Upload folder using huggingface_hub

.gitignore

@@ -13,3 +13,4 @@ chromadb-store/
 chromadb-store.zip
 outputs/
 chromadb-store_20251112.zip
+chromadb-store_20251118.zip

copy_chromadb.py

@@ -98,6 +98,11 @@ db_config = {
         "source_collection_name": "narayaneeyam",
         "destination_collection_name": "narayaneeyam",
     },
+   "bhagavata_purana": {
+        "source_db_path": "../bhagavata_purana_ai/chromadb_store",
+        "source_collection_name": "bhagavata_purana",
+        "destination_collection_name": "bhagavata_purana",
+    },    
 }
 
 parser = argparse.ArgumentParser(description="My app with database parameter")

modules/config/bhagavata_purana.py

@@ -4,32 +4,72 @@ from modules.google_drive.google_drive_utils import convert_drive_url_to_direct
 
 bhagavata_purana_config = {
     "name": "bhagavata_purana",
-    "title": "Bhagavatha Puranam",
+    "title": "Srimad Bhagavatham",
     "banner_url": convert_drive_url_to_direct(
         "https://drive.google.com/file/d/1APDzjcF-G4tn6qHHn0QFmA-odcrQS2-e/view?usp=drive_link"
     ),
     "category" : ScriptureCategoryConstants.PURANAM,
     "output_dir": "./output/bhagavata_purana",
     "collection_name": "bhagavata_purana",
-    "unit": "page",
+    "collection_embedding_fn": "openai",
+    "unit": "sloka",
+    "unit_field": "_global_index",
+    "field_mapping": {
+        "text": "sanskrit",
+        "unit_index": "sloka_number",
+        "chapter_name": lambda doc: (
+            f"Book {doc.get('book_number')}:Chapter {doc.get('chapter_number')}"
+        ),
+        "author": lambda doc: f" Sage Vyasa",
+        "word_by_word_native": "wbw",
+        "translation": "translation",
+        "relative_path": lambda doc: (
+            f"Book#{doc.get('book_number')}: Chapter#{doc.get('chapter_number')}:Sloka#{doc.get('sloka_number')} || {doc.get('_global_index')} ||"
+        ),
+    },
     "metadata_fields": [
         {
-            "name": "file",
-            "label": "File Name",
+            "name": "book_number",
             "datatype": "str",
-            "description": "name of the file from which the information was extracted",
+            "label": "Book Number",
+            "description": "Book Number",
+            "show_as_filter": True,
+            "is_unique": True,
+        },
+        {
+            "name": "chapter_number",
+            "datatype": "str",
+            "label": "Chapter Number",
+            "description": "Chapter Number",
+            "show_as_filter": True,
+            "is_unique": True,
+        },
+        {
+            "name": "sloka_number",
+            "datatype": "int",
+            "label": "Relative Verse Number",
+            "description": "Relative Verse Number",
+            "show_as_filter": True,
+            "is_unique": True,
         },
         {
-            "name": "page",
+            "name": "_global_index",
             "datatype": "int",
-            "label": "Page Number",
-            "description": "Page number from the source",
+            "label": "Absolute Verse Number",
+            "description": "Absolute Verse Number",
             "show_as_filter": True,
             "is_unique": True,
         },
+        {
+            "name": "sanskrit",
+            "label": "Lyrics in Sanskrit",
+            "datatype": "str",
+            "description": "The original verse in Sanskrit.",
+            "show_as_filter": True,
+        },
     ],
     "pdf_path": "./data/bhagavata_purana.pdf",
-    "source": "https://dn790003.ca.archive.org/0/items/bhagavatapuranagitapress_201907/Bhagavata%20Purana%20-%20Gita%20Press_text.pdf",
+    "source": "https://bhaagavatam-firststep.org/",
     "language": "san+eng",
     "example_labels": ["Gajendra Moksham", "Prahalad"],
     "examples": [
@@ -37,4 +77,24 @@ bhagavata_purana_config = {
         "State some verses that showcase the devotion of Prahlada",
     ],
     "llm_hints": [],
+    "credits": {
+        "art": [
+            {
+                "name": "Sri TNCN",
+                "url": "https://www.linkedin.com/in/narashiman-chakravarthy-54326111/",
+                "role": "Illustrator",
+                "notes": "Original artwork used with permission",
+                "photo_url": "https://drive.google.com/uc?export=download&id=1D4Vo-Mvkp1RdZR6whUkcjvNBIW1yHSUE",
+            }
+        ],
+        "data": [
+            {
+                "name": "First Step",
+                "url": "https://bhaagavatam-firststep.org/",
+                "role": "Data provider",
+            },
+        ],
+        "audio": [],
+        "video": [],
+    },    
 }