2023-03-11 00:43:40 +01:00
|
|
|
#!/usr/bin/env bash
|
|
|
|
|
2024-04-25 10:55:15 -07:00
|
|
|
# Processes all the files from abfs://container1/ in AZURE_ACCOUNT_NAME account,
|
2023-07-31 13:20:10 -04:00
|
|
|
# using the `unstructured` library.
|
2023-03-11 00:43:40 +01:00
|
|
|
|
|
|
|
# Structured outputs are stored in azure-ingest-output/
|
|
|
|
|
2023-12-11 20:04:15 -05:00
|
|
|
SCRIPT_DIR=$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" &>/dev/null && pwd)
|
2023-03-11 00:43:40 +01:00
|
|
|
cd "$SCRIPT_DIR"/../../.. || exit 1
|
|
|
|
|
|
|
|
PYTHONPATH=. ./unstructured/ingest/main.py \
|
2023-12-18 23:48:21 -08:00
|
|
|
azure \
|
|
|
|
--remote-url abfs://container1/ \
|
2024-04-25 10:55:15 -07:00
|
|
|
--account-name "<AZURE_ACCOUNT_NAME>" \
|
2023-12-18 23:48:21 -08:00
|
|
|
--output-dir azure-ingest-output \
|
|
|
|
--num-processes 2
|