Add a C++ backend for LibertyXDR
Also change the C backend so that it also de/serializes unions without any other fields besides the tag.
This commit is contained in:
@@ -222,7 +222,7 @@ function codegen_struct(name, cg, ctype, funcname) {
|
||||
delete cg[i]
|
||||
}
|
||||
|
||||
function codegen_union_tag(d, cg) {
|
||||
function codegen_union_tag(name, d, cg) {
|
||||
cg["tagtype"] = d["type"]
|
||||
cg["tagname"] = d["name"]
|
||||
append(cg, "fields", "\t" CodegenCType[d["type"]] " " d["name"] ";\n")
|
||||
@@ -281,7 +281,7 @@ function codegen_union(name, cg, exhaustive, f, ctype, funcname) {
|
||||
|
||||
CodegenDispose[name] = "\t" funcname "(&%s);\n"
|
||||
}
|
||||
if (cg["serialize"]) {
|
||||
{
|
||||
funcname = PrefixLower cameltosnake(name) "_serialize"
|
||||
print ""
|
||||
print "static bool\n" \
|
||||
@@ -299,7 +299,7 @@ function codegen_union(name, cg, exhaustive, f, ctype, funcname) {
|
||||
CodegenSerialize[name] = "\tif (!" funcname "(&%s, w))\n" \
|
||||
"\t\treturn false;\n"
|
||||
}
|
||||
if (cg["deserialize"]) {
|
||||
{
|
||||
funcname = PrefixLower cameltosnake(name) "_deserialize"
|
||||
print ""
|
||||
print "static bool\n" \
|
||||
|
||||
67
tools/lxdrgen-cpp-posix.cpp
Normal file
67
tools/lxdrgen-cpp-posix.cpp
Normal file
@@ -0,0 +1,67 @@
|
||||
// lxdrgen-cpp-posix.cpp: POSIX support code for lxdrgen-cpp.awk.
|
||||
//
|
||||
// Copyright (c) 2023, Přemysl Eric Janouch <p@janouch.name>
|
||||
// SPDX-License-Identifier: 0BSD
|
||||
#include <iconv.h>
|
||||
|
||||
#include <cstdint>
|
||||
#include <string>
|
||||
|
||||
// Various BSD derivatives may have a problem here.
|
||||
// Linux defines __STDC_ISO_10646__, but also supports "WCHAR_T".
|
||||
#ifdef APPLE
|
||||
#define ICONV_WCHAR "UTF-32"
|
||||
#else
|
||||
#define ICONV_WCHAR "WCHAR_T"
|
||||
#endif
|
||||
|
||||
namespace LibertyXDR {
|
||||
|
||||
bool utf8_to_wstring(const uint8_t *utf8, size_t length, std::wstring &wide) {
|
||||
iconv_t conv = iconv_open(ICONV_WCHAR, "UTF-8");
|
||||
if (conv == (iconv_t) -1)
|
||||
return false;
|
||||
|
||||
wchar_t buffer[1024] = {};
|
||||
char *start = (char *) buffer, *out = start, *in = (char *) utf8;
|
||||
size_t available = sizeof buffer;
|
||||
wide.clear();
|
||||
while (iconv(conv, &in, &length, &out, &available) == (size_t) -1) {
|
||||
if (errno != E2BIG) {
|
||||
iconv_close(conv);
|
||||
return false;
|
||||
}
|
||||
|
||||
wide.append(buffer, (out - start) / sizeof *buffer);
|
||||
out = start;
|
||||
available = sizeof buffer;
|
||||
}
|
||||
wide.append(buffer, (out - start) / sizeof *buffer);
|
||||
iconv_close(conv);
|
||||
return true;
|
||||
}
|
||||
|
||||
bool wstring_to_utf8(const std::wstring &wide, std::string &utf8) {
|
||||
iconv_t conv = iconv_open("UTF-8", ICONV_WCHAR);
|
||||
if (conv == (iconv_t) -1)
|
||||
return false;
|
||||
|
||||
char buffer[1024] = {}, *out = buffer, *in = (char *) wide.data();
|
||||
size_t available = sizeof buffer, length = wide.size() * sizeof wide[0];
|
||||
utf8.clear();
|
||||
while (iconv(conv, &in, &length, &out, &available) == (size_t) -1) {
|
||||
if (errno != E2BIG) {
|
||||
iconv_close(conv);
|
||||
return false;
|
||||
}
|
||||
|
||||
utf8.append(buffer, out - buffer);
|
||||
out = buffer;
|
||||
available = sizeof buffer;
|
||||
}
|
||||
utf8.append(buffer, out - buffer);
|
||||
iconv_close(conv);
|
||||
return true;
|
||||
}
|
||||
|
||||
} // namespace LibertyXDR
|
||||
47
tools/lxdrgen-cpp-win32.cpp
Normal file
47
tools/lxdrgen-cpp-win32.cpp
Normal file
@@ -0,0 +1,47 @@
|
||||
// lxdrgen-cpp-win32.cpp: Win32 support code for lxdrgen-cpp.awk.
