xbp_cli/commands/

commit.rs

use crate::cli::auto_commit::{print_push_summary, push_current_branch};
use crate::commands::service::load_xbp_config_with_root;
use crate::config::resolve_openrouter_api_key;
use crate::openrouter::{complete_user_prompt, DEFAULT_MODEL};
use crate::utils::command_exists;
use colored::Colorize;
use once_cell::sync::Lazy;
use regex::Regex;
use serde::Deserialize;
use std::collections::{BTreeMap, BTreeSet};
use std::env;
use std::fs;
use std::path::{Path, PathBuf};
use tokio::process::Command;
use uuid::Uuid;

static RUST_FUNCTION_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(r"^\s*(?:pub(?:\([^)]*\))?\s+)?(?:async\s+)?fn\s+([A-Za-z_][A-Za-z0-9_]*)")
        .expect("valid rust function regex")
});
static JS_FUNCTION_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(
        r"^\s*(?:export\s+)?(?:default\s+)?(?:async\s+)?function\s+([A-Za-z_$][A-Za-z0-9_$]*)",
    )
    .expect("valid js function regex")
});
static JS_ARROW_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(
        r"^\s*(?:export\s+)?(?:const|let|var)\s+([A-Za-z_$][A-Za-z0-9_$]*)\s*=\s*(?:async\s*)?(?:\([^)]*\)|[A-Za-z_$][A-Za-z0-9_$]*)\s*=>",
    )
    .expect("valid js arrow regex")
});
static METHOD_CONTEXT_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(r"\b([A-Za-z_][A-Za-z0-9_]*)\s*\(").expect("valid method context regex")
});
static RUST_TYPE_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(r"^\s*(?:pub(?:\([^)]*\))?\s+)?(struct|enum|trait|type)\s+([A-Za-z_][A-Za-z0-9_]*)")
        .expect("valid rust type regex")
});
static TS_TYPE_RE: Lazy<Regex> = Lazy::new(|| {
    Regex::new(r"^\s*(?:export\s+)?(interface|type|enum|class)\s+([A-Za-z_$][A-Za-z0-9_$]*)")
        .expect("valid ts type regex")
});

const AI_DIFF_CHAR_LIMIT: usize = 12_000;
const MAX_PROMPT_FILES: usize = 40;
const MAX_PROMPT_SYMBOLS: usize = 12;
const MAX_PROMPT_CONTEXTS: usize = 12;
const MAX_PRINT_SYMBOLS: usize = 8;
const CONVENTIONAL_TYPES: &[&str] = &[
    "feat", "fix", "docs", "refactor", "chore", "test", "build", "ci", "perf", "style", "revert",
];

#[derive(Debug, Clone)]
pub struct CommitArgs {
    pub dry_run: bool,
    pub no_ai: bool,
    pub model: String,
    pub scope: Option<String>,
}

#[derive(Debug, Clone)]
struct WorktreeAnalysis {
    repo_root: PathBuf,
    repo_name: String,
    branch: Option<String>,
    status_entries: Vec<StatusEntry>,
    files: Vec<FileChangeSummary>,
    total_additions: u32,
    total_deletions: u32,
    diff_text: String,
    new_functions: Vec<String>,
    changed_functions: Vec<String>,
    removed_functions: Vec<String>,
    new_types: Vec<String>,
    changed_types: Vec<String>,
    removed_types: Vec<String>,
    hunk_contexts: Vec<String>,
}

#[derive(Debug, Clone)]
struct StatusEntry {
    code: String,
    path: String,
}

#[derive(Debug, Clone)]
struct FileChangeSummary {
    path: String,
    status: String,
    additions: u32,
    deletions: u32,
}

#[derive(Debug, Clone)]
struct SymbolSummary {
    new_functions: Vec<String>,
    changed_functions: Vec<String>,
    removed_functions: Vec<String>,
    new_types: Vec<String>,
    changed_types: Vec<String>,
    removed_types: Vec<String>,
    hunk_contexts: Vec<String>,
}

#[derive(Debug, Clone, Copy, Eq, PartialEq)]
enum SymbolKind {
    Function,
    Type,
}

#[derive(Debug, Clone)]
struct CommitMessagePlan {
    commit: ConventionalCommit,
    generation_mode: GenerationMode,
}

#[derive(Debug, Clone)]
struct ConventionalCommit {
    commit_type: String,
    scope: Option<String>,
    description: String,
    body: Vec<String>,
    breaking_change: Option<String>,
    footers: Vec<String>,
}

#[derive(Debug, Clone, Copy)]
enum GenerationMode {
    OpenRouter,
    Heuristic,
}

#[derive(Debug, Deserialize)]
struct AiCommitPayload {
    #[serde(rename = "type")]
    commit_type: String,
    scope: Option<String>,
    description: String,
    #[serde(default)]
    body: Vec<String>,
    #[serde(default)]
    breaking_change: bool,
    breaking_description: Option<String>,
    #[serde(default)]
    footers: Vec<String>,
}

pub async fn run_commit(args: CommitArgs) -> Result<(), String> {
    if !command_exists("git") {
        return Err("Git is not installed on this machine.".to_string());
    }

    let invocation_dir =
        env::current_dir().map_err(|e| format!("Failed to read current directory: {}", e))?;
    let analysis = analyze_worktree(&invocation_dir).await?;
    let auto_push_after_commit = resolve_auto_push_on_commit().await;
    let commit_plan = generate_commit_plan(&analysis, &args).await;
    let rendered_message = render_commit_message(&commit_plan.commit);

    print_analysis_summary(&analysis, &commit_plan, &rendered_message);

