hare

commit f679ee1c3a82b2b0b2612e1c38859ff0039f7742
parent 76af771721b0fc611633b2c97b1f0e9bfdcb33e7
Author: Alexey Yerin <yyp@disroot.org>
Date:   Sun,  7 Nov 2021 15:00:55 +0300

shlex: fix whitespace folding

Previously this caused split("hello '' world") to return
["hello", "world"] without the empty string in between.

Signed-off-by: Alexey Yerin <yyp@disroot.org>

Diffstat:
M
shlex/+test.ha
 | 
7
+++++++
M
shlex/split.ha
 | 
42
++++++++++++++++++++++++++++++++----------

2 files changed, 39 insertions(+), 10 deletions(-)
diff --git a/shlex/+test.ha b/shlex/+test.ha
@@ -19,6 +19,13 @@
 	assert(len(s) == 1);
 	assert(s[0] == "hello \"world\"");
 
+	const s = split("hello '' world")!;
+	defer splitfree(s);
+	assert(len(s) == 3);
+	assert(s[0] == "hello");
+	assert(s[1] == "");
+	assert(s[2] == "world");
+
 	const s = split("with\\ backslashes 'single quoted' \"double quoted\"")!;
 	defer splitfree(s);
 	assert(len(s) == 3);
diff --git a/shlex/split.ha b/shlex/split.ha
@@ -24,19 +24,28 @@ export fn split(in: const str) ([]str | syntaxerr) = {
 		};
 
 		switch (r) {
-		case '\\' =>
-			scan_backslash(s, in)?;
-		case '"' =>
-			scan_double(s, in)?;
-		case '\'' =>
-			scan_single(s, in)?;
 		case ' ', '\t', '\n' =>
-			if (len(strio::string(s)) > 0) {
-				append(slice, strio::finish(s));
-				s = strio::dynamic();
+			let r = :out {
+				for (true) match (bufio::scanrune(in)!) {
+				case r: rune =>
+					if (r != ' ' && r != '\t' && r != '\n') {
+						yield :out, r;
+					};
+				case io::EOF =>
+					yield :out, void;
+				};
+				abort(); // Unreachable
+			};
+			append(slice, strio::finish(s));
+			s = strio::dynamic();
+
+			match (r) {
+			case void => void;
+			case r: rune =>
+				process(s, in, r)?;
 			};
 		case =>
-			strio::appendrune(s, r)!;
+			process(s, in, r)?;
 		};
 	};
 
@@ -48,6 +57,19 @@ export fn split(in: const str) ([]str | syntaxerr) = {
 	return slice;
 };
 
+fn process(out: io::handle, in: io::handle, r: rune) (void | syntaxerr) = {
+	switch (r) {
+	case '\\' =>
+		scan_backslash(out, in)?;
+	case '"' =>
+		scan_double(out, in)?;
+	case '\'' =>
+		scan_single(out, in)?;
+	case =>
+		strio::appendrune(out, r)!;
+	};
+};
+
 fn scan_backslash(out: io::handle, in: io::handle) (void | syntaxerr) = {
 	const r = match (bufio::scanrune(in)!) {
 	case r: rune =>