Fix unicode normalization in file suggestions

src/gui/suggesters/FileIndex.test.ts

@@ -5,17 +5,17 @@ import type { App, TFile, Vault, MetadataCache, Workspace } from 'obsidian';
 // Test-specific subclass that allows resetting the singleton
 class TestableFileIndex extends FileIndex {
 	static reset(): void {
-		if (TestableFileIndex.instance) {
+		if (FileIndex.instance) {
 			// Clear any pending timeouts
-			if ((TestableFileIndex.instance as any).reindexTimeout !== null) {
-				clearTimeout((TestableFileIndex.instance as any).reindexTimeout);
+			if ((FileIndex.instance as any).reindexTimeout !== null) {
+				clearTimeout((FileIndex.instance as any).reindexTimeout);
 			}
-			if ((TestableFileIndex.instance as any).fuseUpdateTimeout !== null) {
-				clearTimeout((TestableFileIndex.instance as any).fuseUpdateTimeout);
+			if ((FileIndex.instance as any).fuseUpdateTimeout !== null) {
+				clearTimeout((FileIndex.instance as any).fuseUpdateTimeout);
 			}
-			// Clear the instance
-			TestableFileIndex.instance = null as any;
 		}
+		// Clear the instance
+		FileIndex.instance = null as any;
 	}
 }
 
@@ -189,9 +189,10 @@ describe('FileIndex', () => {
 			}));
 
 			// Initial index – ensure all batched timers run so both files are indexed
-			await freshIndex.ensureIndexed();
+			const indexPromise = freshIndex.ensureIndexed();
 			// Flush any pending timers (including 0-ms ones) used inside performReindex()
 			await vi.runAllTimersAsync();
+			await indexPromise;
 			expect(freshIndex.getIndexedFileCount()).toBeGreaterThanOrEqual(1);
 
 			// Spy on the methods - use proper type assertion
@@ -340,6 +341,28 @@ describe('FileIndex', () => {
 	});
 
 	describe('search functionality', () => {
+		it('should match normalized query against decomposed filenames', async () => {
+			const nfcName = 'Rücken-Fit';
+			const nfdName = nfcName.normalize('NFD');
+			const files = [
+				{
+					path: `${nfdName}.md`,
+					basename: nfdName,
+					extension: 'md',
+					parent: { path: '' },
+					stat: { mtime: Date.now() }
+				}
+			] as TFile[];
+
+			(mockApp.vault.getMarkdownFiles as any).mockReturnValue(files);
+			mockApp.metadataCache.getFileCache = vi.fn(() => ({}));
+
+			await fileIndex.ensureIndexed();
+			const results = fileIndex.search('Rü', {}, 10);
+
+			expect(results.some(result => result.file.basename === nfdName)).toBe(true);
+		});
+
 		it.skip('should return exact matches first', async () => {
 			const files = [
 				{ path: 'test.md', basename: 'test' },

src/gui/suggesters/FileIndex.ts

@@ -1,7 +1,7 @@
 import type { App, Plugin } from "obsidian";
 import { TFile } from "obsidian";
 import Fuse from "fuse.js";
-import { sanitizeHeading } from "./utils";
+import { normalizeForFuse, normalizeForSearch, sanitizeHeading } from "./utils";
 
 export interface IndexedFile {
 	path: string;
@@ -69,6 +69,29 @@ const FUSE_UPDATE_DEBOUNCE_MS = 100;
 // Regex to test if a character is alphanumeric (used for word boundary detection)
 const ALPHANUMERIC_REGEX = /\w/;
 
+const normalizeFuseValue = (value: unknown): string | string[] => {
+	if (typeof value === "string") return normalizeForFuse(value);
+	if (Array.isArray(value)) {
+		return value
+			.filter((entry): entry is string => typeof entry === "string")
+			.map((entry) => normalizeForFuse(entry));
+	}
+	return "";
+};
+
+const resolvePath = (obj: IndexedFile, path: string | string[]): unknown => {
+	if (Array.isArray(path)) {
+		return path.reduce((acc: any, key) => acc?.[key], obj as any);
+	}
+	if (path.includes(".")) {
+		return path.split(".").reduce((acc: any, key) => acc?.[key], obj as any);
+	}
+	return (obj as any)[path];
+};
+
+const getFuseValue = (obj: IndexedFile, path: string | string[]): string | string[] =>
+	normalizeFuseValue(resolvePath(obj, path));
+
 // Configurable search ranking weights
 export const SearchWeights = {
 	base: {
@@ -131,7 +154,8 @@ export class FileIndex {
 			ignoreLocation: true,
 			findAllMatches: true,
 			shouldSort: false, // We'll handle sorting ourselves
-			includeMatches: true // Include match information to detect alias hits
+			includeMatches: true, // Include match information to detect alias hits
+			getFn: getFuseValue
 		};
 
 		this.fuseStrict = new Fuse<IndexedFile>([], {
@@ -459,7 +483,8 @@ export class FileIndex {
 
 	search(query: string, context: SearchContext = {}, limit = 50): SearchResult[] {
 		const results: SearchResult[] = [];
-		const queryLower = query.toLowerCase();
+		const queryLower = normalizeForSearch(query);
+		const fuseQuery = normalizeForFuse(query);
 
