Embed binary files as base64 encoded strings

They need to be broken into smaller lines; yqlib doesn't do this,
and yamlfmt fails with a buffer overflow when it tries to keep
existing line breaks.

Signed-off-by: Jan Dubois <[email protected]>

Author: jandubois

pkg/limatmpl/embed.go

@@ -6,11 +6,14 @@ package limatmpl
 import (
 	"bytes"
 	"context"
+	"encoding/base64"
 	"fmt"
 	"os"
 	"path/filepath"
 	"slices"
+	"strings"
 	"sync"
+	"unicode"
 
 	"github.com/coreos/go-semver/semver"
 	"github.com/lima-vm/lima/pkg/limayaml"
@@ -254,7 +257,7 @@ const mergeDocuments = `
 | $a | (select(.mountTypesUnsupported) | .mountTypesUnsupported) |= unique
 
 # Remove the custom tags again so they do not clutter up the YAML output.
-| $a | .. tag = ""
+| $a | .. | select(tag == "!!tag") tag = ""
 `
 
 // listFields returns dst and src fields like "list[idx].field".
@@ -552,11 +555,49 @@ func (tmpl *Template) combineNetworks() {
 	}
 }
 
+// Break base64 strings into shorter chunks; otherwise the yamlfmt code will fail
+// with a buffer overflow while trying to retain line breaks.
+const lineLength = 76
+
+// binaryString returns a base64 encoded version of the binary string, broken into chunks
+// of at most lineLength characters per line.
+func binaryString(s string) string {
+	encoded := base64.StdEncoding.EncodeToString([]byte(s))
+	if len(encoded) <= lineLength {
+		return encoded
+	}
+
+	// Estimate capacity: encoded length + number of newlines
+	lineCount := (len(encoded) + lineLength - 1) / lineLength
+	builder := strings.Builder{}
+	builder.Grow(len(encoded) + lineCount)
+
+	for i := 0; i < len(encoded); i += lineLength {
+		end := i + lineLength
+		if end > len(encoded) {
+			end = len(encoded)
+		}
+		builder.WriteString(encoded[i:end])
+		builder.WriteByte('\n')
+	}
+
+	return builder.String()
+}
+
 // updateScript replaces a "file" property with the actual script and then renames the field to newName ("script" or "content").
 func (tmpl *Template) updateScript(field string, idx int, newName, script string) {
+	tag := ""
+	for _, r := range script {
+		if !(unicode.IsPrint(r) || r == '\n' || r == '\r' || r == '\t') {
+			script = binaryString(script)
+			tag = "!!binary"
+			break
+		}
+	}
 	entry := fmt.Sprintf("$a.%s[%d].file", field, idx)
 	// Assign script to the "file" field and then rename it to "script".
-	tmpl.expr.WriteString(fmt.Sprintf("| (%s) = %q | (%s | key) = %q\n", entry, script, entry, newName))
+	tmpl.expr.WriteString(fmt.Sprintf("| (%s) = %q | (%s) tag = %q | (%s | key) = %q\n",
+		entry, script, entry, tag, entry, newName))
 }
 
 // embedAllScripts replaces all "provision" and "probes" file references with the actual script.

pkg/limatmpl/embed_test.go

@@ -392,6 +392,27 @@ provision:
 		"base: https://example.com/lima-linux-riscv64.img",
 		"{arch: riscv64, images: [{location: https://example.com/lima-linux-riscv64.img, arch: riscv64}]}",
 	},
+	{
+		"Binary files are base64 encoded",
+		`#
+provision:
+- mode: data
+  file: base1.sh # This comment will move to the "content" key
+  path: /tmp/data
+`,
+		// base1.sh is binary because it contains an audible bell character '\a'
+		"# base0.yaml is ignored\n---\n#!\a123456789012345678901234567890123456789012345678901234567890",
+		`
+provision:
+- mode: data
+  content: !!binary |  # This comment will move to the "content" key
+    IyEHMTIzNDU2Nzg5MDEyMzQ1Njc4OTAxMjM0NTY3ODkwMTIzNDU2Nzg5MDEyMzQ1Njc4OTAxMjM0
+    NTY3ODkw
+  path: /tmp/data
+
+# base0.yaml is ignored
+`,
+	},
 }
 
 func TestEmbed(t *testing.T) {