exclude patterns, not directories

this makes the exclude synax much more flexible

.stylize.yml

@@ -2,7 +2,7 @@
 formatters:
     .py: yapf
     .go: gofmt
-exclude_dirs:
+exclude:
   - testdata
 clang_style: google
 yapf_style: pep8

config.go

@@ -8,7 +8,7 @@ import (
 // This type defines the structure of the yml config file for stylize.
 type Config struct {
 	FormattersByExt map[string]string `yaml:"formatters"`
-	ExcludeDirs     []string          `yaml:"exclude_dirs"`
+	ExcludePatterns []string          `yaml:"exclude"`
 
 	// TODO: do better
 	ClangStyle string `yaml:"clang_style"`

main.go

@@ -14,7 +14,7 @@ func main() {
 	patchFileFlag := flag.String("patch_output", "", "Path to output patch to. If '-', writes to stdout.")
 	configFileFlag := flag.String("config", ".stylize.yml", "Optional config file (defaults to .stylize.yml).")
 	dirFlag := flag.String("dir", ".", "Directory to recursively format.")
-	excludeDirFlag := flag.String("exclude_dirs", "", "Directories to exclude (comma-separated).")
+	excludeFlag := flag.String("exclude", "", "A list of exclude patterns (comma-separated).")
 	diffbaseFlag := flag.String("git_diffbase", "", "If provided, stylize only looks at files that differ from the given commit/branch.")
 	parallelismFlag := flag.Int("j", 8, "Number of files to process in parallel.")
 	flag.Parse()
@@ -31,7 +31,10 @@ func main() {
 		log.Printf("Loaded config from file %s", *configFileFlag)
 	}
 
-	rootDir := absPathOrFail(*dirFlag)
+	rootDir, err := filepath.Abs(*dirFlag)
+	if err != nil {
+		log.Fatal(err)
+	}
 
 	// set style configs from config file
 	// TODO: do better
@@ -44,20 +47,14 @@ func main() {
 		}
 	}
 
-	var excludeDirs []string
+	var excludePatterns []string
 	// exclude dirs from config
 	if cfg != nil {
-		excludeDirs = append(excludeDirs, cfg.ExcludeDirs...)
+		excludePatterns = append(excludePatterns, cfg.ExcludePatterns...)
 	}
 	// exclude dirs from flag
-	if len(*excludeDirFlag) > 0 {
-		excludeDirs = append(excludeDirs, strings.Split(*excludeDirFlag, ",")...)
-	}
-	// make exclude dirs absolute - if they're not already, they're assumed to be relative to the root directory
-	for i, edir := range excludeDirs {
-		if !filepath.IsAbs(edir) {
-			excludeDirs[i] = filepath.Join(rootDir, edir)
-		}
+	if len(*excludeFlag) > 0 {
+		excludePatterns = append(excludePatterns, strings.Split(*excludeFlag, ",")...)
 	}
 
 	// setup formatters
@@ -86,9 +83,9 @@ func main() {
 			defer patchFileOut.Close()
 			log.Printf("Writing patch to file %s", *patchFileFlag)
 		}
-		stats = StylizeMain(formatters, rootDir, excludeDirs, *diffbaseFlag, patchOut, *inPlaceFlag, *parallelismFlag)
+		stats = StylizeMain(formatters, rootDir, excludePatterns, *diffbaseFlag, patchOut, *inPlaceFlag, *parallelismFlag)
 	} else {
-		stats = StylizeMain(formatters, rootDir, excludeDirs, *diffbaseFlag, nil, *inPlaceFlag, *parallelismFlag)
+		stats = StylizeMain(formatters, rootDir, excludePatterns, *diffbaseFlag, nil, *inPlaceFlag, *parallelismFlag)
 	}
 
 	if stats.Error != 0 {

stylize.go

@@ -22,19 +22,10 @@ type FormattingResult struct {
 	Error        error
 }
 
-func fileIsExcluded(file string, excludeDirs []string) bool {
-	for _, eDir := range excludeDirs {
-		if filepath.HasPrefix(file, eDir) {
-			return true
-		}
-	}
-	return false
-}
-
 // Walks the given directory and sends all non-excluded files to the returned channel.
 // @param rootDir absolute path to root directory
 // @return file paths relative to rootDir
-func IterateAllFiles(rootDir string, excludeDirs []string) <-chan string {
+func IterateAllFiles(rootDir string, exclude []string) <-chan string {
 	files := make(chan string)
 
 	go func() {
@@ -44,16 +35,21 @@ func IterateAllFiles(rootDir string, excludeDirs []string) <-chan string {
 				return nil
 			}
 
-			excludeDirs = append(excludeDirs, absPathOrFail(".git"), absPathOrFail(".hg"))
-			if fi.IsDir() && fileIsExcluded(path, excludeDirs) {
+			relPath, _ := filepath.Rel(rootDir, path)
+
+			exclude = append(exclude, ".git", ".hg")
+			if fi.IsDir() && fileIsExcluded(relPath, exclude) {
 				return filepath.SkipDir
 			}
 