    if args.dry_run {
        println!(
            "\n{}",
            "Dry run only. No git commit was created."
                .bright_yellow()
                .bold()
        );
        return Ok(());
    }

    stage_and_commit(&analysis.repo_root, &rendered_message).await?;

    let short_sha = git_output(&analysis.repo_root, &["rev-parse", "--short", "HEAD"]).await?;
    let full_sha = git_output(&analysis.repo_root, &["rev-parse", "HEAD"]).await?;

    println!(
        "\n{} {} {}",
        "Committed".bright_green().bold(),
        short_sha.bright_white().bold(),
        format!("({})", full_sha).dimmed()
    );

    if auto_push_after_commit {
        match push_current_branch(&analysis.repo_root).await {
            Ok(Some(outcome)) => print_push_summary(&outcome),
            Ok(None) => println!(
                "{}",
                "Push skipped because the current HEAD is detached.".bright_yellow()
            ),
            Err(error) => {
                return Err(format!(
                    "Created local commit {} but failed to push it: {}",
                    short_sha, error
                ));
            }
        }
    } else {
        println!(
            "{}",
            "Auto-push disabled by xbp config (`github.auto_push_on_commit: false`)."
                .bright_yellow()
        );
    }

    Ok(())
}

async fn resolve_auto_push_on_commit() -> bool {
    load_xbp_config_with_root()
        .await
        .map(|(_, config)| config.auto_push_on_commit_enabled())
        .unwrap_or(true)
}

async fn analyze_worktree(invocation_dir: &Path) -> Result<WorktreeAnalysis, String> {
    let repo_root = PathBuf::from(
        git_output(invocation_dir, &["rev-parse", "--show-toplevel"])
            .await
            .map_err(|_| "Current directory is not inside a git repository.".to_string())?,
    );

    let repo_name = repo_root
        .file_name()
        .and_then(|value| value.to_str())
        .filter(|value| !value.trim().is_empty())
        .unwrap_or("repository")
        .to_string();

    let branch = git_output(&repo_root, &["rev-parse", "--abbrev-ref", "HEAD"])
        .await
        .ok()
        .filter(|value| !value.is_empty() && value != "HEAD");

    let status_output = git_output(
        &repo_root,
        &["status", "--porcelain=v1", "--untracked-files=all"],
    )
    .await?;
    let status_entries = parse_status_entries(&status_output);
    if status_entries.is_empty() {
        return Err("No worktree changes were found to commit.".to_string());
    }

    let temp_index = prepare_temporary_index(&repo_root).await?;
    let temp_index_path = temp_index.path.clone();
    let git_env = [("GIT_INDEX_FILE", temp_index_path.as_os_str())];

    git_output_with_env(&repo_root, &["add", "--all"], &git_env).await?;
    let name_status_output = git_output_with_env(
        &repo_root,
        &["diff", "--cached", "--name-status", "--find-renames"],
        &git_env,
    )
    .await?;
    let numstat_output = git_output_with_env(
        &repo_root,
        &["diff", "--cached", "--numstat", "--find-renames"],
        &git_env,
    )
    .await?;
    let diff_text = git_output_with_env(
        &repo_root,
        &[
            "diff",
            "--cached",
            "--unified=0",
            "--no-color",
            "--no-ext-diff",
            "--find-renames",
        ],
        &git_env,
    )
    .await?;

    if diff_text.trim().is_empty() {
        return Err("No staged diff could be produced from the current worktree.".to_string());
    }

    let files = merge_file_summaries(&name_status_output, &numstat_output);
    let total_additions = files.iter().map(|entry| entry.additions).sum();
    let total_deletions = files.iter().map(|entry| entry.deletions).sum();
    let symbols = summarize_symbols(&diff_text);

    Ok(WorktreeAnalysis {
        repo_root,
        repo_name,
        branch,
        status_entries,
        files,
        total_additions,
        total_deletions,
        diff_text,
        new_functions: symbols.new_functions,
        changed_functions: symbols.changed_functions,
        removed_functions: symbols.removed_functions,
        new_types: symbols.new_types,
        changed_types: symbols.changed_types,
        removed_types: symbols.removed_types,
        hunk_contexts: symbols.hunk_contexts,
    })
}

async fn prepare_temporary_index(repo_root: &Path) -> Result<TemporaryIndex, String> {
    let real_index_path =
        PathBuf::from(git_output(repo_root, &["rev-parse", "--git-path", "index"]).await?);
    let temp_index_path = env::temp_dir().join(format!("xbp-commit-index-{}.tmp", Uuid::new_v4()));

    if real_index_path.exists() {
        fs::copy(&real_index_path, &temp_index_path).map_err(|e| {
            format!(
                "Failed to prepare temporary git index {}: {}",
                temp_index_path.display(),
                e
            )
        })?;
    } else {
        let git_env = [("GIT_INDEX_FILE", temp_index_path.as_os_str())];
        let _ = git_output_with_env(repo_root, &["read-tree", "HEAD"], &git_env).await;
    }

