feat: detect compilation-related vfs changes (#1199)

crates/tinymist-project/src/compiler.rs

@@ -26,7 +26,7 @@ use crate::LspCompilerFeat;
 use tinymist_world::{
     vfs::{
         notify::{FilesystemEvent, MemoryEvent, NotifyMessage, UpstreamUpdateEvent},
-        FsProvider, RevisingVfs,
+        FileId, FsProvider, RevisingVfs,
     },
     CompilerFeat, CompilerUniverse, CompilerWorld, EntryReader, EntryState, TaskInputs, WorldDeps,
 };
@@ -99,6 +99,7 @@ impl<F: CompilerFeat + 'static> CompileSnapshot<F> {
             snap,
             doc,
             warnings,
+            deps: OnceLock::default(),
         }
     }
 }
@@ -122,6 +123,8 @@ pub struct CompiledArtifact<F: CompilerFeat> {
     pub warnings: EcoVec<SourceDiagnostic>,
     /// The compiled document.
     pub doc: SourceResult<Arc<TypstDocument>>,
+    /// The depended files.
+    pub deps: OnceLock<EcoVec<FileId>>,
 }
 
 impl<F: CompilerFeat> std::ops::Deref for CompiledArtifact<F> {
@@ -138,6 +141,7 @@ impl<F: CompilerFeat> Clone for CompiledArtifact<F> {
             snap: self.snap.clone(),
             doc: self.doc.clone(),
             warnings: self.warnings.clone(),
+            deps: self.deps.clone(),
         }
     }
 }
@@ -150,6 +154,17 @@ impl<F: CompilerFeat> CompiledArtifact<F> {
             .cloned()
             .or_else(|| self.snap.success_doc.clone())
     }
+
+    pub fn depended_files(&self) -> &EcoVec<FileId> {
+        self.deps.get_or_init(|| {
+            let mut deps = EcoVec::default();
+            self.world.iter_dependencies(&mut |f| {
+                deps.push(f);
+            });
+
+            deps
+        })
+    }
 }
 
 pub trait CompileHandler<F: CompilerFeat, Ext>: Send + Sync + 'static {
@@ -225,6 +240,14 @@ impl CompileReasons {
     pub fn any(&self) -> bool {
         self.by_memory_events || self.by_fs_events || self.by_entry_update
     }
+
+    pub fn exclude(&self, excluded: Self) -> Self {
+        Self {
+            by_memory_events: self.by_memory_events && !excluded.by_memory_events,
+            by_fs_events: self.by_fs_events && !excluded.by_fs_events,
+            by_entry_update: self.by_entry_update && !excluded.by_entry_update,
+        }
+    }
 }
 
 fn no_reason() -> CompileReasons {

crates/tinymist-vfs/src/lib.rs

@@ -187,7 +187,11 @@ pub struct Vfs<M: PathAccessModel + Sized> {
 
 impl<M: PathAccessModel + Sized> fmt::Debug for Vfs<M> {
     fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        f.debug_struct("Vfs").finish()
+        f.debug_struct("Vfs")
+            .field("revision", &self.revision)
+            .field("managed", &self.managed.lock().entries.size())
+            .field("paths", &self.paths.lock().paths.len())
+            .finish()
     }
 }
 
@@ -211,10 +215,25 @@ impl<M: PathAccessModel + Clone + Sized> Vfs<M> {
             source_cache: self.source_cache.clone(),
             managed: Arc::new(Mutex::new(EntryMap::default())),
             paths: Arc::new(Mutex::new(PathMap::default())),
-            revision: NonZeroUsize::new(1).expect("initial revision is 1"),
+            revision: NonZeroUsize::new(2).expect("initial revision is 2"),
             access_model: self.access_model.clone(),
         }
     }
+
+    pub fn is_clean_compile(&self, rev: usize, file_ids: &[FileId]) -> bool {
+        let mut m = self.managed.lock();
+        for id in file_ids {
+            let Some(entry) = m.entries.get_mut(id) else {
+                log::info!("Vfs(dirty, {id:?}): file id not found");
+                return false;
+            };
+            if entry.changed_at == 0 || entry.changed_at >= rev {
+                log::info!("Vfs(dirty, {id:?}): rev {rev:?} => {:?}", entry.changed_at);
+                return false;
+            }
+        }
+        true
+    }
 }
 
