python interpreter refactor to sandbox (#1956)

* python interpreter refactor to sandbox

* add deno to tests workflow

.github/workflows/run_tests.yml

@@ -41,6 +41,17 @@ jobs:
         python-version: ["3.9"]
     steps:
       - uses: actions/checkout@v4
+      - name: Install Deno
+        run: |
+          curl -fsSL https://deno.land/install.sh | sh
+          echo "Deno installed"
+
+      - name: Add Deno to PATH
+        run: echo "${HOME}/.deno/bin" >> $GITHUB_PATH
+
+      - name: Verify Deno installation
+        run: deno --version
+
       - name: Load cached Poetry installation
         id: cached-poetry
         uses: actions/cache@v3

docs/docs/deep-dive/modules/program-of-thought.md

@@ -19,7 +19,7 @@ Program of Thought is instantiated based on a user-defined DSPy Signature, which
 import dsp
 import dspy
 from ..primitives.program import Module
-from ..primitives.python_interpreter import CodePrompt, PythonInterpreter
+from ..primitives.python_interpreter import PythonInterpreter
 import re
 
 class ProgramOfThought(Module):
@@ -58,7 +58,7 @@ Executes the last stored generated code and outputs the final answer, with the s
 - **Code Parsing:**
     Program of Thought internally processes each code generation as a string and filters out extraneous bits to ensure the code block conforms to executable Python syntax. If the code is empty or does not match these guidelines, the parser returns an error string, signaling the PoT process for regeneration.
 - **Code Execution:**
-    Program of Thought relies on a Python interpreter adapted by CAMEL-AI to execute code generated by LLMs. The final code generation is formatted as a CodePrompt instance and executed by the PythonInterpreter. This adaptation is present in [DSPy primitives](https://github.com/stanfordnlp/dspy/blob/main/dspy/primitives/python_interpreter.py).
+    Program of Thought relies on a sandboxed environment Python interpreter using Deno and Pyodide adapted by [this tutorial](https://til.simonwillison.net/deno/pyodide-sandbox). This adaptation is present in [DSPy primitives](https://github.com/stanfordnlp/dspy/blob/main/dspy/primitives/python_interpreter.py).
 
 ## Tying It All Together
 Using ProgramOfThought mirrors the simplicity of the base `Predict` and `ChainOfThought` modules. Here is an example call:

dspy/predict/program_of_thought.py

@@ -4,15 +4,14 @@
 from dspy.signatures.signature import ensure_signature
 
 from ..primitives.program import Module
-from ..primitives.python_interpreter import CodePrompt, PythonInterpreter
+from ..primitives.python_interpreter import PythonInterpreter
 
 
 class ProgramOfThought(Module):
-    def __init__(self, signature, max_iters=3, import_white_list=None):
+    def __init__(self, signature, max_iters=3):
         super().__init__()
         self.signature = signature = ensure_signature(signature)
         self.max_iters = max_iters
-        self.import_white_list = import_white_list
 
         self.input_fields = signature.input_fields
         self.output_fields = signature.output_fields
@@ -152,10 +151,9 @@ def parse_code(self, code_data):
     def execute_code(self, code):
         if not code:
             return code, None, "Error: Empty code before execution."
-        code_prompt = CodePrompt(code, code_type="python")
-        interpreter = PythonInterpreter(action_space={"print": print}, import_white_list=self.import_white_list)
+        interpreter = PythonInterpreter()
         try:
-            output = str(code_prompt.execute(interpreter=interpreter)[0])
+            output = str(interpreter.execute(code))
             return code, output, None
         except Exception as e:
             return code, None, str(e)