This article covers connecting Unstructured to Delta Tables in Amazon S3. For information about connecting Unstructured to Delta Tables in Databricks instead, see Delta Tables in Databricks.

Batch process all your records to store structured outputs in a Delta Table in an Amazon S3 bucket.

The requirements are as follows.

The following video shows how to fulfill the minimum set of Amazon S3 requirements to store Delta Tables:

The preceding video does not show how to create an AWS account.

For more information about requirements, see the following:

  • An AWS account. Create an AWS account.

  • An S3 bucket. Create an S3 bucket. Additional approaches are in the following video and in the how-to sections at the end of this page.

  • For authenticated bucket read access, the authenticated AWS IAM user must have at minimum the permissions of s3:ListBucket and s3:GetObject for that bucket. Learn how.

  • For bucket write access, authenticated access to the bucket must be enabled (anonymous access must not be enabled), and the authenticated AWS IAM user must have at minimum the permission of s3:PutObject for that bucket. Learn how.

  • For authenticated access, an AWS access key and secret access key for the authenticated AWS IAM user in the account. Create an AWS access key and secret access key.

  • If the target files are in the root of the bucket, the path to the bucket, formatted as protocol://bucket/ (for example, s3://my-bucket/). If the target files are in a folder, the path to the target folder in the S3 bucket, formatted as protocol://bucket/path/to/folder/ (for example, s3://my-bucket/my-folder/).

  • If the target files are in a folder, make sure the authenticated AWS IAM user has authenticated access to the folder as well. Enable authenticated folder access.

​
Add an access policy to an existing bucket

To use the Amazon S3 console to add an access policy that allows all authenticated AWS IAM users in the corresponding AWS account to read and write to an existing S3 bucket, do the following.

Your organization might have stricter bucket policy requirements. Check with your AWS account administrator if you are unsure.

  1. Sign in to the AWS Management Console.

  2. Open the Amazon S3 Console.

  3. Browse to the existing bucket and open it.

  4. Click the Permissions tab.

  5. In the Bucket policy area, click Edit.

  6. In the Policy text area, copy the following JSON-formatted policy. To change the following policy to restrict it to a specific user in the AWS account, change root to that specific username.

    In this policy, replace the following:

    • Replace <my-account-id> with your AWS account ID.
    • Replace <my-bucket-name> in two places with the name of your bucket.
    {
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "AllowAuthenticatedUsersInAccountReadWrite",
            "Effect": "Allow",
            "Principal": {
                "AWS": "arn:aws:iam::<my-account-id>:root"
            },
            "Action": [
                "s3:GetObject",
                "s3:PutObject",
                "s3:ListBucket",
                "s3:DeleteObject"
            ],
            "Resource": [
                "arn:aws:s3:::<my-bucket-name>",
                "arn:aws:s3:::<my-bucket-name>/*"
            ],
            "Condition": {
                "StringEquals": {
                    "aws:PrincipalType": "IAMUser"
                }
            }
        }
    ]
}

  7. Click Save changes.

​
Create a bucket with AWS CloudFormation

To use the AWS CloudFormation console to create an Amazon S3 bucket that allows all authenticated AWS IAM users in the corresponding AWS account to read and write to the bucket, do the following.

Your organization might have stricter bucket policy requirements. Check with your AWS account administrator if you are unsure.

  1. Save the following YAML to a file on your local machine, for example create-s3-bucket.yaml. To change the following bucket policy to restrict it to a specific user in the AWS account, change root to that specific username.

    AWSTemplateFormatVersion: '2010-09-09'
Description: 'CloudFormation template to create an S3 bucket with specific permissions for account users.'

Parameters:
  BucketName:
    Type: String
    Description: 'Name of the S3 bucket to create'

Resources:
  MyS3Bucket:
    Type: 'AWS::S3::Bucket'
    Properties:
      BucketName: !Ref BucketName
      PublicAccessBlockConfiguration:
        BlockPublicAcls: true
        BlockPublicPolicy: false
        IgnorePublicAcls: true
        RestrictPublicBuckets: true

  BucketPolicy:
    Type: 'AWS::S3::BucketPolicy'
    Properties:
      Bucket: !Ref MyS3Bucket
      PolicyDocument:
        Version: '2012-10-17'
        Statement:
          - Sid: AllowAllAuthenticatedUsersInAccount
            Effect: Allow
            Principal:
              AWS: !Sub 'arn:aws:iam::${AWS::AccountId}:root'
            Action:
              - 's3:GetObject'
              - 's3:PutObject'
              - 's3:ListBucket'
              - 's3:DeleteObject'
            Resource:
              - !Sub 'arn:aws:s3:::${BucketName}'
              - !Sub 'arn:aws:s3:::${BucketName}/*'

Outputs:
  BucketName:
    Description: 'Name of the created S3 bucket'
    Value: !Ref MyS3Bucket

  2. Sign in to the AWS Management Console.

  3. Open the AWS CloudFormation Console.

  4. Click Create stack > With new resources (standard).

  5. On the Create stack page, with Choose an existing template already selected, select Upload a template file.

  6. Click Choose file, and browse to and select the YAML file from your local machine.

  7. Click Next.

  8. Enter a unique Stack name and BucketName.

  9. Click Next two times.

  10. Click Submit.

  11. Wait until the Status changes to CREATE_COMPLETE.

  12. After the bucket is created, you can delete the YAML file, if you want.

​
Create a bucket with the AWS CLI

To use the AWS CLI to create an Amazon S3 bucket that allows all authenticated AWS IAM users in the corresponding AWS account to read and write to the bucket, do the following.

