Instagram Saved Posts Export Feature

Implementation Complete ✅

The Instagram saved posts export feature has been implemented to allow users to extract their saved Instagram content and import it into treeper trips.

What Was Implemented

Backend (TypeScript)

• apps/backend/src/imports/dto/create-import.dto.ts
  • Added 'instagram' to ImportSourceType enum
  • Updated validators to accept Instagram exports

Workers (Python)

• apps/workers/src/treeper_workers/ai/ig_parser.py (NEW)

  • Parses Instagram export ZIPs containing saved_posts.json
  • Converts saved posts into activities with:
    • Title from caption (truncated to 200 chars)
    • Notes with full caption
    • Photos from static posts or carousels
    • Video URLs from reels
    • Dates extracted from post timestamps
  • Handles edge cases: missing captions, corrupt files, empty exports

• apps/workers/src/treeper_workers/ai/pipeline.py

  • Integrated Instagram parser into source dispatcher
  • Maps source_type='instagram' to ZIP parsing

Tests

• apps/workers/tests/test_ig_parser.py (NEW)
  • 16 unit tests covering:
    • Empty and single posts
    • Photos, carousels, and reels
    • Date parsing (multiple ISO formats)
    • Caption truncation and snippet generation
    • Error handling (corrupt ZIPs, missing files)
    • Edge cases (no captions, non-dict entries)

How It Works

User Flow

1. User exports Instagram data: Settings → Security → Download Data
2. Instagram emails ZIP after 24-48 hours
3. User uploads ZIP via treeper web UI:
   • POST /v1/trips/:tripId/imports/sign-upload → get signed URL
   • Upload ZIP to Supabase storage
4. Backend creates import: POST /v1/trips/:tripId/imports
   • source_type: 'instagram'
   • source_uri: 's3://bucket/trip-attachments/_instagram/:tripId/:importId.zip'
5. Worker parses ZIP → creates ItineraryDraft
6. User reviews draft, selects activities: POST /imports/:id/commit
7. Activities stored in trip with dates, captions, and media URLs

Data Extraction

Instagram Export Format:

[
  {
    "caption": "Beautiful sunset...",
    "taken_at": "2026-07-15T06:00:00",
    "photo": [{"uri": "https://instagram.com/photo.jpg"}],
    "carousel": [...],
    "video_list": [{"uri": "https://video.mp4"}]
  }
]

Parsed Activity:

• title: Caption (max 200 chars) or “Saved Post #N”
• kind: ‘sight’ (can be reclassified later)
• date: Extracted from taken_at timestamp
• notes: Full caption text
• image_urls: Photos from post
• video_url: Reel video link
• location: null (user can add later)

Enrichment & Commit

Activities are created with:

• ✅ Titles and notes from captions
• ✅ Media URLs from Instagram export
• ❌ Coordinates (null — user adds locations later or uses reenrich)
• ❌ Media enrichment (already have URLs from export)

User can later:

• Edit titles, add locations
• Trigger POST /v1/trips/:tripId/reenrich to geocode locations
• Adjust activity kinds via the UI

API Contract

// Create Instagram Import
POST /v1/trips/:tripId/imports
{
  "sources": [{
    "source_type": "instagram",
    "source_uri": "s3://bucket/trip-attachments/_instagram/:tripId/:importId.zip",
    "source_filename": "instagram_export.zip",
    "source_bytes": 15000000,
    "source_hash": "sha256..."
  }],
  "user_context": "My saved posts for Japan trip"
}

// Response: Import queued, worker parses ZIP → returns ItineraryDraft
GET /v1/imports/:id
→ status: 'parsing' | 'ready' | 'failed'
→ payload: { destinations: [], activities: [...] }

// Commit: User selects activities
POST /v1/imports/:id/commit
{
  "selected_activity_indexes": [0, 1, 3]
}
→ Activities created in trip with dates, notes, links

File Structure

apps/
├── backend/src/imports/dto/
│   └── create-import.dto.ts ← Updated with 'instagram'
└── workers/src/treeper_workers/
    ├── ai/
    │   ├── ig_parser.py ← NEW
    │   └── pipeline.py ← Updated
    └── tests/
        └── test_ig_parser.py ← NEW (16 tests, all passing)

Verification

All 173 worker tests pass including:

• 16 new Instagram parser unit tests
• All existing import pipeline tests
• Full integration with enrichment system

Backend TypeScript compiles without errors.

Edge Cases Handled

Case	Behavior
Empty saved posts	Returns empty draft (user can discard import)
Missing caption	Default title: “Saved Post #1”, #2, etc.
Corrupt ZIP	Error: “Invalid ZIP file”
Invalid JSON	Error: JSON parse failure
No coordinates	Activities created with location_lat=null
Very large ZIP	Supported (Supabase limit 50GB)
Duplicate imports	Creates separate sets (user can delete via UI)
Enrichment failures	Graceful degradation (activities created without geo)

Future Enhancements

• Incremental imports: Track already-imported posts to avoid duplicates
• Geocoding from captions: Auto-extract location hints from caption text
• Kind classification: Use LLM to guess activity types from captions
• Realtime sync: Webhook for new saved posts (would need API approval)

Testing Instructions

Run Instagram parser tests:

cd apps/workers
uv run pytest tests/test_ig_parser.py -v

Run full test suite:

uv run pytest tests/ -q

Integration test (local):

1. Prepare test ZIP:

   mkdir -p ig_export && cat > ig_export/saved_posts.json <<'EOF'
   [{"caption": "Test post https://example.com", "taken_at": "2026-07-15T06:00:00"}]
   EOF
   zip -r test.zip ig_export

2. Upload via local API:

   # Sign upload URL
   curl -X POST http://localhost:3000/v1/trips/:tripId/imports/sign-upload \
     -H "Authorization: Bearer $JWT" \
     -d '{"filename":"ig_export.zip","bytes":1000}' | jq .url
   
   # Upload ZIP
   curl -X PUT "$SIGNED_URL" --data-binary @test.zip
   
   # Create import
   curl -X POST http://localhost:3000/v1/trips/:tripId/imports \
     -H "Authorization: Bearer $JWT" \
     -d '{"sources":[{"source_type":"instagram","source_uri":"...zip"}]}'

3. Verify in Supabase:

   SELECT status, error FROM imports WHERE id = '...';
   SELECT payload FROM import_drafts WHERE import_id = '...';