scripts/test_form4_pipeline.py

# test_form4_pipeline.py
# Quick smoke-test for the Form 4 collection pipeline.
# Runs the full pipeline (metadata fetch → XML download → parse) on
# 5 well-known S&P 500 companies for 2024 only (~2-5 minutes).
#
# Checks:
#   1. EDGAR submissions API is reachable and returns Form 4 metadata
#   2. XML download and parsing produces valid transaction rows
#   3. All expected columns are present
#   4. Values are sensible (dates, numeric fields, A/D codes)
#
# Run: python scripts/test_form4_pipeline.py

import time
import threading
import requests
import pandas as pd
from xml.etree import ElementTree as ET

# ── Test config ────────────────────────────────────────────────────────────────
TEST_CIKS = {
    320193:  'AAPL',    # Apple
    789019:  'MSFT',    # Microsoft
    1045810: 'NVDA',    # NVIDIA
    1018724: 'AMZN',    # Amazon
    1652044: 'GOOGL',   # Alphabet
}
DATE_START = '2024-01-01'
DATE_END   = '2024-12-31'
MAX_XML_PER_CIK = 3      # only parse first 3 XMLs per company — keeps test fast

HEADERS = {'User-Agent': 'S2880814 University of Edinburgh s.g.vishnu@sms.ed.ac.uk'}
TIMEOUT = 15

EXPECTED_INDEX_COLS = {'cik', 'ticker', 'filing_date', 'accession', 'primary_doc'}
EXPECTED_TXN_COLS   = {
    'cik', 'ticker', 'filing_date', 'accession',
    'owner_name', 'is_director', 'is_officer', 'is_ten_pct_owner', 'officer_title',
    'security_title', 'transaction_date', 'transaction_code',
    'acquired_disposed', 'shares', 'price_per_share', 'shares_after_txn',
}

PASS = '  [PASS]'
FAIL = '  [FAIL]'
INFO = '  [INFO]'


# ── Rate limiter (same as main script) ────────────────────────────────────────
class RateLimiter:
    def __init__(self, rate):
        self._lock    = threading.Lock()
        self._min_gap = 1.0 / rate
        self._last    = 0.0
    def wait(self):
        with self._lock:
            gap = self._min_gap - (time.monotonic() - self._last)
            if gap > 0:
                time.sleep(gap)
            self._last = time.monotonic()

limiter = RateLimiter(8.0)


def safe_get(url, retries=3):
    for attempt in range(retries):
        limiter.wait()
        try:
            r = requests.get(url, headers=HEADERS, timeout=TIMEOUT)
            if r.status_code == 200:
                return r
            elif r.status_code == 429:
                print(f"    [429] Rate limited, sleeping 10s...")
                time.sleep(10)
        except requests.RequestException as e:
            print(f"    [WARN] Request error (attempt {attempt+1}): {e}")
            time.sleep(2)
    return None


# ── Test 1: EDGAR API reachability ─────────────────────────────────────────────
print("=" * 60)
print("TEST 1 — EDGAR API reachability")
print("=" * 60)

r = safe_get("https://data.sec.gov/submissions/CIK0000320193.json")  # Apple
if r:
    data = r.json()
    company_name = data.get('name', 'unknown')
    print(f"{PASS} EDGAR submissions API reachable")
    print(f"{INFO} Response: {company_name} (CIK 320193 = Apple)")
else:
    print(f"{FAIL} Could not reach EDGAR API — check internet connection")
    exit(1)


# ── Test 2: Form 4 metadata fetch ──────────────────────────────────────────────
print()
print("=" * 60)
print("TEST 2 — Form 4 metadata fetch (5 companies, 2024 only)")
print("=" * 60)

def get_form4_meta(cik):
    r = safe_get(f"https://data.sec.gov/submissions/CIK{cik:010d}.json")
    if not r:
        return []
    data   = r.json()
    recent = data.get('filings', {}).get('recent', {})

    forms = recent.get('form', [])
    accs  = recent.get('accessionNumber', [])
    dates = recent.get('filingDate', [])
    docs  = recent.get('primaryDocument', [])

    result = []
    for form, acc, date, doc in zip(forms, accs, dates, docs):
        if form == '4' and DATE_START <= date <= DATE_END:
            result.append({'accession': acc, 'filing_date': date, 'primary_doc': doc})
    return result

index_rows = []
all_test_filings = []

for cik, ticker in TEST_CIKS.items():
    filings = get_form4_meta(cik)
    print(f"{INFO} {ticker:6s} (CIK {cik}): {len(filings):3d} Form 4 filings in 2024")
    if filings:
        print(f"{PASS} Metadata fetch OK — sample accession: {filings[0]['accession']}")
    else:
        print(f"{FAIL} No Form 4 filings found for {ticker} in 2024")

    for f in filings:
        index_rows.append({
            'cik': cik, 'ticker': ticker,
            'filing_date': f['filing_date'],
            'accession':   f['accession'],
            'primary_doc': f['primary_doc'],
        })
        all_test_filings.append((cik, ticker, f))

print(f"\n{INFO} Total filings collected across 5 companies: {len(index_rows)}")