+			if fileIsExcluded(relPath, exclude) {
+				return nil
+			}
+
 			if fi.IsDir() {
 				return nil
 			}
 
-			relPath, _ := filepath.Rel(rootDir, path)
 			files <- relPath
 
 			return nil
@@ -66,7 +62,7 @@ func IterateAllFiles(rootDir string, excludeDirs []string) <-chan string {
 // Finds files that have been modified since the common ancestor of HEAD and
 // diffbase and sends them onto the returned channel.
 // @return file paths relative to rootDir
-func IterateGitChangedFiles(rootDir string, excludeDirs []string, diffbase string) (<-chan string, error) {
+func IterateGitChangedFiles(rootDir string, exclude []string, diffbase string) (<-chan string, error) {
 	changedFiles, err := gitChangedFiles(rootDir, diffbase)
 	if err != nil {
 		return nil, err
@@ -90,8 +86,14 @@ func IterateGitChangedFiles(rootDir string, excludeDirs []string, diffbase strin
 
 		for _, file := range changedFiles {
 			absPath := filepath.Join(gitRoot, file)
-			if fileIsExcluded(absPath, excludeDirs) {
-				// log.Printf("Excluding file: %s", absPath)
+
+			// get file path relative to root directory
+			relPath, err := filepath.Rel(rootDir, absPath)
+			if err != nil {
+				log.Fatal(err)
+			}
+
+			if fileIsExcluded(relPath, exclude) {
 				continue
 			}
 
@@ -102,12 +104,6 @@ func IterateGitChangedFiles(rootDir string, excludeDirs []string, diffbase strin
 				continue
 			}
 
-			// get file path relative to root directory
-			relPath, err := filepath.Rel(rootDir, absPath)
-			if err != nil {
-				log.Fatal(err)
-			}
-
 			files <- relPath
 		}
 	}()
@@ -263,30 +259,30 @@ func LogActionsAndCollectStats(results <-chan FormattingResult, inPlace bool) Ru
 //     diffbase. Otherwise looks at all files.
 // @param formatters A map of file extension -> formatter
 // @return (changeCount, totalCount, errCount)
-func StylizeMain(formatters map[string]Formatter, rootDir string, excludeDirs []string, gitDiffbase string, patchOut io.Writer, inPlace bool, parallelism int) RunStats {
+func StylizeMain(formatters map[string]Formatter, rootDir string, exclude []string, gitDiffbase string, patchOut io.Writer, inPlace bool, parallelism int) RunStats {
 	if inPlace && patchOut != nil {
 		log.Fatal("Patch output writer should only be provided in non-inplace runs")
 	}
 	if !filepath.IsAbs(rootDir) {
 		log.Fatalf("root directory should be an absolute path: '%s'", rootDir)
 	}
 
-	for _, excl := range excludeDirs {
-		if !filepath.IsAbs(excl) {
-			log.Fatal("exclude directories should be absolute")
+	for _, excl := range exclude {
+		if filepath.IsAbs(excl) {
+			log.Fatal("exclude directories should not be absolute")
 		}
 	}
 
 	// setup file source
 	var err error
 	var fileChan <-chan string
 	if len(gitDiffbase) > 0 {
-		fileChan, err = IterateGitChangedFiles(rootDir, excludeDirs, gitDiffbase)
+		fileChan, err = IterateGitChangedFiles(rootDir, exclude, gitDiffbase)
 		if err != nil {
 			log.Fatal(err)
 		}
 	} else {
-		fileChan = IterateAllFiles(rootDir, excludeDirs)
+		fileChan = IterateAllFiles(rootDir, exclude)
 	}
 
 	// run formatter on all files

stylize_test.go

@@ -24,6 +24,27 @@ func TestMain(m *testing.M) {
 	os.Exit(m.Run())
 }
 
+func expectMatch(t *testing.T, match bool, pattern, file string) {
+	m := filePatternMatch(pattern, file)
+	if m != match {
+		t.Logf("'%s' '%s'", pattern, file)
+		if match {
+			t.Error("Expected match, but didn't get one")
+		} else {
+			t.Error("It matched, but wasnt supposed to")
+		}
+	}
+}
+
+func TestMatch(t *testing.T) {
+	expectMatch(t, true, "exclude/", "exclude/file.cpp")
+	expectMatch(t, true, "exclude", "exclude/file.cpp")
+	expectMatch(t, false, "exclud", "exclude/file.cpp")
+	expectMatch(t, false, "bbb", "bbb.cpp")
+	expectMatch(t, true, "*", "bad.cpp")
+	expectMatch(t, true, "*", "files/bad.cpp")
+}
+
 func TestCreatePatch(t *testing.T) {
 	goldenFile := "testdata/patch.golden"
 
@@ -42,7 +63,8 @@ func TestCreatePatch(t *testing.T) {
 		patchOut = &patchBuffer
 	}
 