    Ok(TemporaryIndex {
        path: temp_index_path,
    })
}

async fn generate_commit_plan(analysis: &WorktreeAnalysis, args: &CommitArgs) -> CommitMessagePlan {
    let forced_scope = sanitize_scope(args.scope.as_deref());
    let heuristic = build_heuristic_commit(analysis, forced_scope.clone());

    if args.no_ai {
        return CommitMessagePlan {
            commit: heuristic,
            generation_mode: GenerationMode::Heuristic,
        };
    }

    let Some(api_key) = resolve_openrouter_api_key() else {
        return CommitMessagePlan {
            commit: heuristic,
            generation_mode: GenerationMode::Heuristic,
        };
    };

    let prompt = build_commit_prompt(analysis, forced_scope.as_deref(), &heuristic);
    let model = if args.model.trim().is_empty() {
        DEFAULT_MODEL
    } else {
        args.model.trim()
    };

    let ai_message = complete_user_prompt(&api_key, model, &prompt, Some("XBP Commit Generator"))
        .await
        .and_then(|raw| parse_ai_commit_payload(&raw, forced_scope.as_deref()));

    if let Some(commit) = ai_message {
        CommitMessagePlan {
            commit,
            generation_mode: GenerationMode::OpenRouter,
        }
    } else {
        CommitMessagePlan {
            commit: heuristic,
            generation_mode: GenerationMode::Heuristic,
        }
    }
}

fn build_commit_prompt(
    analysis: &WorktreeAnalysis,
    forced_scope: Option<&str>,
    heuristic: &ConventionalCommit,
) -> String {
    let mut lines = Vec::new();
    lines.push("You are generating one git commit message for the current worktree.".to_string());
    lines.push(String::new());
    lines.push("Return strict JSON only. No markdown, no code fences, no explanation.".to_string());
    lines.push(String::new());
    lines.push("Schema:".to_string());
    lines.push("{".to_string());
    lines.push(
        r#"  "type": "feat|fix|docs|refactor|chore|test|build|ci|perf|style|revert","#.to_string(),
    );
    lines.push(r#"  "scope": "short lowercase scope or null","#.to_string());
    lines.push(
        r#"  "description": "imperative summary under 72 chars, no trailing period","#.to_string(),
    );
    lines.push(r#"  "body": ["optional paragraph", "optional paragraph"],"#.to_string());
    lines.push(r#"  "breaking_change": false,"#.to_string());
    lines.push(
        r#"  "breaking_description": "required only when breaking_change=true","#.to_string(),
    );
    lines.push(r#"  "footers": ["optional footer line"]"#.to_string());
    lines.push("}".to_string());
    lines.push(String::new());
    lines.push("Rules:".to_string());
    lines.push("- Follow Conventional Commits 1.0.0 exactly.".to_string());
    lines.push("- Prefer feat for new capabilities, fix for bug repair, docs for docs-only changes, refactor for internal restructuring without behavior change, and ci/build/test/perf/style/chore when clearly better.".to_string());
    lines.push("- Use a scope only when it is clearly anchored in the codebase.".to_string());
    lines.push(
        "- The description must be specific to the changed behavior or module, not generic."
            .to_string(),
    );
    lines.push("- Mention a breaking change only when the diff clearly changes a public or operator-facing contract.".to_string());
    lines.push("- Use the file list, symbol list, and diff excerpt. Do not invent files, functions, types, or breaking changes.".to_string());
    lines.push(String::new());
    lines.push(format!("Repository: {}", analysis.repo_name));
    if let Some(branch) = &analysis.branch {
        lines.push(format!("Branch: {}", branch));
    }
    if let Some(scope) = forced_scope {
        lines.push(format!("Forced scope: {}", scope));
    } else if let Some(scope) = heuristic.scope.as_deref() {
        lines.push(format!("Suggested scope: {}", scope));
    }
    lines.push(format!(
        "Heuristic fallback subject: {}",
        render_commit_subject(heuristic)
    ));
    lines.push(String::new());
    lines.push("Worktree stats:".to_string());
    lines.push(format!("- files changed: {}", analysis.files.len()));
    lines.push(format!(
        "- lines: +{} -{}",
        analysis.total_additions, analysis.total_deletions
    ));
    if !analysis.status_entries.is_empty() {
        lines.push("- worktree statuses:".to_string());
        for entry in analysis.status_entries.iter().take(MAX_PROMPT_FILES) {
            lines.push(format!("  - {} {}", entry.code, entry.path));
        }
    }
    if !analysis.files.is_empty() {
        lines.push("- file diffs:".to_string());
        for file in analysis.files.iter().take(MAX_PROMPT_FILES) {
            lines.push(format!(
                "  - [{}] {} (+{} -{})",
                file.status, file.path, file.additions, file.deletions
            ));
        }
    }
    if !analysis.new_functions.is_empty() {
        lines.push(format!(
            "- new functions: {}",
            analysis
                .new_functions
                .iter()
                .take(MAX_PROMPT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.changed_functions.is_empty() {
        lines.push(format!(
            "- changed functions: {}",
            analysis
                .changed_functions
                .iter()
                .take(MAX_PROMPT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.new_types.is_empty() {
        lines.push(format!(
            "- new types: {}",
            analysis
                .new_types
                .iter()
                .take(MAX_PROMPT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.changed_types.is_empty() {
        lines.push(format!(
            "- changed types: {}",
            analysis
                .changed_types
                .iter()
                .take(MAX_PROMPT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.hunk_contexts.is_empty() {
        lines.push(format!(
            "- hunk contexts: {}",
            analysis
                .hunk_contexts
                .iter()
                .take(MAX_PROMPT_CONTEXTS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    lines.push(String::new());
    lines.push("Diff excerpt:".to_string());
    lines.push(truncate_for_prompt(&analysis.diff_text, AI_DIFF_CHAR_LIMIT));
    lines.join("\n")
}

fn parse_ai_commit_payload(raw: &str, forced_scope: Option<&str>) -> Option<ConventionalCommit> {
    let payload: AiCommitPayload = serde_json::from_str(raw).ok()?;
    let commit_type = sanitize_commit_type(&payload.commit_type)?;
    let description = sanitize_description(&payload.description)?;
    let mut body = payload
        .body
        .into_iter()
        .map(|entry| entry.trim().to_string())
        .filter(|entry| !entry.is_empty())
        .collect::<Vec<_>>();
    if body.len() > 3 {
        body.truncate(3);
    }

    let breaking_change = if payload.breaking_change {
        payload
            .breaking_description
            .as_deref()
            .and_then(sanitize_footer_value)
    } else {
        None
    };

    let mut footers = payload
        .footers
        .into_iter()
        .map(|entry| entry.trim().to_string())
        .filter(|entry| !entry.is_empty())
        .collect::<Vec<_>>();
    if breaking_change.is_some()
        && !footers
            .iter()
            .any(|entry| entry.starts_with("BREAKING CHANGE:"))
    {
        if let Some(breaking) = &breaking_change {
            footers.push(format!("BREAKING CHANGE: {}", breaking));
        }
    }

