| 1 | #include <Functions/IFunctionImpl.h> |
| 2 | #include <Functions/FunctionFactory.h> |
| 3 | #include <DataTypes/DataTypesNumber.h> |
| 4 | #include <Formats/FormatSettings.h> |
| 5 | #include <Columns/ColumnsNumber.h> |
| 6 | #include <IO/WriteBufferFromString.h> |
| 7 | #include <Common/UTF8Helpers.h> |
| 8 | #include <Common/assert_cast.h> |
| 9 | |
| 10 | |
| 11 | namespace DB |
| 12 | { |
| 13 | |
| 14 | /** visibleWidth(x) - calculates the approximate width when outputting the value in a text form to the console. |
| 15 | * In fact it calculate the number of Unicode code points. |
| 16 | * It does not support zero width and full width characters, combining characters, etc. |
| 17 | */ |
| 18 | class FunctionVisibleWidth : public IFunction |
| 19 | { |
| 20 | public: |
| 21 | static constexpr auto name = "visibleWidth" ; |
| 22 | static FunctionPtr create(const Context &) |
| 23 | { |
| 24 | return std::make_shared<FunctionVisibleWidth>(); |
| 25 | } |
| 26 | |
| 27 | bool useDefaultImplementationForNulls() const override { return false; } |
| 28 | ColumnNumbers getArgumentsThatDontImplyNullableReturnType(size_t /*number_of_arguments*/) const override { return {0}; } |
| 29 | |
| 30 | /// Get the name of the function. |
| 31 | String getName() const override |
| 32 | { |
| 33 | return name; |
| 34 | } |
| 35 | |
| 36 | size_t getNumberOfArguments() const override |
| 37 | { |
| 38 | return 1; |
| 39 | } |
| 40 | |
| 41 | DataTypePtr getReturnTypeImpl(const DataTypes & /*arguments*/) const override |
| 42 | { |
| 43 | return std::make_shared<DataTypeUInt64>(); |
| 44 | } |
| 45 | |
| 46 | bool useDefaultImplementationForConstants() const override { return true; } |
| 47 | |
| 48 | /// Execute the function on the block. |
| 49 | void executeImpl(Block & block, const ColumnNumbers & arguments, size_t result, size_t input_rows_count) override |
| 50 | { |
| 51 | auto & src = block.getByPosition(arguments[0]); |
| 52 | size_t size = input_rows_count; |
| 53 | |
| 54 | auto res_col = ColumnUInt64::create(size); |
| 55 | auto & res_data = assert_cast<ColumnUInt64 &>(*res_col).getData(); |
| 56 | |
| 57 | /// For simplicity reasons, function is implemented by serializing into temporary buffer. |
| 58 | |
| 59 | String tmp; |
| 60 | FormatSettings format_settings; |
| 61 | for (size_t i = 0; i < size; ++i) |
| 62 | { |
| 63 | { |
| 64 | WriteBufferFromString out(tmp); |
| 65 | src.type->serializeAsText(*src.column, i, out, format_settings); |
| 66 | } |
| 67 | |
| 68 | res_data[i] = UTF8::countCodePoints(reinterpret_cast<const UInt8 *>(tmp.data()), tmp.size()); |
| 69 | } |
| 70 | |
| 71 | block.getByPosition(result).column = std::move(res_col); |
| 72 | } |
| 73 | }; |
| 74 | |
| 75 | |
| 76 | void registerFunctionVisibleWidth(FunctionFactory & factory) |
| 77 | { |
| 78 | factory.registerFunction<FunctionVisibleWidth>(); |
| 79 | } |
| 80 | |
| 81 | } |
| 82 | |