Extract text from PDF files without any problems

Introduction

In the ever-evolving landscape of data processing, extracting structured information from PDF files remains a formidable challenge, even in 2024. While numerous models excel at question-answering tasks, the real complexity lies in transforming unstructured PDF content into organized, actionable data. Let’s explore this challenge and find out how Indexify and PaddleOCR can be the tools we need to extract text from PDF files seamlessly.

Revelation: We really did fix it! cmd/ctrl+f And search for the term spotlight to find out how!

PDF file extraction is essential in several areas. Let's look at some common use cases:

• Invoices and receiptsThese documents vary widely in format and contain complex layouts, tables, and sometimes handwritten notes. Accurate analysis is essential to automating accounting processes.
• Academic papers and thesesThese typically include a combination of text, graphics, tables, and formulas. The challenge is to correctly convert not only text, but also mathematical equations and scientific notation.
• Legal documentsContracts and court documents are often riddled with formatting nuances. Maintaining the integrity of the original format when extracting text is critical for reviews and legal compliance.
• Historical archives and manuscriptsThese documents present unique challenges due to paper degradation, variations in historical handwriting, and archaic language. OCR technology must handle these variations for effective research and archival purposes.
• Medical history and prescriptionsThese documents often contain handwritten notes and important medical terminology. Accurately capturing this information is vital for patient care and medical research.

Of course! Here is the revised text in active voice:

Indexify is an open-source data framework that addresses the complexities of extracting unstructured data from any source, as shown in Figure 1. Its architecture supports:

• Ingesting millions of unstructured data points.
• Real-time indexing and extraction pipelines.
• Horizontal scalability to adapt to growing data volumes.
• Rapid extraction times (within seconds after ingestion).
• Flexible deployment on multiple hardware platforms (GPU, TPU, and CPU).

If you're interested in reading more about indexify and how to set it up for extraction, read our 2-minute article 'ai/getting_started/” target=”_blank” rel=”noreferrer noopener nofollow”>starting' guide.

At the heart of Indexify are its extractors (as shown in Figure 2) – compute functions that transform unstructured data or extract insights from it. These extractors can be deployed to run on any hardware, and a single Indexify deployment supports tens of thousands of extractors on a cluster.

Currently, Indexify supports several extractors for multiple modalities (as shown in Figure 3). The complete list of Indexify extractors along with their use cases can be found at ai/apis/extractors/” target=”_blank” rel=”noreferrer noopener nofollow”>documentation.

PaddleOCR PDF Extractor, based on the PaddleOCR library, is a powerful tool in the Indexify ecosystem. It integrates several OCR algorithms for text detection (DB, EAST, SAST) and recognition (CRNN, RARE, StarNet, Rosetta, SRN).

Let's review the setup and usage of the PaddleOCR extractor:

Below is an example of creating a pipeline that extracts text, tables, and images from a PDF document.

You will need three different terminals open to complete this tutorial:

• Terminal 1 to download and run the Indexify server.
• Terminal 2 to run our Indexify extractors that will handle structured extraction, chunking, and embedding of the ingested pages.
• Terminal 3 to run our Python scripts to help load and query data from our Indexify server.

Step 1: Start the Indexify server

First, let's start by downloading the Indexify server and running it.

Terminal 1

curl https://getindexify.ai | sh
./indexify server -d

Let's start by creating a new virtual environment before installing the necessary packages into our virtual environment.

Terminal 2

python3 -m venv venv
source venv/bin/activate
pip3 install indexify-extractor-sdk indexify

Then we can run all the available extractors using the following command.

!indexify-extractor download tensorlake/paddleocr_extractor
!indexify-extractor join-server

Terminal 3

!python3 -m venv venv
!source venv/bin/activate

Create a Python script that defines the extraction graph and run it. Steps 3 through 5 in this subsection should be part of the same Python file that should be run after activating venv in Terminal 3.

from indexify import IndexifyClient, ExtractionGraph
client = IndexifyClient()
extraction_graph_spec = """
name: 'pdflearner'
extraction_policies:
   - extractor: 'tensorlake/paddleocr_extractor'
     name: 'pdf_to_text'
"""
extraction_graph = ExtractionGraph.from_yaml(extraction_graph_spec)
client.create_extraction_graph(extraction_graph)

This code sets up an extraction graph called 'pdflearner' that uses the PaddleOCR extractor to convert PDF files to text.

