changepacks_core/

project_finder.rs

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

use crate::project::Project;
use anyhow::{Context, Result};
use async_trait::async_trait;

/// Generates `projects()`, `projects_mut()`, `project_count()`,
/// `extend_projects()`, `extend_projects_mut()`, and `contains_project()` for
/// finders backed by a `projects: HashMap<PathBuf, Project>` field.
///
/// The `extend_projects` / `extend_projects_mut` bodies drain
/// `self.projects.values()` / `self.projects.values_mut()` straight into the
/// caller's buffer, so the intermediate `Vec` that `projects()` /
/// `projects_mut()` has to materialize is never built. Yield order is
/// `HashMap::values()` / `HashMap::values_mut()` in both bodies, so overriding
/// is order-preserving with respect to the defaulted
/// [`ProjectFinder::extend_projects`] and
/// [`ProjectFinder::extend_projects_mut`].
///
/// The `contains_project` body is the O(1) hashed probe the map already
/// offers, replacing the defaulted linear scan over `projects()` — see
/// [`ProjectFinder::contains_project`]. `HashMap<PathBuf, Project>` borrows
/// its keys as `Path`, so the probe neither allocates nor clones.
#[macro_export]
macro_rules! impl_projects_hashmap_accessors {
    () => {
        fn projects(&self) -> ::std::vec::Vec<&$crate::Project> {
            self.projects.values().collect::<::std::vec::Vec<_>>()
        }
        fn projects_mut(&mut self) -> ::std::vec::Vec<&mut $crate::Project> {
            self.projects.values_mut().collect::<::std::vec::Vec<_>>()
        }
        fn project_count(&self) -> ::std::primitive::usize {
            self.projects.len()
        }
        fn extend_projects<'a>(&'a self, out: &mut ::std::vec::Vec<&'a $crate::Project>) {
            out.extend(self.projects.values());
        }
        fn extend_projects_mut<'a>(
            &'a mut self,
            out: &mut ::std::vec::Vec<&'a mut $crate::Project>,
        ) {
            out.extend(self.projects.values_mut());
        }
        fn contains_project(&self, path: &::std::path::Path) -> ::std::primitive::bool {
            self.projects.contains_key(path)
        }
    };
}

/// Generates `dependencies()` and `add_dependency()` for types with a
/// `dependencies: HashSet<String>` field.
///
/// `add_dependency` probes membership with `contains` before allocating.
/// `HashSet<String>` borrows its keys as `str`, so the probe is
/// allocation-free, while the previous unconditional
/// `insert(dependency.to_string())` heap-allocated a fresh `String` that
/// `insert` immediately dropped whenever the name was already present.
/// Callers hit that duplicate path routinely: a manifest that lists the same
/// package in more than one dependency section (e.g. `dependencies` and
/// `peerDependencies` in `package.json`) is walked section by section, and
/// every section after the first re-adds a name already in the set. The set
/// contents are unchanged either way — a `HashSet` keeps its existing key on
/// a duplicate insert — so this is purely allocation elision.
#[macro_export]
macro_rules! impl_dependencies_accessors {
    () => {
        fn dependencies(&self) -> &::std::collections::HashSet<::std::string::String> {
            &self.dependencies
        }
        fn add_dependency(&mut self, dependency: &str) {
            if !self.dependencies.contains(dependency) {
                self.dependencies
                    .insert(::std::string::ToString::to_string(dependency));
            }
        }
    };
}

/// Generates `is_publishable_by_default()` for package/workspace structs with a
/// `publishable_by_default: bool` field.
///
/// Contract: the implementing struct MUST own a field named exactly
/// `publishable_by_default` of type `bool` — the same field
/// [`impl_discovered_new!`] initializes. Language crates whose publishability
/// is derived from a differently named field (e.g. Java's `has_publish_task`)
/// must keep their hand-rolled body instead.
#[macro_export]
macro_rules! impl_publishable_by_default {
    () => {
        fn is_publishable_by_default(&self) -> ::std::primitive::bool {
            self.publishable_by_default
        }
    };
}

/// Generates const-backed publish command defaults.
///
/// Two arguments return `Some($dry_run.to_string())`; one argument returns
/// `None` for ecosystems without a built-in dry-run command.
#[macro_export]
macro_rules! impl_const_publish_commands {
    ($publish:path, $dry_run:path) => {
        fn default_publish_command(&self) -> ::std::string::String {
            $publish.to_string()
        }
        fn default_dry_run_publish_command(&self) -> ::std::option::Option<::std::string::String> {
            ::std::option::Option::Some($dry_run.to_string())
        }
    };
    // CSharp variant: `dotnet nuget push` has no built-in `--dry-run`
    // mode, so the default returns `None`. The actual dry-run flow
    // lives in `CSharpPackage::dry_run_publish` (see
    // `crates/csharp/src/dry_run.rs::resolve_and_run_dry_run`), which
    // honors `config.publishDryRun` overrides first and falls back to a
    // managed `dotnet pack` + `dotnet nuget push` against ephemeral
    // `tempfile::TempDir` directories when no override is set.
    ($publish:path) => {
        fn default_publish_command(&self) -> ::std::string::String {
            $publish.to_string()
        }
        fn default_dry_run_publish_command(&self) -> ::std::option::Option<::std::string::String> {
            ::std::option::Option::None
        }
    };
}

/// Generates the `get_publish_command` / `get_dry_run_publish_command`
/// trait defaults shared by [`Package`](crate::Package) and
/// [`Workspace`](crate::Workspace).
///
/// Both traits resolve their publish commands through the exact same
/// [`crate::publish`] ladder — only the surrounding doc prose used to
/// differ — so the bodies live here once instead of being kept
/// byte-identical by hand in two files.
///
/// Contract: the invoking trait MUST already declare `relative_path()`,
/// `language()`, `default_publish_command()`, and
/// `default_dry_run_publish_command()`.
///
/// The sibling `publish` / `dry_run_publish` defaults live in
/// [`impl_publish_flows!`], which takes the differing missing-directory
/// message constant as an argument.
#[macro_export]
macro_rules! impl_publish_command_resolvers {
    () => {
        /// Get the publish command for this project, checking config first.
        ///
        /// The `default_publish_command()` closure is `FnOnce`, so the
        /// project's language-specific default (e.g. Node's
        /// `detect_package_manager_recursive`, which walks the ancestor chain
        /// with sync filesystem stats) is only invoked when config supplies
        /// neither a per-path nor a per-language override — the common case
        /// where the user configures a custom publish command in
        /// `.changepacks/config.json` now avoids one `String` allocation and,
        /// for Node, the ancestor-walking probe.
        fn get_publish_command(&self, config: &$crate::Config) -> ::std::string::String {
            $crate::publish::resolve_publish_command(
                self.relative_path(),
                self.language(),
                || self.default_publish_command(),
                config,
            )
        }

        /// Get the dry-run publish command for this project, checking config
        /// first, then falling back to the project's
        /// `default_dry_run_publish_command`.
        ///
        /// Mirrors `get_publish_command` — the default closure is `FnOnce` so
        /// it is only invoked on the cache-miss path.
        fn get_dry_run_publish_command(
            &self,
            config: &$crate::Config,
        ) -> ::std::option::Option<::std::string::String> {
            $crate::publish::resolve_dry_run_publish_command(
                self.relative_path(),
                self.language(),
                || self.default_dry_run_publish_command(),
                config,
            )
        }
    };
}

/// Generates the `publish` / `dry_run_publish` trait defaults shared by
/// [`Package`](crate::Package) and [`Workspace`](crate::Workspace).
///
/// `$dir_not_found` is the only thing that differed between the two hand-kept
/// copies: [`crate::publish::PACKAGE_DIR_NOT_FOUND`] for `Package`,
/// [`crate::publish::WORKSPACE_DIR_NOT_FOUND`] for `Workspace`.
///
/// Contract: the invoking trait MUST already declare `path()`,
/// `get_publish_command()`, and `get_dry_run_publish_command()` — the latter
/// two come from [`impl_publish_command_resolvers!`].
///
/// The two methods are emitted in the shape `#[async_trait]` would produce
/// rather than as `async fn`, because an attribute macro cannot see through a
/// `macro_rules!` invocation in a trait body: `#[async_trait]` runs BEFORE
/// this macro expands, so an `async fn` emitted here would survive as a
/// native RPITIT method and make `Package` / `Workspace` dyn-incompatible,
/// breaking `Box<dyn Package>` in `crate::Project`. Emitting the boxed-future
/// signature keeps the defaults object safe and keeps them overridable by
/// `#[async_trait]` impls in the language crates (Node, Java, C#), whose
/// generated signatures this shape matches.
#[macro_export]
macro_rules! impl_publish_flows {
    ($dir_not_found:path) => {
        /// Publish this project using the configured command or default.
        ///
        /// # Errors
        /// Returns error if the publish command fails to spawn or the project
        /// directory is missing. A non-zero exit code is reported via
        /// `PublishOutput::success = false`.
        fn publish<'life0, 'life1, 'async_trait>(
            &'life0 self,
            config: &'life1 $crate::Config,
        ) -> ::core::pin::Pin<
            ::std::boxed::Box<
                dyn ::core::future::Future<
                        Output = ::anyhow::Result<$crate::publish::PublishOutput>,
                    > + ::core::marker::Send
                    + 'async_trait,
            >,
        >
        where
            'life0: 'async_trait,
            'life1: 'async_trait,
            Self: 'async_trait,
        {
            ::std::boxed::Box::pin(async move {
                let command = self.get_publish_command(config);
                $crate::publish::run_publish_flow(&command, self.path(), &[], $dir_not_found).await
            })
        }