# Column check on index
index_df = pd.DataFrame(index_rows)
missing_cols = EXPECTED_INDEX_COLS - set(index_df.columns)
if not missing_cols:
    print(f"{PASS} Index columns all present: {sorted(index_df.columns.tolist())}")
else:
    print(f"{FAIL} Missing index columns: {missing_cols}")


# ── Test 3: XML download and parsing ──────────────────────────────────────────
print()
print("=" * 60)
print(f"TEST 3 — XML download + parse (first {MAX_XML_PER_CIK} filings per company)")
print("=" * 60)

def find_xml_in_filing(cik, accession):
    """Fetch filing directory and return the XML data filename."""
    import re
    acc_clean = accession.replace('-', '')
    r = safe_get(f"https://www.sec.gov/Archives/edgar/data/{cik}/{acc_clean}/")
    if not r:
        return None
    matches = re.findall(r'href="([^"]+\.xml)"', r.text, re.IGNORECASE)
    for m in matches:
        fname = m.split('/')[-1]
        if fname and 'index' not in fname.lower():
            return fname
    return None


def parse_form4(cik, ticker, accession, primary_doc, filing_date):
    """
    Download Form 4 XML and extract non-derivative transactions.

    Resolution order (no extra API calls in the common case):
      1. If primaryDoc is .htm, try swapping extension to .xml first
      2. If that fails, fall back to directory listing (1 extra call)
      3. If all else fails, return error
    """
    acc_clean = accession.replace('-', '')

    candidates = [primary_doc]
    if primary_doc.lower().endswith(('.htm', '.html')):
        xml_variant = primary_doc.rsplit('.', 1)[0] + '.xml'
        candidates = [xml_variant, primary_doc]
        print(f"      [INFO] HTML primaryDoc — trying XML variant first: {xml_variant}")

    root = None
    tried = []
    for doc in candidates:
        url = f"https://www.sec.gov/Archives/edgar/data/{cik}/{acc_clean}/{doc}"
        r   = safe_get(url)
        tried.append(doc)
        if not r:
            print(f"      [WARN] HTTP failed for {doc}")
            continue
        try:
            root = ET.fromstring(r.content)
            print(f"      [INFO] Parsed successfully: {doc}")
            break
        except ET.ParseError as e:
            print(f"      [WARN] XML parse error on {doc}: {e}")
            continue

    # Last resort: directory listing
    if root is None:
        print(f"      [INFO] Trying directory listing fallback...")
        xml_fname = find_xml_in_filing(cik, accession)
        if xml_fname and xml_fname not in tried:
            url = f"https://www.sec.gov/Archives/edgar/data/{cik}/{acc_clean}/{xml_fname}"
            r   = safe_get(url)
            if r:
                try:
                    root = ET.fromstring(r.content)
                    print(f"      [INFO] Directory fallback succeeded: {xml_fname}")
                except ET.ParseError as e:
                    return None, f"All fallbacks exhausted. Last error: {e}"
            else:
                return None, "HTTP failed on directory fallback"
        elif xml_fname in tried:
            return None, "Directory fallback returned same filename already tried"
        else:
            return None, "No XML found in filing directory"

    if root is None:
        return None, "Could not obtain valid XML root"

    owner_name = ''
    is_dir = is_off = is_10pct = '0'
    officer_title = ''
    for rpt in root.findall('.//reportingOwner'):
        owner_name    = rpt.findtext('.//rptOwnerName', '').strip()
        rel = rpt.find('.//reportingOwnerRelationship')
        if rel is not None:
            is_dir        = rel.findtext('isDirector',        '0').strip()
            is_off        = rel.findtext('isOfficer',         '0').strip()
            is_10pct      = rel.findtext('isTenPercentOwner', '0').strip()
            officer_title = rel.findtext('officerTitle',      '').strip()

    records = []
    for txn in root.findall('.//nonDerivativeTransaction'):
        records.append({
            'cik':               cik,
            'ticker':            ticker,
            'filing_date':       filing_date,
            'accession':         accession,
            'owner_name':        owner_name,
            'is_director':       is_dir,
            'is_officer':        is_off,
            'is_ten_pct_owner':  is_10pct,
            'officer_title':     officer_title,
            'security_title':    txn.findtext('.//securityTitle/value',  '').strip(),
            'transaction_date':  txn.findtext('.//transactionDate/value', '').strip(),
            'transaction_code':  txn.findtext('.//transactionCoding/transactionCode', '').strip(),
            'acquired_disposed': txn.findtext(
                './/transactionAmounts/transactionAcquiredDisposedCode/value', '').strip(),
            'shares':            txn.findtext(
                './/transactionAmounts/transactionShares/value', ''),
            'price_per_share':   txn.findtext(
                './/transactionAmounts/transactionPricePerShare/value', ''),
            'shares_after_txn':  txn.findtext(
                './/postTransactionAmounts/sharesOwnedFollowingTransaction/value', ''),
        })
    return records, None

