fix: improve path completion in Markdown files

org.eclipse.wildwebdeveloper.tests/src/org/eclipse/wildwebdeveloper/tests/TestMarkdown.java

@@ -18,6 +18,7 @@
 
 import java.nio.charset.StandardCharsets;
 import java.util.ArrayList;
+import java.util.Arrays;
 import java.util.Collections;
 import java.util.concurrent.atomic.AtomicReference;
 import java.util.stream.Collectors;
@@ -27,13 +28,17 @@
 import org.eclipse.core.resources.IResource;
 import org.eclipse.core.resources.ResourcesPlugin;
 import org.eclipse.core.runtime.CoreException;
+import org.eclipse.jface.text.IDocument;
+import org.eclipse.jface.text.contentassist.ICompletionProposal;
 import org.eclipse.lsp4e.LSPEclipseUtils;
 import org.eclipse.lsp4e.LanguageServerWrapper;
 import org.eclipse.lsp4e.LanguageServiceAccessor;
+import org.eclipse.lsp4e.operations.completion.LSContentAssistProcessor;
 import org.eclipse.ui.PlatformUI;
 import org.eclipse.ui.editors.text.TextEditor;
 import org.eclipse.ui.ide.IDE;
 import org.eclipse.ui.tests.harness.util.DisplayHelper;
+import org.eclipse.wildwebdeveloper.Activator;
 import org.junit.jupiter.api.Test;
 import org.junit.jupiter.api.extension.ExtendWith;
 
@@ -123,4 +128,61 @@ void diagnosticsCoverTypicalMarkdownIssues() throws Exception {
 
 		assertTrue(markerTests.isEmpty(), "The following markers were not found: " + markerTests);
 	}
+
+	@Test
+	void workspaceHeaderCompletionsRespectExcludeGlobs() throws Exception {
+		var project = ResourcesPlugin.getWorkspace().getRoot().getProject(getClass().getName() + ".hdr" + System.nanoTime());
+		project.create(null);
+		project.open(null);
+
+		// Configure exclusion: exclude docs/generated/** from workspace header completions
+		Activator.getDefault().getPreferenceStore().setValue("markdown.suggest.paths.excludeGlobs", "docs/generated/**");
+
+		// Create markdown files with unique headers
+		// Ensure folders exist
+		var docsFolder = project.getFolder("docs");
+		if (!docsFolder.exists())
+			docsFolder.create(true, true, null);
+		var genFolder = docsFolder.getFolder("generated");
+		if (!genFolder.exists())
+			genFolder.create(true, true, null);
+
+		IFile excluded = project.getFile("docs/generated/excluded.md");
+		excluded.create("# Excluded Only\n".getBytes(StandardCharsets.UTF_8), true, false, null);
+
+		IFile included = project.getFile("docs/included.md");
+		included.create("# Included Only\n".getBytes(StandardCharsets.UTF_8), true, false, null);
+
+		// File where we'll trigger completions (double hash to respect default preference)
+		IFile index = project.getFile("index.md");
+		index.create("[](##)\n".getBytes(StandardCharsets.UTF_8), true, false, null);
+
+		var editor = (TextEditor) IDE.openEditor(PlatformUI.getWorkbench().getActiveWorkbenchWindow().getActivePage(), index);
+		var display = editor.getSite().getShell().getDisplay();
+		IDocument document = editor.getDocumentProvider().getDocument(editor.getEditorInput());
+
+		// Ensure Markdown Language Server is started and connected
+		var markdownLS = new AtomicReference<LanguageServerWrapper>();
+		assertTrue(DisplayHelper.waitForCondition(display, 10_000, () -> {
+			markdownLS.set(LanguageServiceAccessor.getStartedWrappers(document, null, false).stream() //
+					.filter(w -> "org.eclipse.wildwebdeveloper.markdown".equals(w.serverDefinition.id)) //
+					.findFirst().orElse(null));
+			return markdownLS.get() != null //
+					&& markdownLS.get().isActive() //
+					&& markdownLS.get().isConnectedTo(LSPEclipseUtils.toUri(document));
+		}), "Markdown LS did not start");
+
+		// Trigger content assist at the end of '##'
+		int offset = document.get().indexOf("##") + 2;
+		var cap = new LSContentAssistProcessor();
+
+		assertTrue(DisplayHelper.waitForCondition(display, 15_000, () -> {
+			ICompletionProposal[] proposals = cap.computeCompletionProposals(Utils.getViewer(editor), offset);
+			if (proposals == null || proposals.length == 0)
+				return false;
+			boolean hasIncluded = Arrays.stream(proposals).anyMatch(p -> "#included-only".equals(p.getDisplayString()));
+			boolean hasExcluded = Arrays.stream(proposals).anyMatch(p -> "#excluded-only".equals(p.getDisplayString()));
+			return hasIncluded && !hasExcluded;
+		}), "Workspace header completions did not respect exclude globs");
+	}
 }

