Caching¶

Three-level cache system for fast repeated queries.

Cache Levels¶

Level	Location	Content
1	`osm_data.csv`	Final processed output
2	`processed_units.json`	Processed Unit objects
3	`*_dc/` directories	Raw OSM elements

Cache Location¶

OS	Path
Linux	`~/.cache/osm-powerplants/`
macOS	`~/Library/Caches/osm-powerplants/`
Windows	`%LOCALAPPDATA%\osm-powerplants\Cache\`

Custom location:

cache_dir: /path/to/cache

Force Refresh¶

Re-download all data from Overpass API:

osm-powerplants process Germany --force-refresh

Note: For large countries (Germany, France), this can take several minutes due to API queries.

Update from API Cache¶

When analyzing rejections or debugging, you may run processing with --force-refresh which updates the API cache. To then update the CSV output without re-downloading:

osm-powerplants process Germany --update -o germany.csv

This skips the CSV cache but reuses the API cache, saving time for large countries.

Typical workflow:

# 1. Force refresh to get latest OSM data (slow for large countries)
osm-powerplants process Germany --force-refresh -o germany.csv

# 2. Analyze rejections, tune config...

# 3. Reprocess with updated config (fast - uses API cache)
osm-powerplants process Germany --update -o germany.csv

Clear Cache¶

rm -rf ~/.cache/osm-powerplants/*

Config Hash¶

Cache is automatically invalidated when processing parameters change. The config hash is computed from:

source_mapping, technology_mapping
plants_only, missing_*_allowed
capacity_extraction, units_clustering, units_reconstruction

Performance¶

Cache Hit	Speedup
CSV	50-100x
Units	10-30x
API	2-5x