utils.py

import re
import pandas as pd
from atomic_weights import atomic_weights


def parse_formula(formula):
    pattern = r'([A-Z][a-z]?)(\d*)'
    matches = re.findall(pattern, formula)

    if not matches:
        raise ValueError("Invalid chemical formula.")

    composition = {}

    for element, count in matches:

        if element not in atomic_weights:
            raise ValueError(f"{element} not available in database.")

        count = int(count) if count else 1

        composition[element] = composition.get(element, 0) + count

    return composition


def molecular_weight(comp):

    total = 0

    for element, count in comp.items():
        total += atomic_weights[element] * count

    return round(total,3)


def classify(formula):

    if formula.startswith("H"):
        return "Acid"

    elif formula.endswith("OH"):
        return "Base"

    elif formula.endswith("O"):
        return "Oxide"

    elif "C" in formula:
        return "Organic Compound"

    else:
        return "Salt"


def analyze_compound(formula):

    comp = parse_formula(formula)

    mw = molecular_weight(comp)

    compound_type = classify(formula)

    df = pd.DataFrame({
        "Element": list(comp.keys()),
        "Atoms": list(comp.values())
    })

    summary = f"""
### Analysis Report

**Formula:** {formula}

**Compound Type:** {compound_type}

**Molecular Weight:** {mw} g/mol
"""

    return summary, df