Upload folder using huggingface_hub

config.py

@@ -11,6 +11,7 @@ from modules.scripture_helpers.kamba_ramayanam_helper import (
     get_list_of_padalams_in_tamil,
 )
 from modules.scripture_helpers.pancha_sooktham_helper import get_chapter_order_from_pancha_sooktham, get_pancha_sooktham_chapters
+from modules.scripture_helpers.taitriya_upanishad_helper import get_chapter_order_from_taitriya_upanishad, get_taitriya_upanishad_chapters
 from modules.scripture_helpers.vishnu_sahasranamam_helper import (
     get_chapter_order_from_sahasranamam,
     get_chapters_from_sahasranamam,
@@ -840,6 +841,60 @@ class SanatanConfig:
             ],
             "llm_hints": [],
         },
+        {
+            "name": "taitriya_upanishad",
+            "title": "Taitriya Upanishad",
+            "output_dir": "./output/taitriya_upanishad",
+            "collection_name": "taitriya_upanishad",
+            "collection_embedding_fn": "openai",
+            "unit": "verse",
+            "unit_field": "relative_verse_number",
+            "chapter_order": lambda: get_chapter_order_from_taitriya_upanishad(),
+            "audio_storage": "external",
+            "field_mapping": {
+                "text": "sanskrit",
+                "unit_index": "relative_verse_number",
+                "transliteration": lambda doc: f"English: \n{doc.get('english')}\n\nதமிழ்: \n{doc.get('tamil')}\n\nతెలుగు: \n{doc.get('telugu')}\n\nಕನ್ನಡ: \n{doc.get('kannada')}",
+                "chapter_name": "chapter_title",
+                "relative_path": lambda doc: f"{doc.get('chapter_title',"")}-{doc.get('relative_verse_number',"")}",
+            },
+            "metadata_fields": [
+                {
+                    "name": "chapter_title",
+                    "datatype": "str",
+                    "label": "Chapter Name",
+                    "description": "Name of the Chapter",
+                    "show_as_filter": True,
+                    "component": "dropdown",
+                    "lov": lambda: get_taitriya_upanishad_chapters(),
+                    "is_unique": True,
+                },
+                {
+                    "name": "sanskrit",
+                    "label": "Lyrics in sanskrit",
+                    "datatype": "str",
+                    "description": "The original sloka in sanskrit.",
+                },
+                {
+                    "name": "relative_verse_number",
+                    "label": "Relative Verse Number",
+                    "datatype": "int",
+                    "description": "The relative verse number of the sloka inside the chapter.",
+                    "show_as_filter": True,
+                    "is_unique": True,
+                },
+            ],
+            "pdf_path": "./data/taitriya_upanishad.pdf",
+            "source": "https://stotranidhi.com/",
+            "language": "san+eng",
+            "example_labels": [
+                "Seeksha Valli",
+            ],
+            "examples": [
+                "Show some verses from Seeksha Valli",
+            ],
+            "llm_hints": [],
+        },        
     ]
 
     def get_scripture_by_collection(self, collection_name: str):

copy_chromadb.py

@@ -28,6 +28,11 @@ db_config = {
         "source_collection_name": "pancha_sooktham",
         "destination_collection_name": "pancha_sooktham",
     },
+    "taitriya_upanishad": {
+        "source_db_path": "../taitriya_upanishad_ai/chromadb_store",
+        "source_collection_name": "taitriya_upanishad",
+        "destination_collection_name": "taitriya_upanishad",
+    },
 }
 
 parser = argparse.ArgumentParser(description="My app with database parameter")

modules/scripture_helpers/taitriya_upanishad_helper.py

@@ -0,0 +1,17 @@
+import json
+
+
+def get_chapter_order_from_taitriya_upanishad():
+    return {
+        "Seekashaavalli": 1,
+        "Aanandavalli": 2,
+        "Bruguvalli": 3,
+    }
+
+
+def get_taitriya_upanishad_chapters() -> list[str]:
+    """
+    Get a list of chapter names from taitriya upanishad
+    """
+
+    return sorted(get_chapter_order_from_taitriya_upanishad().keys())