feat: Docker runtime improvements and srsly dependency removal

[dstengle]

Summary

This PR significantly improves Docker runtime configuration and removes the problematic srsly dependency that was causing compilation issues with cloud native buildpacks.

Problem Solved

The original issue was that srsly (a spaCy dependency) contains C extensions that require compilation, causing failures in cloud native buildpack environments. Additionally, the CLI was hardcoded to specific paths making Docker usage difficult.

Key Changes

🔧 Dependency Management

• Removed spacy dependency - Eliminates srsly compilation issues completely
• Moved SPARQLWrapper from test to main dependencies (required at runtime)
• Updated poetry.lock with clean dependency tree

🐳 Docker Runtime Configuration

• Environment variable support for all paths:
  • KBP_WORK_DIR - Working directory
  • KBP_HOME - Configuration directory (replaces hardcoded ~/.kbp)
  • KBP_CONFIG_PATH - Custom config file location
  • KBP_KNOWLEDGE_BASE_PATH - Documents directory
  • KBP_METADATA_STORE_PATH - Metadata storage location

🛠️ New Docker Tooling

• scripts/docker-run.sh - User-friendly wrapper script with host networking
• docker-compose.app.yml - Production-ready compose configuration
• README-docker.md - Comprehensive Docker usage guide
• Multi-stage Dockerfile - Optimized for both compilation and runtime

📁 Path Flexibility

• Configurable .kbp location - No longer hardcoded to specific paths
• Volume mount support - Works with any directory structure
• Config file detection - Multiple fallback locations with env var priority

Testing

```bash

Build and test Docker image

docker build -t knowledgebase-processor:latest .

Test with environment variables

docker run --rm -v "$(pwd):/workspace" \
-e KBP_WORK_DIR=/workspace \
-e KBP_HOME=/workspace/.kbp \
-w /workspace \
knowledgebase-processor:latest kb --help

Test wrapper script

./scripts/docker-run.sh init
./scripts/docker-run.sh scan
```

Usage Examples

Quick Start

```bash

Initialize and scan documents

./scripts/docker-run.sh init
./scripts/docker-run.sh scan

With custom directory

./scripts/docker-run.sh -w ~/Documents init

Continuous monitoring with host network access

./scripts/docker-run.sh publish --watch
```

Docker Compose

```bash

Interactive mode

docker-compose -f docker-compose.app.yml up kbp

Watch mode with Fuseki

docker-compose -f docker-compose.app.yml up fuseki kbp-watch
```

Benefits

✅ No more compilation issues - Cloud native buildpacks now work
✅ Flexible Docker deployment - Works with any directory structure
✅ Host network connectivity - Can access local SPARQL endpoints
✅ Configurable paths - Adapts to different runtime environments
✅ Production ready - Complete tooling and documentation

Breaking Changes

• spacy removed - EntityRecognizer was disabled by default anyway
• Config resolution order - Environment variables now take priority

Migration Guide

For users currently using spacy/entity recognition:

1. This feature was disabled by default (`analyze_entities: false`)
2. If needed, can be re-enabled by installing spacy separately
3. Most users won't be affected as this was an optional feature

For Docker users:

1. Use the new wrapper script: `./scripts/docker-run.sh`
2. Set environment variables for custom paths if needed
3. See `README-docker.md` for comprehensive examples

🤖 Generated with Claude Code