Merge pull request #5 from b-cubed-eu/add-manual-resource

Add manual resource

Author: wlangera

.gitignore

@@ -17,6 +17,7 @@
 *.xls*
 *_files
 *.gz
+*.geojson
 docs
 libs
 output

data/b3data_package/datapackage.json

@@ -11,8 +11,15 @@
       "encoding": "utf-8",
       "title": "Occurrence cube for birds in Belgium (MGRS 10 km)",
       "description": "Occurrence cube for birds in Belgium between 2000 en 2024. The taxonomical resolution is 'species' and the temporal resolution is 'year' Spatial aggregation is done using the MGRS grid at 10 km scale. Only grid cells that fall within the 10 km MGRS reference grid for mainland Belgium (see b3data: `mgrs10_refgrid_belgium.geojson`) are included.",
-      "sources": ["GBIF Occurrence Download", "https://doi.org/10.15468/dl.y3wpwk"],
-      "licenses": [" CC BY-NC 4.0", "https://creativecommons.org/licenses/by-nc/4.0/", "Creative Commons Attribution-NonCommercial 4.0 International"],
+      "sources": {
+        "title": "GBIF Occurrence Download",
+        "path": "https://doi.org/10.15468/dl.y3wpwk"
+      },
+      "licenses": {
+        "name": "CC BY-NC 4.0",
+        "path": "https://creativecommons.org/licenses/by-nc/4.0/",
+        "title": "Creative Commons Attribution-NonCommercial 4.0 International"
+      },
       "schema": {
         "fields": [
           {
@@ -49,27 +56,47 @@
           }
         ]
       }
+    },
+    {
+      "name": "mgrs10_refgrid_belgium",
+      "path": "mgrs10_refgrid_belgium.geojson",
+      "profile": "spatial-data-resource",
+      "format": "geojson",
+      "title": "MGRS 10 Km reference grid Belgium",
+      "description": "MGRS 10 Km reference grid for the mainland of Belgium.",
+      "licenses": {
+        "name": "CC0 1.0",
+        "path": "https://creativecommons.org/publicdomain/zero/1.0/",
+        "title": "Creative Commons Zero v1.0 Universal"
+      }
     }
   ],
   "title": "b3data: Data resources for the b3verse",
   "description": "This data package contains data resources to be used across the b3verse (https://docs.b-cubed.eu/guides/b3verse/). This includes example datasets (occurrence cubes) as well as spatial resources like reference grids or raster data.",
-  "keywords1": "data cubes",
-  "licenses.name": " CC BY-NC 4.0",
+  "keywords": ["data cubes", "b3verse", "frictionless", "biodiversity"],
+  "licenses": {
+    "name": "CC BY-NC 4.0",
+    "path": "https://creativecommons.org/licenses/by-nc/4.0/",
+    "title": "Creative Commons Attribution-NonCommercial 4.0 International"
+  },
   "version": "0.1.0",
-  "sources.title": "b3data-scripts",
-  "contributors.title": "Ward Langeraert",
-  "contributors.path": "https://orcid.org/0000-0002-5900-8109",
-  "contributors.email": "[email protected]",
-  "contributors.role": "author",
-  "contributors.organization": "Research Institute for Nature and Forest (INBO)",
-  "contributors.title.1": "Toon Van Daele",
-  "contributors.path.1": "https://orcid.org/0000-0002-1362-853X",
-  "contributors.role.1": "contributor",
-  "contributors.organization.1": "Research Institute for Nature and Forest (INBO)",
-  "sources.path": "https://github.com/b-cubed-eu/b3data-scripts",
-  "licenses.path": "https://creativecommons.org/licenses/by-nc/4.0/",
-  "licenses.title": "Creative Commons Attribution-NonCommercial 4.0 International",
-  "keywords2": "b3verse",
-  "keywords3": "frictionless",
-  "keywords4": "biodiversity"
+  "sources": {
+    "title": "b3data-scripts",
+    "path": "https://github.com/b-cubed-eu/b3data-scripts"
+  },
+  "contributors": [
+    {
+      "title": "Ward Langeraert",
+      "path": "https://orcid.org/0000-0002-5900-8109",
+      "email": "[email protected]",
+      "role": "author",
+      "organization": "Research Institute for Nature and Forest (INBO)"
+    },
+    {
+      "title": "Toon Van Daele",
+      "path": "https://orcid.org/0000-0002-1362-853X",
+      "role": "contributor",
+      "organization": "Research Institute for Nature and Forest (INBO)"
+    }
+  ]
 }

source/R/add_manual_resource.R

@@ -0,0 +1,23 @@
+add_manual_resource <- function(package, new_resource, replace = TRUE) {
+  if (replace) {
+    # Filter out any resource with same name or path
+    package$resources <- Filter(function(res) {
+      !(res$name == new_resource$name && res$path == new_resource$path)
+    }, package$resources)
+  } else {
+    # If not overwriting, check for duplicates and warn
+    conflict <- any(vapply(package$resources, function(res) {
+      res$name == new_resource$name && res$path == new_resource$path
+    }, logical(1)))
+
+    if (conflict) {
+      warning(paste("Resource with same name and path already exists and",
+                    "`replace = FALSE`. Skipping."))
+      return(package)
+    }
+  }
+
+  # Append new resource
+  package$resources <- append(package$resources, list(new_resource))
+  return(package)
+}

source/add_spatial_resources.Rmd

@@ -22,12 +22,15 @@ library(tidyverse)     # Data wrangling and visualisation
 library(frictionless)  # Create frictionless data package
 library(sf)            # Spatial objects
 
+# Source functions
+source(here::here("source", "R", "add_manual_resource.R"))
+
 # Data path and create directory if necessary
 data_path <- here::here("data", "raw")
 dir.create(data_path, showWarnings = FALSE, recursive = TRUE)
 
