".repeat(60));
+
+        // Last 1 day
+        if let Ok(range) = DateRange::new(today - chrono::Duration::days(1), today) {
+            let est = estimator.estimate_single(instrument, &range);
+            println!(
+                "{:<20} {:>12} {:>12} {:>12}",
+                "Last 1 day",
+                Estimator::format_bytes(est.estimated_compressed_bytes),
+                Estimator::format_bytes(est.estimated_output_bytes),
+                Estimator::format_duration(est.estimated_duration),
+            );
+        }
+
+        // Last 1 week
+        if let Ok(range) = DateRange::new(today - chrono::Duration::days(7), today) {
+            let est = estimator.estimate_single(instrument, &range);
+            println!(
+                "{:<20} {:>12} {:>12} {:>12}",
+                "Last 1 week",
+                Estimator::format_bytes(est.estimated_compressed_bytes),
+                Estimator::format_bytes(est.estimated_output_bytes),
+                Estimator::format_duration(est.estimated_duration),
+            );
+        }
+
+        // Last 1 month
+        if let Ok(range) = DateRange::new(today - chrono::Duration::days(30), today) {
+            let est = estimator.estimate_single(instrument, &range);
+            println!(
+                "{:<20} {:>12} {:>12} {:>12}",
+                "Last 1 month",
+                Estimator::format_bytes(est.estimated_compressed_bytes),
+                Estimator::format_bytes(est.estimated_output_bytes),
+                Estimator::format_duration(est.estimated_duration),
+            );
+        }
+
+        // Last 1 year
+        if let Ok(range) = DateRange::new(today - chrono::Duration::days(365), today) {
+            let est = estimator.estimate_single(instrument, &range);
+            println!(
+                "{:<20} {:>12} {:>12} {:>12}",
+                "Last 1 year",
+                Estimator::format_bytes(est.estimated_compressed_bytes),
+                Estimator::format_bytes(est.estimated_output_bytes),
+                Estimator::format_duration(est.estimated_duration),
+            );
+        }
+
+        // Full history (from start to today)
+        let start_date = start.date_naive();
+        if let Ok(range) = DateRange::new(start_date, today) {
+            let est = estimator.estimate_single(instrument, &range);
+            let years = (today - start_date).num_days() as f64 / 365.25;
+            println!(
+                "{:<20} {:>12} {:>12} {:>12}",
+                format!("Full history ({:.1}y)", years),
+                Estimator::format_bytes(est.estimated_compressed_bytes),
+                Estimator::format_bytes(est.estimated_output_bytes),
+                Estimator::format_duration(est.estimated_duration),
+            );
+        }
+
+        println!("\nNote: Estimates are based on historical averages and may vary.");
+    }
+
+    Ok(())
+}

bin/src/commands/list.rs

@@ -0,0 +1,41 @@
+//! List command implementation.
+//!
+//! This module handles listing available instruments with optional filtering.
+
+use crate::display::parse_category;
+use anyhow::Result;
+use paracas_lib::prelude::*;
+
+/// List available instruments with optional category filter or search pattern.
+pub(crate) fn list_instruments(category: Option<&str>, search: Option<&str>) -> Result<()> {
+    let registry = InstrumentRegistry::global();
+
+    let instruments: Vec<_> = match (category, search) {
+        (Some(cat), _) => {
+            let category = parse_category(cat)?;
+            registry.by_category(category).collect()
+        }
+        (_, Some(pattern)) => registry.search(pattern),
+        (None, None) => registry.all().collect(),
+    };
+
+    if instruments.is_empty() {
+        println!("No instruments found.");
+        return Ok(());
+    }
+
+    println!("{:<15} {:<20} {:<10}", "ID", "NAME", "CATEGORY");
+    println!("{}", "-".repeat(50));
+
+    for instrument in &instruments {
+        println!(
+            "{:<15} {:<20} {:<10}",
+            instrument.id(),
+            instrument.name(),
+            instrument.category()
+        );
+    }
+
+    println!("\nTotal: {} instruments", instruments.len());
+    Ok(())
+}

bin/src/commands/mod.rs

@@ -0,0 +1,8 @@
+//! CLI command implementations.
+
+pub(crate) mod daemon_run;
+pub(crate) mod download;
+pub(crate) mod download_all;
+pub(crate) mod info;
+pub(crate) mod list;
+pub(crate) mod status;

bin/src/commands/status.rs

@@ -0,0 +1,192 @@
+//! Background job status command.
+
+use anyhow::{Context, Result};
+use paracas_daemon::{DownloadJob, JobStatus, StateManager};
+
+/// Execute the status command.
+pub(crate) fn status(
+    job_id: Option<&str>,
+    running_only: bool,
+    show_all: bool,
+    follow: Option<u64>,
+    cancel_id: Option<&str>,
+) -> Result<()> {
+    let state_manager =
+        StateManager::with_default_path().context("Failed to initialize state manager")?;
+
+    // Handle cancellation request
+    if let Some(id) = cancel_id {
+        return cancel_job(&state_manager, id);
+    }
+
+    // Handle follow/watch mode
+    if let Some(interval) = follow {
+        return watch_jobs(&state_manager, job_id, interval);
+    }
+
+    // Show specific job or list jobs
+    #[allow(clippy::option_if_let_else)]
+    match job_id {
+        Some(id) => show_job_detail(&state_manager, id),
+        None => list_jobs(&state_manager, running_only, show_all),
+    }
+}
+
+fn show_job_detail(state: &StateManager, job_id: &str) -> Result<()> {
+    let id = job_id.parse().context("Invalid job ID format")?;
+
+    let job = state.load_job(id).context("Job not found")?;
+
+    println!("Job: {}", job.id);
+    println!("Status: {:?}", job.status);
+    println!("Created: {}", job.created_at.format("%Y-%m-%d %H:%M:%S"));
+
+    if let Some(started) = job.started_at {
+        println!("Started: {}", started.format("%Y-%m-%d %H:%M:%S"));
+    }
+    if let Some(completed) = job.completed_at {
+        println!("Completed: {}", completed.format("%Y-%m-%d %H:%M:%S"));
+    }
+
+    println!("Progress: {:.1}%", job.progress_percent());
+    println!(
+        "PID: {}",
+        job.pid
+            .map(|p| p.to_string())
+            .unwrap_or_else(|| "N/A".into())
+    );
+    println!(
+        "Log: {}",
+        job.log_file
+            .as_ref()
+            .map(|p| p.display().to_string())
+            .unwrap_or_else(|| "N/A".into())
+    );
+
+    println!("\nTasks:");
+    for (i, task) in job.tasks.iter().enumerate() {
+        let progress = if task.hours_total > 0 {
+            (task.hours_completed as f64 / task.hours_total as f64) * 100.0
+        } else {
+            0.0
+        };
+        println!(
+            "  {}. {} [{:?}] {:.1}% ({}/{} hours)",
+            i + 1,
+            task.instrument_id,
+            task.status,
+            progress,
+            task.hours_completed,
+            task.hours_total,
+        );
+        if let Some(ref err) = task.error_message {
+            println!("     Error: {}", err);
+        }
+    }
+
+    Ok(())
+}
+
+fn list_jobs(state: &StateManager, running_only: bool, show_all: bool) -> Result<()> {
+    let jobs = state.list_jobs()?;
+
+    let filtered: Vec<_> = jobs
+        .into_iter()
+        .filter(|job| {
+            if running_only {
+                matches!(job.status, JobStatus::Running | JobStatus::Pending)
+            } else if show_all {
+                true
+            } else {
+                // Default: show recent (last 24h) or active
+                let is_recent = job.created_at > chrono::Utc::now() - chrono::Duration::hours(24);
+                is_recent || matches!(job.status, JobStatus::Running | JobStatus::Pending)
+            }
+        })
+        .collect();
+
+    if filtered.is_empty() {
+        println!("No jobs found.");
+        if !show_all {
+            println!("Use --all to show all historical jobs.");
+        }
+        return Ok(());
+    }
+
+    println!(
+        "{:<36} {:<12} {:<10} {:<20}",
+        "JOB ID", "STATUS", "PROGRESS", "CREATED"
+    );
+    println!("{}", "-".repeat(80));
+
+    for job in &filtered {
+        println!(
+            "{:<36} {:<12} {:>8.1}% {:<20}",
+            job.id,
+            format!("{:?}", job.status),
+            job.progress_percent(),
+            job.created_at.format("%Y-%m-%d %H:%M"),
+        );
+    }
+
+    println!("\nTotal: {} jobs", filtered.len());
+    Ok(())
+}
+
+fn cancel_job(state: &StateManager, job_id: &str) -> Result<()> {
+    let id = job_id.parse().context("Invalid job ID format")?;
+
+    let mut job: DownloadJob = state.load_job(id).context("Job not found")?;
+
+    if !matches!(job.status, JobStatus::Running | JobStatus::Pending) {
+        anyhow::bail!("Job is not running (status: {:?})", job.status);
+    }
+
+    // Send SIGTERM to the process if running
+    if let Some(pid) = job.pid {
+        #[cfg(unix)]
+        {
+            use std::process::Command;
+            let _ = Command::new("kill")
+                .args(["-TERM", &pid.to_string()])
+                .status();
+        }
+        #[cfg(windows)]
+        {
+            use std::process::Command;
+            let _ = Command::new("taskkill")
+                .args(["/PID", &pid.to_string()])
+                .status();
+        }
+    }
+
+    job.status = JobStatus::Cancelled;
+    state.save_job(&job)?;
+
+    println!("Job {} cancelled.", id);
+    Ok(())
+}
+
+fn watch_jobs(state: &StateManager, job_id: Option<&str>, interval_secs: u64) -> Result<()> {
+    use std::io::Write;
+
+    let interval = std::time::Duration::from_secs(interval_secs);
+
+    loop {
+        // Clear screen
+        print!("\x1B[2J\x1B[1;1H");
+        std::io::stdout().flush()?;
+
+        println!(
+            "Watching jobs (refresh every {}s, Ctrl+C to exit)\n",
+            interval_secs
+        );
+
+        match job_id {
+            Some(id) => show_job_detail(state, id)?,
+            None => list_jobs(state, true, false)?,
+        }
+
+        std::thread::sleep(interval);
+    }
+}

bin/src/display.rs

@@ -0,0 +1,138 @@
+//! Display utilities and output formatting for the paracas CLI.
+
+use anyhow::{Result, bail};
+use clap::ValueEnum;
+use paracas_lib::prelude::*;
+use std::fs::File;
+use std::io::BufWriter;
+use std::path::PathBuf;
+
+/// Output format for downloaded data.
+#[derive(Clone, Copy, ValueEnum)]
+pub(crate) enum Format {
+    Csv,
+    Json,
+    Ndjson,
+    Parquet,
+}
+
+impl Format {
+    /// Returns the file extension for this format.
+    pub(crate) const fn extension(&self) -> &'static str {
+        match self {
+            Self::Csv => "csv",
+            Self::Json => "json",
+            Self::Ndjson => "ndjson",
+            Self::Parquet => "parquet",
+        }
+    }
+}
+
+impl std::fmt::Display for Format {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", self.extension())
+    }
+}
+
+/// Aggregate ticks into OHLCV bars using the given timeframe.
+pub(crate) fn aggregate_ticks(ticks: &[Tick], timeframe: Timeframe) -> Vec<Ohlcv> {
+    let mut aggregator = TickAggregator::new(timeframe);
+    let mut bars = Vec::new();
+
+    for tick in ticks {
+        if let Some(bar) = aggregator.process(*tick) {
+            bars.push(bar);
+        }
+    }
+
+    if let Some(bar) = aggregator.finish() {
+        bars.push(bar);
+    }
+
+    bars
+}
+
+/// Write ticks to a file in the specified format.
+pub(crate) fn write_ticks(ticks: &[Tick], output: &PathBuf, format: Format) -> Result<()> {
+    let file = File::create(output)?;
+    let writer = BufWriter::new(file);
+
+    match format {
+        Format::Csv => {
+            let formatter = CsvFormatter::new();
+            formatter.write_ticks(ticks, writer)?;
+        }
+        Format::Json => {
+            let formatter = JsonFormatter::new();
+            formatter.write_ticks(ticks, writer)?;
+        }
+        Format::Ndjson => {
+            let formatter = JsonFormatter::ndjson();
+            formatter.write_ticks(ticks, writer)?;
+        }
+        Format::Parquet => {
+            #[cfg(feature = "parquet")]
+            {
+                let formatter = ParquetFormatter::new();
+                formatter.write_ticks(ticks, writer)?;
+            }
+            #[cfg(not(feature = "parquet"))]
+            {
+                bail!("Parquet support not compiled in");
+            }
+        }
+    }
+
+    Ok(())
+}
+
+/// Write OHLCV bars to a file in the specified format.
+pub(crate) fn write_ohlcv(bars: &[Ohlcv], output: &PathBuf, format: Format) -> Result<()> {
+    let file = File::create(output)?;
+    let writer = BufWriter::new(file);
+
+    match format {
+        Format::Csv => {
+            let formatter = CsvFormatter::new();
+            formatter.write_ohlcv(bars, writer)?;
+        }
+        Format::Json => {
+            let formatter = JsonFormatter::new();
+            formatter.write_ohlcv(bars, writer)?;
+        }
+        Format::Ndjson => {
+            let formatter = JsonFormatter::ndjson();
+            formatter.write_ohlcv(bars, writer)?;
+        }
+        Format::Parquet => {
+            #[cfg(feature = "parquet")]
+            {
+                let formatter = ParquetFormatter::new();
+                formatter.write_ohlcv(bars, writer)?;
+            }
+            #[cfg(not(feature = "parquet"))]
+            {
+                bail!("Parquet support not compiled in");
+            }
+        }
+    }
+
+    Ok(())
+}
+
+/// Parse a category string into a Category enum.
+pub(crate) fn parse_category(s: &str) -> Result<Category> {
+    match s.to_lowercase().as_str() {
+        "forex" => Ok(Category::Forex),
+        "crypto" => Ok(Category::Crypto),
+        "index" => Ok(Category::Index),
+        "stock" => Ok(Category::Stock),
+        "commodity" => Ok(Category::Commodity),
+        "etf" => Ok(Category::Etf),
+        "bond" => Ok(Category::Bond),
+        _ => bail!(
+            "Unknown category: {}. Valid options: forex, crypto, index, stock, commodity, etf, bond",
+            s
+        ),
+    }
+}

bin/src/main.rs

@@ -1,22 +1,21 @@
 //! paracas CLI - High-performance Dukascopy tick data downloader.
 
