refactor: revisions data file to be smaller

Author: limulus

11ty/revisions.ts

@@ -2,16 +2,21 @@ import { execFileSync } from 'node:child_process'
 import { createHash } from 'node:crypto'
 import { readFile, writeFile } from 'node:fs/promises'
 
-export const HASH_ALGORITHM = 'shake128'
-export const HASH_OUTPUT_BYTES = 6
 export const REVISIONS_PATH = 'www/_data/revisions.json'
 
-export interface Revision {
-  hash: string
+const HASH_ALGORITHM = 'shake128'
+const HASH_OUTPUT_BYTES = 6
+
+export interface RevisionVersion {
   algorithm: string
   outputBytes: number
   fields: string[]
   separator: string
+}
+
+export interface Revision {
+  v: number
+  hash: string
   date: string
   commit: string
   message?: string
@@ -23,7 +28,10 @@ export interface RevisionEntry {
   generatedTeaserHash?: string
 }
 
-export type RevisionsFile = Record<string, RevisionEntry>
+export interface RevisionsFile {
+  versions: Record<string, RevisionVersion>
+  entries: Record<string, RevisionEntry>
+}
 
 export function computeHash(
   title: string,
@@ -39,10 +47,6 @@ export function computeHash(
   return hash.digest('hex')
 }
 
-export function hashFields(subhead: string | undefined): string[] {
-  return subhead != null ? ['title', 'subhead', 'content'] : ['title', 'content']
-}
-
 export function computeTeaserHash(teaser: string): string {
   const hash = createHash(HASH_ALGORITHM, {
     outputLength: HASH_OUTPUT_BYTES,
@@ -79,17 +83,21 @@ export async function loadRevisions(path: string = REVISIONS_PATH): Promise<Revi
     const content = await readFile(path, 'utf-8')
     return JSON.parse(content) as RevisionsFile
   } catch {
-    return {}
+    return { versions: {}, entries: {} }
   }
 }
 
 export async function saveRevisions(
   revisions: RevisionsFile,
   path: string = REVISIONS_PATH
 ): Promise<void> {
-  const sorted: RevisionsFile = {}
-  for (const key of Object.keys(revisions).sort()) {
-    sorted[key] = revisions[key]
+  const sortedEntries: Record<string, RevisionEntry> = {}
+  for (const key of Object.keys(revisions.entries).sort()) {
+    sortedEntries[key] = revisions.entries[key]
+  }
+  const output: RevisionsFile = {
+    versions: revisions.versions,
+    entries: sortedEntries,
   }
-  await writeFile(path, JSON.stringify(sorted, null, 2) + '\n')
+  await writeFile(path, JSON.stringify(output, null, 2) + '\n')
 }

11ty/update-revisions.ts

@@ -3,12 +3,9 @@ import matter from 'gray-matter'
 import { readFile } from 'node:fs/promises'
 
 import {
-  HASH_ALGORITHM,
-  HASH_OUTPUT_BYTES,
   computeHash,
   getCommitMessage,
   getCurrentCommit,
-  hashFields,
   loadRevisions,
   saveRevisions,
 } from './revisions.ts'
@@ -32,24 +29,20 @@ async function main() {
     const content = parsed.content
 
     const hash = computeHash(title, subhead, content)
-    const fields = hashFields(subhead)
 
-    const entry = revisions[filePath] ?? { revisions: [] }
+    const entry = revisions.entries[filePath] ?? { revisions: [] }
     const latestRevision = entry.revisions[entry.revisions.length - 1]
 
     if (!latestRevision || latestRevision.hash !== hash) {
       const message = getCommitMessage(commit)
       entry.revisions.push({
+        v: 1,
         hash,
-        algorithm: HASH_ALGORITHM,
-        outputBytes: HASH_OUTPUT_BYTES,
-        fields,
-        separator: '\\n',
         date: new Date().toISOString(),
         commit,
         ...(message && { message }),
       })
-      revisions[filePath] = entry
+      revisions.entries[filePath] = entry
       revisionCount++
     }
   }

bin/backfill-revision-messages.ts

@@ -9,7 +9,7 @@ async function main() {
   const revisions = await loadRevisions()
   let updated = 0
 
-  for (const entry of Object.values(revisions)) {
+  for (const entry of Object.values(revisions.entries)) {
     for (const revision of entry.revisions) {
       if (!revision.message) {
         const message = getCommitMessage(revision.commit)

bin/backfill-revisions.ts

@@ -5,11 +5,8 @@ import { readFile } from 'node:fs/promises'
 
 import {
   type RevisionsFile,
-  HASH_ALGORITHM,
-  HASH_OUTPUT_BYTES,
   REVISIONS_PATH,
   computeHash,
-  hashFields,
   saveRevisions,
 } from '../11ty/revisions.ts'
 
