Rust: Prefer richer feature configurations when extracting dependencies

rust/extractor/src/configuration.rs

@@ -0,0 +1,307 @@
+use crate::util::cmp_option_by;
+use itertools::Itertools;
+use ra_ap_base_db::{SourceDatabase, SourceRootId, all_crates};
+use ra_ap_cfg::{CfgAtom, CfgOptions};
+use ra_ap_hir::Semantics;
+use ra_ap_ide_db::RootDatabase;
+use ra_ap_project_model::ProjectManifest;
+use ra_ap_vfs::{FileId, Vfs};
+use std::cmp::Ordering;
+use std::collections::{BTreeMap, HashMap, HashSet, btree_map::Entry};
+use std::path::{Path, PathBuf};
+
+/// Returns the enclosing directory of the manifest file.
+pub fn manifest_parent(manifest: &ProjectManifest) -> &Path {
+    manifest.manifest_path().parent().as_ref()
+}
+
+/// Reads the file's source-root ID in `db`.
+///
+/// A source root is rust-analyzer's group of files within which relative paths
+/// are resolved. It need not correspond to one crate: a package's library and
+/// build script can share a source root.
+fn source_root_id(db: &RootDatabase, file_id: FileId) -> SourceRootId {
+    db.file_source_root(file_id).source_root_id(db)
+}
+
+/// Maps source root IDs to cfg options shared by all their crates, using only
+/// metadata.
+///
+/// In order for a file to be included in a crate the crate's root file and the
+/// file must share the same source root. However, this condition is not
+/// sufficient. The construction here relies on this to identify the cases where
+/// this requirement _is_ sufficient to determine the cfg options of any crate
+/// that could contain the file. Either because only one crate can contain it or
+/// because all such crates share the same cfg options.
+///
+/// Roots whose crates disagree map to `None` and need per-file semantic lookup.
+fn root_configurations(db: &RootDatabase) -> HashMap<SourceRootId, Option<&CfgOptions>> {
+    let mut configurations = HashMap::new();
+    for &krate in all_crates(db).iter() {
+        let root = source_root_id(db, krate.data(db).root_file_id);
+        configurations
+            .entry(root)
+            .and_modify(|configuration: &mut Option<&CfgOptions>| {
+                // We're in the case where multiple crates share a source root,
+                // for example a library and its build script. If their cfg
+                // options differ, we mark the configuration as `None`.
+                if configuration.is_some_and(|cfg| cfg != krate.cfg_options(db)) {
+                    *configuration = None;
+                }
+            })
+            .or_insert_with(|| Some(krate.cfg_options(db)));
+    }
+    configurations
+}
+
+/// Returns an iterator over the dependency files in `vfs`.
+///
+/// These are files that correspond to libraries and that are not in `source_files`.
+pub fn dependency_files<'a>(
+    db: &'a RootDatabase,
+    vfs: &'a Vfs,
+    source_files: &'a HashSet<PathBuf>,
+) -> impl Iterator<Item = (FileId, &'a Path)> {
+    vfs.iter().filter_map(|(file_id, file)| {
+        let file = file.as_path().map(<_ as AsRef<Path>>::as_ref)?;
+        (file.extension().is_some_and(|ext| ext == "rs")
+            && !source_files.contains(file)
+            && db
+                .source_root(source_root_id(db, file_id))
+                .source_root(db)
+                .is_library)
+            .then_some((file_id, file))
+    })
+}
+
+/// Compare two `CfgOptions`. Feature count wins, sorted cfg options break ties.
+fn compare_cfg_options(left: &CfgOptions, right: &CfgOptions) -> Ordering {
+    let feature_count = |cfg_options: &CfgOptions| {
+        cfg_options
+            .into_iter()
+            .filter(|cfg| matches!(cfg, CfgAtom::KeyValue { key, .. } if key.as_str() == "feature"))
+            .count()
+    };
+    feature_count(left)
+        .cmp(&feature_count(right))
+        .then_with(|| left.into_iter().sorted().cmp(right.into_iter().sorted()))
+}
+
+#[derive(PartialEq, Eq)]
+struct DependencySource<'a> {
+    cfg_options: Option<CfgOptions>,
+    manifest: &'a ProjectManifest,
+}
+
+impl Ord for DependencySource<'_> {
+    fn cmp(&self, other: &Self) -> Ordering {
+        // Compare by cfg options first, then by manifest parent directory
+        // (arbitrary but deterministic).
+        cmp_option_by(&self.cfg_options, &other.cfg_options, compare_cfg_options)
+            .then_with(|| manifest_parent(self.manifest).cmp(manifest_parent(other.manifest)))
+    }
+}
+
+impl PartialOrd for DependencySource<'_> {
+    fn partial_cmp(&self, other: &Self) -> Option<Ordering> {
+        Some(self.cmp(other))
+    }
+}
+
+/// Tracks extraction candidates for each dependency file. Keeps one preferred
+/// manifest per distinct set of cfg options, allowing lower-priority
+/// configurations to be tried if the preferred one disables a file.
+#[derive(Default)]
+pub struct DependencyFiles<'a> {
+    files: BTreeMap<PathBuf, Vec<DependencySource<'a>>>,
+}
+
+impl<'a> DependencyFiles<'a> {
+    /// Considers dependency files loaded from this manifest as extraction
+    /// candidates. The lexicographically largest manifest directory wins for
+    /// each configuration.
+    pub fn add_manifest(
+        &mut self,
+        manifest: &'a ProjectManifest,
+        semantics: &Semantics<'_, RootDatabase>,
+        vfs: &Vfs,
+        source_files: &HashSet<PathBuf>,
+    ) {
+        let db = semantics.db;
+        let root_configurations = root_configurations(db);
+        for (file_id, file) in dependency_files(db, vfs, source_files) {
+            let root = source_root_id(db, file_id);
+            let cfg_options = match root_configurations.get(&root) {
+                // There's a unique cfg configuration for this file.
+                Some(Some(configuration)) => Some((*configuration).clone()),
+                // There's no unique cfg configuration for this file. In this
+                // case we fall back to the expensive `file_to_module_def`
+                // approach which requires rust_analyzer to calculate semantic
+                // information.
+                Some(None) => semantics
+                    .file_to_module_def(file_id)
