Refactor sanitize_ai_text() (#649)

Co-authored-by: UltralyticsAssistant <[email protected]>

actions/review_pr.py

@@ -6,7 +6,15 @@ import json
 import re
 from pathlib import Path
 
-from .utils import ACTIONS_CREDIT, GITHUB_API_URL, MAX_PROMPT_CHARS, Action, get_response, remove_html_comments
+from .utils import (
+    ACTIONS_CREDIT,
+    GITHUB_API_URL,
+    MAX_PROMPT_CHARS,
+    Action,
+    get_response,
+    remove_html_comments,
+    sanitize_ai_text,
+)
 
 REVIEW_MARKER = "## 🔍 PR Review"
 ERROR_MARKER = "⚠️ Review generation encountered an error"
@@ -31,13 +39,6 @@ SKIP_PATTERNS = [
 ]
 
 
-def _sanitize_ai_text(s: str) -> str:
-    """Strip private-use citation tokens like '' and normalize whitespace."""
-    return re.sub(
-        r"[\uE000-\uF8FF]*\bcite[\uE000-\uF8FF]*(turn\d+(?:search|view)\d+|[\w\d]+)[\uE000-\uF8FF]*", "", s or ""
-    )
-
-
 def parse_diff_files(diff_text: str) -> tuple[dict, str]:
     """Parse diff and return file mapping with line numbers AND augmented diff with explicit line numbers.
 
@@ -237,10 +238,10 @@ def generate_pr_review(
         )
 
         # Sanitize leaked tool-citation tokens from model output
-        response["summary"] = _sanitize_ai_text(response.get("summary", ""))
+        response["summary"] = sanitize_ai_text(response.get("summary", ""))
         for c in response.get("comments", []):
             if "message" in c:
-                c["message"] = _sanitize_ai_text(c["message"])
+                c["message"] = sanitize_ai_text(c["message"])
 
         print(json.dumps(response, indent=2))
 

actions/utils/__init__.py

@@ -17,6 +17,7 @@ from .openai_utils import (
     get_pr_summary_guidelines,
     get_pr_summary_prompt,
     get_response,
+    sanitize_ai_text,
 )
 from .version_utils import check_pubdev_version, check_pypi_version
 
@@ -39,5 +40,6 @@ __all__ = (
     "get_pr_summary_prompt",
     "get_response",
     "remove_html_comments",
+    "sanitize_ai_text",
     "ultralytics_actions_info",
 )

actions/utils/openai_utils.py

@@ -4,6 +4,7 @@ from __future__ import annotations
 
 import json
 import os
+import re
 import time
 
 import requests
@@ -33,6 +34,13 @@ SYSTEM_PROMPT_ADDITION = """Guidance:
 """
 
 
+def sanitize_ai_text(s: str) -> str:
+    """Strip private-use citation tokens like '' and normalize whitespace."""
+    return re.sub(
+        r"[\uE000-\uF8FF]*\bcite[\uE000-\uF8FF]*(turn\d+(?:search|view)\d+|[\w\d]+)[\uE000-\uF8FF]*", "", s or ""
+    )
+
+
 def remove_outer_codeblocks(string):
     """Removes outer code block markers and language identifiers from a string while preserving inner content."""
     string = string.strip()