Fix Jinja2 template errors from invalid parameter/block names with special characters (SKY-7356) (#4644)

skyvern/utils/strings.py

@@ -1,5 +1,6 @@
 import os
 import random
+import re
 import string
 import uuid
 
@@ -18,3 +19,30 @@ def is_uuid(string: str) -> bool:
         return True
     except ValueError:
         return False
+
+
+def sanitize_identifier(value: str, default: str = "identifier") -> str:
+    """Sanitizes a string to be a valid Python/Jinja2 identifier.
+
+    Replaces non-alphanumeric characters (except underscores) with underscores,
+    collapses consecutive underscores, strips leading/trailing underscores,
+    and prepends an underscore if the result starts with a digit.
+
+    Args:
+        value: The raw value to sanitize.
+        default: Fallback value if everything is stripped.
+
+    Returns:
+        A sanitized string that is a valid Python identifier.
+    """
+    sanitized = re.sub(r"[^a-zA-Z0-9_]", "_", value)
+    sanitized = re.sub(r"_+", "_", sanitized)
+    sanitized = sanitized.strip("_")
+
+    if sanitized and sanitized[0].isdigit():
+        sanitized = "_" + sanitized
+
+    if not sanitized:
+        sanitized = default
+
+    return sanitized

skyvern/utils/templating.py

@@ -7,6 +7,28 @@ class Constants:
     MissingVariablePattern = var_pattern = r"\{\{\s*([a-zA-Z_][a-zA-Z0-9_.\[\]'\"]*)\s*\}\}"
 
 
+def replace_jinja_reference(text: str, old_key: str, new_key: str) -> str:
+    """Replaces jinja-style references in a string.
+
+    Handles patterns like {{oldKey}}, {{oldKey.field}}, {{oldKey | filter}}, {{oldKey[0]}}
+
+    Args:
+        text: The text to search in
+        old_key: The key to replace (without braces)
+        new_key: The new key to use (without braces)
+
+    Returns:
+        The text with references replaced
+    """
+    # Match {{oldKey}} or {{oldKey.something}} or {{oldKey | filter}} or {{oldKey[0]}} etc.
+    # Use negative lookahead to ensure key is not followed by identifier characters,
+    # which prevents matching {{keyOther}} when searching for {{key}}
+    # Capture whitespace after {{ to preserve formatting (e.g., "{{ key }}" stays "{{ newKey }}")
+    escaped_old_key = re.escape(old_key)
+    pattern = rf"\{{\{{(\s*){escaped_old_key}(?![a-zA-Z0-9_])"
+    return re.sub(pattern, rf"{{{{\1{new_key}", text)
+
+
 def get_missing_variables(template_source: str, template_data: dict) -> set[str]:
     # quick check - catch top-level undefineds
     env = Environment(undefined=StrictUndefined)