        /// Run the publish command in dry-run mode to verify the pre-release
        /// flow works without actually publishing.
        ///
        /// Returns `Ok(Some(output))` with the captured command output, or
        /// `Ok(None)` when the language does not support a dry-run mode and
        /// the user has not provided an override in `config.publish_dry_run`.
        ///
        /// # Errors
        /// Returns error if the dry-run command fails to spawn or the project
        /// directory is missing. A non-zero exit code is reported via
        /// `PublishOutput::success = false`.
        fn dry_run_publish<'life0, 'life1, 'async_trait>(
            &'life0 self,
            config: &'life1 $crate::Config,
        ) -> ::core::pin::Pin<
            ::std::boxed::Box<
                dyn ::core::future::Future<
                        Output = ::anyhow::Result<
                            ::std::option::Option<$crate::publish::PublishOutput>,
                        >,
                    > + ::core::marker::Send
                    + 'async_trait,
            >,
        >
        where
            'life0: 'async_trait,
            'life1: 'async_trait,
            Self: 'async_trait,
        {
            ::std::boxed::Box::pin(async move {
                let command = self.get_dry_run_publish_command(config);
                $crate::publish::run_dry_run_publish_flow(
                    command.as_deref(),
                    self.path(),
                    &[],
                    $dir_not_found,
                )
                .await
            })
        }
    };
}

/// Generates the `check_changed`, `is_publishable_by_default`, and
/// `is_dry_run_publishable_by_default` trait defaults shared by
/// [`Package`](crate::Package) and [`Workspace`](crate::Workspace).
///
/// These three bodies were byte-identical hand-kept copies in `package.rs`
/// and `workspace.rs`; they live here once for the same reason
/// [`impl_publish_flows!`] and [`impl_publish_command_resolvers!`] do.
///
/// Contract: the invoking trait MUST already declare `is_changed()`,
/// `set_changed()`, and `path()` (the manifest path — its parent is the
/// project directory). Like its two publish siblings, this macro emits trait
/// *default methods*, so it is only meaningful inside the `Package` /
/// `Workspace` trait definitions in this crate.
///
/// `check_changed` is a sync default, so unlike [`impl_publish_flows!`] it
/// needs no `Pin<Box<dyn Future>>` desugaring and does not interact with
/// `#[async_trait]`.
///
/// The emitted `check_changed` is monotonic: it early-returns once the
/// project is already changed and only ever flips `changed` to `true` via the
/// pure, stateless `should_mark_changed`. That is the invariant
/// [`ProjectFinder::check_changed_many`] relies on for its project-major loop
/// order and early `break`.
#[macro_export]
macro_rules! impl_shared_project_defaults {
    () => {
        /// # Errors
        /// Returns error if the parent path cannot be determined.
        fn check_changed(&mut self, path: &::std::path::Path) -> ::anyhow::Result<()> {
            if self.is_changed() {
                return ::core::result::Result::Ok(());
            }
            if $crate::change_detection::should_mark_changed(path, self.path())? {
                self.set_changed(true);
            }
            ::core::result::Result::Ok(())
        }

        /// Whether this project should be included in publish runs when no
        /// project-path or language command override is configured.
        fn is_publishable_by_default(&self) -> ::std::primitive::bool {
            true
        }

        /// Whether this project should be included in dry-run publish runs when no
        /// project-path or language command override is configured.
        fn is_dry_run_publishable_by_default(&self) -> ::std::primitive::bool {
            self.is_publishable_by_default()
        }
    };
}

/// Generates the shared basic accessors for package/workspace structs with
/// `name`, `version`, `path`, `relative_path`, and `is_changed` fields.
#[macro_export]
macro_rules! impl_basic_accessors {
    () => {
        fn name(&self) -> ::std::option::Option<&::std::primitive::str> {
            self.name.as_deref()
        }
        fn version(&self) -> ::std::option::Option<&::std::primitive::str> {
            self.version.as_deref()
        }
        fn path(&self) -> &::std::path::Path {
            &self.path
        }
        fn relative_path(&self) -> &::std::path::Path {
            &self.relative_path
        }
        fn is_changed(&self) -> ::std::primitive::bool {
            self.is_changed
        }
        fn set_changed(&mut self, changed: ::std::primitive::bool) {
            self.is_changed = changed;
        }
        fn set_name(&mut self, name: ::std::string::String) {
            self.name = ::std::option::Option::Some(name);
        }
    };
}

/// Asserts the `name()` / `set_name()` round trip generated by
/// [`impl_basic_accessors!`] for the project value produced by `$project`.
///
/// Every language crate had a byte-identical `test_set_name` body — build a
/// name-less package/workspace, assert `name()` is `None`, `set_name` it, then
/// assert `name()` observes the new value — differing only in the concrete type
/// and its manifest path. All eleven exercise the exact same
/// [`impl_basic_accessors!`] expansion, so the assertions live here once and
/// each site supplies only its own constructor.
///
/// Contract: `$project` must evaluate to a value whose `name` starts as `None`,
/// and the `Package` or `Workspace` trait supplying `name()` / `set_name()`
/// must be in scope at the call site.
#[macro_export]
macro_rules! assert_set_name_roundtrip {
    ($project:expr) => {{
        let mut project = $project;
        ::std::assert_eq!(
            project.name(),
            ::std::option::Option::None,
            "a project constructed with no name must report None"
        );
        project.set_name(::std::string::ToString::to_string("my-project"));
        ::std::assert_eq!(
            project.name(),
            ::std::option::Option::Some("my-project"),
            "set_name must be observable through name()"
        );
    }};
}

/// Asserts the `is_changed()` / `set_changed()` round trip generated by
/// [`impl_basic_accessors!`] for the project value produced by `$project`.
///
/// Every language crate had a byte-identical `set_changed` body — build a
/// project, assert it starts unchanged, flip it to `true`, assert, flip it back
/// to `false`, assert — differing only in the concrete type and its manifest
/// path. All eleven exercise the exact same [`impl_basic_accessors!`]
/// expansion, so the assertions live here once and each site supplies only its
/// own constructor.
///
/// Contract: `$project` must evaluate to a freshly constructed value whose
/// `is_changed` starts as `false`, and the `Package` or `Workspace` trait
/// supplying `is_changed()` / `set_changed()` must be in scope at the call site.
#[macro_export]
macro_rules! assert_set_changed_roundtrip {
    ($project:expr) => {{
        let mut project = $project;
        ::std::assert!(
            !project.is_changed(),
            "a freshly constructed project must start unchanged"
        );
        project.set_changed(true);
        ::std::assert!(
            project.is_changed(),
            "set_changed(true) must be observable through is_changed()"
        );
        project.set_changed(false);
        ::std::assert!(
            !project.is_changed(),
            "set_changed(false) must clear is_changed() again"
        );
    }};
}

/// Asserts the `dependencies()` / `add_dependency()` round trip generated by
/// [`impl_dependencies_accessors!`] for the project value produced by
/// `$project`.
///
/// Every language crate had a byte-identical dependencies body — build a
/// project, assert `dependencies()` starts empty, add two names, assert the set
/// has both and a length of two, re-add the first, assert the length is still
/// two — differing only in the concrete type and the two dependency-name
/// literals. All ten exercise the exact same [`impl_dependencies_accessors!`]
/// expansion, including its `contains`-before-`insert` duplicate path, so the
/// assertions live here once and each site supplies only its own constructor
/// and names.
///
/// Contract: `$project` must evaluate to a freshly constructed value whose
/// `dependencies` set starts empty, `$first` and `$second` must be distinct
/// `&str` names, and the `Package` or `Workspace` trait supplying
/// `dependencies()` / `add_dependency()` must be in scope at the call site.
#[macro_export]
macro_rules! assert_dependencies_roundtrip {
    ($project:expr, $first:expr, $second:expr) => {{
        let mut project = $project;
        ::std::assert!(
            project.dependencies().is_empty(),
            "a freshly constructed project must start with no dependencies"
        );
        project.add_dependency($first);
        project.add_dependency($second);
        let deps = project.dependencies();
        ::std::assert_eq!(deps.len(), 2, "both added dependencies must be recorded");
        ::std::assert!(
            deps.contains($first),
            "add_dependency must be observable through dependencies()"
        );
        ::std::assert!(
            deps.contains($second),
            "add_dependency must be observable through dependencies()"
        );
        project.add_dependency($first);
        ::std::assert_eq!(
            project.dependencies().len(),
            2,
            "re-adding an existing dependency must not grow the set"
        );
    }};
}