 		// Handle global heading search when query contains '#'
 		if (query.includes('#')) {
@@ -474,7 +499,7 @@ export class FileIndex {
 
 		// 1. Exact matches (basename and aliases) - Tier 0
 		for (const file of allFiles) {
-			if (file.basename.toLowerCase() === queryLower) {
+			if (normalizeForSearch(file.basename) === queryLower) {
 				results.push({
 					file,
 					score: this.calculateScore(file, query, context, this.effectiveWeights.base.basenameExact, 'exact'),
@@ -491,7 +516,7 @@ export class FileIndex {
 			if (addedPaths.has(file.path)) continue;
 
 			for (const alias of file.aliases) {
-				if (alias.toLowerCase() === queryLower) {
+				if (normalizeForSearch(alias) === queryLower) {
 					results.push({
 						file,
 						score: this.calculateScore(file, query, context, this.effectiveWeights.base.aliasExact, 'alias'),
@@ -506,8 +531,9 @@ export class FileIndex {
 
 		// 1.5. Prefix matches (basename) - Tier 1
 		for (const file of allFiles) {
-			if (file.basename.toLowerCase().startsWith(queryLower) && 
-				file.basename.toLowerCase() !== queryLower && // not exact (already added)
+			const basenameLower = normalizeForSearch(file.basename);
+			if (basenameLower.startsWith(queryLower) && 
+				basenameLower !== queryLower && // not exact (already added)
 				!addedPaths.has(file.path)) {
 				results.push({
 					file,
@@ -522,8 +548,9 @@ export class FileIndex {
 		// 2. Prefix alias matches - Tier 1
 		for (const file of allFiles) {
 			for (const alias of file.aliases) {
-				if (alias.toLowerCase().startsWith(queryLower) && 
-					alias.toLowerCase() !== queryLower &&  // not exact (already added)
+				const aliasLower = normalizeForSearch(alias);
+				if (aliasLower.startsWith(queryLower) && 
+					aliasLower !== queryLower &&  // not exact (already added)
 					!addedPaths.has(file.path)) {
 					results.push({
 						file,
@@ -540,10 +567,11 @@ export class FileIndex {
 		for (const file of allFiles) {
 			if (addedPaths.has(file.path)) continue;
 
-			const idx = file.basename.toLowerCase().indexOf(queryLower);
+			const basenameLower = normalizeForSearch(file.basename);
+			const idx = basenameLower.indexOf(queryLower);
 			if (idx > 0) { // not at start (that would be prefix)
 				// Check if match starts at word boundary
-				const charBefore = file.basename[idx - 1];
+				const charBefore = basenameLower[idx - 1];
 				if (!ALPHANUMERIC_REGEX.test(charBefore)) { // Previous char is not alphanumeric
 					results.push({
 						file,
@@ -557,11 +585,11 @@ export class FileIndex {
 		}
 
 		// 3. Fuzzy search with adaptive threshold - Tier 3 and below
-		let fuseResults = this.fuseStrict.search(query, { limit: limit * 2 });
+		let fuseResults = this.fuseStrict.search(fuseQuery, { limit: limit * 2 });
 
 		// Relax threshold if we have too few results
 		if (fuseResults.length < this.effectiveWeights.thresholds.fuzzyRelaxCount) {
-			fuseResults = this.fuseRelaxed.search(query, { limit: limit * 2 });
+			fuseResults = this.fuseRelaxed.search(fuseQuery, { limit: limit * 2 });
 		}
 
 		for (const result of fuseResults) {
@@ -599,7 +627,7 @@ export class FileIndex {
 			for (const unresolvedLink of this.unresolvedLinks) {
 				if (unresolvedCount >= unresolvedLimit) break;
 
-				if (unresolvedLink.toLowerCase().includes(queryLower)) {
+				if (normalizeForSearch(unresolvedLink).includes(queryLower)) {
 					results.push({
 						file: {
 							path: unresolvedLink,
@@ -654,14 +682,14 @@ export class FileIndex {
 		}
 
