fallow_engine/

project_config.rs

//! Project config resolution owned by the engine boundary.

use std::ffi::OsString;
use std::path::{Path, PathBuf};

use fallow_config::{
    ConfigLoadOptions, FallowConfig, ProductionAnalysis, ResolvedConfig, WorkspaceDiagnostic,
    WorkspaceInfo,
};
use fallow_types::output_format::OutputFormat;
use rustc_hash::FxHashSet;

use crate::{EngineError, EngineResult};

/// The production flags of one run: the global `--production` override and one
/// override per analysis (`--production-dead-code`, `--production-health`,
/// `--production-dupes`).
///
/// Every command and surface that runs more than one analysis reads the
/// production mode of each analysis from here, so the precedence has one
/// implementation: the flag of the analysis, then the global override, then
/// the config.
#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
pub struct ProductionFlags {
    /// The global override. The CLI sets `Some(true)` for `--production` and
    /// `None` without it. The programmatic API can also pass `Some(false)`.
    pub global: Option<bool>,
    /// Override for the dead-code analysis.
    pub dead_code: Option<bool>,
    /// Override for the health analysis.
    pub health: Option<bool>,
    /// Override for the duplication analysis.
    pub dupes: Option<bool>,
}

impl ProductionFlags {
    /// The flags of a CLI run, where `--production` can only switch the mode
    /// on.
    #[must_use]
    pub const fn from_cli(
        production: bool,
        dead_code: Option<bool>,
        health: Option<bool>,
        dupes: Option<bool>,
    ) -> Self {
        Self {
            global: if production { Some(true) } else { None },
            dead_code,
            health,
            dupes,
        }
    }

    /// The production override of a command that runs one analysis
    /// (`dead-code`, `dupes`, `health`): its own override, else
    /// `--production`. `None` means that the config decides.
    #[must_use]
    pub const fn single_analysis_override(production: bool, own: Option<bool>) -> Option<bool> {
        Self::from_cli(production, own, None, None).override_for(ProductionAnalysis::DeadCode)
    }

    /// The override flag of one analysis, without the global override.
    #[must_use]
    pub const fn own(self, analysis: ProductionAnalysis) -> Option<bool> {
        match analysis {
            ProductionAnalysis::DeadCode => self.dead_code,
            ProductionAnalysis::Health => self.health,
            ProductionAnalysis::Dupes => self.dupes,
        }
    }

    /// The production override of one analysis: its own flag, else the global
    /// override. `None` means that the config decides.
    #[must_use]
    pub const fn override_for(self, analysis: ProductionAnalysis) -> Option<bool> {
        match self.own(analysis) {
            Some(value) => Some(value),
            None => self.global,
        }
    }

    /// The production mode of one analysis as far as the flags decide, before
    /// the config is loaded. A missing flag reads as `false`.
    #[must_use]
    pub const fn mode(self, analysis: ProductionAnalysis) -> bool {
        match self.override_for(analysis) {
            Some(value) => value,
            None => false,
        }
    }

    /// The production mode of each analysis as far as the flags decide.
    #[must_use]
    pub const fn modes(self) -> ProductionModes {
        ProductionModes {
            dead_code: self.mode(ProductionAnalysis::DeadCode),
            health: self.mode(ProductionAnalysis::Health),
            dupes: self.mode(ProductionAnalysis::Dupes),
        }
    }

    /// The effective production mode of one analysis: the flags first, then
    /// the `production` setting of the config.
    #[must_use]
    pub const fn effective(
        self,
        analysis: ProductionAnalysis,
        config: fallow_config::ProductionConfig,
    ) -> bool {
        match self.override_for(analysis) {
            Some(value) => value,
            None => config.for_analysis(analysis),
        }
    }

    /// The effective production mode of each analysis.
    #[must_use]
    pub const fn effective_modes(self, config: fallow_config::ProductionConfig) -> ProductionModes {
        ProductionModes {
            dead_code: self.effective(ProductionAnalysis::DeadCode, config),
            health: self.effective(ProductionAnalysis::Health, config),
            dupes: self.effective(ProductionAnalysis::Dupes, config),
        }
    }
}