-use anyhow::{Context, Result, bail};
-use chrono::NaiveDate;
-use clap::{Parser, Subcommand, ValueEnum};
-use futures::StreamExt;
-use indicatif::{ProgressBar, ProgressStyle};
-use paracas_lib::prelude::*;
-use std::fs::File;
-use std::io::BufWriter;
+use anyhow::{Context, Result};
+use clap::{Parser, Subcommand};
 use std::path::PathBuf;
 
+mod commands;
+mod display;
+
+use display::Format;
+
 #[derive(Parser)]
 #[command(name = "paracas")]
 #[command(about = "High-performance Dukascopy tick data downloader", long_about = None)]
 #[command(version)]
 struct Cli {
     #[command(subcommand)]
-    command: Commands,
+    command: Option<Commands>,
 
     /// Verbosity level (-v, -vv, -vvv)
     #[arg(short, long, action = clap::ArgAction::Count, global = true)]
@@ -25,6 +24,10 @@ struct Cli {
     /// Quiet mode (suppress progress output)
     #[arg(short, long, global = true)]
     quiet: bool,
+
+    /// Hidden: Run as daemon with job ID (internal use only)
+    #[arg(long, hide = true)]
+    daemon_run: Option<String>,
 }
 
 #[derive(Subcommand)]
@@ -57,6 +60,14 @@ enum Commands {
         /// Maximum concurrent downloads
         #[arg(long, default_value = "32")]
         concurrency: usize,
+
+        /// Run in background as daemon
+        #[arg(long)]
+        background: bool,
+
+        /// Skip confirmation prompt (for background mode)
+        #[arg(long)]
+        yes: bool,
     },
 
     /// List available instruments
@@ -75,21 +86,88 @@ enum Commands {
         /// Instrument identifier
         instrument: String,
     },
-}
 
-#[derive(Clone, Copy, ValueEnum)]
-enum Format {
-    Csv,
-    Json,
-    Ndjson,
-    Parquet,
+    /// Check background job status
+    Status {
+        /// Specific job ID to check
+        job_id: Option<String>,
+
+        /// Show only running jobs
+        #[arg(long)]
+        running: bool,
+
+        /// Show all jobs (including completed)
+        #[arg(long)]
+        all: bool,
+
+        /// Follow/watch mode (refresh every N seconds)
+        #[arg(short, long)]
+        follow: Option<u64>,
+
+        /// Cancel a running job
+        #[arg(long)]
+        cancel: Option<String>,
+    },
+
+    /// Download all instruments (or filter by category)
+    DownloadAll {
+        /// Filter by category (forex, crypto, index, commodity)
+        #[arg(short, long)]
+        category: Option<String>,
+
+        /// Start date (YYYY-MM-DD). Defaults to each instrument's earliest data.
+        #[arg(short, long)]
+        start: Option<String>,
+
+        /// End date (YYYY-MM-DD). Defaults to today.
+        #[arg(short, long)]
+        end: Option<String>,
+
+        /// Output directory. Files named <instrument>.<format>
+        #[arg(short, long, default_value = ".")]
+        output_dir: PathBuf,
+
+        /// Output format
+        #[arg(short, long, value_enum, default_value = "csv")]
+        format: Format,
+
+        /// OHLCV aggregation timeframe (omit for raw ticks)
+        #[arg(short, long)]
+        timeframe: Option<String>,
+
+        /// Maximum concurrent instruments to download
+        #[arg(long, default_value = "4")]
+        parallel_instruments: usize,
+
+        /// Maximum concurrent HTTP requests per instrument
+        #[arg(long, default_value = "32")]
+        concurrency: usize,
+
+        /// Run in background as daemon
+        #[arg(long)]
+        background: bool,
+
+        /// Skip confirmation prompt
+        #[arg(long)]
+        yes: bool,
+    },
 }
 
 #[tokio::main]
 async fn main() -> Result<()> {
     let cli = Cli::parse();
 
-    match cli.command {
+    // Check for daemon mode first (internal use)
+    if let Some(job_id) = cli.daemon_run {
+        return commands::daemon_run::daemon_run(&job_id).await;
+    }
+
+    // Require a command otherwise
+    let command = cli
+        .command
+        .context("No command provided. Use --help for usage.")?;
+
+    match command {
         Commands::Download {
             instrument,
             start,
@@ -98,8 +176,10 @@ async fn main() -> Result<()> {
             format,
             timeframe,
             concurrency,
+            background,
+            yes,
         } => {
-            download(
+            commands::download::download(
                 &instrument,
                 start.as_deref(),
                 end.as_deref(),
@@ -107,284 +187,49 @@ async fn main() -> Result<()> {
                 format,
                 timeframe.as_deref(),
                 concurrency,
+                background,
+                yes,
                 cli.quiet,
             )
             .await
         }
         Commands::List { category, search } => {
-            list_instruments(category.as_deref(), search.as_deref())
+            commands::list::list_instruments(category.as_deref(), search.as_deref())
+        }
+        Commands::Info { instrument } => commands::info::show_info(&instrument),
+        Commands::Status {
+            job_id,
+            running,
+            all,
+            follow,
+            cancel,
+        } => commands::status::status(job_id.as_deref(), running, all, follow, cancel.as_deref()),
+        Commands::DownloadAll {
+            category,
+            start,
+            end,
+            output_dir,
+            format,
+            timeframe,
+            parallel_instruments,
+            concurrency,
+            background,
+            yes,
+        } => {
+            commands::download_all::download_all(
+                category.as_deref(),
+                start.as_deref(),
+                end.as_deref(),
+                output_dir,
+                format,
+                timeframe.as_deref(),
+                parallel_instruments,
+                concurrency,
+                background,
+                yes,
+                cli.quiet,
+            )
+            .await
         }
-        Commands::Info { instrument } => show_info(&instrument),
-    }
-}
-
-#[allow(clippy::too_many_arguments)]
-async fn download(
-    instrument_id: &str,
-    start_str: Option<&str>,
-    end_str: Option<&str>,
-    output: Option<PathBuf>,
-    format: Format,
-    timeframe_str: Option<&str>,
-    concurrency: usize,
-    quiet: bool,
-) -> Result<()> {
-    // Lookup instrument
-    let registry = InstrumentRegistry::global();
-    let instrument = registry
-        .get(instrument_id)
-        .with_context(|| format!("Unknown instrument: {instrument_id}"))?;
-
-    // Parse start date (default to instrument's earliest available data)
-    let start = match start_str {
-        Some(s) => NaiveDate::parse_from_str(s, "%Y-%m-%d")
-            .with_context(|| format!("Invalid start date: {s}"))?,
-        None => instrument
-            .start_tick_date()
-            .map(|dt| dt.date_naive())
-            .unwrap_or_else(|| NaiveDate::from_ymd_opt(2003, 5, 5).expect("valid date")),
-    };
-
-    // Parse end date (default to today)
-    let end = match end_str {
-        Some(s) => NaiveDate::parse_from_str(s, "%Y-%m-%d")
-            .with_context(|| format!("Invalid end date: {s}"))?,
-        None => chrono::Utc::now().date_naive(),
-    };
-
-    let range = DateRange::new(start, end)?;
-
-    // Determine output path (default to <instrument>.<format>)
-    let output = output.unwrap_or_else(|| {
-        let ext = match format {
-            Format::Csv => "csv",
-            Format::Json => "json",
-            Format::Ndjson => "ndjson",
-            Format::Parquet => "parquet",
-        };
-        PathBuf::from(format!("{}.{}", instrument_id, ext))
-    });
-
-    // Parse timeframe
-    let timeframe = match timeframe_str {
-        Some(tf) => tf
-            .parse::<Timeframe>()
-            .map_err(|e| anyhow::anyhow!("{e}"))?,
-        None => Timeframe::Tick,
-    };
-
-    // Create client
-    let config = ClientConfig {
-        concurrency,
-        ..Default::default()
-    };
-    let client = DownloadClient::new(config)?;
-
-    // Setup progress bar
-    let total_hours = range.total_hours() as u64;
-    let progress = if quiet {
-        ProgressBar::hidden()
-    } else {
-        let pb = ProgressBar::new(total_hours);
-        pb.set_style(
-            ProgressStyle::default_bar()
-                .template("{spinner:.green} [{elapsed_precise}] [{bar:40.cyan/blue}] {pos}/{len} hours ({percent}%) {msg}")
-                .expect("Invalid progress template")
-                .progress_chars("=>-"),
-        );
-        pb.set_message(format!("{} {} -> {}", instrument.id(), start, end));
-        pb
-    };
-
-    // Download and collect ticks using the resilient stream
-    // This will retry on transient errors and skip hours that fail after retries
-    let mut all_ticks: Vec<Tick> = Vec::new();
-    let mut skipped_hours = 0u64;
-    let mut stream = paracas_lib::tick_stream_resilient(&client, instrument, range);
-
-    while let Some(batch) = stream.next().await {
-        if batch.had_error() {
-            skipped_hours += 1;
-        }
-        all_ticks.extend(batch.ticks);
-        progress.inc(1);
-    }
-
-    let finish_msg = if skipped_hours > 0 {
-        format!(
-            "Downloaded {} ticks ({} hours skipped due to errors)",
-            all_ticks.len(),
-            skipped_hours
-        )
-    } else {
-        format!("Downloaded {} ticks", all_ticks.len())
-    };
-    progress.finish_with_message(finish_msg);
-
-    // Aggregate if needed
-    if timeframe.is_tick() {
-        // Write raw ticks
-        write_ticks(&all_ticks, &output, format)?;
-    } else {
-        // Aggregate to OHLCV
-        let bars = aggregate_ticks(&all_ticks, timeframe);
-        write_ohlcv(&bars, &output, format)?;
-    }
-
-    if !quiet {
-        println!("Output written to: {}", output.display());
-    }
-
-    Ok(())
-}
-
-fn aggregate_ticks(ticks: &[Tick], timeframe: Timeframe) -> Vec<Ohlcv> {
-    let mut aggregator = TickAggregator::new(timeframe);
-    let mut bars = Vec::new();
-
-    for tick in ticks {
-        if let Some(bar) = aggregator.process(*tick) {
-            bars.push(bar);
-        }
-    }
-
-    if let Some(bar) = aggregator.finish() {
-        bars.push(bar);
-    }
-
-    bars
-}
-
-fn write_ticks(ticks: &[Tick], output: &PathBuf, format: Format) -> Result<()> {
-    let file = File::create(output)?;
-    let writer = BufWriter::new(file);
-
-    match format {
-        Format::Csv => {
-            let formatter = CsvFormatter::new();
-            formatter.write_ticks(ticks, writer)?;
-        }
-        Format::Json => {
-            let formatter = JsonFormatter::new();
-            formatter.write_ticks(ticks, writer)?;
-        }
-        Format::Ndjson => {
-            let formatter = JsonFormatter::ndjson();
-            formatter.write_ticks(ticks, writer)?;
-        }
-        Format::Parquet => {
-            #[cfg(feature = "parquet")]
-            {
-                let formatter = ParquetFormatter::new();
-                formatter.write_ticks(ticks, writer)?;
-            }
-            #[cfg(not(feature = "parquet"))]
-            {
-                bail!("Parquet support not compiled in");
-            }
-        }
-    }
-
-    Ok(())
-}
-
-fn write_ohlcv(bars: &[Ohlcv], output: &PathBuf, format: Format) -> Result<()> {
-    let file = File::create(output)?;
-    let writer = BufWriter::new(file);
-
-    match format {
-        Format::Csv => {
-            let formatter = CsvFormatter::new();
-            formatter.write_ohlcv(bars, writer)?;
-        }
-        Format::Json => {
-            let formatter = JsonFormatter::new();
-            formatter.write_ohlcv(bars, writer)?;
-        }
-        Format::Ndjson => {
-            let formatter = JsonFormatter::ndjson();
-            formatter.write_ohlcv(bars, writer)?;
-        }
-        Format::Parquet => {
-            #[cfg(feature = "parquet")]
-            {
-                let formatter = ParquetFormatter::new();
-                formatter.write_ohlcv(bars, writer)?;
-            }
-            #[cfg(not(feature = "parquet"))]
-            {
-                bail!("Parquet support not compiled in");
-            }
-        }
-    }
-
-    Ok(())
-}
-
-fn list_instruments(category: Option<&str>, search: Option<&str>) -> Result<()> {
-    let registry = InstrumentRegistry::global();
-
-    let instruments: Vec<_> = match (category, search) {
-        (Some(cat), _) => {
-            let category = parse_category(cat)?;
-            registry.by_category(category).collect()
-        }
-        (_, Some(pattern)) => registry.search(pattern),
-        (None, None) => registry.all().collect(),
-    };
-
-    if instruments.is_empty() {
-        println!("No instruments found.");
-        return Ok(());
-    }
-
-    println!("{:<15} {:<20} {:<10}", "ID", "NAME", "CATEGORY");
-    println!("{}", "-".repeat(50));
-
-    for instrument in &instruments {
-        println!(
-            "{:<15} {:<20} {:<10}",
-            instrument.id(),
-            instrument.name(),
-            instrument.category()
-        );
-    }
-
-    println!("\nTotal: {} instruments", instruments.len());
-    Ok(())
-}
-
-fn show_info(instrument_id: &str) -> Result<()> {
-    let registry = InstrumentRegistry::global();
-    let instrument = registry
-        .get(instrument_id)
-        .with_context(|| format!("Unknown instrument: {instrument_id}"))?;
-
-    println!("Instrument: {}", instrument.name());
-    println!("ID:         {}", instrument.id());
-    println!("Category:   {}", instrument.category());
-    println!("Description: {}", instrument.description());
-    println!("Decimal Factor: {}", instrument.decimal_factor());
-
-    if let Some(start) = instrument.start_tick_date() {
-        println!("Data Available From: {}", start.format("%Y-%m-%d"));
-    }
-
-    Ok(())
-}
-
-fn parse_category(s: &str) -> Result<Category> {
-    match s.to_lowercase().as_str() {
-        "forex" => Ok(Category::Forex),
-        "crypto" => Ok(Category::Crypto),
-        "index" => Ok(Category::Index),
-        "stock" => Ok(Category::Stock),
-        "commodity" => Ok(Category::Commodity),
-        "etf" => Ok(Category::Etf),
-        "bond" => Ok(Category::Bond),
-        _ => bail!(
-            "Unknown category: {}. Valid options: forex, crypto, index, stock, commodity, etf, bond",
-            s
-        ),
     }
 }

