Merge pull request #59 from sonesuke/feat/lsp-realtime-diagnostics

sonesuke · web-flow · commit 0b1f3af632a1 · 2026-02-03T22:34:20.000+09:00
feat(lsp): implement real-time diagnostics and decouple core
diff --git a/.devcontainer/Dockerfile b/.devcontainer/Dockerfile
@@ -35,6 +35,7 @@ COPY Cargo.toml Cargo.lock ./
 RUN echo "fn main() {}" > src/main.rs \
     && echo "fn main() {}" > benches/performance.rs \
     && cargo build --release \
+    && cargo build \
     && cd / && rm -rf /tmp/build
 
 # Install cargo-binstall as vscode
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
@@ -11,6 +11,7 @@ env:
 
 permissions:
   contents: read
+  packages: read
 
 jobs:
   lint:
@@ -19,11 +20,18 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
+      - name: Log in to GitHub Container Registry
+        uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
       - name: Build and Run Dev Container (Lint)
         uses: devcontainers/ci@v0.3
         with:
-          imageName: docgraph-dev
-          cacheFrom: docgraph-dev
+          imageName: ghcr.io/${{ github.repository }}/docgraph-dev
+          cacheFrom: ghcr.io/${{ github.repository }}/docgraph-dev
           push: never
           runCmd: |
             cargo fmt -- --check
@@ -38,11 +46,18 @@ jobs:
     steps:
       - uses: actions/checkout@v4
 
+      - name: Log in to GitHub Container Registry
+        uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
       - name: Build and Run Dev Container (Test)
         uses: devcontainers/ci@v0.3
         with:
-          imageName: docgraph-dev
-          cacheFrom: docgraph-dev
+          imageName: ghcr.io/${{ github.repository }}/docgraph-dev
+          cacheFrom: ghcr.io/${{ github.repository }}/docgraph-dev
           push: never
           runCmd: |
             cargo llvm-cov --all-features --workspace --lcov --output-path lcov.info
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -16,6 +16,7 @@ thiserror = "2.0"
 toml = "0.9.11"
 tower-lsp = "0.20.0"
 tokio = { version = "1.37.0", features = ["full"] }
+dashmap = "6.1.0"
 
 [dev-dependencies]
 tempfile = "3.24.0"
diff --git a/src/cli/handlers/check.rs b/src/cli/handlers/check.rs
@@ -22,7 +22,7 @@ fn try_check(
 ) -> anyhow::Result<ExitCode> {
     let config = config::Config::load(&path).context("failed to load docgraph.toml")?;
 
-    let diagnostics = lint::check_workspace(&path, fix, rule, true, &config);
+    let diagnostics = lint::check_workspace(&path, fix, rule, true, &config, None);
 
     if json {
         let json_out = serde_json::to_string_pretty(&diagnostics)
@@ -89,7 +89,7 @@ pub fn handle_fmt(path: PathBuf, rule: Option<Vec<String>>) -> ExitCode {
 
 fn try_fmt(path: PathBuf, rule: Option<Vec<String>>) -> anyhow::Result<ExitCode> {
     let config = config::Config::load(&path).context("failed to load docgraph.toml")?;
-    let diagnostics = lint::check_workspace(&path, true, rule, false, &config);
+    let diagnostics = lint::check_workspace(&path, true, rule, false, &config, None);
     print_diagnostics(&diagnostics);
     Ok(ExitCode::SUCCESS)
 }
diff --git a/src/cli/handlers/describe.rs b/src/cli/handlers/describe.rs
@@ -15,7 +15,7 @@ pub fn handle_describe(id: String, path: PathBuf) -> ExitCode {
 
 fn try_describe(id: String, path: PathBuf) -> anyhow::Result<ExitCode> {
     let config = config::Config::load(&path).context("failed to load docgraph.toml")?;
-    let (blocks, _refs) = collect::collect_workspace_all(&path, &config.graph.ignore);
+    let (blocks, _refs) = collect::collect_workspace_all(&path, &config.graph.ignore, None);
 
