Optimize webhooks JSON with compact structure and string interning

Author: ericsciple

.github/workflows/buildtest.yml

@@ -64,3 +64,31 @@ jobs:
           echo ""
           exit 1
         fi
+
+  validate-webhooks:
+    runs-on: ubuntu-latest
+    name: Validate webhook optimization
+
+    steps:
+    - uses: actions/checkout@v4
+    - name: Use Node.js 22.x
+      uses: actions/setup-node@v4
+      with:
+        node-version: 22.x
+        cache: 'npm'
+        registry-url: 'https://npm.pkg.github.com'
+    - run: npm ci
+      env:
+        NODE_AUTH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+    - name: Generate full webhooks file
+      run: cd languageservice && npm run update-webhooks
+    - name: Run optimization validation tests
+      run: cd languageservice && npm test -- --testPathPattern=eventPayloads
+    - name: Verify validation tests ran
+      run: |
+        if [ ! -f languageservice/src/context-providers/events/webhooks.full.validation-complete ]; then
+          echo "ERROR: Validation tests did not run!"
+          echo "The webhooks.full.validation-complete marker file was not created."
+          exit 1
+        fi
+        echo "Validation tests completed at: $(cat languageservice/src/context-providers/events/webhooks.full.validation-complete)"

.gitignore

@@ -7,7 +7,8 @@ node_modules
 # Minified JSON (generated at build time)
 *.min.json
 
-# Intermediate JSON for size comparison (generated by update-webhooks --all)
-*.all.json
-*.drop.json
-*.strip.json
+# Full webhooks source (generated by update-webhooks, used for validation tests)
+*.full.json
+
+# Validation marker (generated by tests)
+*.validation-complete

docs/json-data-files.md

@@ -6,8 +6,9 @@ This document describes the JSON data files used by the language service package
 
 The language service uses several JSON files containing schema definitions, webhook payloads, and other metadata. To reduce bundle size, these files are:
 
-1. **Optimized at generation time** — unused events are dropped, unused fields are stripped
-2. **Minified at build time** — whitespace is removed to produce `.min.json` files
+1. **Optimized at generation time** — unused events are dropped, unused fields are stripped, shared objects are deduplicated, property names are interned
+2. **Compacted using a space-efficient format** — params use type-based dispatch arrays instead of objects
+3. **Minified at build time** — whitespace is removed to produce `.min.json` files
 
 The source `.json` files are human-readable and checked into the repository. The `.min.json` files are generated during build and gitignored.
 
@@ -19,6 +20,7 @@ The source `.json` files are human-readable and checked into the repository. The
 |------|-------------|
 | `src/context-providers/events/webhooks.json` | Webhook event payload schemas for autocompletion |
 | `src/context-providers/events/objects.json` | Deduplicated shared object definitions referenced by webhooks |
+| `src/context-providers/events/strings.json` | Interned property names shared by webhooks and objects |
 | `src/context-providers/events/schedule.json` | Schedule event context data |
 | `src/context-providers/events/workflow_call.json` | Reusable workflow call context data |
 | `src/context-providers/descriptions.json` | Context variable descriptions for hover |
@@ -33,7 +35,7 @@ The source `.json` files are human-readable and checked into the repository. The
 
 ### Webhooks and Objects
 
-The `webhooks.json` and `objects.json` files are generated from the [GitHub REST API description](https://github.com/github/rest-api-description):
+The `webhooks.json`, `objects.json`, and `strings.json` files are generated from the [GitHub REST API description](https://github.com/github/rest-api-description):
 
 ```bash
 cd languageservice
@@ -44,9 +46,11 @@ This script:
 1. Fetches webhook schemas from the GitHub API description
 2. **Validates** all events are categorized (fails if new events are found)
 3. **Drops** events that aren't valid workflow triggers (see [Dropped Events](#dropped-events))
-4. **Strips** unused fields like `description` and `summary` (see [Stripped Fields](#stripped-fields))
-5. **Deduplicates** shared object definitions into `objects.json`
-6. Writes the optimized, pretty-printed JSON files
+4. **Strips** unused fields like `type`, `in`, `isRequired` (see [Stripped Fields](#stripped-fields))
+5. **Compacts** params into a space-efficient array format (see [Compact Format](#compact-format))
+6. **Deduplicates** shared object definitions into `objects.json`
+7. **Interns** duplicate property names into `strings.json` (see [String Interning](#string-interning))
+8. Writes the optimized, pretty-printed JSON files
 
 ### Handling New Webhook Events
 
@@ -101,13 +105,15 @@ The code imports the minified versions:
 
