Upload folder using huggingface_hub

config.py

@@ -10,6 +10,7 @@ from modules.scripture_helpers.kamba_ramayanam_helper import (
     get_list_of_padalams,
     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.vishnu_sahasranamam_helper import (
     get_chapter_order_from_sahasranamam,
     get_chapters_from_sahasranamam,
@@ -108,7 +109,7 @@ class SanatanConfig:
             "unit": "verse",
             "unit_field": "relative_verse_number",
             "chapter_order": lambda: get_chapter_order_from_bhagavat_gita(),
-            "audio_storage" : "external",
+            "audio_storage": "external",
             "field_mapping": {
                 "text": "sanskrit",
                 "title": "verse_title",
@@ -785,6 +786,60 @@ class SanatanConfig:
                 "if the user asks for YouTube videos, DO NOT do a web search, instead do a search on this collection."
             ],
         },
+        {
+            "name": "pancha_sooktham",
+            "title": "Pancha Sooktham",
+            "output_dir": "./output/pancha_sooktham",
+            "collection_name": "pancha_sooktham",
+            "collection_embedding_fn": "openai",
+            "unit": "verse",
+            "unit_field": "relative_verse_number",
+            "chapter_order": lambda: get_chapter_order_from_pancha_sooktham(),
+            "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_pancha_sooktham_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/pancha_sooktham.pdf",
+            "source": "https://stotranidhi.com/",
+            "language": "san+eng",
+            "example_labels": [
+                "Purusha Sooktham",
+            ],
+            "examples": [
+                "Show some verses from purusha sooktham",
+            ],
+            "llm_hints": [],
+        },
     ]
 
     def get_scripture_by_collection(self, collection_name: str):

copy_chromadb.py

@@ -23,6 +23,11 @@ db_config = {
         "source_collection_name": "bhagavat_gita",
         "destination_collection_name": "bhagavat_gita_openai",
     },
+    "pancha_sooktham": {
+        "source_db_path": "../sooktham_ai/chromadb_store",
+        "source_collection_name": "pancha_sooktham",
+        "destination_collection_name": "pancha_sooktham",
+    },
 }
 
 parser = argparse.ArgumentParser(description="My app with database parameter")

modules/scripture_helpers/pancha_sooktham_helper.py

@@ -0,0 +1,19 @@
+import json
+
+
+def get_chapter_order_from_pancha_sooktham():
+    return {
+        "Purusha Suktam": 1,
+        "Narayana Suktam": 2,
+        "Sri Suktam": 3,
+        "Bhu Suktam": 4,
+        "Neela Suktam": 5,
+    }
+
+
+def get_pancha_sooktham_chapters() -> list[str]:
+    """
+    Get a list of chapter names from pancha sooktham
+    """
+
+    return sorted(get_chapter_order_from_pancha_sooktham().keys())