crates/paracas-daemon/Cargo.toml

@@ -0,0 +1,25 @@
+[package]
+name = "paracas-daemon"
+version.workspace = true
+edition.workspace = true
+rust-version.workspace = true
+license.workspace = true
+homepage.workspace = true
+repository.workspace = true
+description = "Background job management for paracas tick data downloader"
+
+[lints]
+workspace = true
+
+[dependencies]
+serde = { workspace = true }
+serde_json = { workspace = true }
+chrono = { workspace = true }
+thiserror = { workspace = true }
+uuid = { workspace = true }
+directories = { workspace = true }
+tokio = { workspace = true }
+
+[dev-dependencies]
+tempfile = { workspace = true }
+tokio = { workspace = true, features = ["rt-multi-thread", "macros"] }

crates/paracas-daemon/src/daemon.rs

@@ -0,0 +1,273 @@
+//! Daemon process spawning for background downloads.
+//!
+//! This module provides functionality to spawn detached daemon processes
+//! that can run downloads in the background, even after the parent process exits.
+
+use crate::{DownloadJob, JobId, StateError, StateManager};
+use std::fs::OpenOptions;
+use std::path::PathBuf;
+use std::process::{Command, Stdio};
+
+/// Result type for daemon operations (re-exported from state module).
+pub(crate) type Result<T> = std::result::Result<T, StateError>;
+
+/// Environment variable name for the daemon job ID.
+pub const DAEMON_JOB_ID_ENV: &str = "PARACAS_DAEMON_JOB_ID";
+
+/// Command line argument for daemon mode.
+pub const DAEMON_RUN_ARG: &str = "--daemon-run";
+
+/// Spawns detached daemon processes for background downloads.
+///
+/// The spawner handles all the platform-specific details of creating
+/// a detached background process that will continue running after
+/// the parent process exits.
+#[derive(Debug, Clone)]
+pub struct DaemonSpawner {
+    state_manager: StateManager,
+    executable_path: PathBuf,
+}
+
+impl DaemonSpawner {
+    /// Create a new daemon spawner.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the current executable path cannot be determined.
+    pub fn new(state_manager: StateManager) -> Result<Self> {
+        let executable_path = Self::executable_path()?;
+        Ok(Self {
+            state_manager,
+            executable_path,
+        })
+    }
+
+    /// Create a new daemon spawner with a custom executable path.
+    ///
+    /// This is useful for testing or when spawning a different binary.
+    #[must_use]
+    pub const fn with_executable(state_manager: StateManager, executable_path: PathBuf) -> Self {
+        Self {
+            state_manager,
+            executable_path,
+        }
+    }
+
+    /// Spawn a background download job.
+    ///
+    /// Returns the job ID for tracking. The job's PID and log file path
+    /// will be updated after spawning.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the daemon process cannot be spawned.
+    pub fn spawn(&self, job: &mut DownloadJob) -> Result<JobId> {
+        let job_id = job.id;
+
+        // Set up log file path
+        let log_path = self.state_manager.job_log_path(job_id);
+        job.log_file = Some(log_path.clone());
+
+        // Save job state before spawning
+        self.state_manager.save_job(job)?;
+
+        // Open log file for stdout/stderr redirection
+        let log_file = OpenOptions::new()
+            .create(true)
+            .write(true)
+            .truncate(true)
+            .open(&log_path)
+            .map_err(|e| StateError::WriteFile {
+                path: log_path.clone(),
+                source: e,
+            })?;
+
+        let log_file_stderr = log_file.try_clone().map_err(|e| StateError::WriteFile {
+            path: log_path.clone(),
+            source: e,
+        })?;
+
+        // Spawn the daemon process
+        let child = self.spawn_detached(job_id, log_file, log_file_stderr)?;
+
+        // Update job with PID
+        let pid = child.id();
+        job.pid = Some(pid);
+        self.state_manager.save_job(job)?;
+
+        Ok(job_id)
+    }
+
+    /// Spawn a detached child process.
+    #[cfg(unix)]
+    fn spawn_detached(
+        &self,
+        job_id: JobId,
+        stdout: std::fs::File,
+        stderr: std::fs::File,
+    ) -> Result<std::process::Child> {
+        use std::os::unix::process::CommandExt;
+
+        let child = Command::new(&self.executable_path)
+            .args([DAEMON_RUN_ARG, &job_id.to_string()])
+            .env(DAEMON_JOB_ID_ENV, job_id.to_string())
+            .stdin(Stdio::null())
+            .stdout(stdout)
+            .stderr(stderr)
+            .process_group(0) // Create new process group (detach from parent)
+            .spawn()
+            .map_err(|e| StateError::SpawnDaemon {
+                executable: self.executable_path.clone(),
+                source: e,
+            })?;
+
+        Ok(child)
+    }
+
+    /// Spawn a detached child process on Windows.
+    #[cfg(windows)]
+    fn spawn_detached(
+        &self,
+        job_id: JobId,
+        stdout: std::fs::File,
+        stderr: std::fs::File,
+    ) -> Result<std::process::Child> {
+        use std::os::windows::process::CommandExt;
+
+        // CREATE_NEW_PROCESS_GROUP | DETACHED_PROCESS
+        const CREATE_NEW_PROCESS_GROUP: u32 = 0x00000200;
+        const DETACHED_PROCESS: u32 = 0x00000008;
+
+        let child = Command::new(&self.executable_path)
+            .args([DAEMON_RUN_ARG, &job_id.to_string()])
+            .env(DAEMON_JOB_ID_ENV, job_id.to_string())
+            .stdin(Stdio::null())
+            .stdout(stdout)
+            .stderr(stderr)
+            .creation_flags(CREATE_NEW_PROCESS_GROUP | DETACHED_PROCESS)
+            .spawn()
+            .map_err(|e| StateError::SpawnDaemon {
+                executable: self.executable_path.clone(),
+                source: e,
+            })?;
+
+        Ok(child)
+    }
+
+    /// Spawn a detached child process (fallback for other platforms).
+    #[cfg(not(any(unix, windows)))]
+    fn spawn_detached(
+        &self,
+        job_id: JobId,
+        stdout: std::fs::File,
+        stderr: std::fs::File,
+    ) -> Result<std::process::Child> {
+        let child = Command::new(&self.executable_path)
+            .args([DAEMON_RUN_ARG, &job_id.to_string()])
+            .env(DAEMON_JOB_ID_ENV, job_id.to_string())
+            .stdin(Stdio::null())
+            .stdout(stdout)
+            .stderr(stderr)
+            .spawn()
+            .map_err(|e| StateError::SpawnDaemon {
+                executable: self.executable_path.clone(),
+                source: e,
+            })?;
+
+        Ok(child)
+    }
+
+    /// Get the path to the current executable.
+    fn executable_path() -> Result<PathBuf> {
+        std::env::current_exe().map_err(|e| StateError::ExecutablePath { source: e })
+    }
+
+    /// Returns the state manager reference.
+    #[must_use]
+    pub const fn state_manager(&self) -> &StateManager {
+        &self.state_manager
+    }
+
+    /// Returns the executable path.
+    #[must_use]
+    pub const fn executable(&self) -> &PathBuf {
+        &self.executable_path
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::InstrumentTask;
+    use std::path::PathBuf;
+    use tempfile::TempDir;
+
+    fn create_test_job() -> DownloadJob {
+        let tasks = vec![InstrumentTask::new(
+            "EURUSD".to_string(),
+            "2024-01-01".to_string(),
+            "2024-01-02".to_string(),
+            PathBuf::from("/tmp/eurusd.csv"),
+            "csv".to_string(),
+            "tick".to_string(),
+            48,
+        )];
+        DownloadJob::new(tasks, 4)
+    }
+
+    #[test]
+    fn test_daemon_spawner_creation() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        // Use a known executable for testing
+        let spawner = DaemonSpawner::with_executable(state_manager, PathBuf::from("/bin/echo"));
+
+        assert_eq!(spawner.executable(), &PathBuf::from("/bin/echo"));
+    }
+
+    #[test]
+    fn test_daemon_spawner_with_executable() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        let custom_path = PathBuf::from("/custom/paracas");
+        let spawner = DaemonSpawner::with_executable(state_manager, custom_path.clone());
+
+        assert_eq!(spawner.executable(), &custom_path);
+    }
+
+    #[test]
+    fn test_spawn_sets_log_file() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        // Use /bin/true or /usr/bin/true for a quick-exit process
+        #[cfg(unix)]
+        let exe_path = if PathBuf::from("/bin/true").exists() {
+            PathBuf::from("/bin/true")
+        } else {
+            PathBuf::from("/usr/bin/true")
+        };
+        #[cfg(not(unix))]
+        let exe_path = PathBuf::from("cmd.exe");
+
+        let spawner = DaemonSpawner::with_executable(state_manager.clone(), exe_path);
+
+        let mut job = create_test_job();
+        let job_id = job.id;
+
+        // Spawn the job
+        let result = spawner.spawn(&mut job);
+
+        // On CI or systems where the binary doesn't exist, this might fail
+        if result.is_ok() {
+            assert!(job.log_file.is_some());
+            assert!(job.pid.is_some());
+
+            // Verify log file path is correct
+            let expected_log_path = state_manager.job_log_path(job_id);
+            assert_eq!(job.log_file.as_ref().unwrap(), &expected_log_path);
+        }
+    }
+}

