vtcode_core/tools/handlers/

turn_diff_tracker.rs

//! Turn Diff Tracker (from Codex)
//!
//! Aggregates file diffs across multiple apply_patch tool calls within a turn.
//! This provides a unified view of all changes made during a conversation turn.
//!
//! Supports Agent Trace attribution tracking for AI-generated code.

use hashbrown::HashMap;
use std::fmt::Write as _;
use std::path::PathBuf;
use std::sync::Arc;

use crate::config::constants::diff::MAX_PREVIEW_BYTES;
use serde::{Deserialize, Serialize};
use tokio::sync::RwLock;
use vtcode_diff::{DiffOptions, format_unified_diff};

/// Attribution information for a file change (Agent Trace compatible).
#[derive(Clone, Debug, PartialEq, Eq, Default, Serialize, Deserialize)]
#[serde(rename_all = "snake_case")]
pub struct ChangeAttribution {
    /// Model ID in provider/model format (e.g., "anthropic/claude-opus-4").
    pub model_id: Option<String>,
    /// Provider name (e.g., "anthropic", "openai").
    pub provider: Option<String>,
    /// Session ID linking to the conversation.
    pub session_id: Option<String>,
    /// Turn number within the session.
    pub turn_number: Option<u32>,
    /// Contributor type: "ai", "human", "mixed", "unknown".
    pub contributor_type: String,
}

impl ChangeAttribution {
    /// Create AI attribution with model info.
    pub fn ai(model_id: impl Into<String>, provider: impl Into<String>) -> Self {
        Self {
            model_id: Some(model_id.into()),
            provider: Some(provider.into()),
            session_id: None,
            turn_number: None,
            contributor_type: "ai".to_string(),
        }
    }

    /// Create human attribution.
    pub fn human() -> Self {
        Self {
            contributor_type: "human".to_string(),
            ..Default::default()
        }
    }

    /// Create unknown attribution.
    pub fn unknown() -> Self {
        Self {
            contributor_type: "unknown".to_string(),
            ..Default::default()
        }
    }

    /// Add session context.
    pub fn with_session(mut self, session_id: impl Into<String>, turn: u32) -> Self {
        self.session_id = Some(session_id.into());
        self.turn_number = Some(turn);
        self
    }

    /// Get normalized model ID in provider/model format.
    pub fn normalized_model_id(&self) -> Option<String> {
        match (&self.model_id, &self.provider) {
            (Some(model), Some(provider)) if !model.contains('/') => Some(format!("{provider}/{model}")),
            (Some(model), _) => Some(model.clone()),
            _ => None,
        }
    }
}

/// File change types (from Codex protocol)
#[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize)]
pub struct FileChange {
    /// The type of change.
    #[serde(flatten)]
    pub kind: FileChangeKind,
    /// Attribution information (Agent Trace).
    #[serde(skip_serializing_if = "Option::is_none")]
    pub attribution: Option<ChangeAttribution>,
    /// Line range affected (1-indexed, for Agent Trace).
    #[serde(skip_serializing_if = "Option::is_none")]
    pub line_range: Option<(u32, u32)>,
}

/// Kind of file change.
#[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize)]
#[serde(tag = "type", rename_all = "snake_case")]
pub enum FileChangeKind {
    /// New file added
    Add { content: String },
    /// File deleted
    Delete { original_content: String },
    /// File modified
    Update { old_content: String, new_content: String },
    /// File renamed
    Rename {
        new_path: PathBuf,
        old_content: Option<String>,
        new_content: Option<String>,
    },
}

impl FileChange {
    /// Create a new Add change.
    pub fn add(content: impl Into<String>) -> Self {
        Self {
            kind: FileChangeKind::Add { content: content.into() },
            attribution: None,
            line_range: None,
        }
    }

    /// Create a new Delete change.
    pub fn delete(original_content: impl Into<String>) -> Self {
        Self {
            kind: FileChangeKind::Delete { original_content: original_content.into() },
            attribution: None,
            line_range: None,
        }
    }

