123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458 |
- //
- // Created by Иван_Архипов on 24.11.2017.
- //
- #include "Subfiles/TextSubFile.h"
- #include "BinaryData.h"
- #include "DatFile.h"
- #include "SubfileData.h"
- #include "EasyLogging++/easylogging++.h"
- #include <codecvt>
- #include <locale>
- std::u16string to_utf16(long long x) {
- std::u16string res;
- while (x > 0) {
- res += char16_t(u'0' + x % 10);
- x /= 10ll;
- }
- std::reverse(res.begin(), res.end());
- return res;
- }
- long long from_utf16(const std::u16string &num) {
- long long res = 0;
- for (auto c : num) {
- res = res * 10ll + (c - u'0');
- }
- return res;
- }
- std::string argumentsFromUtf16(const std::u16string &args) {
- std::string res;
- size_t pointer = 0;
- while (pointer < args.length()) {
- size_t pointer1 = args.find(u'-', pointer);
- if (pointer1 == std::u16string::npos)
- pointer1 = args.length();
- if (!res.empty())
- res += "-";
- res += std::to_string(from_utf16(args.substr(pointer, pointer1 - pointer)));
- pointer = pointer1 + 1;
- }
- return res;
- }
- namespace LOTRO_DAT {
- BinaryData TextSubFile::buffer_ = BinaryData(10 * 1024 * 1024);
- TextSubFile::TextSubFile(DatFile &dat, long long dictionary_offset, long long unknown1,
- long long file_id, long long file_offset, long long file_size,
- long long timestamp, long long version, long long block_size, long long unknown2)
- : SubFile(dat, dictionary_offset, unknown1, file_id, file_offset, file_size,
- timestamp, version, block_size, unknown2) {
- }
- FILE_TYPE TextSubFile::FileType() const {
- return TEXT;
- }
- std::string TextSubFile::Extension() const {
- return std::string(".txt");
- }
- SubfileData TextSubFile::PrepareForExport(const BinaryData &file_data) {
- if (file_data.Empty()) {
- patch_fragments_.clear();
- text_pieces_.clear();
- argument_references_.clear();
- argument_strings_.clear();
- return SubfileData();
- }
- SubfileData result;
- long long offset = 9; // first 4 bytes - file_id, then 4 bytes - unknown, then 1 byte - unknown
- long long text_fragment_num = file_data.ToNumber<1>(offset);
- if ((text_fragment_num & 0x80) != 0) {
- text_fragment_num = (((text_fragment_num ^ 0x80) << 8) | file_data.ToNumber<1>(offset + 1));
- offset += 1;
- }
- offset += 1;
- for (long long i = 0; i < text_fragment_num; i++) {
- long long fragment_id = file_data.ToNumber<8>(offset);
- offset += 8;
- MakePieces(file_data, offset);
- MakeArgumentReferences(file_data, offset);
- MakeArgumentStrings(file_data, offset);
- std::u16string text = u"[";
- for (size_t j = 0; j + 1 < text_pieces_.size(); j++)
- text += text_pieces_[j] + u"<--DO_NOT_TOUCH!-->";
- text += text_pieces_[text_pieces_.size() - 1] + u"]";
- std::u16string arguments;
- for (size_t j = 0; j + 1 < argument_references_.size(); j++)
- arguments += to_utf16(argument_references_[j]) + u"-";
- if (!argument_references_.empty())
- arguments += to_utf16(argument_references_[argument_references_.size() - 1]);
- if (result.text_data.length() > 0)
- result.text_data += u"|||";
- result.text_data += to_utf16(fragment_id) + u":::";
- result.text_data += arguments + u":::";
- result.text_data += text;
- }
- result.options["fid"] = file_id();
- result.options["ext"] = Extension();
- patch_fragments_.clear();
- text_pieces_.clear();
- argument_references_.clear();
- argument_strings_.clear();
- return result;
- }
- BinaryData TextSubFile::MakeForImport(const BinaryData &old_data, const SubfileData &data) {
- LOG(DEBUG) << "Preparing text file " << file_id() << " for import.";