crates/paracas-daemon/src/job.rs

@@ -0,0 +1,317 @@
+//! Download job definitions and types.
+
+use chrono::{DateTime, Utc};
+use serde::{Deserialize, Serialize};
+use std::path::PathBuf;
+use uuid::Uuid;
+
+/// Unique identifier for a download job.
+pub type JobId = Uuid;
+
+/// Status of a download job or task.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize, Default)]
+#[serde(rename_all = "lowercase")]
+pub enum JobStatus {
+    /// Job is queued but not yet started.
+    #[default]
+    Pending,
+    /// Job is currently running.
+    Running,
+    /// Job completed successfully.
+    Completed,
+    /// Job failed with an error.
+    Failed,
+    /// Job was cancelled by the user.
+    Cancelled,
+}
+
+impl JobStatus {
+    /// Returns true if the job is in a terminal state.
+    #[must_use]
+    pub const fn is_finished(&self) -> bool {
+        matches!(self, Self::Completed | Self::Failed | Self::Cancelled)
+    }
+
+    /// Returns the status as a string identifier.
+    #[must_use]
+    pub const fn as_str(&self) -> &'static str {
+        match self {
+            Self::Pending => "pending",
+            Self::Running => "running",
+            Self::Completed => "completed",
+            Self::Failed => "failed",
+            Self::Cancelled => "cancelled",
+        }
+    }
+}
+
+impl std::fmt::Display for JobStatus {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", self.as_str())
+    }
+}
+
+/// A download task for a single instrument within a job.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct InstrumentTask {
+    /// The instrument identifier (e.g., "EURUSD").
+    pub instrument_id: String,
+    /// Start date for the download (inclusive).
+    pub start_date: String,
+    /// End date for the download (inclusive).
+    pub end_date: String,
+    /// Output file path for this instrument's data.
+    pub output_path: PathBuf,
+    /// Output format (e.g., "csv", "json", "parquet").
+    pub format: String,
+    /// Timeframe for aggregation (e.g., "tick", "m1", "h1").
+    pub timeframe: String,
+    /// Current status of this task.
+    pub status: JobStatus,
+    /// Number of hours completed for this task.
+    pub hours_completed: u32,
+    /// Total number of hours to download.
+    pub hours_total: u32,
+    /// Number of ticks downloaded so far.
+    pub ticks_downloaded: u64,
+    /// Number of bytes written to output file.
+    pub bytes_written: u64,
+    /// Error message if the task failed.
+    pub error_message: Option<String>,
+}
+
+impl InstrumentTask {
+    /// Creates a new instrument task.
+    #[must_use]
+    pub const fn new(
+        instrument_id: String,
+        start_date: String,
+        end_date: String,
+        output_path: PathBuf,
+        format: String,
+        timeframe: String,
+        hours_total: u32,
+    ) -> Self {
+        Self {
+            instrument_id,
+            start_date,
+            end_date,
+            output_path,
+            format,
+            timeframe,
+            status: JobStatus::Pending,
+            hours_completed: 0,
+            hours_total,
+            ticks_downloaded: 0,
+            bytes_written: 0,
+            error_message: None,
+        }
+    }
+
+    /// Returns the progress percentage for this task.
+    #[must_use]
+    pub fn progress_percent(&self) -> f64 {
+        if self.hours_total == 0 {
+            return 0.0;
+        }
+        (self.hours_completed as f64 / self.hours_total as f64) * 100.0
+    }
+}
+
+/// A complete download job containing one or more instrument tasks.
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub struct DownloadJob {
+    /// Unique identifier for this job.
+    pub id: JobId,
+    /// Timestamp when the job was created.
+    pub created_at: DateTime<Utc>,
+    /// Timestamp when the job started running.
+    pub started_at: Option<DateTime<Utc>>,
+    /// Timestamp when the job completed (success, failure, or cancellation).
+    pub completed_at: Option<DateTime<Utc>>,
+    /// Current status of the job.
+    pub status: JobStatus,
+    /// List of instrument download tasks.
+    pub tasks: Vec<InstrumentTask>,
+    /// Number of concurrent downloads.
+    pub concurrency: usize,
+    /// Process ID of the daemon running this job.
+    pub pid: Option<u32>,
+    /// Path to the log file for this job.
+    pub log_file: Option<PathBuf>,
+}
+
+impl DownloadJob {
+    /// Creates a new download job with the given tasks.
+    #[must_use]
+    pub fn new(tasks: Vec<InstrumentTask>, concurrency: usize) -> Self {
+        Self {
+            id: Uuid::new_v4(),
+            created_at: Utc::now(),
+            started_at: None,
+            completed_at: None,
+            status: JobStatus::Pending,
+            tasks,
+            concurrency,
+            pid: None,
+            log_file: None,
+        }
+    }
+
+    /// Returns the overall progress percentage across all tasks.
+    #[must_use]
+    pub fn progress_percent(&self) -> f64 {
+        let total_hours: u32 = self.tasks.iter().map(|t| t.hours_total).sum();
+        let completed_hours: u32 = self.tasks.iter().map(|t| t.hours_completed).sum();
+
+        if total_hours == 0 {
+            return 0.0;
+        }
+        (completed_hours as f64 / total_hours as f64) * 100.0
+    }
+
+    /// Returns true if the job is in a terminal state.
+    #[must_use]
+    pub const fn is_finished(&self) -> bool {
+        self.status.is_finished()
+    }
+
+    /// Marks the job as started with the current timestamp and process ID.
+    pub fn mark_started(&mut self, pid: u32) {
+        self.status = JobStatus::Running;
+        self.started_at = Some(Utc::now());
+        self.pid = Some(pid);
+    }
+
+    /// Marks the job as completed successfully.
+    pub fn mark_completed(&mut self) {
+        self.status = JobStatus::Completed;
+        self.completed_at = Some(Utc::now());
+    }
+
+    /// Marks the job as failed with an optional error message.
+    pub fn mark_failed(&mut self, error: Option<String>) {
+        self.status = JobStatus::Failed;
+        self.completed_at = Some(Utc::now());
+
+        // If an error message is provided, set it on any running tasks
+        if let Some(ref msg) = error {
+            for task in &mut self.tasks {
+                if task.status == JobStatus::Running {
+                    task.status = JobStatus::Failed;
+                    task.error_message = Some(msg.clone());
+                }
+            }
+        }
+    }
+
+    /// Marks the job as cancelled.
+    pub fn mark_cancelled(&mut self) {
+        self.status = JobStatus::Cancelled;
+        self.completed_at = Some(Utc::now());
+
+        // Cancel any pending or running tasks
+        for task in &mut self.tasks {
+            if !task.status.is_finished() {
+                task.status = JobStatus::Cancelled;
+            }
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_job_status_is_finished() {
+        assert!(!JobStatus::Pending.is_finished());
+        assert!(!JobStatus::Running.is_finished());
+        assert!(JobStatus::Completed.is_finished());
+        assert!(JobStatus::Failed.is_finished());
+        assert!(JobStatus::Cancelled.is_finished());
+    }
+
+    #[test]
+    fn test_instrument_task_progress() {
+        let mut task = InstrumentTask::new(
+            "EURUSD".to_string(),
+            "2024-01-01".to_string(),
+            "2024-01-02".to_string(),
+            PathBuf::from("/tmp/eurusd.csv"),
+            "csv".to_string(),
+            "tick".to_string(),
+            48,
+        );
+
+        assert_eq!(task.progress_percent(), 0.0);
+
+        task.hours_completed = 24;
+        assert!((task.progress_percent() - 50.0).abs() < 0.001);
+
+        task.hours_completed = 48;
+        assert!((task.progress_percent() - 100.0).abs() < 0.001);
+    }
+
+    #[test]
+    fn test_download_job_progress() {
+        let tasks = vec![
+            InstrumentTask::new(
+                "EURUSD".to_string(),
+                "2024-01-01".to_string(),
+                "2024-01-02".to_string(),
+                PathBuf::from("/tmp/eurusd.csv"),
+                "csv".to_string(),
+                "tick".to_string(),
+                48,
+            ),
+            InstrumentTask::new(
+                "GBPUSD".to_string(),
+                "2024-01-01".to_string(),
+                "2024-01-02".to_string(),
+                PathBuf::from("/tmp/gbpusd.csv"),
+                "csv".to_string(),
+                "tick".to_string(),
+                48,
+            ),
+        ];
+
+        let mut job = DownloadJob::new(tasks, 4);
+        assert_eq!(job.progress_percent(), 0.0);
+
+        job.tasks[0].hours_completed = 48;
+        assert!((job.progress_percent() - 50.0).abs() < 0.001);
+
+        job.tasks[1].hours_completed = 48;
+        assert!((job.progress_percent() - 100.0).abs() < 0.001);
+    }
+
+    #[test]
+    fn test_download_job_lifecycle() {
+        let tasks = vec![InstrumentTask::new(
+            "EURUSD".to_string(),
+            "2024-01-01".to_string(),
+            "2024-01-02".to_string(),
+            PathBuf::from("/tmp/eurusd.csv"),
+            "csv".to_string(),
+            "tick".to_string(),
+            48,
+        )];
+
+        let mut job = DownloadJob::new(tasks, 4);
+        assert_eq!(job.status, JobStatus::Pending);
+        assert!(job.started_at.is_none());
+        assert!(!job.is_finished());
+
+        job.mark_started(12345);
+        assert_eq!(job.status, JobStatus::Running);
+        assert!(job.started_at.is_some());
+        assert_eq!(job.pid, Some(12345));
+        assert!(!job.is_finished());
+
+        job.mark_completed();
+        assert_eq!(job.status, JobStatus::Completed);
+        assert!(job.completed_at.is_some());
+        assert!(job.is_finished());
+    }
+}

crates/paracas-daemon/src/lib.rs

@@ -0,0 +1,27 @@
+//! Background job management for paracas tick data downloader.
+//!
+//! This crate provides state management and job tracking for background
+//! download operations:
+//!
+//! - [`JobId`] - Unique identifier for download jobs
+//! - [`JobStatus`] - Current status of a job
+//! - [`InstrumentTask`] - Download task for a single instrument
+//! - [`DownloadJob`] - Complete download job with multiple tasks
+//! - [`StateManager`] - Persistent state storage and retrieval
+//! - [`DaemonSpawner`] - Spawns detached daemon processes for background downloads
+//! - [`DaemonProgress`] - Thread-safe progress tracking for daemon jobs
+
+#![doc(issue_tracker_base_url = "https://github.com/factordynamics/paracas/issues/")]
+#![cfg_attr(docsrs, feature(doc_cfg, doc_auto_cfg))]
+#![warn(missing_docs)]
+#![forbid(unsafe_code)]
+
+mod daemon;
+mod job;
+mod progress;
+mod state;
+
+pub use daemon::{DAEMON_JOB_ID_ENV, DAEMON_RUN_ARG, DaemonSpawner};
+pub use job::{DownloadJob, InstrumentTask, JobId, JobStatus};
+pub use progress::DaemonProgress;
+pub use state::{Result, StateError, StateManager};