txn_rows   = []
xml_ok     = 0
xml_fail   = 0
xml_empty  = 0

# Group by CIK and limit to MAX_XML_PER_CIK each
from collections import defaultdict
by_cik = defaultdict(list)
for cik, ticker, f in all_test_filings:
    by_cik[cik].append((ticker, f))

for cik, items in by_cik.items():
    ticker  = items[0][0]
    sample  = items[:MAX_XML_PER_CIK]
    print(f"\n  {ticker} — parsing {len(sample)} XML(s):")
    for _, f in sample:
        records, err = parse_form4(cik, ticker, f['accession'], f['primary_doc'], f['filing_date'])
        if err:
            print(f"    {FAIL} {f['accession']}: {err}")
            xml_fail += 1
        elif records is None:
            print(f"    {FAIL} {f['accession']}: no result returned")
            xml_fail += 1
        elif len(records) == 0:
            print(f"    {INFO} {f['accession']}: parsed OK but 0 non-derivative transactions (options only filing)")
            xml_empty += 1
            xml_ok += 1
        else:
            r0 = records[0]
            print(f"    {PASS} {f['accession']}: {len(records)} transaction(s)")
            print(f"         Insider : {r0['owner_name']} | Title: {r0['officer_title']}")
            print(f"         Date    : {r0['transaction_date']} | Code: {r0['transaction_code']} | A/D: {r0['acquired_disposed']}")
            print(f"         Shares  : {r0['shares']} @ ${r0['price_per_share']}")
            xml_ok += 1
            txn_rows.extend(records)


# ── Test 4: Output validation ──────────────────────────────────────────────────
print()
print("=" * 60)
print("TEST 4 — Output validation")
print("=" * 60)

if txn_rows:
    txn_df = pd.DataFrame(txn_rows)

    # Column check
    missing = EXPECTED_TXN_COLS - set(txn_df.columns)
    if not missing:
        print(f"{PASS} All expected transaction columns present")
    else:
        print(f"{FAIL} Missing columns: {missing}")

    # A/D code check
    valid_ad = txn_df['acquired_disposed'].isin(['A', 'D', ''])
    if valid_ad.all():
        print(f"{PASS} acquired_disposed values all valid (A/D/blank)")
    else:
        bad = txn_df[~valid_ad]['acquired_disposed'].unique()
        print(f"{FAIL} Unexpected acquired_disposed values: {bad}")

    # Transaction code check
    known_codes = {'P', 'S', 'A', 'D', 'F', 'M', 'G', 'C', 'E', 'H', 'I', 'J', 'L', 'O', 'U', 'W', 'X', 'Z', ''}
    bad_codes = set(txn_df['transaction_code'].unique()) - known_codes
    if not bad_codes:
        print(f"{PASS} Transaction codes all recognised: {sorted(txn_df['transaction_code'].unique())}")
    else:
        print(f"{FAIL} Unknown transaction codes: {bad_codes}")

    # Date format check
    try:
        pd.to_datetime(txn_df['transaction_date'])
        print(f"{PASS} transaction_date parses as dates OK")
    except Exception:
        print(f"{FAIL} transaction_date has unparseable values")

    # Sample rows
    print(f"\n{INFO} Sample transactions (first 5 rows):")
    print(txn_df[['ticker', 'owner_name', 'officer_title',
                  'transaction_date', 'transaction_code',
                  'acquired_disposed', 'shares', 'price_per_share']].head().to_string(index=False))
else:
    print(f"{INFO} No non-derivative transactions found in sampled filings (may be options-only batch)")


# ── Summary ────────────────────────────────────────────────────────────────────
print()
print("=" * 60)
print("SUMMARY")
print("=" * 60)
print(f"  XMLs parsed successfully : {xml_ok}")
print(f"  XMLs failed              : {xml_fail}")
print(f"  Filings with 0 non-deriv : {xml_empty}")
print(f"  Transaction rows parsed  : {len(txn_rows)}")
print()
if xml_fail == 0:
    print("  ALL TESTS PASSED — pipeline is working correctly.")
    print("  Safe to run the full collect_form4.py overnight.")
else:
    print(f"  {xml_fail} XML(s) failed — check errors above before full run.")
print("=" * 60)