 ```ts
 import webhooks from "./events/webhooks.min.json"
+import objects from "./events/objects.min.json"
+import strings from "./events/strings.min.json"
 ```
 
 ## CI Verification
 
 CI verifies that generated source files are up-to-date:
 
-1. Runs `npm run update-webhooks` to regenerate webhooks.json and objects.json
+1. Runs `npm run update-webhooks` to regenerate webhooks.json, objects.json, and strings.json
 2. Checks for uncommitted changes with `git diff --exit-code`
 
 The `.min.json` files are generated at build time and are not committed to the repository.
@@ -147,4 +153,95 @@ Only `name`, `description`, and `childParamsGroups` are kept — these are used
 
 To compare all fields vs stripped, run `npm run update-webhooks -- --all` and diff the `.all.json` files against the regular ones.
 
-See `EVENT_ACTION_FIELDS` and `BODY_PARAM_FIELDS` in `script/webhooks/index.ts` to modify what gets stripped.
+See `EVENT_ACTION_FIELDS` and `BODY_PARAM_FIELDS` in `script/webhooks/update-webhooks.ts` to modify what gets stripped.
+
+## Compact Format
+
+After stripping, params are further compacted from objects into arrays using type-based dispatch:
+
+| Format | Meaning |
+|--------|---------|
+| `"name"` | Name only (no description, no children) |
+| `[name, desc]` | Name + description (arr[1] is a string) |
+| `[name, children]` | Name + children (arr[1] is an array) |
+| `[name, desc, children]` | Name + description + children |
+
+The reader uses `typeof arr[1]` to determine the format: if it's a string, it's a description; if it's an array, it's children.
+
+**Example:**
+
+```json
+// Before (object format)
+{
+  "name": "issue",
+  "description": "The issue itself.",
+  "childParamsGroups": [
+    { "name": "id" },
+    { "name": "title", "description": "Issue title" }
+  ]
+}
+
+// After (compact format)
+["issue", "The issue itself.", [
+  "id",
+  ["title", "Issue title"]
+]]
+```
+
+## String Interning
+
+Property names that appear 2+ times are "interned" into a shared string table (`strings.json`). In the compact arrays, these names are replaced with numeric indices:
+
+```json
+// strings.json
+["url", "id", "name", ...]  // Index 0 = "url", 1 = "id", 2 = "name"
+
+// webhooks.json - uses indices instead of strings
+["push", [
+  [0, "The URL..."],  // 0 = "url" from string table
+  [1, "Unique ID"],   // 1 = "id"
+  2                   // 2 = "name" (name-only, no description)
+]]
+```
+
+**How to distinguish indices from other values:**
+
+- **Top-level numbers in `p` arrays** → Object indices (references into `objects.json`)
+- **Nested numbers inside compact arrays** → String indices (references into `strings.json`)
+- **Literal strings** → Singletons (names appearing only once, not interned)
+
+Singletons are kept as literal strings for readability and to avoid the overhead of adding rarely-used names to the string table.
+
+## Deduplication
+
+Shared object definitions are extracted into `objects.json` and referenced by index:
+
+```json
+// objects.json
+[
+  ["url", "The URL"],           // Index 0
+  ["id", "Unique identifier"],  // Index 1
+  [...]
+]
+
+// webhooks.json - top-level numbers reference objects
+{
+  "push": {
+    "default": {
+      "p": [0, 1, ["ref", "The git ref"]]  // 0 and 1 are object indices
+    }
+  }
+}
+```
+
+This reduces duplication when the same object structure appears in multiple events (e.g., `repository`, `sender`, `organization`).
+
+## Size Reduction
+
+The optimizations achieve approximately 97% file size reduction:
+
+| Stage | Minified | Gzip |
+|-------|----------|------|
+| Original (webhooks.full.json) | 6.7 MB | 310 KB |
+| After optimization | 209 KB | 22 KB |
+| **Reduction** | **97%** | **93%** |

languageservice/package.json

@@ -37,13 +37,13 @@
     "format-check": "prettier --check '**/*.ts'",
     "lint": "eslint 'src/**/*.ts'",
     "lint-fix": "eslint --fix 'src/**/*.ts'",
-    "minify-json": "node ../script/minify-json.js src/context-providers/descriptions.json src/context-providers/events/webhooks.json src/context-providers/events/objects.json src/context-providers/events/schedule.json src/context-providers/events/workflow_call.json",
