privacyforms-pdf

Python library for parsing and filling PDF forms using pypdf.

Features

• Parse fillable PDFs into a canonical PDFRepresentation schema
• Fill PDF forms from simple JSON key/value data
• Extract layout hints and visual row groupings
• Validate representation JSON against the schema
• Verify sample data keys against parsed field IDs
• Extend the CLI through pluggy command entry points

Requirements

• Python 3.12+
• pypdf >= 5

Installation

git clone <repo-url>
cd privacyforms.pdf
uv sync

CLI Quick Start

Parse A PDF

pdf-forms parse form.pdf -o representation.json

This writes a compact PDFRepresentation JSON document.

Verify A Representation JSON File

pdf-forms verify-json representation.json

Verify Sample Data Keys Against Parsed Field IDs

pdf-forms verify-data --form-json representation.json --data-json sample-data.json

Preferred format:

• use field IDs such as f-0 for canonical machine-facing data
• field names such as Candidate Name remain supported for convenience

Compatibility modes:

• fill-form accepts --field-keys name|id|auto
• verify-data accepts --key-mode name|id|auto
• auto accepts a mixture of field IDs and field names

Fill A PDF Form

pdf-forms fill-form form.pdf data.json -o filled.pdf
pdf-forms fill-form form.pdf data.json -o filled.pdf --no-validate
pdf-forms fill-form form.pdf data.json -o filled.pdf --strict
pdf-forms fill-form form.pdf data.json -o filled.pdf --field-keys id

Recommended fill-form payloads are keyed by field IDs:

{
  "f-0": "John Smith",
  "f-1": "Software Engineer",
  "f-2": "2025-06-01",
  "f-3": true
}

Field names and mixed key styles are still supported through --field-keys name and --field-keys auto.

Check Whether A PDF Contains A Form

pdf-forms info form.pdf

Python API

The package currently exposes two main API layers:

• read/parse APIs via parse_pdf() and extract_pdf_form()
• higher-level read/fill/validate APIs via PDFFormService

Parse A PDF Into PDFRepresentation

from privacyforms_pdf import extract_pdf_form

representation = extract_pdf_form("form.pdf")

print(representation.spec_version)
print(representation.source)
print(len(representation.fields))
print(len(representation.rows))

for field in representation.fields:
    print(field.id, field.name, field.type, field.value)

Extract Labels And Nearby Text

Install the optional labels dependency (requires PyMuPDF):

pip install privacyforms.pdf[labels]

CLI:

pdf-forms parse form.pdf --labels -o representation.json

Python API:

from privacyforms_pdf import PDFFormService

service = PDFFormService()
representation = service.extract("form.pdf", extract_labels=True)

for field in representation.fields:
    print(field.title)  # best inferred label
    for block in field.text_blocks:
        print(block.role, block.direction, block.text)

You can also call parse_pdf() directly:

from privacyforms_pdf import parse_pdf

representation = parse_pdf("form.pdf")
json_text = representation.to_compact_json()

Fill And Validate Forms

from privacyforms_pdf import PDFFormService

service = PDFFormService()

has_form = service.has_form("form.pdf")

form_data = {
    "f-0": "John Smith",
    "f-3": True,
}

errors = service.validate_form_data("form.pdf", form_data, key_mode="id")
if errors:
    print(errors)
else:
    service.fill_form("form.pdf", form_data, "filled.pdf", key_mode="id")

You can also fill from a JSON file:

from privacyforms_pdf import PDFFormService

service = PDFFormService()
service.fill_form_from_json("form.pdf", "data.json", "filled.pdf", key_mode="id")

The class also exposes read helpers:

from privacyforms_pdf import PDFFormService

service = PDFFormService()
representation = service.extract("form.pdf")
fields = service.list_fields("form.pdf")
field = service.get_field_by_id("form.pdf", "f-0")
value = service.get_field_value("form.pdf", "Candidate Name")
service.extract_to_json("form.pdf", "representation.json")

Public Objects

Primary exports from privacyforms_pdf:

• PDFFormService
• FormFiller
• parse_pdf
• extract_pdf_form
• PDFRepresentation
• PDFField
• FieldFlags
• FieldLayout
• FieldTextBlock
• FieldTextRole
• FieldTextDirection
• ChoiceOption
• RowGroup
• PDFFormError
• PDFFormNotFoundError
• FormValidationError
• FieldNotFoundError

