gitdataai/libs/service/git_tools/diff.rs

//! Git diff and blame tools.

use super::ctx::GitToolCtx;
use agent::{ToolDefinition, ToolHandler, ToolParam, ToolRegistry, ToolSchema};
use std::collections::HashMap;

async fn git_diff_exec(ctx: GitToolCtx, args: serde_json::Value) -> Result<serde_json::Value, String> {
    let p: serde_json::Map<String, serde_json::Value> = serde_json::from_value(args).map_err(|e| e.to_string())?;
    let project_name = p.get("project_name").and_then(|v| v.as_str()).ok_or("missing project_name")?;
    let repo_name = p.get("repo_name").and_then(|v| v.as_str()).ok_or("missing repo_name")?;
    let base = p.get("base").and_then(|v| v.as_str()).map(|s| s.to_string());
    let head = p.get("head").and_then(|v| v.as_str()).map(|s| s.to_string());
    let paths = p.get("paths").and_then(|v| v.as_array()).map(|a| {
        a.iter().filter_map(|v| v.as_str().map(|s| s.to_string())).collect::<Vec<_>>()
    });

    let domain = ctx.open_repo(project_name, repo_name).await?;

    let resolve = |rev: &str| -> Result<git::commit::types::CommitOid, String> {
        if rev.len() >= 40 {
            Ok(git::commit::types::CommitOid::new(rev))
        } else {
            domain.commit_get_prefix(rev).map_err(|e| e.to_string()).map(|m| m.oid)
        }
    };

    let base_oid = match &base {
        Some(b) => Some(resolve(b)?),
        None => None,
    };
    let head_oid = match &head {
        Some(h) => Some(resolve(h)?),
        None => None,
    };

    let opts = paths.map(|ps| {
        let mut o = git::diff::types::DiffOptions::new();
        for p in ps { o = o.pathspec(&p); }
        Some(o)
    }).flatten();

    let result = match (&base_oid, &head_oid) {
        (None, None) => {
            let head_meta = domain.commit_get_prefix("HEAD").map_err(|e| e.to_string())?;
            domain.diff_commit_to_workdir(&head_meta.oid, opts).map_err(|e| e.to_string())?
        }
        (Some(base), None) => {
            domain.diff_commit_to_workdir(base, opts).map_err(|e| e.to_string())?
        }
        (Some(base), Some(head_oid_val)) => {
            domain.diff_tree_to_tree(Some(base), Some(head_oid_val), opts).map_err(|e| e.to_string())?
        }
        (None, Some(_)) => {
            return Err("base revision required when head is specified".into());
        }
    };

    let files: Vec<_> = result.deltas.iter().map(|d| {
        serde_json::json!({ "path": d.new_file.path, "status": format!("{:?}", d.status), "is_binary": d.new_file.is_binary })
    }).collect();

    Ok(serde_json::json!({
        "stats": { "files_changed": result.stats.files_changed, "insertions": result.stats.insertions, "deletions": result.stats.deletions },
        "files": files
    }))
}

async fn git_diff_stats_exec(ctx: GitToolCtx, args: serde_json::Value) -> Result<serde_json::Value, String> {
    let p: serde_json::Map<String, serde_json::Value> = serde_json::from_value(args).map_err(|e| e.to_string())?;
    let project_name = p.get("project_name").and_then(|v| v.as_str()).ok_or("missing project_name")?;
    let repo_name = p.get("repo_name").and_then(|v| v.as_str()).ok_or("missing repo_name")?;
    let base = p.get("base").and_then(|v| v.as_str()).ok_or("missing base")?;
    let head = p.get("head").and_then(|v| v.as_str()).ok_or("missing head")?;

    let domain = ctx.open_repo(project_name, repo_name).await?;

    let stats = if base.len() >= 40 || head.len() >= 40 {
        domain.diff_stats(&git::commit::types::CommitOid::new(base), &git::commit::types::CommitOid::new(head))
            .map_err(|e| e.to_string())?
    } else {
        let b = domain.commit_get_prefix(base).map_err(|e| e.to_string())?.oid;
        let h = domain.commit_get_prefix(head).map_err(|e| e.to_string())?.oid;
        domain.diff_stats(&b, &h).map_err(|e| e.to_string())?
    };

    Ok(serde_json::json!({
        "files_changed": stats.files_changed,
        "insertions": stats.insertions,
        "deletions": stats.deletions
    }))
}

async fn git_blame_exec(ctx: GitToolCtx, args: serde_json::Value) -> Result<serde_json::Value, String> {
    let p: serde_json::Map<String, serde_json::Value> = serde_json::from_value(args).map_err(|e| e.to_string())?;
    let project_name = p.get("project_name").and_then(|v| v.as_str()).ok_or("missing project_name")?;
    let repo_name = p.get("repo_name").and_then(|v| v.as_str()).ok_or("missing repo_name")?;
    let path = p.get("path").and_then(|v| v.as_str()).ok_or("missing path")?;
    let rev = p.get("rev").and_then(|v| v.as_str()).map(|s| s.to_string()).unwrap_or_else(|| "HEAD".to_string());
    let from_line = p.get("from_line").and_then(|v| v.as_u64().map(|n| n as u32));
    let to_line = p.get("to_line").and_then(|v| v.as_u64().map(|n| n as u32));

    let domain = ctx.open_repo(project_name, repo_name).await?;
    let oid = if rev.len() >= 40 {
        git::commit::types::CommitOid::new(&rev)
    } else {
        domain.commit_get_prefix(&rev).map_err(|e| e.to_string())?.oid
    };