/// The production mode of each analysis of one run.
#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
pub struct ProductionModes {
    /// Production mode of the dead-code analysis.
    pub dead_code: bool,
    /// Production mode of the health analysis.
    pub health: bool,
    /// Production mode of the duplication analysis.
    pub dupes: bool,
}

impl ProductionModes {
    /// Whether health can reuse the dead-code parse: both run in the same mode.
    #[must_use]
    pub const fn dead_code_matches_health(self) -> bool {
        self.dead_code == self.health
    }

    /// Whether duplication can reuse the dead-code files: both run in the same
    /// mode.
    #[must_use]
    pub const fn dead_code_matches_dupes(self) -> bool {
        self.dead_code == self.dupes
    }

    /// Whether all three analyses run in the same mode.
    #[must_use]
    pub const fn all_match(self) -> bool {
        self.dead_code_matches_health() && self.dead_code_matches_dupes()
    }
}

/// Resolved project config plus the config file path when one was loaded.
#[derive(Debug)]
pub struct ProjectConfig {
    /// Fully resolved config for the project.
    pub config: ResolvedConfig,
    /// Path of the loaded config file; `None` when defaults were used.
    pub path: Option<PathBuf>,
    /// Workspace metadata discovered under the project root.
    pub workspaces: Vec<WorkspaceInfo>,
    /// Diagnostics from workspace discovery (undeclared or invalid members).
    pub workspace_diagnostics: Vec<WorkspaceDiagnostic>,
    /// Workspace discovery wall time in milliseconds, when measured.
    pub workspace_discovery_ms: Option<f64>,
    /// The plugin files, rule packs and `autoDiscover` directories that
    /// config resolution read.
    pub inputs: fallow_config::ConfigInputs,
    /// The content of [`Self::inputs`] just before resolution read them.
    pub inputs_before_resolve: fallow_config::ConfigInputsSnapshot,
}

/// Scalar config-loading knobs for one analysis family.
#[derive(Debug, Clone, Copy)]
pub struct ProjectConfigOptions {
    /// Output format the resolved config carries into rendering.
    pub output: OutputFormat,
    /// Bypass the parse cache for this run.
    pub no_cache: bool,
    /// Worker thread count recorded on the resolved config.
    pub threads: usize,
    /// Tri-state production override: `Some` forces production-only analysis
    /// on or off regardless of config, `None` defers to the config value.
    pub production_override: Option<bool>,
    /// Suppress progress notes on stderr.
    pub quiet: bool,
    /// Which analysis family's per-analysis production config to flatten.
    pub analysis: ProductionAnalysis,
    /// Permit `extends` config inheritance from remote URLs.
    pub allow_remote_extends: bool,
}

/// Resolved project configuration plus doctor-specific plugin diagnostics.
#[derive(Debug)]
#[non_exhaustive]
pub struct ProjectConfigReadiness {
    /// The same resolved configuration returned by [`config_for_project_analysis`].
    pub project: ProjectConfig,
    /// Unresolved resources named explicitly by the `plugins` config field.
    pub configured_plugin_diagnostics: Vec<fallow_config::ConfiguredPluginDiagnostic>,
}

/// Resolve the analysis config for a project.
///
/// # Errors
///
/// Returns an error when an explicit config cannot be loaded or automatic
/// config discovery finds an invalid config.
pub fn config_for_project(root: &Path, config_path: Option<&Path>) -> EngineResult<ProjectConfig> {
    config_for_project_with_load_options(root, config_path, ConfigLoadOptions::default())
}

/// Resolve project config with an explicit inheritance trust policy.
///
/// # Errors
///
/// Returns an error when config loading or validation fails.
pub fn config_for_project_with_load_options(
    root: &Path,
    config_path: Option<&Path>,
    load_options: ConfigLoadOptions,
) -> EngineResult<ProjectConfig> {
    let user_config = load_user_config(root, config_path, load_options)?;
    let (mut config, path) = match user_config {
        Some((config, path)) => (config, Some(path)),
        None => (FallowConfig::default(), None),
    };
    if path.is_some() {
        config.production = config
            .production
            .for_analysis(ProductionAnalysis::DeadCode)
            .into();
        validate_boundaries_and_rule_packs(root, &config)?;
    }
    let threads = std::thread::available_parallelism().map_or(1, std::num::NonZeroUsize::get);
    let inputs = fallow_config::ConfigInputs::new(root, &config);
    let inputs_before_resolve = inputs.snapshot();
    let mut resolved = config.resolve(
        root.to_path_buf(),
        OutputFormat::Human,
        threads,
        false,
        true,
        None,
    );
    apply_env_overrides(&mut resolved);
    let (workspaces, workspace_diagnostics, workspace_discovery_ms) =
        collect_workspace_metadata(&resolved)?;
    Ok(ProjectConfig {
        inputs,
        inputs_before_resolve,
        config: resolved,
        path,
        workspaces,
        workspace_diagnostics,
        workspace_discovery_ms: Some(workspace_discovery_ms),
    })
}

