# -*- coding: utf-8 -*-
"""The Greek e-shop technical audit: classify, then measure.

Sample frame: Tranco top-1M (list PYG5J, 2026-08-06) filtered to .gr, top N by rank.
A site counts as an e-shop if it shows a commerce platform signature, or Product/Offer
structured data, or cart wording. All three are recorded so the rule can be varied.
"""
import collections
import io
import json
import os
import statistics

HERE = os.path.dirname(os.path.abspath(__file__))
rows = [json.loads(l) for l in io.open(os.path.join(HERE, 'gr_clean.jsonl'), encoding='utf-8') if l.strip()]

reached = [r for r in rows if r.get('status') == 200]
shops = [r for r in reached if r.get('platform') or r.get('jsonld_product') or r.get('cart_signal')]
strict = [r for r in reached if r.get('platform') or r.get('jsonld_product')]


def pct(n, d):
    return (100.0 * n / d) if d else 0.0


print('=' * 92)
print('SAMPLE')
print('=' * 92)
print('  domains attempted            %d' % len(rows))
print('  responded HTTP 200           %d  (%.0f%%)' % (len(reached), pct(len(reached), len(rows))))
err = collections.Counter((r.get('error') or '').split(':')[0] for r in rows if r.get('status') != 200)
print('  failures by kind             %s' % dict(err.most_common(6)))
print('  classified as e-shops        %d  (%.0f%% of reachable)  [broad rule]'
      % (len(shops), pct(len(shops), len(reached))))
print('  classified as e-shops        %d  (%.0f%%)  [strict: platform or Product schema]'
      % (len(strict), pct(len(strict), len(reached))))

S = shops
print()
print('=' * 92)
print('WHAT GREEK E-SHOPS ARE BUILT ON   (n=%d)' % len(S))
print('=' * 92)
plat = collections.Counter(r.get('platform') or '(none detected)' for r in S)
for k, v in plat.most_common():
    print('  %-22s %4d  %5.1f%%' % (k, v, pct(v, len(S))))
cms = collections.Counter(r.get('cms') or '(none detected)' for r in S)
print('  --- underlying CMS ---')
for k, v in cms.most_common(6):
    print('  %-22s %4d  %5.1f%%' % (k, v, pct(v, len(S))))

print()
print('=' * 92)
print('SECURITY HEADERS   (share of e-shops sending each)')
print('=' * 92)
H = [('strict-transport-security', 'HSTS'), ('content-security-policy', 'CSP'),
     ('x-frame-options', 'X-Frame-Options'), ('x-content-type-options', 'X-Content-Type-Options'),
     ('referrer-policy', 'Referrer-Policy'), ('permissions-policy', 'Permissions-Policy')]
for k, label in H:
    n = sum(1 for r in S if (r.get('headers') or {}).get(k))
    print('  %-26s %4d / %d  %5.1f%%' % (label, n, len(S), pct(n, len(S))))
nz = sum(1 for r in S if not any((r.get('headers') or {}).get(k) for k, _ in H))
print('  %-26s %4d / %d  %5.1f%%' % ('NONE OF THE SIX', nz, len(S), pct(nz, len(S))))

print()
print('=' * 92)
print('TRANSPORT')
print('=' * 92)
tls = collections.Counter(r.get('tls_version') or '(handshake failed)' for r in S)
for k, v in tls.most_common():
    print('  %-22s %4d  %5.1f%%' % (k, v, pct(v, len(S))))
bad = sum(1 for r in S if not r.get('tls_valid'))
print('  certificate not valid to a default trust store: %d (%.1f%%)' % (bad, pct(bad, len(S))))

print()
print('=' * 92)
print('BASELINE HYGIENE')
print('=' * 92)
for label, f in [('mobile viewport declared', lambda r: r.get('viewport')),
                 ('html lang attribute', lambda r: r.get('lang_attr')),
                 ('exactly one <h1>', lambda r: r.get('h1') == 1),
                 ('no <h1> at all', lambda r: r.get('h1') == 0),
                 ('Product/Offer structured data', lambda r: r.get('jsonld_product')),
                 ('cookie-consent machinery', lambda r: r.get('consent')),
                 ('response compressed', lambda r: 'gzip' in ((r.get('headers') or {}).get('content-encoding') or '')
                  or 'br' in ((r.get('headers') or {}).get('content-encoding') or ''))]:
    n = sum(1 for r in S if f(r))
    print('  %-32s %4d / %d  %5.1f%%' % (label, n, len(S), pct(n, len(S))))

print()
print('=' * 92)
print('IMAGE ALT TEXT   (accessibility, on the homepage only)')
print('=' * 92)
withimg = [r for r in S if r.get('imgs')]
cov = [100.0 * r['imgs_alt'] / r['imgs'] for r in withimg]
if cov:
    cov.sort()
    print('  e-shops with any <img>        %d' % len(withimg))
    print('  median alt coverage           %.0f%%' % statistics.median(cov))
    print('  e-shops at 100%% coverage      %d  (%.1f%%)'
          % (sum(1 for c in cov if c >= 99.5), pct(sum(1 for c in cov if c >= 99.5), len(cov))))
    print('  e-shops below 50%% coverage    %d  (%.1f%%)'
          % (sum(1 for c in cov if c < 50), pct(sum(1 for c in cov if c < 50), len(cov))))
    print('  e-shops at 0%%                 %d  (%.1f%%)'
          % (sum(1 for c in cov if c == 0), pct(sum(1 for c in cov if c == 0), len(cov))))

print()
print('=' * 92)
print('WEIGHT AND SPEED   (homepage HTML only, single unwarmed request)')
print('=' * 92)
for label, key in [('response ms', 'elapsed_ms'), ('bytes on the wire', 'bytes'),
                   ('HTML bytes after decompression', 'html_bytes'),
                   ('<script> tags', 'scripts')]:
    v = sorted(r[key] for r in S if r.get(key) is not None)
    if v:
        print('  %-32s median %8.0f   p90 %8.0f   max %9.0f'
              % (label, statistics.median(v), v[int(len(v) * 0.9)], v[-1]))

print()
print('=' * 92)
print('THE COMPOUND FAILURE   (e-shops failing more than one basic check)')
print('=' * 92)


def fails(r):
    f = []
    if not any((r.get('headers') or {}).get(k) for k, _ in H):
        f.append('no security headers')
    if not r.get('viewport'):
        f.append('no mobile viewport')
    if not r.get('tls_valid'):
        f.append('TLS not valid')
    if r.get('imgs') and 100.0 * r['imgs_alt'] / r['imgs'] < 50:
        f.append('alt text under 50%')
    if not r.get('jsonld_product'):
        f.append('no product structured data')
    return f


dist = collections.Counter(len(fails(r)) for r in S)
for k in sorted(dist):
    print('  %d of 5 checks failed        %4d  %5.1f%%' % (k, dist[k], pct(dist[k], len(S))))