os

An operating system
git clone https://erai.gay/code/os/
Log | Files | Refs | README | LICENSE
commit 28476526f86abe770d02adb0c6cc86867e363133
parent ef9f830ae2d8d4f17258b50c16c7a75304637077
Author: erai <erai@omiltem.net>
Date:   Sat, 14 Sep 2024 18:22:18 -0400

basic peg syntax error location

Diffstat:

Mcc3.c | 2+-


Mparsepeg.c | 36++++++++++++++++++------------------


Mpeg.c | 40++++++++++++++++++++++++++++++++++------


Mpeglib.c | 83++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++---


Dpxe.sh | 5-----


Mwatch.sh | 2+-


6 files changed, 134 insertions(+), 34 deletions(-)

diff --git a/cc3.c b/cc3.c
@@ -18,7 +18,7 @@ main(argc: int, argv: **byte, envp: **byte) {
 	src = freadall(f, &len);
 	fclose(f);
 
-	peg = peg_new(src, len, &a);
+	peg = peg_new(argv[1], src, len, &a);
 	node = peg_parse(peg);
 	peg_free(peg);
 
diff --git a/parsepeg.c b/parsepeg.c
@@ -37,7 +37,7 @@ tag_to_str(tag: int): *byte {
 
 p_grammar(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_grammar);
     ok = p_sp(c);
     if ok {
     ok = p_rule(c);
@@ -61,7 +61,7 @@ p_grammar(c: *peg): int {
 
 p_rule(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_rule);
     ok = p_identifier(c);
     if ok {
     ok = p_sp(c);
@@ -81,7 +81,7 @@ p_rule(c: *peg): int {
 
 p_pattern(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_pattern);
     ok = p_alternative(c);
     if ok {
     loop {
@@ -108,7 +108,7 @@ p_pattern(c: *peg): int {
 
 p_alternative(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_alternative);
     ok = p_lookahead(c);
     if ok {
     loop {
@@ -124,7 +124,7 @@ p_alternative(c: *peg): int {
 
 p_lookop(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_lookop);
     ok = charset(c, "!&");
     if ok { leave(c, P_lookop); } else { fail(c); }
     return ok;
@@ -132,7 +132,7 @@ p_lookop(c: *peg): int {
 
 p_lookahead(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_lookahead);
     choice(c);
     ok = p_lookop(c);
     if ok {
@@ -148,7 +148,7 @@ p_lookahead(c: *peg): int {
 
 p_countop(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_countop);
     ok = charset(c, "*+?");
     if ok { leave(c, P_countop); } else { fail(c); }
     return ok;
@@ -156,7 +156,7 @@ p_countop(c: *peg): int {
 
 p_suffix(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_suffix);
     ok = p_primary(c);
     if ok {
     loop {
@@ -175,7 +175,7 @@ p_suffix(c: *peg): int {
 
 p_primary(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_primary);
     choice(c);
     ok = literal(c, "(");
     if ok {
@@ -221,7 +221,7 @@ p_primary(c: *peg): int {
 
 p_any(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_any);
     ok = literal(c, ".");
     if ok { leave(c, P_any); } else { fail(c); }
     return ok;
@@ -229,13 +229,13 @@ p_any(c: *peg): int {
 
 p_literal(c: *peg): int {
     var ok: int;
-    enter(c);
-    ok = charset(c, "'");
+    enter(c, P_literal);
+    ok = literal(c, "'");
     if ok {
     loop {
     choice(c);
     choice(c);
-    ok = charset(c, "'");
+    ok = literal(c, "'");
     if ok { fail(c); fail(c); ok = 0; } else { ok = 1; }
     if ok {
     ok = any(c);
@@ -245,7 +245,7 @@ p_literal(c: *peg): int {
     }
     }
     if ok {
-    ok = charset(c, "'");
+    ok = literal(c, "'");
     }
     if ok { leave(c, P_literal); } else { fail(c); }
     return ok;
@@ -253,7 +253,7 @@ p_literal(c: *peg): int {
 
 p_class(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_class);
     ok = literal(c, "[");
     if ok {
     loop {
@@ -291,7 +291,7 @@ p_class(c: *peg): int {
 
 p_call(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_call);
     ok = p_identifier(c);
     if ok {
     choice(c);
@@ -307,7 +307,7 @@ p_call(c: *peg): int {
 
 p_identifier(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_identifier);
     ok = charset(c, "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ_abcdefghijklmnopqrstuvwxyz");
     if ok {
     loop {
@@ -323,7 +323,7 @@ p_identifier(c: *peg): int {
 
 p_sp(c: *peg): int {
     var ok: int;
-    enter(c);
+    enter(c, P_sp);
     loop {
     choice(c);
     choice(c);
diff --git a/peg.c b/peg.c
@@ -77,6 +77,7 @@ translate_literal(c: *compiler, n: *peg_node) {
 
 	hex = "0123456789abcdef";
 
