fix formatting files directly, add tests, improve code generation det…

…ection

config.go

@@ -14,6 +14,25 @@ const (
 	defaultConfigFile = ".gimps.yaml"
 )
 
+var (
+	defaultExcludes = []string{
+		// to not break 3rd party code
+		"vendor/**",
+
+		// to not muck with generated files
+		"**/zz_generated.**",
+		"**/zz_generated_**",
+		"**/generated.pb.go",
+		"**/generated.proto",
+		"**/*_generated.go",
+
+		// for performance
+		".git/**",
+		"_build/**",
+		"node_modules/**",
+	}
+)
+
 type Config struct {
 	gimps.Config         `yaml:",inline"`
 	Exclude              []string `yaml:"exclude"`
@@ -42,24 +61,7 @@ func loadConfiguration(filename string, moduleRoot string) (*Config, error) {
 	}
 
 	if c.Exclude == nil || len(c.Exclude) == 0 {
-		// vendor is because we never want to modify vendor, the
-		// others are just to save time while scanning bigger
-		// repositories that maybe also contain non-Go stuff
-		c.Exclude = []string{
-			// to not break 3rd party code
-			"vendor/**",
-
-			// to not muck with generated files
-			"**/zz_generated.**",
-			"**/zz_generated_**",
-			"**/generated.pb.go",
-			"**/*_generated.go",
-
-			// for performance
-			".git/**",
-			"_build/**",
-			"node_modules/**",
-		}
+		c.Exclude = defaultExcludes
 	}
 
 	if c.DetectGeneratedFiles == nil {

fs.go

@@ -0,0 +1,132 @@
+package main
+
+import (
+	"errors"
+	"go/parser"
+	"go/token"
+	"io/fs"
+	"io/ioutil"
+	"os"
+	"path/filepath"
+	"regexp"
+	"strings"
+
+	doublestar "github.com/bmatcuk/doublestar/v4"
+)
+
+// listFiles takes a filename or directory as its start argument and returns
+// a list of absolute file paths. If a filename is given, the list contains
+// exactly one element, otherwise the directory is scanned recursively.
+// Note that if start is a file, the skip rules are not evaluated. This allows
+// users to force-format an otherwise skipped file.
+func listFiles(start string, moduleRoot string, skips []string) ([]string, error) {
+	result := []string{}
+
+	info, err := os.Stat(start)
+	if err != nil {
+		return nil, err
+	}
+
+	if !info.IsDir() {
+		return []string{start}, nil
+	}
+
+	err = filepath.WalkDir(start, func(path string, d fs.DirEntry, err error) error {
+		relPath, err := filepath.Rel(moduleRoot, path)
+		if err != nil {
+			return err
+		}
+
+		if isSkipped(relPath, skips) {
+			if d.IsDir() {
+				return filepath.SkipDir
+			} else {
+				return nil
+			}
+		}
+
+		if !d.IsDir() && strings.HasSuffix(path, ".go") {
+			result = append(result, path)
+		}
+
+		return nil
+	})
+	if err != nil {
+		return nil, err
+	}
+
+	return result, nil
+}
+
+func isSkipped(relPath string, skips []string) bool {
+	for _, skip := range skips {
+		if match, _ := doublestar.Match(skip, relPath); match {
+			return true
+		}
+	}
+
+	return false
+}
+
+func goModRootPath(path string) (string, error) {
+	// turn path into directory, if it's a file
+	if info, err := os.Stat(path); err == nil && !info.IsDir() {
+		path = filepath.Dir(path)
+	}
+
+	for {
+		if fi, err := os.Stat(filepath.Join(path, "go.mod")); err == nil && !fi.IsDir() {
+			return path, nil
+		}
+
+		d := filepath.Dir(path)
+		if d == path {
+			break
+		}
+
+		path = d
+	}
+
+	return "", errors.New("no go.mod found")
+}
+
+var (
+	// detect generated files by presence if this string in the first non-stripped line
+	generatedRe = regexp.MustCompile("(been generated|generated by|do not edit)")
+)
+
+func isGeneratedFile(filename string) (bool, error) {
+	content, err := ioutil.ReadFile(filename)
+	if err != nil {
+		return false, err
+	}
+
+	return isGeneratedCode(content)
+}
+
+func isGeneratedCode(sourceCode []byte) (bool, error) {
+	fset := token.NewFileSet()
+
+	file, err := parser.ParseFile(fset, "", sourceCode, parser.ParseComments)
+	if err != nil {
+		return false, err
+	}
+
+	// go through all comments until we reach the package declaration
+outer:
+	for _, commentGroup := range file.Comments {
+		for _, comment := range commentGroup.List {
+			// found the package declaration
+			if comment.Slash > file.Package {
+				break outer
+			}
+
+			text := []byte(strings.ToLower(comment.Text))
+			if generatedRe.Match(text) {
+				return true, nil
+			}
+		}
+	}
+
+	return false, nil
+}