Your organization might have stricter bucket policy requirements. Check with your AWS account administrator if you are unsure.

  1. Install the AWS CLI.

  2. Set up the AWS CLI.

  3. Copy the following script to a file on your local machine, for example a file named create-s3-bucket.sh. To change the following bucket policy to restrict it to a specific user in the AWS account, change root to that specific username.

    In this script, replace the following:

    • Replace <my-account-id> with your AWS account ID.
    • Replace <my-unique-bucket-name> with the name of your bucket.
    • Replace <us-east-1> with your AWS Region.
    #!/bin/bash

# Set variables for the AWS account ID, Amazon S3 bucket name, and AWS Region.
ACCOUNT_ID="<my-account-id>"
BUCKET_NAME="<my-unique-bucket-name>"
REGION="<us-east-1>"

# Temporary filename for the bucket policy.
# Do not change this variable.
POLICY_FILE="bucket_policy.json"

# Create the bucket.
aws s3api create-bucket --bucket $BUCKET_NAME --region $REGION

# Wait for the bucket to exist.
echo "Waiting for bucket '$BUCKET_NAME' to be fully created..."
aws s3api wait bucket-exists --bucket $BUCKET_NAME

# Check if the wait command was successful.
if [ $? -eq 0 ]; then
    echo "The bucket '$BUCKET_NAME' has been fully created."
else
    echo "Error: Timed out waiting for bucket '$BUCKET_NAME' to be created."
    exit 1
fi

# Remove the "block public policy" bucket access setting.
aws s3api put-public-access-block \
    --bucket $BUCKET_NAME \
    --public-access-block-configuration \
    '{"BlockPublicPolicy": false, "IgnorePublicAcls": false, "BlockPublicAcls": false, "RestrictPublicBuckets": false}'

# Check if the operation was successful.
if [ $? -eq 0 ]; then
    echo "The block public policy access setting was removed from '$BUCKET_NAME'."
else
    echo "Error: Failed to remove the block public policy access setting from '$BUCKET_NAME'."
    exit 1
fi

# Create the bucket policy.
cat << EOF > $POLICY_FILE
{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "AllowAuthenticatedUsersInAccountReadWrite",
            "Effect": "Allow",
            "Principal": {
                "AWS": "arn:aws:iam::$ACCOUNT_ID:root"
            },
            "Action": [
                "s3:GetObject",
                "s3:PutObject",
                "s3:ListBucket",
                "s3:DeleteObject"
            ],
            "Resource": [
                "arn:aws:s3:::$BUCKET_NAME",
                "arn:aws:s3:::$BUCKET_NAME/*"
            ],
            "Condition": {
                "StringEquals": {
                    "aws:PrincipalType": "IAMUser"
                }
            }
        }
    ]
}
EOF

# Apply the bucket policy.
aws s3api put-bucket-policy --bucket $BUCKET_NAME --policy file://$POLICY_FILE

# Check if the policy application was successful.
if [ $? -eq 0 ]; then
    echo "The bucket policy was applied to '$BUCKET_NAME'."
else
    echo "Error: Failed to apply the bucket policy to '$BUCKET_NAME'."
    exit 1
fi

# Verify the applied policy.
echo "Verifying the applied policy:"
aws s3api get-bucket-policy --bucket $BUCKET_NAME --query Policy --output text

# Remove the temporary bucket policy file.
rm $POLICY_FILE

  4. Run the script, for example:

    sh create-s3-bucket.sh

  5. After the bucket is created, you can delete the script file, if you want.

​
Delta table output format

A Delta table consists of Parquet files that contain data and a transaction log that stores metadata about the transactions. Learn more.

The Delta Tables in Amazon S3 destination connector generates the following output within the specified path to the S3 bucket (or the specified folder within the bucket):

  • Initially, one Parquet (.parquet) file per file in the source location. For example, for a file in the source location named my-file.pdf, an associated file with the extension .parquet is generated. Various kinds of file transactions can result in additional Parquet files being generated. These Parquet filenames are automatically generated by the Delta Lake engine and are not meant to be manually modified.
  • A folder named _delta_log that contains metadata and change history about the .parquet files. As Parquet files are added to, changed, or removed from the specified bucket or folder path, the _delta_log folder is updated with any related metadata and change history details.