crates/paracas-daemon/src/progress.rs

@@ -0,0 +1,490 @@
+//! Progress tracking for daemon downloads.
+//!
+//! This module provides thread-safe progress tracking for daemon jobs,
+//! including periodic checkpointing to disk for crash recovery.
+
+use crate::{DownloadJob, JobStatus, StateError, StateManager};
+use std::sync::Arc;
+use std::time::{Duration, Instant};
+use tokio::sync::RwLock;
+
+/// Thread-safe progress tracker for daemon jobs.
+///
+/// The `DaemonProgress` struct provides a way to track download progress
+/// from multiple concurrent tasks while ensuring periodic checkpoints
+/// are saved to disk for crash recovery.
+#[derive(Debug)]
+pub struct DaemonProgress {
+    /// State manager for persistence.
+    state_manager: StateManager,
+    /// The job being tracked (protected by RwLock for concurrent access).
+    job: Arc<RwLock<DownloadJob>>,
+    /// Minimum interval between saves.
+    save_interval: Duration,
+    /// Last time state was saved to disk.
+    last_save: std::sync::Mutex<Instant>,
+}
+
+impl DaemonProgress {
+    /// Default save interval for checkpointing (10 seconds).
+    pub const DEFAULT_SAVE_INTERVAL: Duration = Duration::from_secs(10);
+
+    /// Create a new progress tracker.
+    ///
+    /// The tracker will periodically save checkpoints to disk at the
+    /// default interval of 10 seconds.
+    #[must_use]
+    pub fn new(state_manager: StateManager, job: DownloadJob) -> Self {
+        Self {
+            state_manager,
+            job: Arc::new(RwLock::new(job)),
+            save_interval: Self::DEFAULT_SAVE_INTERVAL,
+            last_save: std::sync::Mutex::new(Instant::now()),
+        }
+    }
+
+    /// Create a new progress tracker with a custom save interval.
+    #[must_use]
+    pub fn with_save_interval(
+        state_manager: StateManager,
+        job: DownloadJob,
+        save_interval: Duration,
+    ) -> Self {
+        Self {
+            state_manager,
+            job: Arc::new(RwLock::new(job)),
+            save_interval,
+            last_save: std::sync::Mutex::new(Instant::now()),
+        }
+    }
+
+    /// Update progress for a specific task.
+    ///
+    /// This updates the hours completed and ticks downloaded for the task
+    /// at the given index. If enough time has passed since the last save,
+    /// the state will be checkpointed to disk.
+    ///
+    /// # Arguments
+    ///
+    /// * `task_idx` - Index of the task to update
+    /// * `hours` - Number of hours completed
+    /// * `ticks` - Number of ticks downloaded
+    pub async fn update_task_progress(&self, task_idx: usize, hours: u64, ticks: u64) {
+        {
+            let mut job = self.job.write().await;
+            if let Some(task) = job.tasks.get_mut(task_idx) {
+                task.hours_completed = hours as u32;
+                task.ticks_downloaded = ticks;
+                if task.status == JobStatus::Pending {
+                    task.status = JobStatus::Running;
+                }
+            }
+        }
+
+        // Check if we should save
+        self.maybe_save_checkpoint().await;
+    }
+
+    /// Mark a task as completed.
+    ///
+    /// This updates the task status to `Completed` and records the
+    /// final byte count.
+    ///
+    /// # Arguments
+    ///
+    /// * `task_idx` - Index of the task to mark as completed
+    /// * `bytes` - Total bytes written for this task
+    pub async fn mark_task_completed(&self, task_idx: usize, bytes: u64) {
+        {
+            let mut job = self.job.write().await;
+            if let Some(task) = job.tasks.get_mut(task_idx) {
+                task.status = JobStatus::Completed;
+                task.bytes_written = bytes;
+                task.hours_completed = task.hours_total;
+            }
+        }
+
+        // Always save on task completion
+        let _ = self.save_checkpoint().await;
+    }
+
+    /// Mark a task as failed.
+    ///
+    /// This updates the task status to `Failed` and records the error message.
+    ///
+    /// # Arguments
+    ///
+    /// * `task_idx` - Index of the task to mark as failed
+    /// * `error` - Error message describing the failure
+    pub async fn mark_task_failed(&self, task_idx: usize, error: &str) {
+        {
+            let mut job = self.job.write().await;
+            if let Some(task) = job.tasks.get_mut(task_idx) {
+                task.status = JobStatus::Failed;
+                task.error_message = Some(error.to_string());
+            }
+        }
+
+        // Always save on task failure
+        let _ = self.save_checkpoint().await;
+    }
+
+    /// Mark a task as running.
+    ///
+    /// This updates the task status to `Running`.
+    ///
+    /// # Arguments
+    ///
+    /// * `task_idx` - Index of the task to mark as running
+    pub async fn mark_task_running(&self, task_idx: usize) {
+        {
+            let mut job = self.job.write().await;
+            if let Some(task) = job.tasks.get_mut(task_idx) {
+                task.status = JobStatus::Running;
+            }
+        }
+
+        // Save when task starts
+        let _ = self.save_checkpoint().await;
+    }
+
+    /// Mark the entire job as completed.
+    ///
+    /// Call this when all tasks have finished successfully.
+    pub async fn mark_job_completed(&self) {
+        {
+            let mut job = self.job.write().await;
+            job.mark_completed();
+        }
+
+        // Always save on job completion
+        let _ = self.save_checkpoint().await;
+    }
+
+    /// Mark the entire job as failed.
+    ///
+    /// Call this when the job fails due to a critical error.
+    ///
+    /// # Arguments
+    ///
+    /// * `error` - Error message describing the failure
+    pub async fn mark_job_failed(&self, error: &str) {
+        {
+            let mut job = self.job.write().await;
+            job.mark_failed(Some(error.to_string()));
+        }
+
+        // Always save on job failure
+        let _ = self.save_checkpoint().await;
+    }
+
+    /// Save current progress to disk (called periodically).
+    ///
+    /// This forces a checkpoint save regardless of the save interval.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the state cannot be saved to disk.
+    pub async fn save_checkpoint(&self) -> Result<(), StateError> {
+        let job = self.job.read().await;
+        self.state_manager.save_job(&job)?;
+
+        // Update last save time
+        if let Ok(mut last_save) = self.last_save.lock() {
+            *last_save = Instant::now();
+        }
+
+        Ok(())
+    }
+
+    /// Check if enough time has passed and save if needed.
+    async fn maybe_save_checkpoint(&self) {
+        let should_save = self
+            .last_save
+            .lock()
+            .map_or(true, |last_save| last_save.elapsed() >= self.save_interval);
+
+        if should_save {
+            let _ = self.save_checkpoint().await;
+        }
+    }
+
+    /// Get current job state.
+    ///
+    /// Returns a clone of the current job state.
+    pub async fn job(&self) -> DownloadJob {
+        self.job.read().await.clone()
+    }
+
+    /// Get the number of completed tasks.
+    pub async fn completed_tasks(&self) -> usize {
+        let job = self.job.read().await;
+        job.tasks
+            .iter()
+            .filter(|t| t.status == JobStatus::Completed)
+            .count()
+    }
+
+    /// Get the number of failed tasks.
+    pub async fn failed_tasks(&self) -> usize {
+        let job = self.job.read().await;
+        job.tasks
+            .iter()
+            .filter(|t| t.status == JobStatus::Failed)
+            .count()
+    }
+
+    /// Get the total number of tasks.
+    pub async fn total_tasks(&self) -> usize {
+        let job = self.job.read().await;
+        job.tasks.len()
+    }
+
+    /// Get the current progress percentage.
+    pub async fn progress_percent(&self) -> f64 {
+        let job = self.job.read().await;
+        job.progress_percent()
+    }
+
+    /// Check if all tasks are finished.
+    pub async fn all_tasks_finished(&self) -> bool {
+        let job = self.job.read().await;
+        job.tasks.iter().all(|t| t.status.is_finished())
+    }
+
+    /// Returns a reference to the state manager.
+    #[must_use]
+    pub const fn state_manager(&self) -> &StateManager {
+        &self.state_manager
+    }
+}
+
+impl Clone for DaemonProgress {
+    fn clone(&self) -> Self {
+        Self {
+            state_manager: self.state_manager.clone(),
+            job: Arc::clone(&self.job),
+            save_interval: self.save_interval,
+            last_save: std::sync::Mutex::new(self.last_save.lock().map_or(Instant::now(), |g| *g)),
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::InstrumentTask;
+    use std::path::PathBuf;
+    use tempfile::TempDir;
+
+    fn create_test_job() -> DownloadJob {
+        let tasks = vec![
+            InstrumentTask::new(
+                "EURUSD".to_string(),
+                "2024-01-01".to_string(),
+                "2024-01-02".to_string(),
+                PathBuf::from("/tmp/eurusd.csv"),
+                "csv".to_string(),
+                "tick".to_string(),
+                48,
+            ),
+            InstrumentTask::new(
+                "GBPUSD".to_string(),
+                "2024-01-01".to_string(),
+                "2024-01-02".to_string(),
+                PathBuf::from("/tmp/gbpusd.csv"),
+                "csv".to_string(),
+                "tick".to_string(),
+                48,
+            ),
+        ];
+        DownloadJob::new(tasks, 4)
+    }
+
+    #[tokio::test]
+    async fn test_progress_tracker_creation() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress = DaemonProgress::new(state_manager, job);
+        let current = progress.job().await;
+
+        assert_eq!(current.tasks.len(), 2);
+        assert_eq!(current.status, JobStatus::Pending);
+    }
+
+    #[tokio::test]
+    async fn test_update_task_progress() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress = DaemonProgress::new(state_manager, job);
+
+        progress.update_task_progress(0, 24, 1_000_000).await;
+
+        let current = progress.job().await;
+        assert_eq!(current.tasks[0].hours_completed, 24);
+        assert_eq!(current.tasks[0].ticks_downloaded, 1_000_000);
+        assert_eq!(current.tasks[0].status, JobStatus::Running);
+    }
+
+    #[tokio::test]
+    async fn test_mark_task_completed() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+        let job_id = job.id;
+
+        let progress = DaemonProgress::new(state_manager.clone(), job);
+
+        progress.mark_task_completed(0, 1024 * 1024).await;
+
+        let current = progress.job().await;
+        assert_eq!(current.tasks[0].status, JobStatus::Completed);
+        assert_eq!(current.tasks[0].bytes_written, 1024 * 1024);
+        assert_eq!(current.tasks[0].hours_completed, 48); // Should be set to total
+
+        // Verify saved to disk
+        let loaded = state_manager.load_job(job_id).unwrap();
+        assert_eq!(loaded.tasks[0].status, JobStatus::Completed);
+    }
+
+    #[tokio::test]
+    async fn test_mark_task_failed() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+        let job_id = job.id;
+
+        let progress = DaemonProgress::new(state_manager.clone(), job);
+
+        progress.mark_task_failed(0, "Connection timeout").await;
+
+        let current = progress.job().await;
+        assert_eq!(current.tasks[0].status, JobStatus::Failed);
+        assert_eq!(
+            current.tasks[0].error_message,
+            Some("Connection timeout".to_string())
+        );
+
+        // Verify saved to disk
+        let loaded = state_manager.load_job(job_id).unwrap();
+        assert_eq!(loaded.tasks[0].status, JobStatus::Failed);
+    }
+
+    #[tokio::test]
+    async fn test_mark_job_completed() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress = DaemonProgress::new(state_manager, job);
+
+        progress.mark_job_completed().await;
+
+        let current = progress.job().await;
+        assert_eq!(current.status, JobStatus::Completed);
+        assert!(current.completed_at.is_some());
+    }
+
+    #[tokio::test]
+    async fn test_mark_job_failed() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress = DaemonProgress::new(state_manager, job);
+
+        progress.mark_job_failed("Fatal error").await;
+
+        let current = progress.job().await;
+        assert_eq!(current.status, JobStatus::Failed);
+        assert!(current.completed_at.is_some());
+    }
+
+    #[tokio::test]
+    async fn test_completed_tasks_count() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress = DaemonProgress::new(state_manager, job);
+
+        assert_eq!(progress.completed_tasks().await, 0);
+
+        progress.mark_task_completed(0, 1024).await;
+        assert_eq!(progress.completed_tasks().await, 1);
+
+        progress.mark_task_completed(1, 2048).await;
+        assert_eq!(progress.completed_tasks().await, 2);
+    }
+
+    #[tokio::test]
+    async fn test_all_tasks_finished() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress = DaemonProgress::new(state_manager, job);
+
+        assert!(!progress.all_tasks_finished().await);
+
+        progress.mark_task_completed(0, 1024).await;
+        assert!(!progress.all_tasks_finished().await);
+
+        progress.mark_task_failed(1, "error").await;
+        assert!(progress.all_tasks_finished().await);
+    }
+
+    #[tokio::test]
+    async fn test_progress_percent() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress = DaemonProgress::new(state_manager, job);
+
+        // Initially 0%
+        assert!((progress.progress_percent().await - 0.0).abs() < 0.001);
+
+        // Update first task to 50%
+        progress.update_task_progress(0, 24, 100).await;
+        // 24/96 total hours = 25%
+        assert!((progress.progress_percent().await - 25.0).abs() < 0.001);
+    }
+
+    #[tokio::test]
+    async fn test_custom_save_interval() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress =
+            DaemonProgress::with_save_interval(state_manager, job, Duration::from_secs(1));
+
+        assert_eq!(progress.save_interval, Duration::from_secs(1));
+    }
+
+    #[tokio::test]
+    async fn test_progress_clone() {
+        let temp_dir = TempDir::new().unwrap();
+        let state_manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+        let job = create_test_job();
+
+        let progress = DaemonProgress::new(state_manager, job);
+        let cloned = progress.clone();
+
+        // Both should share the same job state
+        progress.update_task_progress(0, 10, 100).await;
+
+        let original_job = progress.job().await;
+        let cloned_job = cloned.job().await;
+
+        assert_eq!(
+            original_job.tasks[0].hours_completed,
+            cloned_job.tasks[0].hours_completed
+        );
+    }
+}