org.eclipse.wildwebdeveloper/META-INF/MANIFEST.MF

@@ -35,7 +35,8 @@ Require-Bundle: org.eclipse.ui,
 Bundle-RequiredExecutionEnvironment: JavaSE-21
 Bundle-ActivationPolicy: lazy
 Eclipse-BundleShape: dir
-Export-Package: org.eclipse.wildwebdeveloper.debug;x-internal:=true,
+Export-Package: org.eclipse.wildwebdeveloper;x-friends:="org.eclipse.wildwebdeveloper.tests",
+ org.eclipse.wildwebdeveloper.debug;x-internal:=true,
  org.eclipse.wildwebdeveloper.debug.node;x-internal:=true,
  org.eclipse.wildwebdeveloper.debug.npm;x-internal:=true,
  org.eclipse.wildwebdeveloper.markdown;x-friends:="org.eclipse.wildwebdeveloper.tests"

org.eclipse.wildwebdeveloper/src/org/eclipse/wildwebdeveloper/markdown/MarkdownLanguageClient.java

@@ -18,11 +18,12 @@
 import java.net.URISyntaxException;
 import java.net.URL;
 import java.nio.charset.StandardCharsets;
+import java.nio.file.FileSystems;
 import java.nio.file.Files;
 import java.nio.file.Path;
+import java.nio.file.PathMatcher;
 import java.nio.file.Paths;
 import java.util.ArrayList;
-import java.util.Collections;
 import java.util.HashMap;
 import java.util.List;
 import java.util.Map;
@@ -38,6 +39,7 @@
 import org.eclipse.core.resources.ResourcesPlugin;
 import org.eclipse.core.runtime.FileLocator;
 import org.eclipse.core.runtime.ILog;
+import org.eclipse.core.runtime.IPath;
 import org.eclipse.lsp4e.LSPEclipseUtils;
 import org.eclipse.lsp4e.client.DefaultLanguageClient;
 import org.eclipse.lsp4j.ConfigurationItem;
@@ -140,7 +142,7 @@ public void resourceChanged(final IResourceChangeEvent event) {
 					// Notify server of the file system event for this resource
 					final var payload = new HashMap<String, Object>();
 					payload.put("id", Integer.valueOf(id));
-					payload.put("uri", uri.toString());
+					payload.put("uri", normalizeFileUriForLanguageServer(uri));
 					payload.put("kind", kind);
 					server.fsWatcherOnChange(payload);
 					return true;
@@ -212,7 +214,7 @@ public CompletableFuture<List<Map<String, Object>>> parseMarkdown(final Map<Stri
 					}
 				}
 				if (!hasText && uriPath == null) {
-					return Collections.emptyList();
+					return List.of();
 				}
 				final String argPath;
 				if (hasText) {
@@ -229,7 +231,7 @@ public CompletableFuture<List<Map<String, Object>>> parseMarkdown(final Map<Stri
 				if (exit != 0) {
 					final var err = new String(proc.getErrorStream().readAllBytes(), StandardCharsets.UTF_8);
 					ILog.get().warn("markdown-it parser failed (" + exit + "): " + err, null);
-					return Collections.emptyList();
+					return List.of();
 				}
 				final var gson = new Gson();
 				final List<Map<String, Object>> tokens = gson.fromJson(out, List.class);
@@ -242,7 +244,7 @@ public CompletableFuture<List<Map<String, Object>>> parseMarkdown(final Map<Stri
 					}
 				} catch (final Exception ignore) {
 				}
-				return tokens != null ? tokens : Collections.emptyList();
+				return tokens != null ? tokens : List.of();
 			} catch (final InterruptedException ex) {
 				ILog.get().warn(ex.getMessage(), ex);
 				/* Clean up whatever needs to be handled before interrupting  */
@@ -256,7 +258,7 @@ public CompletableFuture<List<Map<String, Object>>> parseMarkdown(final Map<Stri
 					} catch (final Exception ignore) {
 					}
 			}
-			return Collections.emptyList();
+			return List.of();
 		});
 	}
 