    Some(ConventionalCommit {
        commit_type,
        scope: forced_scope
            .and_then(|scope| sanitize_scope(Some(scope)))
            .or_else(|| sanitize_scope(payload.scope.as_deref())),
        description,
        body,
        breaking_change,
        footers,
    })
}

fn build_heuristic_commit(
    analysis: &WorktreeAnalysis,
    forced_scope: Option<String>,
) -> ConventionalCommit {
    let commit_type = infer_commit_type(analysis).to_string();
    let scope = forced_scope.or_else(|| infer_scope(analysis));
    let description = infer_description(analysis, &commit_type, scope.as_deref());
    let body = build_heuristic_body(analysis);

    ConventionalCommit {
        commit_type,
        scope,
        description,
        body,
        breaking_change: None,
        footers: Vec::new(),
    }
}

fn infer_commit_type(analysis: &WorktreeAnalysis) -> &'static str {
    let lowered_paths = analysis
        .files
        .iter()
        .map(|file| file.path.to_ascii_lowercase())
        .collect::<Vec<_>>();

    let docs_only = !lowered_paths.is_empty()
        && lowered_paths.iter().all(|path| {
            path.ends_with(".md")
                || path.ends_with(".mdx")
                || path.ends_with(".txt")
                || path.starts_with("docs/")
        });
    if docs_only {
        return "docs";
    }

    let test_only = !lowered_paths.is_empty()
        && lowered_paths.iter().all(|path| {
            path.contains("/tests/")
                || path.contains("\\tests\\")
                || path.contains(".test.")
                || path.contains(".spec.")
        });
    if test_only {
        return "test";
    }

    let ci_only = !lowered_paths.is_empty()
        && lowered_paths.iter().all(|path| {
            path.starts_with(".github/")
                || path.contains("workflow")
                || path.ends_with(".yml")
                || path.ends_with(".yaml")
        });
    if ci_only {
        return "ci";
    }

    let build_only = !lowered_paths.is_empty()
        && lowered_paths.iter().all(|path| {
            path.ends_with("cargo.toml")
                || path.ends_with("cargo.lock")
                || path.ends_with("package.json")
                || path.ends_with("pnpm-lock.yaml")
                || path.ends_with("package-lock.json")
                || path.ends_with("wrangler.toml")
                || path.ends_with(".nix")
        });
    if build_only {
        return "build";
    }

    let has_new_files = analysis.files.iter().any(|file| file.status == "added");
    let has_new_symbols = !analysis.new_functions.is_empty() || !analysis.new_types.is_empty();
    if has_new_files || has_new_symbols {
        return "feat";
    }

    if !analysis.changed_functions.is_empty()
        || !analysis.changed_types.is_empty()
        || analysis.total_additions >= analysis.total_deletions
    {
        return "fix";
    }

    "chore"
}

fn infer_scope(analysis: &WorktreeAnalysis) -> Option<String> {
    let mut scopes = BTreeSet::new();

    for file in &analysis.files {
        let path = file.path.replace('\\', "/");
        let inferred = if path.starts_with("crates/cli/") {
            Some("cli")
        } else if path.starts_with("crates/api/") {
            Some("api")
        } else if path.starts_with("crates/github/") {
            Some("github")
        } else if path.starts_with("crates/runtime/") {
            Some("runtime")
        } else if path.starts_with("apps/web/") {
            Some("web")
        } else if path.starts_with("docs/") {
            Some("docs")
        } else if path.starts_with(".github/") {
            Some("ci")
        } else {
            None
        };

        if let Some(value) = inferred {
            scopes.insert(value.to_string());
        }
    }

    if scopes.len() == 1 {
        scopes.into_iter().next()
    } else if scopes.is_empty() {
        None
    } else if scopes.contains("cli") {
        Some("cli".to_string())
    } else {
        None
    }
}

fn infer_description(
    analysis: &WorktreeAnalysis,
    commit_type: &str,
    scope: Option<&str>,
) -> String {
    let interesting_names = analysis
        .files
        .iter()
        .filter_map(|file| interesting_file_stem(&file.path))
        .collect::<Vec<_>>();

    if interesting_names.iter().any(|name| name == "commit") {
        return match scope {
            Some("cli") => "add worktree commit generator".to_string(),
            _ => "add conventional commit generator".to_string(),
        };
    }

    if commit_type == "docs" {
        if let Some(name) = interesting_names.first() {
            return format!("document {}", name.replace('_', "-"));
        }
        return "update command documentation".to_string();
    }

    if let Some(name) = interesting_names.first() {
        return match commit_type {
            "feat" => format!("add {}", name.replace('_', "-")),
            "fix" => format!("improve {}", name.replace('_', "-")),
            "refactor" => format!("refactor {}", name.replace('_', "-")),
            "test" => format!("cover {}", name.replace('_', "-")),
            "ci" => format!("update {}", name.replace('_', "-")),
            "build" => format!("adjust {}", name.replace('_', "-")),
            _ => format!("update {}", name.replace('_', "-")),
        };
    }

