Build document workflows with Python

01 · GET STARTED

Install eDocAPI

Use Python 3.10–3.13 and a virtual environment. Install the core package first; add optional extras only when your project needs them.

python -m venv .venv
# Windows PowerShell
.\.venv\Scripts\Activate.ps1
# macOS / Linux
# source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install edocapi

eDocAPI has two main building blocks: Document processes files in Python, while App exposes operations as HTTP endpoints.

02 · THE DOCUMENT TOOLKIT

Process a file directly

Create a Document from a path, inspect it, and call the operation you need. File outputs are returned as Documents; text and metadata operations return Python values.

from edocapi import Document

doc = Document("report.docx")
print(doc.type, doc.size)

pdf = doc.to_pdf()
print("PDF:", pdf.path)

text = doc.extract_text()
details = doc.info()
print(text[:500])
print(details)

03 · FORMATS AND OPTIONAL EXTRAS

Choose the operation that fits the file

Capabilities depend on input format. Check Document.supported_types() in code and install extras for enhanced rendering or OCR.

python -m pip install "edocapi[html]"        # enhanced HTML / Markdown PDF
python -m pip install "edocapi[pdf-images]"  # PDF page images (also install Poppler)
python -m pip install "edocapi[ml]"          # Tesseract OCR helpers
python -m pip install "edocapi[structured-ocr]" # PaddleOCR; Python 3.13 or earlier
python -m pip install "edocapi[server]"      # Uvicorn server

04 · BUILD AN HTTP API

Expose your workflow with App

Decorate Python functions with @app.get or @app.post. Upload fields such as file and files are provided to your handler; returning a Document sends a file response, and returning a dictionary sends JSON.

from edocapi import App, Document

app = App(max_file_size="10MB")

@app.post("/convert")
def convert(file):
    return Document(file).to_pdf()

@app.post("/extract")
def extract(file):
    return {"text": Document(file).extract_text()}

@app.post("/compress")
def compress(file, target_size: int | None = None):
    return Document(file).compress(target_size=target_size)

@app.post("/ocr")
def ocr(file, lang: str = "eng"):
    return {"text": Document(file).ocr(lang=lang)}

OCR’s HTTP route is disabled by default; enable it with App(enable_ocr=True). The library OCR method remains available independently of that route setting.

05 · LIMITS AND SAFE HANDLING

Set boundaries before you serve uploads

The framework validates file content, checks file signatures, sanitizes names, and cleans managed temporary files. Configure a smaller upload limit and rate limit for your own workload.

app = App(max_file_size="10MB", max_request_size="30MB")
app.limits(100, per="minute")

@app.post("/ocr", limit="5/minute")
def ocr(file):
    return {"text": Document(file).ocr()}

• 400 — missing upload or invalid document.
• 413 — file or full request body exceeds its configured limit.
• 415 — unsupported file type or signature mismatch.
• 429 — rate limit exceeded; check the Retry-After header.

Treat uploads as untrusted input. Renderers block local and remote resource fetching. Don’t raise size limits or disable validation unless your service has a clear reason and the capacity to handle it.

06 · CLI AND DEPLOYMENT

Use the same tools from your terminal

edocapi --version
edocapi convert report.docx -o report.pdf
edocapi extract report.pdf
edocapi info image.png
edocapi ocr scan.png --lang eng
edocapi compress large.pdf --target 500KB

# Start your App locally
uvicorn main:app --reload

For deployment, install edocapi[server] and run your ASGI app with Uvicorn bound to 0.0.0.0 and the port supplied by your host. Keep debug mode off in production and set proxy request limits to match the app.

07 · CONTINUE LEARNING

Work through the complete course

The textbook in the assets folder is a hands-on course based on eDocAPI 0.2.0. It covers the Document class, supported formats, OCR, App routes, limits, errors, the CLI, security, testing, and three capstone projects. Type and run the exercises as you go.