     let target_block = blocks
         .iter()
diff --git a/src/cli/handlers/graph.rs b/src/cli/handlers/graph.rs
@@ -15,7 +15,7 @@ pub fn handle_graph(path: PathBuf) -> ExitCode {
 
 fn try_graph(path: PathBuf) -> anyhow::Result<ExitCode> {
     let config = config::Config::load(&path).context("failed to load docgraph.toml")?;
-    let (blocks, _refs) = collect::collect_workspace_all(&path, &config.graph.ignore);
+    let (blocks, _refs) = collect::collect_workspace_all(&path, &config.graph.ignore, None);
     let json_out =
         serde_json::to_string_pretty(&blocks).context("failed to serialize graph to JSON")?;
     println!("{}", json_out);
diff --git a/src/cli/handlers/list.rs b/src/cli/handlers/list.rs
@@ -16,7 +16,7 @@ pub fn handle_list(query: String, path: PathBuf) -> ExitCode {
 
 fn try_list(query: String, path: PathBuf) -> anyhow::Result<ExitCode> {
     let config = config::Config::load(&path).context("failed to load docgraph.toml")?;
-    let (blocks, _refs) = collect::collect_workspace_all(&path, &config.graph.ignore);
+    let (blocks, _refs) = collect::collect_workspace_all(&path, &config.graph.ignore, None);
     let regex_str = glob_to_regex(&query);
     let re = regex::Regex::new(&regex_str)
         .with_context(|| format!("Invalid query pattern: '{}'", query))?;
diff --git a/src/cli/handlers/trace.rs b/src/cli/handlers/trace.rs
@@ -21,7 +21,7 @@ fn try_trace(
     direction: String,
 ) -> anyhow::Result<ExitCode> {
     let config = config::Config::load(&path).context("failed to load docgraph.toml")?;
-    let (blocks, _refs) = collect::collect_workspace_all(&path, &config.graph.ignore);
+    let (blocks, _refs) = collect::collect_workspace_all(&path, &config.graph.ignore, None);
     let target_regex_str = glob_to_regex(&to);
     let target_re = regex::Regex::new(&target_regex_str)
         .with_context(|| format!("Invalid target pattern: '{}'", to))?;
diff --git a/src/core/collect.rs b/src/core/collect.rs
@@ -1,20 +1,33 @@
 use crate::core::parse::extract_all;
 use crate::core::types::{RefUse, SpecBlock};
 use crate::core::walk::find_markdown_files;
+use std::collections::HashMap;
 use std::fs;
-use std::path::Path;
+use std::path::{Path, PathBuf};
 
 /// Collect all SpecBlocks and standalone RefUses from the workspace
 pub fn collect_workspace_all(
     root: &Path,
     ignore_patterns: &[String],
+    overrides: Option<&HashMap<PathBuf, String>>,
 ) -> (Vec<SpecBlock>, Vec<RefUse>) {
     let files = find_markdown_files(root, ignore_patterns);
     let mut all_blocks = Vec::new();
     let mut all_refs = Vec::new();
 
     for file_path in files {
-        match fs::read_to_string(&file_path) {
+        // Canonicalize the path for lookup to match the keys in overrides
+        let lookup_path = fs::canonicalize(&file_path).unwrap_or_else(|_| file_path.clone());
+
+        let content_result = if let Some(map) = overrides
+            && let Some(content) = map.get(lookup_path.as_path())
+        {
+            Ok(content.clone())
+        } else {
+            fs::read_to_string(&file_path)
+        };
+
+        match content_result {
             Ok(content) => {
                 let (blocks, refs) = extract_all(&content, &file_path);
                 all_blocks.extend(blocks);
@@ -40,7 +53,7 @@ mod tests {
         let mut file = File::create(&file_path).unwrap();
         writeln!(file, "<a id=\"ID-1\"></a>\n# Heading\n[Ref](#ID-2)").unwrap();
 