/// Resolve the parse-cache size limit for a resolved config.
#[must_use]
pub(crate) fn resolve_cache_max_size_bytes(config: &ResolvedConfig) -> usize {
    config
        .cache_max_size_mb
        .map_or(fallow_extract::cache::DEFAULT_CACHE_MAX_SIZE, |mb| {
            (mb as usize).saturating_mul(1024 * 1024)
        })
}

pub(crate) fn default_project_config(root: &Path) -> ProjectConfig {
    let threads = std::thread::available_parallelism().map_or(1, std::num::NonZeroUsize::get);
    let inputs = fallow_config::ConfigInputs::new(root, &FallowConfig::default());
    let inputs_before_resolve = inputs.snapshot();
    let mut config = FallowConfig::default().resolve(
        root.to_path_buf(),
        OutputFormat::Human,
        threads,
        false,
        true,
        None,
    );
    apply_env_overrides(&mut config);
    let (workspaces, workspace_diagnostics, workspace_discovery_ms) =
        collect_workspace_metadata_lossy(&config);
    ProjectConfig {
        inputs,
        inputs_before_resolve,
        config,
        path: None,
        workspaces,
        workspace_diagnostics,
        workspace_discovery_ms: Some(workspace_discovery_ms),
    }
}

/// Resolve config for a specific analysis without depending on the CLI crate.
///
/// This mirrors the CLI's core config semantics: explicit production overrides
/// are applied before resolution, per-analysis production config is flattened
/// for the requested analysis, and boundary / external plugin / rule-pack
/// validation happens before the resolved config reaches the engine.
///
/// # Errors
///
/// Returns an engine error when config loading or validation fails.
pub fn config_for_project_analysis(
    root: &Path,
    config_path: Option<&Path>,
    options: ProjectConfigOptions,
) -> EngineResult<ProjectConfig> {
    resolve_project_config_analysis(root, config_path, options).map(|(project, _)| project)
}

/// Resolve project configuration and collect typed readiness diagnostics for
/// plugin resources named explicitly by the user configuration.
///
/// # Errors
///
/// Returns an engine error when config loading or validation fails.
pub fn config_for_project_readiness(
    root: &Path,
    config_path: Option<&Path>,
    options: ProjectConfigOptions,
) -> EngineResult<ProjectConfigReadiness> {
    let (project, configured_plugin_paths) =
        resolve_project_config_analysis(root, config_path, options)?;
    let configured_plugin_diagnostics =
        fallow_config::diagnose_configured_external_plugins(root, &configured_plugin_paths);
    Ok(ProjectConfigReadiness {
        project,
        configured_plugin_diagnostics,
    })
}

fn resolve_project_config_analysis(
    root: &Path,
    config_path: Option<&Path>,
    options: ProjectConfigOptions,
) -> EngineResult<(ProjectConfig, Vec<String>)> {
    let user_config = load_user_config(
        root,
        config_path,
        ConfigLoadOptions {
            allow_remote_extends: options.allow_remote_extends,
        },
    )?;
    let loaded_user_config = user_config.is_some();
    let (mut config, path) = match user_config {
        Some((config, path)) => (config, Some(path)),
        None => (
            FallowConfig {
                production: options.production_override.unwrap_or(false).into(),
                ..FallowConfig::default()
            },
            None,
        ),
    };

    if loaded_user_config {
        let production = ProductionFlags {
            global: options.production_override,
            ..ProductionFlags::default()
        }
        .effective(options.analysis, config.production);
        config.production = production.into();
    }
    validate_config(root, &config)?;
    let configured_plugin_paths = config.plugins.clone();
    let inputs = fallow_config::ConfigInputs::new(root, &config);
    let inputs_before_resolve = inputs.snapshot();
    let mut resolved = config.resolve(
        root.to_path_buf(),
        options.output,
        options.threads,
        options.no_cache,
        options.quiet,
        None,
    );
    apply_env_overrides(&mut resolved);
    let (workspaces, workspace_diagnostics, workspace_discovery_ms) =
        collect_workspace_metadata(&resolved)?;
    Ok((
        ProjectConfig {
            inputs,
            inputs_before_resolve,
            config: resolved,
            path,
            workspaces,
            workspace_diagnostics,
            workspace_discovery_ms: Some(workspace_discovery_ms),
        },
        configured_plugin_paths,
    ))
}

