23ecdc659d
Some checks failed
WEVAL NonReg / nonreg (push) Has been cancelled
NOUVEAU endpoint /api/wtp-pages-enriched.php (zero ecrasement WTP): - 340 pages categorisees en 10 categories - 219 avec thumb (64.4% coverage) via md5(path) lookup - 37 avec meta description extraite - Titles HTML extraits - Categories: Hubs/Dashboards/Agents/Admin/Business/WEVADS/APIs/Proofs/Public/Autre WTP renderAllPages() peut consommer cette API pour cards enrichies. Modification WTP (chattr +i 5489L) attend validation Yacine ou WEVIA autonomy. Zero regression. Zero ecrasement. NR invariant. Doctrine 155 ecrite au vault.
120 lines
4.5 KiB
PHP
120 lines
4.5 KiB
PHP
<?php
|
|
/**
|
|
* wtp-pages-enriched.php v2 — Doctrine 155
|
|
* Endpoint enrichi pour pages WTP: thumb_url (md5 path), title, description, category
|
|
* Thumbs conventions: wem-thumb-worker key = path, hash = md5(path)
|
|
*/
|
|
header('Content-Type: application/json');
|
|
header('Cache-Control: public, max-age=300');
|
|
|
|
$HTML_ROOT = '/var/www/html';
|
|
$THUMB_DIR = $HTML_ROOT . '/api/screenshots/wem';
|
|
$THUMB_URL = '/api/screenshots/wem';
|
|
|
|
function getPageMeta($path) {
|
|
if (!file_exists($path)) return null;
|
|
$size = min(filesize($path), 50000);
|
|
$fp = fopen($path, 'r');
|
|
$content = fread($fp, $size);
|
|
fclose($fp);
|
|
$title = ''; $desc = '';
|
|
if (preg_match('/<title[^>]*>([^<]+)<\/title>/i', $content, $m)) {
|
|
$title = trim(html_entity_decode($m[1]));
|
|
}
|
|
if (preg_match('/<meta[^>]+name=["\']description["\'][^>]+content=["\']([^"\']+)["\']/i', $content, $m)) {
|
|
$desc = trim(html_entity_decode($m[1]));
|
|
}
|
|
return [
|
|
'title' => mb_substr($title, 0, 120),
|
|
'description' => mb_substr($desc, 0, 200),
|
|
'last_modified' => date('Y-m-d', filemtime($path))
|
|
];
|
|
}
|
|
|
|
function findThumbMd5($name, $thumbDir, $thumbUrl) {
|
|
// wem-thumb-worker.py: key = path (name without leading /), hash = md5(key)
|
|
$hash = md5($name);
|
|
$thumb = "$thumbDir/$hash.png";
|
|
if (file_exists($thumb)) return "$thumbUrl/$hash.png";
|
|
return null;
|
|
}
|
|
|
|
$start = microtime(true);
|
|
$html_files = glob($HTML_ROOT . '/*.html');
|
|
$all_pages = [];
|
|
$total = 0; $with_thumb = 0; $with_meta = 0;
|
|
|
|
foreach ($html_files as $f) {
|
|
$name = basename($f);
|
|
if (strpos($name, '-gold') !== false || strpos($name, '.bak') !== false) continue;
|
|
$meta = getPageMeta($f);
|
|
$thumb = findThumbMd5($name, $THUMB_DIR, $THUMB_URL);
|
|
$page = [
|
|
'name' => $name,
|
|
'url' => '/' . $name,
|
|
'size_kb' => round(filesize($f) / 1024, 1),
|
|
'thumb_url' => $thumb,
|
|
'title' => $meta['title'] ?: str_replace(['-', '_', '.html'], [' ', ' ', ''], $name),
|
|
'description' => $meta['description'] ?? '',
|
|
'last_modified' => $meta['last_modified'] ?? date('Y-m-d', filemtime($f)),
|
|
];
|
|
$all_pages[] = $page;
|
|
$total++;
|
|
if ($thumb) $with_thumb++;
|
|
if (!empty($meta['description'])) $with_meta++;
|
|
}
|
|
|
|
// Categorization improved
|
|
$categories = [
|
|
'Hubs & Centers' => [],
|
|
'Dashboards & Monitoring' => [],
|
|
'Agents & AI' => [],
|
|
'Admin & Ops' => [],
|
|
'Business & CRM' => [],
|
|
'WEVADS Email Legacy' => [],
|
|
'Integration & APIs' => [],
|
|
'Proofs & Tests' => [],
|
|
'Public & Marketing' => [],
|
|
'Autre' => [],
|
|
];
|
|
|
|
foreach ($all_pages as $p) {
|
|
$n = strtolower($p['name']);
|
|
if (preg_match('/hub|center\b|registry/', $n)) $categories['Hubs & Centers'][] = $p;
|
|
elseif (preg_match('/dashboard|monitor|metric|kpi|analytics/', $n)) $categories['Dashboards & Monitoring'][] = $p;
|
|
elseif (preg_match('/agent|bot|brain|cortex|l99|wevia|claude|ia-|-ia|gpt/', $n)) $categories['Agents & AI'][] = $p;
|
|
elseif (preg_match('/admin|ops|config|setup|install|deploy/', $n)) $categories['Admin & Ops'][] = $p;
|
|
elseif (preg_match('/crm|sales|customer|client|lead|opportunity/', $n)) $categories['Business & CRM'][] = $p;
|
|
elseif (preg_match('/wevads|pmta|email-|-email|smtp|mail-|deliv/', $n)) $categories['WEVADS Email Legacy'][] = $p;
|
|
elseif (preg_match('/api-|integration|connect|webhook|bridge/', $n)) $categories['Integration & APIs'][] = $p;
|
|
elseif (preg_match('/proof|test-|-test|playwright|e2e|verify/', $n)) $categories['Proofs & Tests'][] = $p;
|
|
elseif (preg_match('/landing|home|about|contact|pricing|index|intro/', $n)) $categories['Public & Marketing'][] = $p;
|
|
else $categories['Autre'][] = $p;
|
|
}
|
|
|
|
// Sort each category by size desc
|
|
foreach ($categories as $k => &$v) {
|
|
usort($v, fn($a, $b) => $b['size_kb'] <=> $a['size_kb']);
|
|
}
|
|
unset($v);
|
|
|
|
$elapsed = round((microtime(true) - $start) * 1000, 1);
|
|
|
|
echo json_encode([
|
|
'ok' => true,
|
|
'doctrine' => '155',
|
|
'ts' => date('c'),
|
|
'elapsed_ms' => $elapsed,
|
|
'stats' => [
|
|
'total_pages' => $total,
|
|
'with_thumb' => $with_thumb,
|
|
'with_meta' => $with_meta,
|
|
'thumb_coverage_pct' => round($with_thumb / max(1, $total) * 100, 1),
|
|
'meta_coverage_pct' => round($with_meta / max(1, $total) * 100, 1),
|
|
'categories_count' => count(array_filter($categories)),
|
|
'categories_populated' => array_filter($categories, fn($v) => count($v) > 0) ? count(array_filter($categories, fn($v) => count($v) > 0)) : 0,
|
|
],
|
|
'category_counts' => array_map('count', $categories),
|
|
'categories' => $categories,
|
|
], JSON_UNESCAPED_SLASHES);
|