+	fputs(c.out, "    ok = literal(c, \"");
 	i = 1;
 	len = n.len - 1;
 	loop {
@@ -97,6 +98,7 @@ translate_literal(c: *compiler, n: *peg_node) {
 
 		i = i + 1;
 	}
+	fputs(c.out, "\");\n");
 }
 
 hexdig(c: byte): int {
@@ -166,6 +168,7 @@ translate_charset(c: *compiler, n: *peg_node) {
 	var a: int;
 	var b: int;
 	var hex: *byte;
+	var count: int;
 
 	hex = "0123456789abcdef";
 
@@ -217,6 +220,29 @@ translate_charset(c: *compiler, n: *peg_node) {
 		}
 	}
 
+	count = 0;
+	i = 1;
+	loop {
+		if i == 256 {
+			break;
+		}
+		count = count + c.scratch[i]:int;
+		i = i + 1;
+	}
+
+	if count == 0 {
+		fputs(c.out, "    fail(c);\n");
+		fputs(c.out, "    ok = 0;\n");
+		return;
+	} else if count >= 255 {
+		fputs(c.out, "    ok = any(c);\n");
+		return;
+	} else if count == 1 {
+		fputs(c.out, "    ok = literal(c, \"");
+	} else {
+		fputs(c.out, "    ok = charset(c, \"");
+	}
+
 	i = 1;
 	loop {
 		if i == 256 {
@@ -236,6 +262,7 @@ translate_charset(c: *compiler, n: *peg_node) {
 
 		i = i + 1;
 	}
+	fputs(c.out, "\");\n");
 }
 
 translate_pattern(c: *compiler, n: *peg_node) {
@@ -333,13 +360,9 @@ translate_pattern(c: *compiler, n: *peg_node) {
 		} else if n.tag == P_any {
 			fputs(c.out, "    ok = any(c);\n");
 		} else if n.tag == P_literal {
-			fputs(c.out, "    ok = literal(c, \"");
 			translate_literal(c, n);
-			fputs(c.out, "\");\n");
 		} else if n.tag == P_class {
-			fputs(c.out, "    ok = charset(c, \"");
 			translate_charset(c, n);
-			fputs(c.out, "\");\n");
 		} else if n.tag == P_call {
 			fputs(c.out, "    ok = p_");
 			fputb(c.out, n.child.str, n.child.len);
@@ -410,7 +433,9 @@ translate(c: *compiler, n: *peg_node) {
 			fputb(c.out, v.child.str, v.child.len);
 			fputs(c.out, "(c: *peg): int {\n");
 			fputs(c.out, "    var ok: int;\n");
-			fputs(c.out, "    enter(c);\n");
+			fputs(c.out, "    enter(c, P_");
+			fputb(c.out, v.child.str, v.child.len);
+			fputs(c.out, ");\n");
 			translate_pattern(c, v.child.next);
 			fputs(c.out, "    if ok { leave(c, P_");
 			fputb(c.out, v.child.str, v.child.len);
@@ -434,10 +459,12 @@ main(argc: int, argv: **byte, envp: **byte) {
 	var src: *byte;
 	var len: int;
 	var node: *peg_node;
+	var filename: *byte;
 	setup_alloc(&a);
 
 	ifd = 0;
 	ofd = 1;
+	filename = "-";
 
 	i = 1;
 	loop {
@@ -470,6 +497,7 @@ main(argc: int, argv: **byte, envp: **byte) {
 			die("too many inputs");
 		}
 
+		filename = argv[i];
 		ifd = open(argv[i], 0, 0);
 		if ifd < 0 {
 			die("failed to open input");
@@ -488,7 +516,7 @@ main(argc: int, argv: **byte, envp: **byte) {
 	out = fopen(ofd, c.a);
 	c.out = out;
 
-	c.p = peg_new(src, len, c.a);
+	c.p = peg_new(filename, src, len, c.a);
 	node = peg_parse(c.p);
 
 	translate(&c, node);
diff --git a/peglib.c b/peglib.c
@@ -2,6 +2,9 @@ struct peg_frame {
 	pos: int;
 	depth: int;
 	op: int;
+	tag: int;
+	line: int;
+	col: int;
 }
 
 struct peg_op {
@@ -14,9 +17,20 @@ struct peg_op {
 struct peg {
 	a: *alloc;
 
+	filename: *byte;
+
 	src: *byte;
 	size: int;
 	pos: int;
+	line: int;
+	col: int;
+	tag: int;
+
+	fail_depth: int;
+	fail_tag: int;
+	fail_line: int;
+	fail_col: int;
+	fail_literal: *byte;
 
 	stack: *peg_frame;
 	sp: int;
@@ -47,6 +61,9 @@ choice(c: *peg) {
 	c.stack[c.sp].pos = c.pos;
 	c.stack[c.sp].depth = c.depth;
 	c.stack[c.sp].op = c.op;
+	c.stack[c.sp].tag = c.tag;
+	c.stack[c.sp].line = c.line;
+	c.stack[c.sp].col = c.col;
 	c.sp = c.sp + 1;
 }
 