- ParsePatchFragments(data);
- if (file_size() <= 10 + 8) {// File is empty, nothing to do;
- patch_fragments_.clear();
- text_pieces_.clear();
- argument_references_.clear();
- argument_strings_.clear();
- return old_data;
- }
- BinaryData new_data;
- long long offset = 9 + 8; // first 8 bytes - file_info. After them:
- // first 4 bytes - file_id, then 4 bytes - unknown, then 1 byte - unknown
- long long text_fragment_num = old_data.ToNumber<1>(offset);
- if ((text_fragment_num & 0x80) != 0) {
- text_fragment_num = (((text_fragment_num ^ 0x80) << 8) | old_data.ToNumber<1>(offset + 1));
- offset += 1;
- }
- offset += 1;
- new_data = new_data + old_data.CutData(0, offset); // Adding file info
- for (long long i = 0; i < text_fragment_num; i++) {
- long long fragment_id = old_data.ToNumber<8>(offset);
- offset += 8;
- new_data = new_data + old_data.CutData(offset - 8, offset);
- TextFragment id_comp;
- id_comp.fragment_id = fragment_id;
- auto fragment_iterator = std::lower_bound(patch_fragments_.begin(), patch_fragments_.end(), id_comp);
- if (fragment_iterator == patch_fragments_.end()) {
- // Retrieving old pieces
- new_data = new_data + GetPieceData(old_data, offset);
- // Retrieving old references
- new_data = new_data + GetArgumentReferenceData(old_data, offset);
- // Retrieving old ref_strings
- new_data = new_data + GetArgumentStringsData(old_data, offset);
- } else {
- // Making and adding new pieces
- new_data = new_data + BuildPieces(old_data, *fragment_iterator, offset);
- // Making and adding new references
- new_data = new_data + BuildArgumentReferences(old_data, *fragment_iterator, offset);
- // Making and adding new strings
- new_data = new_data + BuildArgumentStrings(old_data, *fragment_iterator, offset);
- }
- }
- new_data = new_data + old_data.CutData(offset); // Adding elapsed file data
- patch_fragments_.clear();
- text_pieces_.clear();
- argument_references_.clear();
- argument_strings_.clear();
- return new_data;
- }
- void TextSubFile::ParsePatchFragments(const SubfileData &data) {
- LOG(DEBUG) << "Started parsing patch fragments";
- size_t pointer = 0;
- while (pointer < data.text_data.length()) {
- // Parsing fragment_id
- size_t pointer1 = data.text_data.find(u":::", pointer);
- if (pointer1 == std::u16string::npos) {
- LOG(ERROR) << "Unable to parse fragment id! Cannot find '...' divider. File_id = " << file_id_;
- return;
- }
- long long fragment_id = from_utf16(data.text_data.substr(pointer, pointer1 - pointer));
- pointer = pointer1 + 3;
- TextFragment fragment;
- fragment.fragment_id = fragment_id;
- // Parsing arguments
- pointer1 = data.text_data.find(u":::", pointer);
- if (pointer1 == std::u16string::npos) {
- LOG(ERROR) << "Unable to parse arguments! Cannot find '...' divider. File_id = " << file_id_;
- return;
- }
- std::u16string arguments = data.text_data.substr(pointer, pointer1 - pointer);
- pointer = pointer1 + 3;
- if (arguments.length() > 0) {
- fragment.args = argumentsFromUtf16(arguments);
- }
- // Parsing text
- pointer1 = data.text_data.find(u"|||", pointer);
- if (pointer1 == std::u16string::npos)
- pointer1 = data.text_data.length();
- fragment.text = data.text_data.substr(pointer, pointer1 - pointer);
- pointer = pointer1 + 3;
- patch_fragments_.push_back(fragment);
- }
- std::sort(patch_fragments_.begin(), patch_fragments_.end());
- LOG(DEBUG) << "Finished parsing text patch fragments";
- }
- // Make pieces/arguments/argument strings functions
- void TextSubFile::MakePieces(const BinaryData &data, long long &offset) {
