Sports Highlights Pipeline

from mixpeek import Mixpeek
import requests

client = Mixpeek(api_key="YOUR_API_KEY")
API_URL = "https://api.mixpeek.com"
headers = {"Authorization": "Bearer YOUR_API_KEY", "X-Namespace": "your-namespace"}

# 1. Create a bucket for game footage
bucket = requests.post(f"{API_URL}/v1/buckets", headers=headers, json={
    "bucket_name": "sports-footage",
    "schema": {
        "properties": {
            "video_url": {"type": "url", "required": True},
            "sport": {"type": "text"},
            "game_id": {"type": "text"},
            "broadcast_date": {"type": "text"}
        }
    }
}).json()

# 2. Create a collection with multi-modal extraction
collection = requests.post(f"{API_URL}/v1/collections", headers=headers, json={
    "collection_name": "game-scenes",
    "source": {"type": "bucket", "bucket_id": bucket["bucket_id"]},
    "feature_extractor": {
        "feature_extractor_name": "video_extractor",
        "version": "v1",
        "input_mappings": {"video_url": "video_url"},
        "parameters": {
            "scene_detection_threshold": 0.3,
            "keyframe_interval": 2,
            "max_scenes": 500,
            "extract_audio": True,
            "transcription_model": "whisper-large-v3"
        },
        "field_passthrough": [
            {"source_path": "sport"},
            {"source_path": "game_id"},
            {"source_path": "broadcast_date"}
        ]
    }
}).json()

# 3. Ingest a match recording
requests.post(f"{API_URL}/v1/buckets/{bucket['bucket_id']}/objects", headers=headers, json={
    "key_prefix": "/soccer/champions-league",
    "metadata": {
        "sport": "soccer",
        "game_id": "cl-2026-final",
        "broadcast_date": "2026-05-25"
    },
    "blobs": [{"property": "video_url", "type": "video", "url": "s3://my-bucket/games/cl-final.mp4"}]
})

# 4. Create a sport-specific highlight taxonomy
taxonomy = requests.post(f"{API_URL}/v1/taxonomies", headers=headers, json={
    "taxonomy_name": "soccer_events",
    "taxonomy_type": "flat",
    "nodes": [
        {"node_id": "goal", "collection_id": "col_goal_examples", "description": "Ball crosses goal line"},
        {"node_id": "save", "collection_id": "col_save_examples", "description": "Goalkeeper saves shot"},
        {"node_id": "foul", "collection_id": "col_foul_examples", "description": "Referee stops play for foul"},
        {"node_id": "celebration", "collection_id": "col_celebration_examples", "description": "Player celebration"},
        {"node_id": "var_review", "collection_id": "col_var_examples", "description": "VAR check sequence"}
    ]
}).json()

# 5. Create a retriever that ranks moments by highlight score
retriever = requests.post(f"{API_URL}/v1/retrievers", headers=headers, json={
    "retriever_name": "soccer-highlights",
    "collection_ids": [collection["collection_id"]],
    "input_schema": {
        "properties": {
            "game_id": {"type": "text"},
            "min_confidence": {"type": "number"},
            "event_types": {"type": "array"}
        }
    },
    "stages": [
        {
            "stage_name": "feature_search",
            "version": "v1",
            "parameters": {
                "queries": [
                    {
                        "feature_address": "mixpeek://video_extractor@v1/scene_embedding",
                        "input_mapping": {"text": "highlight action moment goal celebration"},
                        "weight": 0.6
                    },
                    {
                        "feature_address": "mixpeek://audio_extractor@v1/transcript_embedding",
                        "input_mapping": {"text": "goal score amazing incredible crowd roars"},
                        "weight": 0.4
                    }
                ],
                "fusion_method": "rrf",
                "limit": 100
            }
        },
        {
            "stage_name": "taxonomy_enrich",
            "version": "v1",
            "parameters": {"taxonomy_id": taxonomy["taxonomy_id"]}
        },
        {
            "stage_name": "attribute_filter",
            "version": "v1",
            "parameters": {
                "filters": [
                    {"field": "metadata.game_id", "operator": "eq", "value": "{{inputs.game_id}}"},
                    {"field": "taxonomy_enrichment.confidence", "operator": "gte", "value": "{{inputs.min_confidence}}"}
                ]
            }
        },
        {"stage_name": "limit", "version": "v1", "parameters": {"limit": 50}}
    ]
}).json()

# 6. Retrieve highlights for a specific match
highlights = requests.post(
    f"{API_URL}/v1/retrievers/{retriever['retriever_id']}/execute",
    headers=headers,
    json={
        "inputs": {"game_id": "cl-2026-final", "min_confidence": 0.7, "event_types": ["goal", "save", "celebration"]},
        "limit": 20
    }
).json()

# 7. Build highlight manifest
print("\n🎬 Top Highlights:")
for i, doc in enumerate(highlights["documents"], 1):
    start = doc["metadata"].get("start_time", 0)
    end = doc["metadata"].get("end_time", 0)
    event = doc.get("taxonomy_enrichment", {}).get("node_id", "highlight")
    score = doc["score"]
    print(f"  {i}. [{event.upper()}] {start:.1f}s - {end:.1f}s (score: {score:.3f})")
    print(f"     Keyframe: {doc['metadata'].get('keyframe_url', '')}")