Skip to content

Evaporate

DFEvaporateProgram #

Bases: BaseEvaporateProgram[DataFrameRowsOnly]

蒸发 DF 程序。

给定一组字段，从一组节点中提取数据框。 每个节点对应数据框中的一行 - 行中的每个值对应一个字段值。

Source code in llama_index/program/evaporate/base.py 
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
class DFEvaporateProgram(BaseEvaporateProgram[DataFrameRowsOnly]):
    """蒸发 DF 程序。

    给定一组字段，从一组节点中提取数据框。
    每个节点对应数据框中的一行 - 行中的每个值对应一个字段值。"""

    def fit(
        self,
        nodes: List[BaseNode],
        field: str,
        field_context: Optional[Any] = None,
        expected_output: Optional[Any] = None,
        inplace: bool = True,
    ) -> str:
        """给定输入的节点和字段，合成python代码。"""
        fn = self._extractor.extract_fn_from_nodes(nodes, field)
        logger.debug(f"Extracted function: {fn}")
        if inplace:
            self._field_fns[field] = fn
        return fn

    def _inference(
        self, nodes: List[BaseNode], fn_str: str, field_name: str
    ) -> List[Any]:
        """给定输入，调用python代码并返回结果。"""
        results = self._extractor.run_fn_on_nodes(nodes, fn_str, field_name)
        logger.debug(f"Results: {results}")
        return results

    @property
    def output_cls(self) -> Type[DataFrameRowsOnly]:
        """输出类。"""
        return DataFrameRowsOnly

    def __call__(self, *args: Any, **kwds: Any) -> DataFrameRowsOnly:
        """在推断数据上调用蒸发函数。"""
        # TODO: either specify `nodes` or `texts` in kwds
        if "nodes" in kwds:
            nodes = kwds["nodes"]
        elif "texts" in kwds:
            nodes = [TextNode(text=t) for t in kwds["texts"]]
        else:
            raise ValueError("Must provide either `nodes` or `texts`.")

        col_dict = {}
        for field in self._fields:
            col_dict[field] = self._inference(nodes, self._field_fns[field], field)

        df = pd.DataFrame(col_dict, columns=self._fields)

        # convert pd.DataFrame to DataFrameRowsOnly
        df_row_objs = []
        for row_arr in df.values:
            df_row_objs.append(DataFrameRow(row_values=list(row_arr)))
        return DataFrameRowsOnly(rows=df_row_objs)

output_cls property #

output_cls: Type[DataFrameRowsOnly]

输出类。

fit #

fit(
    nodes: List[BaseNode],
    field: str,
    field_context: Optional[Any] = None,
    expected_output: Optional[Any] = None,
    inplace: bool = True,
) -> str

给定输入的节点和字段，合成python代码。

Source code in llama_index/program/evaporate/base.py 
132
133
134
135
136
137
138
139
140
141
142
143
144
145
def fit(
    self,
    nodes: List[BaseNode],
    field: str,
    field_context: Optional[Any] = None,
    expected_output: Optional[Any] = None,
    inplace: bool = True,
) -> str:
    """给定输入的节点和字段，合成python代码。"""
    fn = self._extractor.extract_fn_from_nodes(nodes, field)
    logger.debug(f"Extracted function: {fn}")
    if inplace:
        self._field_fns[field] = fn
    return fn