|
||||
//
|
||||
// Copyright (c) 2023, Přemysl Eric Janouch <p@janouch.name>
|
||||
// SPDX-License-Identifier: 0BSD
|
||||
#include <windows.h>
|
||||
|
||||
#include <climits>
|
||||
#include <cstdint>
|
||||
#include <string>
|
||||
|
||||
namespace LibertyXDR {
|
||||
|
||||
bool utf8_to_wstring(const uint8_t *utf8, size_t length, std::wstring &wide) {
|
||||
wide.clear();
|
||||
if (!length)
|
||||
return true;
|
||||
if (length > INT_MAX)
|
||||
return false;
|
||||
|
||||
int size = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS,
|
||||
(LPCCH) utf8, length, nullptr, 0);
|
||||
if (size <= 0)
|
||||
return false;
|
||||
|
||||
wide.resize(size);
|
||||
return !MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS,
|
||||
(LPCCH) utf8, length, wide.data(), size);
|
||||
}
|
||||
|
||||
bool wstring_to_utf8(const std::wstring &wide, std::string &utf8) {
|
||||
utf8.clear();
|
||||
if (wide.empty())
|
||||
return true;
|
||||
if (wide.size() > INT_MAX)
|
||||
return false;
|
||||
|
||||
int size = WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS,
|
||||
(LPCWCH) wide.data(), wide.size(), nullptr, 0, NULL, NULL);
|
||||
if (size <= 0)
|
||||
return false;
|
||||
|
||||
utf8.resize(size);
|
||||
return !WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS,
|
||||
(LPCWCH) wide.data(), wide.size(), utf8.data(), size, NULL, NULL);
|
||||
}
|
||||
|
||||
} // namespace LibertyXDR
|
||||
350
tools/lxdrgen-cpp.awk
Normal file
350
tools/lxdrgen-cpp.awk
Normal file
@@ -0,0 +1,350 @@
|
||||
# lxdrgen-cpp.awk: C++ backend for lxdrgen.awk.
|
||||
#
|
||||
# This backend is intended for Windows, it just happens to have a fallback
|
||||
# that will probably work on Unices, of which we make use in tests.
|
||||
#
|
||||
# Copyright (c) 2023, Přemysl Eric Janouch <p@janouch.name>
|
||||
# SPDX-License-Identifier: 0BSD
|
||||
|
||||
function define_internal(name, ctype) {
|
||||
Types[name] = "internal"
|
||||
CodegenCType[name] = ctype
|
||||
CodegenSerialize[name] = \
|
||||
"\tw.append(%s);\n"
|
||||
CodegenDeserialize[name] = \
|
||||
"\tif (!r.read(%s))\n" \
|
||||
"\t\treturn false;\n"
|
||||
}
|
||||
|
||||
function define_int(shortname, ctype) {
|
||||
define_internal(shortname, ctype)
|
||||
}
|
||||
|
||||
function define_sint(size) { define_int("i" size, "int" size "_t") }
|
||||
function define_uint(size) { define_int("u" size, "uint" size "_t") }
|
||||
|
||||
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("string", "std::wstring")
|
||||
define_internal("bool", "bool")
|
||||
|
||||
CodegenSerialize["string"] = \
|
||||
"\tif (!w.append(%s))\n" \
|
||||
"\t\treturn false;\n"
|
||||
|
||||
print "// Code generated from " FILENAME ". DO NOT EDIT."