/// Apply the environment variables that every host honors. The CLI, the LSP,
/// the MCP server and the Node bindings all load config through this module,
/// so a variable applied here has the same meaning on each surface.
fn apply_env_overrides(config: &mut ResolvedConfig) {
    apply_env_overrides_from(config, |name| std::env::var_os(name));
}

fn apply_env_overrides_from(
    config: &mut ResolvedConfig,
    lookup: impl Fn(&str) -> Option<OsString>,
) {
    let max_file_size_mb = lookup("FALLOW_MAX_FILE_SIZE")
        .and_then(|raw| raw.to_str().and_then(|raw| raw.trim().parse::<u32>().ok()));
    if let Some(max_file_size_mb) = max_file_size_mb {
        config.max_file_size_bytes =
            fallow_config::resolve_max_file_size_bytes(Some(max_file_size_mb));
    }
    if let Some(dir) =
        lookup(fallow_config::CACHE_DIR_ENV).and_then(fallow_config::cache_dir_from_env_value)
    {
        config.override_cache_dir(dir);
    }
    if let Some(mb) = lookup(fallow_config::CACHE_MAX_SIZE_ENV)
        .as_deref()
        .and_then(fallow_config::cache_max_size_from_env_value)
    {
        config.cache_max_size_mb = Some(mb);
    }
    config.apply_package_baselines_env(lookup(fallow_config::PACKAGE_BASELINES_ENV).as_deref());
}

pub(crate) fn collect_workspace_metadata(
    config: &ResolvedConfig,
) -> EngineResult<(Vec<WorkspaceInfo>, Vec<WorkspaceDiagnostic>, f64)> {
    let start = std::time::Instant::now();
    let (workspaces, diagnostics) =
        fallow_config::discover_workspaces_with_diagnostics(&config.root, &config.ignore_patterns)
            .map_err(|err| EngineError::new(err.to_string()))?;
    let diagnostics = with_undeclared_workspace_diagnostics(config, &workspaces, diagnostics);
    let elapsed_ms = start.elapsed().as_secs_f64() * 1000.0;
    Ok((workspaces, diagnostics, elapsed_ms))
}

fn collect_workspace_metadata_lossy(
    config: &ResolvedConfig,
) -> (Vec<WorkspaceInfo>, Vec<WorkspaceDiagnostic>, f64) {
    collect_workspace_metadata(config).unwrap_or_default()
}

fn with_undeclared_workspace_diagnostics(
    config: &ResolvedConfig,
    workspaces: &[WorkspaceInfo],
    mut diagnostics: Vec<WorkspaceDiagnostic>,
) -> Vec<WorkspaceDiagnostic> {
    let mut existing: FxHashSet<PathBuf> = diagnostics
        .iter()
        .map(|diagnostic| {
            dunce::canonicalize(&diagnostic.path).unwrap_or_else(|_| diagnostic.path.clone())
        })
        .collect();
    for diagnostic in fallow_config::find_undeclared_workspaces_with_ignores(
        &config.root,
        workspaces,
        &config.ignore_patterns,
    ) {
        let canonical =
            dunce::canonicalize(&diagnostic.path).unwrap_or_else(|_| diagnostic.path.clone());
        if existing.insert(canonical) {
            diagnostics.push(diagnostic);
        }
    }
    diagnostics
}

fn load_user_config(
    root: &Path,
    config_path: Option<&Path>,
    options: ConfigLoadOptions,
) -> EngineResult<Option<(FallowConfig, PathBuf)>> {
    if let Some(path) = config_path {
        let config = FallowConfig::load_with_options(path, options)
            .map_err(|err| EngineError::new(format!("invalid config: {err:#}")))?;
        return Ok(Some((config, path.to_path_buf())));
    }
    FallowConfig::find_and_load_with_options(root, options)
        .map_err(|err| EngineError::new(format!("invalid config: {err}")))
}

