Obsidian Data Handling
Overview
Data management patterns for Obsidian plugins: plugin config with loadData/saveData, vault file I/O, frontmatter parsing via metadataCache, handling renames and deletes, cross-device sync considerations, and IndexedDB fallback for large datasets.
Prerequisites
- Working Obsidian plugin (
export default class extends Plugin) - Understanding of Obsidian's
VaultandMetadataCacheAPIs - TypeScript compilation configured
Instructions
Step 1: Plugin Config with loadData / saveData
Obsidian stores plugin data in .obsidian/plugins/<plugin-id>/data.json. Use loadData() and saveData() — never read that file directly.
interface PluginConfig {
version: number;
apiEndpoint: string;
syncInterval: number;
excludedFolders: string[];
}
const DEFAULT_CONFIG: PluginConfig = {
version: 1,
apiEndpoint: 'https://api.example.com',
syncInterval: 300,
excludedFolders: [],
};
export default class DataPlugin extends Plugin {
config: PluginConfig;
async onload() {
await this.loadConfig();
}
async loadConfig() {
const saved = await this.loadData();
this.config = Object.assign({}, DEFAULT_CONFIG, saved);
// Migrate from older config versions
if (this.config.version < 1) {
this.config.excludedFolders = [];
this.config.version = 1;
await this.saveConfig();
}
}
async saveConfig() {
await this.saveData(this.config);
}
}
loadData() returns null on first run — Object.assign onto defaults handles this cleanly.
Step 2: Reading and Writing Vault Files
import { TFile, TFolder, normalizePath } from 'obsidian';
// Read a markdown file
async readNote(path: string): Promise<string | null> {
const file = this.app.vault.getAbstractFileByPath(normalizePath(path));
if (file instanceof TFile) {
return await this.app.vault.read(file);
}
return null;
}
// Write or create a markdown file
async writeNote(path: string, content: string): Promise<TFile> {
const normalized = normalizePath(path);
const existing = this.app.vault.getAbstractFileByPath(normalized);
if (existing instanceof TFile) {
await this.app.vault.modify(existing, content);
return existing;
}
// Ensure parent folder exists
const dir = normalized.substring(0, normalized.lastIndexOf('/'));
if (dir && !this.app.vault.getAbstractFileByPath(dir)) {
await this.app.vault.createFolder(dir);
}
return await this.app.vault.create(normalized, content);
}
// Append to a file (e.g., a log or journal)
async appendToNote(path: string, text: string): Promise<void> {
const file = this.app.vault.getAbstractFileByPath(normalizePath(path));
if (file instanceof TFile) {
await this.app.vault.append(file, '\n' + text);
}
}
Use vault.cachedRead() instead of vault.read() when you don't need the absolute latest content — it avoids hitting disk on every call.
Step 3: Working with Frontmatter via MetadataCache
Never parse YAML frontmatter manually. Obsidian's metadataCache keeps a parsed cache of every file's frontmatter.
import { TFile, CachedMetadata } from 'obsidian';
// Read frontmatter from a file
getFrontmatter(file: TFile): Record<string, any> | null {
const cache: CachedMetadata | null = this.app.metadataCache.getFileCache(file);
return cache?.frontmatter ?? null;
}
// Update frontmatter using processFrontMatter (Obsidian 1.4+)
async setStatus(file: TFile, status: string): Promise<void> {
await this.app.fileManager.processFrontMatter(file, (fm) => {
fm.status = status;
fm.updated = new Date().toISOString();
});
}
// Bulk query: find all files with a specific tag
getFilesWithTag(tag: string): TFile[] {
const files: TFile[] = [];
for (const file of this.app.vault.getMarkdownFiles()) {
const cache = this.app.metadataCache.getFileCache(file);
const tags = cache?.tags?.map(t => t.tag) ?? [];
const fmTags = cache?.frontmatter?.tags ?? [];
if (tags.includes(tag) || fmTags.includes(tag.replace('#', ''))) {
files.push(file);
}
}
return files;
}
processFrontMatter handles YAML serialization correctly — it preserves comments and formatting, and is the only safe way to update frontmatter programmatically.
Step 4: Handling File Renames and Deletes
Plugins that index file paths must update their state when files move or disappear.
async onload() {
// Track renames to update internal references
this.registerEvent(
this.app.vault.on('rename', (file, oldPath) => {
if (file instanceof TFile) {
this.onFileRenamed(file, oldPath);
}
})
);
// Clean up when files are deleted
this.registerEvent(
this.app.vault.on('delete', (file) => {
if (file instanceof TFile) {
this.onFileDeleted(file.path);
}
})
);
}
private onFileRenamed(file: TFile, oldPath: string) {
// Update any stored path references
if (this.config.pinnedFiles?.includes(oldPath)) {
const idx = this.config.pinnedFiles.indexOf(oldPath);
this.config.pinnedFiles[idx] = file.path;
this.saveConfig();
}
}
private onFileDeleted(path: string) {
// Remove from any indexes
if (this.config.pinnedFiles?.includes(path)) {
this.config.pinnedFiles = this.config.pinnedFiles.filter(p => p !== path);
this.saveConfig();
}
}
Always use registerEvent — it automatically cleans up the listener when the plugin unloads.