|
||||
print ""
|
||||
print "#include <cstdint>"
|
||||
print "#include <memory>"
|
||||
print "#include <string>"
|
||||
print "#include <vector>"
|
||||
print ""
|
||||
print "namespace LibertyXDR {"
|
||||
print ""
|
||||
print "bool utf8_to_wstring("
|
||||
print "\tconst uint8_t *utf8, size_t length, std::wstring &wide);"
|
||||
print "bool wstring_to_utf8("
|
||||
print "\tconst std::wstring &wide, std::string &utf8);"
|
||||
print ""
|
||||
print "struct Reader {"
|
||||
print "\tconst uint8_t *data = {};"
|
||||
print "\tsize_t length = {};"
|
||||
print ""
|
||||
print "\ttemplate<typename T> bool read(T &number) {"
|
||||
print "\t\tif (length < sizeof number)"
|
||||
print "\t\t\treturn false;"
|
||||
print ""
|
||||
print "\t\tnumber = 0;"
|
||||
print "\t\tfor (size_t i = 0; i < sizeof number; i++) {"
|
||||
print "\t\t\tnumber = number << 8 | *data++;"
|
||||
print "\t\t\tlength--;"
|
||||
print "\t\t}"
|
||||
print "\t\treturn true;"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tbool read(bool &boolean) {"
|
||||
print "\t\tuint8_t number = 0;"
|
||||
print "\t\tif (!read(number))"
|
||||
print "\t\t\treturn false;"
|
||||
print ""
|
||||
print "\t\tboolean = number != 0;"
|
||||
print "\t\treturn true;"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tbool read(std::wstring &string) {"
|
||||
print "\t\tuint32_t size = 0;"
|
||||
print "\t\tif (!read(size) || size > length)"
|
||||
print "\t\t\treturn false;"
|
||||
print "\t\tif (!utf8_to_wstring(data, size, string))"
|
||||
print "\t\t\treturn false;"
|
||||
print ""
|
||||
print "\t\tdata += size;"
|
||||
print "\t\tlength -= size;"
|
||||
print "\t\treturn true;"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tbool read(std::vector<uint8_t> &vector) {"
|
||||
print "\t\tuint32_t size = 0;"
|
||||
print "\t\tif (!read(size) || size > length)"
|
||||
print "\t\t\treturn false;"
|
||||
print "\t\tvector.assign(data, data + size);"
|
||||
print ""
|
||||
print "\t\tdata += size;"
|
||||
print "\t\tlength -= size;"
|
||||
print "\t\treturn true;"
|
||||
print "\t}"
|
||||
print "};"
|
||||
print ""
|
||||
print "struct Writer {"
|
||||
print "\tstd::vector<uint8_t> data;"
|
||||
print ""
|
||||
print "\ttemplate<typename T> bool append(T number) {"
|
||||
print "\t\tuint8_t buffer[sizeof number], *p = buffer + sizeof buffer;"
|
||||
print "\t\twhile (p != buffer) {"
|
||||
print "\t\t\t*--p = number;"
|
||||
print "\t\t\tnumber >>= 8;"
|
||||
print "\t\t}"
|
||||
print "\t\tdata.insert(data.end(), buffer, buffer + sizeof buffer);"
|
||||
print "\t\treturn true;"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tbool append(int8_t number) {"
|
||||
print "\t\tdata.push_back(number);"
|
||||
print "\t\treturn true;"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tbool append(uint8_t number) {"
|
||||
print "\t\tdata.push_back(number);"
|
||||
print "\t\treturn true;"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tbool append(bool boolean) {"
