X-Git-Url: http://pileus.org/git/?p=~andy%2Frhawk;a=blobdiff_plain;f=json.awk;h=cfb8eac56c680460b9e12d4c6d7a1e7c3f45a61c;hp=59b5eb5c5f7c53ae9f9b0af5762160bd96d38867;hb=HEAD;hpb=6fd3e11f6de61947f4f7c704b09ce77f827de4e2

diff --git a/json.awk b/json.awk
index 59b5eb5..cfb8eac 100644
--- a/json.awk
+++ b/json.awk
@@ -1,3 +1,12 @@
+# usage:
+# 	@include "json.awk"
+# 	BEGIN {
+# 		src["hello"] = "world"
+# 		json_save("test.json", src)
+# 		json_load("test.json", dst)
+# 		print dst["hello"]
+# 	}
+#
 # value:
 # 	object: { string : value, .. }
 # 	array:  [ value, .. ]
@@ -19,7 +28,6 @@
 # 	\t
 # 	\u four-hex-digits
 
-
 # Helpers
 function json_gettype(value,   key, sort, n, i)
 {
@@ -33,6 +41,8 @@ function json_gettype(value,   key, sort, n, i)
 				return "object"
 		return "array"
 	} else {
+		if (value == 0 && value == "")
+			return "null"
 		if (value == value + 0)
 			return "number"
 		if (value == value "")
@@ -71,6 +81,7 @@ function json_write_value(value, pad)
 		case "array":  return json_write_array(value, pad)
 		case "number": return json_write_number(value)
 		case "string": return json_write_string(value)
+		case "null":   return "null"
 		default:       return "error"
 	}
 }
@@ -106,29 +117,41 @@ function json_write_number(number)
 
 function json_write_string(string)
 {
-	# todo: special characters
+	gsub(/\\/, "\\\\", string)
+	gsub(/"/,  "\\\"", string)
+	gsub(/\b/, "\\b",  string)
+	gsub(/\f/, "\\f",  string)
+	gsub(/\n/, "\\n",  string)
+	gsub(/\r/, "\\r",  string)
+	gsub(/\t/, "\\t",  string)
+
 	return "\"" string "\""
 }
 
 
 # Read functions
-function json_tokenize(str, tokens,   i, items, table, type, found)
+function json_tokenize(str, tokens,   i, line, items, table, type, found)
 {
 	table["term"]  = "^[\\[\\]{}:,]"
-	table["str"]   = "^\"[^\"]*\""
-	table["num"]   = "^[0-9]+"
+	table["str"]   = "^\"([^\\\\\"]|\\\\.)*\""
+	table["num"]   = "^[+-]?[0-9]+(.[0-9]+)?"
 	table["var"]   = "^(true|false|null)"
-	table["space"] = "^[ \\t\\n]+"
-	i = 0
+	table["space"] = "^[ \\t]+"
+	table["line"]  = "^\n"
+	i = 0;
+	line = 1;
 	while (length(str) > 0) {
 		found = 0
 		for (type in table) {
 			#print "match: str=["str"] type="type" regex=/"table[type]"/"
 			if (match(str, table[type], items) > 0) {
 				#print "       len="RLENGTH" item=["items[0]"]"
+				if (type == "line")
+					line++;
 				if (type == "term")
 					type = items[0]
-				if (type != "space") {
+				if (type != "space" && type != "line") {
+					tokens[i]["line"] = line
 					tokens[i]["type"] = type
 					tokens[i]["text"] = items[0]
 					i++
@@ -139,22 +162,24 @@ function json_tokenize(str, tokens,   i, items, table, type, found)
 			}
 		}
 		if (!found) {
-			debug("error tokenizing")
+			debug("line " line ": error tokenizing")
 			return 0
 		}
 	}
-	return i
 
 	#for (i = 0; i < length(tokens); i++)
 	#	printf "%-3s %-5s [%s]\n", i":",
 	#		tokens[i]["type"], tokens[i]["text"]
+
+	return i
 }
 
-function json_parse_value(tokens, i, value, key,   type, text)
+function json_parse_value(tokens, i, value, key,   line, type, text)
 {
 	if (!i    ) i = 0;
 	if (!depth) depth = 0
 	depth++
+	line = tokens[i]["line"]
 	type = tokens[i]["type"]
 	text = tokens[i]["text"]
 	#printf "parse %d: i=%-2d type=%-3s text=%s\n", depth, i, type, text
@@ -164,7 +189,7 @@ function json_parse_value(tokens, i, value, key,   type, text)
 		case "str": i = json_parse_string(tokens, i, value, key); break
 		case "num": i = json_parse_number(tokens, i, value, key); break
 		case "var": i = json_parse_var(tokens,    i, value, key); break
-		default:    debug("error: type="type" text="text); return 0
+		default:    debug("line "line": error type="type" text="text); return 0
 	}
 	depth--
 	return i
@@ -175,18 +200,20 @@ function json_parse_object(tokens, i, value, key,   object, k, v)
 	if (tokens[i++]["text"] != "{")
 		return 0;
 
