Merge branch 'main' into dev

README.md

@@ -101,6 +101,13 @@ docker run \
 skohub/skohub-vocabs-docker:latest
 ```
 
+## Validation of SKOS files
+
+During the build the provided SKOS files are validated against this [SkoHub SHACL shape](https://github.com/skohub-io/shapes/blob/main/skohub.shacl.ttl).
+The SkoHub SHACL shape is an opinionated and more restrictive version of the generic SKOS shape you can find in the same repo: https://github.com/skohub-io/shapes
+
+If the SKOS file violates the shape, the build will error and stop.
+If you just hit warnings you will see them in the console output.
 
 ## Serve from other location than root (`/`)
 
@@ -117,6 +124,7 @@ To start configuring copy the default file `cp config.default.yaml config.yaml`.
 You can configure the following settings:
 
 - Tokenizer used for searching
+- Custom Domain
 - UI Configurations
     - Title
     - Logo
@@ -128,10 +136,18 @@ The settings are explained in the following sections.
 ### Tokenizer
 
 SkoHub Vocabs uses Flexsearch v0.6.32 for its searching capabilities.
-Flexsearch offers [different tokenizers](https://github.com/nextapps-de/flexsearch/tree/0.6.32#tokenizer) for indexing your vocabularies.
+Flexsearch offers [different tokenizers](https://github.com/nextapps-de/flexsearch#tokenizer-prefix-search) for indexing your vocabularies.
 The chosen tokenizer directly affects the required memory and size of your index file.
 SkoHub Vocabs defaults to `full` tokenizer.
 
+### Custom Domain
+
+If you want to host your vocabularies under a custom domain (so no W3 perma-id or purl.org redirect), you have to provide that domain in the config.
+Example:
+
+The base of your concept scheme is: `http://my-awesome-domain.org/my-vocab`
+Then provide `http://my-awesome-domain.org` as `custom_domain` in your `config.yaml`
+
 ### UI
 
 The following customizations can be made:

config.default.yaml

@@ -1,6 +1,7 @@
 ---
-# see https://github.com/nextapps-de/flexsearch/tree/0.6.32#tokenizer for options
+# see https://github.com/nextapps-de/flexsearch#tokenizer-prefix-search for options
 tokenizer: "full" # strict, forward, reverse, full
+custom_domain: ""
 searchableAttributes:
   - "prefLabel" # you should not delete this one
   - "notation"

gatsby-config.js

@@ -13,6 +13,7 @@ module.exports = {
     logo: config.logo,
     fonts: config.fonts,
     searchableAttributes: config.searchableAttributes,
+    customDomain: config.customDomain,
   },
   pathPrefix: `${process.env.BASEURL || ""}`,
   plugins: [

gatsby-node.js

@@ -12,14 +12,21 @@ const path = require("path")
 const fs = require("fs-extra")
 const { Index, Document } = require("flexsearch")
 const omitEmpty = require("omit-empty")
-const { i18n, getFilePath, parseLanguages } = require("./src/common")
+const {
+  i18n,
+  getFilePath,
+  parseLanguages,
+  loadConfig,
+} = require("./src/common")
 const context = require("./src/context")
 const queries = require("./src/queries")
 const types = require("./src/types")
+const { validate } = require("./src/validate.js")
 
 require("dotenv").config()
 require("graceful-fs").gracefulify(require("fs"))
 
+const config = loadConfig("./config.yaml", "./config.default.yaml")
 const languages = new Set()
 const languagesByCS = {}
 const inverses = {
@@ -84,7 +91,8 @@ const exportIndex = (index, conceptScheme, language) => {
       (conceptScheme.id.endsWith("/")
         ? conceptScheme.id.slice(0, -1)
         : conceptScheme.id) + `-cs/search/${language}/${key}`,
-      `json`
+      `json`,
+      config.customDomain
     )
     createData({
       path,
@@ -104,6 +112,13 @@ exports.onPreBootstrap = async ({ createContentDigest, actions, getNode }) => {
   console.info(`Found these turtle files:`)
   ttlFiles.forEach((e) => console.info(e))
   for (const f of ttlFiles) {
+    try {
+      console.info("Validating: ", f)
+      await validate("shapes/skohub.shacl.ttl", f)
+    } catch (e) {
+      console.error(e)
+      throw e
+    }
     const ttlString = fs.readFileSync(f).toString()