crates/paracas-daemon/src/state.rs

@@ -0,0 +1,499 @@
+//! State management for persistent job storage.
+
+use crate::{DownloadJob, JobId, JobStatus};
+use directories::ProjectDirs;
+use std::fs;
+use std::path::{Path, PathBuf};
+use thiserror::Error;
+
+/// Errors that can occur during state management operations.
+#[derive(Error, Debug)]
+pub enum StateError {
+    /// Failed to determine the application data directory.
+    #[error("Failed to determine application data directory")]
+    NoDataDir,
+
+    /// Failed to create a directory.
+    #[error("Failed to create directory '{path}': {source}")]
+    CreateDir {
+        /// The path that could not be created.
+        path: PathBuf,
+        /// The underlying I/O error.
+        source: std::io::Error,
+    },
+
+    /// Failed to read a file.
+    #[error("Failed to read file '{path}': {source}")]
+    ReadFile {
+        /// The path that could not be read.
+        path: PathBuf,
+        /// The underlying I/O error.
+        source: std::io::Error,
+    },
+
+    /// Failed to write a file.
+    #[error("Failed to write file '{path}': {source}")]
+    WriteFile {
+        /// The path that could not be written.
+        path: PathBuf,
+        /// The underlying I/O error.
+        source: std::io::Error,
+    },
+
+    /// Failed to delete a file.
+    #[error("Failed to delete file '{path}': {source}")]
+    DeleteFile {
+        /// The path that could not be deleted.
+        path: PathBuf,
+        /// The underlying I/O error.
+        source: std::io::Error,
+    },
+
+    /// Failed to parse JSON.
+    #[error("Failed to parse job file '{path}': {source}")]
+    ParseJson {
+        /// The path that could not be parsed.
+        path: PathBuf,
+        /// The underlying JSON error.
+        source: serde_json::Error,
+    },
+
+    /// Failed to serialize JSON.
+    #[error("Failed to serialize job: {0}")]
+    SerializeJson(#[from] serde_json::Error),
+
+    /// Job not found.
+    #[error("Job not found: {0}")]
+    JobNotFound(JobId),
+
+    /// Failed to read directory.
+    #[error("Failed to read directory '{path}': {source}")]
+    ReadDir {
+        /// The path that could not be read.
+        path: PathBuf,
+        /// The underlying I/O error.
+        source: std::io::Error,
+    },
+
+    /// Failed to spawn daemon process.
+    #[error("Failed to spawn daemon process '{executable}': {source}")]
+    SpawnDaemon {
+        /// The executable that could not be spawned.
+        executable: PathBuf,
+        /// The underlying I/O error.
+        source: std::io::Error,
+    },
+
+    /// Failed to determine executable path.
+    #[error("Failed to determine executable path: {source}")]
+    ExecutablePath {
+        /// The underlying I/O error.
+        source: std::io::Error,
+    },
+}
+
+/// Result type for state operations.
+pub type Result<T> = std::result::Result<T, StateError>;
+
+/// Manages persistent state for download jobs.
+///
+/// Jobs are stored as JSON files in `~/.paracas/jobs/` with log files
+/// stored in `~/.paracas/logs/`.
+#[derive(Debug, Clone)]
+pub struct StateManager {
+    /// Base directory for state storage.
+    base_path: PathBuf,
+    /// Directory for job JSON files.
+    jobs_path: PathBuf,
+    /// Directory for job log files.
+    logs_path: PathBuf,
+}
+
+impl StateManager {
+    /// Creates a new state manager with the given base path.
+    ///
+    /// Creates the necessary subdirectories if they don't exist.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the directories cannot be created.
+    pub fn new(base_path: PathBuf) -> Result<Self> {
+        let jobs_path = base_path.join("jobs");
+        let logs_path = base_path.join("logs");
+
+        // Create directories if they don't exist
+        for path in [&base_path, &jobs_path, &logs_path] {
+            if !path.exists() {
+                fs::create_dir_all(path).map_err(|e| StateError::CreateDir {
+                    path: path.clone(),
+                    source: e,
+                })?;
+            }
+        }
+
+        Ok(Self {
+            base_path,
+            jobs_path,
+            logs_path,
+        })
+    }
+
+    /// Returns the default path for paracas state storage.
+    ///
+    /// Uses the `directories` crate to find the appropriate location:
+    /// - Linux: `~/.local/share/paracas/`
+    /// - macOS: `~/Library/Application Support/paracas/`
+    /// - Windows: `C:\Users\<User>\AppData\Roaming\paracas\`
+    ///
+    /// Falls back to `~/.paracas/` if the platform-specific location
+    /// cannot be determined.
+    #[must_use]
+    pub fn default_path() -> PathBuf {
+        ProjectDirs::from("", "", "paracas").map_or_else(dirs_fallback, |proj_dirs| {
+            proj_dirs.data_dir().to_path_buf()
+        })
+    }
+
+    /// Creates a state manager at the default path.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the directories cannot be created.
+    pub fn with_default_path() -> Result<Self> {
+        Self::new(Self::default_path())
+    }
+
+    /// Returns the base path for state storage.
+    #[must_use]
+    pub fn base_path(&self) -> &Path {
+        &self.base_path
+    }
+
+    /// Returns the path to a job's state file.
+    #[must_use]
+    pub fn job_state_path(&self, job_id: JobId) -> PathBuf {
+        self.jobs_path.join(format!("{job_id}.json"))
+    }
+
+    /// Returns the path to a job's log file.
+    #[must_use]
+    pub fn job_log_path(&self, job_id: JobId) -> PathBuf {
+        self.logs_path.join(format!("{job_id}.log"))
+    }
+
+    /// Saves a job to persistent storage.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the job cannot be serialized or written to disk.
+    pub fn save_job(&self, job: &DownloadJob) -> Result<()> {
+        let path = self.job_state_path(job.id);
+        let json = serde_json::to_string_pretty(job)?;
+
+        fs::write(&path, json).map_err(|e| StateError::WriteFile { path, source: e })
+    }
+
+    /// Loads a job from persistent storage.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the job file cannot be read or parsed.
+    pub fn load_job(&self, job_id: JobId) -> Result<DownloadJob> {
+        let path = self.job_state_path(job_id);
+
+        if !path.exists() {
+            return Err(StateError::JobNotFound(job_id));
+        }
+
+        let content = fs::read_to_string(&path).map_err(|e| StateError::ReadFile {
+            path: path.clone(),
+            source: e,
+        })?;
+
+        serde_json::from_str(&content).map_err(|e| StateError::ParseJson { path, source: e })
+    }
+
+    /// Lists all jobs in persistent storage.
+    ///
+    /// Returns jobs sorted by creation time (newest first).
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the jobs directory cannot be read.
+    pub fn list_jobs(&self) -> Result<Vec<DownloadJob>> {
+        let entries = fs::read_dir(&self.jobs_path).map_err(|e| StateError::ReadDir {
+            path: self.jobs_path.clone(),
+            source: e,
+        })?;
+
+        let mut jobs = Vec::new();
+
+        for entry in entries {
+            let entry = entry.map_err(|e| StateError::ReadDir {
+                path: self.jobs_path.clone(),
+                source: e,
+            })?;
+
+            let path = entry.path();
+            if path.extension().is_some_and(|ext| ext == "json") {
+                let content = fs::read_to_string(&path).map_err(|e| StateError::ReadFile {
+                    path: path.clone(),
+                    source: e,
+                })?;
+
+                match serde_json::from_str::<DownloadJob>(&content) {
+                    Ok(job) => jobs.push(job),
+                    Err(e) => {
+                        // Log warning but continue - don't fail on corrupt files
+                        eprintln!("Warning: Failed to parse job file {:?}: {}", path, e);
+                    }
+                }
+            }
+        }
+
+        // Sort by creation time, newest first
+        jobs.sort_by(|a, b| b.created_at.cmp(&a.created_at));
+
+        Ok(jobs)
+    }
+
+    /// Deletes a job from persistent storage.
+    ///
+    /// Also deletes the associated log file if it exists.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the job file cannot be deleted.
+    pub fn delete_job(&self, job_id: JobId) -> Result<()> {
+        let state_path = self.job_state_path(job_id);
+
+        if !state_path.exists() {
+            return Err(StateError::JobNotFound(job_id));
+        }
+
+        fs::remove_file(&state_path).map_err(|e| StateError::DeleteFile {
+            path: state_path,
+            source: e,
+        })?;
+
+        // Also delete log file if it exists
+        let log_path = self.job_log_path(job_id);
+        if log_path.exists() {
+            let _ = fs::remove_file(&log_path); // Ignore errors for log file
+        }
+
+        Ok(())
+    }
+
+    /// Returns all active (pending or running) jobs.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if jobs cannot be listed.
+    pub fn active_jobs(&self) -> Result<Vec<DownloadJob>> {
+        let jobs = self.list_jobs()?;
+        Ok(jobs.into_iter().filter(|j| !j.is_finished()).collect())
+    }
+
+    /// Checks if a process with the given PID is still running.
+    #[must_use]
+    pub fn is_process_running(pid: u32) -> bool {
+        // Use kill with signal 0 to check if process exists
+        // This doesn't actually send a signal, just checks if the process exists
+        #[cfg(unix)]
+        {
+            use std::process::Command;
+            Command::new("kill")
+                .args(["-0", &pid.to_string()])
+                .output()
+                .map(|output| output.status.success())
+                .unwrap_or(false)
+        }
+
+        #[cfg(windows)]
+        {
+            // On Windows, use tasklist to check if process exists
+            use std::process::Command;
+            Command::new("tasklist")
+                .args(["/FI", &format!("PID eq {}", pid)])
+                .output()
+                .map(|output| {
+                    let stdout = String::from_utf8_lossy(&output.stdout);
+                    stdout.contains(&pid.to_string())
+                })
+                .unwrap_or(false)
+        }
+
+        #[cfg(not(any(unix, windows)))]
+        {
+            // On other platforms, assume the process is not running
+            false
+        }
+    }
+
+    /// Cleans up stale jobs where the process is no longer running.
+    ///
+    /// Marks running jobs as failed if their daemon process has died.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if jobs cannot be listed or updated.
+    pub fn cleanup_stale_jobs(&self) -> Result<Vec<JobId>> {
+        let jobs = self.list_jobs()?;
+        let mut cleaned = Vec::new();
+
+        for mut job in jobs {
+            if job.status == JobStatus::Running {
+                let is_stale = job.pid.is_none_or(|pid| !Self::is_process_running(pid));
+
+                if is_stale {
+                    job.mark_failed(Some("Daemon process died unexpectedly".to_string()));
+                    self.save_job(&job)?;
+                    cleaned.push(job.id);
+                }
+            }
+        }
+
+        Ok(cleaned)
+    }
+}
+
+/// Fallback for determining home directory.
+fn dirs_fallback() -> PathBuf {
+    std::env::var("HOME")
+        .map(PathBuf::from)
+        .unwrap_or_else(|_| PathBuf::from("."))
+        .join(".paracas")
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::InstrumentTask;
+    use tempfile::TempDir;
+
+    fn create_test_job() -> DownloadJob {
+        let tasks = vec![InstrumentTask::new(
+            "EURUSD".to_string(),
+            "2024-01-01".to_string(),
+            "2024-01-02".to_string(),
+            PathBuf::from("/tmp/eurusd.csv"),
+            "csv".to_string(),
+            "tick".to_string(),
+            48,
+        )];
+        DownloadJob::new(tasks, 4)
+    }
+
+    #[test]
+    fn test_state_manager_creation() {
+        let temp_dir = TempDir::new().unwrap();
+        let manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        assert!(manager.base_path().exists());
+        assert!(temp_dir.path().join("jobs").exists());
+        assert!(temp_dir.path().join("logs").exists());
+    }
+
+    #[test]
+    fn test_save_and_load_job() {
+        let temp_dir = TempDir::new().unwrap();
+        let manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        let job = create_test_job();
+        let job_id = job.id;
+
+        manager.save_job(&job).unwrap();
+
+        let loaded = manager.load_job(job_id).unwrap();
+        assert_eq!(loaded.id, job_id);
+        assert_eq!(loaded.status, JobStatus::Pending);
+        assert_eq!(loaded.tasks.len(), 1);
+    }
+
+    #[test]
+    fn test_list_jobs() {
+        let temp_dir = TempDir::new().unwrap();
+        let manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        let job1 = create_test_job();
+        let job2 = create_test_job();
+
+        manager.save_job(&job1).unwrap();
+        manager.save_job(&job2).unwrap();
+
+        let jobs = manager.list_jobs().unwrap();
+        assert_eq!(jobs.len(), 2);
+    }
+
+    #[test]
+    fn test_delete_job() {
+        let temp_dir = TempDir::new().unwrap();
+        let manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        let job = create_test_job();
+        let job_id = job.id;
+
+        manager.save_job(&job).unwrap();
+        assert!(manager.load_job(job_id).is_ok());
+
+        manager.delete_job(job_id).unwrap();
+        assert!(matches!(
+            manager.load_job(job_id),
+            Err(StateError::JobNotFound(_))
+        ));
+    }
+
+    #[test]
+    fn test_active_jobs() {
+        let temp_dir = TempDir::new().unwrap();
+        let manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        let mut pending_job = create_test_job();
+        let mut completed_job = create_test_job();
+        completed_job.mark_completed();
+
+        manager.save_job(&pending_job).unwrap();
+        manager.save_job(&completed_job).unwrap();
+
+        pending_job.mark_started(12345);
+        manager.save_job(&pending_job).unwrap();
+
+        let active = manager.active_jobs().unwrap();
+        assert_eq!(active.len(), 1);
+        assert_eq!(active[0].id, pending_job.id);
+    }
+
+    #[test]
+    fn test_job_not_found() {
+        let temp_dir = TempDir::new().unwrap();
+        let manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        let result = manager.load_job(uuid::Uuid::new_v4());
+        assert!(matches!(result, Err(StateError::JobNotFound(_))));
+    }
+
+    #[test]
+    fn test_job_state_path() {
+        let temp_dir = TempDir::new().unwrap();
+        let manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        let job_id = uuid::Uuid::new_v4();
+        let path = manager.job_state_path(job_id);
+
+        assert!(path.to_string_lossy().contains("jobs"));
+        assert!(path.to_string_lossy().ends_with(".json"));
+    }
+
+    #[test]
+    fn test_job_log_path() {
+        let temp_dir = TempDir::new().unwrap();
+        let manager = StateManager::new(temp_dir.path().to_path_buf()).unwrap();
+
+        let job_id = uuid::Uuid::new_v4();
+        let path = manager.job_log_path(job_id);
+
+        assert!(path.to_string_lossy().contains("logs"));
+        assert!(path.to_string_lossy().ends_with(".log"));
+    }
+}

