comparison script between vidyut-prakriya and ashtadhyayi.com (kartari tinantas)

compare.py

import csv
import json
from dataclasses import dataclass
from indic_transliteration import sanscript
from collections import Counter

@dataclass
class Tinanta:
    gana: int
    number: int
    lakara: str
    text: str


def load_vidyut_padas() -> list[Tinanta]:
    print("Loading vidyut padas")
    padas = []
    with open("tinantas-basic-kartari.csv") as f:
        reader = csv.DictReader(f)
        for row in reader:
            for pada in row['padas'].split('|'):
                padas.append(Tinanta(
                    gana=int(row['gana']),
                    number=int(row['number']),
                    lakara=row['lakara'],
                    text=pada,
                    ))
    return padas


def load_acom_padas() -> list[Tinanta]:
    print("Loading ashtadhyayi.com padas")
    la_map = {
        "plat": "lat",
        "alat": "lat",
        "plit": "lit",
        "alit": "lit",
        "plut": "lut",
        "alut": "lut",
        "plrut": "lrt",
        "alrut": "lrt",
        "plot": "lot",
        "alot": "lot",
        "plang": "lan",
        "alang": "lan",
        "pvidhiling": "vidhi-lin",
        "avidhiling": "vidhi-lin",
        "pashirling": "ashir-lin",
        "aashirling": "ashir-lin",
        "plung": "lun",
        "alung": "lun",
        "plrung": "lrn",
        "alrung": "lrn",
    }
    with open("dhatu/dhatuforms.txt") as f:
        data = json.load(f)

    padas = []
    for key, dhatu_padas in data.items():
        gana, number = key.split('.')
        gana = int(gana)
        number = int(number)
        for raw_lakara, raw_la_padas in dhatu_padas.items():
            lakara = la_map[raw_lakara]
            raw_la_padas_slp1 = sanscript.transliterate(raw_la_padas, "devanagari", "slp1")
            for cell in raw_la_padas_slp1.split(';'):
                for pada in cell.split(','):
                    if pada:
                        padas.append(Tinanta(
                            gana=gana,
                            number=number,
                            lakara=lakara,
                            text=pada
                            ))
    return padas


def _to_dhatu_map(padas: list[Tinanta]) -> dict[[tuple[int, int]], set[Tinanta]]:
    ret = {}
    for p in padas:
        key = (p.gana, p.number)
        if key not in ret:
            ret[key] = set()

        # Drop final d from all padas for ease of comparison
        if not p.text.endswith("d"):
            ret[key].add(p.text)

    return ret


def compare(vidyut_padas: list[Tinanta], acom_padas: list[Tinanta]):
    # group by dhatu
    vidyut_map = _to_dhatu_map(vidyut_padas)
    acom_map = _to_dhatu_map(acom_padas)

    c = Counter()
    for key in vidyut_map:
        if key not in acom_map:
            c["Vidyut key missing in ashtadhyayi.com"] += 1
            continue

        v_padas = vidyut_map[key]
        a_padas = acom_map[key]
        if v_padas == a_padas:
            c["Dhatu exact match"] += 1
        else:
            c["Dhatu mismatch"] += 1
            v_not_a = v_padas - a_padas
            a_not_v = a_padas - v_padas
            gana, number = key
            print(f"{gana}.{number} mismatch:")
            print("  ", v_not_a)
            print("  ", a_not_v)
            print()

    print("Counters:")
    print(c)


def main():
    vidyut_padas = load_vidyut_padas()
    acom_padas = load_acom_padas()
    compare(vidyut_padas, acom_padas)

main()