git_perf/git/

size_ops.rs

use anyhow::{Context, Result};
use std::collections::HashMap;
use std::io::{BufRead, BufReader, BufWriter, Read, Write};
use std::path::Path;
use std::process::{Command, Stdio};
use std::thread;

use super::git_interop::{create_consolidated_read_branch, get_repository_root};

/// Information about the size of a specific measurement
pub struct MeasurementSizeInfo {
    /// Total bytes for this measurement
    pub total_bytes: u64,
    /// Number of occurrences
    pub count: usize,
}

/// Information about measurement storage size
pub struct NotesSizeInfo {
    /// Total size in bytes
    pub total_bytes: u64,
    /// Number of commits with measurements
    pub note_count: usize,
    /// Optional breakdown by measurement name
    pub by_measurement: Option<HashMap<String, MeasurementSizeInfo>>,
}

/// Get size information for all measurement notes
pub fn get_notes_size(detailed: bool, disk_size: bool) -> Result<NotesSizeInfo> {
    let repo_root =
        get_repository_root().map_err(|e| anyhow::anyhow!("Failed to get repo root: {}", e))?;

    // Create a consolidated read branch to include pending writes
    let read_branch = create_consolidated_read_branch()?;

    let batch_format = if disk_size {
        "%(objectsize:disk)"
    } else {
        "%(objectsize)"
    };

    // Spawn git notes list process using the temporary read branch
    let mut list_notes = Command::new("git")
        .args(["notes", "--ref", read_branch.ref_name(), "list"])
        .current_dir(&repo_root)
        .stdout(Stdio::piped())
        .spawn()
        .context("Failed to spawn git notes list")?;

    let notes_out = list_notes
        .stdout
        .take()
        .context("Failed to take stdout from git notes list")?;

    // Spawn git cat-file process
    let mut cat_file = Command::new("git")
        .args(["cat-file", &format!("--batch-check={}", batch_format)])
        .current_dir(&repo_root)
        .stdin(Stdio::piped())
        .stdout(Stdio::piped())
        .spawn()
        .context("Failed to spawn git cat-file")?;

    let cat_file_in = cat_file
        .stdin
        .take()
        .context("Failed to take stdin from git cat-file")?;
    let cat_file_out = cat_file
        .stdout
        .take()
        .context("Failed to take stdout from git cat-file")?;

    // Spawn a thread to pipe note OIDs from git notes list to git cat-file
    // Also collect the note OIDs for later use in detailed breakdown
    let note_oids_handle = thread::spawn(move || -> Result<Vec<String>> {
        let reader = BufReader::new(notes_out);
        let mut writer = BufWriter::new(cat_file_in);
        let mut note_oids = Vec::new();

        for line in reader.lines() {
            let line = line.context("Failed to read line from git notes list")?;
            if let Some(note_oid) = line.split_whitespace().next() {
                writeln!(writer, "{}", note_oid).context("Failed to write OID to git cat-file")?;
                note_oids.push(note_oid.to_string());
            }
        }
        // writer is dropped here, closing stdin to cat-file
        Ok(note_oids)
    });

    // Read sizes from git cat-file output
    let reader = BufReader::new(cat_file_out);
    let mut sizes = Vec::new();

    for line in reader.lines() {
        let line = line.context("Failed to read line from git cat-file")?;
        let size = line
            .trim()
            .parse::<u64>()
            .with_context(|| format!("Failed to parse size from: {}", line))?;
        sizes.push(size);
    }

    // Wait for processes to complete
    let note_oids = note_oids_handle
        .join()
        .map_err(|_| anyhow::anyhow!("Thread panicked"))?
        .context("Failed to collect note OIDs")?;

    list_notes
        .wait()
        .context("Failed to wait for git notes list")?;
    let cat_file_status = cat_file.wait().context("Failed to wait for git cat-file")?;

    if !cat_file_status.success() {
        anyhow::bail!("git cat-file process failed");
    }

    let note_count = note_oids.len();
    if note_count == 0 {
        return Ok(NotesSizeInfo {
            total_bytes: 0,
            note_count: 0,
            by_measurement: if detailed { Some(HashMap::new()) } else { None },
        });
    }

    if sizes.len() != note_count {
        anyhow::bail!("Expected {} sizes but got {}", note_count, sizes.len());
    }

    let total_bytes: u64 = sizes.iter().sum();

    let mut by_measurement = if detailed { Some(HashMap::new()) } else { None };