Step 4: Upload PDF files from your app

content_id = client.upload_file("pdflearner", "/path/to/pdf.file")

client.wait_for_extraction(content_id)

extracted_content = client.get_extracted_content(content_id=content_id, graph_name="pdflearner", policy_name="pdf_to_text")
print(extracted_content)

This snippet loads a PDF, waits for the extraction to complete, and then retrieves and prints the extracted content.

We did not believe it was possible to meaningfully extract all the textual information with a simple, few-step process, so we tested it with an actual tax invoice (as shown in Figure 4).

(Content(content_type="text/plain", data=b"Form 1040\nForms W-2 & W-2G Summary\n2023\nKeep for your records\n Name(s) Shown on Return\nSocial Security Number\nJohn H & Jane K Doe\n321-12-3456\nEmployer\nSP\nFederal Tax\nState Wages\nState Tax\nForm W-2\nWages\nAcmeware Employer\n143,433.\n143,433.\n1,000.\nTotals.\n143,433.\n143,433.\n1,000.\nForm W-2 Summary\nBox No.\nDescription\nTaxpayer\nSpouse\nTotal\nTotal wages, tips and compensation:\n1\na\nW2 box 1 statutory wages reported on Sch C\nW2 box 1 inmate or halfway house wages .\n6\nc\nAll other W2 box 1 wages\n143,433.\n143,433.\nd\nForeign wages included in total wages\ne\n0.\n0.\n2\nTotal federal tax withheld\n 3 & 7 Total social security wages/tips .\n143,566.\n143,566.\n4\nTotal social security tax withheld\n8,901.\n8,901.\n5\nTotal Medicare wages and tips\n143,566.\n143,566.\n6\nTotal Medicare tax withheld . :\n2,082.\n2,082.\n8\nTotal allocated tips .\n9\nNot used\n10 a\nTotal dependent care benefits\nb\nOffsite dependent care benefits\nc\nOnsite dependent care benefits\n11\n Total distributions from nonqualified plans\n12 a\nTotal from Box 12\n3,732.\n3,732.\nElective deferrals to qualified plans\n133.\n133.\nc\nRoth contrib. to 401(k), 403(b), 457(b) plans .\n.\n1 Elective deferrals to government 457 plans\n2 Non-elective deferrals to gov't 457 plans .\ne\nDeferrals to non-government 457 plans\nf\nDeferrals 409A nonqual deferred comp plan .\n6\nIncome 409A nonqual deferred comp plan .\nh\nUncollected Medicare tax :\nUncollected social security and RRTA tier 1\nj\nUncollected RRTA tier 2 . . .\nk\nIncome from nonstatutory stock options\nNon-taxable combat pay\nm\nQSEHRA benefits\nTotal other items from box 12 .\nn\n3,599.\n3,599.\n14 a\n Total deductible mandatory state tax .\nb\nTotal deductible charitable contributions\nc\nTotal state deductible employee expenses .\nd\n Total RR Compensation .\ne\nTotal RR Tier 1 tax .\nf\nTotal RR Tier 2 tax . -\nTotal RR Medicare tax .\ng\nh\nTotal RR Additional Medicare tax .\ni\nTotal RRTA tips. : :\nj\nTotal other items from box 14\nk\nTotal sick leave subject to $511 limit\nTotal sick leave subject to $200 limit\nm\nTotal emergency family leave wages\n16\nTotal state wages and tips .\n143,433.\n143,433.\n17\nTotal state tax withheld\n1,000.\n1,000.\n19\nTotal local tax withheld .", features=(Feature(feature_type="metadata", name="metadata", value={'type': 'text'}, comment=None)), labels={}))

While extracting text is useful, we often need to parse this text to turn it into structured data. Here’s how you can use Indexify to extract specific fields from your PDF files (the full workflow is shown in Figure 5).

