hare

commit 5101d99b1600acebd796c730bb4ef2d3ff627fb3
parent 991b9323d6c52e02b96f54a16b7032e39560e53b
Author: Byron Torres <b@torresjrjr.com>
Date:   Mon, 23 May 2022 11:33:00 +0100

strings: add fromrunes()

Signed-off-by: Byron Torres <b@torresjrjr.com>

Diffstat:
M
scripts/gen-stdlib
 | 
1
+
M
stdlib.mk
 | 
2
++
M
strings/iter.ha
 | 
15
---------------
A
strings/runes.ha
 | 
66
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

4 files changed, 69 insertions(+), 15 deletions(-)
diff --git a/scripts/gen-stdlib b/scripts/gen-stdlib
@@ -1210,6 +1210,7 @@ strings() {
 		cstrings.ha \
 		dup.ha \
 		iter.ha \
+		runes.ha \
 		sub.ha \
 		suffix.ha \
 		tokenize.ha \
diff --git a/stdlib.mk b/stdlib.mk
@@ -1825,6 +1825,7 @@ stdlib_strings_any_srcs = \
 	$(STDLIB)/strings/cstrings.ha \
 	$(STDLIB)/strings/dup.ha \
 	$(STDLIB)/strings/iter.ha \
+	$(STDLIB)/strings/runes.ha \
 	$(STDLIB)/strings/sub.ha \
 	$(STDLIB)/strings/suffix.ha \
 	$(STDLIB)/strings/tokenize.ha \
@@ -3976,6 +3977,7 @@ testlib_strings_any_srcs = \
 	$(STDLIB)/strings/cstrings.ha \
 	$(STDLIB)/strings/dup.ha \
 	$(STDLIB)/strings/iter.ha \
+	$(STDLIB)/strings/runes.ha \
 	$(STDLIB)/strings/sub.ha \
 	$(STDLIB)/strings/suffix.ha \
 	$(STDLIB)/strings/tokenize.ha \
diff --git a/strings/iter.ha b/strings/iter.ha
@@ -145,18 +145,3 @@ export fn iterstr(iter: *iterator) str = {
 	assert(next(&s) is void);
 	assert(prev(&s) as rune == 'に');
 };
-
-// Returns a slice of runes for a string in O(n). The caller must free the
-// return value.
-export fn runes(s: str) []rune = {
-	let sl: []rune = alloc([], len(s));
-	let iter = iter(s);
-	for (true) {
-		match (next(&iter)) {
-		case void => break;
-		case let r: rune =>
-			append(sl, r);
-		};
-	};
-	return sl;
-};
diff --git a/strings/runes.ha b/strings/runes.ha
@@ -0,0 +1,66 @@
+// License: MPL-2.0
+// (c) 2022 Byron Torres <b@torresjrjr.com>
+use encoding::utf8;
+
+// Returns a slice of runes for a string in O(n). The caller must free the
+// return value.
+export fn runes(s: str) []rune = {
+	let sl: []rune = alloc([], len(s));
+	let iter = iter(s);
+	for (true) {
+		match (next(&iter)) {
+		case void => break;
+		case let r: rune =>
+			append(sl, r);
+		};
+	};
+	return sl;
+};
+
+// Returns a string from a slice of runes. The caller must free the return value.
+export fn fromrunes(rs: []rune) str = {
+	let bytes: []u8 = [];
+	for (let i = 0z; i < len(rs); i += 1) {
+		const bs = encoding::utf8::encoderune(rs[i]);
+		append(bytes, bs...);
+	};
+	return fromutf8(bytes);
+};
+
+@test fn fromrunes() void = {
+	const rs = ['H', 'a', 'r', 'r', 'i', 'e', 't'];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == "Harriet", "strings::fromrunes() error (Harriet)");
+
+	const rs: []rune = [];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == "", "strings::fromrunes() error (empty)");
+
+	const rs = ['.'];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == ".", "strings::fromrunes() error (period)");
+
+	const rs = ['\a', '\b', '\f', '\n', '\r', '\t', '\v'];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == "\a\b\f\n\r\t\v", "strings::fromrunes() error (control chars)");
+
+	const rs = ['H', 'e', 'l', 'l', 'o', ',', ' ', 'w', 'o', 'r', 'l', 'd', '!'];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == "Hello, world!", "strings::fromrunes() error (English)");
+
+	const rs = ['¡', 'H', 'o', 'l', 'a', ' ', 'M', 'u', 'n', 'd', 'o', '!'];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == "¡Hola Mundo!", "strings::fromrunes() error (Spanish)");
+
+	const rs = ['Γ', 'ε', 'ι', 'ά', ' ', 'σ', 'ο', 'υ', ' ', 'Κ', 'ό', 'σ', 'μ', 'ε', '!'];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == "Γειά σου Κόσμε!", "strings::fromrunes() error (Greek)");
+
+	const rs = ['П', 'р', 'и', 'в', 'е', 'т', ',', ' ', 'м', 'и', 'р', '!'];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == "Привет, мир!", "strings::fromrunes() error (Russian)");
+
+	const rs = ['こ', 'ん', 'に', 'ち', 'は', '世', '界', '！'];
+	const s = fromrunes(rs); defer free(s);
+	assert(s == "こんにちは世界！", "strings::fromrunes() error (Japanese)");
+};