@@ -270,19 +272,27 @@ public CompletableFuture<List<Map<String, Object>>> parseMarkdown(final Map<Stri
 	public CompletableFuture<List<String>> findMarkdownFilesInWorkspace(final Object unused) {
 		return CompletableFuture.supplyAsync(() -> {
 			final var uris = new ArrayList<String>();
+			// Compile exclude globs from preferences once per request
+			final String[] excludeGlobs = MarkdownPreferences.getSuggestPathsExcludeGlobs();
+			final List<PathMatcher> excludeMatchers = compileGlobMatchers(excludeGlobs);
 			try {
 				final var roots = MarkdownLanguageServer.getServerRoots();
 				if (roots != null && !roots.isEmpty()) {
 					for (String rootUri : roots) {
 						final var containers = ResourcesPlugin.getWorkspace().getRoot()
-								.findContainersForLocationURI(new java.net.URI(rootUri));
+								.findContainersForLocationURI(URI.create(rootUri));
 						if (containers != null && containers.length > 0) {
 							for (final var container : containers) {
+								if (container.isDerived() || container.isHidden())
+									continue;
 								container.accept((final IResource res) -> {
+									if (res.isDerived() || res.isHidden())
+										return false;
 									if (res.getType() == IResource.FILE) {
 										final String name = res.getName().toLowerCase();
-										if (name.endsWith(".md") || name.endsWith(".markdown") || name.endsWith(".mdown")) {
-											uris.add(res.getLocationURI().toString());
+										if ((name.endsWith(".md") || name.endsWith(".markdown") || name.endsWith(".mdown"))
+												&& !isExcludedByGlobs(res, excludeMatchers)) {
+											uris.add(normalizeFileUriForLanguageServer(res.getLocationURI()));
 										}
 										return false; // no children
 									}
@@ -295,10 +305,13 @@ public CompletableFuture<List<String>> findMarkdownFilesInWorkspace(final Object
 					// Fallback: scan entire workspace
 					final IWorkspaceRoot wsRoot = ResourcesPlugin.getWorkspace().getRoot();
 					wsRoot.accept((final IResource res) -> {
+						if (res.isDerived() || res.isHidden())
+							return false;
 						if (res.getType() == IResource.FILE) {
 							final String name = res.getName().toLowerCase();
-							if (name.endsWith(".md") || name.endsWith(".markdown") || name.endsWith(".mdown")) {
-								uris.add(res.getLocationURI().toString());
+							if ((name.endsWith(".md") || name.endsWith(".markdown") || name.endsWith(".mdown"))
+									&& !isExcludedByGlobs(res, excludeMatchers)) {
+								uris.add(normalizeFileUriForLanguageServer(res.getLocationURI()));
 							}
 							return false; // no children
 						}
@@ -312,6 +325,48 @@ public CompletableFuture<List<String>> findMarkdownFilesInWorkspace(final Object
 		});
 	}
 
