Supported I/O Formats

GISPulse uses PyOGRIO for reading and writing. Format detection is automatic based on the file extension.

# List available formats on your installation
gispulse formats

Vector formats

Recommended formats

Extension	Format	Read	Write	Notes
.gpkg	GeoPackage	yes	yes	Recommended — multi-layer, styles, performant
.fgb	FlatGeobuf	yes	yes	Ultra-fast for large volumes
.parquet	GeoParquet	yes	yes	Optimal for wide tabular data
.geojson	GeoJSON	yes	yes	Web standard, interoperable
.geojsonl	GeoJSON Lines	yes	yes	Streaming, large volumes

Common formats

Extension	Format	Read	Write	Notes
.shp	ESRI Shapefile	yes	yes	Legacy — prefer GPKG
.csv	CSV (with lat/lon columns)	yes	yes	No native geometry
.dxf	AutoCAD DXF	yes	yes	CAD
.kml	KML / KMZ	yes	no	Google Earth
.gml	GML	yes	yes	OGC standard
.gpx	GPX	yes	no	GPS tracks

Database formats

Format	Read	Write	Notes
PostGIS	yes	yes	Via GISPULSE_DSN (Pro)
SpatiaLite	yes	yes	Portable mode
ESRI GeoDatabase (.gdb)	yes	no	Read-only
OGC WFS	yes	no	Via WFS URL

Raster formats (with gispulse[raster])

Extension	Format	Read	Write
.tif, .tiff	GeoTIFF	yes	yes
.vrt	GDAL VRT	yes	no
.img	ERDAS Imagine	yes	no
.nc	NetCDF	yes	no

Automatic detection

GISPulse detects the format from the extension:

# Format detected automatically
gispulse run input.fgb --rules rules.json -o output.gpkg
gispulse run input.geojson --rules rules.json -o output.fgb
gispulse run input.shp --rules rules.json -o output.parquet

If the file has no recognized extension, force it with --layer and --crs.

Format recommendations

For large volumes (> 100,000 features)

1. FlatGeobuf (.fgb) — fastest read/write, spatially indexed
2. GeoParquet (.parquet) — excellent when you have dozens of attribute columns
3. GPKG — versatile, supports styles

For GIS desktop interoperability

• GeoPackage (.gpkg) — supports QGIS styles (QML) and SLD, multi-layer
• GISPulse automatically copies styles during an --all-layers pipeline

For web / API

• GeoJSON — universal standard, human-readable
• FlatGeobuf — performant streaming for large client-side datasets

For "modern" spatial data

• GeoParquet — compatible with DuckDB, Pandas, Arrow, cloud-native

Multi-layer (GPKG)

GeoPackage supports multiple layers in a single file:

# Process a specific layer
gispulse run project.gpkg --rules rules.json -o output.gpkg --layer buildings

# Process all layers (styles copied)
gispulse run project.gpkg --rules rules.json -o output.gpkg --all-layers

# Inspect layers in a GPKG
gispulse layers project.gpkg

3 layer(s):
  - parcels
  - buildings
  - roads