From a626b84b96c7b40904960e039f72ff042148a240 Mon Sep 17 00:00:00 2001
From: joelsprunger <43421397+joelsprunger@users.noreply.github.com>
Date: Tue, 22 Aug 2023 03:05:31 -0700
Subject: [PATCH] feat(backend): adds python code parsing (#1003)

Adds a python code parser and invokes it in processors.py
---
 backend/parsers/code_python.py | 14 ++++++++++++++
 backend/utils/processors.py    |  2 ++
 2 files changed, 16 insertions(+)
 create mode 100644 backend/parsers/code_python.py

diff --git a/backend/parsers/code_python.py b/backend/parsers/code_python.py
new file mode 100644
index 000000000..360601b8b
--- /dev/null
+++ b/backend/parsers/code_python.py
@@ -0,0 +1,14 @@
+from langchain.document_loaders import PythonLoader
+from models import File
+
+from .common import process_file
+
+
+async def process_python(file: File, enable_summarization, brain_id, user_openai_api_key):
+    return await process_file(
+        file=file,
+        loader_class=PythonLoader,
+        enable_summarization=enable_summarization,
+        brain_id=brain_id,
+        user_openai_api_key=user_openai_api_key,
+    )
diff --git a/backend/utils/processors.py b/backend/utils/processors.py
index cae09c105..ab3b51a5d 100644
--- a/backend/utils/processors.py
+++ b/backend/utils/processors.py
@@ -12,6 +12,7 @@ from parsers.pdf import process_pdf
 from parsers.powerpoint import process_powerpoint
 from parsers.txt import process_txt
 from parsers.xlsx import process_xlsx
+from parsers.code_python import process_python
 
 file_processors = {
     ".txt": process_txt,
@@ -34,6 +35,7 @@ file_processors = {
     ".xls": process_xlsx,
     ".epub": process_epub,
     ".ipynb": process_ipnyb,
+    ".py": process_python,
 }