feat: add offline support for linguist

Signed-off-by: Jos Craw <me.joscraw@gmail.com>

.changeset/pink-dolls-unite.md

@@ -0,0 +1,5 @@
+---
+'@backstage/plugin-linguist-backend': minor
+---
+
+Added support for offline language data

plugins/linguist-backend/README.md

@@ -85,6 +85,14 @@ return createRouter({ schedule: schedule, age: { days: 30 } }, { ...env });
 
 With the `age` setup like this if the language breakdown is older than 15 days it will get regenerated. It's recommended that if you choose to use this configuration to set it to a large value - 30, 90, or 180 - as this data generally does not change drastically.
 
+## Offline
+
+The default setup will pull the language data from GitHub, by setting `offline` to `true` a packaged offline version of this data is used instead.
+
+```ts
+return createRouter({ schedule: schedule, offline: true }, { ...env });
+```
+
 ## Use Source Location
 
 You may wish to use the `backstage.io/source-location` annotation over using the `backstage.io/linguist` as you may not be able to quickly add that annotation to your Entities. To do this you'll just need to set the `useSourceLocation` boolean to `true` in your `packages/backend/src/plugins/linguist.ts` when you call `createRouter`:

plugins/linguist-backend/api-report.md

@@ -35,6 +35,7 @@ export class LinguistBackendApi {
     batchSize?: number,
     useSourceLocation?: boolean,
     kind?: string[],
+    offline?: boolean,
   );
   // (undocumented)
   getEntityLanguages(entityRef: string): Promise<Languages>;
@@ -82,6 +83,8 @@ export interface PluginOptions {
   // (undocumented)
   kind?: string[];
   // (undocumented)
+  offline?: boolean;
+  // (undocumented)
   schedule?: TaskScheduleDefinition;
   // (undocumented)
   useSourceLocation?: boolean;

plugins/linguist-backend/src/api/LinguistBackendApi.ts

@@ -57,6 +57,7 @@ export class LinguistBackendApi {
   private readonly batchSize?: number;
   private readonly useSourceLocation?: boolean;
   private readonly kind: string[];
+  private readonly offline?: boolean;
   public constructor(
     logger: Logger,
     store: LinguistBackendStore,
@@ -67,6 +68,7 @@ export class LinguistBackendApi {
     batchSize?: number,
     useSourceLocation?: boolean,
     kind?: string[],
+    offline?: boolean,
   ) {
     this.logger = logger;
     this.store = store;
@@ -78,6 +80,7 @@ export class LinguistBackendApi {
     this.age = age;
     this.useSourceLocation = useSourceLocation;
     this.kind = kindOrDefault(kind);
+    this.offline = offline;
   }
 
   public async getEntityLanguages(entityRef: string): Promise<Languages> {
@@ -190,7 +193,7 @@ export class LinguistBackendApi {
     const readTreeResponse = await this.urlReader.readTree(url);
     const dir = await readTreeResponse.dir();
 
-    const results = await linguist(dir);
+    const results = await linguist(dir, { offline: this.offline });
 
     try {
       const totalBytes = results.languages.bytes;

plugins/linguist-backend/src/service/router.ts

@@ -38,6 +38,7 @@ export interface PluginOptions {
   age?: HumanDuration;
   batchSize?: number;
   useSourceLocation?: boolean;
+  offline?: boolean;
   kind?: string[];
 }
 
@@ -57,7 +58,8 @@ export async function createRouter(
   pluginOptions: PluginOptions,
   routerOptions: RouterOptions,
 ): Promise<express.Router> {
-  const { schedule, age, batchSize, useSourceLocation, kind } = pluginOptions;
+  const { schedule, age, batchSize, useSourceLocation, kind, offline } =
+    pluginOptions;
 
   const { logger, reader, database, discovery, scheduler, tokenManager } =
     routerOptions;
@@ -78,6 +80,7 @@ export async function createRouter(
       batchSize,
       useSourceLocation,
       kind,
+      offline,
     );
 
   if (scheduler && schedule) {