+	private static List<PathMatcher> compileGlobMatchers(String... globs) {
+		if (globs == null || globs.length == 0)
+			return List.of();
+
+		var fs = FileSystems.getDefault();
+		var matchers = new ArrayList<PathMatcher>();
+
+		for (String glob : globs) {
+			if (glob == null || (glob = glob.trim()).isEmpty())
+				continue;
+
+			// If pattern starts with "**/", also add a root-level variant without it.
+			// This makes "**/node_modules/**" also match "node_modules/**".
+			List<String> patterns = glob.startsWith("**/") && glob.length() > 3
+					? List.of(glob, glob.substring(3))
+					: List.of(glob);
+
+			for (String pattern : patterns) {
+				try {
+					matchers.add(fs.getPathMatcher("glob:" + pattern));
+				} catch (Exception ex) {
+					ILog.get().warn(ex.getMessage(), ex);
+				}
+			}
+		}
+		return matchers;
+	}
+
+	private static boolean isExcludedByGlobs(final IResource res, final List<PathMatcher> matchers) {
+		if (matchers == null || matchers.isEmpty())
+			return false;
+		final IPath pr = res.getProjectRelativePath();
+		if (pr == null)
+			return false;
+		final Path p = pr.toPath();
+		for (final PathMatcher m : matchers) {
+			if (m.matches(p))
+				return true;
+		}
+		return false;
+	}
+
 	/**
 	 * <pre>
 	 * Request: { uri: string }
@@ -409,10 +464,10 @@ public CompletableFuture<Void> fsWatcherCreate(final Map<String, Object> params)
 			@SuppressWarnings("unchecked")
 			final Map<String, Object> options = params.get("options") instanceof Map //
 					? (Map<String, Object>) params.get("options")
-					: Collections.emptyMap();
-			final var watcher = new Watcher((MarkdownLanguageServerAPI) getLanguageServer(), id, path,
+					: Map.of();
+			final var watcher = new Watcher((MarkdownLanguageServerAPI) getLanguageServer(), id, path, //
 					Boolean.TRUE.equals(options.get("ignoreCreate")), //
-					Boolean.TRUE.equals(options.get("ignoreChange")),
+					Boolean.TRUE.equals(options.get("ignoreChange")), //
 					Boolean.TRUE.equals(options.get("ignoreDelete")));
 			ResourcesPlugin.getWorkspace().addResourceChangeListener(watcher, IResourceChangeEvent.POST_CHANGE);
 			watchersById.put(Integer.valueOf(id), watcher);
@@ -439,4 +494,35 @@ public CompletableFuture<Void> fsWatcherDelete(final Map<String, Object> params)
 			return null;
 		});
 	}
+
+	/**
+	 * Normalize Windows file URIs to match vscode-uri's URI.toString() form used by the server.
+	 * <br>
+	 * Examples:
+	 * <li>file:/D:/path -> file:///d%3A/path
+	 * <li>file:///D:/path -> file:///d%3A/path
+	 */
+	private static String normalizeFileUriForLanguageServer(final URI uri) {
+		if (uri == null)
+			return null;
+		if (!FileUtils.FILE_SCHEME.equalsIgnoreCase(uri.getScheme()))
+			return uri.toString();
+		final String uriAsString = uri.toString();
+
+		// Ensure triple slash prefix
+		String withoutScheme = uriAsString.substring("file:".length()); // could be :/, :///
+		while (withoutScheme.startsWith("/")) {
+			withoutScheme = withoutScheme.substring(1);
+		}
+
+		// Expect leading like D:/ or d:/ on Windows
+		if (withoutScheme.length() >= 2 && Character.isLetter(withoutScheme.charAt(0)) && withoutScheme.charAt(1) == ':') {
+			final char drive = Character.toLowerCase(withoutScheme.charAt(0));
+			final String rest = withoutScheme.substring(2); // drop ':'
+			return "file:///" + drive + "%3A" + (rest.startsWith("/") ? rest : "/" + rest);
+		}
+
+		// Already in a normalized or UNC form; fall back to original
+		return uriAsString;
+	}
 }

org.eclipse.wildwebdeveloper/src/org/eclipse/wildwebdeveloper/markdown/md-lsp-proxy.js

@@ -31,6 +31,14 @@
 //             server's `URI.toString()` form. This guarantees the server tracks the open document under
 //             the key it uses during diagnostics, enabling `computeDiagnostics()` and `markdown/fs/*` requests.
 //
+// 3) Windows path suggestions (client→server) producing incorrect absolute drive paths:
+//    Problem: When resolving workspace header/path completions, the current document URI and target
+//             document URIs sometimes differ in Windows drive case/encoding (e.g., `file:///D:/...` vs
+//             `file:///d%3A/...`). This causes relative path computation to fall back to odd absolute
+//             paths such as `../../../../d:/...`.
+//    Fix:     Normalize the document URI in `textDocument/completion` requests so that the server sees a
+//             consistent Windows-encoded form and computes clean relative paths (e.g., `GUIDE.md#...`).
+//
 //    Note: On non-Windows URIs, normalization is a no-op; messages are forwarded unchanged.
 