    // If detailed breakdown requested, parse measurement names
    if let Some(ref mut by_name) = by_measurement {
        batch_accumulate_measurement_sizes(Path::new(&repo_root), &note_oids, &sizes, by_name)?;
    }

    Ok(NotesSizeInfo {
        total_bytes,
        note_count,
        by_measurement,
    })
}

/// Parse note contents in batch and accumulate sizes by measurement name.
/// Uses a single `git cat-file --batch` process instead of one process per note.
fn batch_accumulate_measurement_sizes(
    repo_root: &Path,
    note_oids: &[String],
    note_sizes: &[u64],
    by_name: &mut HashMap<String, MeasurementSizeInfo>,
) -> Result<()> {
    use crate::serialization::deserialize;

    debug_assert_eq!(note_oids.len(), note_sizes.len());

    if note_oids.is_empty() {
        return Ok(());
    }

    let mut cat_file = Command::new("git")
        .args(["cat-file", "--batch"])
        .current_dir(repo_root)
        .stdin(Stdio::piped())
        .stdout(Stdio::piped())
        .spawn()
        .context("Failed to spawn git cat-file --batch")?;

    let cat_file_in = cat_file
        .stdin
        .take()
        .context("Failed to take stdin from git cat-file --batch")?;
    let cat_file_out = cat_file
        .stdout
        .take()
        .context("Failed to take stdout from git cat-file --batch")?;

    let oids_owned = note_oids.to_vec();
    let writer_handle = thread::spawn(move || -> Result<()> {
        let mut writer = BufWriter::new(cat_file_in);
        for oid in &oids_owned {
            writeln!(writer, "{}", oid).context("Failed to write OID to git cat-file --batch")?;
        }
        Ok(())
    });

    let mut reader = BufReader::new(cat_file_out);
    for &note_size in note_sizes {
        let mut header = String::new();
        reader
            .read_line(&mut header)
            .context("Failed to read header from git cat-file --batch")?;
        let header = header.trim();
        let content_len: usize = header
            .split_whitespace()
            .nth(2)
            .and_then(|s| s.parse().ok())
            .with_context(|| {
                format!(
                    "Failed to parse object size from git cat-file --batch header: {}",
                    header
                )
            })?;

        let mut content_buf = vec![0u8; content_len];
        reader
            .read_exact(&mut content_buf)
            .context("Failed to read object content from git cat-file --batch")?;
        // Consume the trailing newline separator after each object
        let mut newline = [0u8; 1];
        reader
            .read_exact(&mut newline)
            .context("Failed to read trailing newline from git cat-file --batch")?;

        let content = String::from_utf8_lossy(&content_buf);
        let measurements = deserialize(&content);

        if measurements.is_empty() {
            continue;
        }

        // Distribute note size evenly among measurements in this note
        let size_per_measurement = note_size / measurements.len() as u64;

        for measurement in measurements {
            let entry = by_name
                .entry(measurement.name)
                .or_insert(MeasurementSizeInfo {
                    total_bytes: 0,
                    count: 0,
                });
            entry.total_bytes += size_per_measurement;
            entry.count += 1;
        }
    }

    writer_handle
        .join()
        .map_err(|_| anyhow::anyhow!("Writer thread panicked"))??;

    let status = cat_file
        .wait()
        .context("Failed to wait for git cat-file --batch")?;
    if !status.success() {
        anyhow::bail!("git cat-file --batch failed");
    }

