Merge pull request #857 from okcoker/feature/tag-normalization

feat: Clean up tags

gatsby-node.js

@@ -1,26 +1,40 @@
 import people from './src/data.js';
-import { tags, countries, devices } from './src/util/stats';
+import { tags, countries, devices, normalizeTag } from './src/util/stats';
+
+function unique(arr) {
+  return Array.from(new Set(arr));
+}
 
 function sourceNodes({ actions, createNodeId, createContentDigest }) {
+  const normalizedTagMap = tags().reduce((acc, tag) => {
+    const normalizedTag = normalizeTag(tag.name);
+    acc[normalizedTag] = tag.name;
+    return acc;
+  }, {});
   // Add People to the GraphQL API, we randomize the data on each build so no one gets their feelings hurt
   people
     .sort(() => Math.random() - 0.5)
-    .forEach(p => {
-      // Remove duplicated tags.
-      const person = { ...p, tags: [...new Set(p.tags)] };
+    .forEach(person => {
+      const normalizedPerson = {
+        ...person,
+        // Clean out people that added basically the same tags twice
+        tags: unique(
+          person.tags.map(tag => normalizedTagMap[normalizeTag(tag)] || tag)
+        ),
+      };
       const nodeMeta = {
-        id: createNodeId(`person-${person.name}`),
+        id: createNodeId(`person-${normalizedPerson.name}`),
         parent: null,
         children: [],
         internal: {
           type: `Person`,
           mediaType: `text/html`,
-          content: JSON.stringify(person),
-          contentDigest: createContentDigest(person),
+          content: JSON.stringify(normalizedPerson),
+          contentDigest: createContentDigest(normalizedPerson),
         },
       };
 
-      actions.createNode({ ...person, ...nodeMeta });
+      actions.createNode({ ...normalizedPerson, ...nodeMeta });
     });
 
   // Add tags to GraphQL API

src/util/stats.js

@@ -14,6 +14,23 @@ function countInstances(acc, tag) {
   return acc;
 }
 
+export function normalizeTag(tag) {
+  return (
+    tag
+      // Common mispellings currently seen in the data
+      // Do we want to go this far?
+      .replace(/frontend/i, 'Front End')
+      .replace(/backend/i, 'Back End')
+      .replace(/fullstack/i, 'Full Stack')
+      .replace(/a11y/i, 'Accessibility')
+      .replace(/next.?js/i, 'Next')
+      .replace(/react.?js/i, 'React')
+
+      // Or is lowercase enough?
+      .toLowerCase()
+  );
+}
+
 export function countries() {
   const data = people
     .map(person => ({
@@ -50,7 +67,25 @@ export function tags() {
     .filter(([, count]) => count >= 3)
     .map(([name, count]) => ({ name, count }));
 
-  return [{ name: 'all', count: people.length }, ...tags];
+  const lowercaseTagMap = tags.reduce((acc, tag) => {
+    const normalizedName = normalizeTag(tag.name);
+    const currentCount = acc[normalizedName] || 0;
+    acc[normalizedName] = currentCount + tag.count;
+    return acc;
+  }, {});
+
+  // Merge tags like "JavaScript" and "Javascript" based on the
+  // count… Event though it's obviously JavaScript!
+  const normalizedTags = tags.reduce((acc, { name }) => {
+    const normalizedName = normalizeTag(name);
+    if (typeof lowercaseTagMap[normalizedName] !== 'undefined') {
+      acc.push({ name, count: lowercaseTagMap[normalizedName] });
+      delete lowercaseTagMap[normalizedName];
+    }
+    return acc;
+  }, []);
+
+  return [{ name: 'all', count: people.length }, ...normalizedTags];
 }
 
 export function devices() {