File size: 6,321 Bytes
d7d1d4e 1f6b1ac d7d1d4e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 |
import uuid
import matplotlib.pyplot as plt
from pathlib import Path
from typing import Dict, Any, List, Optional
import pandas as pd
import json
import io
import contextlib
import traceback
from pydantic import BaseModel
class CodeResponse(BaseModel):
"""Container for code-related responses"""
language: str = "python"
code: str
class ChartSpecification(BaseModel):
"""Details about requested charts"""
image_description: str
code: Optional[str] = None
class AnalysisOperation(BaseModel):
"""Container for a single analysis operation with its code and result"""
code: CodeResponse
description: str
class CsvChatResult(BaseModel):
"""Structured response for CSV-related AI interactions"""
response_type: str # Literal["casual", "data_analysis", "visualization", "mixed"]
casual_response: str
analysis_operations: List[AnalysisOperation]
charts: Optional[List[ChartSpecification]] = None
class PythonExecutor:
"""Handles execution of Python code and dummy image generation for CSV analysis"""
def __init__(self, df: pd.DataFrame, charts_folder: str = "generated_charts"):
"""
Initialize the PythonExecutor with a DataFrame
Args:
df (pd.DataFrame): The DataFrame to operate on
charts_folder (str): Folder to save charts in
"""
self.df = df
self.charts_folder = Path(charts_folder)
self.charts_folder.mkdir(exist_ok=True)
def execute_code(self, code: str) -> Dict[str, Any]:
"""
Execute Python code and return the output and any generated plots
Args:
code (str): Python code to execute
Returns:
dict: Dictionary containing execution results and any generated plots
"""
output = ""
error = None
plots = []
# Capture stdout
stdout = io.StringIO()
# Monkey patch plt.show() to save figures
original_show = plt.show
def custom_show():
"""Custom show function that saves plots instead of displaying them"""
for i, fig in enumerate(plt.get_fignums()):
figure = plt.figure(fig)
# Save plot to bytes buffer
buf = io.BytesIO()
figure.savefig(buf, format='png', bbox_inches='tight')
buf.seek(0)
plots.append(buf.read())
plt.close('all')
try:
# Create execution context with common libraries and the DataFrame
exec_globals = {
'pd': pd,
'plt': plt,
'json': json,
'df': self.df, # Include the DataFrame in the execution context
'__builtins__': __builtins__,
}
# Replace plt.show with custom implementation
plt.show = custom_show
# Execute code and capture output
with contextlib.redirect_stdout(stdout):
exec(code, exec_globals)
output = stdout.getvalue()
except Exception as e:
error = {
"message": str(e),
"traceback": traceback.format_exc()
}
finally:
# Restore original plt.show
plt.show = original_show
return {
'output': output,
'error': error,
'plots': plots
}
def save_plot_dummy(self, plot_data: bytes, description: str) -> str:
"""
Save plot to charts folder and return a dummy URL
Args:
plot_data (bytes): Image data in bytes
description (str): Description of the plot
Returns:
str: Dummy URL for the chart
"""
# Generate unique filename
filename = f"chart_{uuid.uuid4().hex}.png"
filepath = self.charts_folder / filename
# Save the plot (even though we're using dummy URLs, we still save it)
with open(filepath, 'wb') as f:
f.write(plot_data)
# Return a dummy URL
return f"https://example.com/charts/{filename}"
def process_response(self, response: CsvChatResult) -> str:
"""
Process the CsvChatResult response and generate formatted output
Args:
response (CsvChatResult): Response from CSV analysis
Returns:
str: Formatted output with results and dummy image URLs
"""
output_parts = []
# Add casual response
output_parts.append(response.casual_response)
# Process analysis operations
for operation in response.analysis_operations:
# Execute the code
result = self.execute_code(operation.code.code)
# Add operation description
output_parts.append(f"\n{operation.description}:")
# Add output or error
if result['error']:
output_parts.append(f"Error: {result['error']['message']}")
else:
output_parts.append(result['output'].strip())
# Process charts if they exist
if response.charts:
output_parts.append("\nVisualizations:")
for chart in response.charts:
if chart.code:
# Execute the chart code
result = self.execute_code(chart.code)
if result['plots']:
# Save each generated plot and get dummy URL
for plot_data in result['plots']:
dummy_url = self.save_plot_dummy(plot_data, chart.image_description)
output_parts.append(f"\n{chart.image_description}")
output_parts.append(f"")
elif result['error']:
output_parts.append(f"\nError generating {chart.image_description}: {result['error']['message']}")
return "\n".join(output_parts) |