add unitxt encode/decode actions

This commit is contained in:
Martin Michelsen
2023-10-21 14:33:39 -07:00
parent 81edf93e3b
commit 53ebd42612
4 changed files with 81 additions and 2 deletions
+1
View File
@@ -105,6 +105,7 @@ add_executable(newserv
src/StaticGameData.cc src/StaticGameData.cc
src/Text.cc src/Text.cc
src/TextArchive.cc src/TextArchive.cc
src/UnicodeTextSet.cc
src/Version.cc src/Version.cc
src/WordSelectTable.cc src/WordSelectTable.cc
) )
+33 -2
View File
@@ -37,6 +37,7 @@
#include "StaticGameData.hh" #include "StaticGameData.hh"
#include "Text.hh" #include "Text.hh"
#include "TextArchive.hh" #include "TextArchive.hh"
#include "UnicodeTextSet.hh"
using namespace std; using namespace std;
@@ -317,6 +318,8 @@ enum class Behavior {
EXTRACT_BML, EXTRACT_BML,
DECODE_TEXT_ARCHIVE, DECODE_TEXT_ARCHIVE,
ENCODE_TEXT_ARCHIVE, ENCODE_TEXT_ARCHIVE,
DECODE_UNICODE_TEXT_SET,
ENCODE_UNICODE_TEXT_SET,
FORMAT_RARE_ITEM_SET, FORMAT_RARE_ITEM_SET,
CONVERT_ITEMRT_REL_TO_JSON, CONVERT_ITEMRT_REL_TO_JSON,
CONVERT_ITEMRT_GSL_TO_JSON, CONVERT_ITEMRT_GSL_TO_JSON,
@@ -375,6 +378,8 @@ static bool behavior_takes_input_filename(Behavior b) {
(b == Behavior::EXTRACT_BML) || (b == Behavior::EXTRACT_BML) ||
(b == Behavior::DECODE_TEXT_ARCHIVE) || (b == Behavior::DECODE_TEXT_ARCHIVE) ||
(b == Behavior::ENCODE_TEXT_ARCHIVE) || (b == Behavior::ENCODE_TEXT_ARCHIVE) ||
(b == Behavior::DECODE_UNICODE_TEXT_SET) ||
(b == Behavior::ENCODE_UNICODE_TEXT_SET) ||
(b == Behavior::DESCRIBE_ITEM) || (b == Behavior::DESCRIBE_ITEM) ||
(b == Behavior::ENCODE_ITEM) || (b == Behavior::ENCODE_ITEM) ||
(b == Behavior::PARSE_OBJECT_GRAPH) || (b == Behavior::PARSE_OBJECT_GRAPH) ||
@@ -415,7 +420,9 @@ static bool behavior_takes_output_filename(Behavior b) {
(b == Behavior::EXTRACT_GSL) || (b == Behavior::EXTRACT_GSL) ||
(b == Behavior::EXTRACT_BML) || (b == Behavior::EXTRACT_BML) ||
(b == Behavior::DECODE_TEXT_ARCHIVE) || (b == Behavior::DECODE_TEXT_ARCHIVE) ||
(b == Behavior::ENCODE_TEXT_ARCHIVE); (b == Behavior::ENCODE_TEXT_ARCHIVE) ||
(b == Behavior::DECODE_UNICODE_TEXT_SET) ||
(b == Behavior::ENCODE_UNICODE_TEXT_SET);
} }
int main(int argc, char** argv) { int main(int argc, char** argv) {
@@ -656,6 +663,10 @@ int main(int argc, char** argv) {
behavior = Behavior::DECODE_TEXT_ARCHIVE; behavior = Behavior::DECODE_TEXT_ARCHIVE;
} else if (!strcmp(argv[x], "encode-text-archive")) { } else if (!strcmp(argv[x], "encode-text-archive")) {
behavior = Behavior::ENCODE_TEXT_ARCHIVE; behavior = Behavior::ENCODE_TEXT_ARCHIVE;
} else if (!strcmp(argv[x], "decode-unicode-text-set")) {
behavior = Behavior::DECODE_UNICODE_TEXT_SET;
} else if (!strcmp(argv[x], "encode-unicode-text-set")) {
behavior = Behavior::ENCODE_UNICODE_TEXT_SET;
} else if (!strcmp(argv[x], "generate-dc-serial-number")) { } else if (!strcmp(argv[x], "generate-dc-serial-number")) {