 impl<M: PathAccessModel + Sized> Vfs<M> {
@@ -247,7 +266,7 @@ impl<M: PathAccessModel + Sized> Vfs<M> {
             source_cache: SourceCache::default(),
             managed: Arc::default(),
             paths: Arc::default(),
-            revision: NonZeroUsize::new(1).expect("initial revision is 1"),
+            revision: NonZeroUsize::new(2).expect("initial revision is 2"),
             access_model,
         }
     }
@@ -326,17 +345,20 @@ impl<M: PathAccessModel + Sized> Vfs<M> {
 
     /// Reads a file.
     pub fn read(&self, fid: TypstFileId) -> FileResult<Bytes> {
-        let bytes = self.managed.lock().slot(fid, |entry| entry.bytes.clone());
+        let bytes = self.managed.lock().slot(fid, |entry| {
+            entry.changed_at = entry.changed_at.max(1);
+            entry.bytes.clone()
+        });
 
         self.read_content(&bytes, fid).clone()
     }
 
     /// Reads a source.
     pub fn source(&self, file_id: TypstFileId) -> FileResult<Source> {
-        let (bytes, source) = self
-            .managed
-            .lock()
-            .slot(file_id, |entry| (entry.bytes.clone(), entry.source.clone()));
+        let (bytes, source) = self.managed.lock().slot(file_id, |entry| {
+            entry.changed_at = entry.changed_at.max(1);
+            (entry.bytes.clone(), entry.source.clone())
+        });
 
         let source = source.get_or_init(|| {
             let content = self
@@ -442,6 +464,7 @@ impl<M: PathAccessModel + Sized> RevisingVfs<'_, M> {
     fn invalidate_file_id(&mut self, file_id: TypstFileId) {
         self.view_changed = true;
         self.managed.slot(file_id, |e| {
+            e.changed_at = self.inner.revision.get();
             e.bytes = Arc::default();
             e.source = Arc::default();
         });
@@ -524,6 +547,7 @@ type BytesQuery = Arc<OnceLock<(Option<ImmutPath>, usize, FileResult<Bytes>)>>;
 
 #[derive(Debug, Clone, Default)]
 struct VfsEntry {
+    changed_at: usize,
     bytes: BytesQuery,
     source: Arc<OnceLock<FileResult<Source>>>,
 }

crates/tinymist-world/src/source.rs

@@ -1,31 +1,20 @@
 // use std::sync::Arc;
 
 use core::fmt;
-use std::{num::NonZeroUsize, sync::Arc};
+use std::sync::Arc;
 
 use parking_lot::Mutex;
-use tinymist_std::hash::FxHashMap;
-use tinymist_std::QueryRef;
+use tinymist_std::{hash::FxHashMap, QueryRef};
 use tinymist_vfs::{Bytes, FsProvider, TypstFileId};
-use typst::{
-    diag::{FileError, FileResult},
-    syntax::Source,
-};
-
-/// incrementally query a value from a self holding state
-type IncrQueryRef<S, E> = QueryRef<S, E, Option<S>>;
+use typst::diag::{FileError, FileResult};
+use typst::syntax::Source;
 
 type FileQuery<T> = QueryRef<T, FileError>;
-type IncrFileQuery<T> = IncrQueryRef<T, FileError>;
-
-pub trait Revised {
-    fn last_accessed_rev(&self) -> NonZeroUsize;
-}
 
 pub struct SourceCache {
     touched_by_compile: bool,
     fid: TypstFileId,
-    source: IncrFileQuery<Source>,
+    source: FileQuery<Source>,
     buffer: FileQuery<Bytes>,
 }
 