    /// Create a new Update change.
    pub fn update(old_content: impl Into<String>, new_content: impl Into<String>) -> Self {
        Self {
            kind: FileChangeKind::Update {
                old_content: old_content.into(),
                new_content: new_content.into(),
            },
            attribution: None,
            line_range: None,
        }
    }

    /// Create a new Rename change.
    pub fn rename(new_path: PathBuf, old_content: Option<String>, new_content: Option<String>) -> Self {
        Self {
            kind: FileChangeKind::Rename { new_path, old_content, new_content },
            attribution: None,
            line_range: None,
        }
    }

    /// Add attribution to the change.
    pub fn with_attribution(mut self, attribution: ChangeAttribution) -> Self {
        self.attribution = Some(attribution);
        self
    }

    /// Add line range to the change.
    pub fn with_line_range(mut self, start: u32, end: u32) -> Self {
        self.line_range = Some((start, end));
        self
    }

    /// Get the new content if any
    pub fn new_content(&self) -> Option<&str> {
        match &self.kind {
            FileChangeKind::Add { content } => Some(content),
            FileChangeKind::Update { new_content, .. } => Some(new_content),
            FileChangeKind::Rename { new_content, .. } => new_content.as_deref(),
            FileChangeKind::Delete { .. } => None,
        }
    }

    /// Get the old content if any
    pub fn old_content(&self) -> Option<&str> {
        match &self.kind {
            FileChangeKind::Delete { original_content } => Some(original_content),
            FileChangeKind::Update { old_content, .. } => Some(old_content),
            FileChangeKind::Rename { old_content, .. } => old_content.as_deref(),
            FileChangeKind::Add { .. } => None,
        }
    }

    /// Check if this is an add operation.
    pub fn is_add(&self) -> bool {
        matches!(self.kind, FileChangeKind::Add { .. })
    }

    /// Check if this is a delete operation.
    pub fn is_delete(&self) -> bool {
        matches!(self.kind, FileChangeKind::Delete { .. })
    }

    /// Check if this is an update operation.
    pub fn is_update(&self) -> bool {
        matches!(self.kind, FileChangeKind::Update { .. })
    }

    /// Check if this is a rename operation.
    pub fn is_rename(&self) -> bool {
        matches!(self.kind, FileChangeKind::Rename { .. })
    }

    /// Compute line count for the new content.
    pub fn new_line_count(&self) -> usize {
        self.new_content().map(|c| c.lines().count()).unwrap_or(0)
    }

    /// Convert from legacy FileChange enum (from tool_handler.rs).
    ///
    /// This provides backward compatibility with the older FileChange format.
    pub fn from_legacy(legacy: &super::tool_handler::FileChange, attribution: Option<ChangeAttribution>) -> Self {
        let kind = match legacy {
            super::tool_handler::FileChange::Add { content } => FileChangeKind::Add { content: content.clone() },
            super::tool_handler::FileChange::Delete => FileChangeKind::Delete {
                original_content: String::new(), // Legacy doesn't preserve content
            },
            super::tool_handler::FileChange::Update { old_content, new_content } => FileChangeKind::Update {
                old_content: old_content.clone(),
                new_content: new_content.clone(),
            },
            super::tool_handler::FileChange::Rename { new_path, content } => FileChangeKind::Rename {
                new_path: new_path.clone(),
                old_content: None,
                new_content: content.clone(),
            },
        };
        Self { kind, attribution, line_range: None }
    }

    /// Convert to legacy FileChange enum (for backward compatibility).
    pub fn to_legacy(&self) -> super::tool_handler::FileChange {
        match &self.kind {
            FileChangeKind::Add { content } => super::tool_handler::FileChange::Add { content: content.clone() },
            FileChangeKind::Delete { .. } => super::tool_handler::FileChange::Delete,
            FileChangeKind::Update { old_content, new_content } => super::tool_handler::FileChange::Update {
                old_content: old_content.clone(),
                new_content: new_content.clone(),
            },
            FileChangeKind::Rename { new_path, new_content, .. } => super::tool_handler::FileChange::Rename {
                new_path: new_path.clone(),
                content: new_content.clone(),
            },
        }
    }
}

