Add popular collection script proof of concept

tests/integration/popular_collections/NSIDC_ECS.txt

@@ -0,0 +1,100 @@
+C2559919423-NSIDC_ECS
+C2565090645-NSIDC_ECS
+C2564427300-NSIDC_ECS
+C2564625052-NSIDC_ECS
+C1646610417-NSIDC_ECS
+C1646610079-NSIDC_ECS
+C2650116584-NSIDC_ECS
+C2776463943-NSIDC_ECS
+C1646610415-NSIDC_ECS
+C2531308461-NSIDC_ECS
+C2317038246-NSIDC_ECS
+C1600510471-NSIDC_ECS
+C2541211133-NSIDC_ECS
+C1000000443-NSIDC_ECS
+C2750966856-NSIDC_ECS
+C1000000461-NSIDC_ECS
+C2607017115-NSIDC_ECS
+C2738530540-NSIDC_ECS
+C2561045326-NSIDC_ECS
+C2399557265-NSIDC_ECS
+C2136471727-NSIDC_ECS
+C2560378689-NSIDC_ECS
+C2136471705-NSIDC_ECS
+C2776464171-NSIDC_ECS
+C1931663473-NSIDC_ECS
+C2567856357-NSIDC_ECS
+C1646610394-NSIDC_ECS
+C1431539277-NSIDC_ECS
+C2666857908-NSIDC_ECS
+C2176562253-NSIDC_ECS
+C2537927247-NSIDC_ECS
+C1000000400-NSIDC_ECS
+C2776463935-NSIDC_ECS
+C1542606326-NSIDC_ECS
+C2623698025-NSIDC_ECS
+C1000000445-NSIDC_ECS
+C2737912334-NSIDC_ECS
+C2776464127-NSIDC_ECS
+C1431413941-NSIDC_ECS
+C2737997483-NSIDC_ECS
+C2509060594-NSIDC_ECS
+C1000000460-NSIDC_ECS
+C1646609734-NSIDC_ECS
+C1631093696-NSIDC_ECS
+C189991864-NSIDC_ECS
+C1449515322-NSIDC_ECS
+C2737997243-NSIDC_ECS
+C189991870-NSIDC_ECS
+C1000000420-NSIDC_ECS
+C1646610247-NSIDC_ECS
+C189991869-NSIDC_ECS
+C2420522159-NSIDC_ECS
+C189991871-NSIDC_ECS
+C1000000464-NSIDC_ECS
+C1365767075-NSIDC_ECS
+C189991863-NSIDC_ECS
+C1371883515-NSIDC_ECS
+C1646609754-NSIDC_ECS
+C1646610390-NSIDC_ECS
+C1646583410-NSIDC_ECS
+C1452437262-NSIDC_ECS
+C2548345108-NSIDC_ECS
+C1646609808-NSIDC_ECS
+C1574621139-NSIDC_ECS
+C2794540918-NSIDC_ECS
+C2776463773-NSIDC_ECS
+C1601063219-NSIDC_ECS
+C2776463679-NSIDC_ECS
+C2534576405-NSIDC_ECS
+C2037494637-NSIDC_ECS
+C1646610101-NSIDC_ECS
+C2253727823-NSIDC_ECS
+C179014688-NSIDC_ECS
+C2692731693-NSIDC_ECS
+C1542606320-NSIDC_ECS
+C1432250096-NSIDC_ECS
+C1723866830-NSIDC_ECS
+C2817412114-NSIDC_ECS
+C179014696-NSIDC_ECS
+C2076118670-NSIDC_ECS
+C179014698-NSIDC_ECS
+C128599377-NSIDC_ECS
+C2519306057-NSIDC_ECS
+C1454773262-NSIDC_ECS
+C130038008-NSIDC_ECS
+C179014694-NSIDC_ECS
+C1646610212-NSIDC_ECS
+C1995259960-NSIDC_ECS
+C179014690-NSIDC_ECS
+C1997893797-NSIDC_ECS
+C1489170961-NSIDC_ECS
+C2559364353-NSIDC_ECS
+C2295286903-NSIDC_ECS
+C2137003017-NSIDC_ECS
+C2136471686-NSIDC_ECS
+C186584407-NSIDC_ECS
+C1000001740-NSIDC_ECS
+C1442092309-NSIDC_ECS
+C1703031106-NSIDC_ECS
+C2776463734-NSIDC_ECS

tests/integration/popular_collections/generate.py

@@ -0,0 +1,41 @@
+"""Generate lists of most popular collections for each of the hardcoded DAACs."""
+
+from pathlib import Path
+
+import requests
+
+THIS_DIR = Path(__file__).parent
+
+
+def top_collections(*, provider: str, num: int = 100) -> list[str]:
+    response = requests.post(
+        "https://cmr.earthdata.nasa.gov/search/collections.json",
+        data={
+            "provider": provider,
+            "has_granules_or_cwic": True,
+            "include_facets": "v2",
+            "include_granule_counts": True,
+            "include_has_granules": True,
+            "include_tags": "edsc.*,opensearch.granule.osdd",
+            "page_num": 1,
+            "page_size": 100,
+            "sort_key[]": "-usage_score",
+        },
+    )
+    collection_ids = [
+        collection["id"] for collection in response.json()["feed"]["entry"]
+    ]
+    return collection_ids
+
+
+def main():
+    for provider in ["NSIDC_ECS"]:
+        collection_ids = top_collections(provider="NSIDC_ECS")
+
+        output = THIS_DIR / f"{provider}.txt"
+        with output.open("w") as f:
+            f.write("\n".join(collection_ids))
+
+
+if __name__ == "__main__":
+    main()