hare

commit 5399f1779bee7067612f09f560cbccf4c387b0c4
parent 01ed35b2e7b8e6f45f2d4fe88562db5a66959a14
Author: Bor Grošelj Simić <bgs@turminal.net>
Date:   Fri,  8 Apr 2022 21:36:14 +0200

add bytes::cut and use it in strings::cut

and document that calling strings::cut with empty delimiter is invalid

Signed-off-by: Bor Grošelj Simić <bgs@turminal.net>

Diffstat:
M
bytes/tokenize.ha
 | 
34
++++++++++++++++++++++++++++++++++
M
strings/tokenize.ha
 | 
13
+++----------

2 files changed, 37 insertions(+), 10 deletions(-)
diff --git a/bytes/tokenize.ha b/bytes/tokenize.ha
@@ -186,3 +186,37 @@ export fn remaining_tokens(s: *tokenizer) []u8 = {
 	assert(peek_token(&t) is void);
 	assert(next_token(&t) is void);
 };
+
+// Returns a the slice "cut" along the first instance of a delimiter, returning
+// everything up to the delimiter, and everything after the delimiter, in a
+// tuple.
+//
+// Caller must ensure delimiter is not an empty slice.
+export fn cut(in: []u8, delim: ([]u8 | u8)) ([]u8, []u8) = {
+	let ln = if (delim is u8) {
+		yield 1z;
+	} else {
+		let ln = len(delim: []u8);
+		assert(ln > 0, "bytes::cut called with empty delimiter");
+		yield ln;
+	};
+	match (index(in, delim)) {
+	case let i: size =>
+		return (in[..i], in[i + ln..]);
+	case void =>
+		return (in, []);
+	};
+};
+
+@test fn cut() void = {
+	const c = cut(['a', 'b', 'c'], ['b']);
+	assert(equal(c.0, ['a']) && equal(c.1, ['c']));
+	const c = cut(['a', 'b', 'c'], 'b');
+	assert(equal(c.0, ['a']) && equal(c.1, ['c']));
+	const c = cut(['a', 'b', 'c', 'b', 'a'], 'b');
+	assert(equal(c.0, ['a']) && equal(c.1, ['c', 'b', 'a']));
+	const c = cut(['a', 'b', 'c'], 'x');
+	assert(equal(c.0, ['a', 'b', 'c']) && equal(c.1, []));
+	const c = cut([], 'x');
+	assert(equal(c.0, []) && equal(c.1, []));
+};
diff --git a/strings/tokenize.ha b/strings/tokenize.ha
@@ -143,17 +143,10 @@ export fn split(in: str, delim: str) []str = splitn(in, delim, types::SIZE_MAX);
 // 	strings::cut("hello world", "=")	// ("hello world", "")
 //
 // The return value is borrowed from the 'in' parameter.
+// Caller must ensure delimiter is not an empty string
 export fn cut(in: str, delim: str) (str, str) = {
-	const tok = tokenize(in, delim);
-	let res = ("", "");
-	match (next_token(&tok)) {
-	case let s: str =>
-		res.0 = s;
-	case void =>
-		return res;
-	};
-	res.1 = remaining_tokens(&tok);
-	return res;
+	let c = bytes::cut(toutf8(in), toutf8(delim));
+	return (fromutf8_unsafe(c.0), fromutf8_unsafe(c.1));
 };
 
 @test fn cut() void = {