2023-08-11 22:02:51 +03:00
|
|
|
#!/usr/bin/env bash
|
|
|
|
set -e
|
|
|
|
|
|
|
|
# Description: This test checks if all the processed content is the same as the expected outputs.
|
|
|
|
# Also checks if a large table can be ingested properly.
|
|
|
|
|
2023-11-01 15:23:44 -04:00
|
|
|
SRC_PATH=$(dirname "$(realpath "$0")")
|
|
|
|
SCRIPT_DIR=$(dirname "$SRC_PATH")
|
2023-08-11 22:02:51 +03:00
|
|
|
cd "$SCRIPT_DIR"/.. || exit 1
|
|
|
|
|
|
|
|
OUTPUT_FOLDER_NAME=airtable-diff
|
2023-11-02 16:41:56 -05:00
|
|
|
OUTPUT_ROOT=${OUTPUT_ROOT:-$SCRIPT_DIR}
|
|
|
|
OUTPUT_DIR=$OUTPUT_ROOT/structured-output/$OUTPUT_FOLDER_NAME
|
|
|
|
WORK_DIR=$OUTPUT_ROOT/workdir/$OUTPUT_FOLDER_NAME
|
2023-08-11 22:02:51 +03:00
|
|
|
DOWNLOAD_DIR=$SCRIPT_DIR/download/$OUTPUT_FOLDER_NAME
|
2023-10-02 16:47:24 -04:00
|
|
|
CI=${CI:-"false"}
|
2023-09-28 21:48:19 -07:00
|
|
|
max_processes=${MAX_PROCESSES:=$(python3 -c "import os; print(os.cpu_count())")}
|
2023-08-11 22:02:51 +03:00
|
|
|
|
2023-09-21 14:51:08 -04:00
|
|
|
# shellcheck disable=SC1091
|
|
|
|
source "$SCRIPT_DIR"/cleanup.sh
|
2023-10-02 16:47:24 -04:00
|
|
|
function cleanup() {
|
2023-12-18 23:48:21 -08:00
|
|
|
cleanup_dir "$OUTPUT_DIR"
|
|
|
|
cleanup_dir "$WORK_DIR"
|
|
|
|
if [ "$CI" == "true" ]; then
|
|
|
|
cleanup_dir "$DOWNLOAD_DIR"
|
|
|
|
fi
|
2023-10-02 16:47:24 -04:00
|
|
|
}
|
|
|
|
trap cleanup EXIT
|
2023-09-21 14:51:08 -04:00
|
|
|
|
2023-08-11 22:02:51 +03:00
|
|
|
VARIED_DATA_BASE_ID="app5YQxSfp220fWtm"
|
|
|
|
VARIED_DATA_BASE_ID_2="appJ43QmP8I17zu88"
|
|
|
|
|
|
|
|
if [ -z "$AIRTABLE_PERSONAL_ACCESS_TOKEN" ]; then
|
2023-12-18 23:48:21 -08:00
|
|
|
echo "Skipping Airtable ingest test because the AIRTABLE_PERSONAL_ACCESS_TOKEN is not set."
|
|
|
|
exit 8
|
2023-08-11 22:02:51 +03:00
|
|
|
fi
|
|
|
|
|
2023-11-02 16:41:56 -05:00
|
|
|
RUN_SCRIPT=${RUN_SCRIPT:-./unstructured/ingest/main.py}
|
|
|
|
PYTHONPATH=${PYTHONPATH:-.} "$RUN_SCRIPT" \
|
2023-12-18 23:48:21 -08:00
|
|
|
airtable \
|
|
|
|
--download-dir "$DOWNLOAD_DIR" \
|
|
|
|
--personal-access-token "$AIRTABLE_PERSONAL_ACCESS_TOKEN" \
|
|
|
|
--list-of-paths "$VARIED_DATA_BASE_ID $VARIED_DATA_BASE_ID_2" \
|
|
|
|
--metadata-exclude filename,file_directory,metadata.data_source.date_processed,metadata.date,metadata.last_modified,metadata.detection_class_prob,metadata.parent_id,metadata.category_depth,metadatda.languages \
|
|
|
|
--num-processes "$max_processes" \
|
|
|
|
--preserve-downloads \
|
|
|
|
--reprocess \
|
|
|
|
--output-dir "$OUTPUT_DIR" \
|
|
|
|
--work-dir "$WORK_DIR" \
|
|
|
|
--max-retry-time 10 \
|
|
|
|
--verbose
|
2023-08-11 22:02:51 +03:00
|
|
|
|
2023-09-21 14:51:08 -04:00
|
|
|
"$SCRIPT_DIR"/check-diff-expected-output.sh $OUTPUT_FOLDER_NAME
|