diff --git a/ignorefile/ignorefile.go b/ignorefile/ignorefile.go
new file mode 100644
index 0000000..94ea5a0
--- /dev/null
+++ b/ignorefile/ignorefile.go
@@ -0,0 +1,73 @@
+package ignorefile
+
+import (
+	"bufio"
+	"bytes"
+	"io"
+	"path/filepath"
+	"strings"
+)
+
+// ReadAll reads an ignore file from a reader and returns the list of file
+// patterns to ignore, applying the following rules:
+//
+//   - An UTF8 BOM header (if present) is stripped.
+//   - Lines starting with "#" are considered comments and are skipped.
+//
+// For remaining lines:
+//
+//   - Leading and trailing whitespace is removed from each ignore pattern.
+//   - It uses [filepath.Clean] to get the shortest/cleanest path for
+//     ignore patterns.
+//   - Leading forward-slashes ("/") are removed from ignore patterns,
+//     so "/some/path" and "some/path" are considered equivalent.
+func ReadAll(reader io.Reader) ([]string, error) {
+	if reader == nil {
+		return nil, nil
+	}
+
+	var excludes []string
+	currentLine := 0
+	utf8bom := []byte{0xEF, 0xBB, 0xBF}
+
+	scanner := bufio.NewScanner(reader)
+	for scanner.Scan() {
+		scannedBytes := scanner.Bytes()
+		// We trim UTF8 BOM
+		if currentLine == 0 {
+			scannedBytes = bytes.TrimPrefix(scannedBytes, utf8bom)
+		}
+		pattern := string(scannedBytes)
+		currentLine++
+		// Lines starting with # (comments) are ignored before processing
+		if strings.HasPrefix(pattern, "#") {
+			continue
+		}
+		pattern = strings.TrimSpace(pattern)
+		if pattern == "" {
+			continue
+		}
+		// normalize absolute paths to paths relative to the context
+		// (taking care of '!' prefix)
+		invert := pattern[0] == '!'
+		if invert {
+			pattern = strings.TrimSpace(pattern[1:])
+		}
+		if len(pattern) > 0 {
+			pattern = filepath.Clean(pattern)
+			pattern = filepath.ToSlash(pattern)
+			if len(pattern) > 1 && pattern[0] == '/' {
+				pattern = pattern[1:]
+			}
+		}
+		if invert {
+			pattern = "!" + pattern
+		}
+
+		excludes = append(excludes, pattern)
+	}
+	if err := scanner.Err(); err != nil {
+		return nil, err
+	}
+	return excludes, nil
+}
diff --git a/ignorefile/ignorefile_test.go b/ignorefile/ignorefile_test.go
new file mode 100644
index 0000000..c35291d
--- /dev/null
+++ b/ignorefile/ignorefile_test.go
@@ -0,0 +1,54 @@
+package ignorefile
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestReadAll(t *testing.T) {
+	actual, err := ReadAll(nil)
+	if err != nil {
+		t.Errorf("Expected no error, got %v", err)
+	}
+	if entries := len(actual); entries != 0 {
+		t.Fatalf("Expected to have zero entries, got %d", entries)
+	}
+
+	const content = `test1
+/test2
+/a/file/here
+
+lastfile
+# this is a comment
+! /inverted/abs/path
+!
+! `
+
+	expected := []string{
+		"test1",
+		"test2",       // according to https://docs.docker.com/engine/reference/builder/#dockerignore-file, /foo/bar should be treated as foo/bar
+		"a/file/here", // according to https://docs.docker.com/engine/reference/builder/#dockerignore-file, /foo/bar should be treated as foo/bar
+		"lastfile",
+		"!inverted/abs/path",
+		"!",
+		"!",
+	}
+
+	actual, err = ReadAll(strings.NewReader(content))
+	if err != nil {
+		t.Error(err)
+	}
+
+	if len(actual) != len(expected) {
+		t.Errorf("Expected %d entries, got %v", len(expected), len(actual))
+	}
+	for i, expectedLine := range expected {
+		if i >= len(actual) {
+			t.Errorf(`missing line %d: expected: "%s", got none`, i+1, expectedLine)
+			continue
+		}
+		if actual[i] != expectedLine {
+			t.Errorf(`line %d: expected: "%s", got: "%s"`, i+1, expectedLine, actual[i])
+		}
+	}
+}