| |
| """ |
| Comprehensive Basic Arithmetic & BODMAS Shard Generator |
| Generates: |
| 1. Complete Multiplication Tables 1 to 100 (English & Hindi pahade) |
| 2. Direct 1 to 4 digit Addition, Subtraction, Multiplication, Division (+ - * /) |
| 3. Step-by-Step BODMAS Multi-operator Solutions |
| 4. Percentage, Discount, GST, Fractions, Square roots, Cubes |
| """ |
|
|
| import random |
| import pandas as pd |
|
|
| random.seed(42) |
|
|
| def generate_tables(): |
| entries = [] |
| hindi_ordinals = ["ekam", "duni", "tiya", "chauke", "panche", "chhakke", "satte", "atthe", "namme", "daham"] |
| for num in range(1, 101): |
| lines = [f"### Multiplication Table of {num}:"] |
| for m in range(1, 11): |
| lines.append(f"{num} x {m} = {num * m}") |
| full_text_en = "\n".join(lines) |
| entries.append({"text": full_text_en, "category": "table_english", "num": num}) |
|
|
| |
| if num <= 30: |
| lines_hi = [f"### {num} का पहाड़ा (Multiplication Table in Hindi):"] |
| for m in range(1, 11): |
| ans = num * m |
| ord_name = hindi_ordinals[m - 1] |
| lines_hi.append(f"{num} {ord_name} {ans} ({num} x {m} = {ans})") |
| full_text_hi = "\n".join(lines_hi) |
| entries.append({"text": full_text_hi, "category": "table_hindi", "num": num}) |
| return entries |
|
|
| def generate_direct_arithmetic(count=15000): |
| entries = [] |
| for _ in range(count): |
| op = random.choice(["+", "-", "*", "/"]) |
| if op == "+": |
| digits = random.choice([1, 2, 3, 4]) |
| a = random.randint(10**(digits-1), 10**digits - 1) |
| b = random.randint(10**(digits-1), 10**digits - 1) |
| res = a + b |
| text = ( |
| f"Question: Calculate {a} + {b}.\n" |
| f"<soch>\nJod (Addition):\n {a}\n+ {b}\n------\n {res}\n</soch>\n" |
| f"Answer: {a} + {b} = {res}." |
| ) |
| entries.append({"text": text, "category": "addition"}) |
| elif op == "-": |
| digits = random.choice([1, 2, 3, 4]) |
| a = random.randint(10**(digits-1), 10**digits - 1) |
| b = random.randint(10**(digits-1), 10**digits - 1) |
| res = a - b |
| text = ( |
| f"Question: Calculate {a} - {b}.\n" |
| f"<soch>\nGhatana (Subtraction):\n {a}\n- {b}\n------\n {res}\n</soch>\n" |
| f"Answer: {a} - {b} = {res}." |
| ) |
| entries.append({"text": text, "category": "subtraction"}) |
| elif op == "*": |
| a = random.randint(2, 999) |
| b = random.randint(2, 99) |
| res = a * b |
| text = ( |
| f"Question: Calculate {a} * {b} (Multiplication / Guna).\n" |
| f"<soch>\nStep-by-step multiplication:\n{a} multiplied by {b} is calculated as:\n" |
| f"{a} * {b} = {res}\n</soch>\n" |
| f"Answer: {a} * {b} = {res}." |
| ) |
| entries.append({"text": text, "category": "multiplication"}) |
| elif op == "/": |
| b = random.randint(2, 50) |
| q = random.randint(2, 200) |
| r = random.choice([0, random.randint(1, b - 1)]) |
| a = b * q + r |
| if r == 0: |
| text = ( |
| f"Question: Calculate {a} / {b} (Division / Bhag).\n" |
| f"<soch>\n{a} divided by {b}:\nQuotient = {q}, Remainder = 0.\nExact division: {a} / {b} = {q}\n</soch>\n" |
| f"Answer: {a} / {b} = {q}." |
| ) |
| else: |
| dec = round(a / b, 4) |
| text = ( |
| f"Question: Calculate {a} / {b} (Division / Bhag).\n" |
| f"<soch>\n{a} divided by {b}:\nQuotient (Bhagphal) = {q}, Remainder (Sheshphal) = {r}.\nIn decimals: {a} / {b} = {dec}\n</soch>\n" |
| f"Answer: {a} / {b} = {q} remainder {r} (approx {dec})." |
| ) |
| entries.append({"text": text, "category": "division"}) |
| return entries |
|
|
| def generate_bodmas(count=5000): |
| entries = [] |
| for _ in range(count): |
| a = random.randint(5, 50) |
| b = random.randint(2, 12) |
| c = random.randint(2, 8) |
| d = random.randint(1, 10) |
| e = random.randint(2, 6) |
|
|
| pattern = random.choice([1, 2, 3]) |
| if pattern == 1: |
| |
| bracket = c + d |
| mult = b * bracket |
| ans = a + mult |
| expr = f"{a} + {b} * ({c} + {d})" |
| text = ( |
| f"Question: Solve the expression using BODMAS: {expr}\n" |
| f"<soch>\nBODMAS Rule (Brackets, Orders, Division/Multiplication, Addition/Subtraction):\n" |
| f"Step 1: Solve inside Bracket: ({c} + {d}) = {bracket}\n" |
| f"Expression becomes: {a} + {b} * {bracket}\n" |