/// Generates constructors for discovered package/workspace structs with a
/// `publishable_by_default` field.
#[macro_export]
macro_rules! impl_discovered_new {
    () => {
        #[must_use]
        pub fn new(
            name: ::std::option::Option<::std::string::String>,
            version: ::std::option::Option<::std::string::String>,
            path: ::std::path::PathBuf,
            relative_path: ::std::path::PathBuf,
        ) -> Self {
            Self::new_discovered(name, version, path, relative_path, true)
        }

        #[must_use]
        pub(crate) fn new_discovered(
            name: ::std::option::Option<::std::string::String>,
            version: ::std::option::Option<::std::string::String>,
            path: ::std::path::PathBuf,
            relative_path: ::std::path::PathBuf,
            publishable_by_default: ::std::primitive::bool,
        ) -> Self {
            Self {
                name,
                version,
                path,
                relative_path,
                is_changed: false,
                publishable_by_default,
                dependencies: ::std::collections::HashSet::new(),
            }
        }
    };
}

/// Declares a discovered package/workspace struct plus its constructors.
///
/// Five language types — `PythonPackage`, `PythonWorkspace`, `DartPackage`,
/// `DartWorkspace` and `CSharpPackage` — declared the exact same seven private
/// fields (`name`, `version`, `path`, `relative_path`, `is_changed`,
/// `publishable_by_default`, `dependencies`) and each immediately followed the
/// declaration with an inherent impl containing
/// [`impl_discovered_new!`](crate::impl_discovered_new). Those field names are
/// already hard-coded by that constructor macro, so the declarations were not
/// independently variable: this macro makes the coupling explicit and keeps the
/// layout in one place.
///
/// Node, Rust and Java are intentionally NOT expressible here — they carry
/// extra fields (`package_manager`, the workspace-inheritance trio) or lack
/// `publishable_by_default` entirely — and keep their hand-written
/// declarations.
///
/// Additional inherent methods stay in a separate `impl` block beside the
/// invocation (see `CSharpPackage`'s command-runner helpers).
///
/// ```ignore
/// changepacks_core::declare_discovered_project!(
///     /// Doc comments and other outer attributes pass through.
///     pub struct PythonPackage
/// );
/// ```
#[macro_export]
macro_rules! declare_discovered_project {
    ($(#[$meta:meta])* pub struct $name:ident) => {
        $(#[$meta])*
        #[derive(::std::fmt::Debug)]
        pub struct $name {
            name: ::std::option::Option<::std::string::String>,
            version: ::std::option::Option<::std::string::String>,
            path: ::std::path::PathBuf,
            relative_path: ::std::path::PathBuf,
            is_changed: ::std::primitive::bool,
            publishable_by_default: ::std::primitive::bool,
            dependencies: ::std::collections::HashSet<::std::string::String>,
        }

        impl $name {
            $crate::impl_discovered_new!();
        }
    };
}

/// Builds the [`Project`](crate::Project) a finder just discovered, choosing
/// the `Workspace` or `Package` variant from `$is_workspace`.
///
/// Four finders — Node, Python, Dart and Java — ended their `visit()` with the
/// same ten-line `if is_workspace { Project::Workspace(Box::new(WsCtor(..))) }
/// else { Project::Package(Box::new(PkgCtor(..))) }`, and in every copy the
/// argument list was byte-identical in both arms because the finder had
/// already hoisted the shared `name` / `version` / `path_key` bindings above
/// the branch. Only the two constructor paths and that argument list actually
/// vary, so they are what this macro takes.
///
/// Evaluation is unchanged from the hand-written shape: `$is_workspace` is
/// evaluated once, and each `$arg` is evaluated exactly once and only inside
/// the branch that is taken. A `path_key.clone()` argument therefore still
/// performs exactly one `PathBuf` allocation per visit, not two, and the
/// moved-once `name` / `version` bindings keep type-checking because only one
/// arm ever runs.
///
/// Deliberately NOT used by two finders:
/// - `crates/csharp/src/finder.rs` has no workspace variant — it always
///   constructs a `Project::Package`, so there is no branch to factor out.
/// - `crates/rust/src/finder.rs` builds its two variants from separate
///   control-flow blocks with different argument lists, not from one `if/else`
///   over a shared list.
///
/// ```ignore
/// let mut project = changepacks_core::discovered_project!(
///     is_workspace,
///     NodeWorkspace::new_discovered,
///     NodePackage::new_discovered,
///     name,
///     version,
///     path_key.clone(),
///     relative_path_key,
///     package_manager,
///     publishable_by_default,
/// );
/// ```
#[macro_export]
macro_rules! discovered_project {
    ($is_workspace:expr, $ws:path, $pkg:path, $($arg:expr),* $(,)?) => {
        if $is_workspace {
            $crate::Project::Workspace(::std::boxed::Box::new($ws($($arg),*)))
        } else {
            $crate::Project::Package(::std::boxed::Box::new($pkg($($arg),*)))
        }
    };
}

/// Generates `fn language(&self) -> Language` for a fixed language variant.
#[macro_export]
macro_rules! impl_language {
    ($lang:expr) => {
        fn language(&self) -> $crate::Language {
            $lang
        }
    };
}

/// Returns `true` when `path`'s extension matches `ext` case-insensitively
/// (ASCII only).
///
/// Mirrors the `path.extension().and_then(|e| e.to_str()).is_some_and(|e|
/// e.eq_ignore_ascii_case(ext))` idiom used across language crates so the
/// predicate lives in exactly one place. Returns `false` when the path has
/// no extension (including dotfiles such as `.json`, where
/// [`std::path::Path::extension`] returns `None`).
///
/// Public so cross-crate callers (e.g. `changepacks-csharp`,
/// `changepacks-java`, `changepacks-utils`) can reuse it via the re-export
/// from `changepacks_core::lib.rs`.
#[must_use]
pub fn has_extension_ignore_ascii_case(path: &Path, ext: &str) -> bool {
    path.extension()
        .and_then(|e| e.to_str())
        .is_some_and(|e| e.eq_ignore_ascii_case(ext))
}

/// Returns `Ok(true)` when `path` refers to an existing regular file.
///
/// Boolean shorthand over [`regular_file_metadata`], which owns the triage:
/// a missing path or directory returns `Ok(false)`, and other metadata errors
/// are propagated with the failing path in their context.
///
/// Shared between `ProjectFinder::matches_project_file` (name-based match
/// used by every language) and `CSharpProjectFinder::visit` (extension-based
/// match) so the byte-identical stat + `is_file()` fallthrough lives in ONE
/// place. Public so cross-crate callers (e.g. `changepacks-csharp`) can
/// reuse it via the re-export from `changepacks_core::lib.rs`.
///
/// # Errors
/// Propagates any [`regular_file_metadata`] error, i.e. a metadata read that
/// fails for a reason other than the path being absent.
pub async fn is_regular_file(path: &Path) -> Result<bool> {
    Ok(regular_file_metadata(path).await?.is_some())
}

/// Returns `Ok(Some(metadata))` when `path` refers to an existing regular
/// file, and `Ok(None)` when it is missing or is not a regular file.
///
/// This is the single owner of the metadata triage ladder shared by every
/// language crate: `is_file()` decides regular-vs-other,
/// [`std::io::ErrorKind::NotFound`] is normalized to "absent", and any other
/// metadata error is propagated with the failing path in its context.
///
/// [`is_regular_file`] is the boolean shorthand over this function. Callers
/// that additionally need something *from* the same
/// [`std::fs::Metadata`] — e.g. `changepacks-java` reading the Unix
/// permission bits of a `java` candidate — must use this function rather
/// than `is_regular_file` followed by their own `metadata` call, because a
/// second stat is both a wasted syscall and a TOCTOU window.
///
/// AGENTS.md rule: never blocking I/O in async — uses `tokio::fs::metadata`.
///
/// # Errors
/// Returns the underlying `io::Error`, annotated with the failing path, when
/// the metadata read fails for any reason other than
/// [`std::io::ErrorKind::NotFound`].
pub async fn regular_file_metadata(path: &Path) -> Result<Option<std::fs::Metadata>> {
    match tokio::fs::metadata(path).await {
        Ok(metadata) if metadata.is_file() => Ok(Some(metadata)),
        Ok(_) => Ok(None),
        Err(error) if error.kind() == std::io::ErrorKind::NotFound => Ok(None),
        Err(error) => {
            Err(error).with_context(|| format!("Failed to read metadata for {}", path.display()))
        }
    }
}

