'
@@ -184,6 +209,76 @@ function trunc(s, n) {
el.innerHTML = html;
}());
+// ── Liste consolidée de tous les agents ──────────────────────────────────────
+(function () {
+ var el = document.getElementById('stats-agents-container');
+ var badge = document.getElementById('agents-count');
+ var ipData = (typeof FOLIO_IP_DATA !== 'undefined') ? FOLIO_IP_DATA : {};
+ if (!el) { return; }
+
+ // Agréger toutes les UAs depuis FOLIO_IP_DATA
+ var uaCounts = {};
+ Object.keys(ipData).forEach(function (ip) {
+ (ipData[ip].agents || []).forEach(function (ua) {
+ uaCounts[ua] = (uaCounts[ua] || 0) + (ipData[ip].hits || 0);
+ });
+ });
+
+ var agents = Object.keys(uaCounts).map(function (ua) {
+ return { ua: ua, hits: uaCounts[ua], bot: isBot(ua) };
+ }).sort(function (a, b) {
+ // Bots d'abord, puis par hits desc
+ if (a.bot !== b.bot) { return a.bot ? -1 : 1; }
+ return b.hits - a.hits;
+ });
+
+ if (!agents.length) {
+ el.innerHTML = '
Aucun agent détecté.
';
+ return;
+ }
+
+ var bots = agents.filter(function (a) { return a.bot; });
+ var unknown = agents.filter(function (a) { return !a.bot; });
+ if (badge) { badge.textContent = '— ' + bots.length + ' bot(s) détecté(s) sur ' + agents.length; }
+
+ function agentRow(a) {
+ return '
'
+ + '| '
+ + (a.bot ? '🤖' : '?') + ' | '
+ + ''
+ + '' + esc(a.ua) + ' | '
+ + ''
+ + a.hits.toLocaleString('fr-FR') + ' | '
+ + '
';
+ }
+
+ var botsHtml = bots.map(agentRow).join('');
+ var unknownHtml = unknown.map(agentRow).join('');
+
+ var html = '
'
+ + '
'
+ + ''
+ + ' | '
+ + 'User-Agent | '
+ + 'Req. | '
+ + '
'
+ + '';
+
+ if (botsHtml) {
+ html += '| '
+ + 'Bots connus (' + bots.length + ') |
'
+ + botsHtml;
+ }
+ if (unknownHtml) {
+ html += '| '
+ + 'Agents non classés (' + unknown.length + ') |
'
+ + unknownHtml;
+ }
+
+ html += '
';
+ el.innerHTML = html;
+}());
+
// ── Pages les plus visitées (RSS XML + sparklines) ───────────────────────────
(function () {
var container = document.getElementById('stats-pages-container');
@@ -303,9 +398,7 @@ function trunc(s, n) {
+ '
'
+ '
'
- + dots
- + yLabels
- + xLabels
+ + dots + yLabels + xLabels
+ '';
}
diff --git a/public/index.php b/public/index.php
index 20a0b73..d469912 100644
--- a/public/index.php
+++ b/public/index.php
@@ -2768,6 +2768,37 @@ switch ($action) {
$adminData['as_groups'] = asGroups();
$adminData['stats_pages_by_day'] = $statsRaw['pages_by_day'] ?? [];
$adminData['stats_ip_data'] = $statsRaw['ip_data'] ?? [];
+
+ // Patterns de bots — initialisation si absent
+ $botsFile = DATA_PATH . '/bots.json';
+ if (!file_exists($botsFile)) {
+ $defaultBots = [
+ 'Googlebot','Googlebot-Image','Google-InspectionTool','Google-Extended',
+ 'bingbot','BingPreview','msnbot',
+ 'DuckDuckBot','DuckDuckGo-Favicons-Bot',
+ 'Baiduspider','YandexBot','YandexImages','YandexMetrika',
+ 'Applebot',
+ 'facebookexternalhit','facebot',
+ 'Twitterbot','LinkedInBot','Slackbot','TelegramBot','WhatsApp','Discordbot',
+ 'PetalBot','Bytespider','SogouSpider','SeznamBot','Exabot',
+ 'AhrefsBot','SemrushBot','MJ12bot','DotBot','rogerbot','BLEXBot','DataForSeoBot',
+ 'Screaming Frog SEO Spider',
+ 'ClaudeBot','GPTBot','Google-Extended','PerplexityBot','cohere-ai','anthropic-ai',
+ 'meta-externalagent','OAI-SearchBot','Amazonbot',
+ 'CCBot','ia_archiver','archive.org_bot',
+ 'NetcraftSurveyAgent',
+ 'python-requests','python-urllib','Python/',
+ 'curl/','wget/','Wget/',
+ 'Go-http-client/1','Java/','Apache-HttpClient','okhttp/',
+ 'Scrapy','HeadlessChrome','PhantomJS','Puppeteer','Playwright','Selenium',