/// Turn diff tracker for aggregating changes (from Codex)
#[derive(Default)]
pub struct TurnDiffTracker {
    changes: HashMap<PathBuf, FileChange>,
    pending_changes: Option<HashMap<PathBuf, FileChange>>,
    /// Current attribution context for new changes.
    current_attribution: Option<ChangeAttribution>,
}

impl FileChange {
    /// Whether any stored content side exceeds `limit` bytes.
    pub fn exceeds_content_limit(&self, limit: usize) -> bool {
        match &self.kind {
            FileChangeKind::Add { content } => content.len() > limit,
            FileChangeKind::Delete { original_content } => original_content.len() > limit,
            FileChangeKind::Update { old_content, new_content } => {
                old_content.len() > limit || new_content.len() > limit
            }
            FileChangeKind::Rename { old_content, new_content, .. } => {
                old_content.as_deref().is_some_and(|content| content.len() > limit)
                    || new_content.as_deref().is_some_and(|content| content.len() > limit)
            }
        }
    }
}

impl TurnDiffTracker {
    pub fn new() -> Self {
        Self::default()
    }

    /// Set the current attribution context for subsequent changes.
    pub fn set_attribution(&mut self, attribution: ChangeAttribution) {
        self.current_attribution = Some(attribution);
    }

    /// Clear the current attribution context.
    pub fn clear_attribution(&mut self) {
        self.current_attribution = None;
    }

    /// Get the current attribution context.
    pub fn current_attribution(&self) -> Option<&ChangeAttribution> {
        self.current_attribution.as_ref()
    }

    /// Called when a patch application begins (from Codex)
    ///
    /// Stores the pending changes until the patch is confirmed
    pub fn on_patch_begin(&mut self, changes: HashMap<PathBuf, FileChange>) {
        // Apply current attribution to all changes
        let changes_with_attribution: HashMap<PathBuf, FileChange> = changes
            .into_iter()
            .map(|(path, mut change)| {
                if change.attribution.is_none() {
                    change.attribution = self.current_attribution.clone();
                }
                (path, change)
            })
            .collect();
        self.pending_changes = Some(changes_with_attribution);
    }

    /// Called when a patch application ends (from Codex)
    ///
    /// If successful, merges pending changes into the main tracker
    pub fn on_patch_end(&mut self, success: bool) {
        if success {
            if let Some(pending) = self.pending_changes.take() {
                for (path, change) in pending {
                    self.merge_change(path, change);
                }
            }
        } else {
            self.pending_changes = None;
        }
    }

    /// Merge a change into the tracker, combining with existing changes
    fn merge_change(&mut self, path: PathBuf, change: FileChange) {
        if let Some(existing) = self.changes.get(&path) {
            // Merge the changes, preserving the latest attribution
            let merged = match (&existing.kind, &change.kind) {
                // Add then Update = Add with new content
                (FileChangeKind::Add { .. }, FileChangeKind::Update { new_content, .. }) => FileChange {
                    kind: FileChangeKind::Add { content: new_content.clone() },
                    attribution: change.attribution.clone().or(existing.attribution.clone()),
                    line_range: change.line_range,
                },
                // Add then Delete = No change (remove from tracker)
                (FileChangeKind::Add { .. }, FileChangeKind::Delete { .. }) => {
                    self.changes.remove(&path);
                    return;
                }
                // Update then Update = Update with combined old/new
                (FileChangeKind::Update { old_content, .. }, FileChangeKind::Update { new_content, .. }) => {
                    FileChange {
                        kind: FileChangeKind::Update {
                            old_content: old_content.clone(),
                            new_content: new_content.clone(),
                        },
                        attribution: change.attribution.clone().or(existing.attribution.clone()),
                        line_range: change.line_range,
                    }
                }
                // Update then Delete = Delete with original old content
                (FileChangeKind::Update { old_content, .. }, FileChangeKind::Delete { .. }) => FileChange {
                    kind: FileChangeKind::Delete { original_content: old_content.clone() },
                    attribution: change.attribution.clone().or(existing.attribution.clone()),
                    line_range: None,
                },
                // Delete then Add = Update
                (FileChangeKind::Delete { original_content }, FileChangeKind::Add { content }) => FileChange {
                    kind: FileChangeKind::Update {
                        old_content: original_content.clone(),
                        new_content: content.clone(),
                    },
                    attribution: change.attribution.clone().or(existing.attribution.clone()),
                    line_range: change.line_range,
                },
                // Default: use the new change
                _ => change,
            };
            self.changes.insert(path, merged);
        } else {
            self.changes.insert(path, change);
        }
    }