@@ -50,7 +47,17 @@ function getFileAtCommit(sha: string, filePath: string): string | null {
 
 async function main() {
   const files = await globby('www/**/*.md')
-  const revisions: RevisionsFile = {}
+  const revisions: RevisionsFile = {
+    versions: {
+      '1': {
+        algorithm: 'shake128',
+        outputBytes: 6,
+        fields: ['title', 'subhead', 'content'],
+        separator: '\n',
+      },
+    },
+    entries: {},
+  }
 
   let totalFiles = 0
   let totalRevisions = 0
@@ -70,7 +77,7 @@ async function main() {
       continue
     }
 
-    const entry: RevisionsFile[string] = { revisions: [] }
+    const entry: RevisionsFile['entries'][string] = { revisions: [] }
     let previousHash: string | null = null
 
     for (const logEntry of logEntries) {
@@ -93,11 +100,8 @@ async function main() {
       if (hash === previousHash) continue
 
       entry.revisions.push({
+        v: 1,
         hash,
-        algorithm: HASH_ALGORITHM,
-        outputBytes: HASH_OUTPUT_BYTES,
-        fields: hashFields(subhead),
-        separator: '\\n',
         date: logEntry.date,
         commit: logEntry.shortSha,
       })
@@ -106,7 +110,7 @@ async function main() {
     }
 
     if (entry.revisions.length > 0) {
-      revisions[filePath] = entry
+      revisions.entries[filePath] = entry
     }
   }
 

bin/generate-teasers.spec.ts

@@ -6,7 +6,7 @@ import {
   insertTeaserIntoFrontmatter,
   wrapText,
 } from './generate-teasers.ts'
-import { computeHash, hashFields } from '../11ty/revisions.ts'
+import { computeHash } from '../11ty/revisions.ts'
 
 describe('computeHash', () => {
   it('produces a 12-character hex string', () => {
@@ -33,16 +33,6 @@ describe('computeHash', () => {
   })
 })
 
-describe('hashFields', () => {
-  it('returns title and content when no subhead', () => {
-    expect(hashFields(undefined)).toEqual(['title', 'content'])
-  })
-
-  it('includes subhead when present', () => {
-    expect(hashFields('A subhead')).toEqual(['title', 'subhead', 'content'])
-  })
-})
-
 describe('wrapText', () => {
   it('returns short text unchanged', () => {
     expect(wrapText('Hello world', 90)).toBe('Hello world')

bin/generate-teasers.ts

@@ -169,12 +169,13 @@ async function main() {
 
     const hash = computeHash(title, subhead, content)
 
-    const inRevisions = filePath in revisions && revisions[filePath].teaserHash != null
-    const hashMatch = inRevisions && revisions[filePath].teaserHash === hash
+    const inRevisions =
+      filePath in revisions.entries && revisions.entries[filePath].teaserHash != null
+    const hashMatch = inRevisions && revisions.entries[filePath].teaserHash === hash
 
     // Check if teaser was manually edited by comparing its hash to the stored generated hash
     const currentTeaserHash = teaser != null ? computeTeaserHash(teaser) : null
-    const storedTeaserHash = revisions[filePath]?.generatedTeaserHash ?? null
+    const storedTeaserHash = revisions.entries[filePath]?.generatedTeaserHash ?? null
     const teaserManuallyEdited =
       currentTeaserHash != null &&
       storedTeaserHash != null &&
@@ -206,9 +207,9 @@ async function main() {
     const updatedContent = insertTeaserIntoFrontmatter(fileContent, newTeaser)
     await writeFile(filePath, updatedContent)
 
-    revisions[filePath] ??= { revisions: [] }
-    revisions[filePath].teaserHash = hash
-    revisions[filePath].generatedTeaserHash = computeTeaserHash(newTeaser)
+    revisions.entries[filePath] ??= { revisions: [] }
+    revisions.entries[filePath].teaserHash = hash
+    revisions.entries[filePath].generatedTeaserHash = computeTeaserHash(newTeaser)
     teaserHashChanged = true
     generatedCount++
   }

www/_data/eleventyComputed.js

@@ -72,7 +72,7 @@ export default {
   // ISO date of the most recent revision, if modified after creation
   lastModified: (data) => {
     const filePath = data.page?.inputPath?.replace(/^\.\//, '')
-    const revs = filePath && data.revisions?.[filePath]?.revisions
+    const revs = filePath && data.revisions?.entries?.[filePath]?.revisions
     if (revs?.length > 1) {
       return revs.at(-1).date
     }