| 350 | if (pos < header.size() && header[pos] == '}') { |
| 351 | ++pos; |
| 352 | break; |
| 353 | } |
| 354 | } |
| 355 | |
| 356 | return index; |
| 357 | } |
| 358 | |
| 359 | void write_safetensors_file( |
| 360 | const std::filesystem::path & path, |
| 361 | const std::vector<SafeTensorWriteEntry> & entries) { |
| 362 | if (entries.empty()) { |
| 363 | throw std::runtime_error("safetensors writer requires at least one tensor"); |
| 364 | } |
| 365 | |
| 366 | std::unordered_set<std::string> names; |
| 367 | std::vector<std::pair<size_t, size_t>> offsets; |
| 368 | offsets.reserve(entries.size()); |
| 369 | size_t cursor = 0; |
| 370 | for (const auto & entry : entries) { |
| 371 | if (entry.name.empty() || entry.dtype.empty()) { |
| 372 | throw std::runtime_error("safetensors writer received an unnamed tensor or dtype"); |
| 373 | } |
| 374 | if (!names.insert(entry.name).second) { |
| 375 | throw std::runtime_error("safetensors writer received duplicate tensor: " + entry.name); |
| 376 | } |
| 377 | for (int64_t dim : entry.shape) { |
| 378 | if (dim < 0) { |
| 379 | throw std::runtime_error("safetensors writer received negative shape dimension: " + entry.name); |
| 380 | } |
| 381 | } |
| 382 | offsets.emplace_back(cursor, cursor + entry.data.size()); |
| 383 | cursor += entry.data.size(); |
| 384 | } |
| 385 | |
| 386 | std::ostringstream header_stream; |
| 387 | header_stream << "{"; |
| 388 | for (size_t i = 0; i < entries.size(); ++i) { |
| 389 | if (i != 0) { |
| 390 | header_stream << ","; |
| 391 | } |
| 392 | header_stream << "\"" << escape_json_string(entries[i].name) << "\":{"; |
| 393 | header_stream << "\"dtype\":\"" << escape_json_string(entries[i].dtype) << "\","; |
| 394 | header_stream << "\"shape\":" << shape_to_json(entries[i].shape) << ","; |
| 395 | header_stream << "\"data_offsets\":[" << offsets[i].first << "," << offsets[i].second << "]"; |
| 396 | header_stream << "}"; |
| 397 | } |
| 398 | header_stream << "}"; |
| 399 | |
| 400 | std::string header = header_stream.str(); |
| 401 | header.append((8 - (header.size() % 8)) % 8, ' '); |
| 402 | |
| 403 | const auto parent = path.parent_path(); |
| 404 | if (!parent.empty()) { |
| 405 | std::filesystem::create_directories(parent); |
| 406 | } |
| 407 | std::ofstream output(path, std::ios::binary | std::ios::trunc); |
| 408 | if (!output) { |
| 409 | throw std::runtime_error("failed to open safetensors output file: " + path.string()); |
no test coverage detected