Config plot libraries #705
Conversation
…lication settings that allows users to define their preferred data visualization library (matplotlib, seaborn, or plotly). With this update, I've eliminated the need for the user to specify in every prompt which library to use, thereby simplifying their interaction with the application and increasing its versatility.
…ferred data visualization library (matplotlib, seaborn, or plotly), simplifying interactions and enhancing the application's versatility.
WalkthroughThe changes introduced in this commit primarily focus on enhancing the data visualization capabilities of the Changes
Poem
TipsChat with CodeRabbit Bot (
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Review Status
Actionable comments generated: 1
Configuration used: CodeRabbit UI
Files selected for processing (9)
- pandasai/assets/prompt_templates/generate_python_code.tmpl (1 hunks)
- pandasai/helpers/viz_library_types/init.py (1 hunks)
- pandasai/helpers/viz_library_types/_viz_library_types.py (1 hunks)
- pandasai/helpers/viz_library_types/base.py (1 hunks)
- pandasai/schemas/df_config.py (2 hunks)
- pandasai/smart_dataframe/init.py (1 hunks)
- pandasai/smart_datalake/init.py (9 hunks)
- tests/prompts/test_generate_python_code_prompt.py (6 hunks)
- tests/test_smartdataframe.py (4 hunks)
Files skipped from review due to trivial changes (4)
- pandasai/assets/prompt_templates/generate_python_code.tmpl
- pandasai/helpers/viz_library_types/base.py
- pandasai/schemas/df_config.py
- pandasai/smart_dataframe/init.py
Additional comments: 22
pandasai/helpers/viz_library_types/__init__.py (1)
- 1-63: The
viz_lib_type_factoryfunction is well implemented. It uses a dictionary to map the string representation of the visualization library to the corresponding class, which is a good practice for a factory function. The function also handles the case where an unknownviz_lib_typeis passed and logs a warning message, which is a good practice for error handling. The function defaults toMatplotlibVizLibraryTypeif noviz_lib_typeis provided or if an unknownviz_lib_typeis passed, which is a reasonable default choice. The function also logs a debug message indicating which visualization library type is going to be used, which can be helpful for debugging. Overall, the function is well designed and implemented.tests/prompts/test_generate_python_code_prompt.py (7)
26-47: The test cases are well defined and cover a variety of scenarios. The use of parameterized tests is a good practice as it allows for easy addition of new test cases in the future.
50-63: The docstring provides a clear explanation of the test cases and the expected behavior. This is good for maintainability and understanding the purpose of the test.
74-78: The variables are set correctly for the test. The use of
set_varis a good practice as it allows for easy modification of the test setup.90-97: > Note: This review was outside of the patch, so it was mapped to the patch with the greatest overlap. Original lines [80-97]
The expected prompt content is well defined and matches the expected output of the function. This is good for ensuring the function behaves as expected.
123-129: The setup for the
test_advanced_reasoning_prompttest is correct. The use of a fake LLM and a SmartDataframe is a good practice for isolating the test from external dependencies.137-140: The variables are set correctly for the test. The use of
set_varis a good practice as it allows for easy modification of the test setup.152-159: > Note: This review was outside of the patch, so it was mapped to the patch with the greatest overlap. Original lines [142-159]
The expected prompt content is well defined and matches the expected output of the function. This is good for ensuring the function behaves as expected.
pandasai/helpers/viz_library_types/_viz_library_types.py (3)
- 1-88: The code looks good overall. However, there's a minor issue with the
template_hintproperty in thePlotlyVizLibraryTypeandSeabornVizLibraryTypeclasses. The hint suggests using the 'agg' backend for rendering plots, which is specific to matplotlib and may not apply to Plotly or Seaborn. Please verify and update the hints accordingly.* 64-70: - @property - def template_hint(self): - return """When a user requests to create a chart, utilize the Python plotly - library to generate high-quality graphics that will be saved - directly to a file. - If you import matplotlib use the 'agg' backend for rendering plots.""" + @property + def template_hint(self): + return """When a user requests to create a chart, utilize the Python plotly + library to generate high-quality graphics that will be saved + directly to a file.""" * 77-83: - @property - def template_hint(self): - return """When a user requests to create a chart, utilize the Python Seaborn - library to generate high-quality graphics that will be saved - directly to a file. - If you import matplotlib use the 'agg' backend for rendering plots.""" + @property + def template_hint(self): + return """When a user requests to create a chart, utilize the Python Seaborn + library to generate high-quality graphics that will be saved + directly to a file."""pandasai/smart_datalake/__init__.py (7)
28-34: The import of
viz_lib_type_factoryis correctly placed in the import section.46-52: The import of
VisualizationLibraryis correctly placed in the import section.70-74: The
_viz_libattribute is correctly initialized toNone.121-129: The
_viz_libattribute is correctly set based on thedata_viz_libraryconfiguration.198-207: The
_load_data_viz_librarymethod is correctly called if thedata_viz_libraryconfiguration is set.358-364: The
viz_lib_type_factoryis correctly used to initializeviz_lib_helper.375-379: The
"viz_library_type"key is correctly added to thedefault_valuesdictionary.tests/test_smartdataframe.py (4)
28-32: The import statements are correctly placed and the imported modules are used in the test cases.
222-226: This comment seems to be a part of a conversation or a prompt for a test case. However, it's not clear how it's used in the test. Please ensure that it's used correctly or remove it if it's not necessary.
288-292: This comment is identical to the one in lines 222-226. If it's not used, consider removing it to avoid redundancy.
