Coverage for src/ipyvizzu/data/converters/df/converter.py: 100%

1""" 

2This module provides the `DataFrameConverter` abstract class. 

3""" 

4 

5from abc import abstractmethod 

6from typing import Dict, List, Optional 

7import warnings 

8 

9from ipyvizzu.data.converters.converter import ToSeriesListConverter 

10from ipyvizzu.data.converters.df.type_alias import DataFrame 

11from ipyvizzu.data.type_alias import ( 

12 DimensionValue, 

13 MeasureValue, 

14 Series, 

15) 

16 

17 

18class DataFrameConverter(ToSeriesListConverter): 

19 """ 

20 Converts data frame into a list of dictionaries representing series. 

21 Each dictionary contains information about the series `name`, `values` and `type`. 

22 """ 

23 

24 # pylint: disable=too-few-public-methods 

25 

26 def __init__( 

27 self, 

28 default_measure_value: MeasureValue, 

29 default_dimension_value: DimensionValue, 

30 max_rows: int, 

31 units: Optional[Dict[str, str]] = None, 

32 ) -> None: 

33 super().__init__(default_measure_value, default_dimension_value) 

34 self._max_rows = max_rows 

35 self._units = units or {} 

36 

37 def get_series_list(self) -> List[Series]: 

38 """ 

39 Convert the `DataFrame` columns to a list of dictionaries representing series. 

40 

41 Returns: 

42 A list of dictionaries representing series, 

43 where each dictionary has `name`, `values` and `type` keys. 

44 """ 

45 

46 series_list = [] 

47 for name in self._get_columns(): 

48 series_list.append(self._get_series_from_column(name)) 

49 return series_list 

50 

51 def _get_series_from_column(self, column_name: str) -> Series: 

52 values, infer_type = self._convert_to_series_values_and_type(column_name) 

53 unit = self._units.get(column_name, None) 

54 return self._convert_to_series(column_name, values, infer_type, unit) 

55 

56 def _is_max_rows_exceeded(self, row_number: int) -> bool: 

57 if row_number > self._max_rows: 

58 warnings.warn( 

59 "The number of rows of the dataframe exceeds the set `max_rows`, " 

60 f"the dataframe is randomly sampled to the set value ({self._max_rows}).", 

61 UserWarning, 

62 stacklevel=2, 

63 ) 

64 return True 

65 return False 

66 

67 @abstractmethod 

68 def _get_sampled_df(self, df: DataFrame) -> DataFrame: 

69 """ 

70 Returns a sampled data frame for the maximum number of rows. 

71 """ 

72 

73 @abstractmethod 

74 def _get_columns(self) -> List[str]: 

75 """ 

76 Return column names of the data frame. 

77 """