     const doc = await jsonld.fromRDF(ttlString, { format: "text/turtle" })
     const compacted = await jsonld.compact(doc, context.jsonld)
@@ -234,20 +249,25 @@ exports.createPages = async ({ graphql, actions: { createPage } }) => {
       const jsonld = omitEmpty(Object.assign({}, collection, context.jsonld))
       languages.forEach((language) =>
         createPage({
-          path: getFilePath(collection.id, `${language}.html`),
+          path: getFilePath(
+            collection.id,
+            `${language}.html`,
+            config.customDomain
+          ),
           component: path.resolve(`./src/components/Collection.jsx`),
           context: {
             language,
             node: collection,
+            customDomain: config.customDomain,
           },
         })
       )
       createData({
-        path: getFilePath(collection.id, "json"),
+        path: getFilePath(collection.id, "json", config.customDomain),
         data: JSON.stringify(json, null, 2),
       })
       createData({
-        path: getFilePath(collection.id, "jsonld"),
+        path: getFilePath(collection.id, "jsonld", config.customDomain),
         data: JSON.stringify(jsonld, null, 2),
       })
     })
@@ -313,23 +333,28 @@ exports.createPages = async ({ graphql, actions: { createPage } }) => {
             // create pages and data
             languagesOfCS.forEach((language) =>
               createPage({
-                path: getFilePath(concept.id, `${language}.html`),
+                path: getFilePath(
+                  concept.id,
+                  `${language}.html`,
+                  config.customDomain
+                ),
                 component: path.resolve(`./src/components/Concept.jsx`),
                 context: {
                   language,
                   node: concept,
                   collections: memberOf.hasOwnProperty(concept.id)
                     ? memberOf[concept.id]
                     : [],
+                  customDomain: config.customDomain,
                 },
               })
             )
             createData({
-              path: getFilePath(concept.id, "json"),
+              path: getFilePath(concept.id, "json", config.customDomain),
               data: JSON.stringify(json, null, 2),
             })
             createData({
-              path: getFilePath(concept.id, "jsonld"),
+              path: getFilePath(concept.id, "jsonld", config.customDomain),
               data: JSON.stringify(jsonld, null, 2),
             })
           }
@@ -361,23 +386,28 @@ exports.createPages = async ({ graphql, actions: { createPage } }) => {
         })
         languagesOfCS.forEach((language) =>
           createPage({
-            path: getFilePath(conceptScheme.id, `${language}.html`),
+            path: getFilePath(
+              conceptScheme.id,
+              `${language}.html`,
+              config.customDomain
+            ),
             component: path.resolve(`./src/components/ConceptScheme.jsx`),
             context: {
               language,
               node: conceptScheme,
               embed: embeddedConcepts,
+              customDomain: config.customDomain,
             },
           })
         )
         createData({
-          path: getFilePath(conceptScheme.id, "json"),
+          path: getFilePath(conceptScheme.id, "json", config.customDomain),
           data: JSON.stringify(
             omitEmpty(Object.assign({}, conceptScheme, context.jsonld), null, 2)
           ),
         })
         createData({
-          path: getFilePath(conceptScheme.id, "jsonld"),
+          path: getFilePath(conceptScheme.id, "jsonld", config.customDomain),
           data: JSON.stringify(
             omitEmpty(Object.assign({}, conceptScheme, context.jsonld), null, 2)
           ),
@@ -405,6 +435,7 @@ exports.createPages = async ({ graphql, actions: { createPage } }) => {
             return [key, Array.from(value)]
           })
         ),
+        customDomain: config.customDomain,
       },
     })
   )