From 3ac57b77107e3217a4879f69cb4fe478c8f9c2e8 Mon Sep 17 00:00:00 2001
From: Javier <160494147+javier-cohere@users.noreply.github.com>
Date: Fri, 8 Nov 2024 11:32:20 +0100
Subject: [PATCH] Add parameters for the new pptx parser (#35)

<!-- begin-generated-description -->

This PR introduces a new class, `PresentationParsingStrategy`, which
extends the `StrEnum` class. This class provides two parsing strategies
for presentations: `Unstructured` and `ImageToMarkdown`. The
`Unstructured` strategy is set as the default.

The `ParserConfig` class is also updated to include a new parameter,
`presentation_parsing_strategy`, which is set to the `Unstructured`
strategy by default.

## Changes:
- Added the `PresentationParsingStrategy` class.
- Added the `presentation_parsing_strategy` parameter to the
`ParserConfig` class.

<!-- end-generated-description -->
---
 compass_sdk/__init__.py | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/compass_sdk/__init__.py b/compass_sdk/__init__.py
index d0e0ed6..abcd4bf 100644
--- a/compass_sdk/__init__.py
+++ b/compass_sdk/__init__.py
@@ -266,6 +266,15 @@ def _missing_(cls, value):
         return cls.QuickText
 
 
+class PresentationParsingStrategy(StrEnum):
+    Unstructured = "Unstructured"
+    ImageToMarkdown = "ImageToMarkdown"
+
+    @classmethod
+    def _missing_(cls, value):
+        return cls.Unstructured
+
+
 class ParserConfig(BaseModel):
     """
     CompassParser configuration. Important parameters:
@@ -319,6 +328,7 @@ class ParserConfig(BaseModel):
     horizontal_table_crop_margin: int = 100
 
     pdf_parsing_strategy: PDFParsingStrategy = PDFParsingStrategy.QuickText
+    presentation_parsing_strategy: PresentationParsingStrategy = PresentationParsingStrategy.Unstructured
 
 
 ### Document indexing