-	StylizeMain(LoadDefaultFormatters(), absPathOrFail("testdata"), []string{absPathOrFail("testdata/exclude")}, "", patchOut, false, PARALLELISM)
+	absDirPath, _ := filepath.Abs("testdata")
+	StylizeMain(LoadDefaultFormatters(), absDirPath, []string{"exclude"}, "", patchOut, false, PARALLELISM)
 
 	if !*generateGoldens {
 		assertGoldenMatch(t, goldenFile, patchBuffer.String())
@@ -60,7 +82,7 @@ func TestInPlace(t *testing.T) {
 	tmp := mktmp(t)
 	dir := copyTestData(t, tmp)
 
-	exclude := []string{path.Join(dir, "exclude")}
+	exclude := []string{"exclude"}
 	t.Log("exclude: " + strings.Join(exclude, ","))
 
 	// run in-place formatting
@@ -83,30 +105,26 @@ func TestInPlaceWithConfig(t *testing.T) {
 	dir := copyTestData(t, tmp)
 
 	cfgPath := path.Join(dir, ".stylize.yml")
-	err := ioutil.WriteFile(cfgPath, []byte("---\nformatters:\n  .py: yapf\nexclude_dirs:\n  - exclude"), 0644)
+	err := ioutil.WriteFile(cfgPath, []byte("---\nformatters:\n  .py: yapf\nexclude:\n  - exclude"), 0644)
 	tCheckErr(t, err)
 	t.Logf("Wrote config file: %s", cfgPath)
 
 	cfg, err := LoadConfig(cfgPath)
 	tCheckErr(t, err)
 	t.Log("Read config file")
-
-	for i, edir := range cfg.ExcludeDirs {
-		cfg.ExcludeDirs[i] = filepath.Join(dir, edir)
-	}
-	t.Log("exclude: " + strings.Join(cfg.ExcludeDirs, ","))
+	t.Log("exclude: " + strings.Join(cfg.ExcludePatterns, ","))
 
 	formatters := LoadFormattersFromMapping(cfg.FormattersByExt)
 
 	// run in-place formatting
-	stats := StylizeMain(formatters, dir, cfg.ExcludeDirs, "", nil, true, PARALLELISM)
+	stats := StylizeMain(formatters, dir, cfg.ExcludePatterns, "", nil, true, PARALLELISM)
 	t.Logf("Stylize results: %d, %d, %d", stats.Change, stats.Total, stats.Error)
 
 	if stats.Change != 1 {
 		t.Fatal("One file should have changed")
 	}
 
-	stats = StylizeMain(formatters, dir, cfg.ExcludeDirs, "", nil, true, PARALLELISM)
+	stats = StylizeMain(formatters, dir, cfg.ExcludePatterns, "", nil, true, PARALLELISM)
 	t.Logf("Stylize results: %d, %d, %d", stats.Change, stats.Total, stats.Error)
 
 	if stats.Change != 0 {
@@ -135,7 +153,7 @@ func TestGitDiffbase(t *testing.T) {
 	runCmd(t, dir, "git", "add", ".")
 	runCmd(t, dir, "git", "commit", "-m", "added files")
 
-	exclude := []string{path.Join(dir, "exclude")}
+	exclude := []string{"exclude"}
 	t.Log("exclude: " + strings.Join(exclude, ","))
 
 	// run stylize with diffbase provided

util.go

@@ -3,13 +3,12 @@ package main
 import (
 	"bytes"
 	"fmt"
+	"github.com/danwakefield/fnmatch"
 	"github.com/pkg/errors"
 	"golang.org/x/crypto/ssh/terminal"
 	"io"
-	"log"
 	"os"
 	"os/exec"
-	"path/filepath"
 	"strings"
 	"syscall"
 	"unsafe"
@@ -30,14 +29,6 @@ func runIOCommand(args []string, in io.Reader, out io.Writer) error {
 	return nil
 }
 
-func absPathOrFail(path string) string {
-	absPath, err := filepath.Abs(path)
-	if err != nil {
-		log.Fatal(err)
-	}
-	return absPath
-}
-
 type winsize struct {
 	Row    uint16
 	Col    uint16
@@ -90,3 +81,35 @@ func gitChangedFiles(rootDir, diffbase string) ([]string, error) {
 
 	return changedFiles, nil
 }
+
+// TODO: this implementation has a lot of flaws
+// It would be nice to do something similar to gitignore
+func filePatternMatch(pattern, file string) bool {
+	if fnmatch.Match(pattern, file, fnmatch.FNM_PATHNAME|fnmatch.FNM_LEADING_DIR) {
+		return true
+	}
+
+	// TODO: this is a hack
+	hasGlobChars := strings.ContainsAny(pattern, "*?")
+	if hasGlobChars {
+		return false
+	}
+
+	// If pattern ends in '/', ignore all files in that directory recursively
+	if strings.HasSuffix(pattern, "/") {
+		if strings.HasPrefix(file, pattern) {
+			return true
+		}
+	}
+
+	return false
+}
+
+func fileIsExcluded(file string, exclude []string) bool {
+	for _, e := range exclude {
+		if filePatternMatch(e, file) {
+			return true
+		}
+	}
+	return false
+}