Complete installation instructions for the DocumentToMarkdown pipelines (poster, patent, talk, presentation, paper, CI, outlook).
- Go to https://docs.conda.io/en/latest/miniconda.html
- Download Miniconda3 Windows 64-bit
- Run installer
- Install for: Just Me (recommended)
- Location: Default (
C:\Users\YourName\miniconda3) - Add to PATH: ☑ (optional but recommended)
conda --version
# Should output: conda 23.x.x or newer# Create environment with Python 3.11
conda create -n ds_env python=3.11 -y
# Activate
conda activate ds_env
# Verify
python --version
# Should output: Python 3.11.xconda activate ds_env
# PDF processing
conda install -c conda-forge pdfplumber pymupdf pillow -y
# Data handling
conda install pandas openpyxl -y
# OCR support (used by poster and talk pipelines)
conda install -c conda-forge tesseract pytesseract -y
# Presentation pipeline (PPTX support)
pip install python-pptx
# Outlook pipeline (DOCX attachment extraction + contact YAML files)
pip install python-docx
pip install pyyaml
# Anthropic API (Claude Vision)
pip install anthropicpython -c "import pdfplumber, fitz, pandas, openpyxl, anthropic, pptx, docx, yaml; print('All packages OK')"
tesseract --version
# Should output: tesseract 5.x.xOr use the provided environment check script:
python scripts/check_environment.pyAll three pipelines use the Anthropic API (Claude) for vision analysis. Set credentials as Windows User environment variables:
# Set authentication token (PowerShell)
[Environment]::SetEnvironmentVariable("ANTHROPIC_AUTH_TOKEN", "your-token", "User")
# Set base URL (e.g. corporate proxy)
[Environment]::SetEnvironmentVariable("ANTHROPIC_BASE_URL", "https://your-proxy-url/api/proxy/anthropic", "User")
# Verify
[Environment]::GetEnvironmentVariable('ANTHROPIC_AUTH_TOKEN', 'User')
[Environment]::GetEnvironmentVariable('ANTHROPIC_BASE_URL', 'User')IMPORTANT: Restart your terminal after setting variables.
Alternatively, if using the Anthropic API directly (no proxy), set ANTHROPIC_API_KEY instead:
[Environment]::SetEnvironmentVariable("ANTHROPIC_API_KEY", "sk-ant-...", "User")# Navigate to project
cd DocumentToMarkdown
# Activate environment
conda activate ds_env
# Process posters
python poster_pipeline.py --sharepoint "path/to/poster_pdfs" --metadata "abstracts.xlsx"
# Process patents
python patent_pipeline.py --input "path/to/patent_pdfs"
# Process talks
python talk_pipeline.py --talks "path/to/talk_pdfs" --metadata "abstracts.xlsx"
# Process presentations
python presentation_pipeline.py --input "path/to/pptx_or_pdf_files"
# Process scientific papers
python paper_pipeline.py --input "path/to/papers"
# Process competitive intelligence documents
python ci_pipeline.py --input "path/to/ci_docs"
# Process Outlook emails (Outlook must be running)
python outlook_pipeline.py --folder "Inbox/CI Reports"See the README for full CLI options, or the pipeline-specific guides for architecture details:
Restart terminal or add to PATH manually:
set PATH=%PATH%;C:\Users\YourName\miniconda3\Scripts# Verify environment is active
conda activate ds_env
where python
# Should show: C:\Users\YourName\miniconda3\envs\ds_env\python.exe
# Reinstall a package
conda install -c conda-forge <package-name> -y# Add to PATH
[Environment]::SetEnvironmentVariable("Path", $env:Path + ";C:\Program Files\Tesseract-OCR", "User")
# Restart terminalIf Tesseract language data is missing, run the automated setup:
python scripts/setup_tessdata.py# Check current values
[Environment]::GetEnvironmentVariable('ANTHROPIC_AUTH_TOKEN', 'User')
# Re-set if empty
[Environment]::SetEnvironmentVariable("ANTHROPIC_AUTH_TOKEN", "your-token", "User")
[Environment]::SetEnvironmentVariable("ANTHROPIC_BASE_URL", "https://your-proxy-url/api/proxy/anthropic", "User")
# MUST restart terminal after changes- Install Miniconda
- Create
ds_envenvironment (Python 3.11) - Install pdfplumber, pymupdf, pandas, openpyxl, anthropic, pytesseract, python-pptx, python-docx, pyyaml
- Set
ANTHROPIC_AUTH_TOKENandANTHROPIC_BASE_URL(orANTHROPIC_API_KEY) - Restart terminal
- Run a pipeline to verify everything works
| Package | Purpose | Install Command |
|---|---|---|
| pdfplumber | Native PDF text extraction | conda install -c conda-forge pdfplumber |
| pymupdf (fitz) | PDF rendering to images | conda install -c conda-forge pymupdf |
| pillow | Image processing | conda install pillow |
| pandas | Metadata handling | conda install pandas |
| openpyxl | Excel file support | conda install openpyxl |
| python-pptx | PPTX text/table extraction | pip install python-pptx |
| python-docx | DOCX text extraction (Outlook attachments) | pip install python-docx |
| pyyaml | YAML read/write for Outlook contact files | pip install pyyaml |
| pytesseract | OCR interface (Tesseract) | conda install -c conda-forge pytesseract |
| anthropic | Claude Vision API client | pip install anthropic |