|
||||
print "\t\treturn append(uint8_t(boolean));"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tbool append(const std::wstring &string) {"
|
||||
print "\t\tif (string.size() > UINT32_MAX)"
|
||||
print "\t\t\treturn false;"
|
||||
print ""
|
||||
print "\t\tstd::string utf8;"
|
||||
print "\t\tif (!wstring_to_utf8(string, utf8))"
|
||||
print "\t\t\treturn false;"
|
||||
print ""
|
||||
print "\t\tappend<uint32_t>(utf8.size());"
|
||||
print "\t\tdata.insert(data.end(), utf8.begin(), utf8.end());"
|
||||
print "\t\treturn true;"
|
||||
print "\t}"
|
||||
print "};"
|
||||
print ""
|
||||
print "} // namespace LibertyXDR"
|
||||
print "namespace " PrefixCamel " {"
|
||||
}
|
||||
|
||||
END {
|
||||
print ""
|
||||
print "} // namespace " PrefixCamel
|
||||
}
|
||||
|
||||
function codegen_constant(name, value) {
|
||||
print ""
|
||||
print "enum { " name " = " value " };"
|
||||
}
|
||||
|
||||
function codegen_enum_value(name, subname, value, cg) {
|
||||
append(cg, "fields", "\t" subname " = " value ",\n")
|
||||
}
|
||||
|
||||
function codegen_enum(name, cg) {
|
||||
print ""
|
||||
print "enum struct " name " : int8_t {"
|
||||
print cg["fields"] "};"
|
||||
|
||||
# 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] = \
|
||||
"\tw.append(static_cast<int8_t>(%s));\n"
|
||||
CodegenDeserialize[name] = \
|
||||
"\t{\n" \
|
||||
"\t\tint8_t v = 0;\n" \
|
||||
"\t\tif (!r.read(v) || !v)\n" \
|
||||
"\t\t\treturn false;\n" \
|
||||
"\t\t%s = static_cast<" name ">(v);\n" \
|
||||
"\t}\n"
|
||||
|
||||
CodegenCType[name] = name
|
||||
for (i in cg)
|
||||
delete cg[i]
|
||||
}
|
||||
|
||||
# Some identifiers do not pose a problem in C, but do in our C++.
|
||||
function codegen_struct_sanitize(name) {
|
||||
if (name ~ /^(serialize|deserialize)_*$/ ||
|
||||
name ~ /^(catch|class|delete|except|finally|friend|new|operator)_*$/ ||
|
||||
name ~ /^(private|protected|public|template|this|throw|try|virtual)_*$/)
|
||||
return name "_"
|
||||
return name
|
||||
}
|
||||
|
||||
function codegen_struct_tag(d, cg, name, f) {
|
||||
name = codegen_struct_sanitize(d["name"])
|
||||
f = "this->" name
|
||||
|
||||
append(cg, "serialize", sprintf(CodegenSerialize[d["type"]], f))
|
||||
# Do not deserialize here, that would be out of order.
|
||||
}
|
||||
|
||||
function codegen_struct_field(d, cg, name, f, serialize, deserialize) {
|
||||
name = codegen_struct_sanitize(d["name"])
|
||||
f = "this->" name
|
||||
|
||||
serialize = CodegenSerialize[d["type"]]
|
||||
deserialize = CodegenDeserialize[d["type"]]
|
||||
if (!d["isarray"]) {
|
||||
append(cg, "fields",
|
||||
"\t" CodegenCType[d["type"]] " " name " = {};\n")
|
||||
append(cg, "serialize", sprintf(serialize, f))
|
||||
append(cg, "deserialize", sprintf(deserialize, f))
|
||||
return
|
||||
}
|
||||
|
||||
append(cg, "fields",
|
||||
"\tstd::vector<" CodegenCType[d["type"]] "> " name ";\n")