Step 5: Cross-Device Sync Considerations
Obsidian vaults synced via iCloud, Dropbox, or Obsidian Sync introduce eventual consistency issues.
// Problem: two devices modify data.json simultaneously
// Solution: merge-friendly data structures
interface SyncSafeConfig {
// Use a map keyed by unique IDs instead of arrays
// Maps merge better than arrays across sync conflicts
items: Record<string, { value: string; updatedAt: number }>;
}
// Timestamp-based last-write-wins merge
mergeConfigs(local: SyncSafeConfig, remote: SyncSafeConfig): SyncSafeConfig {
const merged: SyncSafeConfig = { items: {} };
const allKeys = new Set([
...Object.keys(local.items),
...Object.keys(remote.items),
]);
for (const key of allKeys) {
const l = local.items[key];
const r = remote.items[key];
if (!l) merged.items[key] = r;
else if (!r) merged.items[key] = l;
else merged.items[key] = l.updatedAt >= r.updatedAt ? l : r;
}
return merged;
}
Guidelines for sync-friendly plugins:
- Avoid storing file paths in
data.json— they differ across devices with different vault locations - Use file content hashes or frontmatter IDs for identity instead of paths
- Keep
data.jsonsmall — large files cause sync conflicts and slow sync
Step 6: IndexedDB Fallback for Large Datasets
When plugin data exceeds what's practical for data.json (more than ~1MB), use IndexedDB.
class PluginDatabase {
private db: IDBDatabase | null = null;
private dbName: string;
constructor(pluginId: string) {
this.dbName = `obsidian-${pluginId}`;
}
async open(): Promise<void> {
return new Promise((resolve, reject) => {
const request = indexedDB.open(this.dbName, 1);
request.onupgradeneeded = (event) => {
const db = (event.target as IDBOpenDBRequest).result;
if (!db.objectStoreNames.contains('cache')) {
db.createObjectStore('cache', { keyPath: 'id' });
}
};
request.onsuccess = (event) => {
this.db = (event.target as IDBOpenDBRequest).result;
resolve();
};
request.onerror = () => reject(request.error);
});
}
async put(id: string, data: any): Promise<void> {
if (!this.db) throw new Error('Database not open');
return new Promise((resolve, reject) => {
const tx = this.db!.transaction('cache', 'readwrite');
tx.objectStore('cache').put({ id, data, updatedAt: Date.now() });
tx.oncomplete = () => resolve();
tx.onerror = () => reject(tx.error);
});
}
async get(id: string): Promise<any | null> {
if (!this.db) throw new Error('Database not open');
return new Promise((resolve, reject) => {
const tx = this.db!.transaction('cache', 'readonly');
const request = tx.objectStore('cache').get(id);
request.onsuccess = () => resolve(request.result?.data ?? null);
request.onerror = () => reject(request.error);
});
}
close() {
this.db?.close();
this.db = null;
}
}
// Usage in plugin
async onload() {
this.db = new PluginDatabase(this.manifest.id);
await this.db.open();
}
onunload() {
this.db?.close();
}
IndexedDB is per-device and does not sync across devices. Use it for caches and derived data that can be rebuilt, not for primary user data.
Output
- Plugin config loading with version migration
- Safe vault file read/write/append operations
- Frontmatter access via metadataCache
- Rename and delete event handlers
- Sync-friendly data structures
- IndexedDB storage for large datasets
Error Handling
| Issue | Cause | Solution |
|-------|-------|----------|
| loadData() returns null | First run, no data.json yet | Object.assign onto defaults |
| Frontmatter returns undefined | File not yet indexed by cache | Listen for metadataCache.on('resolved') |
| File write fails | Parent folder doesn't exist | Create folder with vault.createFolder() first |
| Settings lost after sync | Concurrent writes from two devices | Use merge-friendly data structures with timestamps |
| data.json too large / slow | Storing too much data | Move large data to IndexedDB |
| stale cache after modify | cachedRead returns old content | Use vault.read() when freshness matters |
Examples
Export All Notes with Tag to JSON
async exportTaggedNotes(tag: string): Promise<string> {
const files = this.getFilesWithTag(tag);
const notes = await Promise.all(
files.map(async (f) => ({
path: f.path,
content: await this.app.vault.read(f),
frontmatter: this.getFrontmatter(f),
}))
);
return JSON.stringify(notes, null, 2);
}
Atomic Config Update
async updateConfig<K extends keyof PluginConfig>(
key: K,
value: PluginConfig[K]
): Promise<void> {
this.config[key] = value;
await this.saveConfig();
}
Resources
Next Steps
For team access control patterns, see obsidian-enterprise-rbac. For performance with large vaults, see obsidian-rate-limits.