Performance provision quality dlp

Dockerfile

@@ -21,5 +21,5 @@ RUN pip install awscli
 RUN pip install --upgrade pip
 RUN pip3 install --upgrade -r requirements.txt
 
-ENTRYPOINT ["./run.sh"]
+CMD ["./run.sh"]
 

task/build-performance.sh

@@ -0,0 +1,25 @@
+#!/bin/bash
+
+set -e
+
+ENTRY_DATE=$(date +%Y-%m-%d)
+echo "Performance build for entry date: $ENTRY_DATE"
+
+# run the script to generate the provision-quality dataset
+echo "Running generate_provision_quality.py for entry date: $ENTRY_DATE"
+python3 -c "from digital_land.commands import generate_provision_quality; generate_provision_quality()"
+OUTPUT_FILE="/tmp/performance/provision-quality/entry-date=${ENTRY_DATE}/provision-quality.parquet"
+
+if [[ -n "$COLLECTION_DATASET_BUCKET_NAME" ]]; then
+  if [[ -f "$OUTPUT_FILE" ]]; then
+    echo "Uploading $OUTPUT_FILE to s3"
+    aws s3 cp "$OUTPUT_FILE" "s3://${COLLECTION_DATASET_BUCKET_NAME}/performance/provision-quality/entry-date=${ENTRY_DATE}/provision-quality.parquet"
+  else
+    echo "ERROR: Output file $OUTPUT_FILE not found. Skipping S3 upload."
+    exit 1
+  fi
+else
+  echo "COLLECTION_DATASET_BUCKET_NAME not set. Skipping S3 upload."
+fi
+
+echo "Performance build complete"

task/requirements.txt

@@ -1 +1 @@
--e git+https://github.com/digital-land/pipeline.git@main#egg=digital-land
+-e git+https://github.com/digital-land/pipeline.git@provision-quality-dataset#egg=digital-land