-	do {
-		delete k
-		delete v
-		if (!(i=json_parse_value(tokens, i, k, 0)))
-			return 0
-		if (tokens[i++]["text"] != ":")
-			return 0
-		if (!(i=json_parse_value(tokens, i, v, 0)))
-			return 0
-		json_copy(object, k[0], v[0]) 
-	} while (tokens[i++]["text"] == ",")
-	i--
+	if (tokens[i]["text"] != "}") {
+		do {
+			delete k
+			delete v
+			if (!(i=json_parse_value(tokens, i, k, 0)))
+				return 0
+			if (tokens[i++]["text"] != ":")
+				return 0
+			if (!(i=json_parse_value(tokens, i, v, 0)))
+				return 0
+			json_copy(object, k[0], v[0]) 
+		} while (tokens[i++]["text"] == ",")
+		i--
+	}
 
 	if (tokens[i++]["text"] != "}")
 		return 0;
@@ -197,17 +224,18 @@ function json_parse_object(tokens, i, value, key,   object, k, v)
 
 function json_parse_array(tokens, i, value, key,   array, k, v)
 {
-	#print "parse_array: .."
 	if (tokens[i++]["text"] != "[")
 		return 0;
 
-	do {
-		delete v
-		if (!(i=json_parse_value(tokens, i, v, 0)))
-			return 0
-		json_copy(array, k++, v[0]) 
-	} while (tokens[i++]["text"] == ",")
-	i--
+	if (tokens[i]["text"] != "]") {
+		do {
+			delete v
+			if (!(i=json_parse_value(tokens, i, v, 0)))
+				return 0
+			json_copy(array, k++, v[0]) 
+		} while (tokens[i++]["text"] == ",")
+		i--
+	}
 
 	if (tokens[i++]["text"] != "]")
 		return 0;
@@ -227,28 +255,38 @@ function json_parse_number(tokens, i, value, key,   text)
 function json_parse_string(tokens, i, value, key,   text)
 {
 	text = tokens[i++]["text"]
-	text = substr(text, 2, length(text)-2)
+	len  = length(text);
+	text = len == 2 ? "" : substr(text, 2, len-2)
+
+	gsub(/\\\\/, "\\", text)
+	gsub(/\\"/,  "\"", text)
+	gsub(/\\b/,  "\b", text)
+	gsub(/\\f/,  "\f", text)
+	gsub(/\\n/,  "\n", text)
+	gsub(/\\r/,  "\r", text)
+	gsub(/\\t/,  "\t", text)
+
 	json_copy(value, key, text)
-	#print "parse_string: " text
+	#print "parse_string: [" text "]"
 	return i
 }
 
-function json_parse_var(tokens, i, value, key,   text)
+function json_parse_var(tokens, i, value, key,   text, null)
 {
-	text = tokens[i++]["text"]
-	json_copy(value, key, text == "true")
+	switch (tokens[i++]["text"]) {
+		case "true":  json_copy(value, key, 1==1); break;
+		case "false": json_copy(value, key, 1==2); break;
+		case "null":  json_copy(value, key, null); break;
+	}
 	#print "parse_var: " text " -> " text == "true"
 	return i
 }
 
 
-# Nice API?
-function json_load(file, var,   line, text, tokens, data, key)
+# String API
+function json_decode(string, var,   tokens, data, key)
 {
-	while ((getline line < file) > 0)
-		text = text line
-	close(file)
-	if (!json_tokenize(text, tokens))
+	if (!json_tokenize(string, tokens))
 		return ""
 	if (!json_parse_value(tokens, 0, data, 0))
 		return ""
@@ -256,6 +294,21 @@ function json_load(file, var,   line, text, tokens, data, key)
 		return data[0]
 	for (key in data[0])
 		json_copy(var, key, data[0][key])
+	return 1
+}
+
+function json_encode(var, pad)
+{
+	return json_write_value(var, pad)
+}
+
+# File API
+function json_load(file, var,   line, string)
+{
+	delete var
+	while ((getline line < file) > 0)
+		string = string line "\n"
+	return json_decode(string, var)
 }
 