1052-1129: The test case
test_run_passing_viz_library_typeis well written. It tests the behavior of theSmartDataframeclass when a visualization library type is passed. It uses parameterized inputs to test different visualization library types and checks if the expected prompt matches the last prompt.
pandasai/smart_datalake/__init__.py
Outdated
|
|
||
| self._llm = llm | ||
|
|
||
| def _load_data_viz_library(self, data_viz_library: str): | ||
| """ | ||
| Load the appropriate instance for viz library type to use. | ||
|
|
||
| Args: | ||
| data_viz_library (enum): TODO | ||
|
|
||
| Raises: | ||
| TODO | ||
| """ | ||
|
|
||
| self._data_viz_library = VisualizationLibrary.DEFAULT.value | ||
| if data_viz_library in (item.value for item in VisualizationLibrary): | ||
| self._data_viz_library = data_viz_library | ||
|
|
||
| def add_middlewares(self, *middlewares: Optional[Middleware]): |
There was a problem hiding this comment.
The _load_data_viz_library method correctly sets the _data_viz_library attribute based on the data_viz_library argument. However, the method's docstring needs to be updated to reflect its functionality and the possible exceptions it might raise.
def _load_data_viz_library(self, data_viz_library: str):
"""
Load the appropriate instance for viz library type to use.
Args:
data_viz_library (str): The name of the visualization library to use.
Raises:
ValueError: If the provided visualization library is not supported.
"""
self._data_viz_library = VisualizationLibrary.DEFAULT.value
if data_viz_library in (item.value for item in VisualizationLibrary):
self._data_viz_library = data_viz_library
* feat(pipeline): Add pipeline to generate synthetic dataframe * chore(pipeline): maintain documentation and other flows * feat(Pipeline): test case for pipeline * feat(cache): adding cache in pipeline context and fix leftovers * chore(pipeline): rename and add dependency * update poetry lock file * refactor: minor changes from the code review Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com> * chore: update pipeline_context.py * chore: use PandasAI logger instead of default one * refactor: prompt for synthetic data now accepts the amount params * remove extra print statement * 'Refactored by Sourcery' (#703) Co-authored-by: Sourcery AI <> * chore(pipeline): improve pipeline usage remove passing config to pipeline * feat: config plot libraries (#705) * In this commit, I introduced a new configuration parameter in our application settings that allows users to define their preferred data visualization library (matplotlib, seaborn, or plotly). With this update, I've eliminated the need for the user to specify in every prompt which library to use, thereby simplifying their interaction with the application and increasing its versatility. * This commit adds a configuration parameter for users to set their preferred data visualization library (matplotlib, seaborn, or plotly), simplifying interactions and enhancing the application's versatility. * viz_library_type' in test_generate_python_code_prompt.py, resolved failing tests --------- Co-authored-by: sabatino.severino <qrxqfspfibrth6nxywai2qifza6jmskt222howzew43risnx4kva> Co-authored-by: Gabriele Venturi <[email protected]> * build: use ruff for formatting * feat: add add_message method to the agent * Release v1.4.3 * feat: workspace env (#717) * fix(chart): charts to save to save_chart_path * refactor sourcery changes * 'Refactored by Sourcery' * refactor chart save code * fix: minor leftovers * feat(workspace_env): add workspace env to store cache, temp chart and config * add error handling and comments --------- Co-authored-by: Sourcery AI <> --------- Co-authored-by: Gabriele Venturi <[email protected]> Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com> Co-authored-by: sourcery-ai[bot] <58596630+sourcery-ai[bot]@users.noreply.github.com> Co-authored-by: Sab Severino <[email protected]>
* feat: config plot libraries (#705) * In this commit, I introduced a new configuration parameter in our application settings that allows users to define their preferred data visualization library (matplotlib, seaborn, or plotly). With this update, I've eliminated the need for the user to specify in every prompt which library to use, thereby simplifying their interaction with the application and increasing its versatility. * This commit adds a configuration parameter for users to set their preferred data visualization library (matplotlib, seaborn, or plotly), simplifying interactions and enhancing the application's versatility. * viz_library_type' in test_generate_python_code_prompt.py, resolved failing tests --------- Co-authored-by: sabatino.severino <qrxqfspfibrth6nxywai2qifza6jmskt222howzew43risnx4kva> Co-authored-by: Gabriele Venturi <[email protected]> * build: use ruff for formatting * feat: add add_message method to the agent * Release v1.4.3 * feat: workspace env (#717) * fix(chart): charts to save to save_chart_path * refactor sourcery changes * 'Refactored by Sourcery' * refactor chart save code * fix: minor leftovers * feat(workspace_env): add workspace env to store cache, temp chart and config * add error handling and comments --------- Co-authored-by: Sourcery AI <> * fix: hallucinations was plotting when not asked * Release v1.4.4 * feat(sqlConnector): add direct config run sql at runtime * feat(DirectSqlConnector): add sql test cases * fix: minor leftovers * fix(orders): check examples of different tables * 'Refactored by Sourcery' * chore(sqlprompt): add description only when we have it --------- Co-authored-by: Sab Severino <[email protected]> Co-authored-by: Gabriele Venturi <[email protected]> Co-authored-by: Sourcery AI <>
Summary by CodeRabbit
New Features:
data_viz_libraryin the configuration, allowing users to specify their preferred visualization library.Refactor:
__len__method in theSmartDataFrameclass for better code readability, without altering its functionality.Tests:
Style:
{viz_library_type}in the prompt templates for better context and information.