    /// Get all tracked changes
    pub fn changes(&self) -> &HashMap<PathBuf, FileChange> {
        &self.changes
    }

    /// Get pending changes (not yet confirmed)
    pub fn pending_changes(&self) -> Option<&HashMap<PathBuf, FileChange>> {
        self.pending_changes.as_ref()
    }

    /// Check if there are any changes
    pub fn has_changes(&self) -> bool {
        !self.changes.is_empty()
    }

    /// Get unified diff for all tracked changes (from Codex)
    ///
    /// Output is deterministic: entries are emitted in sorted path order.
    /// Entries whose content exceeds [`MAX_PREVIEW_BYTES`] render as a
    /// one-line summary instead of a full diff to keep previews bounded.
    pub fn get_unified_diff(&self) -> String {
        let mut diff = String::new();

        let mut entries: Vec<(&PathBuf, &FileChange)> = self.changes.iter().collect();
        entries.sort_by(|(left, _), (right, _)| (*left).cmp(*right));

        for (path, change) in entries {
            let path_str = path.display();
            if change.exceeds_content_limit(MAX_PREVIEW_BYTES) {
                let _ = write!(
                    diff,
                    "diff --git a/{path_str} b/{path_str}\n@@ summary: content exceeds {MAX_PREVIEW_BYTES} bytes; full diff suppressed\n"
                );
                diff.push('\n');
                continue;
            }
            match &change.kind {
                FileChangeKind::Add { content } => {
                    let new_label = path_str.to_string();
                    diff.push_str(&compute_unified_diff_with_labels("", content, "/dev/null", &new_label));
                }
                FileChangeKind::Delete { original_content } => {
                    let old_label = path_str.to_string();
                    diff.push_str(&compute_unified_diff_with_labels(original_content, "", &old_label, "/dev/null"));
                }
                FileChangeKind::Update { old_content, new_content } => {
                    let filename = path_str.to_string();
                    diff.push_str(&compute_unified_diff_with_labels(old_content, new_content, &filename, &filename));
                }
                FileChangeKind::Rename { new_path, old_content, new_content } => {
                    if let (Some(old), Some(new)) = (old_content, new_content) {
                        let old_label = path_str.to_string();
                        let new_label = new_path.to_string_lossy();
                        diff.push_str(&compute_unified_diff_with_labels(old, new, &old_label, &new_label));
                    }
                }
            }
            diff.push('\n');
        }

        diff
    }

    /// Clear all tracked changes
    pub fn clear(&mut self) {
        self.changes.clear();
        self.pending_changes = None;
    }
}

/// Shared turn diff tracker (thread-safe) (from Codex)
pub type SharedTurnDiffTracker = Arc<RwLock<TurnDiffTracker>>;

/// Create a new shared diff tracker
pub fn new_shared_tracker() -> SharedTurnDiffTracker {
    Arc::new(RwLock::new(TurnDiffTracker::new()))
}

/// Compute unified diff between old and new content
fn compute_unified_diff_with_labels(old: &str, new: &str, old_label: &str, new_label: &str) -> String {
    let old_label = format!("a/{old_label}");
    let new_label = format!("b/{new_label}");
    format_unified_diff(
        old,
        new,
        DiffOptions {
            context_lines: 3,
            old_label: Some(&old_label),
            new_label: Some(&new_label),
            missing_newline_hint: false,
            ..DiffOptions::default()
        },
    )
}

