Spaces:
Sleeping
Sleeping
File size: 6,014 Bytes
eb2c608 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 | from __future__ import annotations
import json
import subprocess
import sys
import tempfile
from pathlib import Path
from typing import Any
import pandas as pd
from autodata_agent.core.errors import ExecutionAppError
from autodata_agent.core.schemas import ChartSpec, ExecutionResult
BANNED_TOKENS = (
"import ",
"__import__",
"open(",
"exec(",
"eval(",
"compile(",
"subprocess",
"socket",
"requests",
"httpx",
"urllib",
"pathlib",
"os.",
"sys.",
"shutil",
"pickle",
"globals(",
"locals(",
"__",
)
RUNNER_CODE = r"""
import contextlib
import io
import json
import math
import statistics
import traceback
import duckdb
import numpy as np
import pandas as pd
with open(INPUT_JSON, "r", encoding="utf-8") as handle:
payload = json.load(handle)
df = pd.DataFrame(payload["records"])
code = payload["code"]
allowed_builtins = {
"abs": abs,
"all": all,
"any": any,
"bool": bool,
"dict": dict,
"enumerate": enumerate,
"float": float,
"int": int,
"len": len,
"list": list,
"max": max,
"min": min,
"pow": pow,
"range": range,
"round": round,
"set": set,
"sorted": sorted,
"str": str,
"sum": sum,
"tuple": tuple,
"zip": zip,
}
scope = {
"__builtins__": allowed_builtins,
"df": df,
"pd": pd,
"np": np,
"duckdb": duckdb,
"math": math,
"statistics": statistics,
}
stdout = io.StringIO()
try:
with contextlib.redirect_stdout(stdout):
exec(code, scope, scope)
result_df = scope.get("result_df")
chart_spec = scope.get("chart_spec")
if result_df is None:
raise ValueError("Generated code must define result_df.")
if not isinstance(result_df, pd.DataFrame):
result_df = pd.DataFrame(result_df)
if chart_spec is None:
raise ValueError("Generated code must define chart_spec.")
if not isinstance(chart_spec, dict):
raise ValueError("chart_spec must be a dictionary.")
result_df = result_df.head(200)
output = {
"success": True,
"result_columns": [str(column) for column in result_df.columns],
"result_rows": json.loads(
result_df.where(pd.notnull(result_df), None).to_json(orient="records")
),
"chart_spec": chart_spec,
"stdout": stdout.getvalue(),
}
except Exception:
output = {
"success": False,
"error": traceback.format_exc(limit=8),
"stdout": stdout.getvalue(),
}
with open(OUTPUT_JSON, "w", encoding="utf-8") as handle:
json.dump(output, handle, ensure_ascii=True)
"""
class CodeExecutor:
def __init__(self, timeout_seconds: float) -> None:
self.timeout_seconds = timeout_seconds
def execute(self, df: pd.DataFrame, code: str) -> ExecutionResult:
self._validate_code(code)
records = json.loads(
df.where(pd.notnull(df), None).to_json(orient="records", date_format="iso")
)
with tempfile.TemporaryDirectory(prefix="autodata_exec_") as tmp:
tmp_path = Path(tmp)
input_path = tmp_path / "input.json"
output_path = tmp_path / "output.json"
runner_path = tmp_path / "runner.py"
input_path.write_text(json.dumps({"records": records, "code": code}), encoding="utf-8")
runner_path.write_text(
"INPUT_JSON = " + repr(str(input_path)) + "\n"
"OUTPUT_JSON = " + repr(str(output_path)) + "\n"
+ RUNNER_CODE,
encoding="utf-8",
)
try:
subprocess.run(
[sys.executable, str(runner_path)],
capture_output=True,
text=True,
timeout=self.timeout_seconds,
check=False,
)
except subprocess.TimeoutExpired as exc:
raise ExecutionAppError(
"analysis_timeout",
"Generated analysis code exceeded the execution timeout.",
details={"timeout_seconds": self.timeout_seconds},
) from exc
if not output_path.exists():
raise ExecutionAppError(
"analysis_runner_failed",
"Analysis runner did not return a result.",
)
payload = json.loads(output_path.read_text(encoding="utf-8"))
if payload.get("success"):
chart = ChartSpec.model_validate(payload["chart_spec"])
return ExecutionResult(
success=True,
result_columns=payload.get("result_columns", []),
result_rows=payload.get("result_rows", []),
chart_spec=chart,
stdout=payload.get("stdout", ""),
)
return ExecutionResult(
success=False,
error=payload.get("error", "Unknown execution error."),
stdout=payload.get("stdout", ""),
)
def _validate_code(self, code: str) -> None:
lowered = code.lower()
for token in BANNED_TOKENS:
if token in lowered:
raise ExecutionAppError(
"unsafe_generated_code",
"Generated code contains a blocked operation.",
details={"blocked_token": token.strip()},
)
if len(code) > 20_000:
raise ExecutionAppError(
"generated_code_too_large",
"Generated code is too large to execute safely.",
)
def compact_result_for_prompt(result: ExecutionResult) -> dict[str, Any]:
return {
"success": result.success,
"columns": result.result_columns,
"sample_rows": result.result_rows[:10],
"chart_spec": result.chart_spec.model_dump() if result.chart_spec else None,
"error": result.error,
}
|