epo-oa-cli 0.2.3

EPO patent prosecution history CLI — download, parse, and analyze European patent documents for AI-assisted analysis

epo-oa-cli

EPO patent prosecution history CLI — Download, parse, and analyze European Patent Office (EPO) prosecution documents for AI-assisted patent analysis.

pip install epo-oa-cli
epo-oa run EP21841218

---

Overview

epo-oa fetches the complete prosecution history of any EP patent from the EPO Register, extracts PDF text (with optional OCR), and generates a structured prosecution.md file ready for AI analysis (Claude, GPT-4, etc.).

epo-oa run EP21841218
  → Downloads 40 documents as ZIP
  → Extracts & parses toc.xml
  → Generates file/EP21841218/EP21841218_prosecution.md

---

Installation

pip install epo-oa-cli

# With OCR support (for image-based PDFs)
pip install "epo-oa-cli[ocr]"

Requires Python 3.12+.

---

Quick Start

# 1. List all documents
epo-oa list EP21841218

# 2. Download as ZIP + extract
epo-oa download EP21841218

# 3. Parse PDFs → prosecution.md
epo-oa extract EP21841218

# 4. All-in-one
epo-oa run EP21841218

With OCR (for image-based PDFs)

EPO PDFs are full-page image scans. Run OCR first to embed text into the analysis file:

# OCR key documents only
epo-oa ocr EP21841218 --codes 1703,1224,ABEX

# OCR all documents
epo-oa ocr EP21841218

# Extract with OCR text embedded
epo-oa extract EP21841218 --with-ocr

---

Commands

Command	Description
epo-oa list <EP>	List prosecution documents from EPO Register
epo-oa download <EP>	Download all documents as ZIP archive
epo-oa extract <EP>	Parse PDFs → prosecution.md / prosecution.json
epo-oa ocr <EP>	OCR image-based PDFs → searchable *_ocr.pdf
epo-oa run <EP>	Download + extract in one step
epo-oa configure	Set proxy / CA-cert options

Options

epo-oa list EP21841218 --format json          # JSON output
epo-oa download EP21841218 --force            # Re-download
epo-oa extract EP21841218 --format json       # JSON output
epo-oa extract EP21841218 --with-ocr          # Embed OCR text
epo-oa ocr EP21841218 --codes 1703,ABEX       # Selective OCR
epo-oa ocr EP21841218 --in-place              # Overwrite originals

---

Proxy & SSL Configuration

For corporate networks or environments that require a proxy or custom CA certificate:

epo-oa configure

This interactively prompts for:

• HTTPS proxy URL — e.g. http://proxy.corp.example.com:8080
• HTTP proxy URL — e.g. http://proxy.corp.example.com:8080
• CA bundle file path — path to a custom .pem / .crt file

Settings are saved to ~/.epo-oa.toml:

[proxy]
https = "http://proxy.corp.example.com:8080"
http  = "http://proxy.corp.example.com:8080"

[ssl]
ca_bundle = "/etc/ssl/certs/corp-ca.pem"

If no config file exists, requests falls back to the standard environment variables (HTTPS_PROXY, HTTP_PROXY, REQUESTS_CA_BUNDLE).

---

Output: prosecution.md

The generated markdown file is structured for AI agents:

# EPO Prosecution Analysis — EP21841218

## Summary
| Item | Count |
|------|-------|
| Total documents | 40 |
| 🔴 Office Actions | 2 |
| 🔵 Amendments | 13 |
| ✅ Grant / Decision | 8 |

## Timeline
| Date | Cat | Document | File |
|------|-----|----------|------|
| 2023-10-30 | 🔍 | European Search Opinion (1703) 🖼️ | ... |
| 2024-02-15 | 🔵 | Amended Claims (CLMSABEX) 🖼️ | ... |
| 2026-02-05 | ✅ | Decision to Grant (2006A) 🖼️ | ... |

## 🔴 Office Action Documents
### European Search Opinion — 2023-10-30
**OCR Text:**
```text
D1 WO 2020/138918 A1 (SAMSUNG ELECTRONICS CO LTD)
1.1 D1 discloses an electronic device with the following features...
` `` `

---

Politeness & Rate Limiting

This tool accesses a public EPO server. It enforces:

• Random delays (1.5–3.0s) between requests
• Browser-like headers
• ZIP archive download (minimises HTTP requests)

Please do not run this tool in tight loops or CI pipelines without appropriate throttling.

---

Document Categories

Icon	Category	Description
🔴	Office Action	Examination notices, search opinions (1224, 1703, 2003–2006, etc.)
🔵	Amendment	Amendments, observations, responses (CLMSABEX, DESCABEX, ABEX, etc.)
✅	Grant	Grant decisions, certificates (2006A, 2066, 2047, etc.)
🔍	Search	Search reports (1503, 1503SS, ISR, IPRP, etc.)
💬	Interview	Interview summaries (INTERV, EXIN)
⚪	Other	Receipts, administrative notices, miscellaneous

---

Notes for AI Agents

• Image-only PDFs show 🖼️ — provide the path field directly to vision-capable models
• Run epo-oa ocr + --with-ocr to embed text for language models
• JSON output (--format json) includes full path and text fields for programmatic access
• The prosecution.md is designed to fit within typical LLM context windows for smaller dockets

---

License

MIT