@@ -103,22 +92,7 @@ impl SourceDb {
     /// See `Vfs::resolve_with_f` for more information.
     pub fn source(&self, fid: TypstFileId, p: &impl FsProvider) -> FileResult<Source> {
         self.slot(fid, |slot| {
-            slot.source
-                .compute_with_context(|prev| {
-                    let content = p.read(fid)?;
-                    let next = from_utf8_or_bom(&content)?.to_owned();
-
-                    // otherwise reparse the source
-                    match prev {
-                        Some(mut source) => {
-                            source.replace(&next);
-                            Ok(source)
-                        }
-                        // Return a new source if we don't have a reparse feature or no prev
-                        _ => Ok(Source::new(fid, next)),
-                    }
-                })
-                .cloned()
+            slot.source.compute(|| p.read_source(fid)).cloned()
         })
     }
 
@@ -129,7 +103,7 @@ impl SourceDb {
             let entry = slots.entry(fid).or_insert_with(|| SourceCache {
                 touched_by_compile: self.is_compiling,
                 fid,
-                source: IncrFileQuery::with_context(None),
+                source: FileQuery::default(),
                 buffer: FileQuery::default(),
             });
             if self.is_compiling && !entry.touched_by_compile {
@@ -150,23 +124,3 @@ impl SourceDb {
         }
     }
 }
-
-pub trait MergeCache: Sized {
-    fn merge(self, _other: Self) -> Self {
-        self
-    }
-}
-
-pub struct FontDb {}
-pub struct PackageDb {}
-
-/// Convert a byte slice to a string, removing UTF-8 BOM if present.
-fn from_utf8_or_bom(buf: &[u8]) -> FileResult<&str> {
-    Ok(std::str::from_utf8(if buf.starts_with(b"\xef\xbb\xbf") {
-        // remove UTF-8 BOM
-        &buf[3..]
-    } else {
-        // Assume UTF-8
-        buf
-    })?)
-}

crates/tinymist-world/src/world.rs

@@ -121,7 +121,9 @@ impl<F: CompilerFeat> CompilerUniverse<F> {
     pub fn increment_revision<T>(&mut self, f: impl FnOnce(&mut RevisingUniverse<F>) -> T) -> T {
         f(&mut RevisingUniverse {
             vfs_revision: self.vfs.revision(),
+            font_changed: false,
             font_revision: self.font_resolver.revision(),
+            registry_changed: false,
             registry_revision: self.registry.revision(),
             view_changed: false,
             inner: self,
@@ -265,7 +267,9 @@ impl<F: CompilerFeat> EntryManager for CompilerUniverse<F> {
 pub struct RevisingUniverse<'a, F: CompilerFeat> {
     view_changed: bool,
     vfs_revision: NonZeroUsize,
+    font_changed: bool,
     font_revision: Option<NonZeroUsize>,
+    registry_changed: bool,
     registry_revision: Option<NonZeroUsize>,
     pub inner: &'a mut CompilerUniverse<F>,
 }
@@ -298,6 +302,7 @@ impl<F: CompilerFeat> Drop for RevisingUniverse<'_, F> {
             view_changed = true;
 
             // The registry has changed affects the vfs cache.
+            log::info!("resetting shadow registry_changed");
             self.vfs().reset_cache();
         }
         let view_changed = view_changed || self.vfs_changed();
@@ -316,10 +321,12 @@ impl<F: CompilerFeat> RevisingUniverse<'_, F> {
     }
 
     pub fn set_fonts(&mut self, fonts: Arc<F::FontResolver>) {
+        self.font_changed = true;
         self.inner.font_resolver = fonts;
     }
 
     pub fn set_package(&mut self, packages: Arc<F::Registry>) {
+        self.registry_changed = true;
         self.inner.registry = packages;
     }
 
@@ -340,6 +347,7 @@ impl<F: CompilerFeat> RevisingUniverse<'_, F> {
         // Resets the cache if the workspace root has changed.
         let root_changed = self.inner.entry.workspace_root() == state.workspace_root();
         if root_changed {
+            log::info!("resetting shadow root_changed");
             self.vfs().reset_cache();
         }
 