 		// Length penalty - calculate first as it's used by alias penalty
-		const queryLower = query.toLowerCase();
+		const queryLower = normalizeForSearch(query);
 		let titleLength = file.basename.length;
 
 		// For alias matches, find the actual matched alias to get correct length
 		if (matchType === 'alias' && file.aliases.length > 0) {
 			// Find which alias was matched
 			const matchedAlias = file.aliases.find(alias => 
-				alias.toLowerCase().includes(queryLower)
+				normalizeForSearch(alias).includes(queryLower)
 			);
 			if (matchedAlias) {
 				titleLength = matchedAlias.length;
@@ -683,15 +711,15 @@ export class FileIndex {
 		}
 
 		// Position bonus - earlier matches are better
-		let textToSearch = file.basename.toLowerCase();
+		let textToSearch = normalizeForSearch(file.basename);
 
 		// For alias matches, find the actual matched alias for position calculation
 		if (matchType === 'alias' && file.aliases.length > 0) {
 			const matchedAlias = file.aliases.find(alias => 
-				alias.toLowerCase().includes(queryLower)
+				normalizeForSearch(alias).includes(queryLower)
 			);
 			if (matchedAlias) {
-				textToSearch = matchedAlias.toLowerCase();
+				textToSearch = normalizeForSearch(matchedAlias);
 			}
 		}
 
@@ -706,7 +734,7 @@ export class FileIndex {
 
 	private searchWithHeadings(query: string, context: SearchContext, limit: number): SearchResult[] {
 		const [filePart, headingPartRaw] = query.split('#');
-		const headingPart = headingPartRaw.toLowerCase();
+		const headingPart = normalizeForSearch(headingPartRaw ?? "");
 		const results: SearchResult[] = [];
 
 		// Prevent infinite recursion by doing a simple search if no file part
@@ -722,7 +750,7 @@ export class FileIndex {
 				for (const heading of file.headings) {
 					if (resultCount >= maxResults) break;
 
-					if (heading.toLowerCase().includes(headingPart)) {
+					if (normalizeForSearch(heading).includes(headingPart)) {
 						results.push({
 							file,
 							score: this.calculateScore(file, query, context, 0.1),
@@ -740,7 +768,7 @@ export class FileIndex {
 
 			for (const fileResult of fileResults) {
 				for (const heading of fileResult.file.headings) {
-					if (heading.toLowerCase().includes(headingPart)) {
+					if (normalizeForSearch(heading).includes(headingPart)) {
 						results.push({
 							file: fileResult.file,
 							score: fileResult.score + 0.05,
@@ -760,15 +788,16 @@ export class FileIndex {
 	private searchFiles(query: string, context: SearchContext, limit: number): SearchResult[] {
 		// Direct file search without heading handling to avoid recursion
 		const results: SearchResult[] = [];
-		const queryLower = query.toLowerCase();
+		const queryLower = normalizeForSearch(query);
+		const fuseQuery = normalizeForFuse(query);
 		const addedPaths = new Set<string>();
 
 		// Pre-create array from fileMap for better performance
 		const allFiles = Array.from(this.fileMap.values());
 
 		// 1. Exact matches (basename and aliases) - Tier 0
 		for (const file of allFiles) {
-			if (file.basename.toLowerCase() === queryLower) {
+			if (normalizeForSearch(file.basename) === queryLower) {
 				results.push({
 					file,
 					score: this.calculateScore(file, query, context, -1000, 'exact'),
@@ -785,7 +814,7 @@ export class FileIndex {
 			if (addedPaths.has(file.path)) continue;
 
 			for (const alias of file.aliases) {
-				if (alias.toLowerCase() === queryLower) {
+				if (normalizeForSearch(alias) === queryLower) {
 					results.push({
 						file,
 						score: this.calculateScore(file, query, context, this.effectiveWeights.base.aliasExact, 'alias'),
@@ -800,8 +829,9 @@ export class FileIndex {
 
 		// 1.5. Prefix matches (basename) - Tier 1
 		for (const file of allFiles) {
-			if (file.basename.toLowerCase().startsWith(queryLower) && 
-				file.basename.toLowerCase() !== queryLower &&
+			const basenameLower = normalizeForSearch(file.basename);
+			if (basenameLower.startsWith(queryLower) && 
+				basenameLower !== queryLower &&
 				!addedPaths.has(file.path)) {
 				results.push({
 					file,
@@ -816,8 +846,9 @@ export class FileIndex {
 		// 2. Prefix alias matches - Tier 1
 		for (const file of allFiles) {
 			for (const alias of file.aliases) {
-				if (alias.toLowerCase().startsWith(queryLower) && 
-					alias.toLowerCase() !== queryLower &&
+				const aliasLower = normalizeForSearch(alias);
+				if (aliasLower.startsWith(queryLower) && 
+					aliasLower !== queryLower &&
 					!addedPaths.has(file.path)) {
 					results.push({
 						file,
@@ -834,10 +865,11 @@ export class FileIndex {
 		for (const file of allFiles) {
 			if (addedPaths.has(file.path)) continue;
 