+    "minify-json": "node ../script/minify-json.js src/context-providers/descriptions.json src/context-providers/events/webhooks.json src/context-providers/events/objects.json src/context-providers/events/strings.json src/context-providers/events/schedule.json src/context-providers/events/workflow_call.json",
     "prebuild": "npm run minify-json",
     "prepublishOnly": "npm run build && npm run test",
     "pretest": "npm run minify-json",
     "test": "NODE_OPTIONS=\"--experimental-vm-modules\" jest",
     "test-watch": "NODE_OPTIONS=\"--experimental-vm-modules\" jest --watch",
-    "update-webhooks": "npx tsx script/webhooks/index.ts",
+    "update-webhooks": "npx tsx script/webhooks/update-webhooks.ts",
     "watch": "tsc --build tsconfig.build.json --watch"
   },
   "dependencies": {

languageservice/script/webhooks/deduplicate.ts

@@ -1,5 +1,38 @@
 import Webhook from "./webhook";
 
+/**
+ * Get the name from a param.
+ * Formats: "name" (string), or [name, ...] (array)
+ */
+function getParamName(param: any): string {
+  if (typeof param === "string") {
+    return param;
+  }
+  if (Array.isArray(param)) {
+    return param[0];
+  }
+  return param.name;
+}
+
+/**
+ * Get params from a webhook action.
+ * Uses 'p' (short key) if present, falls back to 'bodyParameters'
+ */
+function getParams(webhook: any): any[] {
+  return webhook.p || webhook.bodyParameters || [];
+}
+
+/**
+ * Set params on a webhook action using the short key 'p'
+ */
+function setParams(webhook: any, params: any[]): void {
+  if (webhook.p !== undefined) {
+    webhook.p = params;
+  } else {
+    webhook.bodyParameters = params;
+  }
+}
+
 // Store any repeated body parameters in an array
 // and replace them in the webhook with an index in the array
 export function deduplicateWebhooks(webhooks: Record<string, Record<string, Webhook>>): any[] {
@@ -10,10 +43,11 @@ export function deduplicateWebhooks(webhooks: Record<string, Record<string, Webh
   const objectCount: Record<string, number> = {};
 
   for (const webhook of iterateWebhooks(webhooks)) {
-    for (const param of webhook.bodyParameters) {
-      objectsByName[param.name] ||= [];
-      const index = findOrAdd(param, objectsByName[param.name]);
-      const key = `${param.name}:${index}`;
+    for (const param of getParams(webhook)) {
+      const name = getParamName(param);
+      objectsByName[name] ||= [];
+      const index = findOrAdd(param, objectsByName[name]);
+      const key = `${name}:${index}`;
       objectCount[key] ||= 0;
       objectCount[key]++;
     }
@@ -27,18 +61,19 @@ export function deduplicateWebhooks(webhooks: Record<string, Record<string, Webh
 
   for (const webhook of iterateWebhooks(webhooks)) {
     const newParams: any[] = [];
-    for (const param of webhook.bodyParameters) {
-      const index = find(param, objectsByName[param.name]);
-      const key = `${param.name}:${index}`;
+    for (const param of getParams(webhook)) {
+      const name = getParamName(param);
+      const index = find(param, objectsByName[name]);
+      const key = `${name}:${index}`;
       if (objectCount[key] > 1) {
-        newParams.push(indexForParam(param, index, bodyParamIndexMap, duplicatedBodyParams));
+        newParams.push(indexForParam(param, name, index, bodyParamIndexMap, duplicatedBodyParams));
       } else {
         // If an object is only used once, keep it inline
         newParams.push(param);
       }
     }
 
-    webhook.bodyParameters = newParams;
+    setParams(webhook, newParams);
   }
 
   return duplicatedBodyParams;
@@ -74,11 +109,12 @@ function find(param: any, objects: any[]): number {
 
 function indexForParam(
   param: any,
+  paramName: string,
   paramNameIndex: number,
   objectIndexMap: Record<string, number>,
   duplicatedBodyParams: any[]
 ): number {
-  const key = `${param.name}:${paramNameIndex}`;
+  const key = `${paramName}:${paramNameIndex}`;
 
   const existingIndex = objectIndexMap[key];
   if (existingIndex !== undefined) {