/// Visitor pattern for discovering projects by walking the git tree.
///
/// Each language implements this trait to detect its project files (package.json, Cargo.toml, etc.)
/// and build a collection of projects. The `visit` method is called for each file in the git tree.
#[async_trait]
pub trait ProjectFinder: std::fmt::Debug + Send + Sync {
    fn projects(&self) -> Vec<&Project>;
    fn projects_mut(&mut self) -> Vec<&mut Project>;
    /// Number of projects held by this finder.
    ///
    /// Required rather than defaulted: a `self.projects().len()` default would
    /// allocate and immediately drop a whole `Vec<&Project>` just to read a
    /// length. Every implementor already owns an O(1), allocation-free count
    /// (the language finders get one from
    /// [`impl_projects_hashmap_accessors!`]), so the trait demands it instead
    /// of offering a lossy shortcut.
    fn project_count(&self) -> usize;
    /// Append every project held by this finder onto `out`.
    ///
    /// Exists for the same reason [`ProjectFinder::project_count`] does:
    /// callers that merge several finders into one buffer (the CLI's
    /// `collect_projects`) would otherwise pay one throwaway `Vec<&Project>`
    /// per finder — allocated by `projects()` and dropped one line later —
    /// on every `check`, `update`, `publish`, and default-changepack run.
    /// Pushing into the caller's buffer removes that per-finder allocation.
    ///
    /// The default body is the compatibility path for external implementors:
    /// it forwards to `projects()`, so an implementor that only supplies the
    /// required accessors keeps compiling and keeps identical behaviour, and
    /// merely forfeits the allocation elision. Implementors backed by a
    /// `HashMap<PathBuf, Project>` get the elided override for free from
    /// [`impl_projects_hashmap_accessors!`].
    ///
    /// Contract for overrides: append in exactly `projects()` order and never
    /// clear or reorder what `out` already holds — callers rely on the merged
    /// order for their output (e.g. `changepacks check`).
    fn extend_projects<'a>(&'a self, out: &mut Vec<&'a Project>) {
        out.extend(self.projects());
    }
    /// Mutable counterpart of [`ProjectFinder::extend_projects`]: append every
    /// project held by this finder onto `out` as `&mut Project`.
    ///
    /// Exists for the same allocation reason: `find_project_dirs`'s no-name
    /// fallback merges every finder's projects into one buffer, and
    /// `flat_map(|f| f.projects_mut())` paid one throwaway `Vec<&mut Project>`
    /// per finder — six of them on every CLI run — allocated by `projects_mut()`
    /// and dropped as soon as the flattening consumed it. Pushing into the
    /// caller's buffer removes that per-finder allocation.
    ///
    /// The default body is the compatibility path for external implementors:
    /// it forwards to `projects_mut()`, so an implementor that only supplies
    /// the required accessors keeps compiling and keeps identical behaviour,
    /// and merely forfeits the allocation elision. Implementors backed by a
    /// `HashMap<PathBuf, Project>` get the elided override for free from
    /// [`impl_projects_hashmap_accessors!`].
    ///
    /// Contract for overrides: append in exactly `projects_mut()` order and
    /// never clear or reorder what `out` already holds.
    fn extend_projects_mut<'a>(&'a mut self, out: &mut Vec<&'a mut Project>) {
        out.extend(self.projects_mut());
    }
    /// Whether a project keyed by exactly `path` has already been discovered
    /// by this finder.
    ///
    /// `path` is the manifest path a `visit()` call was handed — the same
    /// value every finder uses as its storage key — so this is the
    /// "already visited, do not re-parse" probe. It exists so the six
    /// language finders stop reaching into their private `projects` field
    /// from inside `visit()`; the gate now belongs to the trait that defines
    /// the visit protocol.
    ///
    /// The default body is the compatibility path for external implementors,
    /// exactly like [`ProjectFinder::extend_projects`]: it linearly scans
    /// `projects()` for a project whose [`Project::path`] equals `path`, so
    /// an implementor that only supplies the required accessors keeps
    /// compiling and keeps identical behaviour, and merely forfeits the
    /// hashed lookup. Implementors backed by a `HashMap<PathBuf, Project>`
    /// get the O(1) override for free from
    /// [`impl_projects_hashmap_accessors!`].
    fn contains_project(&self, path: &Path) -> bool {
        self.projects()
            .into_iter()
            .any(|project| project.path() == path)
    }
    /// Whether `visit()` should parse `path`, or bail out early.
    ///
    /// This is the two-guard prelude every language finder open-coded at the
    /// top of its `visit()`: `path` must be a manifest this finder claims AND
    /// must not have been discovered already.
    ///
    /// Guard order is deliberate and preserved from the hand-rolled copies:
    /// the name/stat gate ([`ProjectFinder::matches_project_file`]) runs
    /// FIRST and the map probe ([`ProjectFinder::contains_project`]) second.
    /// `&&` short-circuits, so a path that is not a manifest never pays for
    /// the second probe — and, more importantly, the ordering keeps the
    /// error surface unchanged: a metadata error on a recognized manifest
    /// name still propagates even when that path is already known.
    ///
    /// Only finders whose [`ProjectFinder::project_files`] uses the bare
    /// file-name form may gate on this. An extension-based finder (the
    /// `".csproj"` form) must keep its own
    /// [`has_extension_ignore_ascii_case`] + [`is_regular_file`] check and
    /// call [`ProjectFinder::contains_project`] directly, for the reason
    /// spelled out on [`ProjectFinder::matches_project_file`].
    ///
    /// # Errors
    /// Propagates whatever [`ProjectFinder::matches_project_file`] returns.
    ///
    /// Written in the boxed-future shape `#[async_trait]` would produce rather
    /// than as a defaulted `async fn`, for the same reason
    /// [`impl_publish_flows!`] is: the spans `#[async_trait]` puts on a
    /// *defaulted* body are not attributed back to this file by `llvm-cov`, so
    /// such a default reads as permanently unexecuted in coverage even while
    /// tests drive every branch of it. This shape is object safe, stays
    /// overridable by the `#[async_trait]` impls in the language crates, and is
    /// measurable. The three defaults below follow the same rule.
    fn should_visit_manifest<'life0, 'life1, 'async_trait>(
        &'life0 self,
        path: &'life1 Path,
    ) -> ::core::pin::Pin<
        ::std::boxed::Box<
            dyn ::core::future::Future<Output = Result<bool>> + ::core::marker::Send + 'async_trait,
        >,
    >
    where
        'life0: 'async_trait,
        'life1: 'async_trait,
        Self: ::core::marker::Sync + 'async_trait,
    {
        ::std::boxed::Box::pin(async move {
            Ok(self.matches_project_file(path).await? && !self.contains_project(path))
        })
    }
    /// The manifest patterns this finder claims, in one of TWO forms.
    ///
    /// 1. A bare **file name** (`"package.json"`, `"Cargo.toml"`,
    ///    `"pyproject.toml"`, `"pubspec.yaml"`, `"build.gradle.kts"`). Five of
    ///    the six language finders use only this form.
    /// 2. A leading-dot **extension** (`".csproj"`), used when the manifest
    ///    name varies per project. `CSharpProjectFinder` is the only
    ///    in-tree implementor of this form.
    ///
    /// Only the discovery walk understands both: `find_project_dirs`'s
    /// `project_files_can_visit_path` (in `changepacks-utils`) first compares
    /// the entry to `path.file_name()` and, on a miss, retries any entry that
    /// starts with `.` as a case-insensitive extension match.
    ///
    /// The defaulted [`ProjectFinder::matches_project_file`] deliberately
    /// implements ONLY form 1 — it compares `path.file_name()` against this
    /// list — so an extension entry such as `".csproj"` can never match there
    /// (`Path::new("App.csproj").file_name()` is `"App.csproj"`, never
    /// `".csproj"`). An implementor that returns extension entries therefore
    /// MUST NOT gate `visit()` on `matches_project_file`; it must do its own
    /// extension check with [`has_extension_ignore_ascii_case`] plus
    /// [`is_regular_file`], exactly as `CSharpProjectFinder::visit` does.
    ///
    /// The two forms also differ in case sensitivity: a file-name entry is
    /// compared byte-for-byte, while an extension entry matches
    /// case-insensitively (`App.CSPROJ` is accepted).
    fn project_files(&self) -> &[&str];
    /// # Errors
    /// Returns error if the file visitation fails.
    async fn visit(&mut self, path: &Path, relative_path: &Path) -> Result<()>;
    /// Whether `path` is a project manifest file recognized by this finder.
    ///
    /// Returns `false` for directories and files whose name is not in
    /// `project_files()`. Used by language-specific `visit()` implementations
    /// to gate manifest parsing on file-name matching.
    ///
    /// This gate implements ONLY the bare-file-name form of
    /// [`ProjectFinder::project_files`]. A leading-dot extension entry such as
    /// `".csproj"` can never match here, because `path.file_name()` yields the
    /// whole name (`"App.csproj"`) and never the extension alone. That is why
    /// `CSharpProjectFinder` gates `visit()` on
    /// [`has_extension_ignore_ascii_case`] + [`is_regular_file`] instead of
    /// calling this method — any future extension-based finder must do the
    /// same.
    ///
    /// Check order is name-first, stat-last: on a monorepo with N tracked
    /// files where only K match any recognized manifest name (typically
    /// K ≪ N), the previous stat-then-name shape issued 5 × N async
    /// `tokio::fs::metadata` syscalls across every non-CSharp language
    /// finder invocation. Reversing the order collapses that to ~K stats
    /// total. Missing/non-UTF-8 file names cannot possibly match ASCII
    /// manifest names anyway, so returning `Ok(false)` early is
    /// semantically identical to the previous `with_context` error paths
    /// — which were unreachable for git-index-derived paths.
    ///
    /// # Errors
    /// Returns an error when metadata for a recognized manifest path cannot be
    /// read for a reason other than the path not existing.
    fn matches_project_file<'life0, 'life1, 'async_trait>(
        &'life0 self,
        path: &'life1 Path,
    ) -> ::core::pin::Pin<
        ::std::boxed::Box<
            dyn ::core::future::Future<Output = Result<bool>> + ::core::marker::Send + 'async_trait,
        >,
    >
    where
        'life0: 'async_trait,
        'life1: 'async_trait,
        Self: ::core::marker::Sync + 'async_trait,
    {
        ::std::boxed::Box::pin(async move {
            let Some(name_os) = path.file_name() else {
                return Ok(false);
            };
            let Some(name) = name_os.to_str() else {
                return Ok(false);
            };
            if !self.project_files().contains(&name) {
                return Ok(false);
            }
            is_regular_file(path).await
        })
    }
    /// Mark every project against every path in `paths` from ONE
    /// `projects_mut()` call.
    ///
    /// The driver dispatches every changed file to every finder. Rebuilding
    /// the `Vec<&mut Project>` via `projects_mut()` once per file would cost
    /// `F` changed files × `M` finders fresh Vec allocations. Collecting the
    /// paths once and looping project-major here collapses that to one Vec per
    /// finder (`M` total).
    ///
    /// The project-major / path-major order flip is behavior-preserving:
    /// [`Project::check_changed`] is monotonic — it early-returns once the
    /// project is already changed and only ever sets `changed = true` via the
    /// pure, stateless `should_mark_changed`. A project ends up changed iff
    /// *any* path matches, an order-independent logical OR, so visiting all
    /// paths for one project before moving to the next yields an identical
    /// result to a path-major traversal.
    ///
    /// # Errors
    /// Returns error if checking changed status fails for any project.
    fn check_changed_many(&mut self, paths: &[PathBuf]) -> Result<()> {
        for project in self.projects_mut() {
            for path in paths {
                project.check_changed(path)?;
                // Early break: check_changed is monotonic, so once changed, remaining paths are redundant.
                if project.is_changed() {
                    break;
                }
            }
        }
        Ok(())
    }
    /// Post-visit processing hook for resolving deferred state (e.g., workspace-inherited versions).
    /// Called once after all `visit()` calls complete.
    /// # Errors
    /// Returns error if finalization fails.
    fn finalize<'life0, 'async_trait>(
        &'life0 mut self,
    ) -> ::core::pin::Pin<
        ::std::boxed::Box<
            dyn ::core::future::Future<Output = Result<()>> + ::core::marker::Send + 'async_trait,
        >,
    >
    where
        'life0: 'async_trait,
        Self: ::core::marker::Send + 'async_trait,
    {
        ::std::boxed::Box::pin(async move { Ok(()) })
    }
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::test_support::{MockPackage, MockWorkspace};
    use crate::{Package, Workspace};
    use async_trait::async_trait;
    use rstest::rstest;
    use std::path::PathBuf;