+                    .map(|module| module.krate(db).cfg(db).clone()),
+                None => None,
+            };
+            self.add_candidate(file, cfg_options, manifest);
+        }
+    }
+
+    /// Adds `cfg_options` and `manifest` as an extraction candidate for the
+    /// given file unless a higher-priority candidate with the same cfg options
+    /// already exists.
+    fn add_candidate(
+        &mut self,
+        file: &Path,
+        cfg_options: Option<CfgOptions>,
+        manifest: &'a ProjectManifest,
+    ) {
+        let candidate = DependencySource {
+            cfg_options,
+            manifest,
+        };
+        match self.files.entry(file.to_owned()) {
+            Entry::Occupied(mut entry) => {
+                let sources = entry.get_mut();
+                if let Some(source) = sources
+                    .iter_mut()
+                    .find(|source| source.cfg_options == candidate.cfg_options)
+                {
+                    if manifest_parent(manifest) > manifest_parent(source.manifest) {
+                        source.manifest = manifest;
+                    }
+                } else {
+                    sources.push(candidate);
+                }
+            }
+            Entry::Vacant(entry) => {
+                entry.insert(vec![candidate]);
+            }
+        }
+    }
+
+    /// Takes the highest-priority remaining context for each file, grouped by
+    /// manifest. Lower-priority configurations remain available if the selected
+    /// one disables the file.
+    pub fn by_manifests(&mut self) -> HashMap<&'a ProjectManifest, Vec<PathBuf>> {
+        self.files
+            .iter_mut()
+            .filter_map(|(file, sources)| {
+                sources.sort_unstable();
+                sources.pop().map(|source| (source.manifest, file.clone()))
+            })
+            .into_group_map()
+    }
+
+    /// Returns `true` if there are alternative configurations left for the
+    /// given file.
+    pub fn has_alternative(&self, file: &Path) -> bool {
+        self.files
+            .get(file)
+            .is_some_and(|sources| !sources.is_empty())
+    }
+
+    /// Marks a file as extracted, removing it from the map of files.
+    pub fn extracted(&mut self, file: &Path) {
+        self.files.remove(file);
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use ra_ap_base_db::{
+        CrateGraphBuilder, CrateOrigin, CrateWorkspaceData, FileChange, FileSet, SourceRoot,
+    };
+    use ra_ap_intern::Symbol;
+    use ra_ap_span::Edition;
+    use ra_ap_vfs::VfsPath;
+    use triomphe::Arc;
+
+    fn cfg(features: &[&str]) -> CfgOptions {
+        let mut cfg = CfgOptions::default();
+        for feature in features {
+            cfg.insert_key_value(Symbol::intern("feature"), Symbol::intern(feature));
+        }
+        cfg
+    }
+
+    fn workspace(files: &[&str], crates: &[(&str, &[&str])]) -> (RootDatabase, Vfs) {
+        let root = crate::cwd().unwrap();
+        let mut vfs = Vfs::default();
+        let mut file_set = FileSet::default();
+        let mut change = FileChange::default();
+        for &name in files {
+            let path = VfsPath::from(root.join(name));
+            vfs.set_file_contents(path.clone(), Some(Vec::new()));
+            let (id, _) = vfs.file_id(&path).unwrap();
+            file_set.insert(id, path);
+        }
+        let mut graph = CrateGraphBuilder::default();
+        for &(name, features) in crates {
+            let (id, _) = vfs.file_id(&VfsPath::from(root.join(name))).unwrap();
+            graph.add_crate_root(
+                id,
+                Edition::Edition2024,
+                None,
+                None,
+                cfg(features),
+                None,
+                Default::default(),
+                CrateOrigin::Library {
+                    repo: None,
+                    name: Symbol::intern(name),
+                },
+                Vec::new(),
+                false,
+                Arc::new(root.clone()),
+                Arc::new(CrateWorkspaceData {
+                    target: Err("unused in these tests".into()),
+                    toolchain: None,
+                }),
+            );
+        }
+        change.set_roots(vec![SourceRoot::new_library(file_set)]);
+        change.set_crate_graph(graph);
+        let mut db = RootDatabase::default();
+        change.apply(&mut db);
+        (db, vfs)
+    }
+
+    #[test]
+    fn selects_configuration_without_reading_dependency_sources() {
+        for crates in [
+            vec![("lib.rs", &["a"][..])],
+            vec![("lib.rs", &["a"][..]), ("build.rs", &["a"][..])],
+        ] {
+            let (db, vfs) = workspace(&["lib.rs", "build.rs", "module.rs"], &crates);
+            let root = crate::cwd().unwrap();
+            let manifest = ProjectManifest::from_manifest_file(root.join("Cargo.toml")).unwrap();
+            let mut plan = DependencyFiles::default();
+            plan.add_manifest(&manifest, &Semantics::new(&db), &vfs, &HashSet::new());
+            assert_eq!(plan.files.len(), 3);
+            for sources in plan.files.values() {
+                assert_eq!(sources[0].cfg_options, Some(cfg(&["a"])));
+            }
+        }
+    }
+
+    #[test]
+    fn selects_highest_priority_in_either_order() {
+        let root = crate::cwd().unwrap();
+        let manifest = |path| ProjectManifest::from_manifest_file(root.join(path)).unwrap();
+        let candidates = [
+            (manifest("z/Cargo.toml"), cfg(&["z"])),
+            (manifest("a/Cargo.toml"), cfg(&["a", "b"])),
+            (manifest("b/Cargo.toml"), cfg(&["a", "b"])),
+        ];
+        for order in [[0, 1, 2], [2, 1, 0]] {
+            let mut plan = DependencyFiles::default();
+            for (manifest, cfg) in order.map(|index| &candidates[index]) {
+                plan.add_candidate(Path::new("lib.rs"), Some(cfg.clone()), manifest);
+            }
+            let expected = HashMap::from([(&candidates[2].0, vec![PathBuf::from("lib.rs")])]);
+            assert_eq!(plan.by_manifests(), expected);
+        }
+    }
+}

