cargo 0.100.0

use crate::util::data_structures::HashMap;
use std::path::Path;
use std::path::PathBuf;

use anyhow::Context as _;
use cargo_util::paths;
use serde::Deserialize;
use serde::Serialize;

use crate::CargoResult;
use crate::compiler::BuildRunner;
use crate::compiler::CompileKind;

/// JSON Schema of the [`RustdocFingerprint`] file.
#[derive(Debug, Serialize, Deserialize)]
struct RustdocFingerprintJson {
    /// `rustc -vV` verbose version output.
    pub rustc_vv: String,

    /// Relative paths to cross crate info JSON files from previous `cargo doc` invocations.
    #[serde(default, skip_serializing_if = "Vec::is_empty")]
    pub doc_parts: Vec<PathBuf>,
}

/// Structure used to deal with Rustdoc fingerprinting
///
/// This is important because the `.js`/`.html` & `.css` files
/// that are generated by Rustdoc don't have any versioning yet
/// (see <https://github.com/rust-lang/cargo/issues/8461>).
/// Therefore, we can end up with weird bugs and behaviours
/// if we mix different versions of these files.
///
/// We need to make sure that if there were any previous docs already compiled,
/// they were compiled with the same Rustdoc version that we're currently using.
/// Otherwise we must remove the `doc/` folder and compile again forcing a rebuild.
#[derive(Debug)]
pub struct RustdocFingerprint {
    /// Path to the fingerprint file.
    path: PathBuf,
    /// `rustc -vV` verbose version output for the current session.
    rustc_vv: String,
    /// Absolute paths to new cross crate info JSON files generated in the current session.
    doc_parts: Vec<PathBuf>,
    /// The fingerprint file on disk.
    on_disk: Option<RustdocFingerprintJson>,
}

impl RustdocFingerprint {
    /// Checks whether the latest version of rustc used to compile this workspace's docs
    /// was the same as the one is currently being used in this `cargo doc` call.
    ///
    /// In case it's not,
    /// it takes care of removing the `<build-dir>/doc/` folder
    /// as well as overwriting the rustdoc fingerprint info.
    /// This is to guarantee that we won't end up with mixed versions of the `js/html/css` files
    /// which `rustdoc` autogenerates without any versioning.
    ///
    /// Each requested target platform maintains its own fingerprint file.
    /// That is, if you run `cargo doc` and then `cargo doc --target wasm32-wasip1`,
    /// you will have two separate fingerprint files:
    ///
    /// * `<build-dir>/.rustdoc_fingerprint.json` for host
    /// * `<build-dir>/wasm32-wasip1/.rustdoc_fingerprint.json`
    pub fn check_rustdoc_fingerprint(build_runner: &BuildRunner<'_, '_>) -> CargoResult<()> {
        if build_runner
            .bcx
            .gctx
            .cli_unstable()
            .skip_rustdoc_fingerprint
        {
            return Ok(());
        }
        let new_fingerprint = RustdocFingerprintJson {
            rustc_vv: build_runner.bcx.rustc().verbose_version.clone(),
            doc_parts: Vec::new(),
        };

        for kind in &build_runner.bcx.build_config.requested_kinds {
            check_fingerprint(build_runner, &new_fingerprint, *kind)?;
        }

        Ok(())
    }

    /// Creates a new fingerprint with given doc parts paths.
    pub fn new(
        build_runner: &BuildRunner<'_, '_>,
        kind: CompileKind,
        doc_parts: Vec<PathBuf>,
    ) -> Self {
        let path = fingerprint_path(build_runner, kind);
        let rustc_vv = build_runner.bcx.rustc().verbose_version.clone();
        let on_disk = load_on_disk(&path);
        Self {
            path,
            rustc_vv,
            doc_parts,
            on_disk,
        }
    }

    /// Persists the fingerprint.
    ///
    /// The closure will run before persisting the fingerprint,
    /// and will be given a list of doc parts directories for passing to
    /// `rustdoc --read-doc-meta-dir`.
    pub fn persist<F>(&self, exec: F) -> CargoResult<()>
    where
        // 1. paths for `--read-doc-meta-dir`
        F: Fn(&[&Path]) -> CargoResult<()>,
    {
        // Dedupe crate with the same name by file stem (which is effectively crate name),
        // since rustdoc doesn't distinguish different crate versions.
        //
        // Rules applied here:
        //
        // * If name collides, favor the one selected via CLI over cached ones
        //   (done by the insertion order)
        let base = self.path.parent().unwrap();
        let on_disk_doc_parts: Vec<_> = self
            .on_disk
            .iter()
            .flat_map(|on_disk| {
                on_disk
                    .doc_parts
                    .iter()
                    // Make absolute so that we can pass to rustdoc
                    .map(|p| base.join(p))
                    // Doc parts may be selectively cleaned by `cargo clean -p <doc>`.
                    // We should stop caching those no-exist.
                    .filter(|p| p.exists())
            })
            .collect();
        let dedup_map = on_disk_doc_parts
            .iter()
            .chain(self.doc_parts.iter())
            .map(|p| (p.file_stem(), p))
            .collect::<HashMap<_, _>>();
        let mut doc_parts: Vec<_> = dedup_map.into_values().collect();
        doc_parts.sort_unstable();