    // `Path::new(".json").extension()` returns `None` in Rust — dotfiles have
    // no extension — so `has_extension_ignore_ascii_case(Path::new(".json"), "json")`
    // is `false`. This matches the behaviour of every call site that wraps a
    // bare filename with `Path::new(file_name)`.
    #[rstest]
    #[case("foo.json", "json", true)]
    #[case("foo.JSON", "json", true)]
    #[case("foo.Json", "json", true)]
    #[case("foo", "json", false)]
    #[case(".json", "json", false)]
    #[case("foo.jsonx", "json", false)]
    fn test_has_extension_ignore_ascii_case(
        #[case] file: &str,
        #[case] ext: &str,
        #[case] expected: bool,
    ) {
        assert_eq!(
            has_extension_ignore_ascii_case(Path::new(file), ext),
            expected,
            "has_extension_ignore_ascii_case(Path::new({file:?}), {ext:?})"
        );
    }

    // `add_dependency` now probes `contains` before allocating. These cases
    // lock the observable contract the probe must not change: a repeated name
    // is still stored exactly once, and distinct names all still land.
    #[test]
    fn test_add_dependency_deduplicates_repeated_names() {
        let mut package = MockPackage::same_path("pkg", "/project/package.json");

        // The duplicate path: the same name arrives from two manifest sections.
        package.add_dependency("serde");
        package.add_dependency("serde");
        package.add_dependency("serde");

        assert_eq!(
            package.dependencies().len(),
            1,
            "repeated add_dependency must keep exactly one entry"
        );
        assert!(package.dependencies().contains("serde"));

        // Distinct names still insert normally (the miss path is unchanged).
        package.add_dependency("tokio");
        package.add_dependency("anyhow");
        package.add_dependency("tokio");

        let mut names = package.dependencies().iter().cloned().collect::<Vec<_>>();
        names.sort();
        assert_eq!(names, vec!["anyhow", "serde", "tokio"]);
    }

    #[test]
    fn test_add_dependency_deduplicates_on_workspace_too() {
        // The macro backs both the Package and the Workspace impls of all six
        // language crates, so pin the behaviour at the Workspace surface as well.
        let mut workspace = MockWorkspace::same_path("root", "/project/package.json");

        workspace.add_dependency("left-pad");
        workspace.add_dependency("left-pad");

        assert_eq!(workspace.dependencies().len(), 1);
        assert!(workspace.dependencies().contains("left-pad"));
    }

    #[derive(Debug)]
    struct MockProjectFinder {
        projects: Vec<Project>,
    }

    impl MockProjectFinder {
        fn new() -> Self {
            Self { projects: vec![] }
        }

        fn with_package(mut self, package: MockPackage) -> Self {
            self.projects.push(Project::Package(Box::new(package)));
            self
        }

        fn with_workspace(mut self, workspace: MockWorkspace) -> Self {
            self.projects.push(Project::Workspace(Box::new(workspace)));
            self
        }
    }

    #[async_trait]
    impl ProjectFinder for MockProjectFinder {
        fn projects(&self) -> Vec<&Project> {
            self.projects.iter().collect()
        }

        fn projects_mut(&mut self) -> Vec<&mut Project> {
            self.projects.iter_mut().collect()
        }

        fn project_count(&self) -> usize {
            self.projects.len()
        }

        fn project_files(&self) -> &[&str] {
            &["package.json"]
        }

        async fn visit(&mut self, _path: &Path, _relative_path: &Path) -> Result<()> {
            Ok(())
        }
    }

    /// HashMap-backed finder that takes its accessors from
    /// [`impl_projects_hashmap_accessors!`], so the macro's `extend_projects`
    /// override is exercised inside `core` (the six language finders use the
    /// exact same expansion).
    #[derive(Debug)]
    struct HashMapProjectFinder {
        projects: std::collections::HashMap<PathBuf, Project>,
    }

    impl HashMapProjectFinder {
        fn with_packages(names: &[(&str, &str)]) -> Self {
            let mut projects = std::collections::HashMap::new();
            for (name, path) in names {
                projects.insert(
                    PathBuf::from(*path),
                    Project::Package(Box::new(MockPackage::same_path(name, path))),
                );
            }
            Self { projects }
        }
    }

    #[async_trait]
    impl ProjectFinder for HashMapProjectFinder {
        crate::impl_projects_hashmap_accessors!();

        fn project_files(&self) -> &[&str] {
            &["package.json"]
        }

        async fn visit(&mut self, _path: &Path, _relative_path: &Path) -> Result<()> {
            Ok(())
        }
    }

