Building REP File Import in VS Code

What We’re Building

The ability to drag a REP file from VS Code’s file explorer directly onto the map panel, and have the track data appear. No wizards, no dialogs — just drop and see.

This sounds simple, but it’s the kind of flow that makes tools feel native rather than bolted-on. Analysts receive new data files constantly. Getting that data into a plot should take seconds, not minutes of clicking through import dialogs.

We’re also adding a right-click “Load into Debrief…” option for cases where you want to pick the destination plot explicitly. Same import, different trigger.

How It Fits

This builds on three services we’ve already shipped: debrief-io for parsing REP files, debrief-stac for catalog storage, and the VS Code extension’s map panel. The loader mini-app already handles file loading via Electron — this brings that capability directly into the editor where analysts already spend their time.

The architecture is deliberately simple: the extension orchestrates, the Python services do the work. No new parsing code, no new storage logic. Just wiring.

Key Decisions

• Drag-drop uses HTML5 events in the webview — VS Code’s tree drag-drop API only handles tree-to-tree operations, so we use standard browser APIs for the map panel.

• Duplicate detection by filename — If you try to import a file that’s already an asset on the plot, you get a warning. We considered content hashing but it’s overkill for single-file imports.

• Two-step picker for context menu — First select catalog, then select plot. Mirrors the loader mini-app pattern. Uses VS Code’s native QuickPick so it’s keyboard-navigable.

• Auto-zoom after import — The map adjusts bounds to show the newly imported tracks. Small touch, but it answers “did it work?” immediately.

REP file import now works inside the VS Code extension. Two ways to get data in:

1. Drag-and-drop — drop a .rep file directly onto the map panel
2. Context menu — right-click any REP file in the Explorer, pick your target plot

Either way, the file is parsed, stored as a STAC asset (for provenance), and the tracks appear on the map. The view automatically zooms to fit everything.

The Flow

Drop a file. Watch the progress:

1. Checking for duplicates…
2. Parsing REP file…
3. Storing asset…
4. Storing features…

Then: tracks on screen, map zoomed to fit. For a typical REP file, under 5 seconds total.

If something goes wrong, you get a message that actually helps. Invalid format at line 42? The error says so. Already imported that file? Warning dialog asks what you want to do. Tried to drop three files at once? Polite rejection with explanation.

Architecture Decision

IoService is storage-agnostic. It parses REP files and returns GeoJSON features. Full stop.

The VS Code extension acts as orchestrator:

• Call IoService to parse
• Call StacService to store the asset
• Call StacService to store the features
• Update the map

This separation matters. When we add different storage backends later (local files, S3, whatever), IoService stays unchanged. Parsing is parsing. Storage is storage. The orchestrator decides where data goes.

User drops file
    |
    v
MapPanel (orchestrator)
    |
    +-- IoService.parseRep() --> GeoJSON features
    |
    +-- StacService.addAsset() --> store original file
    |
    +-- StacService.addFeatures() --> merge into plot
    |
    v
Map updates, zooms to fit

Error Handling

We wrote 24 tests just for error messages. Each error code has a template:

Line numbers matter. Field names matter. When an analyst sees an error, they should know what to fix.

By The Numbers

• 114 tests passing across 11 test files
• 24 new tests for bounds calculation, error messages, IoService types
• 522ms total test runtime
• 0 network calls — works completely offline

What’s Next

This completes the tracer bullet for VS Code extension data import. The pipeline now runs from REP file to STAC storage to map visualization. Next milestone: analysis tools integration, so imported data becomes the input for calculations and queries.

View the spec

See evidence