<?php
/**
 * Technical SEO audit of a sample of live Nepali websites.
 *
 * Written to support an article. Every claim published from this has to be
 * something the reader could re-measure themselves, so this only records
 * signals that are objectively present or absent in the HTTP response and the
 * HTML — status codes, headers, meta tags, heading counts, script counts,
 * transfer sizes. It makes no judgement about content quality, keyword choice
 * or ranking, because those are not measurable this way.
 *
 * Sample construction: sites were taken from public search results across six
 * sectors. It is a convenience sample, not a random one, and the write-up must
 * say so.
 *
 *   php audit-nepali-sites.php            # run and write results.json
 *   php audit-nepali-sites.php --report   # summarise an existing results.json
 */

const OUT = __DIR__ . '/nepali-site-audit.json';

$SITES = [
    // sector => [urls]
    'trekking / tourism' => [
        'https://www.nepalhikingteam.com/',
        'https://www.sevensummittreks.com/',
        'https://www.welcomenepaltreks.com/',
        'https://www.greenvalleynepaltreks.com/',
        'https://www.thenepaltrekkingcompany.com/',
        'https://taan.org.np/',
        'https://ntb.gov.np/',
    ],
    'health / clinics' => [
        'https://www.kdh.com.np/',
        'https://dent-inn.com/',
        'https://www.smilenepal.com/',
        'https://kantipurdental.edu.np/',
        'https://www.phectnepal.org/',
    ],
    'hospitality' => [
        'https://hotelthamel.com/',
        'https://www.hotelyakandyeti.com/',
        'https://www.hotelhimalaya.com/',
        'https://www.dwarikas.com/',
    ],
    'news / media' => [
        'https://www.onlinekhabar.com/',
        'https://ekantipur.com/',
        'https://www.setopati.com/',
        'https://myrepublica.nagariknetwork.com/',
        'https://kathmandupost.com/',
        'https://english.insurancekhabar.com/',
        'https://bankingnews.com.np/',
    ],
    'finance / commerce' => [
        'https://www.sharesansar.com/',
        'https://www.nrb.org.np/',
        'https://daraz.com.np/',
        'https://www.hamrobazaar.com/',
        'https://esewa.com.np/',
        'https://khalti.com/',
    ],
    'education / institutions' => [
        'https://tribhuvan-university.edu.np/',
        'https://ku.edu.np/',
        'https://www.pu.edu.np/',
        'https://moe.gov.np/',
    ],
    'seo & digital agencies' => [
        'https://seoexpertinnepal.com/',
        'https://aviralacharya.com.np/',
        'https://www.clickbenz.com/',
        'https://krizmatic.com/',
        'https://kathmanduinfotech.com/',
        'https://dipakpaneru.com.np/',
        'https://baamaconsultant.com.np/',
        'https://pigeonwebsolutions.com/',
    ],
];

/* ------------------------------------------------------------------ */

function fetchOne(string $url): array {
    $ch = curl_init($url);
    curl_setopt_array($ch, [
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_FOLLOWLOCATION => true,
        CURLOPT_MAXREDIRS      => 5,
        CURLOPT_HEADER         => true,
        CURLOPT_TIMEOUT        => 30,
        CURLOPT_CONNECTTIMEOUT => 12,
        CURLOPT_SSL_VERIFYPEER => false,   // we record TLS problems, not enforce them
        CURLOPT_ENCODING       => '',      // advertise gzip/br so we can see if it is used
        CURLOPT_USERAGENT      => 'Mozilla/5.0 (Linux; Android 12; SM-A125F) AppleWebKit/537.36 '
                                . '(KHTML, like Gecko) Chrome/120 Mobile Safari/537.36',
    ]);
    $raw = curl_exec($ch);
    $err = curl_error($ch);
    $info = curl_getinfo($ch);
    curl_close($ch);

    if ($raw === false) {
        return ['ok' => false, 'error' => $err, 'url' => $url];
    }

    $hsize = $info['header_size'];
    return [
        'ok'          => true,
        'url'         => $url,
        'final_url'   => $info['url'],
        'code'        => $info['http_code'],
        'redirects'   => $info['redirect_count'],
        'ttfb'        => $info['starttransfer_time'],
        'total_time'  => $info['total_time'],
        'size'        => $info['size_download'],
        'headers'     => substr($raw, 0, $hsize),
        'body'        => substr($raw, $hsize),
    ];
}

function analyse(array $r): array {
    $h = $r['headers'];
    $b = $r['body'];
    $out = [];

    $out['url']        = $r['url'];
    $out['final_url']  = $r['final_url'];
    $out['code']       = $r['code'];
    $out['redirects']  = $r['redirects'];
    $out['ttfb_ms']    = (int)round($r['ttfb'] * 1000);
    $out['html_bytes'] = $r['size'];

    // HTTPS actually enforced (the final URL after following redirects)
    $out['https'] = str_starts_with($r['final_url'], 'https://');