crates/paracas-estimate/Cargo.toml

@@ -0,0 +1,18 @@
+[package]
+name = "paracas-estimate"
+version.workspace = true
+edition.workspace = true
+rust-version.workspace = true
+license.workspace = true
+homepage.workspace = true
+repository.workspace = true
+description = "Download size and time estimation for paracas tick data downloader"
+
+[lints]
+workspace = true
+
+[dependencies]
+paracas-types = { workspace = true }
+serde = { workspace = true }
+serde_json = { workspace = true }
+chrono = { workspace = true }

crates/paracas-estimate/data/size_estimates.json

@@ -0,0 +1,39 @@
+{
+  "categories": {
+    "forex": {
+      "avg_compressed_bytes_per_hour": 75000,
+      "avg_ticks_per_hour": 5000,
+      "peak_multiplier": 2.5
+    },
+    "crypto": {
+      "avg_compressed_bytes_per_hour": 150000,
+      "avg_ticks_per_hour": 10000,
+      "peak_multiplier": 3.0
+    },
+    "index": {
+      "avg_compressed_bytes_per_hour": 50000,
+      "avg_ticks_per_hour": 3000,
+      "peak_multiplier": 2.0
+    },
+    "commodity": {
+      "avg_compressed_bytes_per_hour": 40000,
+      "avg_ticks_per_hour": 2500,
+      "peak_multiplier": 2.0
+    },
+    "stock": {
+      "avg_compressed_bytes_per_hour": 30000,
+      "avg_ticks_per_hour": 2000,
+      "peak_multiplier": 1.5
+    },
+    "etf": {
+      "avg_compressed_bytes_per_hour": 25000,
+      "avg_ticks_per_hour": 1500,
+      "peak_multiplier": 1.5
+    },
+    "bond": {
+      "avg_compressed_bytes_per_hour": 20000,
+      "avg_ticks_per_hour": 1000,
+      "peak_multiplier": 1.5
+    }
+  }
+}

crates/paracas-estimate/src/data.rs

@@ -0,0 +1,186 @@
+//! Estimate database with historical averages.
+
+use std::collections::HashMap;
+use std::sync::OnceLock;
+
+use serde::{Deserialize, Serialize};
+
+/// Embedded JSON data with historical size estimates.
+const SIZE_ESTIMATES_JSON: &str = include_str!("../data/size_estimates.json");
+
+/// Static estimate database instance.
+static ESTIMATES: OnceLock<EstimateDatabase> = OnceLock::new();
+
+/// Size estimate for a single instrument category.
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub struct CategoryEstimate {
+    /// Category name (e.g., "forex", "crypto").
+    pub category: String,
+    /// Average compressed bytes per hour of data.
+    pub avg_compressed_bytes_per_hour: u64,
+    /// Average number of ticks per hour.
+    pub avg_ticks_per_hour: u64,
+    /// Multiplier for peak trading hours.
+    pub peak_multiplier: f64,
+}
+
+impl CategoryEstimate {
+    /// Creates a new category estimate.
+    #[must_use]
+    pub fn new(
+        category: impl Into<String>,
+        avg_compressed_bytes_per_hour: u64,
+        avg_ticks_per_hour: u64,
+        peak_multiplier: f64,
+    ) -> Self {
+        Self {
+            category: category.into(),
+            avg_compressed_bytes_per_hour,
+            avg_ticks_per_hour,
+            peak_multiplier,
+        }
+    }
+
+    /// Returns the maximum compressed bytes per hour (at peak).
+    #[must_use]
+    pub fn max_compressed_bytes_per_hour(&self) -> u64 {
+        (self.avg_compressed_bytes_per_hour as f64 * self.peak_multiplier) as u64
+    }
+
+    /// Returns the maximum ticks per hour (at peak).
+    #[must_use]
+    pub fn max_ticks_per_hour(&self) -> u64 {
+        (self.avg_ticks_per_hour as f64 * self.peak_multiplier) as u64
+    }
+}
+
+/// Raw JSON structure for deserialization.
+#[derive(Debug, Deserialize)]
+struct RawEstimateData {
+    categories: HashMap<String, RawCategoryEstimate>,
+}
+
+/// Raw category estimate from JSON.
+#[derive(Debug, Deserialize)]
+struct RawCategoryEstimate {
+    avg_compressed_bytes_per_hour: u64,
+    avg_ticks_per_hour: u64,
+    peak_multiplier: f64,
+}
+
+/// Database of historical size estimates per instrument category.
+#[derive(Debug, Clone)]
+pub struct EstimateDatabase {
+    categories: HashMap<String, CategoryEstimate>,
+}
+
+impl EstimateDatabase {
+    /// Returns the global estimate database instance.
+    ///
+    /// This lazily initializes the database from embedded JSON on first access.
+    #[must_use]
+    pub fn global() -> &'static Self {
+        ESTIMATES.get_or_init(|| {
+            Self::from_json(SIZE_ESTIMATES_JSON)
+                .expect("embedded size_estimates.json should be valid")
+        })
+    }
+
+    /// Creates an estimate database from JSON string.
+    ///
+    /// # Errors
+    ///
+    /// Returns an error if the JSON is invalid.
+    pub fn from_json(json: &str) -> Result<Self, serde_json::Error> {
+        let raw: RawEstimateData = serde_json::from_str(json)?;
+        let categories = raw
+            .categories
+            .into_iter()
+            .map(|(name, raw_est)| {
+                let estimate = CategoryEstimate::new(
+                    name.clone(),
+                    raw_est.avg_compressed_bytes_per_hour,
+                    raw_est.avg_ticks_per_hour,
+                    raw_est.peak_multiplier,
+                );
+                (name, estimate)
+            })
+            .collect();
+        Ok(Self { categories })
+    }
+
+    /// Returns the estimate for a category by name.
+    #[must_use]
+    pub fn get(&self, category: &str) -> Option<&CategoryEstimate> {
+        self.categories.get(category)
+    }
+
+    /// Returns all available categories.
+    pub fn categories(&self) -> impl Iterator<Item = &str> {
+        self.categories.keys().map(String::as_str)
+    }
+
+    /// Returns the number of categories in the database.
+    #[must_use]
+    pub fn len(&self) -> usize {
+        self.categories.len()
+    }
+
+    /// Returns true if the database is empty.
+    #[must_use]
+    pub fn is_empty(&self) -> bool {
+        self.categories.is_empty()
+    }
+
+    /// Returns a default estimate for unknown categories.
+    #[must_use]
+    pub fn default_estimate() -> CategoryEstimate {
+        CategoryEstimate::new("unknown", 50000, 3000, 2.0)
+    }
+}
+
+impl Default for EstimateDatabase {
+    fn default() -> Self {
+        Self::global().clone()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_global_database_loads() {
+        let db = EstimateDatabase::global();
+        assert!(!db.is_empty());
+        assert_eq!(db.len(), 7);
+    }
+
+    #[test]
+    fn test_category_estimates_present() {
+        let db = EstimateDatabase::global();
+
+        let forex = db.get("forex").expect("forex should exist");
+        assert_eq!(forex.avg_compressed_bytes_per_hour, 75000);
+        assert_eq!(forex.avg_ticks_per_hour, 5000);
+
+        let crypto = db.get("crypto").expect("crypto should exist");
+        assert_eq!(crypto.avg_compressed_bytes_per_hour, 150000);
+        assert_eq!(crypto.avg_ticks_per_hour, 10000);
+    }
+
+    #[test]
+    fn test_peak_calculations() {
+        let estimate = CategoryEstimate::new("test", 100000, 5000, 2.0);
+
+        assert_eq!(estimate.max_compressed_bytes_per_hour(), 200000);
+        assert_eq!(estimate.max_ticks_per_hour(), 10000);
+    }
+
+    #[test]
+    fn test_default_estimate() {
+        let default = EstimateDatabase::default_estimate();
+        assert_eq!(default.category, "unknown");
+        assert_eq!(default.avg_compressed_bytes_per_hour, 50000);
+    }
+}

