deblob

commit: 4a0166e23b319388c7356c88a3ca2851b7438395
parent 074b50fa51aebb4378bc1e50fac94a4332bd6643
Author: Drew DeVault <sir@cmpwn.com>
Date:   Fri, 18 Feb 2022 10:26:16 +0100

Use os::iter instead of os::readdir

This should be faster and more memory efficient, especially for large
directories, as it skips an allocation and allows the tool to work
incrementally. It also addresses a memory leak in the previous version;
the caller must free the return value of os::readdir.

Diffstat:
M
main.ha
28
+++++++++++++++++-----------

1 file changed, 17 insertions(+), 11 deletions(-)
diff --git a/main.ha b/main.ha
@@ -90,19 +90,25 @@ fn is_excluded(filename: str) bool = {
 };
 
 fn check_dir(dirname: str, ignoring: bool) (ok | invalid) = {
-	const dir_ls = match (os::readdir(dirname)) {
-	case let d: []fs::dirent =>
-		yield d;
-	case let e: fs::error =>
-		fmt::errorf("os::readdir({}): {}\n", dirname, fs::strerror(e))!;
+	const iter = match (os::iter(dirname)) {
+	case let iter: *fs::iterator =>
+		yield iter;
+	case let err: fs::error =>
+		fmt::errorf("os::iter({}): {}\n", dirname, fs::strerror(err))!;
 		return invalid;
 	};
 
-	for (let i = 0z; i < len(dir_ls); i += 1) {
-		const file = dir_ls[i];
-		const filename = path::join(dirname, file.name);
+	for (true) {
+		const ent = match (fs::next(iter)) {
+		case let ent: fs::dirent =>
+			yield ent;
+		case void =>
+			break;
+		};
+
+		const filename = path::join(dirname, ent.name);
 
-		if (file.name == "." || file.name == "..") {
+		if (ent.name == "." || ent.name == "..") {
 			continue;
 		};
 
@@ -110,9 +116,9 @@ fn check_dir(dirname: str, ignoring: bool) (ok | invalid) = {
 			ignoring = is_excluded(filename);
 		};
 
-		if(fs::isdir(file.ftype)) {
+		if (fs::isdir(ent.ftype)) {
 			check_dir(filename, ignoring): void;
-		} else if(fs::isfile(file.ftype)) {
+		} else if(fs::isfile(ent.ftype)) {
 			const is_blob = match (is_blob(filename)) {
 			case let b: bool =>
 				yield b;