from indexify import IndexifyClient, ExtractionGraph, SchemaExtractorConfig, Content, SchemaExtractor
client = IndexifyClient()
schema = {
    'properties': {
        'invoice_number': {'title': 'Invoice Number', 'type': 'string'},
 'date': {'title': 'Date', 'type': 'string'},
        'account_number': {'title': 'Account Number', 'type': 'string'},
        'owner': {'title': 'Owner', 'type': 'string'},
        'address': {'title': 'Address', 'type': 'string'},
        'last_month_balance': {'title': 'Last Month Balance', 'type': 'string'},
        'current_amount_due': {'title': 'Current Amount Due', 'type': 'string'},
        'registration_key': {'title': 'Registration Key', 'type': 'string'},
 'due_date': {'title': 'Due Date', 'type': 'string'}
    },
    'required': ('invoice_number', 'date', 'account_number', 'owner', 'address', 'last_month_balance', 'current_amount_due', 'registration_key', 'due_date')
    'title': 'User',
    'type': 'object'
}
examples = str((
    {
        "type": "object",
        "properties": {
            "employer_name": {"type": "string", "title": "Employer Name"},
            "employee_name": {"type": "string", "title": "Employee Name"},
            "wages": {"type": "number", "title": "Wages"},
            "federal_tax_withheld": {"type": "number", "title": "Federal Tax
Withheld"},
 "state_wages": {"type": "number", "title": "State Wages"},
            "state_tax": {"type": "number", "title": "State Tax"}
        },
        "required": ("employer_name", "employee_name", "wages",
"federal_tax_withheld", "state_wages", "state_tax")
    },
    {
        "type": "object",
        "properties": {
            "booking_reference": {"type": "string", "title": "Booking Reference"},
            "passenger_name": {"type": "string", "title": "Passenger Name"},
            "flight_number": {"type": "string", "title": "Flight Number"},
            "departure_airport": {"type": "string", "title": "Departure Airport"},
            "arrival_airport": {"type": "string", "title": "Arrival Airport"},
            "departure_time": {"type": "string", "title": "Departure Time"},
            "arrival_time": {"type": "string", "title": "Arrival Time"}        },
        "required": ("booking_reference", "passenger_name", "flight_number","departure_airport", "arrival_airport", "departure_time", "arrival_time")
    }
))
extraction_graph_spec = """
name: 'invoice-learner'
extraction_policies:
  - extractor: 'tensorlake/paddleocr_extractor'
    name: 'pdf-extraction'
  - extractor: 'schema_extractor'
    name: 'text_to_json'
    input_params:
      service: 'openai'
      example_text: {examples}
content_source: 'invoice-learner'
"""
extraction_graph = ExtractionGraph.from_yaml(extraction_graph_spec)
client.create_extraction_graph(extraction_graph)
content_id = client.upload_file("invoice-learner", "/path/to/pdf.pdf")
print(content_id)
client.wait_for_extraction(content_id)
extracted_content = client.get_extracted_content(content_id=content_id, graph_name="invoice-learner", policy_name="text_to_json")
print(extracted_content)

This advanced example demonstrates how to chain multiple extractors together. PaddleOCR is first used to extract text from the PDF, and then a schema extractor is applied to parse the text and convert it into structured JSON data according to the defined schema.

The schema extractor is interesting because it allows both using the schema and inferring it from the chosen language model through few-shot learning.

To do this, we pass some examples of how the schema should look like via the example_text parameter. The clearer and more detailed the examples are, the better the inferred schema will be.

Let's investigate the result of this design:

