Upload folder using huggingface_hub

copy_chromadb.py

@@ -207,6 +207,11 @@ db_config = {
     "source_db_path": "../upadesa_rathnamalai/chromadb_store",
     "source_collection_name": "upadesa_rathnamalai",
     "destination_collection_name": "upadesa_rathnamalai"
+  },
+  "thiruvaimozhi_nootrandhadhi": {
+    "source_db_path": "../thiruvaimozhi_nootrandhadhi_ai/chromadb_store",
+    "source_collection_name": "thiruvaimozhi_nootrandhadhi",
+    "destination_collection_name": "thiruvaimozhi_nootrandhadhi"
   }
 
 }

modules/config/__init__.py

@@ -34,6 +34,7 @@ from modules.config.taitriya_aranyakam import taitriya_aranyakam_config
 from modules.config.taitriya_brahmanam import taitriya_brahmanam_config
 from modules.config.taitriya_samhitha import taitriya_samhitha_config
 from modules.config.taitriya_upanishad import taitriya_upanishad_config
+from modules.config.thiruvaimozhi_nootrandhadhi import thiruvaimozhi_nootrandhadhi_config
 from modules.config.upadesa_rathnamalai import upadesa_rathnamalai_config
 from modules.config.vaamanapuraanam import vaamanapuraanam_config
 from modules.config.vaayupuraanam import vaayupuraanam_config
@@ -91,4 +92,5 @@ scripture_configurations = [
     varahapuranam_config,    
     vaazhi_thirunaamams_config,
     upadesa_rathnamalai_config,
+    thiruvaimozhi_nootrandhadhi_config
 ]

modules/config/thiruvaimozhi_nootrandhadhi.py

@@ -0,0 +1,133 @@
+from modules.config.categories import ScriptureCategoryConstants
+from modules.google_drive.google_drive_utils import convert_drive_url_to_direct
+
+thiruvaimozhi_nootrandhadhi_config = {
+    "name": "thiruvaimozhi_nootrandhadhi",
+    "title": "Thiruvaimozhi NootrandhAdhi",
+    "banner_url": convert_drive_url_to_direct(
+        "https://drive.google.com/file/d/1uLPo1Upx1vLJ6InxqOEMspAOz4UT1F4r/view?usp=drive_link"
+    ),
+    "category": ScriptureCategoryConstants.SRI_SOOKTHI,  # adjust if needed
+    "output_dir": "./outputs/thiruvaimozhi_nootrandhadhi",
+    "collection_name": "thiruvaimozhi_nootrandhadhi",
+    "collection_embedding_fn": "openai",
+
+    # Each verse is a unit
+    "unit": "pasuram",
+    "unit_field": "_global_index",
+
+    # === FIELD MAPPING ===
+    "field_mapping": {
+        "text": "text_ta",
+        "chapter_name": "chapter_name",
+        "unit_index": "_global_index",
+        "relative_path": lambda doc: f"{doc.get('chapter_name')} — {doc.get('relative_verse_number')}",
+        "word_by_word_native": "word_by_word_en",
+        "translation": lambda doc: "\n\n".join(
+            filter(None, [
+                f"TAMIL:\n{doc.get('translation_ta')}" if doc.get('translation_ta') else None,
+                f"ENGLISH:\n{doc.get('translation_en')}" if doc.get('translation_en') else None,
+                f"VYAKYANAM:\n{doc.get('highlights')}" if doc.get('highlights') else None,
+            ])
+        ),
+    },
+
+    # === METADATA FIELDS ===
+    "metadata_fields": [
+        {
+            "name": "chapter_name",
+            "datatype": "str",
+            "label": "Chapter Name",
+            "description": "Chapter title",
+            "show_as_filter": True,
+            "component": "dropdown",
+            "lov": lambda: ["THANIYAN","ANDHADHI"], 
+        },
+        {
+            "name": "_global_index",
+            "datatype": "int",
+            "label": "Global Index",
+            "description": "Absolute verse index",
+            "show_as_filter": True,
+            "is_unique": True,
+        },
+        {
+            "name": "chapter_index",
+            "datatype": "int",
+            "label": "Chapter Index",
+            "description": "Chapter index",
+        },
+        {
+            "name": "relative_verse_number",
+            "datatype": "int",
+            "label": "Verse Number (in chapter)",
+            "description": "Sequential verse index inside the chapter",
+            "show_as_filter": True,
+        },
+        {
+            "name": "text_ta",
+            "datatype": "str",
+            "label": "Tamil Verse",
+            "description": "Original verse text in Tamil",
+            "show_as_filter": True,
+        },
+        {
+            "name": "translation_ta",
+            "datatype": "str",
+            "label": "Translation (Tamil)",
+            "description": "Tamil translation",
+        },
+        {
+            "name": "word_by_word_en",
+            "datatype": "str",
+            "label": "Word by Word (English)",
+            "description": "Word by word meanings in English",
+        },
+        {
+            "name": "translation_en",
+            "datatype": "str",
+            "label": "Translation (English)",
+            "description": "Simple English translation",
+        },
+        {
+            "name": "highlights",
+            "datatype": "str",
+            "label": "Highlights",
+            "description": "Key points / commentary",
+        },
+    ],
+
+    # Optional PDF if you have
+    "pdf_path": "./data/thiruvaimozhi_nootrandhadhi.pdf",
+
+    "source": "koyil.org",
+    "language": "tam",
+    "example_labels": [],
+    "examples": [],
+    "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": "koyil.org",
+                "url": "https://divyaprabandham.koyil.org/index.php/2020/09/thiruvaimozhi-nurrandhadhi-tamil-simple/",
+                "role": "Data provider (Tamil)",
+            },
+            {
+                "name": "koyil.org",
+                "url": "https://divyaprabandham.koyil.org/index.php/2018/11/thiruvaimozhi-nurrandhadhi/",
+                "role": "Data provider (English)",
+            },
+        ],
+        "audio": [],
+        "video": [],
+    },
+}