    match commit_type {
        "feat" => "add worktree change support".to_string(),
        "fix" => "improve worktree handling".to_string(),
        "docs" => "update documentation".to_string(),
        _ => "update repository changes".to_string(),
    }
}

fn build_heuristic_body(analysis: &WorktreeAnalysis) -> Vec<String> {
    let mut paragraphs = Vec::new();
    paragraphs.push(format!(
        "Touches {} file{} with +{} and -{} lines across the current worktree.",
        analysis.files.len(),
        if analysis.files.len() == 1 { "" } else { "s" },
        analysis.total_additions,
        analysis.total_deletions
    ));

    let mut detail_parts = Vec::new();
    if !analysis.new_functions.is_empty() {
        detail_parts.push(format!(
            "new functions: {}",
            analysis
                .new_functions
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.changed_functions.is_empty() {
        detail_parts.push(format!(
            "changed functions: {}",
            analysis
                .changed_functions
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.removed_functions.is_empty() {
        detail_parts.push(format!(
            "removed functions: {}",
            analysis
                .removed_functions
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.new_types.is_empty() {
        detail_parts.push(format!(
            "new types: {}",
            analysis
                .new_types
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.changed_types.is_empty() {
        detail_parts.push(format!(
            "changed types: {}",
            analysis
                .changed_types
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !analysis.removed_types.is_empty() {
        detail_parts.push(format!(
            "removed types: {}",
            analysis
                .removed_types
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        ));
    }
    if !detail_parts.is_empty() {
        paragraphs.push(detail_parts.join("; "));
    }

    paragraphs
}

fn render_commit_message(commit: &ConventionalCommit) -> String {
    let mut sections = vec![render_commit_subject(commit)];

    if !commit.body.is_empty() {
        sections.push(commit.body.join("\n\n"));
    }

    let mut footer_lines = commit
        .footers
        .iter()
        .map(|entry| entry.trim().to_string())
        .filter(|entry| !entry.is_empty())
        .collect::<Vec<_>>();
    if let Some(breaking_change) = &commit.breaking_change {
        if !footer_lines
            .iter()
            .any(|entry| entry.starts_with("BREAKING CHANGE:"))
        {
            footer_lines.push(format!("BREAKING CHANGE: {}", breaking_change));
        }
    }
    if !footer_lines.is_empty() {
        sections.push(footer_lines.join("\n"));
    }

    sections.join("\n\n")
}

fn render_commit_subject(commit: &ConventionalCommit) -> String {
    let scope = commit
        .scope
        .as_deref()
        .map(|value| format!("({})", value))
        .unwrap_or_default();
    let bang = if commit.breaking_change.is_some() {
        "!"
    } else {
        ""
    };
    format!(
        "{}{}{}: {}",
        commit.commit_type, scope, bang, commit.description
    )
}

fn print_analysis_summary(
    analysis: &WorktreeAnalysis,
    commit_plan: &CommitMessagePlan,
    rendered_message: &str,
) {
    let branch = analysis
        .branch
        .as_deref()
        .map(|value| format!(" on {}", value.bright_blue()))
        .unwrap_or_default();
    println!(
        "{} {}{}",
        "Commit".bright_green().bold(),
        analysis.repo_name.bright_white().bold(),
        branch
    );
    println!(
        "  {} {}",
        "Mode".bright_cyan().bold(),
        match commit_plan.generation_mode {
            GenerationMode::OpenRouter => "OpenRouter".bright_white(),
            GenerationMode::Heuristic => "Heuristic fallback".bright_white(),
        }
    );
    println!(
        "  {} {} file{} (+{} -{})",
        "Diff".bright_cyan().bold(),
        analysis.files.len(),
        if analysis.files.len() == 1 { "" } else { "s" },
        analysis.total_additions,
        analysis.total_deletions
    );

    if !analysis.new_functions.is_empty() {
        println!(
            "  {} {}",
            "New fn".bright_cyan().bold(),
            analysis
                .new_functions
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        );
    }
    if !analysis.changed_functions.is_empty() {
        println!(
            "  {} {}",
            "Changed fn".bright_cyan().bold(),
            analysis
                .changed_functions
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        );
    }
    if !analysis.new_types.is_empty() {
        println!(
            "  {} {}",
            "New types".bright_cyan().bold(),
            analysis
                .new_types
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        );
    }
    if !analysis.changed_types.is_empty() {
        println!(
            "  {} {}",
            "Changed types".bright_cyan().bold(),
            analysis
                .changed_types
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        );
    }
    if !analysis.removed_functions.is_empty() {
        println!(
            "  {} {}",
            "Removed fn".bright_cyan().bold(),
            analysis
                .removed_functions
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        );
    }
    if !analysis.removed_types.is_empty() {
        println!(
            "  {} {}",
            "Removed types".bright_cyan().bold(),
            analysis
                .removed_types
                .iter()
                .take(MAX_PRINT_SYMBOLS)
                .cloned()
                .collect::<Vec<_>>()
                .join(", ")
        );
    }

    println!("\n{}", "Generated message".bright_magenta().bold());
    println!("{}", "─".repeat(72).bright_black());
    println!("{}", rendered_message);
    println!("{}", "─".repeat(72).bright_black());
}

async fn stage_and_commit(repo_root: &Path, message: &str) -> Result<(), String> {
    git_output(repo_root, &["add", "--all"]).await?;

    let message_path = env::temp_dir().join(format!("xbp-commit-message-{}.txt", Uuid::new_v4()));
    fs::write(&message_path, message).map_err(|e| {
        format!(
            "Failed to write temporary commit message {}: {}",
            message_path.display(),
            e
        )
    })?;