    Ok(())
}

/// Git repository statistics from count-objects
pub struct RepoStats {
    /// Number of loose objects
    pub loose_objects: u64,
    /// Size of loose objects in bytes
    pub loose_size: u64,
    /// Number of packed objects
    pub packed_objects: u64,
    /// Size of pack files in bytes
    pub pack_size: u64,
}

/// Get git repository statistics
pub fn get_repo_stats() -> Result<RepoStats> {
    let repo_root =
        get_repository_root().map_err(|e| anyhow::anyhow!("Failed to get repo root: {}", e))?;

    let output = Command::new("git")
        .args(["count-objects", "-v"])
        .current_dir(&repo_root)
        .output()
        .context("Failed to execute git count-objects")?;

    if !output.status.success() {
        let stderr = String::from_utf8_lossy(&output.stderr);
        anyhow::bail!("git count-objects failed: {}", stderr);
    }

    let stdout = String::from_utf8_lossy(&output.stdout);

    let mut loose_objects = 0;
    let mut loose_size = 0; // in KiB from git
    let mut packed_objects = 0;
    let mut pack_size = 0; // in KiB from git

    for line in stdout.lines() {
        let parts: Vec<&str> = line.split(':').collect();
        if parts.len() != 2 {
            continue;
        }

        let key = parts[0].trim();
        let value = parts[1].trim().parse::<u64>().unwrap_or(0);

        match key {
            "count" => loose_objects = value,
            "size" => loose_size = value,
            "in-pack" => packed_objects = value,
            "size-pack" => pack_size = value,
            _ => {}
        }
    }

    Ok(RepoStats {
        loose_objects,
        loose_size: loose_size * 1024, // Convert KiB to bytes
        packed_objects,
        pack_size: pack_size * 1024, // Convert KiB to bytes
    })
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::test_helpers::with_isolated_cwd_git;

    #[test]
    fn test_get_repo_stats_basic() {
        // Test that get_repo_stats works and returns proper values
        with_isolated_cwd_git(|_git_dir| {
            let stats = get_repo_stats().unwrap();

            // Should have some objects after initial commit
            assert!(stats.loose_objects > 0 || stats.packed_objects > 0);

            // Sizes should be multiples of 1024 (tests * 1024 conversion)
            if stats.loose_size > 0 {
                assert_eq!(
                    stats.loose_size % 1024,
                    0,
                    "loose_size should be multiple of 1024"
                );
            }
            if stats.pack_size > 0 {
                assert_eq!(
                    stats.pack_size % 1024,
                    0,
                    "pack_size should be multiple of 1024"
                );
            }
        });
    }

    #[test]
    fn test_get_notes_size_empty_repo() {
        // Test with a repo that has no notes - exercises the empty case
        with_isolated_cwd_git(|_git_dir| {
            let result = get_notes_size(false, false).unwrap();
            assert_eq!(result.total_bytes, 0);
            assert_eq!(result.note_count, 0);
            assert!(result.by_measurement.is_none());
        });
    }

    #[test]
    fn test_get_repo_stats_conversion_factors() {
        // Test that the * 1024 conversion is correctly applied
        with_isolated_cwd_git(|_git_dir| {
            let stats = get_repo_stats().unwrap();

            // Test that loose_size and pack_size are properly converted from KiB to bytes
            // Both should be multiples of 1024
            assert_eq!(
                stats.loose_size % 1024,
                0,
                "loose_size must be multiple of 1024 (bytes conversion from KiB)"
            );
            assert_eq!(
                stats.pack_size % 1024,
                0,
                "pack_size must be multiple of 1024 (bytes conversion from KiB)"
            );

            // If there are loose objects, the size should be reasonable (not zero, not absurdly large)
            if stats.loose_objects > 0 {
                assert!(
                    stats.loose_size > 0,
                    "loose_size should be > 0 if loose_objects > 0"
                );
                assert!(
                    stats.loose_size < 1_000_000_000,
                    "loose_size should be reasonable"
                );
            }
        });
    }