    // Transfer compression
    $out['compressed'] = (bool)preg_match('#^content-encoding:\s*(gzip|br|deflate|zstd)#mi', $h);

    // <html lang>
    preg_match('#<html[^>]*\blang=["\']([^"\']+)#i', $b, $m);
    $out['lang'] = $m[1] ?? null;

    // Mobile viewport
    $out['viewport'] = (bool)preg_match('#<meta[^>]+name=["\']viewport["\']#i', $b);

    // Title
    preg_match('#<title[^>]*>(.*?)</title>#is', $b, $m);
    $title = isset($m[1]) ? trim(html_entity_decode(strip_tags($m[1]), ENT_QUOTES, 'UTF-8')) : '';
    $out['title'] = $title;
    $out['title_len'] = mb_strlen($title);

    // Meta description
    preg_match('#<meta[^>]+name=["\']description["\'][^>]*content=["\']([^"\']*)#i', $b, $m);
    $desc = isset($m[1]) ? trim(html_entity_decode($m[1], ENT_QUOTES, 'UTF-8')) : '';
    $out['description'] = $desc;
    $out['desc_len'] = mb_strlen($desc);

    // Canonical
    $out['canonical'] = (bool)preg_match('#<link[^>]+rel=["\']canonical["\']#i', $b);

    // Headings
    $out['h1_count'] = preg_match_all('#<h1[\s>]#i', $b);
    $out['h2_count'] = preg_match_all('#<h2[\s>]#i', $b);

    // Open Graph / Twitter
    $out['og'] = (bool)preg_match('#<meta[^>]+property=["\']og:title["\']#i', $b);
    $out['og_image'] = (bool)preg_match('#<meta[^>]+property=["\']og:image["\']#i', $b);
    $out['twitter'] = (bool)preg_match('#<meta[^>]+name=["\']twitter:card["\']#i', $b);

    // Structured data
    preg_match_all('#<script[^>]+application/ld\+json[^>]*>(.*?)</script>#is', $b, $ld);
    $out['jsonld_blocks'] = count($ld[1]);
    $valid = 0; $types = [];
    foreach ($ld[1] as $j) {
        $d = json_decode(trim($j), true);
        if (json_last_error() === JSON_ERROR_NONE) {
            $valid++;
            $flat = json_encode($d);
            preg_match_all('#"@type"\s*:\s*"([A-Za-z]+)"#', $flat, $t);
            $types = array_merge($types, $t[1]);
        }
    }
    $out['jsonld_valid'] = $valid;
    $out['jsonld_invalid'] = count($ld[1]) - $valid;
    $out['jsonld_types'] = array_values(array_unique($types));

    // Images
    preg_match_all('#<img\b[^>]*>#i', $b, $imgs);
    $out['img_count'] = count($imgs[0]);
    $noAlt = 0; $noDim = 0; $lazy = 0;
    foreach ($imgs[0] as $tag) {
        if (!preg_match('#\balt\s*=\s*["\'][^"\']+["\']#i', $tag)) $noAlt++;
        if (!preg_match('#\bwidth\s*=#i', $tag) || !preg_match('#\bheight\s*=#i', $tag)) $noDim++;
        if (preg_match('#\bloading\s*=\s*["\']lazy#i', $tag)) $lazy++;
    }
    $out['img_no_alt'] = $noAlt;
    $out['img_no_dimensions'] = $noDim;
    $out['img_lazy'] = $lazy;

    // Render-blocking resources in <head>
    $head = '';
    if (preg_match('#<head\b.*?</head>#is', $b, $m)) $head = $m[0];
    $out['head_scripts_blocking'] = preg_match_all('#<script\b(?![^>]*\b(async|defer|type=["\']application/ld)\b)[^>]*\bsrc=#i', $head);
    $out['head_stylesheets'] = preg_match_all('#<link[^>]+rel=["\']stylesheet["\']#i', $head);

    // Platform fingerprint
    $out['wordpress'] = (bool)(preg_match('#/wp-content/|/wp-includes/#i', $b));

    return $out;
}

/* ------------------------------------------------------------------ */

if (in_array('--report', $argv, true)) {
    $data = json_decode(file_get_contents(OUT), true);
} else {
    $data = ['generated' => date('c'), 'sites' => []];
    foreach ($SITES as $sector => $urls) {
        foreach ($urls as $u) {
            fwrite(STDERR, "  fetching $u ... ");
            $r = fetchOne($u);
            if (!$r['ok'] || $r['code'] >= 400 || $r['size'] === 0) {
                fwrite(STDERR, "SKIP (" . ($r['error'] ?? 'HTTP ' . $r['code']) . ")\n");
                continue;
            }
            $a = analyse($r);
            $a['sector'] = $sector;
            $data['sites'][] = $a;
            fwrite(STDERR, "ok (HTTP {$a['code']}, {$a['ttfb_ms']}ms)\n");
        }
    }
    file_put_contents(OUT, json_encode($data, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE));
    fwrite(STDERR, "\nwrote " . OUT . "\n\n");
}

