fix(search): escape search excerpts to prevent stored XSS

packages/db/src/search-repo.test.ts

@@ -10,6 +10,7 @@ import {
   PostgresSearchRepository,
   SEARCH_DOCUMENT_VERSION,
   indexedSubjectSql,
+  renderExcerptHtml,
   searchVectorSql,
 } from './search-repo'
 
@@ -256,6 +257,23 @@ describe('matching and ranking', () => {
     expect(results.hits[0]?.excerpt).toMatch(/<b>/)
   })
 
+  it('escapes HTML in a body excerpt so an injected payload is inert', async () => {
+    await seed({ id: 1, message: 'before <img src=x onerror=alert(1)> a kestrel flew past' })
+
+    const excerpt = (await repo.search(query(), scope())).hits[0]?.excerpt ?? ''
+    expect(excerpt).not.toContain('<img')
+    expect(excerpt).toContain('&lt;img src=x onerror=alert(1)&gt;')
+    expect(excerpt).toMatch(/<b>kestrel<\/b>/i)
+  })
+
+  it('leaves no tag in an excerpt other than the highlight marks', async () => {
+    await seed({ id: 1, message: 'kestrel "><svg onload=alert(1)></svg>' })
+
+    const excerpt = (await repo.search(query(), scope())).hits[0]?.excerpt ?? ''
+    expect(excerpt).not.toContain('<svg')
+    expect(excerpt).not.toMatch(/<(?!\/?b>)/)
+  })
+
   it('runs no query at all for empty terms', async () => {
     await seed({ id: 1, message: 'kestrel' })
     expect((await repo.search(query({ terms: '  ' }), scope())).hits).toEqual([])
@@ -536,3 +554,17 @@ describe('reindexing', () => {
     expect(await repo.indexProgress()).toEqual({ indexed: 1, pending: 3 })
   })
 })
+
+describe('renderExcerptHtml', () => {
+  const start = String.fromCharCode(0xe000)
+  const stop = String.fromCharCode(0xe001)
+
+  it('escapes HTML and renders only the highlight sentinels as marks', () => {
+    const raw = `<script>alert(1)</script> ${start}kestrel${stop}`
+    expect(renderExcerptHtml(raw)).toBe('&lt;script&gt;alert(1)&lt;/script&gt; <b>kestrel</b>')
+  })
+
+  it('escapes quotes and ampersands so an attribute cannot break out', () => {
+    expect(renderExcerptHtml('a & b "><svg>')).toBe('a &amp; b &quot;&gt;&lt;svg&gt;')
+  })
+})

packages/db/src/search-repo.ts

@@ -23,6 +23,33 @@ const TOP_AUTHORS = 24
 
 export const SEARCH_DOCUMENT_VERSION = 1
 
+const EXCERPT_HIGHLIGHT_START = String.fromCharCode(0xe000)
+const EXCERPT_HIGHLIGHT_STOP = String.fromCharCode(0xe001)
+
+const HEADLINE_OPTIONS =
+  `StartSel=${EXCERPT_HIGHLIGHT_START}, StopSel=${EXCERPT_HIGHLIGHT_STOP}, ` +
+  `MaxFragments=1, MaxWords=40, MinWords=15`
+
+const EXCERPT_HTML_ESCAPES: Readonly<Record<string, string>> = {
+  '&': '&amp;',
+  '<': '&lt;',
+  '>': '&gt;',
+  '"': '&quot;',
+  "'": '&#39;',
+}
+
+export function renderExcerptHtml(headline: string): string {
+  const escaped = headline.replace(
+    /[&<>"']/g,
+    (character) => EXCERPT_HTML_ESCAPES[character] ?? character,
+  )
+  return escaped
+    .split(EXCERPT_HIGHLIGHT_START)
+    .join('<b>')
+    .split(EXCERPT_HIGHLIGHT_STOP)
+    .join('</b>')
+}
+
 export function searchVectorSql(subject: SQL | string, message: SQL | string): SQL {
   return sql`
     setweight(to_tsvector(${SEARCH_CONFIG}, coalesce(${subject}, '')), 'A') ||
@@ -78,7 +105,7 @@ export class PostgresSearchRepository {
                author_user_id, author_username, created_at, rank,
                ts_headline(${SEARCH_CONFIG}, excerpt_source,
                            websearch_to_tsquery(${SEARCH_CONFIG}, ${query.terms}),
-                           'MaxFragments=1, MaxWords=40, MinWords=15') as excerpt
+                           ${HEADLINE_OPTIONS}) as excerpt
           from ${candidates}
          order by ${finalOrder}
          limit ${query.limit}
@@ -94,7 +121,7 @@ export class PostgresSearchRepository {
       authorUserId: row.author_user_id === null ? null : Number(row.author_user_id),
       authorUsername: String(row.author_username),
       postedAt: row.created_at instanceof Date ? row.created_at : new Date(String(row.created_at)),
-      excerpt: String(row.excerpt ?? ''),
+      excerpt: renderExcerptHtml(String(row.excerpt ?? '')),
       rank: Number(row.rank),
     }))