    let result = git_output(
        repo_root,
        &["commit", "--file", &message_path.to_string_lossy()],
    )
    .await;
    let _ = fs::remove_file(&message_path);
    result.map(|_| ())
}

fn parse_status_entries(output: &str) -> Vec<StatusEntry> {
    output
        .lines()
        .filter_map(|line| {
            if line.len() < 4 {
                return None;
            }
            let code = line[..2].trim().to_string();
            let path = line[3..].trim().replace('\\', "/");
            if path.is_empty() {
                None
            } else {
                Some(StatusEntry { code, path })
            }
        })
        .collect()
}

fn merge_file_summaries(name_status: &str, numstat: &str) -> Vec<FileChangeSummary> {
    let mut status_map = BTreeMap::new();
    let mut display_path_map = BTreeMap::new();

    for raw_line in name_status
        .lines()
        .map(str::trim)
        .filter(|line| !line.is_empty())
    {
        let parts = raw_line.split('\t').collect::<Vec<_>>();
        if parts.is_empty() {
            continue;
        }
        let status = normalize_name_status(parts[0]);
        match parts.as_slice() {
            [_, path] => {
                let key = normalize_path_key(path);
                display_path_map.insert(key.clone(), normalize_display_path(path));
                status_map.insert(key, status);
            }
            [_, old_path, new_path] => {
                let key = normalize_path_key(new_path);
                display_path_map.insert(
                    key.clone(),
                    format!(
                        "{} -> {}",
                        normalize_display_path(old_path),
                        normalize_display_path(new_path)
                    ),
                );
                status_map.insert(key, status);
            }
            _ => {}
        }
    }

    let mut files = Vec::new();
    for raw_line in numstat
        .lines()
        .map(str::trim)
        .filter(|line| !line.is_empty())
    {
        let parts = raw_line.split('\t').collect::<Vec<_>>();
        if parts.len() < 3 {
            continue;
        }
        let additions = parts[0].parse::<u32>().unwrap_or(0);
        let deletions = parts[1].parse::<u32>().unwrap_or(0);
        let raw_path = parts[2..].join("\t");
        let key = normalize_path_key(&raw_path);
        let path = display_path_map
            .get(&key)
            .cloned()
            .unwrap_or_else(|| normalize_display_path(&raw_path));
        let status = status_map
            .get(&key)
            .cloned()
            .unwrap_or_else(|| "modified".to_string());
        files.push(FileChangeSummary {
            path,
            status,
            additions,
            deletions,
        });
    }

    if files.is_empty() {
        for (key, status) in status_map {
            files.push(FileChangeSummary {
                path: display_path_map
                    .get(&key)
                    .cloned()
                    .unwrap_or_else(|| key.clone()),
                status,
                additions: 0,
                deletions: 0,
            });
        }
    }

    files
}

fn summarize_symbols(diff_text: &str) -> SymbolSummary {
    let mut current_file = String::new();
    let mut added_functions = BTreeSet::new();
    let mut removed_functions = BTreeSet::new();
    let mut added_types = BTreeSet::new();
    let mut removed_types = BTreeSet::new();
    let mut changed_functions = BTreeSet::new();
    let mut changed_types = BTreeSet::new();
    let mut hunk_contexts = BTreeSet::new();

    for line in diff_text.lines() {
        if let Some(rest) = line.strip_prefix("+++ b/") {
            current_file = rest.trim().replace('\\', "/");
            continue;
        }
        if line.starts_with("@@") {
            if let Some(context) = parse_hunk_context(line) {
                if is_code_path(&current_file) {
                    if let Some((symbol, kind)) = extract_context_symbol(&context) {
                        match kind {
                            SymbolKind::Function => {
                                changed_functions.insert(symbol);
                            }
                            SymbolKind::Type => {
                                changed_types.insert(symbol);
                            }
                        }
                    }
                }
                hunk_contexts.insert(context);
            }
            continue;
        }

        if current_file.is_empty() || line.starts_with("+++") || line.starts_with("---") {
            continue;
        }

        if let Some(source) = line.strip_prefix('+') {
            if let Some(name) = extract_function_name(source) {
                added_functions.insert(format!("{} ({})", name, current_file));
            }
            if let Some(name) = extract_type_name(source) {
                added_types.insert(format!("{} ({})", name, current_file));
            }
        } else if let Some(source) = line.strip_prefix('-') {
            if let Some(name) = extract_function_name(source) {
                removed_functions.insert(format!("{} ({})", name, current_file));
            }
            if let Some(name) = extract_type_name(source) {
                removed_types.insert(format!("{} ({})", name, current_file));
            }
        }
    }

    let changed_functions_from_dupes = added_functions
        .intersection(&removed_functions)
        .cloned()
        .collect::<BTreeSet<_>>();
    let changed_types_from_dupes = added_types
        .intersection(&removed_types)
        .cloned()
        .collect::<BTreeSet<_>>();

    for entry in &changed_functions_from_dupes {
        changed_functions.insert(entry.clone());
    }
    for entry in &changed_types_from_dupes {
        changed_types.insert(entry.clone());
    }

    let new_functions = added_functions
        .difference(&changed_functions_from_dupes)
        .cloned()
        .collect::<Vec<_>>();
    let removed_functions = removed_functions
        .difference(&changed_functions_from_dupes)
        .cloned()
        .collect::<Vec<_>>();
    let new_types = added_types
        .difference(&changed_types_from_dupes)
        .cloned()
        .collect::<Vec<_>>();
    let removed_types = removed_types
        .difference(&changed_types_from_dupes)
        .cloned()
        .collect::<Vec<_>>();

