astro-relatinator 1.0.4 → 1.2.0

package/README.md

@@ -3,3 +3,72 @@
 An Astro integration for the [humble library for finding related posts and content](https://github.com/DBozhinovski/relatinator/tree/master/packages/relatinator).
 
 ## Installation
+
+Assuming you're running Astro v3 or later, you can install this integration by running:
+
+```
+npx astro add astro-relatinator
+```
+
+OR 
+
+```
+npm install astro-relatinator
+```
+
+## Configuration
+
+The integration takes a single input argument, `paths`, which is an array of paths to the `.md` or `.mdx` files you want to use for training the model. If you're using the [Content Collections API](https://docs.astro.build/en/guides/content-collections/), you can find these at `./src/collections/<collection-name>/`.
+
+### Example:
+
+```ts
+export default defineConfig({
+  // ...
+  integrations: [
+    // ...
+    relatinatorIntegration({
+      paths: ["./src/content/posts"], // path yo your content files
+      schema: ["title", "descriptions", "tags"], // frontmatter fields to use for training
+    }),
+    // ...
+  ],
+});
+```
+
+__Note:__ the fields you specify in the `schema` array must be present in the frontmatter of the files you're training on. If they're not, the integration will throw an error or lead to unexpected results.
+
+## Usage
+
+The integration hooks into the Astro build (or dev) process, and will traing the tf-idf model based on the files you've provided. You can then use the `find` function from the package to find related content.
+
+### Example:
+
+```astro
+---
+import { getCollection } from "astro:content";
+import { find } from "astro-relatinator";
+
+const blogEntries = await getCollection("posts"); // assuming your collection is called posts
+// let's say we want to compare against the first blog entry
+const first = blogEntries[0];
+
+// Concatenate the same fields we used for training and declare how many related posts we want to find
+const related = find(`${first.frontmatter.title} ${first.frontmatter.description} ${first.frontmatter.tags.join(' ')} ${first.body}`, 3)
+
+---
+
+<div>
+<!-- render your related posts here -->
+{ related.map((id) => {
+  const post = posts.find((post) => post.id === id);
+  return (
+    <a href={post.slug}>
+      <h2>{post.frontmatter.title}</h2>
+      <p>{post.frontmatter.description}</p>
+    </a>
+  );
+}) }
+</div>
+
+```

package/integration.ts

@@ -7,7 +7,19 @@ import fs from "node:fs/promises";
 import { TfIdfDocument } from "natural";
 import { Document } from "relatinator";
 
-async function trainModel(paths: string[]) {
+const stringifyData = (data: any): string => {
+  if (typeof data === 'object') {
+    if (Array.isArray(data)) {
+      return data.map(stringifyData).join(' ');
+    } else {
+      return Object.values(data).map(stringifyData).join(' ');
+    }
+  } else {
+    return String(data);
+  }
+}
+
+async function trainModel(paths: string[], schema: string[]) {
   const documents: Document[] = [];
 
   if (tfIdf.documents) {
@@ -26,11 +38,17 @@ async function trainModel(paths: string[]) {
       const idArr = file.split("/");
       const id = idArr[idArr.length - 1];
 
+      let documentContent = '';
+      for (const key of schema) {
+        if (parsed.data[key]) {
+          documentContent += ' ' + stringifyData(parsed.data[key]);
+        }
+      }
+      documentContent += ' ' + parsed.content;
+
       documents.push({
         id,
-        content: `${parsed.data.title} ${
-          parsed.data.description
-        } ${parsed.data.tags.join(" ")} ${parsed.content}`,
+        content: documentContent,
       });
     }
   }
@@ -40,8 +58,10 @@ async function trainModel(paths: string[]) {
 
 const relatinatorIntegration = ({
   paths,
+  schema,
 }: {
   paths: string[];
+  schema: string[];
 }): AstroIntegration => {
   return {
     name: "astro-relatinator",
@@ -50,7 +70,7 @@ const relatinatorIntegration = ({
         logger.info("relatinator server:setup");
 
         logger.info("Training TF-IDF on existing data...");
-        await trainModel(paths);
+        await trainModel(paths, schema);
         logger.info("Initial training done.");
 
         server.watcher.on("all", async (eventName, filePath) => {
@@ -62,7 +82,7 @@ const relatinatorIntegration = ({
             logger.info(
               `Markdown file changed: ${filePath}; Re-training TF-IDF...`
             );
-            await trainModel(paths);
+            await trainModel(paths, schema);
           } else {
             return;
           }
@@ -72,7 +92,7 @@ const relatinatorIntegration = ({
         logger.info("relatinator build:setup");
 
         logger.info("Training TF-IDF on existing data...");
-        await trainModel(paths);
+        await trainModel(paths, schema);
         logger.info("Training done.");
       }
     },

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "astro-relatinator",
-  "version": "1.0.4",
+  "version": "1.2.0",
   "description": "Find related content in your Astro website or app",
   "author": "Darko Bozhinovski <dbozhinovski@gmail.com>",
   "license": "MIT",
@@ -13,7 +13,8 @@
     "words",
     "posts",
     "articles",
-    "static"
+    "static",
+    "astro-integration"
   ],
   "files": [
     "integration.ts"