money/statement_inventory.py

#!/usr/bin/env python3
"""Statement inventory: shows which accounts/months you have statements for, what is missing,
and what to download next. Read-only: never edits Bills_Tracker.xlsx.

Run:  .venv/bin/python statement_inventory.py [folder]      (default: this folder)
Add a new account: add a line to ACCOUNTS below (match text found on its statement).
"""
import sys, re, csv, glob, hashlib, datetime as dt, calendar
from pathlib import Path
import pdfplumber

# name, text that identifies it, statement day-of-month for "next due" (None = month end)
ACCOUNTS = [
    ("Chase Visa",       ["chase.com", "Prime Visa"], 15),
    ("Ally Checking",    ["COMBINED CUSTOMER STATEMENT", "ally.com"], 10),
    ("CVFCU Checking",   ["cvfcu.com"],               None),
]
CSV_ACCOUNTS = {"1494": "Capital One Card"}   # card number (last 4) -> account name

def eom(y, m): return calendar.monthrange(y, m)[1]
def d(s):
    for f in ("%m/%d/%Y", "%m/%d/%y", "%Y-%m-%d"):
        try: return dt.datetime.strptime(s, f).date()
        except ValueError: pass

def read_pdf(p):
    with pdfplumber.open(p) as pdf:
        first = pdf.pages[0].extract_text() or ""
        full = "\n".join((pg.extract_text() or "") for pg in pdf.pages)
    for name, keys, _ in ACCOUNTS:
        if any(k.lower() in first.lower() for k in keys):  # first page only: other pages mention other banks
            m = (re.search(r"Opening/Closing Date\s+\S+ - (\S+)", full)
                 or re.search(r"Statement Ending (\S+)", first)
                 or re.search(r"Statement Date.{0,120}?(\d\d/\d\d/\d{4})", first, re.S))
            end = d(m.group(1)) if m else None
            return name, end
    return None, None

def read_csv(p):
    try:
        rows = list(csv.DictReader(open(p, newline="", encoding="utf-8-sig")))
        dates = [d(r["Transaction Date"]) for r in rows if r.get("Transaction Date")]
        card = {r.get("Card No.", "") for r in rows}
        if not dates: return None
        name = next((CSV_ACCOUNTS[c] for c in card if c in CSV_ACCOUNTS), None)
        return name or f"Unrecognized CSV (card {','.join(sorted(card))})", min(dates), max(dates), len(rows)
    except Exception:
        return None

def main(folder):
    folder = Path(folder)
    seen, found, unknown = {}, {}, []
    for p in sorted(folder.glob("*.pdf")):
        h = hashlib.md5(p.read_bytes()).hexdigest()
        name, end = read_pdf(p)
        if not name or not end: unknown.append(p.name); continue
        key = (name, end.year, end.month)
        if key in found: found[key].append(p.name)
        else: found[key] = [p.name]
    csvs = []
    for p in sorted(folder.glob("*.csv")):
        r = read_csv(p)
        if r: csvs.append((p.name,) + r)
    accts = [a[0] for a in ACCOUNTS] + sorted({c[1] for c in csvs})
    if not found and not csvs: print("No statements found."); return
    months = sorted({(y, m) for (_, y, m) in found})
    first = months[0] if months else (csvs[0][2].year, csvs[0][2].month)
    today = dt.date.today(); last = (today.year, today.month)
    seq, (y, m) = [], first
    while (y, m) <= last:
        seq.append((y, m)); m += 1
        if m > 12: y, m = y + 1, 1
    out = []
    out.append(f"STATEMENT COVERAGE as of {today:%b %d, %Y}  (column = month the statement closes;  X = have it,  . = MISSING,  - = not yet due)\n")
    out.append(f"{'':18}" + " ".join(f"{calendar.month_abbr[m]}{str(y)[2:]:>3}" for y, m in seq))
    notes = []
    for a in accts:
        cells = []
        for y, m in seq:
            if a in {c[1] for c in csvs}:
                inside = any(c[1] == a and (c[2].year, c[2].month) <= (y, m) <= (c[3].year, c[3].month) for c in csvs)
                cells.append("X" if inside else ("." if (y, m) >= (min(c[2] for c in csvs if c[1]==a).year, min(c[2] for c in csvs if c[1]==a).month) else " "))
                continue
            cfg = next(x for x in ACCOUNTS if x[0] == a)
            due = dt.date(y, m, cfg[2] or eom(y, m))
            have = (a, y, m) in found
            mine = [k for k in found if k[0] == a]
            started = (y, m) >= min((k[1], k[2]) for k in mine) if mine else False
            cells.append("X" if have else (" " if not started else ("-" if due > today else ".")))
        out.append(f"{a:18}" + " ".join(f"{c:>6}" for c in cells))
    out.append("")
    for a in accts:
        mine = sorted(k for k in found if k[0] == a)
        if mine:
            cfg = next(x for x in ACCOUNTS if x[0] == a)
            dup = [(k, v) for k, v in found.items() if k[0] == a and len(v) > 1]
            ly, lm = mine[-1][1], mine[-1][2]
            ny, nm = (ly, lm + 1) if lm < 12 else (ly + 1, 1)
            nxt = dt.date(ny, nm, cfg[2] or eom(ny, nm))
            miss = [f"{calendar.month_abbr[m]} {y}" for y, m in seq if (y, m) >= (mine[0][1], mine[0][2]) and (a, y, m) not in found and dt.date(y, m, cfg[2] or eom(y, m)) <= today]
            out.append(f"{a}: {len(mine)} statements, {calendar.month_abbr[mine[0][2]]} {mine[0][1]} – {calendar.month_abbr[lm]} {ly}."
                       f"  Next statement closes ~{nxt:%b %d, %Y}" + ("  (not out yet)" if nxt > today else "  (DOWNLOAD NOW)")
                       + (f"  MISSING in range: {', '.join(miss)}" if miss else "  No gaps."))
            for k, v in dup: out.append(f"    duplicate files for {calendar.month_abbr[k[2]]} {k[1]}: {', '.join(v)} (counted once)")
    for c in csvs:
        out.append(f"{c[1]}: CSV transaction download '{c[0]}', {c[2]:%b %d, %Y} – {c[3]:%b %d, %Y}, {c[4]} rows.  Next download should start {c[3]+dt.timedelta(days=1):%b %d, %Y}.")
    if unknown: out.append("\nUnrecognized files (new account? add it to ACCOUNTS): " + ", ".join(unknown))
    out.append("\nTo add a missed or new statement: drop the PDF/CSV in this folder, re-run this script, then ask Claude to load it into the workbook.")
    text = "\n".join(out); print(text)
    (folder / "Statement_Coverage.txt").write_text(text + "\n")

main(sys.argv[1] if len(sys.argv) > 1 else Path(__file__).parent)