vikramvasudevan's picture
Upload folder using huggingface_hub
7be9035 verified
from modules.config.categories import ScriptureCategoryConstants
from modules.google_drive.google_drive_utils import convert_drive_url_to_direct
katakam_config = {
"name": "katakam",
"title": "Katakam",
"banner_url": convert_drive_url_to_direct(
"https://drive.google.com/file/d/1bGH6JRzt3-BSW5t7J77mcyA9W4T5DeGu/view?usp=drive_link"
),
"category" : ScriptureCategoryConstants.VEDAM,
"output_dir": "./output/katakam",
"collection_name": "katakam",
"collection_embedding_fn": "openai",
"unit": "panchadhi",
"unit_field": "panchadhi",
# "chapter_order": lambda: get_chapter_order_from_katakam(),
"field_mapping": {
"text": "sanskrit",
"unit_index": "panchadhi",
"transliteration": "transliteration",
"chapter_name": lambda doc: (
doc.get("prapaatakam_name")
if doc.get("prapaatakam_name") and doc.get("prapaatakam_name") != "-"
else "॥ काठकम् ॥"
),
"relative_path": lambda doc: (
"॥ काठकम् ॥"
if not doc.get("panchadhi") or doc.get("panchadhi") == "-"
else f"॥ काठकम् ॥ {doc.get('prapaatakam')}.{doc.get('anuvakam')}.{doc.get('panchadhi_index')} || {doc.get('panchadhi_sa')} ||"
),
},
"metadata_fields": [
{
"name": "prapaatakam",
"datatype": "int",
"label": "Prapaatakam Number",
"description": "Prapaatakam Number",
"show_as_filter": True,
"is_unique": True,
},
{
"name": "anuvakam",
"datatype": "int",
"label": "Anuvakam Number",
"description": "Anuvakam Number",
"show_as_filter": True,
"is_unique": True,
},
{
"name": "panchadhi_index",
"datatype": "int",
"label": "Relative Panchadhi Number",
"description": "Relative Panchadhi Number",
"show_as_filter": True,
"is_unique": True,
},
{
"name": "panchadhi",
"datatype": "int",
"label": "Absolute Panchadhi Number",
"description": "Absolute Panchadhi Number",
"show_as_filter": True,
"is_unique": True,
},
{
"name": "sanskrit",
"label": "Lyrics in sanskrit",
"datatype": "str",
"description": "The original sloka in sanskrit.",
"show_as_filter": True,
},
{
"name": "transliteration",
"label": "Transliteration in english",
"datatype": "str",
"description": "The original sloka transliterated in English.",
"show_as_filter": True,
},
],
"pdf_path": "./data/katakam.pdf",
"source": "https://sanskritdocuments.org/doc_veda/taittirIyabrAhmaNam.html",
"language": "san+eng",
"example_labels": [
"Katakam",
],
"examples": [
"Show some verses from Katakam",
],
"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": "sanskritdocuments.org",
"url": "https://sanskritdocuments.org/doc_veda/taittirIyabrAhmaNam.html",
"role": "Data provider",
}
],
"audio": [
{
"name": "Srirangam Sri Vasudevan Srinivasachariar",
"url": "https://www.youtube.com/@akivasudev",
"role": "Recitation audio provider",
"photo_url": "https://drive.google.com/uc?export=download&id=1HB1Nr39j2-55rIeOK_wYNbZZndcGZIaf",
},
],
"video": [],
},
}