Release v0.5.0: network_cost (connectivity to ports), osm/dem sources…

CHANGELOG.md

@@ -8,6 +8,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+---
+
+## [0.5.0] - 2026-10-05
+
 ### Added
 - **`sum` over vector sources**, with `params = {area = true}` for areal weighting: each polygon's
   attribute (e.g. census population) is shared among the cells in proportion to the intersected
@@ -16,6 +20,20 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   touches (a point to its own cell). `params = {column = …}` picks the column (default: the target
   name). Vector sources are no longer clipped to the grid for `sum`, so a polygon crossing the border
   keeps its full denominator.
+- **`network_cost` operator** (alias `gpm_network`): least transport cost over a road network, from
+  each cell to the nearest destination, by an exact multi-source Dijkstra over the line graph
+  (`scipy.sparse.csgraph`). Destinations come from `params = {targets = …}` (a vector file, found
+  in the download cache or as given, or a list of `[x, y]`; polygons count by their centroid).
+  The cost of a segment is its length times a factor: `cost_column` (a per-line multiplier),
+  or `status_column` with `inside_paved` / `inside_unpaved`; the legs from a cell to the network
+  cost `outside` times the distance. `entrance = "vertex"` joins a cell and a destination at the
+  nearest vertex, as TerraME's GPM does; `"segment"` (default) at the nearest point of the nearest
+  line. `unit_scale` rescales the result (`1e-3` for km) and `crs` measures in metres. Cells that
+  reach no destination are NaN, with a warning. Reproduces the regional pattern of TerraME's GPM
+  `Network` (generalized transport cost to ports, r = 0.996 on the 25 km Brazil grid); the two
+  methods do not agree cell by cell (`LambdaGeo/disscube-benchmark`, `connectivity`). Example:
+  `examples/04_network_connectivity.py`. First use: `e_connport`, the connectivity to ports of
+  LuccME-BR (`LambdaGeo/disscube-recipes`, `cases/luccme_br`).
 - **`median` operator**: per-cell median of the valid pixels, from the fine-aligned source.
 - **`std` `ddof` param** (`0` default, `1` for the sample estimate). The default is unchanged, so
   existing `spec_hash`es hold.
@@ -41,6 +59,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   The notes saying TerraME measures from the cell polygon and divides by the whole cell were removed,
   and so was the "half a cell diagonal" remark in the `DistanceOperator` docstring.
 
+### Fixed
+- The CLI reports a missing optional extra (e.g. `disscube[dissmodel]`) as an error message, not a
+  traceback.
+
 ## [0.4.0] - 2026-09-30
 
 ### Changed

CITATION.cff

@@ -6,8 +6,8 @@ authors:
     orcid: "https://orcid.org/0000-0002-0232-4549"
 title: "DisSCube: Declarative spatial data cubes"
 abstract: "An open-source, declarative engine for constructing cellular spatial data cubes for dynamic modeling, spatial simulation, and environmental analysis."
-version: 0.4.0
-date-released: "2026-09-29"
+version: 0.5.0
+date-released: "2026-10-05"
 url: "https://github.com/DisSModel/disscube"
 repository-code: "https://github.com/DisSModel/disscube"
 license: MIT

README.md

@@ -303,7 +303,7 @@ If you use DisSCube in your research, dynamic modeling, or spatial data pipeline
   author       = {Costa, S{\'e}rgio Souza},
   title        = {{DisSCube: Declarative spatial data cubes}},
   year         = {2026},
-  version      = {0.4.0},
+  version      = {0.5.0},
   url          = {https://github.com/DisSModel/disscube}
 }
 ```

docs/architecture/operators.md

@@ -69,6 +69,7 @@ The last three also produce `coverage_purity` and `dominance_purity` as coordina
 | `distance` | Exact Euclidean distance (CRS units) from each cell centre to the nearest feature; the source is not clipped to the grid. `params = {"crs": …}` measures in another CRS (metres in a projected one) | no |
 | `min_distance` | Raster approximation of the distance to the nearest feature inside the grid (NaN, with a warning, if none is inside) | no |
 | `count` | Number of features whose centroid falls in each cell | no |
+| `network_cost` (alias `gpm_network`) | Least transport cost over a road network (exact multi-source Dijkstra) from each cell to the nearest destination: `targets` (file or `[x, y]` list), `cost_column` or `status_column` + `inside_paved` / `inside_unpaved`, `outside` (off-road factor), `entrance` (`segment` or `vertex`), `unit_scale`, `crs`. NaN where no destination is reachable | no |
 | `area` | Share (0..1) of each cell covered by polygons (intersection area / cell area; overlaps count once) — TerraME's `area` | no |
 
 ## The `compute()` contract

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "disscube"
-version = "0.4.0"
+version = "0.5.0"
 description = "Declarative spatial data cubes: describe sources, grid and derived variables in TOML; get a cataloged, reproducible cube."
 readme = "README.md"
 requires-python = ">=3.11"