Support more properties.

src/tar.toit

@@ -10,6 +10,17 @@ A tar archiver.
 Writes the given files into the writer in tar file format.
 */
 class TarWriter:
+  static TYPE-REGULAR-FILE ::= '0'
+  static TYPE-LINK         ::= '1'
+  static TYPE-SYMBOLIC-LINK ::= '2'
+  static TYPE-CHARACTER-DEVICE ::= '3'
+  static TYPE-BLOCK-DEVICE ::= '4'
+  static TYPE-DIRECTORY    ::= '5'
+  static TYPE-FIFO         ::= '6'
+
+  static TYPE-NORMAL_    ::= '0'
+  static TYPE-LONG-LINK_ ::= 'L'
+
   writer_/io.Writer
 
   /**
@@ -24,8 +35,27 @@ class TarWriter:
   This function sets all file attributes to some default values. For example, the
     modification date is set to 0 (epoch time).
   */
-  add file-name/string content/io.Data --permissions/int=((6 << 6) | (6 << 3) | 6) -> none:
-    add_ file-name content --permissions=permissions
+  add file-name/string content/io.Data
+      --permissions/int=((6 << 6) | (6 << 3) | 6)
+      --uid/int=0
+      --gid/int=0
+      --mtime/Time?=null
+      --type/int=TYPE-REGULAR-FILE
+      --user-name/string=""
+      --group-name/string=""
+      --device-major/int=0
+      --device-minor/int=0
+      -> none:
+    add_ file-name content
+        --permissions=permissions
+        --uid=uid
+        --gid=gid
+        --mtime=mtime
+        --type=type
+        --user-name=user-name
+        --group-name=group-name
+        --device-major=device-major
+        --device-minor=device-minor
 
   /**
   Closes the tar stream, but does not close the writer.
@@ -44,10 +74,30 @@ class TarWriter:
   The $type parameter must be one of the constants below: $TYPE-NORMAL_ or
     $TYPE-LONG-LINK_.
   */
-  add_ file-name/string content/io.Data --type/int=TYPE-NORMAL_ --permissions/int -> none:
+  add_ file-name/string content/io.Data
+      --permissions/int
+      --uid/int=0
+      --gid/int=0
+      --mtime/Time?=null
+      --type/int=TYPE-REGULAR-FILE
+      --user-name/string=""
+      --group-name/string=""
+      --device-major/int=0
+      --device-minor/int=0
+      -> none:
     if file-name.size > 100:
       // The file-name is encoded a separate "file".
-      add_ "././@LongLink" file-name --type=TYPE-LONG-LINK_ --permissions=permissions
+      add_ "././@LongLink" file-name
+          --permissions=permissions
+          --uid=uid
+          --gid=gid
+          --mtime=mtime
+          --type=TYPE-LONG-LINK_
+          --user-name=user-name
+          --group-name=group-name
+          --device-major=device-major
+          --device-minor=device-minor
+
       file-name = file-name.copy 0 100
 
     file-size := content.byte-size
@@ -56,15 +106,38 @@ class TarWriter:
     permissions-in-octal/string := permissions.stringify 8
     permissions-in-octal = permissions-in-octal.pad --left 7 '0'
 
+    uid-in-octal/string := uid.stringify 8
+    uid-in-octal = uid-in-octal.pad --left 7 '0'
+
+    gid-in-octal/string := gid.stringify 8
+    gid-in-octal = gid-in-octal.pad --left 7 '0'
+
+    mtime-val := mtime ? (mtime.ms-since-epoch / 1000) : 0
+    mtime-in-octal/string := mtime-val.stringify 8
+    mtime-in-octal = mtime-in-octal.pad --left 11 '0'
+
+    devmajor-in-octal/string := device-major.stringify 8
+    devmajor-in-octal = devmajor-in-octal.pad --left 7 '0'
+
+    devminor-in-octal/string := device-minor.stringify 8
+    devminor-in-octal = devminor-in-octal.pad --left 7 '0'
+
     header := ByteArray 512
     // See https://en.wikipedia.org/wiki/Tar_(computing)#File_format for the format.
     header.replace 0 file-name
     header.replace 100 permissions-in-octal
+    header.replace 108 uid-in-octal
+    header.replace 116 gid-in-octal
     header.replace 124 file-size-in-octal
+    header.replace 136 mtime-in-octal
     // The checksum is computed using spaces. Later it is replaced with the actual values.
     header.replace 148 "        "
     header[156] = type
     header.replace 257 "ustar  "
+    header.replace 265 (limit-string_ user-name 32)
+    header.replace 297 (limit-string_ group-name 32)
+    header.replace 329 devmajor-in-octal
+    header.replace 337 devminor-in-octal
 
     checksum := 0
     for i := 0; i < 512; i++:
@@ -91,5 +164,6 @@ class TarWriter:
         header[i] = '\0'
       writer_.write header 0 missing
 
