app/parser.py

import re
from typing import List, Dict, Optional
from .schemas import TransactionRow


AMOUNT_RE = re.compile(r"(\d{1,3}(?:,\d{3})*(?:\.\d{2})?)")


def normalize_spaces(text: str) -> str:
    return " ".join(text.split())


def parse_amount(value: str) -> Optional[float]:
    try:
        return float(value.replace(",", ""))
    except:
        return None


def looks_like_bank_statement(text: str) -> bool:
    if not text or len(text.strip()) < 50:
        return False

    return True


def is_datetime_line(line: str) -> bool:
    return bool(re.match(r"^\d{4}-\d{2}-\d{2}T\d{2}:", line.strip()))


def extract_amounts(text: str) -> List[float]:
    matches = AMOUNT_RE.findall(text)
    values = [parse_amount(m) for m in matches if parse_amount(m) is not None]
    return values


def parse_statement_pages(page_texts: List[Dict]) -> List[TransactionRow]:
    rows: List[TransactionRow] = []

    for page in page_texts:
        page_no = page["page"]
        text = page["text"] or ""

        lines = [normalize_spaces(l) for l in text.splitlines() if l.strip()]

        i = 0
        while i < len(lines):
            line = lines[i]

            if not is_datetime_line(line):
                i += 1
                continue

            current_date = line.split("T")[0]

            block_lines = [line]
            i += 1

            while i < len(lines) and not is_datetime_line(lines[i]):
                block_lines.append(lines[i])
                i += 1

            block_text = " ".join(block_lines)

            amounts = extract_amounts(block_text)

            if len(amounts) < 3:
                continue

            debit = amounts[-3]
            credit = amounts[-2]
            balance = amounts[-1]

            description = re.sub(r"^\d{4}-\d{2}-\d{2}T\d{2}:", "", block_text)

            raw_amount_strings = AMOUNT_RE.findall(block_text)
            for amt in raw_amount_strings[-3:]:
                description = description.replace(amt, " ")

            description = normalize_spaces(description)

            if not description or len(description) < 3:
                continue

            row = TransactionRow(
                date=current_date,
                description=description,
                debit=debit if debit != 0 else None,
                credit=credit if credit != 0 else None,
                balance=balance,
                source_page=page_no,
                confidence="high",
            )

            rows.append(row)

    return rows


def dedupe_rows(rows: List[TransactionRow]) -> List[TransactionRow]:
    seen = set()
    deduped: List[TransactionRow] = []

    for row in rows:
        key = (
            row.date,
            row.description,
            row.debit,
            row.credit,
            row.balance,
            row.source_page,
        )

        if key in seen:
            continue

        seen.add(key)
        deduped.append(row)

    return deduped