-        let (blocks, refs) = collect_workspace_all(dir.path(), &[]);
+        let (blocks, refs) = collect_workspace_all(dir.path(), &[], None);
 
         assert_eq!(blocks.len(), 1);
         assert_eq!(blocks[0].id, "ID-1");
diff --git a/src/core/lint.rs b/src/core/lint.rs
@@ -2,8 +2,9 @@ use crate::core::types::Diagnostic;
 use rumdl_lib::fix_coordinator::FixCoordinator;
 use rumdl_lib::workspace_index::WorkspaceIndex;
 
+use std::collections::HashMap;
 use std::fs;
-use std::path::Path;
+use std::path::{Path, PathBuf};
 
 /// Check if a rumdl warning should be skipped.
 /// - MD013 (line length): Always skip
@@ -24,6 +25,7 @@ pub fn check_workspace(
     rule_filter: Option<Vec<String>>,
     use_docgraph_filter: bool,
     config: &crate::core::config::Config,
+    overrides: Option<&HashMap<PathBuf, String>>,
 ) -> Vec<Diagnostic> {
     let mut diagnostics = Vec::new();
     let files = crate::core::walk::find_markdown_files(path, &config.graph.ignore);
@@ -60,7 +62,18 @@ pub fn check_workspace(
 
     // Pass 1: Lint individual files and build index
     for file_path in &files {
-        match fs::read_to_string::<&Path>(file_path) {
+        // Canonicalize the path for lookup to match the keys in overrides
+        let lookup_path = fs::canonicalize(file_path).unwrap_or_else(|_| file_path.clone());
+
+        let content_result = if let Some(map) = overrides
+            && let Some(content) = map.get(lookup_path.as_path())
+        {
+            Ok(content.clone())
+        } else {
+            fs::read_to_string::<&Path>(file_path)
+        };
+
+        match content_result {
             Ok(content) => {
                 let mut working_content = content.clone();
 
@@ -210,7 +223,7 @@ pub fn check_workspace(
     // Pass 3: Run custom docgraph workspace-level checks (DG005, DG006, DG004)
     // Collect docgraph's own SpecBlock data
     let (spec_blocks, _refs) =
-        crate::core::collect::collect_workspace_all(path, &config.graph.ignore);
+        crate::core::collect::collect_workspace_all(path, &config.graph.ignore, overrides);
 
     // DG005: Strict Node Types
     let dg005_diags = crate::core::rules::dg005::check_strict_node_types(&spec_blocks, config);
@@ -348,7 +361,7 @@ Text instead of heading"#;
 
         let config = Config::default();
         // Enable fix=true, use_docgraph_filter=false (to ensure we run standard rules)
-        let _diagnostics = check_workspace(dir.path(), true, None, false, &config);
+        let _diagnostics = check_workspace(dir.path(), true, None, false, &config, None);
 
         // Content should be updated
         let new_content = std::fs::read_to_string(&f1).unwrap();
@@ -371,6 +384,7 @@ Text instead of heading"#;
             Some(vec!["MD033".to_string()]),
             false,
             &config,
+            None,
         );
         assert!(diagnostics.iter().any(|d| d.code == "MD033"));
 
@@ -381,6 +395,7 @@ Text instead of heading"#;
             Some(vec!["MD001".to_string()]),
             false,
             &config,
+            None,
         );
         assert!(!diagnostics_empty.iter().any(|d| d.code == "MD033"));
     }
@@ -408,7 +423,7 @@ Text instead of heading"#;
             // We need rules that trigger a fix. MD047 adds a newline.
             // Rumdl default rules include MD047.
             // use_docgraph_filter=false ensures standard rules run
-            let diags = check_workspace(dir.path(), true, None, false, &config);
+            let diags = check_workspace(dir.path(), true, None, false, &config, None);
 