/// Format a diff for newly added content (for testing)
#[cfg(test)]
fn format_addition_diff(content: &str) -> String {
    compute_unified_diff_with_labels("", content, "file", "file")
}

/// Format a diff for deleted content (for testing)
#[cfg(test)]
fn format_deletion_diff(content: &str) -> String {
    compute_unified_diff_with_labels(content, "", "file", "file")
}

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

    #[test]
    fn test_on_patch_begin_and_end_success() {
        let mut tracker = TurnDiffTracker::new();

        let mut changes = HashMap::new();
        changes.insert(PathBuf::from("test.txt"), FileChange::add("hello"));

        tracker.on_patch_begin(changes);
        assert!(tracker.pending_changes().is_some());
        assert!(!tracker.has_changes());

        tracker.on_patch_end(true);
        assert!(tracker.pending_changes().is_none());
        assert!(tracker.has_changes());
    }

    #[test]
    fn test_on_patch_end_failure() {
        let mut tracker = TurnDiffTracker::new();

        let mut changes = HashMap::new();
        changes.insert(PathBuf::from("test.txt"), FileChange::add("hello"));

        tracker.on_patch_begin(changes);
        tracker.on_patch_end(false);

        assert!(tracker.pending_changes().is_none());
        assert!(!tracker.has_changes());
    }

    #[test]
    fn test_merge_add_then_update() {
        let mut tracker = TurnDiffTracker::new();

        // First patch: Add file
        let mut changes1 = HashMap::new();
        changes1.insert(PathBuf::from("test.txt"), FileChange::add("hello"));
        tracker.on_patch_begin(changes1);
        tracker.on_patch_end(true);

        // Second patch: Update file
        let mut changes2 = HashMap::new();
        changes2.insert(PathBuf::from("test.txt"), FileChange::update("hello", "world"));
        tracker.on_patch_begin(changes2);
        tracker.on_patch_end(true);

        // Result should be Add with new content
        let change = tracker.changes().get(&PathBuf::from("test.txt")).unwrap();
        assert!(change.is_add());
        assert_eq!(change.new_content(), Some("world"));
    }

    #[test]
    fn test_merge_add_then_delete() {
        let mut tracker = TurnDiffTracker::new();

        // First patch: Add file
        let mut changes1 = HashMap::new();
        changes1.insert(PathBuf::from("test.txt"), FileChange::add("hello"));
        tracker.on_patch_begin(changes1);
        tracker.on_patch_end(true);

        // Second patch: Delete file
        let mut changes2 = HashMap::new();
        changes2.insert(PathBuf::from("test.txt"), FileChange::delete("hello"));
        tracker.on_patch_begin(changes2);
        tracker.on_patch_end(true);

        // Result: no change (add then delete cancels out)
        assert!(!tracker.has_changes());
    }

    #[test]
    fn test_get_unified_diff() {
        let mut tracker = TurnDiffTracker::new();

        let mut changes = HashMap::new();
        changes.insert(PathBuf::from("new.txt"), FileChange::add("line1\nline2"));
        tracker.on_patch_begin(changes);
        tracker.on_patch_end(true);

        let diff = tracker.get_unified_diff();
        assert!(diff.contains("--- "));
        assert!(diff.contains("+++ "));
        assert!(diff.contains("new.txt"));
        assert!(diff.contains("+line1"));
        assert!(diff.contains("+line2"));
    }

    #[test]
    fn test_attribution_propagation() {
        let mut tracker = TurnDiffTracker::new();
        tracker.set_attribution(ChangeAttribution::ai("claude-opus-4", "anthropic"));

        let mut changes = HashMap::new();
        changes.insert(PathBuf::from("test.txt"), FileChange::add("hello"));
        tracker.on_patch_begin(changes);
        tracker.on_patch_end(true);

        let change = tracker.changes().get(&PathBuf::from("test.txt")).unwrap();
        assert!(change.attribution.is_some());
        let attr = change.attribution.as_ref().unwrap();
        assert_eq!(attr.contributor_type, "ai");
        assert_eq!(attr.model_id, Some("claude-opus-4".to_string()));
    }