    // Accepts both `&[&Project]` and `&[&mut Project]` buffers: std provides
    // `Borrow<T>` for `&T` and `&mut T` alike, so the shared/mutable
    // `extend_projects` twins can assert against one helper.
    fn project_names<P: std::borrow::Borrow<Project>>(projects: &[P]) -> Vec<String> {
        projects
            .iter()
            .map(|project| project.borrow().name().unwrap_or_default().to_string())
            .collect()
    }

    // The defaulted `extend_projects` body is the compatibility path for
    // external implementors: `MockProjectFinder` does NOT override it, so this
    // pins that the default appends exactly `projects()`, in `projects()`
    // order, without disturbing what the buffer already holds.
    #[test]
    fn test_extend_projects_default_matches_projects_and_preserves_buffer() {
        let finder = MockProjectFinder::new()
            .with_package(MockPackage::same_path("pkg1", "/project1/package.json"))
            .with_workspace(MockWorkspace::same_path("root", "/project2/package.json"))
            .with_package(MockPackage::same_path("pkg2", "/project3/package.json"));

        let seed = MockProjectFinder::new()
            .with_package(MockPackage::same_path("seed", "/seed/package.json"));
        let mut out = seed.projects();
        finder.extend_projects(&mut out);

        let mut expected = vec!["seed".to_string()];
        expected.extend(project_names(&finder.projects()));
        assert_eq!(project_names(&out), expected);
    }

    #[test]
    fn test_extend_projects_default_on_empty_finder_is_a_no_op() {
        let finder = MockProjectFinder::new();
        let mut out: Vec<&Project> = Vec::new();
        finder.extend_projects(&mut out);
        assert!(out.is_empty());
    }

    // Mutable twin of the test above: `MockProjectFinder` does NOT override
    // `extend_projects_mut`, so this pins that the default appends exactly
    // `projects_mut()`, in `projects_mut()` order, without disturbing what the
    // buffer already holds — the contract `find_project_dirs`'s no-name
    // fallback relies on when it merges every finder into one buffer.
    #[test]
    fn test_extend_projects_mut_default_matches_projects_and_preserves_buffer() {
        let mut finder = MockProjectFinder::new()
            .with_package(MockPackage::same_path("pkg1", "/project1/package.json"))
            .with_workspace(MockWorkspace::same_path("root", "/project2/package.json"))
            .with_package(MockPackage::same_path("pkg2", "/project3/package.json"));

        let mut seed = MockProjectFinder::new()
            .with_package(MockPackage::same_path("seed", "/seed/package.json"));
        let mut out = seed.projects_mut();
        finder.extend_projects_mut(&mut out);
        let names = project_names(&out);
        drop(out);

        let mut expected = vec!["seed".to_string()];
        expected.extend(project_names(&finder.projects()));
        assert_eq!(names, expected);
    }

    #[test]
    fn test_extend_projects_mut_default_on_empty_finder_is_a_no_op() {
        let mut finder = MockProjectFinder::new();
        let mut out: Vec<&mut Project> = Vec::new();
        finder.extend_projects_mut(&mut out);
        assert!(out.is_empty());
    }

    // The macro override skips the intermediate Vec that `projects()` builds;
    // both must still yield the same projects in the same `HashMap::values()`
    // order, so a caller can swap one for the other without reordering output.
    #[test]
    fn test_extend_projects_macro_override_matches_projects_order() {
        let finder = HashMapProjectFinder::with_packages(&[
            ("pkg1", "/project1/package.json"),
            ("pkg2", "/project2/package.json"),
            ("pkg3", "/project3/package.json"),
        ]);

        let mut out: Vec<&Project> = Vec::new();
        finder.extend_projects(&mut out);

        assert_eq!(out.len(), finder.project_count());
        assert_eq!(project_names(&out), project_names(&finder.projects()));
    }

    // Same equivalence for the mutable override: `HashMap::values_mut()` and
    // `HashMap::values()` walk one unmodified map in the same order, so the
    // elided body must yield exactly what `projects_mut()` would have.
    #[test]
    fn test_extend_projects_mut_macro_override_matches_projects_order() {
        let mut finder = HashMapProjectFinder::with_packages(&[
            ("pkg1", "/project1/package.json"),
            ("pkg2", "/project2/package.json"),
            ("pkg3", "/project3/package.json"),
        ]);

        let mut out: Vec<&mut Project> = Vec::new();
        finder.extend_projects_mut(&mut out);
        let out_len = out.len();
        let names = project_names(&out);
        drop(out);

        assert_eq!(out_len, finder.project_count());
        assert_eq!(names, project_names(&finder.projects_mut()));
    }

    // The borrows handed out must alias the finder's own storage: mutating a
    // project through the merged buffer has to be visible on the finder
    // afterwards, which is exactly what the no-name `set_name` fallback does.
    #[test]
    fn test_extend_projects_mut_yields_borrows_that_mutate_the_finder() {
        let mut finder = HashMapProjectFinder::with_packages(&[("pkg1", "/project1/package.json")]);

        let mut out: Vec<&mut Project> = Vec::new();
        finder.extend_projects_mut(&mut out);
        for project in &mut out {
            project.set_name("renamed".to_string());
        }
        drop(out);

        assert_eq!(project_names(&finder.projects()), vec!["renamed"]);
    }

    // `contains_project` has the same two-body shape as `extend_projects`: a
    // defaulted linear scan for external implementors and a hashed override
    // from the macro. `MockProjectFinder` does NOT override it, so this pins
    // the compatibility path — hit on the exact stored manifest path, miss on
    // an unknown one and on a merely-similar one.
    #[test]
    fn test_contains_project_default_scans_projects_by_path() {
        let finder = MockProjectFinder::new()
            .with_package(MockPackage::same_path("pkg1", "/project1/package.json"))
            .with_workspace(MockWorkspace::same_path("root", "/project2/package.json"));

        assert!(finder.contains_project(Path::new("/project1/package.json")));
        // Workspace projects count too, not just packages.
        assert!(finder.contains_project(Path::new("/project2/package.json")));
        assert!(!finder.contains_project(Path::new("/project3/package.json")));
        // The probe keys on the whole manifest path, never on the directory.
        assert!(!finder.contains_project(Path::new("/project1")));
    }

    #[test]
    fn test_contains_project_default_on_empty_finder_is_always_false() {
        let finder = MockProjectFinder::new();
        assert!(!finder.contains_project(Path::new("/project1/package.json")));
    }

    // The macro override must answer identically to the default for every
    // probe — that equivalence is what lets the six language finders swap
    // their open-coded `self.projects.contains_key(path)` for the trait
    // method without changing behaviour.
    #[test]
    fn test_contains_project_macro_override_matches_default_answers() {
        let entries = [
            ("pkg1", "/project1/package.json"),
            ("pkg2", "/project2/package.json"),
        ];
        let hashed = HashMapProjectFinder::with_packages(&entries);
        let mut scanned = MockProjectFinder::new();
        for (name, path) in entries {
            scanned = scanned.with_package(MockPackage::same_path(name, path));
        }

        for probe in [
            "/project1/package.json",
            "/project2/package.json",
            "/project3/package.json",
            "/project1",
            "",
        ] {
            assert_eq!(
                hashed.contains_project(Path::new(probe)),
                scanned.contains_project(Path::new(probe)),
                "hashed and scanned answers diverged for {probe:?}"
            );
        }
    }

    // `should_visit_manifest` is the consolidated two-guard prelude. Its
    // documented order is name/stat gate FIRST, already-discovered probe
    // SECOND, and it returns `true` only when both agree the manifest is new.
    #[tokio::test]
    async fn test_should_visit_manifest_accepts_new_recognized_manifest() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let manifest = temp_dir.path().join("package.json");
        std::fs::write(&manifest, "{}").unwrap();

        let finder = MockProjectFinder::new();
        assert!(finder.should_visit_manifest(&manifest).await.unwrap());
    }

    // The duplicate-visit half: same manifest, but already discovered.
    #[tokio::test]
    async fn test_should_visit_manifest_rejects_already_discovered_manifest() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let manifest = temp_dir.path().join("package.json");
        std::fs::write(&manifest, "{}").unwrap();

        let finder = MockProjectFinder::new()
            .with_package(MockPackage::same_path("pkg", manifest.to_str().unwrap()));
        assert!(
            !finder.should_visit_manifest(&manifest).await.unwrap(),
            "a manifest already in the finder must not be visited twice"
        );
    }

    // The name/stat half: an unrecognized name and a directory that merely
    // shares a manifest name are both rejected before anything is parsed.
    #[tokio::test]
    async fn test_should_visit_manifest_rejects_non_manifest_and_directory() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let other = temp_dir.path().join("Cargo.toml");
        std::fs::write(&other, "[package]\n").unwrap();
        let dir_path = temp_dir.path().join("package.json");
        std::fs::create_dir(&dir_path).unwrap();

        let finder = MockProjectFinder::new();
        assert!(!finder.should_visit_manifest(&other).await.unwrap());
        assert!(!finder.should_visit_manifest(&dir_path).await.unwrap());
    }