rust/extractor/src/diagnostics.rs

@@ -5,7 +5,7 @@ use chrono::{DateTime, Utc};
 use ra_ap_project_model::ProjectManifest;
 use serde::Serialize;
 use serde::ser::SerializeMap;
-use std::collections::HashMap;
+use std::collections::{HashMap, HashSet};
 use std::fmt::Display;
 use std::fs::File;
 use std::path::{Path, PathBuf};
@@ -220,13 +220,15 @@ struct ExtractionSummary {
 type ExtractionDiagnostics = Diagnostics<ExtractionSummary>;
 
 fn summary(start: Instant, steps: &[ExtractionStep]) -> ExtractionSummary {
-    let mut number_of_manifests = 0;
+    // The manifests loaded during extraction. We might load the same manifest
+    // multiple times, so we use a set to keep track of unique manifests.
+    let mut manifests = HashSet::new();
     let mut number_of_files = 0;
     let mut durations = HashMap::new();
     for step in steps {
         match &step.action {
             ExtractionStepKind::LoadManifest => {
-                number_of_manifests += 1;
+                manifests.extend(step.file.as_ref());
             }
             ExtractionStepKind::Parse => {
                 number_of_files += 1;
@@ -244,7 +246,7 @@ fn summary(start: Instant, steps: &[ExtractionStep]) -> ExtractionSummary {
     }
     info!("total duration: {total}");
     ExtractionSummary {
-        number_of_manifests,
+        number_of_manifests: manifests.len(),
         number_of_files,
         durations: DurationsSummary { durations, total },
     }
@@ -284,3 +286,22 @@ pub fn emit_extraction_diagnostics(
         .with_context(|| format!("writing to diagnostics file {}", target.display()))?;
     Ok(())
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn counts_unique_manifests_but_includes_reload_duration() {
+        let steps: Vec<_> = ["a/Cargo.toml", "b/Cargo.toml", "a/Cargo.toml"]
+            .into_iter()
+            .map(|path| ExtractionStep {
+                action: ExtractionStepKind::LoadManifest,
+                file: Some(PathBuf::from(path)),
+                ms: 1,
+            })
+            .collect();
+        let summary = summary(Instant::now(), &steps);
+        assert_eq!(summary.number_of_manifests, 2);
+    }
+}