xK/xC-gen-proto-go.awk

# xC-gen-proto-go.awk: Go backend for xC-gen-proto.awk.
#
# Copyright (c) 2022, Přemysl Eric Janouch <p@janouch.name>
# SPDX-License-Identifier: 0BSD
#
# This backend also enables proxying to other endpoints using JSON.

function define_internal(name, gotype) {
	Types[name] = "internal"
	CodegenGoType[name] = gotype
}

function define_sint(size,    shortname, gotype) {
	shortname = "i" size
	gotype = "int" size
	define_internal(shortname, gotype)

	if (size == 8) {
		CodegenSerialize[shortname] = "\tdata = append(data, uint8(%s))\n"
		CodegenDeserialize[shortname] = \
			"\tif len(data) >= 1 {\n" \
			"\t\t%s, data = int8(data[0]), data[1:]\n" \
			"\t} else {\n" \
			"\t\treturn nil, false\n" \
			"\t}\n"
		return
	}

	CodegenSerialize[shortname] = \
		"\tdata = binary.BigEndian.AppendUint" size "(data, uint" size "(%s))\n"
	CodegenDeserialize[shortname] = \
		"\tif len(data) >= " (size / 8) " {\n" \
		"\t\t%s = " gotype "(binary.BigEndian.Uint" size "(data))\n" \
		"\t\tdata = data[" (size / 8) ":]\n" \
		"\t} else {\n" \
		"\t\treturn nil, false\n" \
		"\t}\n"
}

function define_uint(size,    shortname, gotype) {
	shortname = "u" size
	gotype = "uint" size
	define_internal(shortname, gotype)

	# Both byte and uint8 luckily marshal as base64-encoded JSON strings.
	if (size == 8) {
		CodegenSerialize[shortname] = "\tdata = append(data, %s)\n"
		CodegenDeserialize[shortname] = \
			"\tif len(data) >= 1 {\n" \
			"\t\t%s, data = data[0], data[1:]\n" \
			"\t} else {\n" \
			"\t\treturn nil, false\n" \
			"\t}\n"
		return
	}

	CodegenSerialize[shortname] = \
		"\tdata = binary.BigEndian.AppendUint" size "(data, %s)\n"
	CodegenDeserialize[shortname] = \
		"\tif len(data) >= " (size / 8) " {\n" \
		"\t\t%s = binary.BigEndian.Uint" size "(data)\n" \
		"\t\tdata = data[" (size / 8) ":]\n" \
		"\t} else {\n" \
		"\t\treturn nil, false\n" \
		"\t}\n"
}

function codegen_begin() {
	define_sint("8")
	define_sint("16")
	define_sint("32")
	define_sint("64")
	define_uint("8")
	define_uint("16")
	define_uint("32")
	define_uint("64")

	define_internal("bool", "bool")
	CodegenSerialize["bool"] = \
		"\tif %s {\n" \
		"\t\tdata = append(data, 1)\n" \
		"\t} else {\n" \
		"\t\tdata = append(data, 0)\n" \
		"\t}\n"
	CodegenDeserialize["bool"] = \
		"\tif data, ok = protoConsumeBoolFrom(data, &%s); !ok {\n" \
		"\t\treturn nil, ok\n" \
		"\t}\n"

	define_internal("string", "string")
	CodegenSerialize["string"] = \
		"\tif data, ok = protoAppendStringTo(data, %s); !ok {\n" \
		"\t\treturn nil, ok\n" \
		"\t}\n"
	CodegenDeserialize["string"] = \
		"\tif data, ok = protoConsumeStringFrom(data, &%s); !ok {\n" \
		"\t\treturn nil, ok\n" \
		"\t}\n"

	print "package main"
	print ""
	print "import ("
	print "\t`encoding/binary`"
	print "\t`encoding/json`"
	print "\t`errors`"
	print "\t`math`"
	print "\t`strconv`"
	print "\t`unicode/utf8`"
	print ")"
	print ""

	print "// protoConsumeBoolFrom tries to deserialize a boolean value"
	print "// from the beginning of a byte stream. When successful,"
	print "// it returns a subslice with any data that might follow."
	print "func protoConsumeBoolFrom(data []byte, b *bool) ([]byte, bool) {"
	print "\tif len(data) < 1 {"
	print "\t\treturn nil, false"
	print "\t}"
	print "\tif data[0] != 0 {"
	print "\t\t*b = true"
	print "\t} else {"
	print "\t\t*b = false"
	print "\t}"
	print "\treturn data[1:], true"
	print "}"
	print ""