@@ -61,10 +78,22 @@ fail(c: *peg) {
 	if c.sp == 0 {
 		die("fail underflow");
 	}
+
+	if c.depth > c.fail_depth {
+		c.fail_depth = c.depth;
+		c.fail_tag = c.tag;
+		c.fail_line = c.line;
+		c.fail_col = c.col;
+		c.fail_literal = 0:*byte;
+	}
+
 	c.sp = c.sp - 1;
 	c.pos = c.stack[c.sp].pos;
 	c.depth = c.stack[c.sp].depth;
 	c.op = c.stack[c.sp].op;
+	c.tag = c.stack[c.sp].tag;
+	c.line = c.stack[c.sp].line;
+	c.col = c.stack[c.sp].col;
 }
 
 get(c: *peg): int {
@@ -76,6 +105,16 @@ get(c: *peg): int {
 
 	ch = c.src[c.pos]:int;
 	c.pos = c.pos + 1;
+	c.col = c.col + 1;
+
+	if ch == '\n' {
+		c.col = 1;
+		c.line = c.line + 1;
+	}
+
+	if ch == 0 {
+		die("invalid nul in source");
+	}
 
 	return ch;
 }
@@ -93,6 +132,7 @@ literal(c: *peg, s: *byte): int {
 		ch = get(c);
 		if ch != (s[i]:int) {
 			fail(c);
+			c.fail_literal = s;
 			return 0;
 		}
 
@@ -102,8 +142,9 @@ literal(c: *peg, s: *byte): int {
 	return 1;
 }
 
-enter(c: *peg) {
+enter(c: *peg, tag: int) {
 	choice(c);
+	c.tag = tag;
 }
 
 leave(c: *peg, tag: int) {
@@ -113,6 +154,7 @@ leave(c: *peg, tag: int) {
 	var tmp: *byte;
 
 	commit(c);
+	c.fail_depth = 0;
 
 	nargs = c.depth - c.stack[c.sp].depth;
 	start = c.stack[c.sp].pos;
@@ -226,16 +268,28 @@ construct(c: *peg): *peg_node {
 	}
 }
 
-peg_new(src: *byte, len: int, a: *alloc): *peg {
+peg_new(filename: *byte, src: *byte, len: int, a: *alloc): *peg {
 	var c: *peg;
 
 	c = alloc(a, sizeof(*c)):*peg;
 
 	c.a = a;
 
+	c.filename = filename;
+
 	c.src = src;
 	c.size = len;
 	c.pos = 0;
+	c.tag = 0;
+	c.line = 1;
+	c.col = 1;
+	c.tag = 0;
+
+	c.fail_depth = 0;
+	c.fail_tag = 0;
+	c.fail_line = 0;
+	c.fail_col = 0;
+	c.fail_literal = 0:*byte;
 
 	c.limit = 1024;
 	c.stack = alloc(a, c.limit * sizeof(*c.stack)):*peg_frame;
@@ -256,7 +310,21 @@ peg_new(src: *byte, len: int, a: *alloc): *peg {
 peg_parse(c: *peg): *peg_node {
 	choice(c);
 	if !p_grammar(c) {
-		die("syntax error");
+		fdputs(2, "syntax error at ");
+		fdputs(2, c.filename);
+		fdputs(2, ":");
+		fdputd(2, c.fail_line);
+		fdputs(2, ":");
+		fdputd(2, c.fail_col);
+		fdputs(2, " expected ");
+		fdputs(2, tag_to_str(c.fail_tag));
+		if c.fail_literal {
+			fdputs(2, " '");
+			fdputs(2, c.fail_literal);
+			fdputs(2, "'");
+		}
+		fdputs(2, "\n");
+		exit(1);
 	}
 	commit(c);
 	return construct(c);
@@ -266,6 +334,15 @@ peg_reset(c: *peg, src: *byte, len: int) {
 	c.src = src;
 	c.size = len;
 	c.pos = 0;
+	c.tag = 0;
+	c.line = 1;
+	c.col = 1;
+	c.tag = 0;
+	c.fail_depth = 0;
+	c.fail_tag = 0;
+	c.fail_line = 0;
+	c.fail_col = 0;
+	c.fail_literal = 0:*byte;
 	c.depth = 0;
 	c.sp = 0;
 	c.op = 0;
diff --git a/pxe.sh b/pxe.sh
@@ -1,5 +0,0 @@
-#!/bin/sh
-./bootstrap.sh || exit 1
-nasm -f bin pxe.asm || exit 1
-cp kernel pxe /srv/tftp/ || exit 1
-printf r > /dev/ttyACM0 || exit 1
diff --git a/watch.sh b/watch.sh
@@ -1,6 +1,6 @@
 #!/bin/sh
 clear
-./pxe.sh
+sh -e ./build.sh
 echo status $?
 : < ~/.post
 exec "$0"