unpack-whenwe-json.js -> bin/unpack-whenwe-json.js

Just to be tidy.
2023-05-09 21:23:58 +01:00
parent 5e8be866b4
commit 72a49ac52e
1 changed files with 0 additions and 0 deletions
--- a/bin/unpack-whenwe-json.js
+++ b/bin/unpack-whenwe-json.js
@@ -0,0 +1,76 @@
+/*
+  Reads in the Drupal node dump of WhenWe data from ./whenwe.json
+  Writes it converted to markdown with frontmatter, organised in subdirectories, in out/
+*/
+const data = require('./whenwe.json');
+const fs = require('fs');
+const path = require('path');
+const { NodeHtmlMarkdown, NodeHtmlMarkdownOptions } = require('node-html-markdown');
+const nhm = new NodeHtmlMarkdown({
+  useLinkReferenceDefinitions: true,
+  useInlineLinks: true,
+});
+
+function toYaml(data, body) {
+  const frontmatter = Object.keys(data).sort().map(key => key+': '+(data[key] ? JSON.stringify(data[key]) : '')).join("\n");
+  return "---\n" + frontmatter + "\n---\n" + nhm.translate(body);
+}
+
+function date(datestr) {
+  if (datestr) {
+    const [Y,M,D,h,m,s] = datestr.split(/[^0-9]/)
+    return new Date(Date.UTC.call(null, Y,M-1,D,h,m,s)).toISOString();
+  }
+  else
+    return '';
+}
+
+
+data.forEach((node, ix) => {
+  const lang = 'und';
+  const filepath = path.join('out', node.type);
+  const body = node.body.und[0].value;
+  const filename = `${node.uuid}.md`;
+
+  fs.mkdirSync(path.join(__dirname, filepath), { recursive: true });
+  const item = {
+    ix: ix,
+    nid: Number(node.nid),
+    type: node.type,
+    title: node.title,
+    uuid: node.uuid,
+    created: new Date(Number(node.created)*1000).toISOString(),
+    changed: new Date(Number(node.changed)*1000).toISOString(),
+    path: node.path.alias,
+    comment_count: node.comment_count,
+  };
+  switch(node.type) {
+  case 'article':
+    Object.assign(item, {
+      original_author: node.field_original_author?.und?.[0]?.value,
+      featured_image: node.field_featured_image?.und?.[0]?.filename,
+      images: node.field_basic_image_image?.und?.map(item => item.filename),
+      category: node.field_category?.und?.[0]?.tid,
+    });
+    break;
+
+  case 'person':
+    Object.assign(item, {
+      forename_at_birth: node.field_forename_at_birth?.und?.[0]?.value,
+      surname_at_birth: node.field_surname_at_birth?.und?.[0]?.value,
+      other_surnames: node.field_other_surnames?.und?.[0]?.value,
+      other_forenames: node.field_other_forenames?.und?.[0]?.value,
+      title: node.field_title?.und?.[0]?.value,
+      date_of_birth: date(node.field_date_of_birth?.und?.[0]?.value),
+      date_of_death: date(node.field_date_of_death?.und?.[0]?.value),
+      parent_of: node.field_parent_of?.und?.[0]?.value,
+      child_of: node.field_child_of?.und?.[0]?.value,
+      partner_of: node.field_partner_of?.und?.[0]?.value,
+      //    lifetime: node.field_lifetime?.und?.[0]?.value,
+      featured_image: node.field_featured_image?.und?.[0]?.filename,
+    });
+    break;
+  }
+  fs.writeFileSync(path.join(filepath, filename), toYaml(item, body));
+  console.log(item.title);
+});