- LOG(DEBUG) << "Started making pieces";
- long long num_pieces = data.ToNumber<4>(offset);
- offset += 4;
- text_pieces_.resize(num_pieces);
- for (long long j = 0; j < num_pieces; j++) {
- long long piece_size = data.ToNumber<1>(offset);
- if ((piece_size & 128) != 0) {
- piece_size = (((piece_size ^ 128) << 8) | data.ToNumber<1>(offset + 1));
- offset += 1;
- }
- offset += 1;
- BinaryData piece_data = data.CutData(offset, offset + piece_size * 2);
- std::u16string piece;
- for (long long k = 0; k < piece_size; k++) {
- char16_t c = char16_t(
- ((short(piece_data[2 * unsigned(k) + 1])) << 8) | (short(piece_data[2 * unsigned(k)])));
- piece += c;
- }
- text_pieces_[j] = piece;
- offset += piece_size * 2;
- }
- LOG(DEBUG) << "Finished making pieces";
- }
- void TextSubFile::MakeArgumentReferences(const BinaryData &data, long long &offset) {
- LOG(DEBUG) << "Started making argument references";
- long long num_references = data.ToNumber<4>(offset);
- offset += 4;
- argument_references_.resize(num_references);
- for (long long j = 0; j < num_references; j++) {
- argument_references_[j] = data.ToNumber<4>(offset);
- offset += 4;
- }
- LOG(DEBUG) << "Finished making argument references";
- }
- void TextSubFile::MakeArgumentStrings(const BinaryData &data, long long &offset) {
- LOG(DEBUG) << "Started making argument strings";
- long long num_arg_strings = data.ToNumber<1>(offset);
- offset += 1;
- argument_strings_.resize(num_arg_strings);
- for (long long j = 0; j < num_arg_strings; j++) {
- long long num_args = data.ToNumber<4>(offset);
- offset += 4;
- argument_strings_[j].resize(num_args);
- for (long long k = 0; k < num_args; k++) {
- long long string_size = data.ToNumber<1>(offset);
- if ((string_size & 0x80) != 0) {
- string_size = (((string_size ^ 0x80) << 8) | data.ToNumber<1>(offset + 1));
- offset += 1;
- }
- offset += 1;
- argument_strings_[j][k] = data.CutData(offset, offset + string_size * 2);
- offset += string_size * 2;
- }
- }
- LOG(DEBUG) << "Finished making argument strings";
- }
- // Build pieces/arguments/argument strings functions from fragment SubfileData
- BinaryData TextSubFile::BuildPieces(const BinaryData &data, const TextFragment &new_data, long long &offset) {
- LOG(DEBUG) << "Started building pieces";
- // Moving &offset pointer in &data
- GetPieceData(data, offset);
- std::u16string file_data = u" fid:" + to_utf16(file_id_)
- + u" gid:" + to_utf16(new_data.fragment_id);
- // Deleting '[' and ']' brackets
- std::u16string text_data = new_data.text.substr(1, new_data.text.size() - 2) + file_data;
- text_pieces_.clear();
- const std::u16string DNT = u"<--DO_NOT_TOUCH!-->";
- size_t prev = 0;
- size_t next = text_data.find(DNT, prev);
- while (next != std::string::npos) {
- std::u16string piece = text_data.substr(prev, next - prev);
- text_pieces_.push_back(piece);
- prev = next + DNT.length();
- next = text_data.find(DNT, prev);
- }
- text_pieces_.push_back(text_data.substr(prev));
- // Building BinaryData from pieces
- unsigned buffer_offset = 0;
- buffer_.Append(BinaryData::FromNumber<4>(text_pieces_.size()), buffer_offset);
- buffer_offset += 4;
- for (const std::u16string &piece : text_pieces_) {
- long long piece_size = piece.length();
- if (piece_size < 128) {
- buffer_.Append(BinaryData::FromNumber<1>(piece_size), buffer_offset);
- buffer_offset += 1;
- } else {
- buffer_.Append(BinaryData::FromNumberRAW<2>((piece_size | 32768)), buffer_offset);
- buffer_offset += 2;
- }
- for (long long j = 0; j < piece_size; j++) {
