Upload folder using huggingface_hub

copy_chromadb.py

@@ -283,6 +283,11 @@ db_config = {
     "source_collection_name": "padhuka_sahasram",
     "destination_collection_name": "padhuka_sahasram"
   },
+  "chandogya_upanishad": {
+    "source_db_path": "../chandogya_upanishad_ai/chromadb_store",
+    "source_collection_name": "chandogya_upanishad",
+    "destination_collection_name": "chandogya_upanishad"
+  },
 
 }
 

modules/config/__init__.py

@@ -7,6 +7,7 @@ from modules.config.brahma_sutra import brahma_sutra_config
 from modules.config.brahmandpuranam import brahmandpuranam_config
 from modules.config.brahmapuranam import brahmapuranam_config
 from modules.config.brahmavaivarthapurana import brahmavaivarthapurana_config
+from modules.config.chandogya_upanishad import chandogya_upanishad_config
 from modules.config.chathusloki import chathusloki_config
 from modules.config.desika_prabandham import desika_prabandham_config
 from modules.config.devaraja_ashtakam import devaraja_ashtakam_config
@@ -119,5 +120,6 @@ scripture_configurations = [
     dhati_panchakam_config,
     gadhyathrayam_config,
     aarthi_prabandham_config,
-    padhuka_sahasram_config
+    padhuka_sahasram_config,
+    chandogya_upanishad_config
 ]

modules/config/chandogya_upanishad.py

@@ -0,0 +1,127 @@
+from modules.config.categories import ScriptureCategoryConstants
+from modules.google_drive.google_drive_utils import convert_drive_url_to_direct
+
+chandogya_upanishad_config = {
+    "name": "chandogya_upanishad",
+    "title": "Chāndogya Upaniṣad",
+    "banner_url": convert_drive_url_to_direct(
+        "https://drive.google.com/file/d/1JsD1k976QNaXRTMB1QtX1ASSbFSo_HoZ/view?usp=drive_link"
+    ),
+    "category": ScriptureCategoryConstants.VEDAM,
+    "output_dir": "./outputs/chandogya_upanishad",
+    "collection_name": "chandogya_upanishad",
+    "collection_embedding_fn": "openai",
+
+    # ---- units ----
+    "unit": "sloka",
+    "unit_field": "_global_index",
+
+    # ---- how documents are interpreted ----
+    "field_mapping": {
+        "text": "lyrics_sa",
+        "unit_index": "_global_index",
+        # Human-readable reference
+        "relative_path": lambda doc: (
+            f"{doc['chapter']}.{doc['section']}.{doc['verse']}"
+        ),
+        "author": lambda doc: "Vedic tradition (Sāmaveda)",
+        "translation" : lambda doc: (
+            f"Simple meaning: \n{doc['simple_meaning_en']}\n"
+            f"Commentary: \n {doc['commentary_en']}"
+        ),
+        "word_by_word_native" : "word_by_word_en",
+    },
+
+    # ---- metadata exposed to UI / filters ----
+    "metadata_fields": [
+        {
+            "name": "chapter",
+            "datatype": "int",
+            "label": "Chapter",
+            "description": "Chapter (Adhyāya) number",
+            "show_as_filter": True,
+        },
+        {
+            "name": "section",
+            "datatype": "int",
+            "label": "Section",
+            "description": "Section (Khanda) number within chapter",
+            "show_as_filter": True,
+        },
+        {
+            "name": "verse",
+            "datatype": "int",
+            "label": "Verse",
+            "description": "Verse number within section",
+            "show_as_filter": True,
+        },
+        {
+            "name": "_global_index",
+            "datatype": "int",
+            "label": "Global Index",
+            "description": "Absolute verse index across the entire Upanishad",
+            "show_as_filter": True,
+            "is_unique": True,
+        },
+        {
+            "name": "lyrics_sa",
+            "datatype": "str",
+            "label": "Sanskrit Text",
+            "description": "Original Sanskrit verse",
+            "show_as_filter": True,
+        },
+        {
+            "name": "lyrics_en",
+            "datatype": "str",
+            "label": "IAST Transliteration",
+            "description": "Roman transliteration of Sanskrit",
+            "show_as_filter": True,
+        },
+        {
+            "name": "simple_meaning_en",
+            "datatype": "str",
+            "label": "Simple Meaning",
+            "description": "Prose English explanation",
+            "show_as_filter": False,
+        },
+        {
+            "name": "word_by_word_en",
+            "datatype": "str",
+            "label": "Word-by-word Meaning",
+            "description": "Gloss-based translation",
+            "show_as_filter": False,
+        },
+        {
+            "name": "commentary_en",
+            "datatype": "str",
+            "label": "Commentary",
+            "description": "Traditional explanatory commentary",
+            "show_as_filter": False,
+        },
+    ],
+
+    # ---- source & attribution ----
+    "pdf_path": None,
+    "source": "wisdomlib.org",
+    "language": "san",
+
+    "example_labels": [],
+    "examples": [],
+    "llm_hints": [],
+
+    "credits": {
+        "data": [
+            {
+                "name": "Wisdom Library",
+                "url": "https://www.wisdomlib.org/hinduism/book/chandogya-upanishad-english",
+                "role": "Data Provider",
+                "photo_url" : "https://www.wisdomlib.org/images/logo.gif",
+            }
+        ],
+        "art": [],
+        "audio": [],
+        "video": [],
+    },
+
+    "audio_modes": {},
+}