	print "// protoAppendStringTo tries to serialize a string value,"
	print "// appending it to the end of a byte stream."
	print "func protoAppendStringTo(data []byte, s string) ([]byte, bool) {"
	print "\tif len(s) > math.MaxUint32 {"
	print "\t\treturn nil, false"
	print "\t}"
	print "\tdata = binary.BigEndian.AppendUint32(data, uint32(len(s)))"
	print "\treturn append(data, s...), true"
	print "}"
	print ""

	print "// protoConsumeStringFrom tries to deserialize a string value"
	print "// from the beginning of a byte stream. When successful,"
	print "// it returns a subslice with any data that might follow."
	print "func protoConsumeStringFrom(data []byte, s *string) ([]byte, bool) {"
	print "\tif len(data) < 4 {"
	print "\t\treturn nil, false"
	print "\t}"
	print "\tlength := binary.BigEndian.Uint32(data)"
	print "\tif data = data[4:]; uint64(len(data)) < uint64(length) {"
	print "\t\treturn nil, false"
	print "\t}"
	print "\t*s = string(data[:length])"
	print "\tif !utf8.ValidString(*s) {"
	print "\t\treturn nil, false"
	print "\t}"
	print "\treturn data[length:], true"
	print "}"
	print ""

	print "// protoUnmarshalEnumJSON converts a JSON fragment to an integer,"
	print "// ensuring that it's within the expected range of enum values."
	print "func protoUnmarshalEnumJSON(data []byte) (int64, error) {"
	print "\tvar n int64"
	print "\tif err := json.Unmarshal(data, &n); err != nil {"
	print "\t\treturn 0, err"
	print "\t} else if n > math.MaxInt8 || n < math.MinInt8 {"
	print "\t\treturn 0, errors.New(`integer out of range`)"
	print "\t} else {"
	print "\t\treturn n, nil"
	print "\t}"
	print "}"
	print ""
}

function codegen_constant(name, value) {
	print "const " PrefixCamel snaketocamel(name) " = " value
	print ""
}

function codegen_enum_value(name, subname, value, cg,    goname) {
	goname = PrefixCamel name snaketocamel(subname)
	append(cg, "fields",
		"\t" goname " = " value "\n")
	append(cg, "stringer",
		"\tcase " goname ":\n" \
		"\t\treturn `" snaketocamel(subname) "`\n")
	append(cg, "marshal",
		goname ",\n")
	append(cg, "unmarshal",
		"\tcase `" snaketocamel(subname) "`:\n" \
		"\t\t*v = " goname "\n")
}

function codegen_enum(name, cg,    gotype, fields) {
	gotype = PrefixCamel name
	print "type " gotype " int8"
	print ""

	print "const ("
	print cg["fields"] ")"
	print ""

	print "func (v " gotype ") String() string {"
	print "\tswitch v {"
	print cg["stringer"] "\tdefault:"
	print "\t\treturn strconv.Itoa(int(v))"
	print "\t}"
	print "}"
	print ""

	CodegenIsMarshaler[name] = 1
	fields = cg["marshal"]
	sub(/,\n$/, ":", fields)
	gsub(/\n/, "\n\t", fields)
	print "func (v " gotype ") MarshalJSON() ([]byte, error) {"
	print "\tswitch v {"
	print indent("case " fields)
	print "\t\treturn []byte(`\"` + v.String() + `\"`), nil"
	print "\t}"
	print "\treturn json.Marshal(int(v))"
	print "}"
	print ""

	print "func (v *" gotype ") UnmarshalJSON(data []byte) error {"
	print "\tvar s string"
	print "\tif json.Unmarshal(data, &s) == nil {"
	print "\t\t// Handled below."
	print "\t} else if n, err := protoUnmarshalEnumJSON(data); err != nil {"
	print "\t\treturn err"
	print "\t} else {"
	print "\t\t*v = " gotype "(n)"
	print "\t\treturn nil"
	print "\t}"
	print ""
	print "\tswitch s {"
	print cg["unmarshal"] "\tdefault:"
	print "\t\treturn errors.New(`unrecognized value: ` + s)"
	print "\t}"
	print "\treturn nil"
	print "}"
	print ""

	# XXX: This should also check if it isn't out-of-range for any reason,
	# but our usage of sprintf() stands in the way a bit.
	CodegenSerialize[name] = "\tdata = append(data, uint8(%s))\n"
	CodegenDeserialize[name] = \
		"\tif len(data) >= 1 {\n" \
		"\t\t%s, data = " gotype "(data[0]), data[1:]\n" \
		"\t} else {\n" \
		"\t\treturn nil, false\n" \
		"\t}\n"

	CodegenGoType[name] = gotype
	for (i in cg)
		delete cg[i]
}

function codegen_struct_field_marshal(d, cg,    camel, f, marshal) {
	camel = snaketocamel(d["name"])
	f = "s." camel