 function json_save(file, var,   cmd, tmp)
@@ -263,7 +316,7 @@ function json_save(file, var,   cmd, tmp)
 	cmd = "mktemp " file ".XXX"
 	cmd | getline tmp
 	close(cmd)
-	print json_write_value(var) > tmp
+	print json_encode(var) > tmp
 	close(tmp)
 	system("mv " tmp " " file)
 }
@@ -272,6 +325,7 @@ function json_save(file, var,   cmd, tmp)
 # Test functions
 function json_test_write()
 {
+	print "Testing JSON Write:"
 	num      = 42
 	str      = "hello, world"
 	arr[0]   = "zero"
@@ -286,54 +340,60 @@ function json_test_write()
 	json_copy(mix, "obj",    obj);
 	json_copy(dub, "first",  mix);
 	json_copy(dub, "second", mix);
-	print json_write_value(num)
-	print json_write_value(str)
-	print json_write_value(arr)
-	print json_write_value(obj)
-	print json_write_value(mix)
-	print json_write_value(dub)
+	print json_encode(num)
+	print json_encode(str)
+	print json_encode(arr)
+	print json_encode(obj)
+	print json_encode(mix)
+	print json_encode(dub)
 }
 
 function json_test_read()
 {
-	json_tokenize("[8, \"abc\", 9]", tokens)
-	json_parse_value(tokens, 0, array, 0)
-	json_print write_value(array[0])
-
-	json_tokenize("{\"abc\": 1, \"def\": 2}", tokens)
-	json_parse_value(tokens, 0, array, 0)
-	json_print write_value(array[0])
-
-	json = "{ \"first\":  { \"arr\":    [ \"zero\",         " \
-	       "                              \"one\",          " \
-	       "                              \"two\" ],        " \
-	       "                \"number\": 42,                 " \
-	       "                \"obj\":    { \"A\": \"a!\",    " \
-	       "                              \"B\": \"b!\",    " \
-	       "                              \"C\": \"c!\" },  " \
-	       "                \"str\":    \"hello, world\" }, " \
-	       "  \"second\": { \"arr\":    [ \"zero\",         " \
-	       "                              \"one\",          " \
-	       "                              \"two\" ],        " \
-	       "                \"number\": 42,                 " \
-	       "                \"obj\":    { \"A\": \"a!\",    " \
-	       "                              \"B\": \"b!\",    " \
-	       "                              \"C\": \"c!\" },  " \
-	       "                \"str\":    \"hello, world\" } }"
-
-	json_tokenize(json, tokens)
-	json_parse_value(tokens, 0, array, 0)
-	print json_write_value(array[0])
+	print "Testing JSON Read:"
+
+	json_decode("[8, \"abc\", 9]", array);
+	print json_encode(array)
+
+	json_decode("{\"abc\": 1, \"def\": 2}", array)
+	print json_encode(array)
+
+	json = "{ \"first\":  { \"arr\":    [ \"\",             \n" \
+	       "                              -1,               \n" \
+	       "                              1.2,              \n" \
+	       "                              true,             \n" \
+	       "                              false,            \n" \
+	       "                              null    ],        \n" \
+	       "                \"number\": 42,                 \n" \
+	       "                \"eobj\":   [ ],                \n" \
+	       "                \"earr\":   { },                \n" \
+	       "                \"obj\":    { \"A\": \"a!\",    \n" \
+	       "                              \"B\": \"b!\",    \n" \
+	       "                              \"C\": \"c!\" },  \n" \
+	       "                \"str\":    \"hello, world\" }, \n" \
+	       "  \"second\": { \"arr\":    [ \"zero\",         \n" \
+	       "                              \"one\",          \n" \
+	       "                              \"two\",          \n" \
+	       "                              \"\\\"\t\r\b\" ], \n" \
+	       "                \"number\": 42,                 \n" \
+	       "                \"obj\":    { \"A\": \"a!\",    \n" \
+	       "                              \"B\": \"b!\",    \n" \
+	       "                              \"C\": \"c!\" },  \n" \
+	       "                \"str\":    \"hello, world\" } }\n"
+
+	json_decode(json, array)
+	print json_encode(array)
 }
 
 function json_test_files()
 {
+	print "Testing JSON Files:"
 	print "load: [" json_load("email.txt", mail) "]"
-	print "mail: "  json_write_value(mail, "      ")
+	print "mail: "  json_encode(mail, "      ")
 	mail["andy753421"] = "andy753421@gmail.com"
 	mail["andy"]       = "andy@gmail.com"
 	mail["somebody"]   = "foo@example.com"
-	print "mail: "  json_write_value(mail, "      ")
+	print "mail: "  json_encode(mail, "      ")
 	print "save: [" json_save("email.txt", mail) "]"
 }