fs_test.go

@@ -0,0 +1,147 @@
+package main
+
+import (
+	"fmt"
+	"strings"
+	"testing"
+)
+
+func TestDefaultExcludeFilterAgainstFilenames(t *testing.T) {
+	testcases := []struct {
+		filename string
+		expected bool
+	}{
+		{
+			filename: "main.go",
+			expected: false,
+		},
+		{
+			filename: "zz_generated.deepcopy.go",
+			expected: true,
+		},
+		{
+			filename: "zz_generated.go",
+			expected: true,
+		},
+		{
+			filename: "generated.pb.go",
+			expected: true,
+		},
+	}
+
+	for _, tt := range testcases {
+		t.Run(tt.filename, func(t *testing.T) {
+			skipped := isSkipped(tt.filename, defaultExcludes)
+			if skipped != tt.expected {
+				t.Errorf("Expected %v but got %v", tt.expected, skipped)
+			}
+
+			tt.filename = "pkg/" + tt.filename
+
+			skipped = isSkipped(tt.filename, defaultExcludes)
+			if skipped != tt.expected {
+				t.Errorf("Expected %v for %q, but got %v", tt.expected, tt.filename, skipped)
+			}
+		})
+	}
+}
+
+func TestIsGeneratedCode(t *testing.T) {
+	testcases := []struct {
+		comment  string
+		expected bool
+	}{
+		{
+			comment:  "",
+			expected: false,
+		},
+		{
+			comment:  "// This file has been generated.",
+			expected: true,
+		},
+		{
+			comment:  "// Code generated by MockGen. DO NOT EDIT.",
+			expected: true,
+		},
+		{
+			comment:  "// Code generated by generate-imagename-constants.sh. DO NOT EDIT.",
+			expected: true,
+		},
+		{
+			comment:  "// This file has been generated with Velero v1.5.3. Do not edit.",
+			expected: true,
+		},
+	}
+
+	for i, tt := range testcases {
+		code := fmt.Sprintf(`
+%s
+package main
+
+func main() {
+
+}
+`, tt.comment)
+		t.Run(fmt.Sprintf("#%d vanilla", i+1), runGeneratedCodeTest(code, tt.expected))
+
+		code = fmt.Sprintf(`
+// +build foo
+
+%s
+
+package main
+
+func main() {
+
+}
+`, tt.comment)
+		t.Run(fmt.Sprintf("#%d with build constraint", i+1), runGeneratedCodeTest(code, tt.expected))
+
+		code = fmt.Sprintf(`
+// +build foo
+/*
+ I am a license header.
+*/
+
+%s
+
+package main
+
+func main() {
+
+}
+`, tt.comment)
+		t.Run(fmt.Sprintf("#%d with build constraint and license header", i+1), runGeneratedCodeTest(code, tt.expected))
+
+		code = fmt.Sprintf(`
+// +build foo
+/*
+ I am a license header.
+*/
+
+package main
+
+%s
+
+func main() {
+
+}
+`, tt.comment)
+		t.Run(fmt.Sprintf("#%d, but too late, so ignore it", i+1), runGeneratedCodeTest(code, false))
+	}
+}
+
+func runGeneratedCodeTest(code string, expected bool) func(t *testing.T) {
+	return func(t *testing.T) {
+		b := []byte(strings.TrimSpace(code))
+
+		generated, err := isGeneratedCode(b)
+		if err != nil {
+			t.Errorf("should not have errored, but got %v", err)
+		}
+
+		if generated != expected {
+			t.Errorf("Expected %v but got %v", expected, generated)
+		}
+	}
+}