crates/paracas-estimate/src/estimator.rs

@@ -0,0 +1,425 @@
+//! Download estimation logic.
+
+use std::sync::OnceLock;
+use std::time::Duration;
+
+use paracas_types::{DateRange, Instrument};
+
+use crate::data::EstimateDatabase;
+
+/// Default download speed assumption in Mbps.
+const DEFAULT_DOWNLOAD_SPEED_MBPS: f64 = 10.0;
+
+/// Compression ratio (uncompressed / compressed).
+const COMPRESSION_RATIO: f64 = 10.0;
+
+/// Static estimator instance.
+static ESTIMATOR: OnceLock<Estimator> = OnceLock::new();
+
+/// Confidence level of the estimate.
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub enum EstimateConfidence {
+    /// High confidence - well-known instrument category with good historical data.
+    High,
+    /// Medium confidence - known category but less data or more variability.
+    Medium,
+    /// Low confidence - unknown category or limited historical data.
+    Low,
+}
+
+impl EstimateConfidence {
+    /// Returns the confidence as a string slice.
+    #[must_use]
+    pub const fn as_str(&self) -> &'static str {
+        match self {
+            Self::High => "high",
+            Self::Medium => "medium",
+            Self::Low => "low",
+        }
+    }
+}
+
+impl std::fmt::Display for EstimateConfidence {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(f, "{}", self.as_str())
+    }
+}
+
+/// Estimated download metrics.
+#[derive(Debug, Clone, PartialEq)]
+pub struct DownloadEstimate {
+    /// Total hours of data to download.
+    pub total_hours: usize,
+    /// Estimated compressed bytes to download.
+    pub estimated_compressed_bytes: u64,
+    /// Estimated uncompressed bytes (compressed * compression ratio).
+    pub estimated_uncompressed_bytes: u64,
+    /// Estimated output file size in bytes.
+    pub estimated_output_bytes: u64,
+    /// Estimated number of ticks.
+    pub estimated_ticks: u64,
+    /// Estimated download duration.
+    pub estimated_duration: Duration,
+    /// Confidence level of the estimate.
+    pub confidence: EstimateConfidence,
+}
+
+impl DownloadEstimate {
+    /// Creates a new download estimate.
+    #[must_use]
+    #[allow(clippy::too_many_arguments)]
+    pub const fn new(
+        total_hours: usize,
+        estimated_compressed_bytes: u64,
+        estimated_uncompressed_bytes: u64,
+        estimated_output_bytes: u64,
+        estimated_ticks: u64,
+        estimated_duration: Duration,
+        confidence: EstimateConfidence,
+    ) -> Self {
+        Self {
+            total_hours,
+            estimated_compressed_bytes,
+            estimated_uncompressed_bytes,
+            estimated_output_bytes,
+            estimated_ticks,
+            estimated_duration,
+            confidence,
+        }
+    }
+
+    /// Creates an empty estimate (zero hours, zero bytes).
+    #[must_use]
+    pub const fn empty() -> Self {
+        Self {
+            total_hours: 0,
+            estimated_compressed_bytes: 0,
+            estimated_uncompressed_bytes: 0,
+            estimated_output_bytes: 0,
+            estimated_ticks: 0,
+            estimated_duration: Duration::ZERO,
+            confidence: EstimateConfidence::High,
+        }
+    }
+}
+
+/// Download size and time estimator.
+#[derive(Debug, Clone)]
+pub struct Estimator {
+    /// Assumed download speed in Mbps.
+    assumed_download_speed_mbps: f64,
+}
+
+impl Estimator {
+    /// Creates a new estimator with the specified download speed.
+    #[must_use]
+    pub const fn new(assumed_download_speed_mbps: f64) -> Self {
+        Self {
+            assumed_download_speed_mbps,
+        }
+    }
+
+    /// Returns the global estimator instance with default settings.
+    #[must_use]
+    pub fn global() -> &'static Self {
+        ESTIMATOR.get_or_init(|| Self::new(DEFAULT_DOWNLOAD_SPEED_MBPS))
+    }
+
+    /// Returns the assumed download speed in Mbps.
+    #[must_use]
+    pub const fn download_speed_mbps(&self) -> f64 {
+        self.assumed_download_speed_mbps
+    }
+
+    /// Estimates download metrics for a single instrument and date range.
+    #[must_use]
+    pub fn estimate_single(
+        &self,
+        instrument: &Instrument,
+        date_range: &DateRange,
+    ) -> DownloadEstimate {
+        let total_hours = date_range.total_hours();
+        let category = instrument.category().as_str();
+
+        let db = EstimateDatabase::global();
+        let (cat_estimate, confidence) = db.get(category).map_or_else(
+            || {
+                (
+                    EstimateDatabase::default_estimate(),
+                    EstimateConfidence::Low,
+                )
+            },
+            |est| (est.clone(), EstimateConfidence::High),
+        );
+
+        self.calculate_estimate(total_hours, &cat_estimate, confidence)
+    }
+
+    /// Estimates download metrics for multiple instruments and date range.
+    #[must_use]
+    pub fn estimate_batch(
+        &self,
+        instruments: &[&Instrument],
+        date_range: &DateRange,
+    ) -> DownloadEstimate {
+        if instruments.is_empty() {
+            return DownloadEstimate::empty();
+        }
+
+        let total_hours = date_range.total_hours();
+        let db = EstimateDatabase::global();
+
+        let mut total_compressed_bytes: u64 = 0;
+        let mut total_ticks: u64 = 0;
+        let mut min_confidence = EstimateConfidence::High;
+
+        for instrument in instruments {
+            let category = instrument.category().as_str();
+            let (cat_estimate, confidence) = db.get(category).map_or_else(
+                || {
+                    (
+                        EstimateDatabase::default_estimate(),
+                        EstimateConfidence::Low,
+                    )
+                },
+                |est| (est.clone(), EstimateConfidence::High),
+            );
+
+            total_compressed_bytes +=
+                cat_estimate.avg_compressed_bytes_per_hour * total_hours as u64;
+            total_ticks += cat_estimate.avg_ticks_per_hour * total_hours as u64;
+
+            // Use the lowest confidence among all instruments
+            if matches!(confidence, EstimateConfidence::Low) {
+                min_confidence = EstimateConfidence::Low;
+            } else if matches!(confidence, EstimateConfidence::Medium)
+                && !matches!(min_confidence, EstimateConfidence::Low)
+            {
+                min_confidence = EstimateConfidence::Medium;
+            }
+        }
+
+        let estimated_uncompressed_bytes =
+            (total_compressed_bytes as f64 * COMPRESSION_RATIO) as u64;
+        let estimated_output_bytes = estimated_uncompressed_bytes;
+        let estimated_duration = self.calculate_duration(total_compressed_bytes);
+
+        DownloadEstimate::new(
+            total_hours * instruments.len(),
+            total_compressed_bytes,
+            estimated_uncompressed_bytes,
+            estimated_output_bytes,
+            total_ticks,
+            estimated_duration,
+            min_confidence,
+        )
+    }
+
+    /// Calculates estimate for a given number of hours and category.
+    fn calculate_estimate(
+        &self,
+        total_hours: usize,
+        cat_estimate: &crate::data::CategoryEstimate,
+        confidence: EstimateConfidence,
+    ) -> DownloadEstimate {
+        let estimated_compressed_bytes =
+            cat_estimate.avg_compressed_bytes_per_hour * total_hours as u64;
+        let estimated_uncompressed_bytes =
+            (estimated_compressed_bytes as f64 * COMPRESSION_RATIO) as u64;
+        let estimated_output_bytes = estimated_uncompressed_bytes;
+        let estimated_ticks = cat_estimate.avg_ticks_per_hour * total_hours as u64;
+        let estimated_duration = self.calculate_duration(estimated_compressed_bytes);
+
+        DownloadEstimate::new(
+            total_hours,
+            estimated_compressed_bytes,
+            estimated_uncompressed_bytes,
+            estimated_output_bytes,
+            estimated_ticks,
+            estimated_duration,
+            confidence,
+        )
+    }
+
+    /// Calculates download duration based on compressed bytes and speed.
+    fn calculate_duration(&self, compressed_bytes: u64) -> Duration {
+        // Convert Mbps to bytes per second
+        let bytes_per_second = self.assumed_download_speed_mbps * 1_000_000.0 / 8.0;
+        let seconds = compressed_bytes as f64 / bytes_per_second;
+        Duration::from_secs_f64(seconds)
+    }
+
+    /// Formats an estimate as a human-readable summary.
+    #[must_use]
+    pub fn format_estimate(estimate: &DownloadEstimate) -> String {
+        format!(
+            "Download: {} compressed, {} uncompressed\n\
+             Ticks: ~{}\n\
+             Duration: {} (at assumed speed)\n\
+             Confidence: {}",
+            Self::format_bytes(estimate.estimated_compressed_bytes),
+            Self::format_bytes(estimate.estimated_uncompressed_bytes),
+            Self::format_ticks(estimate.estimated_ticks),
+            Self::format_duration(estimate.estimated_duration),
+            estimate.confidence,
+        )
+    }
+
+    /// Formats bytes in human-readable form (e.g., "1.5 GB", "250 MB").
+    #[must_use]
+    pub fn format_bytes(bytes: u64) -> String {
+        const KB: u64 = 1024;
+        const MB: u64 = 1024 * KB;
+        const GB: u64 = 1024 * MB;
+        const TB: u64 = 1024 * GB;
+
+        if bytes >= TB {
+            format!("{:.2} TB", bytes as f64 / TB as f64)
+        } else if bytes >= GB {
+            format!("{:.2} GB", bytes as f64 / GB as f64)
+        } else if bytes >= MB {
+            format!("{:.2} MB", bytes as f64 / MB as f64)
+        } else if bytes >= KB {
+            format!("{:.2} KB", bytes as f64 / KB as f64)
+        } else {
+            format!("{} B", bytes)
+        }
+    }
+
+    /// Formats duration in human-readable form (e.g., "2h 30m", "45m").
+    #[must_use]
+    pub fn format_duration(duration: Duration) -> String {
+        let total_secs = duration.as_secs();
+        let hours = total_secs / 3600;
+        let minutes = (total_secs % 3600) / 60;
+        let seconds = total_secs % 60;
+
+        if hours > 0 {
+            if minutes > 0 {
+                format!("{}h {}m", hours, minutes)
+            } else {
+                format!("{}h", hours)
+            }
+        } else if minutes > 0 {
+            if seconds > 0 && minutes < 10 {
+                format!("{}m {}s", minutes, seconds)
+            } else {
+                format!("{}m", minutes)
+            }
+        } else {
+            format!("{}s", seconds)
+        }
+    }
+
+    /// Formats tick count in human-readable form.
+    fn format_ticks(ticks: u64) -> String {
+        if ticks >= 1_000_000_000 {
+            format!("{:.2}B", ticks as f64 / 1_000_000_000.0)
+        } else if ticks >= 1_000_000 {
+            format!("{:.2}M", ticks as f64 / 1_000_000.0)
+        } else if ticks >= 1_000 {
+            format!("{:.2}K", ticks as f64 / 1_000.0)
+        } else {
+            format!("{}", ticks)
+        }
+    }
+}
+
+impl Default for Estimator {
+    fn default() -> Self {
+        Self::new(DEFAULT_DOWNLOAD_SPEED_MBPS)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use chrono::NaiveDate;
+    use paracas_types::Category;
+
+    fn create_test_instrument(category: Category) -> Instrument {
+        Instrument::new(
+            "test",
+            "Test Instrument",
+            "Test description",
+            category,
+            100_000,
+            None,
+        )
+    }
+
+    #[test]
+    fn test_estimate_single_forex() {
+        let estimator = Estimator::default();
+        let instrument = create_test_instrument(Category::Forex);
+        let start = NaiveDate::from_ymd_opt(2024, 1, 1).unwrap();
+        let end = NaiveDate::from_ymd_opt(2024, 1, 1).unwrap();
+        let date_range = DateRange::new(start, end).unwrap();
+
+        let estimate = estimator.estimate_single(&instrument, &date_range);
+
+        assert_eq!(estimate.total_hours, 24);
+        assert_eq!(estimate.estimated_compressed_bytes, 75000 * 24);
+        assert_eq!(estimate.estimated_ticks, 5000 * 24);
+        assert_eq!(estimate.confidence, EstimateConfidence::High);
+    }
+
+    #[test]
+    fn test_estimate_batch() {
+        let estimator = Estimator::default();
+        let forex = create_test_instrument(Category::Forex);
+        let crypto = create_test_instrument(Category::Crypto);
+        let instruments: Vec<&Instrument> = vec![&forex, &crypto];
+        let start = NaiveDate::from_ymd_opt(2024, 1, 1).unwrap();
+        let date_range = DateRange::single_day(start);
+
+        let estimate = estimator.estimate_batch(&instruments, &date_range);
+
+        // 24 hours * 2 instruments
+        assert_eq!(estimate.total_hours, 48);
+        // forex: 75000 * 24 + crypto: 150000 * 24
+        assert_eq!(estimate.estimated_compressed_bytes, (75000 + 150000) * 24);
+    }
+
+    #[test]
+    fn test_format_bytes() {
+        assert_eq!(Estimator::format_bytes(500), "500 B");
+        assert_eq!(Estimator::format_bytes(1536), "1.50 KB");
+        assert_eq!(Estimator::format_bytes(1_572_864), "1.50 MB");
+        assert_eq!(Estimator::format_bytes(1_610_612_736), "1.50 GB");
+    }
+
+    #[test]
+    fn test_format_duration() {
+        assert_eq!(Estimator::format_duration(Duration::from_secs(30)), "30s");
+        assert_eq!(
+            Estimator::format_duration(Duration::from_secs(90)),
+            "1m 30s"
+        );
+        assert_eq!(Estimator::format_duration(Duration::from_secs(3600)), "1h");
+        assert_eq!(
+            Estimator::format_duration(Duration::from_secs(5400)),
+            "1h 30m"
+        );
+    }
+
+    #[test]
+    fn test_empty_batch() {
+        let estimator = Estimator::default();
+        let instruments: Vec<&Instrument> = vec![];
+        let start = NaiveDate::from_ymd_opt(2024, 1, 1).unwrap();
+        let date_range = DateRange::single_day(start);
+
+        let estimate = estimator.estimate_batch(&instruments, &date_range);
+
+        assert_eq!(estimate.total_hours, 0);
+        assert_eq!(estimate.estimated_compressed_bytes, 0);
+    }
+
+    #[test]
+    fn test_estimate_confidence() {
+        assert_eq!(EstimateConfidence::High.as_str(), "high");
+        assert_eq!(EstimateConfidence::Medium.as_str(), "medium");
+        assert_eq!(EstimateConfidence::Low.as_str(), "low");
+    }
+}

crates/paracas-estimate/src/lib.rs

@@ -0,0 +1,21 @@
+//! Download size and time estimation for paracas tick data downloader.
+//!
+//! This crate provides utilities for estimating download sizes and times
+//! based on historical averages for different instrument categories:
+//!
+//! - [`EstimateDatabase`] - Database of historical size estimates per category
+//! - [`CategoryEstimate`] - Size estimates for a single category
+//! - [`Estimator`] - Computes download estimates for instruments and date ranges
+//! - [`DownloadEstimate`] - Estimated download metrics
+//! - [`EstimateConfidence`] - Confidence level of the estimate
+
+#![doc(issue_tracker_base_url = "https://github.com/factordynamics/paracas/issues/")]
+#![cfg_attr(docsrs, feature(doc_cfg, doc_auto_cfg))]
+#![warn(missing_docs)]
+#![forbid(unsafe_code)]
+
+mod data;
+mod estimator;
+
+pub use data::{CategoryEstimate, EstimateDatabase};
+pub use estimator::{DownloadEstimate, EstimateConfidence, Estimator};