    // Equivalence with the hand-rolled prelude the language finders used to
    // open-code: `matches_project_file(path)? && !contains_project(path)`.
    #[tokio::test]
    async fn test_should_visit_manifest_equals_the_open_coded_two_guard_prelude() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let manifest = temp_dir.path().join("package.json");
        std::fs::write(&manifest, "{}").unwrap();
        let unrecognized = temp_dir.path().join("Cargo.toml");
        std::fs::write(&unrecognized, "[package]\n").unwrap();
        let missing = temp_dir.path().join("nested").join("package.json");

        let known = MockProjectFinder::new()
            .with_package(MockPackage::same_path("pkg", manifest.to_str().unwrap()));
        let empty = MockProjectFinder::new();

        for finder in [&known, &empty] {
            for probe in [&manifest, &unrecognized, &missing] {
                let open_coded = finder.matches_project_file(probe).await.unwrap()
                    && !finder.contains_project(probe);
                assert_eq!(
                    finder.should_visit_manifest(probe).await.unwrap(),
                    open_coded,
                    "consolidated gate diverged from the open-coded prelude for {}",
                    probe.display()
                );
            }
        }
    }

    #[test]
    fn test_extend_projects_macro_override_preserves_existing_buffer_contents() {
        let first = HashMapProjectFinder::with_packages(&[("pkg1", "/project1/package.json")]);
        let second = HashMapProjectFinder::with_packages(&[("pkg2", "/project2/package.json")]);

        // Mirrors the CLI's `collect_projects`: one buffer, several finders.
        let mut out: Vec<&Project> = Vec::new();
        first.extend_projects(&mut out);
        second.extend_projects(&mut out);

        assert_eq!(project_names(&out), vec!["pkg1", "pkg2"]);
    }

    #[test]
    fn test_extend_projects_mut_macro_override_preserves_existing_buffer_contents() {
        let mut first = HashMapProjectFinder::with_packages(&[("pkg1", "/project1/package.json")]);
        let mut second = HashMapProjectFinder::with_packages(&[("pkg2", "/project2/package.json")]);

        // Mirrors `find_project_dirs`'s no-name fallback: one buffer, several finders.
        let mut out: Vec<&mut Project> = Vec::new();
        first.extend_projects_mut(&mut out);
        second.extend_projects_mut(&mut out);

        assert_eq!(project_names(&out), vec!["pkg1", "pkg2"]);
    }

    #[test]
    fn test_project_finder_check_changed() {
        let package = MockPackage::same_path("test", "/project/package.json");
        let mut finder = MockProjectFinder::new().with_package(package);

        // Check a file that's in the project directory
        finder
            .check_changed_many(&[PathBuf::from("/project/src/index.js")])
            .unwrap();

        // The project should be marked as changed
        assert!(finder.projects()[0].is_changed());
    }

    #[test]
    fn test_project_finder_check_changed_multiple_projects() {
        let package1 = MockPackage::same_path("pkg1", "/project1/package.json");
        let package2 = MockPackage::same_path("pkg2", "/project2/package.json");
        let mut finder = MockProjectFinder::new()
            .with_package(package1)
            .with_package(package2);

        // Check a file in project1 only
        finder
            .check_changed_many(&[PathBuf::from("/project1/src/index.js")])
            .unwrap();

        // Only project1 should be changed
        assert!(finder.projects()[0].is_changed());
        assert!(!finder.projects()[1].is_changed());
    }

    #[test]
    fn test_project_finder_check_changed_many() {
        let package1 = MockPackage::same_path("pkg1", "/project1/package.json");
        let package2 = MockPackage::same_path("pkg2", "/project2/package.json");
        let workspace = MockWorkspace::same_path("root", "/project3/package.json");
        let mut finder = MockProjectFinder::new()
            .with_package(package1)
            .with_package(package2)
            .with_workspace(workspace);

        // One batch: a file under project1 and a file under project3 (the
        // workspace); nothing under project2. `check_changed_many` must mark
        // exactly project1 and project3 — a project is marked changed iff any
        // path matches it, proving the project-major loop order is
        // behavior-preserving across both Package and Workspace variants.
        let paths = [
            PathBuf::from("/project1/src/index.js"),
            PathBuf::from("/project3/lib/mod.rs"),
        ];
        finder.check_changed_many(&paths).unwrap();

        assert!(finder.projects()[0].is_changed());
        assert!(!finder.projects()[1].is_changed());
        assert!(finder.projects()[2].is_changed());
    }

    #[test]
    fn test_project_finder_check_changed_many_matches_per_file_traversal() {
        // The same inputs fed one-at-a-time (each path its own single-element
        // batch, mirroring a per-file traversal) and fed together in ONE batch
        // must land the two finders in an identical changed-state, locking the
        // order/batch equivalence the driver relies on.
        let paths = [
            PathBuf::from("/project1/src/index.js"),
            PathBuf::from("/project2/README.md"),
        ];

        let mut per_path = MockProjectFinder::new()
            .with_package(MockPackage::same_path("pkg1", "/project1/package.json"))
            .with_package(MockPackage::same_path("pkg2", "/project2/package.json"));
        for path in &paths {
            per_path
                .check_changed_many(std::slice::from_ref(path))
                .unwrap();
        }

        let mut batched = MockProjectFinder::new()
            .with_package(MockPackage::same_path("pkg1", "/project1/package.json"))
            .with_package(MockPackage::same_path("pkg2", "/project2/package.json"));
        batched.check_changed_many(&paths).unwrap();

        assert_eq!(
            per_path.projects()[0].is_changed(),
            batched.projects()[0].is_changed()
        );
        assert_eq!(
            per_path.projects()[1].is_changed(),
            batched.projects()[1].is_changed()
        );
        assert!(batched.projects()[0].is_changed());
        assert!(batched.projects()[1].is_changed());
    }

    #[test]
    fn test_project_finder_with_workspace() {
        let workspace = MockWorkspace::same_path("root", "/project/package.json");
        let mut finder = MockProjectFinder::new().with_workspace(workspace);

        finder
            .check_changed_many(&[PathBuf::from("/project/src/index.js")])
            .unwrap();

        assert!(finder.projects()[0].is_changed());
    }

    // Every other `check_changed_many` test ends in `unwrap`, so the `?` on
    // `project.check_changed(path)` inside the defaulted body was never
    // exercised on its failing branch. The only way that call can fail is
    // `should_mark_changed` finding no parent directory for the project
    // manifest, so a project rooted at `/` — whose `Path::parent()` is `None`
    // on both Windows and Unix — reaches it. The batch loop must surface that
    // error instead of swallowing it and reporting the project as unchanged.
    #[test]
    fn test_project_finder_check_changed_many_propagates_check_changed_error() {
        let mut finder = MockProjectFinder::new().with_package(MockPackage::same_path("root", "/"));

        let error = finder
            .check_changed_many(&[PathBuf::from("/src/index.js")])
            .expect_err("a manifest without a parent directory must fail the batch");

        let chain = format!("{error:#}");
        assert!(
            chain.contains("Parent not found"),
            "error chain should carry the missing-parent context, got: {chain}"
        );
        assert!(
            !finder.projects()[0].is_changed(),
            "a project whose check_changed failed must not be reported as changed"
        );
    }

    #[test]
    fn project_finder_entry_points_are_included_in_coverage() {
        assert!(
            !include_str!("project_finder.rs")
                .contains(concat!("#[cfg(not(", "tarpaulin_include))]"))
        );
    }

    #[tokio::test]
    async fn test_default_project_finder_finalize_is_covered_no_op() {
        let mut finder = MockProjectFinder::new();
        let result = finder.finalize().await;
        assert!(result.is_ok());
    }

    #[tokio::test]
    async fn test_is_regular_file_with_existing_file() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let file_path = temp_dir.path().join("test.txt");
        std::fs::write(&file_path, "test content").unwrap();

        let result = is_regular_file(&file_path).await;
        assert!(result.unwrap());
    }

    #[tokio::test]
    async fn test_is_regular_file_with_directory() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let dir_path = temp_dir.path().join("subdir");
        std::fs::create_dir(&dir_path).unwrap();

        let result = is_regular_file(&dir_path).await;
        assert!(!result.unwrap());
    }

    #[tokio::test]
    async fn test_is_regular_file_with_missing_path() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let missing_path = temp_dir.path().join("nonexistent.txt");

        let result = is_regular_file(&missing_path).await;
        assert!(!result.unwrap());
    }