    use git::blame::ops::BlameOptions;
    let mut bopts = BlameOptions::new();
    if let Some(fl) = from_line { bopts = bopts.min_line(fl as usize); }
    if let Some(tl) = to_line { bopts = bopts.max_line(tl as usize); }

    let hunks = domain.blame_file(&oid, path, Some(bopts)).map_err(|e| e.to_string())?;

    let result: Vec<_> = hunks.iter().map(|h| {
        let oid = h.commit_oid.to_string();
        serde_json::json!({
            "commit_oid": oid.clone(),
            "short_oid": oid.get(..7).unwrap_or(&oid).to_string(),
            "final_start_line": h.final_start_line,
            "final_lines": h.final_lines,
            "orig_start_line": h.orig_start_line,
            "orig_path": h.orig_path,
            "boundary": h.boundary
        })
    }).collect();

    Ok(serde_json::to_value(result).map_err(|e| e.to_string())?)
}

pub fn register_git_tools(registry: &mut ToolRegistry) {
    // git_diff
    let p = HashMap::from([
        ("project_name".into(), ToolParam { name: "project_name".into(), param_type: "string".into(), description: Some("Project name (slug)".into()), required: true, properties: None, items: None }),
        ("repo_name".into(), ToolParam { name: "repo_name".into(), param_type: "string".into(), description: Some("Repository name".into()), required: true, properties: None, items: None }),
        ("base".into(), ToolParam { name: "base".into(), param_type: "string".into(), description: Some("Base revision (commit hash or branch). Defaults to HEAD.".into()), required: false, properties: None, items: None }),
        ("head".into(), ToolParam { name: "head".into(), param_type: "string".into(), description: Some("Head revision to diff against base. Requires base to be set.".into()), required: false, properties: None, items: None }),
        ("paths".into(), ToolParam { name: "paths".into(), param_type: "array".into(), description: Some("Filter diff to specific file paths".into()), required: false, properties: None, items: Some(Box::new(ToolParam { name: "".into(), param_type: "string".into(), description: None, required: false, properties: None, items: None })) }),
    ]);
    let schema = ToolSchema { schema_type: "object".into(), properties: Some(p), required: Some(vec!["project_name".into(), "repo_name".into()]) };
    registry.register(
        ToolDefinition::new("git_diff").description("Show file changes between two commits, or between a commit and the working directory.").parameters(schema),
        ToolHandler::new(|ctx, args| {
            let gctx = super::ctx::GitToolCtx::new(ctx);
            Box::pin(async move {
                git_diff_exec(gctx, args).await.map_err(agent::ToolError::ExecutionError)
            })
        }),
    );

    // git_diff_stats
    let p = HashMap::from([
        ("project_name".into(), ToolParam { name: "project_name".into(), param_type: "string".into(), description: Some("Project name (slug)".into()), required: true, properties: None, items: None }),
        ("repo_name".into(), ToolParam { name: "repo_name".into(), param_type: "string".into(), description: Some("Repository name".into()), required: true, properties: None, items: None }),
        ("base".into(), ToolParam { name: "base".into(), param_type: "string".into(), description: Some("Base revision".into()), required: true, properties: None, items: None }),
        ("head".into(), ToolParam { name: "head".into(), param_type: "string".into(), description: Some("Head revision".into()), required: true, properties: None, items: None }),
    ]);
    let schema = ToolSchema { schema_type: "object".into(), properties: Some(p), required: Some(vec!["project_name".into(), "repo_name".into(), "base".into(), "head".into()]) };
    registry.register(
        ToolDefinition::new("git_diff_stats").description("Get aggregated diff statistics (files changed, insertions, deletions) between two revisions.").parameters(schema),
        ToolHandler::new(|ctx, args| {
            let gctx = super::ctx::GitToolCtx::new(ctx);
            Box::pin(async move {
                git_diff_stats_exec(gctx, args).await.map_err(agent::ToolError::ExecutionError)
            })
        }),
    );

    // git_blame
    let p = HashMap::from([
        ("project_name".into(), ToolParam { name: "project_name".into(), param_type: "string".into(), description: Some("Project name (slug)".into()), required: true, properties: None, items: None }),
        ("repo_name".into(), ToolParam { name: "repo_name".into(), param_type: "string".into(), description: Some("Repository name".into()), required: true, properties: None, items: None }),
        ("path".into(), ToolParam { name: "path".into(), param_type: "string".into(), description: Some("File path to blame".into()), required: true, properties: None, items: None }),
        ("rev".into(), ToolParam { name: "rev".into(), param_type: "string".into(), description: Some("Revision to blame from (default: HEAD)".into()), required: false, properties: None, items: None }),
        ("from_line".into(), ToolParam { name: "from_line".into(), param_type: "integer".into(), description: Some("Start line number for blame range".into()), required: false, properties: None, items: None }),
        ("to_line".into(), ToolParam { name: "to_line".into(), param_type: "integer".into(), description: Some("End line number for blame range".into()), required: false, properties: None, items: None }),
    ]);
    let schema = ToolSchema { schema_type: "object".into(), properties: Some(p), required: Some(vec!["project_name".into(), "repo_name".into(), "path".into()]) };
    registry.register(
        ToolDefinition::new("git_blame").description("Show what revision and author last modified each line of a file (git blame).").parameters(schema),
        ToolHandler::new(|ctx, args| {
            let gctx = super::ctx::GitToolCtx::new(ctx);
            Box::pin(async move {
                git_blame_exec(gctx, args).await.map_err(agent::ToolError::ExecutionError)
            })
        }),
    );
}