-			const idx = file.basename.toLowerCase().indexOf(queryLower);
+			const basenameLower = normalizeForSearch(file.basename);
+			const idx = basenameLower.indexOf(queryLower);
 			if (idx > 0) { // not at start (that would be prefix)
 				// Check if match starts at word boundary
-				const charBefore = file.basename[idx - 1];
+				const charBefore = basenameLower[idx - 1];
 				if (!ALPHANUMERIC_REGEX.test(charBefore)) { // Previous char is not alphanumeric
 					results.push({
 						file,
@@ -851,9 +883,9 @@ export class FileIndex {
 		}
 
 		// 3. Fuzzy search - Tier 3 and below
-		let fuseResults = this.fuseStrict.search(query, { limit: limit * 2 });
+		let fuseResults = this.fuseStrict.search(fuseQuery, { limit: limit * 2 });
 		if (fuseResults.length < this.effectiveWeights.thresholds.fuzzyRelaxCount) {
-			fuseResults = this.fuseRelaxed.search(query, { limit: limit * 2 });
+			fuseResults = this.fuseRelaxed.search(fuseQuery, { limit: limit * 2 });
 		}
 
 		for (const result of fuseResults) {

src/gui/suggesters/fileSuggester.ts

@@ -5,6 +5,7 @@ import type { App } from "obsidian";
 import { TFile } from "obsidian";
 import { FILE_LINK_REGEX } from "../../constants";
 import { FileIndex, type SearchResult, type SearchContext, type IndexedFile } from "./FileIndex";
+import { normalizeForSearch } from "./utils";
 import QuickAdd from "../../main";
 
 interface HTMLElementWithTooltipCleanup extends HTMLElement {
@@ -125,6 +126,7 @@ export class FileSuggester extends TextInputSuggest<SearchResult> {
 	private getHeadingSuggestions(input: string): SearchResult[] {
 		const [fileName, headingQuery] = input.split('#');
 		const noFileSpecified = fileName.trim() === '';
+		const headingQueryNormalized = normalizeForSearch(headingQuery ?? "");
 
 		// Determine candidate files based on whether file part was specified
 		let candidateFiles: IndexedFile[] = [];
@@ -148,9 +150,9 @@ export class FileSuggester extends TextInputSuggest<SearchResult> {
 		for (const file of candidateFiles) {
 			const headings = this.fileIndex.getHeadings(file);
 
-			const filteredHeadings = headings
-				.filter(h => headingQuery === '' || h.toLowerCase().includes(headingQuery.toLowerCase()))
-				.slice(0, 20);
+		const filteredHeadings = headings
+			.filter(h => headingQuery === '' || normalizeForSearch(h).includes(headingQueryNormalized))
+			.slice(0, 20);
 
 			for (const heading of filteredHeadings) {
 				results.push({
@@ -213,10 +215,11 @@ export class FileSuggester extends TextInputSuggest<SearchResult> {
 		// Get all files, not just markdown
 		const allFiles = this.app.vault.getFiles();
 		const attachmentExtensions = ['png', 'jpg', 'jpeg', 'gif', 'svg', 'pdf', 'mp4', 'webm', 'mov', 'canvas'];
+		const queryLower = normalizeForSearch(query);
 
 		const attachmentFiles = allFiles.filter(file =>
 			attachmentExtensions.includes(file.extension.toLowerCase()) &&
-			(query === '' || file.basename.toLowerCase().includes(query.toLowerCase()))
+			(query === '' || normalizeForSearch(file.basename).includes(queryLower))
 		);
 
 		return attachmentFiles
@@ -269,7 +272,8 @@ export class FileSuggester extends TextInputSuggest<SearchResult> {
 				const headingQuery = this.lastInput.includes('#')
 					? this.lastInput.split('#')[1]
 					: '';
-				if (headingQuery && heading.toLowerCase().includes(headingQuery.toLowerCase())) {
+				const headingQueryNormalized = normalizeForSearch(headingQuery ?? "");
+				if (headingQuery && normalizeForSearch(heading).includes(headingQueryNormalized)) {
 					const tempEl = document.createElement('span');
 					this.renderMatch(tempEl, heading, headingQuery);
 					mainText = tempEl.innerHTML;