(Content(content_type="text/plain", data=b'{"Form":"1040","Forms W-2 & W-2G Summary":{"Year":2023,"Keep for your records":true,"Name(s) Shown on Return":"John H & Jane K Doe","Social Security Number":"321-12-3456","Employer":{"Name":"Acmeware Employer","Federal Tax":"SP","State Wages":143433,"State Tax":1000},"Totals":{"Wages":143433,"State Wages":143433,"State Tax":1000}},"Form W-2 Summary":{"Box No.":{"Description":{"Taxpayer":"John H Doe","Spouse":"Jane K Doe","Total":"John H & Jane K Doe"}},"Total wages, tips and compensation":{"W2 box 1 statutory wages reported on Sch C":143433,"W2 box 1 inmate or halfway house wages":0,"All other W2 box 1 wages":143433,"Foreign wages included in total wages":0},"Total federal tax withheld":0,"Total social security wages/tips":143566,"Total social security tax withheld":8901,"Total Medicare wages and tips":143566,"Total Medicare tax withheld":2082,"Total allocated tips":0,"Total dependent care benefits":{"Offsite dependent care benefits":0,"Onsite dependent care benefits":0},"Total distributions from nonqualified plans":0,"Total from Box 12":{"Elective deferrals to qualified plans":3732,"Roth contrib. to 401(k), 403(b), 457(b) plans":133,"Elective deferrals to government 457 plans":0,"Non-elective deferrals to gov't 457 plans":0,"Deferrals to non-government 457 plans":0,"Deferrals 409A nonqual deferred comp plan":0,"Income 409A nonqual deferred comp plan":0,"Uncollected Medicare tax":0,"Uncollected social security and RRTA tier 1":0,"Uncollected RRTA tier 2":0,"Income from nonstatutory stock options":0,"Non-taxable combat pay":0,"QSEHRA benefits":0,"Total other items from box 12":3599},"Total deductible mandatory state tax":0,"Total deductible charitable contributions":0,"Total state deductible employee expenses":0,"Total RR Compensation":0,"Total RR Tier 1 tax":0,"Total RR Tier 2 tax":0,"Total RR Medicare tax":0,"Total RR Additional Medicare tax":0,"Total RRTA tips":0,"Total other items from box 14":0,"Total sick leave subject to $511 limit":0,"Total sick leave subject to $200 limit":0,"Total emergency family leave wages":0,"Total state wages and tips":143433,"Total state tax withheld":1000,"Total local tax withheld":0}}, features=(Feature(feature_type="metadata", name="text", value={'model': 'gpt-3.5-turbo-0125', 'completion_tokens': 204, 'prompt_tokens': 692}, comment=None)), labels={}))

Yes, it is difficult to read, so let us expand it for you in Figure 6.

This means that after this step, our textual data is successfully extracted into a structured JSON format. The data can have a complex layout, unevenly spaced, horizontal orientation, vertical orientation, diagonal orientation, large fonts, small fonts… It doesn't matter the layout, it just works!

Well, that pretty much solves the problem we set out to solve. We can finally shout Mission Accomplished, Tom Cruise style!

While the PaddleOCR extractor is powerful for extracting text from PDF files, Indexify’s real strength lies in its ability to chain multiple extractors together, creating sophisticated data processing pipelines. Let’s dive into why you might want to use additional extractors and how Indexify makes this process simple and efficient.

Indexify extraction graphs allow you to apply a sequence of extractors to ingested content on an ongoing basis. Each step in an extraction graph is known as an extraction policy. This approach offers several advantages:

• Modular processing:Break down complex extraction tasks into smaller, more manageable steps.
• Flexibility:Easily modify or replace individual extractors without affecting the entire plumbing.
• Efficiency:Process data in streaming mode, reducing latency and resource usage.

Lineage tracing

Indexify traces the lineage of transformed content and features extracted from the source. This feature is crucial for:

• Governance data:Understand how your data has been processed and transformed.
• Depuration:Easily trace problems back to their source.
• Compliance:Meet regulatory requirements by maintaining a clear audit trail of data transformations.

While PaddleOCR excels at text extraction, other extractors can add significant value to your data processing process.

Why choose Indexify?

Indexify shines in scenarios where:

• You are dealing with a large volume of documents (>1000).
• Your data volume grows over time.
• You need reliable and available ingestion channels.
• You are working with multimodal data or combining multiple models in a single pipeline.
• Your app's user experience depends on up-to-date data.

Conclusion

Extracting structured data from PDF files doesn’t have to be a headache. With Indexify and a variety of powerful extractors like PaddleOCR, you can streamline your workflow, handle large volumes of documents, and extract structured, meaningful data with ease. Whether you’re processing invoices, academic papers, or any other type of PDF document, Indexify gives you the tools you need to turn unstructured data into valuable insights.

Ready to streamline your PDF extraction process? Try Indexify and experience the ease of intelligent, scalable data extraction.