    SymbolSummary {
        new_functions,
        changed_functions: changed_functions.into_iter().collect(),
        removed_functions,
        new_types,
        changed_types: changed_types.into_iter().collect(),
        removed_types,
        hunk_contexts: hunk_contexts.into_iter().collect(),
    }
}

fn parse_hunk_context(line: &str) -> Option<String> {
    let parts = line.split("@@").collect::<Vec<_>>();
    if parts.len() < 3 {
        return None;
    }
    let context = parts[2].trim();
    if context.is_empty() {
        None
    } else {
        Some(context.to_string())
    }
}

fn extract_context_symbol(context: &str) -> Option<(String, SymbolKind)> {
    let trimmed = context.trim();
    if trimmed.is_empty() {
        return None;
    }

    for capture in [
        RUST_FUNCTION_RE.captures(trimmed),
        JS_FUNCTION_RE.captures(trimmed),
    ]
    .into_iter()
    .flatten()
    {
        if let Some(name) = capture.get(1) {
            let symbol = name.as_str().to_string();
            if is_noise_symbol(&symbol) {
                return None;
            }
            return Some((symbol, SymbolKind::Function));
        }
    }

    if let Some(capture) = JS_ARROW_RE.captures(trimmed) {
        if let Some(name) = capture.get(1) {
            let symbol = name.as_str().to_string();
            if is_noise_symbol(&symbol) {
                return None;
            }
            return Some((symbol, SymbolKind::Function));
        }
    }

    for capture in [RUST_TYPE_RE.captures(trimmed), TS_TYPE_RE.captures(trimmed)]
        .into_iter()
        .flatten()
    {
        if let Some(name) = capture.get(2) {
            let symbol = name.as_str().to_string();
            if is_noise_symbol(&symbol) {
                return None;
            }
            return Some((symbol, SymbolKind::Type));
        }
    }

    if let Some(capture) = METHOD_CONTEXT_RE.captures(trimmed) {
        if let Some(name) = capture.get(1) {
            let symbol = name.as_str().to_string();
            if is_noise_symbol(&symbol) {
                return None;
            }
            return Some((symbol, SymbolKind::Function));
        }
    }

    None
}

fn extract_function_name(source: &str) -> Option<String> {
    for capture in [
        RUST_FUNCTION_RE.captures(source),
        JS_FUNCTION_RE.captures(source),
        JS_ARROW_RE.captures(source),
    ]
    .into_iter()
    .flatten()
    {
        if let Some(name) = capture.get(1) {
            return Some(name.as_str().to_string());
        }
    }
    None
}

fn extract_type_name(source: &str) -> Option<String> {
    for capture in [RUST_TYPE_RE.captures(source), TS_TYPE_RE.captures(source)]
        .into_iter()
        .flatten()
    {
        if let Some(name) = capture.get(2) {
            return Some(name.as_str().to_string());
        }
    }
    None
}

fn is_code_path(path: &str) -> bool {
    let normalized = path.to_ascii_lowercase();
    [
        ".rs", ".ts", ".tsx", ".js", ".jsx", ".mjs", ".cjs", ".py", ".go", ".java", ".kt", ".swift",
    ]
    .iter()
    .any(|extension| normalized.ends_with(extension))
}

fn is_noise_symbol(symbol: &str) -> bool {
    matches!(
        symbol,
        "fn" | "pub"
            | "impl"
            | "mod"
            | "use"
            | "struct"
            | "enum"
            | "trait"
            | "type"
            | "class"
            | "interface"
            | "const"
            | "let"
            | "var"
            | "async"
            | "await"
            | "match"
            | "if"
            | "else"
            | "for"
            | "while"
            | "loop"
    )
}

fn sanitize_commit_type(raw: &str) -> Option<String> {
    let normalized = raw.trim().to_ascii_lowercase();
    if CONVENTIONAL_TYPES.contains(&normalized.as_str()) {
        Some(normalized)
    } else {
        None
    }
}

fn sanitize_scope(raw: Option<&str>) -> Option<String> {
    let raw = raw?.trim();
    if raw.is_empty() {
        return None;
    }
    let sanitized = raw
        .chars()
        .filter(|ch| ch.is_ascii_alphanumeric() || matches!(ch, '-' | '_' | '/'))
        .collect::<String>()
        .to_ascii_lowercase();
    if sanitized.is_empty() {
        None
    } else {
        Some(sanitized)
    }
}

fn sanitize_description(raw: &str) -> Option<String> {
    let trimmed = raw.trim().trim_matches('"').trim();
    if trimmed.is_empty() {
        return None;
    }
    let normalized = trimmed.trim_end_matches('.').trim();
    if normalized.is_empty() {
        None
    } else {
        Some(normalized.to_string())
    }
}

fn sanitize_footer_value(raw: &str) -> Option<String> {
    let trimmed = raw.trim();
    if trimmed.is_empty() {
        None
    } else {
        Some(trimmed.to_string())
    }
}

fn interesting_file_stem(path: &str) -> Option<String> {
    let stem = Path::new(path)
        .file_stem()
        .and_then(|value| value.to_str())
        .map(|value| value.trim().to_ascii_lowercase())?;
    if stem.is_empty()
        || matches!(
            stem.as_str(),
            "mod" | "lib" | "main" | "readme" | "index" | "commands" | "router"
        )
    {
        None
    } else {
        Some(stem)
    }
}

fn normalize_name_status(raw: &str) -> String {
    let code = raw.chars().next().unwrap_or('M');
    match code {
        'A' => "added",
        'D' => "deleted",
        'R' => "renamed",
        'C' => "copied",
        'T' => "typechange",
        'U' => "unmerged",
        'M' => "modified",
        _ => "modified",
    }
    .to_string()
}

