From b0e04d51860646dacf6e649b5babbd1087500b18 Mon Sep 17 00:00:00 2001
From: Andrew Nesbitt <andrewnez@gmail.com>
Date: Fri, 31 Jul 2026 12:49:24 +0100
Subject: [PATCH] Use magic for binary detection

---
 go.mod       |  5 ++++-
 go.sum       |  2 ++
 pack.go      | 16 ++++++----------
 pack_test.go | 36 +++++++++++++++++++++++++++---------
 4 files changed, 39 insertions(+), 20 deletions(-)

diff --git a/go.mod b/go.mod
index a82b23b..a0a3e8e 100644
--- a/go.mod
+++ b/go.mod
@@ -2,6 +2,9 @@ module github.com/git-pkgs/outline
 
 go 1.25.6
 
-require github.com/odvcencio/gotreesitter v0.45.0
+require (
+	github.com/git-pkgs/magic v0.1.0
+	github.com/odvcencio/gotreesitter v0.45.0
+)
 
 require github.com/git-pkgs/gitignore v1.2.0
diff --git a/go.sum b/go.sum
index 82631d6..8946676 100644
--- a/go.sum
+++ b/go.sum
@@ -1,4 +1,6 @@
 github.com/git-pkgs/gitignore v1.2.0 h1:7vdR8/SvF31dvXqIdC1bKgCvyySefXuN8aY3xJLuFaM=
 github.com/git-pkgs/gitignore v1.2.0/go.mod h1:Lr0XwhbvP071rZF/zIIhkY1gEhFDoWHH91lngwLpeUg=
+github.com/git-pkgs/magic v0.1.0 h1:xLrqq7CMXB9g5bJnmJyKw17Rvlh0GFiEmO6e5RFsoeY=
+github.com/git-pkgs/magic v0.1.0/go.mod h1:3ndidt+yvFaI1M0aEkkzkOlFnLPkeVQASIUojazcxCI=
 github.com/odvcencio/gotreesitter v0.45.0 h1:tt1iWOfcFglz59BQyOz64z2H5XelucxkKK1Lyg0YesU=
 github.com/odvcencio/gotreesitter v0.45.0/go.mod h1:hBVkghd0paaYAVwd2087vfwdeU984bQbMo9LvpE0moo=
diff --git a/pack.go b/pack.go
index 3c53575..2424187 100644
--- a/pack.go
+++ b/pack.go
@@ -1,22 +1,20 @@
 package outline
 
 import (
-	"bytes"
 	"io/fs"
 	"os"
 	"path/filepath"
 	"runtime"
 	"sort"
 	"sync"
-	"unicode/utf8"
 
 	"github.com/git-pkgs/gitignore"
+	"github.com/git-pkgs/magic"
 )
 
 const (
 	defaultMaxFileSize = 1 << 20
 	defaultMaxFiles    = 10000
-	binarySniffLen     = 8192
 )
 
 // Options configures Pack.
@@ -203,7 +201,7 @@ func readFile(root, path string, opts Options) File {
 		f.Skipped = "unreadable"
 		return f
 	}
-	if isBinary(data) {
+	if !isPackableText(data) {
 		f.Skipped = "binary"
 		return f
 	}
@@ -222,10 +220,8 @@ func readFile(root, path string, opts Options) File {
 	return f
 }
 
-func isBinary(data []byte) bool {
-	n := min(len(data), binarySniffLen)
-	if bytes.IndexByte(data[:n], 0) >= 0 {
-		return true
-	}
-	return !utf8.Valid(data[:n])
+func isPackableText(data []byte) bool {
+	result := magic.Detect(data)
+	return result.Kind == magic.KindText &&
+		(result.Encoding == "" || result.Encoding == "utf-8")
 }
diff --git a/pack_test.go b/pack_test.go
index b554e81..d2ef441 100644
--- a/pack_test.go
+++ b/pack_test.go
@@ -29,6 +29,7 @@ func setupRepo(t *testing.T) string {
 	writeFile(t, root, "node_modules/left-pad/index.js", "module.exports = pad")
 	writeFile(t, root, "vendor/thing.go", "package thing")
 	writeFile(t, root, "logo.png", "\x89PNG\r\n\x1a\n\x00\x00")
+	writeFile(t, root, "document.pdf", "%PDF-1.7\n")
 	writeFile(t, root, "huge.txt", strings.Repeat("x", 2000))
 	return root
 }
@@ -85,6 +86,9 @@ func TestPack(t *testing.T) {
 	if f := files["logo.png"]; f.Skipped != "binary" {
 		t.Errorf("logo.png skip = %q, want binary", f.Skipped)
 	}
+	if f := files["document.pdf"]; f.Skipped != "binary" {
+		t.Errorf("document.pdf skip = %q, want binary", f.Skipped)
+	}
 	if f := files["huge.txt"]; f.Skipped != "too-large" {
 		t.Errorf("huge.txt skip = %q, want too-large", f.Skipped)
 	}
@@ -150,20 +154,34 @@ func TestPackUserIgnore(t *testing.T) {
 	}
 }
 
-func TestIsBinary(t *testing.T) {
+func TestIsPackableText(t *testing.T) {
+	const previousProbeSize = 8 << 10
+
 	cases := []struct {
+		name string
 		data []byte
 		want bool
 	}{
-		{[]byte("hello world"), false},
-		{[]byte("hello\x00world"), true},
-		{[]byte{0xff, 0xfe, 0x00}, true},
-		{[]byte("日本語"), false},
-		{nil, false},
+		{name: "ASCII", data: []byte("hello world"), want: true},
+		{name: "NUL", data: []byte("hello\x00world"), want: false},
+		{name: "PNG without NUL", data: []byte("\x89PNG\r\n\x1a\n"), want: false},
+		{name: "disallowed control", data: []byte("hello\x01world"), want: false},
+		{name: "invalid UTF-8", data: []byte{0xff}, want: false},
+		{
+			name: "NUL after old probe",
+			data: append([]byte(strings.Repeat("a", previousProbeSize)), 0),
+			want: false,
+		},
+		{name: "UTF-16LE", data: []byte{0xff, 0xfe, 't', 0, 'e', 0, 'x', 0, 't', 0}, want: false},
+		{name: "UTF-16BE", data: []byte{0xfe, 0xff, 0, 't', 0, 'e', 0, 'x', 0, 't'}, want: false},
+		{name: "Unicode", data: []byte("日本語"), want: true},
+		{name: "empty", data: nil, want: true},
 	}
 	for _, c := range cases {
-		if got := isBinary(c.data); got != c.want {
-			t.Errorf("isBinary(%q) = %v, want %v", c.data, got, c.want)
-		}
+		t.Run(c.name, func(t *testing.T) {
+			if got := isPackableText(c.data); got != c.want {
+				t.Errorf("isPackableText(%q) = %v, want %v", c.data, got, c.want)
+			}
+		})
 	}
 }