             // Expect DG000 error for write failure
             assert!(
diff --git a/src/lsp/backend.rs b/src/lsp/backend.rs
@@ -1,3 +1,4 @@
+use dashmap::DashMap;
 use std::path::PathBuf;
 use std::sync::Arc;
 use tokio::sync::Mutex;
@@ -13,6 +14,7 @@ pub struct Backend {
     pub workspace_root: Arc<Mutex<Option<PathBuf>>>,
     pub blocks: Arc<Mutex<Vec<types::SpecBlock>>>,
     pub standalone_refs: Arc<Mutex<Vec<types::RefUse>>>,
+    pub documents: Arc<DashMap<Url, String>>,
 }
 
 impl Backend {
@@ -22,17 +24,36 @@ impl Backend {
             workspace_root: Arc::new(Mutex::new(None)),
             blocks: Arc::new(Mutex::new(Vec::new())),
             standalone_refs: Arc::new(Mutex::new(Vec::new())),
+            documents: Arc::new(DashMap::new()),
         }
     }
 
     pub async fn run_lint(&self) {
         let root_opt = self.workspace_root.lock().await.clone();
         if let Some(root) = root_opt {
             let config = config::Config::load(&root).unwrap_or_else(|_| config::Config::default());
-            let diagnostics = lint::check_workspace(&root, false, None, true, &config);
+
+            // Create overrides map (convert DashMap<Url, String> to HashMap<PathBuf, String>)
+            // This decouples core from lsp types
+            let mut overrides = std::collections::HashMap::new();
+            for entry in self.documents.iter() {
+                if let Ok(path) = entry.key().to_file_path() {
+                    // Try to canonicalize the path for consistent lookup
+                    if let Ok(canon_path) = std::fs::canonicalize(&path) {
+                        overrides.insert(canon_path, entry.value().clone());
+                    } else {
+                        overrides.insert(path, entry.value().clone());
+                    }
+                }
+            }
+
+            // Use documents map for overrides (pass reference to HashMap)
+            let diagnostics =
+                lint::check_workspace(&root, false, None, true, &config, Some(&overrides));
 
             // Update index
-            let (blocks, refs) = collect::collect_workspace_all(&root, &config.graph.ignore);
+            let (blocks, refs) =
+                collect::collect_workspace_all(&root, &config.graph.ignore, Some(&overrides));
             {
                 let mut b = self.blocks.lock().await;
                 *b = blocks;
@@ -44,6 +65,21 @@ impl Backend {
             let mut file_diagnostics: std::collections::HashMap<PathBuf, Vec<Diagnostic>> =
                 std::collections::HashMap::new();
 
+            // Initialize with all workspace files + open files to ensure we clear diagnostics for everything
+            let workspace_files =
+                crate::core::walk::find_markdown_files(&root, &config.graph.ignore);
+            for path in workspace_files {
+                file_diagnostics.entry(path).or_default();
+            }
+
+            // Also ensure open files are included (though they should be in workspace_files if on disk)
+            // But if they are new untitled files or outside root? (LSP usually handles workspace files)
+            for entry in self.documents.iter() {
+                if let Ok(path) = entry.key().to_file_path() {
+                    file_diagnostics.entry(path).or_default();
+                }
+            }
+
             for d in diagnostics {
                 let diag = Diagnostic {
                     range: Range {
@@ -129,11 +165,18 @@ impl LanguageServer for Backend {
         Ok(())
     }
 
-    async fn did_open(&self, _: DidOpenTextDocumentParams) {
+    async fn did_open(&self, params: DidOpenTextDocumentParams) {
+        self.documents
+            .insert(params.text_document.uri.clone(), params.text_document.text);
         self.run_lint().await;
     }
 
-    async fn did_change(&self, _: DidChangeTextDocumentParams) {
+    async fn did_change(&self, params: DidChangeTextDocumentParams) {
+        // We assume full text sync (TextDocumentSyncKind::FULL)
+        if let Some(change) = params.content_changes.first() {
+            self.documents
+                .insert(params.text_document.uri.clone(), change.text.clone());
+        }
         self.run_lint().await;
     }
 
@@ -142,6 +185,7 @@ impl LanguageServer for Backend {
     }
 
     async fn did_close(&self, params: DidCloseTextDocumentParams) {
+        self.documents.remove(&params.text_document.uri);
         self.client
             .publish_diagnostics(params.text_document.uri, vec![], None)
             .await;