behavior = Behavior::GENERATE_DC_SERIAL_NUMBER; behavior = Behavior::GENERATE_DC_SERIAL_NUMBER;
} else if (!strcmp(argv[x], "generate-all-dc-serial-numbers")) { } else if (!strcmp(argv[x], "generate-all-dc-serial-numbers")) {
@@ -735,7 +746,7 @@ int main(int argc, char** argv) {
filename += ".bmp"; filename += ".bmp";
} else if (behavior == Behavior::ENCODE_GVM) { } else if (behavior == Behavior::ENCODE_GVM) {
filename += ".gvm"; filename += ".gvm";
} else if (behavior == Behavior::DECODE_TEXT_ARCHIVE) { } else if ((behavior == Behavior::DECODE_TEXT_ARCHIVE) || (behavior == Behavior::DECODE_UNICODE_TEXT_SET)) {
filename += ".json"; filename += ".json";
} else if (behavior == Behavior::DISASSEMBLE_QUEST_SCRIPT) { } else if (behavior == Behavior::DISASSEMBLE_QUEST_SCRIPT) {
filename += ".txt"; filename += ".txt";
@@ -1502,6 +1513,26 @@ int main(int argc, char** argv) {
} }
break; break;
} }
case Behavior::DECODE_UNICODE_TEXT_SET: {
auto strings = parse_unicode_text_set(read_input_data());
JSON j = JSON::list();
for (const u16string& s : strings) {
j.emplace_back(encode_sjis(s));
}
string out_data = j.serialize(JSON::SerializeOption::FORMAT);
write_output_data(out_data.data(), out_data.size());
break;
}
case Behavior::ENCODE_UNICODE_TEXT_SET: {
auto json = JSON::parse(read_input_data());
vector<u16string> strings;
for (const auto& s_json : json.as_list()) {
strings.emplace_back(decode_sjis(s_json->as_string()));
}
string encoded = serialize_unicode_text_set(strings);
write_output_data(encoded.data(), encoded.size());
break;
}
case Behavior::CAT_CLIENT: { case Behavior::CAT_CLIENT: {
shared_ptr<PSOBBEncryption::KeyFile> key; shared_ptr<PSOBBEncryption::KeyFile> key;
+40
View File
@@ -0,0 +1,40 @@
#include "UnicodeTextSet.hh"
#include <phosg/Strings.hh>
#include <string>
#include <vector>
#include "Compression.hh"
using namespace std;
vector<u16string> parse_unicode_text_set(const string& prs_data) {
string data = prs_decompress(prs_data);
StringReader r(data);
r.skip(4);
uint32_t count = r.get_u32l();
vector<u16string> ret;
while (ret.size() < count) {
ret.emplace_back(&r.pget<char16_t>(r.get_u32l()));
}
return ret;
}
string serialize_unicode_text_set(const vector<u16string>& strings) {
StringWriter w;
w.put_u32l(strings.size());
size_t string_offset = (strings.size() * 4) + 4; // Header size
for (const auto& s : strings) {
w.put_u32l(string_offset);
string_offset = (((s.size() + 1) << 1) + 3) & (~3);
}
for (const auto& s : strings) {
u16string uni_s = decode_sjis(s);
w.write(uni_s.c_str(), (uni_s.size() + 1) * 2);
while (w.size() & 3) {
w.put_u8(0);
}
}
return std::move(w.str());
}
+7
View File
@@ -0,0 +1,7 @@
#pragma once
#include <string>
#include <vector>
std::vector<std::u16string> parse_unicode_text_set(const std::string& prs_data);
std::string serialize_unicode_text_set(const std::vector<std::u16string>& strings);