Technical Architecture & File Lifecycle

Data Processing Policy

A transparent engineering breakdown of how Online Metadata Tools processes files across local browser memory and isolated server-side temporary storage.

Effective Date:
Last Reviewed:
Platform: Online Metadata Tools (onlinemetadatatools.com)

1. Purpose & Scope

This Data Processing Policy explains the technical lifecycle of files and metadata processed through Online Metadata Tools (https://onlinemetadatatools.com). It distinguishes clearly between operations performed locally in your web browser, operations executed in isolated temporary server storage, and persistent account/billing records stored in our database.

2. Supported File Types & Metadata Categories Processed

Our platform inspects and cleans container headers across 36 supported file extensions:

  • Standard & Raster Images: JPG, JPEG, PNG, WebP, GIF, HEIC, HEIF, AVIF, TIFF, TIF, BMP.
  • Camera RAW Formats: CR2 (Canon), NEF (Nikon), ARW (Sony), DNG (Adobe Digital Negative).
  • Documents & Office Files: PDF, DOCX, XLSX, PPTX, ODT, ODS, ODP.
  • Video Containers: MP4, MOV, WebM, MKV, AVI, M4V.
  • Audio Containers: MP3, WAV, FLAC, AAC, M4A, OGG, OPUS, AIFF.

Metadata Categories Processed: EXIF IFDs, GPS coordinates, IPTC-IIM datasets, XMP XML packets, ICC color profiles, C2PA / JUMBF provenance manifests, AI generation prompt parameters (Stable Diffusion / ComfyUI chunks), PDF Info dictionaries, Office XML docProps, ID3 tags, Vorbis comments, and QuickTime/MP4 container atoms.

3. Hybrid Processing Architecture: Local vs. Server Processing

Online Metadata Tools uses a hybrid processing model rather than a purely client-only or purely server-only pipeline:

A. Local Browser Processing (lib/client/localProcessor.ts)

  • Formats Handled Locally: JPG, JPEG, PNG, WebP, GIF, PDF, DOCX, XLSX, PPTX, ODT, ODS, and ODP up to 50 MB on desktop browsers and 25 MB on mobile browsers.
  • How It Works: File bytes are read into browser memory (ArrayBuffer / Uint8Array), inspected and stripped locally, and provided for download via transient browser Blob object URLs (URL.createObjectURL) that are revoked when you reset the workspace or leave the page.

B. Server-Side Processing (lib/server/services/processor.ts)

  • When Server Processing Is Used: Video formats, audio formats, Camera RAW formats, HEIC/HEIF/AVIF/TIFF/BMP images, files exceeding browser memory thresholds, multi-file server ZIP batch jobs (/api/v1/batch), signed verification certificates, or automatic fallback when a browser cannot parse a complex container locally.
  • How It Works: The file is transmitted over encrypted HTTPS to our server, validated via magic-byte inspection, processed by format-specific engines (including FFmpeg stream copy -map_metadata -1 -c copy for audio/video, pdf-lib, Office XML sanitizers, and binary segment strippers), and re-scanned to verify metadata removal.

4. Temporary Server File Retention & Automatic Deletion

When server-side processing is used, files are handled strictly as transient processing artifacts:

  • Isolated Temporary Directory: Files are written to /tmp/onlinemetadatatools (or the configured TEMP_DIR) under a randomly generated UUIDv4 identifier. Path-traversal protection enforces that all file operations stay strictly within this directory.
  • 10-Minute Default Time-to-Live (TTL): Each registered output file is assigned an expiration timestamp configured by TEMP_FILE_TTL_MINUTES (default: 10 minutes).
  • Deletion Mechanism: Once a file's TTL expires (and it is not actively mid-processing or mid-download), cleanExpiredRegistryEntries() and the /api/cron/cleanup endpoint unlink both the temporary media file and its companion .meta.json file from the filesystem.

5. Distinction Between Temporary File Processing and Retained Account Data

Data CategoryWhere Processed / StoredRetention Lifecycle
Local Browser FilesBrowser RAM (ArrayBuffer / Blob)Cleared when workspace is reset or browser tab is closed.
Server-Processed Files & ZIP BatchesIsolated temp directory (/tmp/onlinemetadatatools)Default 10-minute TTL (tempFileTtlMinutes = 10), then unlinked from disk.
Extracted Metadata Values & Original FilenamesTransient API response payload onlyNever written to PostgreSQL database tables.
Quota & Operational Job MetricsPostgreSQL (daily_usages, processing_jobs)Retained for quota enforcement and service diagnostics (contains format, byte size, risk score, and SHA-256 IP hash).
User Account & Subscription RecordsPostgreSQL (users, subscriptions, saved_presets)Retained while account is active or as required for billing compliance.

6. Security Measures & Concurrency Controls

  • Magic-Byte & Structural Validation: Uploaded files are verified against expected binary signatures before processing to reject disguised executables or malformed containers.
  • Concurrency & Resource Isolation: Server processing jobs are governed by a global processing semaphore and per-format execution timeouts (30s for images/audio/docs, 60s for PDFs, 180s for video).
  • Zero AI Training: User files and metadata are processed exclusively to fulfill the user's explicit inspection or cleaning request and are never used for machine-learning training or marketing profiling.

7. Technical Service Providers & Sub-Processors

  • Application Hosting & Database Infrastructure: Hosts the Next.js application server, temporary filesystem, and PostgreSQL database.
  • Creem.io: Processes Day Pass ($3.99) and Pro Monthly ($7.99/mo) checkout sessions and billing webhooks.
  • Google YouTube Data API v3: Queried only when a user submits a YouTube URL in the YouTube Metadata Viewer.

8. Contact Information

For questions regarding our data processing architecture or to submit a data privacy request, please email [email protected], review our Privacy Policy, or visit our Contact Support page.