add format. Improve tests.

2 files changed, 115 insertions, 53 deletions

diff --git a/mp b/mp
index 337f316..9f5209a 100755
--- a/mp
+++ b/mp
@@ -1,7 +1,7 @@
 #!/usr/bin/awk -f
 
 BEGIN {
-	version = "2025-07-08"
+	version = "0.1"
 
 	help[""] = "Commands:\n" \
 		"  help [cmd]   prints cmd help or this help text\n"
@@ -16,18 +16,28 @@ BEGIN {
 	if (filename) src = getfile(filename)
 	prompt("mp-" version " meta parser. Type \"help\" for help.\n")
 }
+{ ERROR = "" }
 $1 == "help" {
 	prompt(help[$2])
 	next
 }
 $1 == "src" {
-	prompt(src)
+	printf "%s", src
+	prompt()
+	next
+}
+$1 == "format" {
+	print format_json(v, "", 1)
+	prompt()
 	next
 }
 $1 == "parse" {
 	delete v
-	if ($2) parse_json($2, v); else parse_json(src, v)
-	for (k in v) print "v[" k "]: " v[k]
+	sub(/^[[:space:]]+/, "")
+	$0 = substr($0, 6)
+	if ($0) src = $0
+	parse_json(src, v)
+	print format_json(v)
 	prompt()
 	next
 }
@@ -35,65 +45,112 @@ $1 == "parse" {
 	if (NF) error("invalid command: " $1)
 	prompt()
 }
+END { if (ERROR) exit 1 }
 
-function error(s) { ERROR = s }
+function error(s, n) { ERROR = s; return n }
 
 function prompt(s) {
-	if (ERROR) {
-		print ERROR > "/dev/stderr"
-		ERROR = ""
-	}
-	if (tty) printf("%smp> ", s)
+	if (ERROR) print ERROR > "/dev/stderr"
+	if (tty) printf("%s> ", s)
 }
 
 function getfile(name,  l, r, o) {
 	while (r = getline l < name) {
-		if (r == -1) {
-			error = "getfile " name ": getline error"
-			break
-		}
+		if (r == -1) return error("getfile " name ": getline error")
 		o = o l "\n"
 	}
-	close(name)
 	return o
 }
 