-  static TYPE-NORMAL_    ::= '0'
-  static TYPE-LONG-LINK_ ::= 'L'
+  static limit-string_ str/string limit/int -> string:
+    if str.size <= limit: return str
+    return str[..limit]

tests/properties-test.toit

@@ -0,0 +1,85 @@
+// Copyright (C) 2026 Toit contributors.
+// Use of this source code is governed by a Zero-Clause BSD license that can
+// be found in the tests/TESTS_LICENSE file.
+
+import expect show *
+import tar show *
+import expect show *
+import tar show *
+import io
+
+class MemoryWriter extends io.Writer:
+  buffer_/ByteArray := ByteArray 0
+
+  try-write_ data/io.Data from/int=0 to/int=data.byte-size -> int:
+    // If data is a string, convert to byte array.
+    bytes/ByteArray := ?
+    if data is string:
+      bytes = (data as string).to-byte-array
+    else:
+      bytes = data as ByteArray
+    // Append to buffer.
+    buffer_ = buffer_ + bytes[from..to]
+    return to - from
+
+  close:
+    // Do nothing.
+
+  bytes -> ByteArray:
+    return buffer_
+
+main:
+  test-properties
+
+test-properties:
+  writer := MemoryWriter
+  tar := TarWriter writer
+
+  mtime := Time.epoch + (Duration --s=1234567890)
+
+  tar.add "test.txt" "content"
+      --permissions=420 // 0644
+      --uid=1000
+      --gid=1000
+      --mtime=mtime
+      --type=TarWriter.TYPE-REGULAR-FILE
+      --user-name="user"
+      --group-name="group"
+      --device-major=1
+      --device-minor=2
+
+  tar.close
+
+  bytes := writer.bytes
+  // Header is first 512 bytes.
+  header := bytes[0..512]
+
+  // Helper to parse octal string from header.
+  parse-octal := : |offset length|
+    str-bytes := header[offset..offset+length]
+    // Find null terminator or take full length.
+    end := str-bytes.index-of 0
+    if end == -1: end = length
+    str := str-bytes[0..end].to-string
+    int.parse str --radix=8
+
+  // Helper to parse string from header.
+  parse-string := : |offset length|
+    str-bytes := header[offset..offset+length]
+    end := str-bytes.index-of 0
+    if end == -1: end = length
+    str-bytes[0..end].to-string
+
+  // Verify fields.
+  expect-equals "test.txt" (parse-string.call 0 100)
+  expect-equals 420 (parse-octal.call 100 8)
+  expect-equals 1000 (parse-octal.call 108 8)
+  expect-equals 1000 (parse-octal.call 116 8)
+  expect-equals 7 (parse-octal.call 124 12) // "content".size = 7
+  expect-equals 1234567890 (parse-octal.call 136 12)
+  expect-equals TarWriter.TYPE-REGULAR-FILE header[156]
+  expect-equals "ustar  " (parse-string.call 257 8)
+  expect-equals "user" (parse-string.call 265 32)
+  expect-equals "group" (parse-string.call 297 32)
+  expect-equals 1 (parse-octal.call 329 8)
+  expect-equals 2 (parse-octal.call 337 8)

tests/properties-verification-test.toit

@@ -0,0 +1,57 @@
+import expect show *
+import tar show *
+import .utils
+
+main:
+  test-properties-verification
+
+test-properties-verification:
+  // We use a generator that writes a tar file to the provided writer.
+  entries := list-with-tar-bin: |writer|
+    tar := TarWriter writer
+
+    // File 1: RW-R--R-- (644).
+    tar.add "test-props.txt" "content"
+        --permissions=0b110_100_100  // 644.
+        --uid=1234
+        --gid=5678
+        --user-name="testuser"
+        --group-name="testgroup"
+        --mtime=(Time.epoch + (Duration --s=1000000000))  // 2001-09-09.
+
+    // File 2: OWXRWXRWX (777) but with some mask usually, let's try 755.
+    tar.add "executable.sh" "echo hello"
+        --permissions=0b111_101_101  // 755.
+        --uid=1000
+        --gid=1000
+
+    tar.close
+
+  expect-equals 2 entries.size
+
+  // Verify File 1
+  entry1 := entries[0]
+  expect-equals "test-props.txt" entry1.name
+  expect-equals 7 entry1.size
+  expect-equals 0b110_100_100 entry1.permissions
+  // Note: System tar might display IDs if user/group names don't map to existing users,
+  // or it might display the names from the header.
+  // The 'utils.toit' listing uses 'tar -tvf' which usually prefers names if available in header (ustar).
+  // Let's check what we get. If it fails we might need to adjust expectation to allow IDs.
+  // But since we wrote "testuser" in the header, 'tar' should show it.
+  expect-equals "testuser" entry1.owner
+  expect-equals "testgroup" entry1.group
+  // 2001-09-09.
+  // Linux: 2001-09-09 03:46 (timezone dependent?).
+  // Mac: Sep 9 2001 (or similar).
+  // Just check it contains 2001 or Sep.
+  print "Entry 1 mtime: $entry1.mtime"
+  expect (entry1.mtime.contains "2001")
+
+  // Verify File 2.
+  entry2 := entries[1]
+  expect-equals "executable.sh" entry2.name
+  expect-equals 10 entry2.size
+  expect-equals 0b111_101_101 entry2.permissions
+  expect-equals "1000" entry2.owner
+  expect-equals "1000" entry2.group