| f"Step 2: Multiply: {b} * {bracket} = {mult}\n" |
| f"Expression becomes: {a} + {mult}\n" |
| f"Step 3: Add: {a} + {mult} = {ans}\n" |
| f"</soch>\n" |
| f"Answer: {expr} = {ans}." |
| ) |
| elif pattern == 2: |
| |
| part1 = a * b |
| part2 = c * d |
| ans = part1 - part2 |
| expr = f"({a} * {b}) - ({c} * {d})" |
| text = ( |
| f"Question: Solve the expression using BODMAS: {expr}\n" |
| f"<soch>\nStep 1: Solve first bracket: {a} * {b} = {part1}\n" |
| f"Step 2: Solve second bracket: {c} * {d} = {part2}\n" |
| f"Step 3: Subtract: {part1} - {part2} = {ans}\n" |
| f"</soch>\n" |
| f"Answer: {expr} = {ans}." |
| ) |
| else: |
| |
| mult = a * e |
| ans = mult + b - d |
| expr = f"{a} * {e} + {b} - {d}" |
| text = ( |
| f"Question: Solve the expression: {expr}\n" |
| f"<soch>\nBODMAS:\nStep 1: Multiply: {a} * {e} = {mult}\n" |
| f"Step 2: Add: {mult} + {b} = {mult + b}\n" |
| f"Step 3: Subtract: {mult + b} - {d} = {ans}\n" |
| f"</soch>\n" |
| f"Answer: {expr} = {ans}." |
| ) |
| entries.append({"text": text, "category": "bodmas"}) |
| return entries |
|
|
| def generate_percentage_and_everyday(count=5000): |
| entries = [] |
| gst_slabs = [5, 12, 18, 28] |
| for _ in range(count): |
| choice = random.choice(["gst", "discount", "simple_interest", "fraction", "powers"]) |
| if choice == "gst": |
| amount = random.randint(10, 500) * 100 |
| rate = random.choice(gst_slabs) |
| gst_val = int(amount * (rate / 100)) |
| total = amount + gst_val |
| text = ( |
| f"Question: {amount} rupaye par {rate}% GST kitna hoga aur kul rashi (total amount) kitni hogi?\n" |
| f"<soch>\nGST calculation:\n" |
| f"GST Amount = Amount * (Rate / 100) = {amount} * ({rate} / 100) = {gst_val} rupaye.\n" |
| f"Kul Rashi = Mool Rashi + GST = {amount} + {gst_val} = {total} rupaye.\n" |
| f"</soch>\n" |
| f"Answer: {amount} rupaye par {rate}% GST = {gst_val} rupaye hoga, aur kul rashi {total} rupaye hogi." |
| ) |
| entries.append({"text": text, "category": "gst"}) |
| elif choice == "discount": |
| mrp = random.randint(20, 200) * 50 |
| disc_rate = random.choice([10, 15, 20, 25, 30, 40, 50]) |
| disc_val = int(mrp * (disc_rate / 100)) |
| selling_price = mrp - disc_val |
| text = ( |
| f"Question: MRP Rs. {mrp} par {disc_rate}% discount ke baad bikri muly (Selling Price) kya hoga?\n" |
| f"<soch>\nDiscount = {mrp} * ({disc_rate} / 100) = {disc_val} rupaye.\n" |
| f"Selling Price = MRP - Discount = {mrp} - {disc_val} = {selling_price} rupaye.\n" |
| f"</soch>\n" |
| f"Answer: {disc_rate}% discount ke baad selling price Rs. {selling_price} hoga (bachat: Rs. {disc_val})." |
| ) |
| entries.append({"text": text, "category": "discount"}) |
| elif choice == "powers": |
| base = random.randint(2, 25) |
| exp = 2 if base > 15 else random.randint(2, 4) |
| val = base ** exp |
| text = ( |
| f"Question: What is {base}^{exp} (Square / Cube)?\n" |
| f"<soch>\n{base} raised to the power {exp} = {val}\n</soch>\n" |
| f"Answer: {base}^{exp} = {val}." |
| ) |
| entries.append({"text": text, "category": "powers"}) |
| elif choice == "fraction": |
| n1 = random.randint(1, 4) |
| d1 = random.choice([2, 4, 8]) |
| n2 = random.randint(1, 4) |
| d2 = d1 |
| ans_n = n1 + n2 |
| text = ( |
| f"Question: Solve fraction addition: {n1}/{d1} + {n2}/{d2}\n" |
| f"<soch>\nDenominators are same ({d1}):\n({n1} + {n2}) / {d1} = {ans_n} / {d1}\n</soch>\n" |
| f"Answer: {n1}/{d1} + {n2}/{d2} = {ans_n}/{d1}." |
| ) |
| entries.append({"text": text, "category": "fraction"}) |
| return entries |
|
|
| def main(): |
| print("[*] Generating Basic Arithmetic, Tables, BODMAS, and Everyday Math...") |
| all_data = [] |
| all_data.extend(generate_tables()) |
| all_data.extend(generate_direct_arithmetic(15000)) |
| all_data.extend(generate_bodmas(5000)) |
| all_data.extend(generate_percentage_and_everyday(5000)) |
|
|
| random.shuffle(all_data) |
| df = pd.DataFrame(all_data) |
| out_file = "basic_arithmetic_bodmas_tables.parquet" |
| df.to_parquet(out_file, compression="zstd") |
| print(f"[OK] Generated {out_file}: {len(df)} rows, {df.memory_usage().sum() / (1024**2):.2f} MB") |
|
|
| if __name__ == "__main__": |
| main() |
|
|