	# Complex types are json.Marshalers, there's no need to json.Marshal(&f).
	if (!d["isarray"]) {
		if (CodegenIsMarshaler[d["type"]])
			marshal = f ".MarshalJSON()"
		else
			marshal = "json.Marshal(" f ")"

		append(cg, "marshal",
			"\tb = append(b, `,\"" decapitalize(camel) "\":`...)\n" \
			"\tif j, err := " marshal "; err != nil {\n" \
			"\t\treturn nil, err\n" \
			"\t} else {\n" \
			"\t\tb = append(b, j...)\n" \
			"\t}\n")
		return
	}

	if (CodegenIsMarshaler[d["type"]])
		marshal = f "[i].MarshalJSON()"
	else
		marshal = "json.Marshal(" f "[i])"

	append(cg, "marshal",
		"\tb = append(b, `,\"" decapitalize(camel) "\":[`...)\n" \
		"\tfor i := 0; i < len(" f "); i++ {\n" \
		"\t\tif i > 0 {\n" \
		"\t\t\tb = append(b, ',')\n" \
		"\t\t}\n" \
		"\t\tif j, err := " marshal "; err != nil {\n" \
		"\t\t\treturn nil, err\n" \
		"\t\t} else {\n" \
		"\t\t\tb = append(b, j...)\n" \
		"\t\t}\n" \
		"\t}\n" \
		"\tb = append(b, ']')\n")
}

function codegen_struct_field(d, cg,    camel, f, serialize, deserialize) {
	codegen_struct_field_marshal(d, cg)

	camel = snaketocamel(d["name"])
	f = "s." camel
	serialize = CodegenSerialize[d["type"]]
	deserialize = CodegenDeserialize[d["type"]]
	if (!d["isarray"]) {
		append(cg, "fields", "\t" camel " " CodegenGoType[d["type"]] \
			" `json:\"" decapitalize(camel) "\"`\n")
		append(cg, "serialize", sprintf(serialize, f))
		append(cg, "deserialize", sprintf(deserialize, f))
		return
	}

	append(cg, "fields", "\t" camel " []" CodegenGoType[d["type"]] \
		" `json:\"" decapitalize(camel) "\"`\n")

	# XXX: This should also check if it isn't out-of-range for any reason.
	append(cg, "serialize",
		sprintf(CodegenSerialize["u32"], "uint32(len(" f "))"))
	if (d["type"] == "u8") {
		append(cg, "serialize",
			"\tdata = append(data, " f "...)\n")
	} else {
		append(cg, "serialize",
			"\tfor i := 0; i < len(" f "); i++ {\n" \
			indent(sprintf(serialize, f "[i]")) \
			"\t}\n")
	}

	append(cg, "deserialize",
		"\t{\n" \
		"\t\tvar length uint32\n" \
		indent(sprintf(CodegenDeserialize["u32"], "length")))
	if (d["type"] == "u8") {
		append(cg, "deserialize",
			"\t\tif uint64(len(data)) < uint64(length) {\n" \
			"\t\t\treturn nil, false\n" \
			"\t\t}\n" \
			"\t\t" f ", data = data[:length], data[length:]\n" \
			"\t}\n")
	} else {
		append(cg, "deserialize",
			"\t\t" f " = make([]" CodegenGoType[d["type"]] ", length)\n" \
			"\t}\n" \
			"\tfor i := 0; i < len(" f "); i++ {\n" \
			indent(sprintf(deserialize, f "[i]")) \
			"\t}\n")
	}
}

function codegen_struct_tag(d, cg,    camel, f) {
	codegen_struct_field_marshal(d, cg)

	camel = snaketocamel(d["name"])
	f = "s." camel
	append(cg, "fields", "\t" camel " " CodegenGoType[d["type"]] \
		" `json:\"" decapitalize(camel) "\"`\n")
	append(cg, "serialize", sprintf(CodegenSerialize[d["type"]], f))
	# Do not deserialize here, that is already done by the containing union.
}

function codegen_struct(name, cg,    gotype) {
	gotype = PrefixCamel name
	print "type " gotype " struct {\n" cg["fields"] "}\n"

	if (cg["marshal"]) {
		CodegenIsMarshaler[name] = 1
		print "func (s *" gotype ") MarshalJSON() ([]byte, error) {"
		print "\tb := []byte{}"
		print cg["marshal"] "\tb[0] = '{'"
		print "\treturn append(b, '}'), nil"
		print "}"
		print ""
	}

	if (cg["serialize"]) {
		print "func (s *" gotype ") AppendTo(data []byte) ([]byte, bool) {"
		print "\tok := true"
		print cg["serialize"] "\treturn data, ok"
		print "}"
		print ""

