feat: Add more option to utils to shorten text

phospho-python/phospho/models.py

@@ -12,6 +12,7 @@
 from phospho.utils import (
     generate_timestamp,
     generate_uuid,
+    shorten_text,
 )
 
 # Add other job types here
@@ -483,6 +484,8 @@ def transcript(
         with_previous_messages: bool = False,
         only_previous_messages: bool = False,
         max_previous_messages: Optional[int] = None,
+        message_content_max_len: Optional[int] = None,
+        message_content_shorten_how: Literal["left", "right", "center"] = "left",
     ) -> str:
         """
         Return a string representation of the message.
@@ -508,7 +511,15 @@ def transcript(
             if with_role:
                 transcript += f"{self.role}: {self.content}"
             else:
-                transcript += "\n" + self.content
+                if message_content_max_len is not None:
+                    content = shorten_text(
+                        prompt=self.content,
+                        max_length=message_content_max_len,
+                        how=message_content_shorten_how,
+                    )
+                else:
+                    content = self.content
+                transcript += "\n" + content
         return transcript
 
     def previous_messages_transcript(

phospho-python/phospho/utils.py

@@ -256,10 +256,10 @@ def shorten_text(
     prompt: Optional[str],
     max_length: int,
     margin: int = 20,
-    how: Literal["left", "right"] = "left",
+    how: Literal["left", "right", "center"] = "left",
 ) -> str:
     """
-    Shorten the text to fit in the max_length by only keeping the beginning of the text
+    Shorten the prompt to fit in the max_length by only keeping some part of the text.
     """
     try:
         import tiktoken
@@ -278,7 +278,14 @@ def shorten_text(
     else:
         if how == "left":
             return encoding.decode(tokens[: max_length - margin])
-        if how == "right":
+        elif how == "right":
             return encoding.decode(tokens[-(max_length - margin) :])
+        elif how == "center":
+            # Keep the beginning and the end of the text, with [...] in the middle
+            return (
+                encoding.decode(tokens[: (max_length - margin) // 2])
+                + " [...] "
+                + encoding.decode(tokens[-(max_length - margin) // 2 :])
+            )
         else:
             raise ValueError(f"Unknown value for how: {how}")