+ 'UptimeRobot','Pingdom','StatusCake','Site24x7','GTmetrix',
+ 'Chrome-Lighthouse','PageSpeed','Zabbix','check_http',
+ 'libwww-perl','GuzzleHttp','masscan','zgrab','nuclei',
+ ];
+ @mkdir(dirname($botsFile), 0755, true);
+ @file_put_contents($botsFile, json_encode($defaultBots, JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT));
+ }
+ $adminData['bot_patterns'] = json_decode((string) file_get_contents($botsFile), true) ?: [];
}
if ($tab === 'categories') {
@@ -3206,6 +3237,20 @@ switch ($action) {
header('Location: /admin/stats?' . ($ok ? 'saved=1' : 'error=write'));
exit;
+ case 'admin_save_bots':
+ requireAuth();
+ if (!isAdmin() || $_SERVER['REQUEST_METHOD'] !== 'POST') {
+ http_response_code(403);
+ exit;
+ }
+ $botsFile = DATA_PATH . '/bots.json';
+ $patterns = array_values(array_unique(array_filter(
+ array_map('trim', explode("\n", (string) ($_POST['bot_patterns'] ?? '')))
+ )));
+ $ok = @file_put_contents($botsFile, json_encode($patterns, JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT)) !== false;
+ header('Location: /admin/stats?' . ($ok ? 'saved=1' : 'error=write'));
+ exit;
+
case 'admin_create_role':
requireAuth();
if (!isAdmin() || $_SERVER['REQUEST_METHOD'] !== 'POST') {
diff --git a/public/version.txt b/public/version.txt
index 14781be..599e1a1 100644
--- a/public/version.txt
+++ b/public/version.txt
@@ -1 +1 @@
-1.6.30
+1.6.31
diff --git a/src/AccessLogParser.php b/src/AccessLogParser.php
index 4aaf572..c140236 100644
--- a/src/AccessLogParser.php
+++ b/src/AccessLogParser.php
@@ -30,7 +30,15 @@ class AccessLogParser
}
/**
- * @return array{pages:array
,books:array,ips:array,pages_by_day:array>,ips_by_day:array>,ip_top_paths:array>,ip_agents:array>}
+ * @return array{
+ * pages:array,
+ * books:array,
+ * ips:array,
+ * pages_by_day:array>,
+ * ips_by_day:array>,
+ * ip_top_paths:array>,
+ * ip_agents:array>
+ * }
*/
public function stats(): array
{
@@ -47,15 +55,16 @@ class AccessLogParser
$cutoff = strtotime("-{$this->days} days midnight") ?: (time() - $this->days * 86400);
$pages = [];
$books = [];
- $ips = [];
+ $ips = []; // toutes requêtes publiques (tous chemins, tous statuts)
$dayPages = [];
- $ipDays = []; // [ip => [dayOffset => count]]
- $ipPaths = []; // [ip => [path => count]]
- $ipPathTs = []; // [ip => [path => last_timestamp]]
- $ipAgents = []; // [ip => [ua => count]]
+ $ipPaths = []; // chemins /post/ et /book/ avec statut 200 (pour les ts)
+ $ipPathTs = [];
+ $ipAllPaths = []; // tous chemins, tous statuts
+ $ipAllDays = []; // tous jours, tous statuts
+ $ipAgents = []; // tous user-agents par IP
foreach ($this->logFiles() as $file) {
- $this->parseFile($file, $cutoff, $pages, $books, $ips, $dayPages, $ipDays, $ipPaths, $ipPathTs, $ipAgents);
+ $this->parseFile($file, $cutoff, $pages, $books, $ips, $dayPages, $ipPaths, $ipPathTs, $ipAllPaths, $ipAllDays, $ipAgents);
}
arsort($pages);
@@ -73,27 +82,30 @@ class AccessLogParser
$pagesByDay[$path] = $arr;
}
- // Per-IP daily counts + top paths + top agents, limité aux 200 IPs les plus actives
+ // Top 200 IPs par volume total de requêtes
$topIpKeys = array_keys(array_slice($ips, 0, 200, true));
$ipsByDay = [];
$ipTopPaths = [];
$ipTopAgents = [];
foreach ($topIpKeys as $ip) {
+ // Sparkline : activité totale par jour
$arr = array_fill(0, $this->days, 0);
- foreach ($ipDays[$ip] ?? [] as $offset => $count) {
+ foreach ($ipAllDays[$ip] ?? [] as $offset => $count) {
if ($offset >= 0 && $offset < $this->days) {
$arr[$offset] = $count;
}
}
$ipsByDay[$ip] = $arr;
- $paths = $ipPaths[$ip] ?? [];
- arsort($paths);