Together, this set of Parquet files and their associated _delta_log folder (and its contents) describe a single, versioned Delta table. Because of this, Unstructured recommends the following usage best practices:

  • In the source location, each set of source files that is to be considered as a unit for change management purposes should be controlled by a unique, dedicated Delta Tables in S3 destination connector. This connector should reference a unique, dedicated output folder within the bucket. Having multiple workflows refer to different sets of source files, yet all share the same Delta table, could results in data loss or table corruption.
  • Avoid directly modifying, adding, or deleting Parquet data files or the _delta_log folder within a Delta table’s directory. This can lead to data loss or table corruption.
  • If you need to copy or move a Delta table to a different location, you must move or copy its entire set of Parquet files and its associated _delta_log folder (and its contents) together as a unit. Note that the copied or moved Delta table will no longer be controlled by the original Delta Tables in S3 destination connector.

The Delta Table connector dependencies for Amazon S3:

CLI, Python
pip install "unstructured-ingest[delta-table]"

You might also need to install additional dependencies, depending on your needs. Learn more.

The following environment variables:

  • AWS_S3_URL - The path to the S3 bucket or folder, formatted as s3://my-bucket/ (if the files are in the bucket’s root) or s3://my-bucket/my-folder/, represented by --table-uri (CLI) or table_uri (Python).
  • AWS_ACCESS_KEY_ID - The AWS access key ID for the authenticated AWS IAM user, represented by --aws-access-key-id (CLI) or aws_access_key (Python).
  • AWS_SECRET_ACCESS_KEY - The corresponding AWS secret access key, represented by --aws-secret-access-key (CLI) or aws_secret_access_key (Python).

Now call the Unstructured Ingest CLI or the Unstructured Ingest Python library. The source connector can be any of the ones supported. This example uses the local source connector.

This example sends files to Unstructured for processing by default. To process files locally instead, see the instructions at the end of this page.

#!/usr/bin/env bash

# Chunking and embedding are optional.

unstructured-ingest \
  local \
    --input-path $LOCAL_FILE_INPUT_DIR \
    --partition-by-api \
    --api-key $UNSTRUCTURED_API_KEY \
    --partition-endpoint $UNSTRUCTURED_API_URL \
    --strategy hi_res \
    --additional-partition-args="{\"split_pdf_page\":\"true\", \"split_pdf_allow_failed\":\"true\", \"split_pdf_concurrency_level\": 15}" \
    --chunking-strategy by_title \
    --embedding-provider huggingface \
  delta-table \
    --aws-access-key-id $AWS_ACCESS_KEY_ID \
    --aws-secret-access-key $AWS_SECRET_ACCESS_KEY \
    --table-uri $AWS_S3_URL

For the Unstructured Ingest CLI and the Unstructured Ingest Python library, you can use the --partition-by-api option (CLI) or partition_by_api (Python) parameter to specify where files are processed:

  • To do local file processing, omit --partition-by-api (CLI) or partition_by_api (Python), or explicitly specify partition_by_api=False (Python).

    Local file processing does not use an Unstructured API key or API URL, so you can also omit the following, if they appear:

    • --api-key $UNSTRUCTURED_API_KEY (CLI) or api_key=os.getenv("UNSTRUCTURED_API_KEY") (Python)
    • --partition-endpoint $UNSTRUCTURED_API_URL (CLI) or partition_endpoint=os.getenv("UNSTRUCTURED_API_URL") (Python)
    • The environment variables UNSTRUCTURED_API_KEY and UNSTRUCTURED_API_URL

  • To send files to the Unstructured Partition Endpoint for processing, specify --partition-by-api (CLI) or partition_by_api=True (Python).

    Unstructured also requires an Unstructured API key and API URL, by adding the following:

    • --api-key $UNSTRUCTURED_API_KEY (CLI) or api_key=os.getenv("UNSTRUCTURED_API_KEY") (Python)
    • --partition-endpoint $UNSTRUCTURED_API_URL (CLI) or partition_endpoint=os.getenv("UNSTRUCTURED_API_URL") (Python)
    • The environment variables UNSTRUCTURED_API_KEY and UNSTRUCTURED_API_URL, representing your API key and API URL, respectively.

    You must specify the API URL only if you are not using the default API URL for Unstructured Ingest, for example, if you are using a self-hosted instance of the Unstructured API.

    The default API URL for Unstructured Ingest is https://api.unstructuredapp.io/general/v0/general, which is the API URL for the Unstructured Partition Endpoint. However, you should always use the URL that was provided to you when your Unstructured account was created. If you do not have this URL, contact Unstructured Sales at sales@unstructured.io.

    If you do not have an API key, get one now.

    If the Unstructured API is self-hosted, the process for generating Unstructured API keys, and the Unstructured API URL that you use, are different. For details, contact Unstructured Sales at sales@unstructured.io.