|
||||
|
||||
# XXX: We should probably pedantically check for overflows.
|
||||
append(cg, "serialize",
|
||||
sprintf(CodegenSerialize["u32"], "uint32_t(" f ".size())") \
|
||||
"\tfor (const auto &it : " f ")\n" \
|
||||
indent(sprintf(serialize, "it")))
|
||||
|
||||
if (d["type"] == "u8") {
|
||||
append(cg, "deserialize",
|
||||
"\tif (!r.read(" f "))\n" \
|
||||
"\t\treturn false;\n")
|
||||
} else if (deserialize) {
|
||||
append(cg, "deserialize",
|
||||
"\t{\n" \
|
||||
"\t\tuint32_t size = 0;\n" \
|
||||
indent(sprintf(CodegenDeserialize["u32"], "size")) \
|
||||
"\t\t" f ".resize(size);\n" \
|
||||
"\t}\n" \
|
||||
"\tfor (auto &it : " f ")\n" \
|
||||
indent(sprintf(deserialize, "it")))
|
||||
}
|
||||
}
|
||||
|
||||
function codegen_struct(name, cg) {
|
||||
print ""
|
||||
print "struct " name " {"
|
||||
print cg["fields"]
|
||||
print "\tbool serialize(LibertyXDR::Writer &w) const {"
|
||||
print indent(cg["serialize"]) "\t\treturn true;"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tbool deserialize([[maybe_unused]] LibertyXDR::Reader &r) {"
|
||||
print indent(cg["deserialize"]) "\t\treturn true;"
|
||||
print "\t}"
|
||||
print "};"
|
||||
|
||||
CodegenSerialize[name] = "\tif (!%s->serialize(w))\n" \
|
||||
"\t\treturn false;\n"
|
||||
CodegenDeserialize[name] = "\tif (!%s->deserialize(r))\n" \
|
||||
"\t\treturn false;\n"
|
||||
|
||||
CodegenCType[name] = name
|
||||
for (i in cg)
|
||||
delete cg[i]
|
||||
}
|
||||
|
||||
function codegen_union_tag(name, d, cg, tagname) {
|
||||
cg["tagtype"] = d["type"]
|
||||
cg["tagname"] = tagname = codegen_struct_sanitize(d["name"])
|
||||
|
||||
print ""
|
||||
print "struct " name " {"
|
||||
print "\t" CodegenCType[d["type"]] " " tagname " = {};"
|
||||
print "\tvirtual ~" name "() = 0;"
|
||||
print "\tvirtual bool serialize(LibertyXDR::Writer &w) const = 0;"
|
||||
print "\tvirtual bool deserialize(LibertyXDR::Reader &r) = 0;"
|
||||
print "};"
|
||||
print ""
|
||||
print name "::~" name "() {}"