    #[tokio::test]
    async fn test_is_regular_file_propagates_metadata_error_with_path_context() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        #[cfg(windows)]
        let invalid_path = temp_dir.path().join("invalid\0path");
        #[cfg(unix)]
        let invalid_path = {
            use std::os::unix::fs::symlink;

            let path = temp_dir.path().join("metadata-loop");
            symlink(&path, &path).unwrap();
            path
        };

        let error = is_regular_file(&invalid_path)
            .await
            .expect_err("metadata errors other than NotFound must be propagated");
        let chain = format!("{error:#}");
        assert!(
            chain.contains(&invalid_path.display().to_string()),
            "error chain should name the path whose metadata failed, got: {chain}"
        );
    }

    // `regular_file_metadata` is the ladder `is_regular_file` and the Java
    // executable probe both sit on, and it is the only one of the two that
    // hands the caller the stat'ed `Metadata`. These cases pin that extra
    // guarantee: the returned metadata must describe the file itself, and the
    // non-file exits must stay indistinguishable `None`s.
    #[tokio::test]
    async fn test_regular_file_metadata_returns_metadata_for_existing_file() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let file_path = temp_dir.path().join("test.txt");
        std::fs::write(&file_path, "test content").unwrap();

        let metadata = regular_file_metadata(&file_path)
            .await
            .unwrap()
            .expect("an existing regular file must yield its metadata");
        assert!(metadata.is_file());
        assert_eq!(metadata.len(), "test content".len() as u64);
    }

    #[tokio::test]
    async fn test_regular_file_metadata_returns_none_for_directory_and_missing_path() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let dir_path = temp_dir.path().join("subdir");
        std::fs::create_dir(&dir_path).unwrap();

        assert!(regular_file_metadata(&dir_path).await.unwrap().is_none());
        assert!(
            regular_file_metadata(&temp_dir.path().join("nonexistent.txt"))
                .await
                .unwrap()
                .is_none()
        );
    }

    #[tokio::test]
    async fn test_regular_file_metadata_propagates_error_with_path_context() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        #[cfg(windows)]
        let invalid_path = temp_dir.path().join("invalid\0path");
        #[cfg(unix)]
        let invalid_path = {
            use std::os::unix::fs::symlink;

            let path = temp_dir.path().join("metadata-loop");
            symlink(&path, &path).unwrap();
            path
        };

        let error = regular_file_metadata(&invalid_path)
            .await
            .expect_err("metadata errors other than NotFound must be propagated");
        let chain = format!("{error:#}");
        assert!(
            chain.contains(&format!(
                "Failed to read metadata for {}",
                invalid_path.display()
            )),
            "error chain should carry the shared metadata context, got: {chain}"
        );
    }

    // `matches_project_file` is the defaulted gate every non-CSharp language
    // finder calls before parsing a manifest. `MockProjectFinder::project_files`
    // returns exactly `["package.json"]`, so these cases pin all four exits of
    // its documented name-first / stat-last order.

    // Exit 4 (the only `true`): recognized name AND a real regular file.
    #[tokio::test]
    async fn test_matches_project_file_accepts_recognized_regular_file() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let manifest = temp_dir.path().join("package.json");
        std::fs::write(&manifest, "{}").unwrap();

        let finder = MockProjectFinder::new();
        assert!(
            finder.matches_project_file(&manifest).await.unwrap(),
            "a real file named package.json must be recognized"
        );
    }

    // Exit 4 again, negative half: the name matches but the entry is a
    // DIRECTORY, so the stat must veto it. This is why the stat cannot simply
    // be dropped once the name check is in place.
    #[tokio::test]
    async fn test_matches_project_file_rejects_directory_with_recognized_name() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let dir_path = temp_dir.path().join("package.json");
        std::fs::create_dir(&dir_path).unwrap();

        let finder = MockProjectFinder::new();
        assert!(
            !finder.matches_project_file(&dir_path).await.unwrap(),
            "a directory named package.json must not be treated as a manifest"
        );
    }

    // Exit 3: an unrecognized name is rejected even though the file really
    // exists — the name guard, not the stat, is what filters it out.
    #[tokio::test]
    async fn test_matches_project_file_rejects_unrecognized_name() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let other = temp_dir.path().join("Cargo.toml");
        std::fs::write(&other, "[package]\n").unwrap();

        let finder = MockProjectFinder::new();
        assert!(
            !finder.matches_project_file(&other).await.unwrap(),
            "Cargo.toml is not in this finder's project_files()"
        );
    }

    // Exit 1: `file_name()` is `None` for a path ending in `..`, even though
    // that path resolves to an existing directory. The early return must fire
    // before any stat.
    #[tokio::test]
    async fn test_matches_project_file_rejects_path_without_file_name() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let parent_ref = temp_dir.path().join("..");
        assert!(parent_ref.file_name().is_none());

        let finder = MockProjectFinder::new();
        assert!(
            !finder.matches_project_file(&parent_ref).await.unwrap(),
            "a path with no file name cannot match a manifest name"
        );
    }

    // Exit 2: `to_str()` is `None` for a non-UTF-8 file name. Such a name
    // cannot equal any ASCII manifest name, so the guard must short-circuit to
    // `Ok(false)` before any stat — exactly what the method doc reasons about.
    // The path deliberately does not exist: the early return fires first.
    #[tokio::test]
    async fn test_matches_project_file_rejects_non_utf8_file_name() {
        #[cfg(unix)]
        let name: std::ffi::OsString = {
            use std::os::unix::ffi::OsStrExt;
            std::ffi::OsStr::from_bytes(b"\xFF\xFEpackage.json").to_os_string()
        };
        #[cfg(windows)]
        let name: std::ffi::OsString = {
            use std::os::windows::ffi::OsStringExt;
            // Unpaired high surrogate — unrepresentable in UTF-8.
            std::ffi::OsString::from_wide(&[0xD800, u16::from(b'x')])
        };

        // Stay honest if a platform ever normalizes the name away.
        assert!(
            Path::new(&name)
                .file_name()
                .and_then(std::ffi::OsStr::to_str)
                .is_none(),
            "fixture must really be a non-UTF-8 file name"
        );

        let temp_dir = tempfile::TempDir::new().unwrap();
        let path = temp_dir.path().join(&name);

        let finder = MockProjectFinder::new();
        assert!(
            !finder.matches_project_file(&path).await.unwrap(),
            "a non-UTF-8 file name cannot match an ASCII manifest name"
        );
    }

    /// Finder that returns the leading-dot EXTENSION form of
    /// [`ProjectFinder::project_files`], the same shape `CSharpProjectFinder`
    /// uses.
    #[derive(Debug)]
    struct ExtensionProjectFinder;

    #[async_trait]
    impl ProjectFinder for ExtensionProjectFinder {
        fn projects(&self) -> Vec<&Project> {
            vec![]
        }

        fn projects_mut(&mut self) -> Vec<&mut Project> {
            vec![]
        }

        fn project_count(&self) -> usize {
            0
        }

        fn project_files(&self) -> &[&str] {
            &[".csproj"]
        }

        async fn visit(&mut self, _path: &Path, _relative_path: &Path) -> Result<()> {
            Ok(())
        }
    }

    // Pins the documented half of the dual contract: the defaulted
    // `matches_project_file` implements ONLY the bare-file-name form, so an
    // extension entry can never match through it — not even for a real
    // `.csproj` file on disk, in any casing. This is exactly why
    // `CSharpProjectFinder::visit` gates on `has_extension_ignore_ascii_case`
    // + `is_regular_file` instead of calling this method; if that ever
    // changed, C# discovery would silently stop finding projects.
    #[tokio::test]
    async fn test_matches_project_file_never_matches_an_extension_entry() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let finder = ExtensionProjectFinder;

        for name in ["App.csproj", "App.CSPROJ"] {
            let manifest = temp_dir.path().join(name);
            std::fs::write(&manifest, "<Project/>").unwrap();

            assert!(
                !finder.matches_project_file(&manifest).await.unwrap(),
                "{name} must not match through the file-name-only gate"
            );
            // The extension-aware check is the one that accepts it.
            assert!(has_extension_ignore_ascii_case(&manifest, "csproj"));
            assert!(is_regular_file(&manifest).await.unwrap());
        }
    }

    // Recognized name, nothing on disk: `is_regular_file` maps NotFound to
    // `Ok(false)` rather than an error, so the gate stays quiet for deleted
    // manifests still listed in the git index.
    #[tokio::test]
    async fn test_matches_project_file_rejects_missing_manifest() {
        let temp_dir = tempfile::TempDir::new().unwrap();
        let missing = temp_dir.path().join("package.json");

        let finder = MockProjectFinder::new();
        assert!(
            !finder.matches_project_file(&missing).await.unwrap(),
            "a package.json that does not exist must not match"
        );
    }
}