From fd8619a74cc4f82ed519adf4a78be8651bbddb9b Mon Sep 17 00:00:00 2001
From: Himanshi-Mirosoft <v-agrawalh@microsoft.com>
Date: Wed, 24 Jul 2024 19:39:17 +0530
Subject: [PATCH] fix: Type HTM upload issue fix on explore page under admin 
 (#1172)

Co-authored-by: Himanshi Agrawal <v-himagrawal@microsoft.com>
Co-authored-by: Roopan P M <v-roopanpm@microsoft.com>
---
 .../batch/utilities/helpers/config/config_helper.py   |  1 +
 .../batch/utilities/helpers/config/default.json       | 11 +++++++++++
 code/tests/utilities/helpers/test_config_helper.py    |  9 +++++++--
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/code/backend/batch/utilities/helpers/config/config_helper.py b/code/backend/batch/utilities/helpers/config/config_helper.py
index f6ae638f7..7ad35bbe2 100644
--- a/code/backend/batch/utilities/helpers/config/config_helper.py
+++ b/code/backend/batch/utilities/helpers/config/config_helper.py
@@ -63,6 +63,7 @@ def get_available_document_types(self) -> list[str]:
             "pdf",
             "url",
             "html",
+            "htm",
             "md",
             "jpeg",
             "jpg",
diff --git a/code/backend/batch/utilities/helpers/config/default.json b/code/backend/batch/utilities/helpers/config/default.json
index c3100dff8..dfdbd2201 100644
--- a/code/backend/batch/utilities/helpers/config/default.json
+++ b/code/backend/batch/utilities/helpers/config/default.json
@@ -74,6 +74,17 @@
         "strategy": "web"
       }
     },
+    {
+      "document_type": "htm",
+      "chunking": {
+        "strategy": "layout",
+        "size": 500,
+        "overlap": 100
+      },
+      "loading": {
+        "strategy": "web"
+      }
+    },
     {
       "document_type": "docx",
       "chunking": {
diff --git a/code/tests/utilities/helpers/test_config_helper.py b/code/tests/utilities/helpers/test_config_helper.py
index c5363382e..8ddd6bb03 100644
--- a/code/tests/utilities/helpers/test_config_helper.py
+++ b/code/tests/utilities/helpers/test_config_helper.py
@@ -211,6 +211,11 @@ def test_default_config_when_use_advanced_image_processing(env_helper_mock):
             "chunking": expected_chunking,
             "loading": {"strategy": "web"},
         },
+        {
+            "document_type": "htm",
+            "chunking": expected_chunking,
+            "loading": {"strategy": "web"},
+        },
         {
             "document_type": "docx",
             "chunking": expected_chunking,
@@ -409,7 +414,7 @@ def test_get_available_document_types(config: Config):
 
     # then
     assert sorted(document_types) == sorted(
-        ["txt", "pdf", "url", "html", "md", "jpeg", "jpg", "png", "docx"]
+        ["txt", "pdf", "url", "html", "htm", "md", "jpeg", "jpg", "png", "docx"]
     )
 
 
@@ -424,7 +429,7 @@ def test_get_available_document_types_when_advanced_image_processing_enabled(
 
     # then
     assert sorted(document_types) == sorted(
-        ["txt", "pdf", "url", "html", "md", "jpeg", "jpg", "png", "docx", "tiff", "bmp"]
+        ["txt", "pdf", "url", "html", "htm", "md", "jpeg", "jpg", "png", "docx", "tiff", "bmp"]
     )