|
||||
}
|
||||
|
||||
function codegen_union_struct(name, casename, cg, scg, structname) {
|
||||
# And thus not all generated structs are present in Types.
|
||||
structname = name "_" snaketocamel(casename)
|
||||
|
||||
print ""
|
||||
print "struct " structname " : virtual public " name " {"
|
||||
print scg["fields"]
|
||||
print "\t" structname "() {"
|
||||
print "\t\tthis->" cg["tagname"] " = " \
|
||||
CodegenCType[cg["tagtype"]] "::" casename ";"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tvirtual bool serialize(LibertyXDR::Writer &w) const {"
|
||||
print indent(scg["serialize"]) "\t\treturn true;"
|
||||
print "\t}"
|
||||
print ""
|
||||
print "\tvirtual bool deserialize([[maybe_unused]] LibertyXDR::Reader &r) {"
|
||||
print indent(scg["deserialize"]) "\t\treturn true;"
|
||||
print "\t}"
|
||||
print "};"
|
||||
|
||||
append(cg, "deserialize",
|
||||
"\tcase " CodegenCType[cg["tagtype"]] "::" casename ":\n" \
|
||||
"\t\treturn new " structname "();\n")
|
||||
|
||||
CodegenSerialize[structname] = "\tif (!%s->serialize(w))\n" \
|
||||
"\t\treturn false;\n"
|
||||
CodegenDeserialize[structname] = "\tif (!%s->deserialize(r))\n" \
|
||||
"\t\treturn false;\n"
|
||||
|
||||
CodegenCType[structname] = structname
|
||||
for (i in scg)
|
||||
delete scg[i]
|
||||
}
|
||||
|
||||
function codegen_union(name, cg, exhaustive, ctype) {
|
||||
CodegenSerialize[name] = "\tif (!%s->serialize(w))\n" \
|
||||
"\t\treturn false;\n"
|
||||
|
||||
ctype = "std::unique_ptr<" name ">"
|
||||
if (cg["deserialize"]) {
|
||||
print ""
|
||||
print "static " name " *read" name "(" \
|
||||
CodegenCType[cg["tagtype"]] " " cg["tagname"] ") {"
|
||||
print "\tswitch (" cg["tagname"] ") {"
|
||||
print cg["deserialize"] "\tdefault:"
|
||||
print "\t\treturn nullptr;"
|
||||
print "\t}"
|
||||
print "}"
|
||||
print ""
|
||||
print "static " ctype " read" name "(LibertyXDR::Reader &r) {"
|
||||
print "\tint8_t v = 0;"
|
||||
print "\tif (!r.read(v) || !v)"
|
||||
print "\t\treturn nullptr;"
|
||||
print ""
|
||||
print "\t" ctype " result(read" name "(static_cast<" \
|
||||
CodegenCType[cg["tagtype"]] ">(v)));"
|
||||
print "\tif (!result || !result->deserialize(r))"
|
||||
print "\t\treturn nullptr;"
|
||||
print "\treturn result;"
|
||||
print "}"
|
||||
|
||||
CodegenDeserialize[name] = "\tif (!(%s = read" name "(r)))\n" \
|
||||
"\t\treturn false;\n"
|
||||
}
|
||||
|
||||
CodegenCType[name] = ctype
|
||||
for (i in cg)
|
||||
delete cg[i]
|
||||
}
|
||||
@@ -439,7 +439,7 @@ function codegen_struct(name, cg, gotype) {
|
||||
delete cg[i]
|
||||
}
|
||||
|
||||
function codegen_union_tag(d, cg) {
|
||||
function codegen_union_tag(name, d, cg) {
|
||||
cg["tagtype"] = d["type"]
|
||||
cg["tagname"] = snaketocamel(d["name"])
|
||||
# The tag is implied from the type of struct stored in the interface.
|
||||
|
||||
@@ -183,7 +183,7 @@ function codegen_struct(name, cg) {
|
||||
delete cg[i]
|
||||
}
|
||||
|
||||
function codegen_union_tag(d, cg) {
|
||||
function codegen_union_tag(name, d, cg) {
|
||||
cg["tagtype"] = d["type"]
|
||||
cg["tagname"] = decapitalize(snaketocamel(d["name"]))
|
||||
}
|
||||
|
||||
@@ -215,7 +215,7 @@ function codegen_struct(name, cg, swifttype) {
|
||||
delete cg[i]
|
||||
}
|
||||
|
||||
function codegen_union_tag(d, cg) {
|
||||
function codegen_union_tag(name, d, cg) {
|
||||
cg["tagtype"] = d["type"]
|
||||
cg["tagname"] = decapitalize(snaketocamel(d["name"]))
|
||||
}
|
||||
|
||||
@@ -232,7 +232,7 @@ function defunion( name, tag, tagtype, tagvalue, cg, scg, d, a, i,
|
||||
|
||||
if (Types[tagtype] != "enum")
|
||||
fatal("not an enum type: " tagtype)
|
||||
codegen_union_tag(tag, cg)
|
||||
codegen_union_tag(name, tag, cg)
|
||||
|
||||
split(EnumValues[tagtype], a, SUBSEP)
|
||||
for (i in a)
|
||||
|
||||
Reference in New Issue
Block a user