Earnings from EDGAR

"""Runs in the session's shell: yearly revenue, net income and diluted EPS of a few companies from SEC EDGAR'scompanyfacts API (the numbers the companies filed in XBRL), growth and margins worked out here, a CSV, a chart and ashort summary. Prints the result as JSON.Environment: TICKERS (e.g. AAPL,MSFT), SEC_USER_AGENT (your name and email: SEC asks every automated client to declareone), YEARS (default 4), EDGAR_DATA_URL and EDGAR_TICKERS_URL (SEC's by default).How a yearly value is chosen: facts from 10-K filings whose period is about a year (350-380 days), one per period end;when several filings report the same period (each 10-K repeats the two years before it, and may restate one), the onefiled last wins. Quarterly (10-Q) facts are left out."""import csvimport jsonimport osimport timefrom datetime import dateimport matplotlibmatplotlib.use("Agg")import matplotlib.pyplot as plt  # noqa: E402import requests  # noqa: E402TICKERS = [t.strip().upper() for t in os.environ.get("TICKERS", "AAPL,MSFT").split(",") if t.strip()]YEARS = int(os.environ.get("YEARS", "4"))DATA = os.environ.get("EDGAR_DATA_URL", "https://data.sec.gov").rstrip("/")TICKERS_URL = os.environ.get("EDGAR_TICKERS_URL", "https://www.sec.gov/files/company_tickers.json")http = requests.Session()http.headers.update({"User-Agent": os.environ["SEC_USER_AGENT"], "Accept-Encoding": "gzip, deflate"})REVENUE = ["Revenues", "RevenueFromContractWithCustomerExcludingAssessedTax", "SalesRevenueNet", "RevenueFromContractWithCustomerIncludingAssessedTax"]calls = 0def get(url):    global calls    calls += 1    time.sleep(0.15)  # SEC's fair-access limit is 10 requests a second; this stays well under it    r = http.get(url, timeout=60)    r.raise_for_status()    return r.json()def yearly(facts, concept, unit):    """{period end: (value, filed)} for one concept: full-year 10-K values, the latest filing per period."""    out = {}    for f in facts.get(concept, {}).get("units", {}).get(unit, []):        if not f.get("form", "").startswith("10-K") or "start" not in f:            continue        days = (date.fromisoformat(f["end"]) - date.fromisoformat(f["start"])).days        if not 350 <= days <= 380:            continue        if f["end"] not in out or f["filed"] > out[f["end"]][1]:            out[f["end"]] = (f["val"], f["filed"])    return outciks = {v["ticker"].upper(): (v["cik_str"], v["title"]) for v in get(TICKERS_URL).values()}companies = []for ticker in TICKERS:    if ticker not in ciks:        raise SystemExit(f"{ticker} is not in SEC's ticker list")    cik, title = ciks[ticker]    gaap = get(f"{DATA}/api/xbrl/companyfacts/CIK{cik:010d}.json")["facts"].get("us-gaap", {})    # Companies change the concept they file revenue under: take the one with the most recent year.    options = [(c, yearly(gaap, c, "USD")) for c in REVENUE]    concept, revenue = max(options, key=lambda o: (max(o[1], default=""), len(o[1])))    income = yearly(gaap, "NetIncomeLoss", "USD")    eps = yearly(gaap, "EarningsPerShareDiluted", "USD/shares")    years = []    for end in sorted(revenue)[-YEARS:]:        rev = revenue[end][0]        ni = income.get(end, (None,))[0]        years.append({"end": end, "revenue": rev, "netIncome": ni, "eps": eps.get(end, (None,))[0], "filed": revenue[end][1],                      "netMarginPct": round(100 * ni / rev, 1) if ni is not None and rev else None})    for prev, cur in zip(years, years[1:]):        cur["revenueGrowthPct"] = round(100 * (cur["revenue"] / prev["revenue"] - 1), 1)    companies.append({"ticker": ticker, "cik": cik, "name": title, "revenueConcept": concept, "years": years})with open("earnings.csv", "w", newline="") as f:    w = csv.writer(f)    w.writerow(["ticker", "year_end", "revenue_usd", "net_income_usd", "eps_diluted", "revenue_growth_pct", "net_margin_pct", "filed"])    for c in companies:        for y in c["years"]:            w.writerow([c["ticker"], y["end"], y["revenue"], y["netIncome"], y["eps"], y.get("revenueGrowthPct", ""), y["netMarginPct"], y["filed"]])fig, axes = plt.subplots(1, len(companies), figsize=(5 * len(companies), 4), squeeze=False)for ax, c in zip(axes[0], companies):    labels = [y["end"][:7] for y in c["years"]]    xs = range(len(labels))    ax.bar([x - 0.2 for x in xs], [y["revenue"] / 1e6 for y in c["years"]], 0.4, label="Revenue", color="#1f6feb")    ax.bar([x + 0.2 for x in xs], [(y["netIncome"] or 0) / 1e6 for y in c["years"]], 0.4, label="Net income", color="#2da44e")    ax.axhline(0, color="#888", linewidth=0.8)    ax.set_xticks(list(xs), labels)    ax.set_title(f"{c['ticker']} ({c['name']})", fontsize=10)    ax.set_ylabel("USD millions")    ax.legend(fontsize=8)fig.tight_layout()fig.savefig("chart.png", dpi=120)def money(v):    return "n/a" if v is None else f"${v / 1e9:.2f} billion" if abs(v) >= 1e9 else f"${v / 1e6:.1f} million"lines = ["# Earnings from SEC EDGAR", "", "Yearly figures as filed in 10-K reports (XBRL companyfacts); growth and margins computed here.", ""]for c in companies:    last = c["years"][-1]    growth = last.get("revenueGrowthPct")    lines += [f"## {c['ticker']}: {c['name']}", "",              f"Year ending {last['end']}: revenue {money(last['revenue'])}" + (f" ({growth:+.1f}% on the year before)" if growth is not None else "") +              f", net income {money(last['netIncome'])} ({last['netMarginPct']}% of revenue), diluted EPS {last['eps']}.", "",              "| Year end | Revenue | Net income | EPS | Growth | Margin |", "|---|---|---|---|---|---|"]    lines += [f"| {y['end']} | {money(y['revenue'])} | {money(y['netIncome'])} | {y['eps']} | {y.get('revenueGrowthPct', '')} | {y['netMarginPct']} |" for y in c["years"]]    lines.append("")open("summary.md", "w").write("\n".join(lines))print(json.dumps({"companies": companies, "requests": calls}))