		CodegenSerialize[name] = \
			"\tif data, ok = %s.AppendTo(data); !ok {\n" \
			"\t\treturn nil, ok\n" \
			"\t}\n"
	}
	if (cg["deserialize"]) {
		print "func (s *" gotype ") ConsumeFrom(data []byte) ([]byte, bool) {"
		print "\tok := true"
		print cg["deserialize"] "\treturn data, ok"
		print "}"
		print ""

		CodegenDeserialize[name] = \
			"\tif data, ok = %s.ConsumeFrom(data); !ok {\n" \
			"\t\treturn nil, ok\n" \
			"\t}\n"
	}

	CodegenGoType[name] = gotype
	for (i in cg)
		delete cg[i]
}

function codegen_union_tag(d, cg) {
	cg["tagtype"] = d["type"]
	cg["tagname"] = d["name"]
	# The tag is implied from the type of struct stored in the interface.
}

function codegen_union_struct(name, casename, cg, scg,     structname, init) {
	# And thus not all generated structs are present in Types.
	structname = name snaketocamel(casename)
	codegen_struct(structname, scg)

	init = CodegenGoType[structname] "{" snaketocamel(cg["tagname"]) \
		": " decapitalize(snaketocamel(cg["tagname"])) "}"
	append(cg, "unmarshal",
		"\tcase " CodegenGoType[cg["tagtype"]] snaketocamel(casename) ":\n" \
		"\t\ts := " init "\n" \
		"\t\terr = json.Unmarshal(data, &s)\n" \
		"\t\tu.Interface = &s\n")
	append(cg, "serialize",
		"\tcase *" CodegenGoType[structname] ":\n" \
		indent(sprintf(CodegenSerialize[structname], "union")))
	append(cg, "deserialize",
		"\tcase " CodegenGoType[cg["tagtype"]] snaketocamel(casename) ":\n" \
		"\t\ts := " init "\n" \
		indent(sprintf(CodegenDeserialize[structname], "s")) \
		"\t\tu.Interface = &s\n")
}

function codegen_union(name, cg,    gotype, tagfield, tagvar) {
	gotype = PrefixCamel name
	print "type " gotype " struct {"
	print "\tInterface any"
	print "}"
	print ""

	# This cannot be a pointer method, it wouldn't work recursively.
	CodegenIsMarshaler[name] = 1
	print "func (u " gotype ") MarshalJSON() ([]byte, error) {"
	print "\treturn u.Interface.(json.Marshaler).MarshalJSON()"
	print "}"
	print ""

	tagfield = snaketocamel(cg["tagname"])
	tagvar = decapitalize(tagfield)
	print "func (u *" gotype ") UnmarshalJSON(data []byte) (err error) {"
	print "\tvar t struct {"
	print "\t\t" tagfield " " CodegenGoType[cg["tagtype"]] \
		" `json:\"" tagvar "\"`"
	print "\t}"
	print "\tif err := json.Unmarshal(data, &t); err != nil {"
	print "\t\treturn err"
	print "\t}"
	print ""
	print "\tswitch " tagvar " := t." tagfield "; " tagvar " {"
	print cg["unmarshal"] "\tdefault:"
	print "\t\terr = errors.New(`unsupported value: ` + " tagvar ".String())"
	print "\t}"
	print "\treturn err"
	print "}"
	print ""

	# XXX: Consider changing the interface into an AppendTo/ConsumeFrom one,
	# that would eliminate these type case switches entirely.
	# On the other hand, it would make it possible to send unsuitable structs.
	print "func (u *" gotype ") AppendTo(data []byte) ([]byte, bool) {"
	print "\tok := true"
	print "\tswitch union := u.Interface.(type) {"
	print cg["serialize"] "\tdefault:"
	print "\t\treturn nil, false"
	print "\t}"
	print "\treturn data, ok"
	print "}"
	print ""

	CodegenSerialize[name] = \
		"\tif data, ok = %s.AppendTo(data); !ok {\n" \
		"\t\treturn nil, ok\n" \
		"\t}\n"

	print "func (u *" gotype ") ConsumeFrom(data []byte) ([]byte, bool) {"
	print "\tok := true"
	print "\tvar " tagvar " " CodegenGoType[cg["tagtype"]]
	print sprintf(CodegenDeserialize[cg["tagtype"]], tagvar)
	print "\tswitch " tagvar " {"
	print cg["deserialize"] "\tdefault:"
	print "\t\treturn nil, false"
	print "\t}"
	print "\treturn data, ok"
	print "}"
	print ""

	CodegenDeserialize[name] = \
		"\tif data, ok = %s.ConsumeFrom(data); !ok {\n" \
		"\t\treturn nil, ok\n" \
		"\t}\n"

	CodegenGoType[name] = gotype
	for (i in cg)
		delete cg[i]
}