-out_path <- here::here("data", "processed")
-dir.create(out_path, showWarnings = FALSE, recursive = TRUE)
+package_path <- here::here("data", "b3data_package")
+dir.create(package_path, showWarnings = FALSE, recursive = TRUE)
 ```
 
 # Goal
@@ -64,6 +67,8 @@ Each dataset includes the following metadata (see: [Frictionless resource spec](
 # Datasets
 ## MGRS 10 Km reference grid Belgium
 
+Load the data.
+
 ```{r}
 # Read reference grid
 utm10_bel <- st_read(file.path(data_path, "utm10_bel.shp"))
@@ -72,9 +77,53 @@ utm10_bel <- st_read(file.path(data_path, "utm10_bel.shp"))
 ggplot() + geom_sf(data = utm10_bel)
 ```
 
+Clean data and transform cell codes to MGRS.
+
+```{r}
+mgrs10_refgrid_belgium <- utm10_bel %>%
+  select(utm_tag = TAG, geometry) %>%
+  mutate(mgrscode = ifelse(grepl("^[A-G]", utm_tag), paste0("31U", utm_tag),
+                           paste0("32U", utm_tag))) %>%
+  select(mgrscode, geometry)
+```
+
+Write to data package and add metadata.
+
+```{r}
+st_write(mgrs10_refgrid_belgium,
+         file.path(package_path, "mgrs10_refgrid_belgium.geojson"),
+         delete_dsn = TRUE)
+```
+
+```{r}
+# Read package
+b3data_package <- read_package(file.path(package_path, "datapackage.json"))
+
+# Add resource to data package
+mgrs10_resource <- list(
+  name = "mgrs10_refgrid_belgium",
+  path = "mgrs10_refgrid_belgium.geojson",
+  profile = "spatial-data-resource",
+  format = "geojson",
+  title = "MGRS 10 Km reference grid Belgium",
+  description = "MGRS 10 Km reference grid for the mainland of Belgium.",
+  licenses = list(
+    name = "CC0 1.0",
+    path = "https://creativecommons.org/publicdomain/zero/1.0/",
+    title = "Creative Commons Zero v1.0 Universal"
+    )
+  )
+
+b3data_package <- add_manual_resource(
+  b3data_package, mgrs10_resource, replace = TRUE)
+
+# Write package to directory
+write_package(
+  package = b3data_package,
+  directory = package_path,
+  compress = TRUE)
+```
 
 ## EEA 100 km reference grid Europe
 
 > Coming soon
-
-# Write package

source/create_b3data_package.Rmd

@@ -152,7 +152,7 @@ ggplot() + geom_sf(data = utm10_bel)
 ```{r}
 # Only select grid cells from reference grid
 bird_cube_belgium_mgrs10 <- bird_cube_belgium_mgrs10_full %>%
-  filter(substr(mgrscode, 4, nchar(mgrscode)) %in% utm10_bel$TAG)
+  filter(substring(mgrscode, 4) %in% utm10_bel$TAG)
 ```
 
 We create the package and add the dataset.
@@ -171,12 +171,12 @@ b3data_package <- create_package() %>%
       "Only grid cells that fall within the 10 km MGRS reference grid for",
       "mainland Belgium (see b3data: `mgrs10_refgrid_belgium.geojson`) are",
       "included."),
-    sources = c(
+    sources = list(
       title = "GBIF Occurrence Download",
       path = "https://doi.org/10.15468/dl.y3wpwk"
     ),
-    licenses = c(
-      name = " CC BY-NC 4.0",
+    licenses = list(
+      name = "CC BY-NC 4.0",
       path = "https://creativecommons.org/licenses/by-nc/4.0/",
       title = "Creative Commons Attribution-NonCommercial 4.0 International"
     )
@@ -210,45 +210,47 @@ b3data_package <- append(
     "resources like reference grids or raster data."
   )),
   after = 4)
-b3data_package <- append(b3data_package,
-                         c(keywords = c("data cubes", "b3verse", "frictionless",
-                                        "biodiversity")),
-                         after = 5)
 b3data_package <- append(
   b3data_package,
-  c(licenses = c(
-    name = " CC BY-NC 4.0",
+  c(keywords = list(list(
+    "data cubes", "b3verse", "frictionless", "biodiversity"
+    ))),
+  after = 5)
+b3data_package <- append(
+  b3data_package,
+  c(licenses = list(list(
+    name = "CC BY-NC 4.0",
     path = "https://creativecommons.org/licenses/by-nc/4.0/",
     title = "Creative Commons Attribution-NonCommercial 4.0 International"
-    )),
+    ))),
   after = 6)
 b3data_package <- append(b3data_package,
                          c(version = "0.1.0"),
                          after = 7)
 b3data_package <- append(
   b3data_package,
-  c(sources = c(
+  c(sources = list(list(
     title = "b3data-scripts",
     path = "https://github.com/b-cubed-eu/b3data-scripts"
-    )),
+    ))),
   after = 8)
 b3data_package <- append(
   b3data_package,
-  c(contributors = c(
-    c(
+  c(contributors = list(list(
+    list(
       title = "Ward Langeraert",
       path = "https://orcid.org/0000-0002-5900-8109",
       email = "[email protected]",
       role = "author",
       organization = "Research Institute for Nature and Forest (INBO)"
       ),
-    c(
+    list(
       title = "Toon Van Daele",
       path = "https://orcid.org/0000-0002-1362-853X",
       role = "contributor",
       organization = "Research Institute for Nature and Forest (INBO)"
       )
-    )),
+    ))),
   after = 9)
 
 # Warning: append() drops the custom datapackage class.