Refine file exclusion

site/drop_test.go

@@ -21,13 +21,13 @@ func readTestSiteDrop(t *testing.T) map[string]interface{} {
 
 func TestSite_ToLiquid(t *testing.T) {
 	drop := readTestSiteDrop(t)
-	docs, isTime := drop["documents"].([]pages.Document)
-	require.True(t, isTime, fmt.Sprintf("documents has type %T", drop["documents"]))
+	docs, ok := drop["documents"].([]pages.Document)
+	require.True(t, ok, fmt.Sprintf("documents has type %T", drop["documents"]))
 	require.Len(t, docs, 4)
 }
+
 func TestSite_ToLiquid_time(t *testing.T) {
 	drop := readTestSiteDrop(t)
-
 	_, ok := drop["time"].(time.Time)
 	require.True(t, ok)
 

site/read.go

@@ -3,6 +3,7 @@ package site
 import (
 	"os"
 	"path/filepath"
+	"strings"
 
 	"github.com/osteele/gojekyll/collection"
 	"github.com/osteele/gojekyll/config"
@@ -56,7 +57,11 @@ func (s *Site) readFiles(dir, base string) error {
 		switch {
 		case info.IsDir() && s.Exclude(rel):
 			return filepath.SkipDir
-		case info.IsDir(), s.Exclude(rel):
+		case info.IsDir():
+			return nil
+		case s.Exclude(rel):
+			return nil
+		case strings.HasPrefix(rel, "_"):
 			return nil
 		}
 		defaultFrontmatter := s.config.GetFrontMatterDefaults("", rel)

site/rebuild.go

@@ -3,7 +3,6 @@ package site
 import (
 	"fmt"
 	"os"
-	"path/filepath"
 	"strings"
 	"time"
 
@@ -130,7 +129,7 @@ loop:
 		switch {
 		case s.config.IsConfigPath(path):
 			// break
-		case !s.fileAffectsBuild(path):
+		case s.Exclude(path):
 			continue loop
 		case seen[path]:
 			continue loop
@@ -141,25 +140,6 @@ loop:
 	return result
 }
 
-// Returns true if the file or a parent directory is excluded.
-// Cf. Site.Exclude.
-func (s *Site) fileAffectsBuild(rel string) bool {
-	for rel != "" {
-		switch {
-		case rel == ".":
-			return true
-		case utils.MatchList(s.config.Include, rel):
-			return true
-		case utils.MatchList(s.config.Exclude, rel):
-			return false
-		case strings.HasPrefix(rel, "."):
-			return false
-		}
-		rel = filepath.Dir(rel)
-	}
-	return true
-}
-
 // returns true if changes to the site-relative paths invalidate doc
 func (s *Site) invalidatesDoc(paths map[string]bool, d pages.Document) bool {
 	rel := utils.MustRel(s.SourceDir(), d.SourcePath())

site/site.go

@@ -3,7 +3,7 @@ package site
 import (
 	"fmt"
 	"path/filepath"
-	"strings"
+	"regexp"
 	"sync"
 
 	"github.com/osteele/gojekyll/collection"
@@ -205,23 +205,26 @@ func (s *Site) URLPage(urlpath string) (p pages.Document, found bool) {
 	return
 }
 
+var excludeFileRE = regexp.MustCompile(`^[#~]|^\..|~$`)
+
 // Exclude returns a boolean indicating that the site configuration excludes a file or directory.
-// This function just looks at the base name: it doesn't recognize that a file in an
-// excluded directory is therefore excluded. It also excludes files in _includes, _sass, and _{container}, etc.
-// which aren't copied to the destination but do affect the build.
-// Cf. Site.fileAffectsBuild.
-func (s *Site) Exclude(path string) bool {
-	base := filepath.Base(path)
-	switch {
-	case path == ".":
-		return false
-	case utils.MatchList(s.config.Include, base):
-		return false
-	case utils.MatchList(s.config.Exclude, base):
-		return true
-	case strings.HasPrefix(base, "."), strings.HasPrefix(base, "_"):
-		return true
-	default:
-		return false
+// It does not exclude top-level _underscore files and directories.
+func (s *Site) Exclude(siteRel string) bool {
+	for siteRel != "." {
+		dir, base := filepath.Dir(siteRel), filepath.Base(siteRel)
+		switch {
+		case utils.MatchList(s.config.Include, siteRel):
+			return false
+		case utils.MatchList(s.config.Exclude, siteRel):
+			return true
+		case dir != "." && base[0] == '_':
+			return true
+		default:
+			if excludeFileRE.MatchString(base) {
+				return true
+			}
+		}
+		siteRel = dir
 	}
+	return false
 }

site/site_test.go

@@ -7,10 +7,40 @@ import (
 	"github.com/stretchr/testify/require"
 )
 
-func TestIsMarkdown(t *testing.T) {
+func TestKeepFile(t *testing.T) {
 	s := New(config.Flags{})
 	require.Equal(t, "", s.PathPrefix())
 	require.False(t, s.KeepFile("random"))
 	require.True(t, s.KeepFile(".git"))
 	require.True(t, s.KeepFile(".svn"))
 }
+
+func TestExclude(t *testing.T) {
+	s := New(config.Flags{})
+	s.config.Exclude = append(s.config.Exclude, "exclude/")
+	s.config.Include = append(s.config.Include, ".include/")
+	require.False(t, s.Exclude("."))
+	require.True(t, s.Exclude(".git"))
+	require.True(t, s.Exclude(".dir"))
+	require.True(t, s.Exclude(".dir/file"))
+	require.False(t, s.Exclude(".htaccess"))
+	require.False(t, s.Exclude("dir"))
+	require.False(t, s.Exclude("dir/file"))
+	require.True(t, s.Exclude("dir/.file"))
+	require.True(t, s.Exclude("dir/#file"))
+	require.True(t, s.Exclude("dir/~file"))
+	require.True(t, s.Exclude("dir/file~"))
+	require.True(t, s.Exclude("dir/subdir/.file"))
+	require.False(t, s.Exclude(".include/file"))
+	require.True(t, s.Exclude("exclude/file"))
+	require.False(t, s.Exclude("_posts"))
+	require.False(t, s.Exclude("_posts/file"))
+	require.True(t, s.Exclude("_posts/_file"))
+	require.True(t, s.Exclude("_posts/_dir/file"))
+
+	// The following aren't documented but are evident
+	// TODO submit a doc PR to Jekyll
+	require.True(t, s.Exclude("#file"))
+	require.True(t, s.Exclude("~file"))
+	require.True(t, s.Exclude("file~"))
+}

utils/filepath.go

@@ -15,8 +15,10 @@ func FilenameDate(s string) (time.Time, bool) {
 	return t, err == nil
 }
 
-// MatchList reports whether any glob pattern matches the path.
+// MatchList implement Jekyll include: and exclude: configurations.
+// It reports whether any glob pattern matches the path.
 // It panics with ErrBadPattern if any pattern is malformed.
+// To match Jekyll, a string "dir/" matches that begins with this prefix.
 func MatchList(patterns []string, name string) bool {
 	for _, p := range patterns {
 		match, err := filepath.Match(p, name)
@@ -26,6 +28,9 @@ func MatchList(patterns []string, name string) bool {
 		if match {
 			return true
 		}
+		if strings.HasSuffix(p, "/") && strings.HasPrefix(name, p) {
+			return true
+		}
 	}
 	return false
 }