Add ui agent (#343)

* add uipath agent

* readme update

mm_agents/uipath/utils.py

@@ -0,0 +1,57 @@
+import json
+import re
+
+from json_minify import json_minify
+from json_repair import repair_json
+
+
+class ValidationException(Exception):
+    def __init__(self, message: str):
+        self.message = message
+
+
+def parse_message_json(message: str) -> dict:
+    message = message.strip()
+    code_block_pattern = r"```json\s*([\s\S]+?)```"
+    code_block_match = re.search(code_block_pattern, message, re.DOTALL)
+
+    if code_block_match:
+        json_str = code_block_match.group(1).strip()
+    else:
+        bracket_pattern = r"\{.*\}"
+        bracket_match = re.search(bracket_pattern, message, re.DOTALL)
+        if not bracket_match:
+            raise ValidationException("Response does not have correct json format")
+        json_str = bracket_match.group(0).strip()
+
+    try:
+        json_str = json_minify(json_str)
+        data = json.loads(json_str)
+    except json.JSONDecodeError:
+        try:
+            json_str = repair_json(json_str)
+            data = json.loads(json_str)
+        except json.JSONDecodeError:
+            raise ValidationException("Response does not have correct json format")
+    return data
+
+
+class GroundingOutput:
+    def __init__(
+        self,
+        description: str,
+        position: tuple[int, int],
+        end_position: tuple[int, int] = None,
+    ):
+        self.description = description
+        self.position = position
+        self.end_position = end_position
+
+
+class GroundingRequest:
+    def __init__(
+        self, description: str, image_base64: str, action_type: str | None = None
+    ):
+        self.description = description
+        self.image_base64 = image_base64
+        self.action_type = action_type