tests/utils.toit

@@ -65,11 +65,11 @@ Returns the concatenated contents of all extracted files.
 */
 extract [generator]:
   return run-tar
-      "x"   // extract
-      "-PO" // P for absolute paths, to stdout
+      "x"   // extract.
+      "-PO" // P for absolute paths, to stdout.
       generator
 
-split-fields line/string -> List/*<string>*/:
+split-fields_ line/string -> List/*<string>*/:
   result := []
   start-pos := 0
   last-was-space := true
@@ -85,22 +85,76 @@ split-fields line/string -> List/*<string>*/:
 class TarEntry:
   name/string
   size/int
-  permissions/string
+  permissions/int
+  owner/string
+  group/string
+  mtime/string
 
-  constructor --.name --.size --.permissions:
+  constructor --.name --.size --.permissions --.owner --.group --.mtime:
+
+parse-permissions_ str/string -> int:
+  res := 0
+  // Expected format: -rwxrwxrwx (10 chars).
+  if str.size < 10: return 0
+
+  // User.
+  if str[1] == 'r': res |= 0b100_000_000
+  if str[2] == 'w': res |= 0b010_000_000
+  if str[3] == 'x': res |= 0b001_000_000
+
+  // Group.
+  if str[4] == 'r': res |= 0b000_100_000
+  if str[5] == 'w': res |= 0b000_010_000
+  if str[6] == 'x': res |= 0b000_001_000
+
+  // Other.
+  if str[7] == 'r': res |= 0b000_000_100
+  if str[8] == 'w': res |= 0b000_000_010
+  if str[9] == 'x': res |= 0b000_000_001
+
+  return res
 
 list-with-tar-bin [generator] -> List/*<TarEntry>*/:
   listing := inspect-with-tar-bin generator
   lines := (listing.trim --right "\n").split "\n"
   return lines.map: |line|
     // A line looks something like:
-    // Linux: -rw-rw-r-- 0/0               5 1970-01-01 01:00 /foo
-    // Mac:   -rw-rw-r--  0 0      0           5 Jan  1  1970 /foo
-    permissions-index := 0
-    name-index := system.platform == system.PLATFORM-MACOS ? 8 : 5
-    size-index := system.platform == system.PLATFORM-MACOS ? 4 : 2
-    components := split-fields line
+    // Linux: -rw-rw-r-- 1000/1000          5 1970-01-01 01:00 /foo
+    // Mac:   -rw-rw-r--  0 1000 1000      5 Jan  1  1970 /foo
+    // Mac (recent): -rw-rw-r--  0 1000 1000      5 Jan  1 01:00 /foo
+    components := split-fields_ line
+
+    name-index := 0
+    size-index := 0
+    mtime-string := ""
+    user := ""
+    group := ""
+
+    permissions := parse-permissions_ components[0]
+
+    if system.platform == system.PLATFORM-MACOS:
+      name-index = 8
+      size-index = 4
+      user = components[2]
+      group = components[3]
+      // Date is at 5, 6, 7 (e.g., "Jan", "1", "1970" or "Jan", "1", "01:00").
+      mtime-string = "$components[5] $components[6] $components[7]"
+    else:
+      name-index = 5
+      size-index = 2
+      parts := components[1].split "/"
+      user = parts[0]
+      group = parts[1]
+      // Date is at 3, 4 (e.g., "1970-01-01", "01:00")
+      mtime-string = "$components[3] $components[4]"
+
     file-name := components[name-index]
     size := int.parse components[size-index]
-    permissions := components[permissions-index]
-    TarEntry --name=file-name --size=size --permissions=permissions
+
+    TarEntry
+        --name=file-name
+        --size=size
+        --permissions=permissions
+        --owner=user
+        --group=group
+        --mtime=mtime-string