Make readme generation a little more forgiving and use jsonld cli for…

… jsonld conversion (it works well)
hubmapconsortium · Oct 27, 2023 · bda874c · bda874c
1 parent 7229296
commit bda874c
Show file tree

Hide file tree

Showing 4 changed files with 40 additions and 21 deletions.
diff --git a/scripts/setup-environment.sh b/scripts/setup-environment.sh
@@ -87,6 +87,9 @@ if [ -e "$ENV/bin/activate" ]; then
   # Install the RDF Formatter (for RDFa support)
   npm install -g rdf-formatter
 
+  # Install the JSON-LD CLI (a better JSON-LD converter)
+  npm install -g jsonld-cli
+
   # Download ontologies
   ${ROOT_DIR}/scripts/download-ontologies.sh
 

diff --git a/src/enrichment/enrich-graph.js b/src/enrichment/enrich-graph.js
@@ -2,13 +2,14 @@ import { existsSync, readFileSync } from 'fs';
 import { load } from 'js-yaml';
 import { resolve } from 'path';
 import { error, info } from '../utils/logging.js';
-import { convert, merge } from '../utils/robot.js';
+import { RDF_EXTENSIONS, convert } from '../utils/reify.js';
+import { merge } from '../utils/robot.js';
 import {
   cleanTemporaryFiles,
   convertNormalizedDataToOwl,
   convertNormalizedMetadataToRdf,
+  logOutput,
   runCompleteClosure,
-  logOutput 
 } from './utils.js';
 
 export function enrichGraphMetadata(context) {
@@ -39,7 +40,7 @@ export function enrichGraphData(context) {
       const extension = inputRdfFile.split('.').slice(-1)[0];
       if (extension === 'ttl') {
         toMerge.push(inputRdf);
-      } else if (extension === 'rdf' || extension === 'jsonld' || extension === 'owl'){
+      } else if (RDF_EXTENSIONS.has(extension)) {
         const outputTtl = resolve(obj.path, 'enriched', inputRdfFile + '.ttl');
         convert(inputRdf, outputTtl, 'ttl');
         toMerge.push(outputTtl);
@@ -54,11 +55,10 @@ export function enrichGraphData(context) {
     info(`Creating graph: ${enrichedPath}`);
     convert(enrichedMergePath, enrichedPath, 'ttl');
 
-    info('Optimizing graph...')
+    info('Optimizing graph...');
     const redundantPath = resolve(obj.path, 'enriched/redundant.ttl');
     runCompleteClosure(enrichedPath, redundantPath);
     logOutput(redundantPath);
-
   } catch (e) {
     error(e);
   } finally {

diff --git a/src/finalizing/utils/generate-readme-md.js b/src/finalizing/utils/generate-readme-md.js
@@ -11,17 +11,17 @@ export function renderReadmeMd(templateFile, metadata) {
   const env = new Environment(undefined, { autoescape: false });
 
   env.addFilter('authorList', (list) => {
-    return list.map(a => a.fullName).join('; ');
+    return list?.map(a => a.fullName).join('; ') ?? '';
   });
   env.addFilter('orcidList', (list) => {
-    return list.map(a => `[${a.orcid}](https://orcid.org/${a.orcid})`).join('; ');
+    return list?.map(a => `[${a.orcid}](https://orcid.org/${a.orcid})`).join('; ') ?? '';
   });
   env.addFilter('downloadLinks', (datatable) => {
-    return datatable.map((str) => {
+    return datatable?.map((str) => {
       const ext = str !== undefined ? str.slice(str.replace('.zip', '').lastIndexOf('.') + 1).replace(')', '') : '';
       return `[${ext.toUpperCase()}](assets/${str})`
     }).join(' ');
-  });
+  }) ?? '';
   const template = readFileSync(templateFile).toString();
   return env.renderString(template, { ...TYPE_MAPPINGS, ...metadata });
 }

diff --git a/src/utils/reify.js b/src/utils/reify.js
@@ -9,6 +9,27 @@ const FORMATS = {
   nq: 'application/n-quads',
 };
 
+export const RDF_EXTENSIONS = new Set([
+  'json',
+  'jsonld',
+  'json-ld',
+  'nt',
+  'xml',
+  'rdf',
+  'nq',
+  'n3',
+  '.owl',
+  '.trig',
+  'turtle',
+]);
+
+export const JSONLD_EXTENSIONS = new Set(['json', 'jsonld', 'json-ld']);
+
+export function isJsonLd(fileOrUrl) {
+  const extension = fileOrUrl.split('.').slice(-1)[0];
+  return JSONLD_EXTENSIONS.has(extension);
+}
+
 export function reifyDoTurtle(context, inputPath) {
   const graphName = context.selectedDigitalObject.iri;
   reifyTurtle(inputPath, graphName);
@@ -44,18 +65,13 @@ function reifyTurtle(inputPath, graphName) {
   }
 }
 
-function convert(inputPath, outputPath, outputFormat, graphName) {
+export function convert(inputPath, outputPath, outputFormat, graphName) {
+  let command = `rdfpipe --output-format ${outputFormat} ${inputPath}`;
+  if (isJsonLd(inputPath)) {
+    command = `cat ${inputPath} | jsonld expand | rdfpipe --input-format json-ld --output-format ${outputFormat}`;
+  }
   if (graphName && outputFormat === 'application/n-quads') {
-    throwOnError(
-      `rdfpipe --output-format ${outputFormat} ${inputPath} | \\
-          perl -pe 's|\\Qfile://${inputPath}\\E|${graphName}|g' \\
-          > ${outputPath}`,
-      `Failed to fix n-quads file: ${outputPath}`
-    );
-  } else {
-    throwOnError(
-      `rdfpipe --output-format ${outputFormat} ${inputPath} > ${outputPath}`,
-      `Failed to convert to '${outputFormat}' format.`
-    );
+    command += ` | perl -pe 's|\\Qfile://${inputPath}\\E|${graphName}|g'`;
   }
+  throwOnError(`${command} > ${outputPath}`, `Failed to convert to '${outputFormat}' format.`);
 }