    #[test]
    fn test_format_addition_diff() {
        let diff = format_addition_diff("line1\nline2");
        assert!(diff.contains("@@"));
        assert!(diff.contains("+line1"));
        assert!(diff.contains("+line2"));
    }

    #[test]
    fn test_format_deletion_diff() {
        let diff = format_deletion_diff("line1\nline2");
        assert!(diff.contains("@@"));
        assert!(diff.contains("-line1"));
        assert!(diff.contains("-line2"));
    }

    #[test]
    fn test_compute_unified_diff() {
        let old = "line1\nline2\nline3";
        let new = "line1\nmodified\nline3";
        let diff = compute_unified_diff_with_labels(old, new, "file.txt", "file.txt");

        assert!(diff.contains(" line1"));
        assert!(diff.contains("-line2"));
        assert!(diff.contains("+modified"));
        assert!(diff.contains(" line3"));
        assert!(diff.starts_with("--- a/"));
    }

    #[test]
    fn test_file_change_accessors() {
        let add = FileChange::add("hello");
        assert_eq!(add.new_content(), Some("hello"));
        assert_eq!(add.old_content(), None);
        assert!(add.is_add());

        let delete = FileChange::delete("goodbye");
        assert_eq!(delete.new_content(), None);
        assert_eq!(delete.old_content(), Some("goodbye"));
        assert!(delete.is_delete());

        let update = FileChange::update("old", "new");
        assert_eq!(update.new_content(), Some("new"));
        assert_eq!(update.old_content(), Some("old"));
        assert!(update.is_update());
    }

    #[test]
    fn test_normalized_model_id() {
        let attr = ChangeAttribution::ai("claude-opus-4", "anthropic");
        assert_eq!(attr.normalized_model_id(), Some("anthropic/claude-opus-4".to_string()));

        let attr2 = ChangeAttribution::ai("anthropic/claude-opus-4", "anthropic");
        assert_eq!(attr2.normalized_model_id(), Some("anthropic/claude-opus-4".to_string()));
    }

    #[tokio::test]
    async fn test_shared_tracker() {
        let tracker = new_shared_tracker();

        {
            let mut t = tracker.write().await;
            let mut changes = HashMap::new();
            changes.insert(PathBuf::from("test.txt"), FileChange::add("hello"));
            t.on_patch_begin(changes);
            t.on_patch_end(true);
        }

        {
            let t = tracker.read().await;
            assert!(t.has_changes());
        }
    }

    #[test]
    fn test_file_change_serialization() {
        let change = FileChange::add("fn main() {}")
            .with_attribution(ChangeAttribution::ai("claude-opus-4", "anthropic"))
            .with_line_range(1, 5);

        let json = serde_json::to_string_pretty(&change).unwrap();
        assert!(json.contains("\"type\": \"add\""));
        assert!(json.contains("\"content\": \"fn main() {}\""));
        assert!(json.contains("\"contributor_type\": \"ai\""));
        assert!(json.contains("\"model_id\": \"claude-opus-4\""));

        // Roundtrip
        let restored: FileChange = serde_json::from_str(&json).unwrap();
        assert!(restored.is_add());
        assert_eq!(restored.new_content(), Some("fn main() {}"));
        assert!(restored.attribution.is_some());
    }

    #[test]
    fn test_change_attribution_serialization() {
        let attr = ChangeAttribution::ai("gpt-5", "openai").with_session("session-abc", 3);

        let json = serde_json::to_string(&attr).unwrap();
        let restored: ChangeAttribution = serde_json::from_str(&json).unwrap();

        assert_eq!(restored.model_id, Some("gpt-5".to_string()));
        assert_eq!(restored.provider, Some("openai".to_string()));
        assert_eq!(restored.session_id, Some("session-abc".to_string()));
        assert_eq!(restored.turn_number, Some(3));
        assert_eq!(restored.contributor_type, "ai");
    }
}