Method getScanFunction

src/binder/binder.cpp:231–287

}

TableFunction Binder::getScanFunction(const FileTypeInfo& typeInfo,
    const FileScanInfo& fileScanInfo) const {
    Function* func = nullptr;
    std::vector<LogicalType> inputTypes;
    inputTypes.push_back(LogicalType::STRING());
    auto catalog = Catalog::Get(*clientContext);
    auto transaction = transaction::Transaction::Get(*clientContext);
    switch (typeInfo.fileType) {
    case FileType::PARQUET: {
        auto entry = catalog->getFunctionEntry(transaction, ParquetScanFunction::name);
        func = BuiltInFunctionsUtils::matchFunction(ParquetScanFunction::name, inputTypes,
            entry->ptrCast<FunctionCatalogEntry>());
    } break;
    case FileType::NPY: {
        auto entry = catalog->getFunctionEntry(transaction, NpyScanFunction::name);
        func = BuiltInFunctionsUtils::matchFunction(NpyScanFunction::name, inputTypes,
            entry->ptrCast<FunctionCatalogEntry>());
    } break;
    case FileType::CSV: {
        bool containCompressedCSV = std::any_of(fileScanInfo.filePaths.begin(),
            fileScanInfo.filePaths.end(), [&](const auto& file) {
                return VirtualFileSystem::GetUnsafe(*clientContext)->isCompressedFile(file);
            });
        auto csvConfig = CSVReaderConfig::construct(fileScanInfo.options);
        // Parallel CSV scanning is only allowed:
        // 1. No newline character inside the csv body.
        // 2. The CSV file to scan is not compressed (because we couldn't perform seek in such
        // case).
        // 3. Not explicitly set by the user to use the serial csv reader.
        auto name = (csvConfig.parallel && !containCompressedCSV) ? ParallelCSVScan::name :
                                                                    SerialCSVScan::name;
        auto entry = catalog->getFunctionEntry(transaction, name);
        func = BuiltInFunctionsUtils::matchFunction(name, inputTypes,
            entry->ptrCast<FunctionCatalogEntry>());
    } break;
    case FileType::UNKNOWN: {
        try {
            auto name = std::format("{}_SCAN", typeInfo.fileTypeStr);
            auto entry = catalog->getFunctionEntry(transaction, name);
            func = BuiltInFunctionsUtils::matchFunction(name, inputTypes,
                entry->ptrCast<FunctionCatalogEntry>());
        } catch (...) {
            if (typeInfo.fileTypeStr == "") {
                throw BinderException{"Cannot infer the format of the given file. Please "
                                      "set the file format explicitly by (file_format=<type>)."};
            }
            throw BinderException{
                std::format("Cannot load from file type {}. If this file type is part of a lbug "
                            "extension please load the extension then try again.",
                    typeInfo.fileTypeStr)};
        }
    } break;
    default:
        UNREACHABLE_CODE;
    }
    return *func->ptrCast<TableFunction>();
}