From e680d55ca2d7f6b213e2a8693eba6be39163ba41 Mon Sep 17 00:00:00 2001
From: Samuel Newman <mozzius@protonmail.com>
Date: Thu, 31 Oct 2024 18:09:49 +0000
Subject: [PATCH] Filter out invalid facets in RichText (#2933)

* add failing test

* simplify test

* check facet was removed

* filter out invalid facets

* changeset
---
 .changeset/swift-sheep-confess.md       |  5 +++
 packages/api/src/rich-text/rich-text.ts |  8 +++--
 packages/api/tests/rich-text.test.ts    | 44 +++++++++++++++++++++++++
 3 files changed, 55 insertions(+), 2 deletions(-)
 create mode 100644 .changeset/swift-sheep-confess.md

diff --git a/.changeset/swift-sheep-confess.md b/.changeset/swift-sheep-confess.md
new file mode 100644
index 00000000000..e4723e584ca
--- /dev/null
+++ b/.changeset/swift-sheep-confess.md
@@ -0,0 +1,5 @@
+---
+"@atproto/api": patch
+---
+
+Fix handling of invalid facets in RichText
diff --git a/packages/api/src/rich-text/rich-text.ts b/packages/api/src/rich-text/rich-text.ts
index 261ea7e4e68..a9c037b33bf 100644
--- a/packages/api/src/rich-text/rich-text.ts
+++ b/packages/api/src/rich-text/rich-text.ts
@@ -169,7 +169,7 @@ export class RichText {
       this.facets = entitiesToFacets(this.unicodeText, props.entities)
     }
     if (this.facets) {
-      this.facets.sort(facetSort)
+      this.facets = this.facets.filter(facetFilter).sort(facetSort)
     }
     if (opts?.cleanNewlines) {
       sanitizeRichText(this, { cleanNewlines: true }).copyInto(this)
@@ -378,7 +378,11 @@ export class RichText {
   }
 }
 
-const facetSort = (a, b) => a.index.byteStart - b.index.byteStart
+const facetSort = (a: Facet, b: Facet) => a.index.byteStart - b.index.byteStart
+
+const facetFilter = (facet: Facet) =>
+  // discard negative-length facets. zero-length facets are valid
+  facet.index.byteStart <= facet.index.byteEnd
 
 function entitiesToFacets(text: UnicodeString, entities: Entity[]): Facet[] {
   const facets: Facet[] = []
diff --git a/packages/api/tests/rich-text.test.ts b/packages/api/tests/rich-text.test.ts
index 5c30edcf94f..7683babb3aa 100644
--- a/packages/api/tests/rich-text.test.ts
+++ b/packages/api/tests/rich-text.test.ts
@@ -658,4 +658,48 @@ describe('RichText#segments', () => {
       },
     ])
   })
+
+  it("doesn't duplicate text when negative-length facets are present", () => {
+    const input = {
+      text: 'hello world',
+      facets: [
+        // invalid zero-length
+        {
+          features: [],
+          index: {
+            byteStart: 6,
+            byteEnd: 0,
+          },
+        },
+        // valid normal facet
+        {
+          features: [],
+          index: {
+            byteEnd: 11,
+            byteStart: 6,
+          },
+        },
+        // valid zero-length
+        {
+          features: [],
+          index: {
+            byteEnd: 0,
+            byteStart: 0,
+          },
+        },
+      ],
+    }
+
+    const rt = new RichText(input)
+
+    let output = ''
+    for (const segment of rt.segments()) {
+      output += segment.text
+    }
+
+    expect(output).toEqual(input.text)
+
+    // invalid one should have been removed
+    expect(rt.facets?.length).toEqual(2)
+  })
 })