/* ---------------- summary ---------------- */

$s = $data['sites'];
$n = count($s);
if (!$n) { echo "no sites captured\n"; exit(1); }

function pct(int $k, int $n): string { return sprintf('%d/%d (%d%%)', $k, $n, round($k / $n * 100)); }
function count_where(array $s, callable $f): int { return count(array_filter($s, $f)); }

echo "NEPALI WEBSITE TECHNICAL AUDIT\n";
echo "sample: $n live sites, " . count(array_unique(array_column($s, 'sector'))) . " sectors\n";
echo "run: " . $data['generated'] . "\n";
echo str_repeat('=', 62) . "\n\n";

$checks = [
    'serves over HTTPS'                 => fn($x) => $x['https'],
    'has a mobile viewport tag'         => fn($x) => $x['viewport'],
    'compresses HTML'                   => fn($x) => $x['compressed'],
    'has a <title>'                     => fn($x) => $x['title_len'] > 0,
    'title within 10-60 chars'          => fn($x) => $x['title_len'] >= 10 && $x['title_len'] <= 60,
    'has a meta description'            => fn($x) => $x['desc_len'] > 0,
    'description within 50-160 chars'   => fn($x) => $x['desc_len'] >= 50 && $x['desc_len'] <= 160,
    'has a canonical tag'               => fn($x) => $x['canonical'],
    'has exactly one <h1>'              => fn($x) => $x['h1_count'] === 1,
    'has NO <h1> at all'                => fn($x) => $x['h1_count'] === 0,
    'has an <html lang> attribute'      => fn($x) => !empty($x['lang']),
    'has Open Graph title'              => fn($x) => $x['og'],
    'has an og:image'                   => fn($x) => $x['og_image'],
    'has any JSON-LD'                   => fn($x) => $x['jsonld_blocks'] > 0,
    'has BROKEN JSON-LD'                => fn($x) => $x['jsonld_invalid'] > 0,
    'every image has alt text'          => fn($x) => $x['img_count'] > 0 && $x['img_no_alt'] === 0,
    'every image has width+height'      => fn($x) => $x['img_count'] > 0 && $x['img_no_dimensions'] === 0,
    'runs WordPress'                    => fn($x) => $x['wordpress'],
];

foreach ($checks as $label => $fn) {
    printf("  %-34s %s\n", $label, pct(count_where($s, $fn), $n));
}

/* distributions */
$ttfb = array_column($s, 'ttfb_ms');
sort($ttfb);
$bytes = array_column($s, 'html_bytes');
sort($bytes);
$blocking = array_column($s, 'head_scripts_blocking');
sort($blocking);

function median(array $a) { $c = count($a); return $c % 2 ? $a[intdiv($c,2)] : ($a[$c/2-1]+$a[$c/2])/2; }

echo "\n";
printf("  median TTFB                        %d ms  (range %d-%d)\n", median($ttfb), $ttfb[0], end($ttfb));
printf("  median HTML transferred            %.0f KB (range %.0f-%.0f)\n",
    median($bytes)/1024, $bytes[0]/1024, end($bytes)/1024);
printf("  median render-blocking <head> JS   %d      (worst %d)\n", median($blocking), end($blocking));

$noAlt = array_sum(array_column($s, 'img_no_alt'));
$imgs  = array_sum(array_column($s, 'img_count'));
printf("  images missing alt text            %d of %d (%d%%)\n", $noAlt, $imgs, $imgs ? round($noAlt/$imgs*100) : 0);

$noDim = array_sum(array_column($s, 'img_no_dimensions'));
printf("  images missing width/height        %d of %d (%d%%)\n", $noDim, $imgs, $imgs ? round($noDim/$imgs*100) : 0);

/* language handling */
echo "\n  <html lang> values in use:\n";
$langs = array_count_values(array_filter(array_column($s, 'lang')));
arsort($langs);
foreach ($langs as $l => $c) printf("    %-12s %d\n", $l, $c);

/* schema types */
echo "\n  schema types found across the sample:\n";
$all = [];
foreach ($s as $x) foreach ($x['jsonld_types'] as $t) $all[$t] = ($all[$t] ?? 0) + 1;
arsort($all);
$i = 0;
foreach ($all as $t => $c) { printf("    %-22s %d\n", $t, $c); if (++$i >= 12) break; }

/* per-sector */
echo "\n  by sector (one h1 / canonical / any schema):\n";
$sectors = [];
foreach ($s as $x) $sectors[$x['sector']][] = $x;
foreach ($sectors as $name => $rows) {
    printf("    %-26s n=%-3d h1:%-6s canon:%-6s schema:%s\n", $name, count($rows),
        count_where($rows, fn($x) => $x['h1_count'] === 1) . '/' . count($rows),
        count_where($rows, fn($x) => $x['canonical']) . '/' . count($rows),
        count_where($rows, fn($x) => $x['jsonld_blocks'] > 0) . '/' . count($rows));
}
echo "\n";