Historical Documentation Analyzer

Phase I ESA tool that analyzes ERIS historical documentation packages (aerial photographs, topographic maps, city directories) and produces formatted tables with cross-referencing summary.

Features

• Aerial Photograph Analysis — AI vision identifies land use, structures, and environmental concerns within the property boundary
• Topographic Map Analysis — Extrapolates property location, describes terrain and features
• City Directory Parsing — Extracts occupant listings, flags environmentally concerning uses
• Cross-Reference Summary — Synthesizes findings across all document types, identifies RECs
• Three-zone analysis — Subject Property, Adjoining Properties, Surrounding Properties
• Year range grouping — Consecutive similar years automatically grouped
• Model agnostic — Works with OpenAI (GPT-4o) or Anthropic (Claude Sonnet)

Prerequisites

• Python 3.10+
• poppler (required for PDF processing):
  • macOS: brew install poppler
  • Ubuntu/Debian: apt-get install poppler-utils
  • Windows: Download poppler
  • Or just use Docker (see below) — poppler is included.

Quick Start (Local)

# Clone the repo
git clone https://github.com/kstring99/historical-analyzer.git
cd historical-analyzer

# Install dependencies
pip install -r requirements.txt

# Set your API key (OpenAI or Anthropic — pick one)
export OPENAI_API_KEY=sk-...
# or: export ANTHROPIC_API_KEY=sk-ant-...

# Run
uvicorn app.main:app --port 8000

Open http://localhost:8000

Docker Deployment

# Build
docker build -t esa-historical-analyzer .

# Run with OpenAI enterprise key
docker run -p 8000:8000 -e OPENAI_API_KEY=sk-... esa-historical-analyzer

Environment Variables

Variable	Description	Required
OPENAI_API_KEY	OpenAI API key (enterprise)	Yes (for OpenAI)
ANTHROPIC_API_KEY	Anthropic API key	Yes (for Anthropic)

Output Format

Produces three tables per document type matching the Quire ESA template:

• Year — Individual year or grouped range (e.g., "1946 - 1978")
• Issues Noted — Yes/No with specific evidence cited
• Observations/Occupants — Professional ESA-style descriptions

Plus a cross-referencing summary paragraph that synthesizes findings across all sources.