fn validate_config(root: &Path, config: &FallowConfig) -> EngineResult<()> {
    fallow_config::discover_and_validate_external_plugins(root, &config.plugins)
        .map_err(|errors| joined_config_errors("invalid external plugin definition", &errors))?;
    validate_boundaries_and_rule_packs(root, config)
}

fn validate_boundaries_and_rule_packs(root: &Path, config: &FallowConfig) -> EngineResult<()> {
    config
        .validate_resolved_boundaries(root)
        .map_err(|errors| joined_config_errors("invalid boundary configuration", &errors))?;
    let packs = fallow_config::load_rule_packs(root, &config.rule_packs)
        .map_err(|errors| joined_config_errors("invalid rule pack", &errors))?;
    let zone_errors = fallow_config::validate_rule_pack_zones(
        root,
        &config.boundaries,
        &config.rule_packs,
        &packs,
    );
    if !zone_errors.is_empty() {
        return Err(joined_config_errors("invalid rule pack", &zone_errors));
    }
    Ok(())
}

fn joined_config_errors(label: &str, errors: &[impl ToString]) -> EngineError {
    let joined = errors
        .iter()
        .map(ToString::to_string)
        .collect::<Vec<_>>()
        .join("\n  - ");
    EngineError::new(format!("{label}:\n  - {joined}"))
}

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

    fn lookup_from(
        vars: &'static [(&'static str, &'static str)],
    ) -> impl Fn(&str) -> Option<OsString> {
        move |name| {
            vars.iter()
                .find(|(key, _)| *key == name)
                .map(|(_, value)| OsString::from(value))
        }
    }

    #[test]
    fn env_overrides_reach_every_host_config() {
        let root = Path::new("/repo");
        let mut config = default_project_config(root).config;

        apply_env_overrides_from(
            &mut config,
            lookup_from(&[
                ("FALLOW_MAX_FILE_SIZE", "7"),
                ("FALLOW_CACHE_DIR", ".cache/fallow"),
                ("FALLOW_CACHE_MAX_SIZE", "64"),
            ]),
        );

        assert_eq!(
            config.max_file_size_bytes,
            fallow_config::resolve_max_file_size_bytes(Some(7))
        );
        assert_eq!(config.cache_dir, root.join(".cache/fallow"));
        assert_eq!(config.cache_max_size_mb, Some(64));
    }

    #[test]
    fn invalid_or_empty_env_values_keep_the_config() {
        let root = Path::new("/repo");
        let mut config = default_project_config(root).config;
        let before_dir = config.cache_dir.clone();
        let before_max = config.cache_max_size_mb;

        apply_env_overrides_from(
            &mut config,
            lookup_from(&[("FALLOW_CACHE_DIR", ""), ("FALLOW_CACHE_MAX_SIZE", "0")]),
        );

        assert_eq!(config.cache_dir, before_dir);
        assert_eq!(config.cache_max_size_mb, before_max);
    }

    #[test]
    fn package_baselines_env_turns_the_map_off_only_for_a_false_value() {
        let root = Path::new("/repo");
        let with_map = || {
            let mut config = default_project_config(root).config;
            config
                .workspace_changed_since
                .insert("packages/web".to_owned(), "main".to_owned());
            config
        };

        let mut disabled = with_map();
        apply_env_overrides_from(
            &mut disabled,
            lookup_from(&[("FALLOW_PACKAGE_BASELINES", " Off ")]),
        );
        assert!(disabled.workspace_changed_since.is_empty());

        for value in ["true", "1", "maybe", ""] {
            let mut kept = with_map();
            let vars: &'static [(&'static str, &'static str)] = match value {
                "true" => &[("FALLOW_PACKAGE_BASELINES", "true")],
                "1" => &[("FALLOW_PACKAGE_BASELINES", "1")],
                "maybe" => &[("FALLOW_PACKAGE_BASELINES", "maybe")],
                _ => &[("FALLOW_PACKAGE_BASELINES", "")],
            };
            apply_env_overrides_from(&mut kept, lookup_from(vars));
            assert_eq!(kept.workspace_changed_since.len(), 1, "{value:?}");
        }
    }
}