- buffer_.Append(BinaryData::FromNumber<2>(short(piece[j])), buffer_offset);
- buffer_offset += 2;
- }
- }
- LOG(DEBUG) << "Pieces built successfully";
- return buffer_.CutData(0, buffer_offset);
- }
- BinaryData TextSubFile::BuildArgumentReferences(const BinaryData &data, const TextFragment &new_data,
- long long &offset) {
- LOG(DEBUG) << "Started building argument refs";
- // Moving &offset pointer in &data
- GetArgumentReferenceData(data, offset);
- // If there are no args - making 4 null-bytes and return;
- if (new_data.args.empty()) {
- BinaryData result = BinaryData::FromNumber<4>(0);
- return result;
- }
- // Parsing arguments from list in options["args"]
- std::string args_list = new_data.args;
- argument_references_.clear();
- size_t prev = 0;
- size_t next = args_list.find('-', prev);
- while (next != std::string::npos) {
- std::string argument = args_list.substr(prev, next - prev);
- argument_references_.push_back(std::stoll(argument));
- prev = next + 1;
- next = args_list.find('-', prev);
- }
- std::string argument = args_list.substr(prev);
- argument_references_.push_back(std::stoll(argument));
- BinaryData result;
- BinaryData temp_data = BinaryData::FromNumber<4>(argument_references_.size());
- result = result + temp_data;
- for (const long long &arg_reference : argument_references_) {
- temp_data = BinaryData::FromNumber<4>(arg_reference);
- result = result + temp_data;
- }
- LOG(DEBUG) << "Argument refs built successfully";
- return result;
- }
- BinaryData TextSubFile::BuildArgumentStrings(const BinaryData &data, const TextFragment &, long long &offset) {
- LOG(DEBUG) << "Started building argument strings";
- LOG(DEBUG) << "Built arg strings successfully";
- return GetArgumentStringsData(data, offset);
- }
- // Get BinaryData contents of pieces/arguments/argument strings
- BinaryData TextSubFile::GetPieceData(const BinaryData &data, long long &offset) const {
- LOG(DEBUG) << "Started getting piece data";
- long long old_offset = offset;
- long long num_pieces = data.ToNumber<4>(offset);
- offset += 4;
- for (long long j = 0; j < num_pieces; j++) {
- long long piece_size = data.ToNumber<1>(offset);
- if ((piece_size & 128) != 0) {
- piece_size = (((piece_size ^ 128) << 8) | data.ToNumber<1>(offset + 1));
- offset += 1;
- }
- offset += 1;
- offset += piece_size * 2;
- }
- LOG(DEBUG) << "Got piece data";
- return data.CutData(old_offset, offset);
- }
- BinaryData TextSubFile::GetArgumentReferenceData(const BinaryData &data, long long &offset) const {
- LOG(DEBUG) << "Started getting arg refs data";
- long long old_offset = offset;
- long long num_references = data.ToNumber<4>(offset);
- offset += 4;
- offset += 4 * num_references;
- LOG(DEBUG) << "Finished getting arg refs data";
- return data.CutData(old_offset, offset);
- }
- BinaryData TextSubFile::GetArgumentStringsData(const BinaryData &data, long long &offset) const {
- LOG(DEBUG) << "Started getting arg strings data";
- long long old_offset = offset;
- long long num_arg_strings = data.ToNumber<1>(offset);
- offset += 1;
- for (long long j = 0; j < num_arg_strings; j++) {
- long long num_args = data.ToNumber<4>(offset);
- offset += 4;
- for (long long k = 0; k < num_args; k++) {
- long long string_size = data.ToNumber<1>(offset);
- if ((string_size & 0x80) != 0) {
- string_size = (((string_size ^ 0x80) << 8) | data.ToNumber<1>(offset + 1));
- offset += 1;
- }
- offset += 1;
- offset += string_size * 2;
- }
- }
- LOG(DEBUG) << "Finished getting arg strings data";
- return data.CutData(old_offset, offset);
- }
- };
|