PDFRepresentation Schema

Top-level fields:

• spec_version: str
• source: str | None
• fields: list[PDFField]
• rows: list[RowGroup]

PDFField

Main fields:

• name: str
• title: str | None
• id: str
• type: PDFFieldType
• field_flags: FieldFlags | None
• layout: FieldLayout | None
• default_value: str | bool | list[str] | None
• value: str | bool | list[str] | None
• choices: list[ChoiceOption]
• text_blocks: list[FieldTextBlock] — nearby labels, descriptions, helpers
• format: str | None
• max_length: int | None
• textarea_rows: int | None
• textarea_cols: int | None

Supported field types:

• textfield
• textarea
• datefield
• checkbox
• radiobuttongroup
• combobox
• listbox
• signature

FieldLayout

Layout hints are stored in integer PDF coordinates:

• page: int | None
• x: int | None
• y: int | None
• width: int | None
• height: int | None

FieldTextBlock

Nearby text associated with a field (populated when extract_labels=True):

• text: str
• role: "label" | "description" | "helper" | "instruction" | "unknown"
• direction: "left" | "right" | "above" | "below" | "inside" | "unknown"
• layout: FieldLayout | None
• distance: float | None

RowGroup

Visual rows derived from layout analysis:

• fields: list[PDFField | str]
• page_index: int

When serialized, row fields are emitted as field IDs.

JSON Shape

Example parsed representation:

{
  "source": "form.pdf",
  "fields": [
    {
      "name": "Candidate Name",
      "id": "f-0",
      "type": "textfield",
      "layout": {
        "page": 1,
        "x": 53,
        "y": 1077,
        "width": 361,
        "height": 27
      }
    }
  ],
  "rows": [
    {
      "fields": ["f-0"],
      "page_index": 1
    }
  ]
}

Notes:

• omitted fields are intentionally excluded by compact serialization
• field_flags only serializes flags set to true
• rows reference fields by ID in JSON

Exceptions

• PDFFormError: base exception for form-related failures
• PDFFormNotFoundError: raised when a PDF does not contain a form
• FormValidationError: raised when fill-time validation fails
• FieldNotFoundError: exported for compatibility and field lookup failures

Ratings

Security

• Symlinks are rejected for both reads and writes to prevent path-traversal issues
• PDF files are validated via magic-byte header check (%PDF) before parsing
• Input size limits guard against oversized PDFs (> 50 MB) and JSON (> 10 MB)
• JSON depth limits prevent stack exhaustion from malicious payloads

Architecture

• Clean separation of concerns: schema → parser → filler → extractor → cli
• Canonical PDFRepresentation schema (Pydantic v2) is the single source of truth
• CLI commands are loaded dynamically via pluggy entry points — easy to extend
• Low-level PDF writer (FormFiller) is decoupled from the high-level service (PDFFormService)

API Design

• Two complementary layers: function-based (parse_pdf, extract_pdf_form) and class-based (PDFFormService)
• Field IDs are the canonical key format; field names remain supported for convenience
• key_mode="auto" accepts mixed payloads of IDs and names
• All public methods have complete type hints and Google-style docstrings

Functionality

• Handles all common PDF form types: text, textarea, date, checkbox, radio, combo, listbox, signature
• Radio button state resolution works across different PDF generators
• Listbox filling includes custom appearance streams so selections are visible in viewers
• Graceful fallback when pypdf's appearance-stream generation hits edge cases

Code Quality

• 100% test coverage (426 tests) with pytest and pytest-cov
• Ruff enforces strict linting (E, W, F, I, N, D, UP, B, C4, SIM, TCH)
• ty type checker runs in strict mode — complete type hints throughout
• Bandit security scanner integrated; no high or medium severity issues

Development

Quality Checks

make check
make test
make test-cov

Project Structure

privacyforms.pdf/
├── privacyforms_pdf/
│   ├── __init__.py
│   ├── schema.py
│   ├── schema_layout.py
│   ├── parser.py
│   ├── extractor.py
│   ├── filler.py
│   ├── hooks.py
│   ├── cli.py
│   └── commands/
├── tests/
├── samples/
├── demo/
├── docs/
├── pyproject.toml
└── README.md

License

MIT