an initial implementation of wrb codelists import from csv

code-lists/README.md

@@ -0,0 +1,11 @@
+# WRB codelists
+
+[World Reference Base](https://wrb.isric.org/) working group of IUSS maintains a set of code lists to describe soils, as part of the World Reference Base for Soil Resources (current: 4th edition 2022). This repository presents a home for identification of the concepts in those lists, as well as facilites maintenance of these lists in preparation of upcoming releases.
+
+## csv2skos
+
+The [code lists](./wrb-codelists.ttl) are described in RDF using the SKOS ontology. The RDF is generated from a [CSV](./wrb-codelists.csv) using a [conversion script](./csv2skos/csv2skos.py). MS Excel has been used to prepare the initial version of the CSV (Note that Excel uses ';' as a separator, where this initiative uses ',' as a separator).
+
+## skos2html
+
+[skos2html](./csv2skos/skos2html.py) is a small utility to generate html of the skos file for human readability.

code-lists/csv2skos/README.md

@@ -0,0 +1,11 @@
+# csv2skos
+
+A tool to load csv data into skos
+
+```
+virtualenv .
+. bin/activate
+pip install -r requirements.txt
+python csv2skos.py
+```
+

code-lists/csv2skos/csv2skos.py

@@ -0,0 +1,42 @@
+import csv
+from rdflib import Graph, URIRef, Literal, Namespace
+from rdflib.namespace import DC, DCTERMS, FOAF, OWL, RDF, RDFS, SKOS, XMLNS, XSD
+
+def valuri(data):
+    return "".join(x for x in data.replace(' ','-') if (x.isalnum() or x =='-'))
+
+g = Graph()
+WRB = Namespace("http://iuss-wrb.github.io/wrb#")
+
+ont = URIRef("http://iuss-wrb.github.io/wrb#")
+g.add((ont, RDF.type, OWL.Ontology))
+g.add((ont, DCTERMS.description, Literal("Code lists to describe soil properties as defined by the IUSS WRB working group")))
+g.add((ont, DCTERMS.creator, URIRef("https://orcid.org/0000-0003-1499-618X")))
+g.add((ont, DCTERMS.rights, Literal("This ontology is distributed under Creative Commons Attribution 4.0 License - https://creativecommons.org/licenses/by/4.0")))
+g.add((ont, DCTERMS.source, URIRef("https://wrb.isric.org/files/WRB_fourth_edition_2022-12-18.pdf")))
+g.add((ont, DCTERMS.title, Literal("WRB code lists")))
+g.add((ont, FOAF.logo, URIRef("https://wrb.isric.org/images/logo.png")))
+
+with open('../wrb-codelists.csv', 'r') as csvfile:
+    reader = csv.DictReader(csvfile)
+    cs = ""
+    for row in reader:
+        if cs != valuri(row['attribute']):
+            cs = valuri(row['attribute'])
+            cs2 = URIRef(WRB[valuri(row['attribute'])])
+            g.add((cs2, RDF.type, SKOS.ConceptScheme))
+            g.add((cs2, SKOS.prefLabel, Literal(row['attribute'])))
+
+        concept = URIRef(WRB[valuri(row['attribute'])+'/'+valuri(row['id'])])
+        g.add((concept, RDF.type, SKOS.Concept))
+        g.add((concept, SKOS.prefLabel, Literal(row['label'])))
+        g.add((concept, SKOS.inScheme, cs2))
+        g.add((concept, SKOS.notation, Literal(row['notation'])))
+        g.add((concept, SKOS.definition, Literal(row['definition'])))
+
+g.bind("skos", SKOS) 
+g.bind("dcterms", DCTERMS) 
+g.bind("owl", OWL) 
+g.bind("wrb", WRB)
+g.serialize(destination="../wrb-codelists.ttl")
+

code-lists/csv2skos/requirements.txt

@@ -0,0 +1 @@
+rdflib

code-lists/csv2skos/skos2html.py

@@ -0,0 +1,33 @@
+from rdflib import Graph
+from rdflib.namespace import RDF, SKOS, DCTERMS, OWL
+
+g = Graph()
+g.parse("../wrb-codelists.ttl")
+
+ttl = "Codelists" # default title
+desc = "A series of codelists" # default description
+
+for s, p, o in g.triples((None, RDF.type, OWL.Ontology)):
+    ttl = g.value(s,DCTERMS.title)
+    desc = g.value(s,DCTERMS.description) + "<br/>"
+    for t in ["creator","rights","source"]:
+        if g.value(s,DCTERMS[t]):
+            desc += "<br/><b>"+t+"</b>: " + str(g.value(s,DCTERMS[t]))
+
+html = "<h1>"+ttl+"</h1><p>"+desc+"</p>"
+
+html += "<h2 id='#top'>Contents</h2><p><ul>"
+for s, p, o in g.triples((None, RDF.type, SKOS.ConceptScheme)):    
+    html += "<li><a href='#"+str(s)+"'>" + g.value(s,SKOS.prefLabel) + "</a></li>"
+html += "</ul></p><hr/>\n"
+
+for s, p, o in g.triples((None, RDF.type, SKOS.ConceptScheme)):    
+    html += "<h2 id='"+str(s)+"'>" + g.value(s,SKOS.prefLabel) + "</h2>\n"
+    html += "<table><tr><th>Code</th><th>Label</th><th>Definition</th></tr>\n"
+    for s2, p2, o2 in g.triples((None, SKOS.inScheme, s)): 
+        html += "<tr><td>"+g.value(s2,SKOS.notation)+"</td><td>"+ g.value(s2,SKOS.prefLabel) + "</td><td>" + g.value(s2,SKOS.definition) +"</td></tr>\n"
+    html += "<br/><a href='#top'>&#8657; Index</a><hr/>"
+
+f = open("../index.html", "w")
+f.write("<html>\n<head><title>"+ttl+"</title></head>\n<body>"+html+"</body>\n<style> body {font-family: arial} td, th {border: 1px solid gray;padding:5px} th { background-color: #efefef } table{border-collapse: collapse}</style></html>")
+f.close()