refactoring: remove dependency on SimpleFiles, make tree parsing multithreaded

crates/treehouse/src/state.rs

@@ -3,23 +3,19 @@ use std::{collections::HashMap, ops::Range};
 use anyhow::Context;
 use codespan_reporting::{
     diagnostic::{Diagnostic, Label, LabelStyle, Severity},
-    files::SimpleFiles,
     term::termcolor::{ColorChoice, StandardStream},
 };
+use tracing::instrument;
 use ulid::Ulid;
 
 use crate::{
     tree::{SemaBranchId, SemaRoots, SemaTree},
-    vfs::VPathBuf,
+    vfs::{VPath, VPathBuf},
 };
 
 #[derive(Debug, Clone)]
 pub enum Source {
-    Tree {
-        input: String,
-        tree_path: String,
-        target_path: VPathBuf,
-    },
+    Tree { input: String, tree_path: VPathBuf },
     Other(String),
 }
 
@@ -38,26 +34,54 @@ impl AsRef<str> for Source {
     }
 }
 
-pub type Files = SimpleFiles<String, Source>;
-pub type FileId = <Files as codespan_reporting::files::Files<'static>>::FileId;
+#[derive(Debug, Clone)]
+pub struct File {
+    pub path: VPathBuf,
+    pub source: Source,
+    pub line_starts: Vec<usize>,
+}
+
+impl File {
+    fn line_start(&self, line_index: usize) -> Result<usize, codespan_reporting::files::Error> {
+        use std::cmp::Ordering;
+
+        match line_index.cmp(&self.line_starts.len()) {
+            Ordering::Less => Ok(self
+                .line_starts
+                .get(line_index)
+                .cloned()
+                .expect("failed despite previous check")),
+            Ordering::Equal => Ok(self.source.as_ref().len()),
+            Ordering::Greater => Err(codespan_reporting::files::Error::LineTooLarge {
+                given: line_index,
+                max: self.line_starts.len() - 1,
+            }),
+        }
+    }
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
+pub struct FileId(usize);
 
 /// Treehouse compilation context.
 pub struct Treehouse {
-    pub files: Files,
+    pub files: Vec<File>,
+    pub files_by_tree_path: HashMap<VPathBuf, FileId>,
 
     pub tree: SemaTree,
     pub branches_by_named_id: HashMap<String, SemaBranchId>,
-    pub roots: HashMap<String, SemaRoots>,
+    pub roots: HashMap<FileId, SemaRoots>,
 
     pub branch_redirects: HashMap<String, SemaBranchId>,
 
-    missingno_generator: ulid::Generator,
+    pub missingno_generator: ulid::Generator,
 }
 
 impl Treehouse {
     pub fn new() -> Self {
         Self {
-            files: Files::new(),
+            files: vec![],
+            files_by_tree_path: HashMap::new(),
 
             tree: SemaTree::default(),
             branches_by_named_id: HashMap::new(),
@@ -69,27 +93,34 @@ impl Treehouse {
         }
     }
 
-    pub fn add_file(&mut self, filename: String, source: Source) -> FileId {
-        self.files.add(filename, source)
+    pub fn add_file(&mut self, path: VPathBuf, source: Source) -> FileId {
+        let id = FileId(self.files.len());
+        self.files.push(File {
+            line_starts: codespan_reporting::files::line_starts(source.input()).collect(),
+
+            path,
+            source,
+        });
+        id
+    }
+
+    /// Get the name of a file, assuming it was previously registered.
+    pub fn path(&self, file_id: FileId) -> &VPath {
+        &self.files[file_id.0].path
     }
 
     /// Get the source code of a file, assuming it was previously registered.
     pub fn source(&self, file_id: FileId) -> &Source {
-        self.files
-            .get(file_id)
-            .expect("file should have been registered previously")
-            .source()
+        &self.files[file_id.0].source
     }
 
-    /// Get the name of a file, assuming it was previously registered.
-    pub fn filename(&self, file_id: FileId) -> &str {
-        self.files
-            .get(file_id)
-            .expect("file should have been registered previously")
-            .name()
+    pub fn set_source(&mut self, file_id: FileId, source: Source) {
+        self.files[file_id.0].line_starts =
+            codespan_reporting::files::line_starts(source.input()).collect();
+        self.files[file_id.0].source = source;
     }
 
-    pub fn tree_path(&self, file_id: FileId) -> Option<&str> {
+    pub fn tree_path(&self, file_id: FileId) -> Option<&VPath> {
         match self.source(file_id) {
             Source::Tree { tree_path, .. } => Some(tree_path),
             Source::Other(_) => None,
@@ -109,6 +140,49 @@ impl Default for Treehouse {
     }
 }
 
+impl<'a> codespan_reporting::files::Files<'a> for Treehouse {
+    type FileId = FileId;
+
+    type Name = &'a VPath;
+
+    type Source = &'a str;
+
+    fn name(&'a self, id: Self::FileId) -> Result<Self::Name, codespan_reporting::files::Error> {
+        Ok(self.path(id))
+    }
+
+    fn source(
+        &'a self,
+        id: Self::FileId,
+    ) -> Result<Self::Source, codespan_reporting::files::Error> {
+        Ok(self.source(id).input())
+    }
+
+    fn line_index(
+        &'a self,
+        id: Self::FileId,
+        byte_index: usize,
+    ) -> Result<usize, codespan_reporting::files::Error> {
+        let file = &self.files[id.0];
+        Ok(file
+            .line_starts
+            .binary_search(&byte_index)
+            .unwrap_or_else(|next_line| next_line - 1))
+    }
+
+    fn line_range(
+        &'a self,
+        id: Self::FileId,
+        line_index: usize,
+    ) -> Result<Range<usize>, codespan_reporting::files::Error> {
+        let file = &self.files[id.0];
+        let line_start = file.line_start(line_index)?;
+        let next_line_start = file.line_start(line_index + 1)?;
+
+        Ok(line_start..next_line_start)
+    }
+}
+
 pub struct TomlError {
     pub message: String,
     pub span: Option<Range<usize>>,
@@ -135,7 +209,11 @@ pub fn toml_error_to_diagnostic(error: TomlError) -> Diagnostic<FileId> {
     }
 }
 
-pub fn report_diagnostics(files: &Files, diagnostics: &[Diagnostic<FileId>]) -> anyhow::Result<()> {
+#[instrument(skip(files, diagnostics))]
+pub fn report_diagnostics(
+    files: &Treehouse,
+    diagnostics: &[Diagnostic<FileId>],
+) -> anyhow::Result<()> {
     let writer = StandardStream::stderr(ColorChoice::Auto);
     let config = codespan_reporting::term::Config::default();
     for diagnostic in diagnostics {