        // Prepare args for `rustdoc --read-doc-meta-dir`
        let doc_parts_dirs: Vec<_> = doc_parts.iter().map(|p| p.parent().unwrap()).collect();
        exec(&doc_parts_dirs)?;

        // Persist with relative paths to the directory where fingerprint file is at.
        let json = RustdocFingerprintJson {
            rustc_vv: self.rustc_vv.clone(),
            doc_parts: doc_parts
                .iter()
                .map(|p| p.strip_prefix(base).unwrap_or(p).to_owned())
                .collect(),
        };
        paths::write(&self.path, serde_json::to_string(&json)?)?;

        Ok(())
    }

    /// Checks if the fingerprint is outdated comparing against given doc parts file paths.
    pub fn is_dirty(&self) -> bool {
        let Some(on_disk) = self.on_disk.as_ref() else {
            return true;
        };

        let Some(fingerprint_mtime) = paths::mtime(&self.path).ok() else {
            return true;
        };

        if self.rustc_vv != on_disk.rustc_vv {
            return true;
        }

        for path in &self.doc_parts {
            let parts_mtime = match paths::mtime(&path) {
                Ok(mtime) => mtime,
                Err(e) => {
                    tracing::debug!("failed to read mtime of {}: {e}", path.display());
                    return true;
                }
            };

            if parts_mtime > fingerprint_mtime {
                return true;
            }
        }

        false
    }
}

/// Returns the path to rustdoc fingerprint file for a given [`CompileKind`].
fn fingerprint_path(build_runner: &BuildRunner<'_, '_>, kind: CompileKind) -> PathBuf {
    build_runner
        .files()
        .layout(kind)
        .build_dir()
        .root()
        .join(".rustdoc_fingerprint.json")
}

/// Checks rustdoc fingerprint file for a given [`CompileKind`].
fn check_fingerprint(
    build_runner: &BuildRunner<'_, '_>,
    new_fingerprint: &RustdocFingerprintJson,
    kind: CompileKind,
) -> CargoResult<()> {
    let fingerprint_path = fingerprint_path(build_runner, kind);

    let write_fingerprint = || -> CargoResult<()> {
        paths::write(&fingerprint_path, serde_json::to_string(new_fingerprint)?)
    };

    let Ok(rustdoc_data) = paths::read(&fingerprint_path) else {
        // If the fingerprint does not exist, do not clear out the doc
        // directories. Otherwise this ran into problems where projects
        // like bootstrap were creating the doc directory before running
        // `cargo doc` in a way that deleting it would break it.
        return write_fingerprint();
    };

    match serde_json::from_str::<RustdocFingerprintJson>(&rustdoc_data) {
        Ok(on_disk_fingerprint) => {
            if on_disk_fingerprint.rustc_vv == new_fingerprint.rustc_vv {
                return Ok(());
            } else {
                tracing::debug!(
                    "doc fingerprint changed:\noriginal:\n{}\nnew:\n{}",
                    on_disk_fingerprint.rustc_vv,
                    new_fingerprint.rustc_vv
                );
            }
        }
        Err(e) => {
            tracing::debug!("could not deserialize {:?}: {}", fingerprint_path, e);
        }
    };
    // Fingerprint does not match, delete the doc directories and write a new fingerprint.
    tracing::debug!(
        "fingerprint {:?} mismatch, clearing doc directories",
        fingerprint_path
    );
    let doc_dir = build_runner
        .files()
        .layout(kind)
        .artifact_dir()
        .expect("artifact-dir was not locked")
        .doc();
    if doc_dir.exists() {
        clean_doc(doc_dir)?;
    }

    write_fingerprint()?;

    Ok(())
}

/// Loads an on-disk fingerprint JSON file.
fn load_on_disk(path: &Path) -> Option<RustdocFingerprintJson> {
    let on_disk = match paths::read(path) {
        Ok(data) => data,
        Err(e) => {
            tracing::debug!("failed to read rustdoc fingerprint at {path:?}: {e}");
            return None;
        }
    };

    match serde_json::from_str::<RustdocFingerprintJson>(&on_disk) {
        Ok(on_disk) => Some(on_disk),
        Err(e) => {
            tracing::debug!("could not deserialize {path:?}: {e}");
            None
        }
    }
}

fn clean_doc(path: &Path) -> CargoResult<()> {
    let entries = path
        .read_dir()
        .with_context(|| format!("failed to read directory `{}`", path.display()))?;
    for entry in entries {
        let entry = entry?;
        // Don't remove hidden files. Rustdoc does not create them,
        // but the user might have.
        if entry
            .file_name()
            .to_str()
            .map_or(false, |name| name.starts_with('.'))
        {
            continue;
        }
        let path = entry.path();
        if entry.file_type()?.is_dir() {
            paths::remove_dir_all(path)?;
        } else {
            paths::remove_file(path)?;
        }
    }
    Ok(())
}