-# parse the next json value in s, returns the number of characters read.
-function parse_json(s, a, p,    c, i, n) {
-	if (match(s, /^[[:space:]]+/)) {
+function next_json_token(str,   n) {
+	n = 0
+	TOKEN = TOKSTR = ""
+	if (match(str, /^[[:space:]]+/)) {
 		n = RLENGTH
-		s = substr(s, RLENGTH+1)
+		str = substr(str, RLENGTH+1)
 	}
-	if (!s) return n
-	if (match(s, /^null|true|false/)) {
-		a[p, "type"] = substr(s, 1, RLENGTH)
-		return n + RLENGTH
+	if (match(str, /^(null|true|false|[][}{,:])/)) {
+		TOKEN = substr(str, 1, RLENGTH)
+	} else if (match(str, /^[.0-9Ee+-]+/)) {
+		TOKEN = "number"
+		TOKSTR = substr(str, 1, RLENGTH)
+	} else if (match(str, /^"(\\.|[^\\"])*"/)) {
+		TOKEN = "string"
+		TOKSTR = substr(str, 2, RLENGTH-2)
+	} else {
+		TOKSTR = substr(str, 1, 1)
+		TOKEN = TOKSTR ? "invalid" : ""
 	}
-	if (match(s, /^[.0-9Ee+-]+/)) {
-		a[p, "type"] = "number"
-		a[p, "string"] = substr(s, 1, RLENGTH)
-		return n + RLENGTH
-	}
-	if (match(s, /^"(\\.|[^\\"])*"/)) {
-		a[p, "type"] = "string"
-		a[p, "string"] = substr(s, 2, RLENGTH-2)
-		return n + RLENGTH
+	return n + RLENGTH
+}
+
+function parse_json(str, arr, pk,   i, n) {
+	n = next_json_token(str)
+	if (TOKEN == "[") {
+		arr[pk, "type"] = "array"
+		for (i = 0; TOKEN != "]"; i++) {
+			if (i > 0 && TOKEN != ",") return error("not a ','", n)
+			n += parse_json(substr(str, n+1), arr, pk SUBSEP "value" SUBSEP i)
+			if (ERROR) {
+				if (i == 0 && TOKEN == "]") ERROR = ""
+				break
+			}
+			n += next_json_token(substr(str, n+1))
+		}
+		arr[pk, "size"] = i
+	} else if (TOKEN == "{") {
+		arr[pk, "type"] = "object"
+		for (i = 0; TOKEN != "}"; i++) {
+			if (i > 0 && TOKEN != ",") return error("not a ','", n)
+			n += next_json_token(substr(str, n+1))
+			if (i == 0 && TOKEN == "}") break
+			if (TOKEN != "string") return error("not a string", n)
+			arr[pk, "key", i] = TOKSTR
+			n += next_json_token(substr(str, n+1))
+			if (TOKEN != ":") return error("not a ':'", n)
+			n += parse_json(substr(str, n+1), arr, pk SUBSEP "value" SUBSEP i)
+			if (ERROR) return n
+			n += next_json_token(substr(str, n+1))
+		}
+		arr[pk, "size"] = i
+	} else if (TOKEN ~ /^(null|true|false|string|number)$/) {
+		arr[pk, "type"] = TOKEN
+		if (TOKSTR) arr[pk, "string"] = TOKSTR
+    } else if (TOKEN) ERROR = "invalid token '" TOKEN "'"
+	return n
+}
+
+function format_json(arr, pk, il, i, l, t, s, bl, al, ps) {
+	if (il) {
+		bl = space(il * 2)
+		al = "\n"
+		ps = " "
 	}
-	c = substr(s, 1, 1)
-	if (c == "{") {
-		a[p, "type"] = "object"
-		i = 0
-		n++
-		s = substr(s, 2)
-		if (match(s, /^[[:space:]]+/)) {
-			n += RLENGTH
-			s = substr(s, RLENGTH+1)
+	t = arr[pk, "type"]
+	if (t == "null" || t == "true" || t == "false") return t
+	if (t == "number") return arr[pk, "string"]
+	if (t == "string") return "\"" arr[pk, "string"] "\""
+	if (t == "object") {
+		s = "{" al
+		l = arr[pk, "size"]
+		for (i = 0; i < l; i++) {
+			s = s bl "\"" arr[pk, "key", i] "\":" ps format_json(arr, pk SUBSEP "value" SUBSEP i, il ? il+1 : 0)
+			if (i < l-1) s = s "," al
 		}
-		if (match(s, /^"(\\.|[^\\"])*"/)) {
-			a[p, "key", i] = substr(s, 2, RLENGTH-2)
-		} else {
-			error("invalid key: " s)
-			return n
+		return s al substr(bl, 3) "}"
+	}
+	if (t == "array") {
+		s = "[" al
+		l = arr[pk, "size"]
+		for (i = 0; i < arr[pk, "size"]; i++) {
+			s = s bl format_json(arr, pk SUBSEP "value" SUBSEP i, il ? il+1 : 0)
+			if (i < l-1) s = s "," al
 		}
+		return s al substr(bl, 3) "]"
 	}
 }
+
+function space(n,  i, s) {
+	for (i = 0; i < n; i++) s = s " "
+	return s
+}
diff --git a/tests b/tests
index 1c8dcff..50e1c75 100755
--- a/tests
+++ b/tests
@@ -2,8 +2,9 @@
 # Test suite for mp.
 
 run() {
-	[ "$filter" ] && case $1 in $filter) ;; *) return; esac
-	out=$(echo "$2" | ./mp 2>&1 | tr -d '\034' | awk -v ORS='\\n' 1)
+	# shellcheck disable=SC2254
+	[ "$filter" ] && case $1 in ($filter) ;; (*) return; esac
+	out=$(echo "$2" | ./mp 2>&1)
 	[ "$out" = "$3" ] && pass=$((pass + 1)) && return
 	printf "%s FAIL\nWant: \"%s\"\n Got: \"%s\"\n" "$1" "$3" "$out"
 	fail=$((fail + 1))
@@ -14,8 +15,12 @@ run() {
 pass=0 fail=0 filter="$1"
 trap 'echo "$pass passed, $fail failed"; exit $((fail))' EXIT
 
-run test1 'parse null'    'v[type]: null\n'
-run test2 'parse true'    'v[type]: true\n'
-run test3 'parse false'   'v[type]: false\n'
-run test4 'parse 12'      'v[string]: 12\nv[type]: number\n'
-run test5 'parse "hello"' 'v[string]: hello\nv[type]: string\n'
+run basic1 'parse' ''
+run basic2 'parse null' 'null'
+run basic3 'parse true' 'true'
+run basic4 'parse false' 'false'
+run num1 'parse 12' '12'
+run str1 'parse "hello"' '"hello"'
+run obj1 'parse {}' '{}'
+run arr1 'parse []' '[]'
+