fn normalize_display_path(raw: &str) -> String {
    raw.trim().replace('\\', "/")
}

fn normalize_path_key(raw: &str) -> String {
    let cleaned = raw.trim().replace(['{', '}'], "").replace('\\', "/");
    if let Some((_, tail)) = cleaned.split_once("=>") {
        tail.trim().to_string()
    } else {
        cleaned
    }
}

fn truncate_for_prompt(text: &str, limit: usize) -> String {
    if text.chars().count() <= limit {
        return text.to_string();
    }

    let truncated = text.chars().take(limit).collect::<String>();
    format!(
        "{}\n\n[diff truncated after {} characters]",
        truncated, limit
    )
}

async fn git_output(project_root: &Path, args: &[&str]) -> Result<String, String> {
    git_output_with_env(project_root, args, &[]).await
}

async fn git_output_with_env(
    project_root: &Path,
    args: &[&str],
    envs: &[(&str, &std::ffi::OsStr)],
) -> Result<String, String> {
    let mut command = Command::new("git");
    command.current_dir(project_root).args(args);
    for (key, value) in envs {
        command.env(key, value);
    }

    let output = command
        .output()
        .await
        .map_err(|e| format!("Failed to run `git {}`: {}", args.join(" "), e))?;

    if !output.status.success() {
        let stderr = String::from_utf8_lossy(&output.stderr).trim().to_string();
        if stderr.is_empty() {
            return Err(format!(
                "`git {}` failed with status {}",
                args.join(" "),
                output.status
            ));
        }
        return Err(format!("`git {}` failed: {}", args.join(" "), stderr));
    }

    Ok(String::from_utf8_lossy(&output.stdout).trim().to_string())
}

struct TemporaryIndex {
    path: PathBuf,
}

impl Drop for TemporaryIndex {
    fn drop(&mut self) {
        let _ = fs::remove_file(&self.path);
    }
}

#[cfg(test)]
mod tests {
    use super::{
        infer_commit_type, parse_ai_commit_payload, render_commit_message, sanitize_scope,
        summarize_symbols, ConventionalCommit, FileChangeSummary, StatusEntry, WorktreeAnalysis,
    };
    use std::path::PathBuf;

    #[test]
    fn symbol_summary_tracks_new_and_changed_symbols() {
        let diff = r#"
diff --git a/crates/cli/src/commands/commit.rs b/crates/cli/src/commands/commit.rs
index 1111111..2222222 100644
--- a/crates/cli/src/commands/commit.rs
+++ b/crates/cli/src/commands/commit.rs
@@ -0,0 +1,3 @@
+pub struct CommitArgs {
+}
+pub async fn run_commit() {}
@@ -10,1 +12,1 @@ pub async fn old_name() {}
-pub async fn old_name() {}
+pub async fn old_name() {}
"#;
        let summary = summarize_symbols(diff);
        assert!(summary
            .new_functions
            .iter()
            .any(|item| item.contains("run_commit")));
        assert!(summary
            .new_types
            .iter()
            .any(|item| item.contains("CommitArgs")));
        assert!(summary
            .changed_functions
            .iter()
            .any(|item| item.contains("old_name")));
    }

    #[test]
    fn ai_payload_respects_forced_scope() {
        let raw = r#"{
  "type": "feat",
  "scope": "wrong",
  "description": "add commit command",
  "body": ["Summarize the worktree."],
  "breaking_change": false,
  "footers": []
}"#;

        let commit = parse_ai_commit_payload(raw, Some("cli")).expect("parse");
        assert_eq!(commit.scope.as_deref(), Some("cli"));
        assert_eq!(commit.commit_type, "feat");
    }

    #[test]
    fn renders_breaking_change_footer_once() {
        let rendered = render_commit_message(&ConventionalCommit {
            commit_type: "feat".to_string(),
            scope: Some("cli".to_string()),
            description: "add commit command".to_string(),
            body: vec!["Summarize the worktree.".to_string()],
            breaking_change: Some("existing automation must call xbp commit".to_string()),
            footers: Vec::new(),
        });

        assert!(rendered.contains("feat(cli)!: add commit command"));
        assert!(rendered.contains("BREAKING CHANGE: existing automation must call xbp commit"));
    }

    #[test]
    fn infers_feat_for_new_symbols() {
        let analysis = WorktreeAnalysis {
            repo_root: PathBuf::from("C:/repo"),
            repo_name: "xbp".to_string(),
            branch: Some("main".to_string()),
            status_entries: vec![StatusEntry {
                code: "??".to_string(),
                path: "crates/cli/src/commands/commit.rs".to_string(),
            }],
            files: vec![FileChangeSummary {
                path: "crates/cli/src/commands/commit.rs".to_string(),
                status: "added".to_string(),
                additions: 120,
                deletions: 0,
            }],
            total_additions: 120,
            total_deletions: 0,
            diff_text: String::new(),
            new_functions: vec!["run_commit (crates/cli/src/commands/commit.rs)".to_string()],
            changed_functions: Vec::new(),
            removed_functions: Vec::new(),
            new_types: vec!["CommitArgs (crates/cli/src/commands/commit.rs)".to_string()],
            changed_types: Vec::new(),
            removed_types: Vec::new(),
            hunk_contexts: Vec::new(),
        };

        assert_eq!(infer_commit_type(&analysis), "feat");
    }

    #[test]
    fn scope_sanitizer_keeps_valid_tokens() {
        assert_eq!(
            sanitize_scope(Some("CLI/tools")),
            Some("cli/tools".to_string())
        );
        assert_eq!(sanitize_scope(Some("  ")), None);
    }
}