 const { spawn } = require('child_process');
@@ -45,7 +53,8 @@ const serverArgs = process.argv.slice(3);
 // Launch the wrapped language server; we proxy its stdio
 const child = spawn(process.execPath, [serverMain, ...serverArgs], { stdio: ['pipe', 'pipe', 'inherit'] });
 
-// Client → Server (Problem 2): normalize Windows URIs and mirror lifecycle notifications
+// Client → Server (Problem 2 & 3): normalize Windows URIs, mirror lifecycle notifications,
+// and normalize completion requests
 let inBuffer = Buffer.alloc(0);
 process.stdin.on('data', chunk => {
   inBuffer = Buffer.concat([inBuffer, chunk]);
@@ -123,15 +132,14 @@ function processInboundBuffer() {
   }
 }
 
-// Duplicate lifecycle notifications with a normalized URI so diagnostics can run (Problem 2)
+// Client → Server normalization (Problem 2 & 3)
 function transformInbound(bodyBuf) {
   try {
     const text = bodyBuf.toString('utf8');
     const msg = JSON.parse(text);
     const method = msg && msg.method;
 
-    // For text document lifecycle events, also send a duplicate event
-    // with a normalized file URI so the server's URI.toString() lookups match.
+    // Duplicate lifecycle notifications with a normalized URI (Problem 2)
     if ((method === 'textDocument/didOpen' || method === 'textDocument/didChange' || method === 'textDocument/didClose' || method === 'textDocument/willSave' || method === 'textDocument/didSave') && msg.params && msg.params.textDocument) {
       const origUri = msg.params.textDocument.uri;
       const normUri = normalizeFileUriForServer(origUri);
@@ -142,22 +150,35 @@ function transformInbound(bodyBuf) {
         return [Buffer.from(JSON.stringify(msg), 'utf8'), Buffer.from(JSON.stringify(dup), 'utf8')];
       }
     }
+
+    // Normalize completion requests so relative path suggestions are correct on Windows (Problem 3)
+    if (method === 'textDocument/completion' && msg.params && msg.params.textDocument) {
+      const origUri = msg.params.textDocument.uri;
+      const normUri = normalizeFileUriForServer(origUri);
+      if (normUri && normUri !== origUri) {
+        const req = structuredClone(msg);
+        req.params.textDocument.uri = normUri;
+        return [Buffer.from(JSON.stringify(req), 'utf8')];
+      }
+    }
+
   } catch { }
   return [bodyBuf];
 }
 
 // Normalize Windows file URIs to match vscode-uri’s URI.toString() (Problem 2)
 function normalizeFileUriForServer(uri) {
-  if (typeof uri !== 'string' || !uri.startsWith('file:///'))
-    return undefined;
-
-  // Normalize Windows drive letter and encode colon to match vscode-uri toString()
-  // Example: file:///D:/path -> file:///d%3A/path
-  const after = uri.slice('file:///'.length);
+  if (typeof uri !== 'string' || !uri.startsWith('file:')) return undefined;
+  // Strip scheme and leading slashes to get to drive letter
+  let after = uri.slice('file:'.length);
+  while (after.startsWith('/')) after = after.slice(1);
+  // Example accepted forms: D:/path or d:/path
   if (/^[A-Za-z]:/.test(after)) {
     const drive = after[0].toLowerCase();
-    const rest = after.slice(2); // drop ":"
-    return 'file:///' + drive + '%3A' + rest;
+    const rest = after.slice(2); // drop ':'
+    // Ensure a leading slash for the path segment
+    const pathPart = rest.startsWith('/') ? rest : '/' + rest;
+    return 'file:///' + drive + '%3A' + pathPart;
   }
   return undefined;
 }