@@ -351,11 +359,12 @@ impl<F: CompilerFeat> RevisingUniverse<'_, F> {
     }
 
     pub fn font_changed(&self) -> bool {
-        is_revision_changed(self.font_revision, self.font_resolver.revision())
+        self.font_changed && is_revision_changed(self.font_revision, self.font_resolver.revision())
     }
 
     pub fn registry_changed(&self) -> bool {
-        is_revision_changed(self.registry_revision, self.registry.revision())
+        self.registry_changed
+            && is_revision_changed(self.registry_revision, self.registry.revision())
     }
 
     pub fn vfs_changed(&self) -> bool {
@@ -447,6 +456,10 @@ impl<F: CompilerFeat> CompilerWorld<F> {
         self.source_db.take_state()
     }
 
+    pub fn vfs(&self) -> &Vfs<F::AccessModel> {
+        &self.vfs
+    }
+
     /// Sets flag to indicate whether the compiler is currently compiling.
     /// Note: Since `CompilerWorld` can be cloned, you can clone the world and
     /// set the flag then to avoid affecting the original world.

crates/tinymist/src/project.rs

@@ -83,6 +83,7 @@ impl LspPreviewState {
 #[derive(Default)]
 pub struct ProjectStateExt {
     pub is_compiling: bool,
+    pub last_compilation: Option<CompiledArtifact<LspCompilerFeat>>,
 }
 
 /// LSP project compiler.
@@ -129,6 +130,7 @@ impl Project {
             let proj = self.state.projects().find(|p| p.id == compiled.id);
             if let Some(proj) = proj {
                 proj.ext.is_compiling = false;
+                proj.ext.last_compilation = Some(compiled.clone());
             }
         }
 
@@ -228,6 +230,33 @@ impl CompileHandler<LspCompilerFeat, ProjectStateExt> for CompileHandlerImpl {
                 continue;
             }
 
+            let reason = s.reason;
+
+            const VFS_SUB: CompileReasons = CompileReasons {
+                by_memory_events: true,
+                by_fs_events: true,
+                by_entry_update: false,
+            };
+
+            let is_vfs_sub = reason.any() && !reason.exclude(VFS_SUB).any();
+            let id = &s.id;
+
+            if is_vfs_sub
+                && 'vfs_is_clean: {
+                    let Some(compilation) = &s.ext.last_compilation else {
+                        break 'vfs_is_clean false;
+                    };
+
+                    let last_rev = compilation.world.vfs().revision();
+                    let deps = compilation.depended_files().clone();
+                    s.verse.vfs().is_clean_compile(last_rev.get(), &deps)
+                }
+            {
+                log::info!("Project: skip compilation for {id:?} due to harmless vfs changes");
+                s.reason = CompileReasons::default();
+                continue;
+            }
+
             let Some(compile_fn) = s.may_compile(&c.handler) else {
                 continue;
             };
@@ -287,7 +316,7 @@ impl CompileHandler<LspCompilerFeat, ProjectStateExt> for CompileHandlerImpl {
             let mut n_rev = self.notified_revision.lock();
             if *n_rev >= snap.world.revision().get() {
                 log::info!(
-                    "TypstActor: already notified for revision {} <= {n_rev}",
+                    "Project: already notified for revision {} <= {n_rev}",
                     snap.world.revision(),
                 );
                 return;
@@ -407,11 +436,11 @@ impl QuerySnap {
     pub fn run_analysis<T>(self, f: impl FnOnce(&mut LocalContextGuard) -> T) -> anyhow::Result<T> {
         let world = self.snap.world;
         let Some(main) = world.main_id() else {
-            error!("TypstActor: main file is not set");
+            error!("Project: main file is not set");
             bail!("main file is not set");
         };
         world.source(main).map_err(|err| {
-            info!("TypstActor: failed to prepare main file: {err:?}");
+            info!("Project: failed to prepare main file: {err:?}");
             anyhow::anyhow!("failed to get source: {err}")
         })?;