From 0a8ffe0c63c7f214ac93cd740992b23f9187c181 Mon Sep 17 00:00:00 2001
From: Brian McGee <brian@bmcgee.ie>
Date: Tue, 9 Jul 2024 19:21:09 +0100
Subject: [PATCH] feat: move filtering of directories and symlinks into walker
 implementations

Signed-off-by: Brian McGee <brian@bmcgee.ie>
---
 cache/cache.go          |  5 -----
 cli/format.go           |  9 +++------
 walk/filesystem.go      |  6 ++++++
 walk/filesystem_test.go | 19 -------------------
 walk/git.go             |  6 +++++-
 5 files changed, 14 insertions(+), 31 deletions(-)

diff --git a/cache/cache.go b/cache/cache.go
index a6c743f7..fe1c1ac2 100644
--- a/cache/cache.go
+++ b/cache/cache.go
@@ -218,11 +218,6 @@ func ChangeSet(ctx context.Context, walker walk.Walker, filesCh chan<- *walk.Fil
 			}
 		}
 
-		// ignore symlinks
-		if file.Info.Mode()&os.ModeSymlink == os.ModeSymlink {
-			return nil
-		}
-
 		// open a new read tx if there isn't one in progress
 		// we have to periodically open a new read tx to prevent writes from being blocked
 		if tx == nil {
diff --git a/cli/format.go b/cli/format.go
index 7218a956..95ff37bc 100644
--- a/cli/format.go
+++ b/cli/format.go
@@ -237,12 +237,9 @@ func (f *Format) walkFilesystem(ctx context.Context) func() error {
 				case <-ctx.Done():
 					return ctx.Err()
 				default:
-					// ignore symlinks and directories
-					if !(file.Info.IsDir() || file.Info.Mode()&os.ModeSymlink == os.ModeSymlink) {
-						stats.Add(stats.Traversed, 1)
-						stats.Add(stats.Emitted, 1)
-						f.filesCh <- file
-					}
+					stats.Add(stats.Traversed, 1)
+					stats.Add(stats.Emitted, 1)
+					f.filesCh <- file
 					return nil
 				}
 			})
diff --git a/walk/filesystem.go b/walk/filesystem.go
index 35aec7ce..729a4972 100644
--- a/walk/filesystem.go
+++ b/walk/filesystem.go
@@ -4,6 +4,7 @@ import (
 	"context"
 	"fmt"
 	"io/fs"
+	"os"
 	"path/filepath"
 )
 
@@ -32,6 +33,11 @@ func (f filesystemWalker) Walk(_ context.Context, fn WalkFunc) error {
 			return fmt.Errorf("no such file or directory '%s'", path)
 		}
 
+		// ignore directories and symlinks
+		if info.IsDir() || info.Mode()&os.ModeSymlink == os.ModeSymlink {
+			return nil
+		}
+
 		relPath, err := f.relPath(path)
 		if err != nil {
 			return fmt.Errorf("failed to determine a relative path for %s: %w", path, err)
diff --git a/walk/filesystem_test.go b/walk/filesystem_test.go
index 202f0595..503ec0d3 100644
--- a/walk/filesystem_test.go
+++ b/walk/filesystem_test.go
@@ -10,56 +10,37 @@ import (
 )
 
 var examplesPaths = []string{
-	".",
-	"elm",
 	"elm/elm.json",
-	"elm/src",
 	"elm/src/Main.elm",
-	"go",
 	"go/go.mod",
 	"go/main.go",
-	"haskell",
 	"haskell/CHANGELOG.md",
 	"haskell/Foo.hs",
 	"haskell/Main.hs",
-	"haskell/Nested",
 	"haskell/Nested/Foo.hs",
 	"haskell/Setup.hs",
 	"haskell/haskell.cabal",
 	"haskell/treefmt.toml",
-	"haskell-frontend",
 	"haskell-frontend/CHANGELOG.md",
 	"haskell-frontend/Main.hs",
 	"haskell-frontend/Setup.hs",
 	"haskell-frontend/haskell-frontend.cabal",
-	"html",
 	"html/index.html",
-	"html/scripts",
 	"html/scripts/.gitkeep",
-	"javascript",
-	"javascript/source",
 	"javascript/source/hello.js",
-	"nix",
 	"nix/sources.nix",
 	"nixpkgs.toml",
-	"python",
 	"python/main.py",
 	"python/requirements.txt",
 	"python/virtualenv_proxy.py",
-	"ruby",
 	"ruby/bundler.rb",
-	"rust",
 	"rust/Cargo.toml",
-	"rust/src",
 	"rust/src/main.rs",
-	"shell",
 	"shell/foo.sh",
-	"terraform",
 	"terraform/main.tf",
 	"terraform/two.tf",
 	"touch.toml",
 	"treefmt.toml",
-	"yaml",
 	"yaml/test.yaml",
 }
 
diff --git a/walk/git.go b/walk/git.go
index 58afa4e8..cfe5f12f 100644
--- a/walk/git.go
+++ b/walk/git.go
@@ -50,9 +50,13 @@ func (g *gitWalker) Walk(ctx context.Context, fn WalkFunc) error {
 				case <-ctx.Done():
 					return ctx.Err()
 				default:
-					path := filepath.Join(g.root, entry.Name)
+					// we only want regular files, not directories or symlinks
+					if !entry.Mode.IsRegular() {
+						continue
+					}
 
 					// stat the file
+					path := filepath.Join(g.root, entry.Name)
 					info, err := os.Lstat(path)
 
 					file := File{