    #[test]
    fn test_get_repo_stats_field_assignments() {
        // Test that all fields are properly assigned from git output
        with_isolated_cwd_git(|_git_dir| {
            let stats = get_repo_stats().unwrap();

            // Verify that fields are assigned (not just defaulted to 0)
            // After creating a repo with an initial commit, we should have objects
            let total_objects = stats.loose_objects + stats.packed_objects;
            assert!(
                total_objects > 0,
                "Should have at least one object from initial commit"
            );

            // Verify the match arms are working by checking expected field types
            // loose_objects should be count
            // loose_size should be size * 1024
            // packed_objects should be in-pack
            // pack_size should be size-pack * 1024

            // Verify fields are properly typed as u64 (not negative types)
            // The fact that we can do arithmetic on them proves the match arms worked
            let _sum =
                stats.loose_objects + stats.loose_size + stats.packed_objects + stats.pack_size;
            assert!(
                _sum >= stats.loose_objects,
                "Arithmetic should work on u64 fields"
            );
        });
    }

    #[test]
    fn test_get_notes_size_with_measurements() {
        use crate::measurement_storage;

        // Test the full flow: add measurements -> get size with detailed breakdown
        with_isolated_cwd_git(|_git_dir| {
            // Add measurements using the public API
            measurement_storage::add("test_metric_1", 42.0, &[]).unwrap();
            measurement_storage::add("test_metric_2", 100.0, &[]).unwrap();
            measurement_storage::add("test_metric_1", 84.0, &[]).unwrap();

            // Get size information with detailed breakdown
            let result = get_notes_size(true, false).unwrap();

            // Should have measurements now
            assert!(
                result.total_bytes > 0,
                "total_bytes should be > 0 after adding measurements"
            );
            assert_eq!(
                result.note_count, 1,
                "Should have 1 note (all measurements on HEAD)"
            );

            // Verify detailed breakdown
            let by_measurement = result
                .by_measurement
                .expect("Should have detailed breakdown");

            // Should have entries for both metrics
            assert!(
                by_measurement.contains_key("test_metric_1"),
                "Should have test_metric_1 in breakdown"
            );
            assert!(
                by_measurement.contains_key("test_metric_2"),
                "Should have test_metric_2 in breakdown"
            );

            // Test metric 1 should have count of 2
            let metric1_info = &by_measurement["test_metric_1"];
            assert_eq!(
                metric1_info.count, 2,
                "test_metric_1 should have 2 occurrences"
            );
            assert!(
                metric1_info.total_bytes > 0,
                "test_metric_1 should have non-zero size"
            );

            // Test metric 2 should have count of 1
            let metric2_info = &by_measurement["test_metric_2"];
            assert_eq!(
                metric2_info.count, 1,
                "test_metric_2 should have 1 occurrence"
            );
            assert!(
                metric2_info.total_bytes > 0,
                "test_metric_2 should have non-zero size"
            );

            // Verify that the size is distributed correctly (note_size / num_measurements)
            // In this case, 3 measurements total, so each should get roughly 1/3 of note size
            let total_from_breakdown: u64 =
                by_measurement.values().map(|info| info.total_bytes).sum();

            // The total from breakdown may not exactly equal total_bytes due to integer division
            // For example: 121 / 3 = 40 per measurement, 40 * 3 = 120 (loses 1 byte)
            // So we verify it's within the number of measurements
            let num_measurements = 3u64;
            assert!(
                result.total_bytes.abs_diff(total_from_breakdown) < num_measurements,
                "Sum of breakdown ({}) should be within {} bytes of total_bytes ({}) due to integer division",
                total_from_breakdown,
                num_measurements,
                result.total_bytes
            );

            // Since we have 3 measurements, each gets result.total_bytes / 3
            let expected_per_measurement = result.total_bytes / num_measurements;
            assert!(
                metric1_info.total_bytes >= expected_per_measurement,
                "test_metric_1 appears twice, should have at least 1/3 of total (appears 2/3 times)"
            );
        });
    }
}