zc0: byte string types, u/m/d as names, enhancements in asm

bstr locals, globals, parameters and returns are byte strings for + , the
comparisons, indexing, slicing, len and put_s; b"..." globals; embed inside
a function. u, m and d are ordinary names: a recast is the letter with the
star touching it (u*x), tags .u .m .d are accepted on types and bare. an
enhancement named inside asm { } emits its opcode. zc0's image of the
genome and language server sources is unchanged.

1 files changed, 96 insertions, 0 deletions

diff --git a/tests/programs/zc0bstr.zc b/tests/programs/zc0bstr.zc
new file mode 100644
index 0000000..bb4ffbc
--- /dev/null
+++ b/tests/programs/zc0bstr.zc
@@ -0,0 +1,96 @@
+namespace;
+include rt;
+
+// packed byte strings in the zc0 subset: the same program runs on the resident
+// compiler and, built into a run-only image, on stage 0 (src/tools/zc0.c). it
+// covers every b"..." escape, the packed layout, bstr locals, globals,
+// parameters and returns, + and the six comparisons, indexing, slicing, len
+// and embed. run from the repository root: the embeds read tests/data/embed.
+
+bstr greeting = b"hello";
+bstr shout;
+bstr plain = b"plain";
+embed "tests/data/embed/text.txt" as bstr text;
+embed "tests/data/embed/nul.bin" as bstr nul;
+embed "tests/data/embed/text.txt" as untyped;
+
+bstr tag(bstr s) { return b"<" + s + b">"; }
+bstr twice(bstr s);
+bstr twice(bstr s) { return s + s; }
+bstr pick(int n) {
+	bstr r;
+	if (n) r = b"yes"; else r = b"no";
+	return r;
+}
+int rank(bstr a, bstr b) { return (a < b) * 8 + (a <= b) * 4 + (a == b) * 2 + (a > b); }
+int count(bstr s, int c) {
+	int n = 0, i = 0;
+	while (i < len(s)) { if (s[i] == c) n++; i++; }
+	return n;
+}
+int num(int n) { rt.put_n(n); rt.put_c(32); return 0; }
+int nl() { rt.put_c(10); return 0; }
+int bytes(bstr s) {
+	int i = 0;
+	num(len(s));
+	while (i < len(s)) { num(s[i]); i++; }
+	nl();
+	return 0;
+}
+int words(bstr s) {
+	word* w = rt.bstr_storage(s);
+	int n = 1 + (w[0] >> 3) + 1, i = 0;
+	while (i < n) { num(w[i]); i++; }
+	nl();
+	return 0;
+}
+
+fn main() {
+	bstr a = b"abc", b, c;
+	bstr e = b"";
+	let u = b"x";
+	// every escape the lexer knows: n t r 0 e, and any other character stands for itself
+	bytes(b"\n\t\r\e\\\"\'\q\x41");
+	bytes(b"ab\0cd");
+	bytes(b"\0");
+	bytes(b"café");
+	bytes(b"caf\303\251");
+	bytes(b"quote \" inside, 'single', back\\slash");
+	// the layout: a length word, bytes packed 8 to a word, a 0 word after a full last word
+	words(e);
+	words(b"a");
+	words(b"abcdefg");
+	words(b"abcdefgh");
+	words(b"abcdefghi");
+	words(b"0123456789abcdef");
+	words(b"0123456789abcdefg");
+	words(text);
+	words(nul);
+	// declarations, assignment later, a return value, concatenation
+	b = tag(a);
+	c = twice(b) + e + u;
+	rt.put_s(a); rt.put_c(32); rt.put_s(b); rt.put_c(32); rt.put_s(c); nl();
+	shout = twice(greeting) + b"!";
+	rt.put_s(greeting); rt.put_c(32); rt.put_s(shout); rt.put_c(32); rt.put_s(plain); nl();
+	rt.put_s(tag(twice(tag(b"z")))); nl();
+	rt.put_s(pick(1) + b"/" + pick(0)); nl();
+	// comparisons: below, equal, above, a prefix, the empty string, a 9 byte string
+	num(rank(b"abc", b"abd")); num(rank(b"abc", b"abc")); num(rank(b"abd", b"abc"));
+	num(rank(b"ab", b"abc")); num(rank(b"", b"")); num(rank(b"", b"a"));
+	num(rank(b"abcdefghi", b"abcdefghi")); num(rank(b"abcdefgh", b"abcdefghi")); nl();
+	num(a == b"abc"); num(a != b"abc"); num(a == b); num(a != b);
+	num(tag(a) == b"<abc>"); num(shout > greeting); num(greeting >= shout); num(plain <= plain); nl();
+	// indexing, slicing, len
+	num(a[0]); num(a[2]); num(a[3]); num(len(a)); num(len(e)); num(len(c)); num(len(b"123456789")); nl();
+	rt.put_s(c[1:4]); rt.put_c(32); rt.put_s(a[1:2]); rt.put_c(32); rt.put_s(a[0:0]); rt.put_c('|');
+	rt.put_s(a[2:99]); rt.put_c('|'); rt.put_s(a[3:1]); rt.put_c('|'); rt.put_s((a + a)[2:4]); nl();
+	num(count(c, '<')); num(count(shout, 'l')); num(count(b"mississippi", 's')); num(count(e, 'x')); nl();
+	num(b"xyz"[1]); num(len(b"xyz")); num(len(a + a + a)); num(len(tag(e))); nl();
+	// embed: a global of the file's bytes, with or without the type, and a local
+	rt.put_s(text); num(len(text)); num(text[0]); num(text == untyped); num(len(nul)); num(nul[2]); nl();
+	embed "tests/data/embed/text.txt" as bstr here;
+	embed "tests/data/embed/answer.bin" as int answer;
+	embed "tests/data/embed/text.txt" as char* chars;
+	rt.put_s(here + b"|" + untyped); num(len(here)); num(here == text); num(answer); num(chars[6]); nl();
+	return 0;
+}