Add asset_type field to the assets structures (#60)

This PR adds the `asset_type` field to the assets to make it possible to identify what is contained in the assets.

cohere/compass/models/documents.py

@@ -22,7 +22,20 @@ class CompassDocumentMetadata(ValidatedModel):
     parent_document_id: str = ""
 
 
+class AssetType(str, Enum):
+    def __str__(self) -> str:
+        return self.value
+
+    # A page that has been rendered as an image
+    PAGE_IMAGE = "page_image"
+    # A Markdown representation of a page's content
+    PAGE_MARKDOWN = "page_markdown"
+    # A dump of the text extracted from a document
+    DOCUMENT_TEXT = "document_text"
+
+
 class CompassDocumentChunkAsset(BaseModel):
+    asset_type: AssetType
     content_type: str
     asset_data: str
 

cohere/compass/models/search.py

@@ -5,8 +5,11 @@
 # 3rd party imports
 from pydantic import BaseModel
 
+from cohere.compass.models.documents import AssetType
+
 
 class AssetInfo(BaseModel):
+    asset_type: AssetType
     content_type: str
     presigned_url: str