+ // Top 20 chemins tous types confondus
+ $allPaths = $ipAllPaths[$ip] ?? [];
+ arsort($allPaths);
$ipTopPaths[$ip] = [];
- foreach (array_slice($paths, 0, 10, true) as $p => $cnt) {
+ foreach (array_slice($allPaths, 0, 20, true) as $p => $cnt) {
$ipTopPaths[$ip][$p] = ['n' => $cnt, 'ts' => $ipPathTs[$ip][$p] ?? 0];
}
+ // Top 5 user-agents
$agents = $ipAgents[$ip] ?? [];
arsort($agents);
$ipTopAgents[$ip] = array_keys(array_slice($agents, 0, 5, true));
@@ -163,16 +175,24 @@ class AccessLogParser
return (int) strtotime("{$m[1]} {$m[2]} {$m[3]} {$m[4]} {$m[5]}");
}
- private function parseLine(string $line, int $cutoff, array &$pages, array &$books, array &$ips, array &$dayPages, array &$ipDays, array &$ipPaths, array &$ipPathTs, array &$ipAgents): void
- {
+ private function parseLine(
+ string $line,
+ int $cutoff,
+ array &$pages,
+ array &$books,
+ array &$ips,
+ array &$dayPages,
+ array &$ipPaths,
+ array &$ipPathTs,
+ array &$ipAllPaths,
+ array &$ipAllDays,
+ array &$ipAgents
+ ): void {
if (!preg_match(self::RE, $line, $m)) {
return;
}
[, $ip, $ts, $path, $status, $ua] = $m;
- if ($status !== '200') {
- return;
- }
$tsVal = self::parseTimestamp($ts);
if ($tsVal < $cutoff) {
return;
@@ -181,38 +201,54 @@ class AccessLogParser
$publicIp = filter_var($ip, FILTER_VALIDATE_IP, FILTER_FLAG_NO_PRIV_RANGE | FILTER_FLAG_NO_RES_RANGE) !== false;
$dayOffset = (int) floor(($tsVal - $cutoff) / 86400);
- if (str_starts_with($path, '/post/') && strlen($path) > 6) {
- $pages[$path] = ($pages[$path] ?? 0) + 1;
- if ($publicIp) {
- $ips[$ip] = ($ips[$ip] ?? 0) + 1;
- }
- $dayPages[$path][$dayOffset] = ($dayPages[$path][$dayOffset] ?? 0) + 1;
- $ipDays[$ip][$dayOffset] = ($ipDays[$ip][$dayOffset] ?? 0) + 1;
- $ipPaths[$ip][$path] = ($ipPaths[$ip][$path] ?? 0) + 1;
- if ($tsVal > ($ipPathTs[$ip][$path] ?? 0)) {
- $ipPathTs[$ip][$path] = $tsVal;
- }
+ // Toutes les requêtes publiques : comptage global, chemins, jours, agents
+ if ($publicIp) {
+ $ips[$ip] = ($ips[$ip] ?? 0) + 1;
+ $ipAllPaths[$ip][$path] = ($ipAllPaths[$ip][$path] ?? 0) + 1;
+ $ipAllDays[$ip][$dayOffset] = ($ipAllDays[$ip][$dayOffset] ?? 0) + 1;
if ($ua !== '') {
- $ipAgents[$ip][$ua] = ($ipAgents[$ip][$ua] ?? 0) + 1;
+ $ipAgents[$ip][$ua] = ($ipAgents[$ip][$ua] ?? 0) + 1;
}
- } elseif (str_starts_with($path, '/book/') && strlen($path) > 6) {
+ }
+
+ // Comptage spécifique aux pages de contenu (statut 200, /post/ ou /book/)
+ if ($status !== '200') {
+ return;
+ }
+
+ if (str_starts_with($path, '/post/') && strlen($path) > 6) {
+ $pages[$path] = ($pages[$path] ?? 0) + 1;
+ $dayPages[$path][$dayOffset] = ($dayPages[$path][$dayOffset] ?? 0) + 1;
+ if ($publicIp) {
+ $ipPaths[$ip][$path] = ($ipPaths[$ip][$path] ?? 0) + 1;
+ if ($tsVal > ($ipPathTs[$ip][$path] ?? 0)) {
+ $ipPathTs[$ip][$path] = $tsVal;
+ }
+ }
+ } elseif (str_ends_with($path, '/') === false && str_starts_with($path, '/book/') && strlen($path) > 6) {
$books[$path] = ($books[$path] ?? 0) + 1;
if ($publicIp) {
- $ips[$ip] = ($ips[$ip] ?? 0) + 1;
- }
- $ipDays[$ip][$dayOffset] = ($ipDays[$ip][$dayOffset] ?? 0) + 1;
- $ipPaths[$ip][$path] = ($ipPaths[$ip][$path] ?? 0) + 1;
- if ($tsVal > ($ipPathTs[$ip][$path] ?? 0)) {
- $ipPathTs[$ip][$path] = $tsVal;
- }
- if ($ua !== '') {
- $ipAgents[$ip][$ua] = ($ipAgents[$ip][$ua] ?? 0) + 1;
+ $ipPaths[$ip][$path] = ($ipPaths[$ip][$path] ?? 0) + 1;
+ if ($tsVal > ($ipPathTs[$ip][$path] ?? 0)) {
+ $ipPathTs[$ip][$path] = $tsVal;
+ }
}
}
}
- private function parseFile(array $file, int $cutoff, array &$pages, array &$books, array &$ips, array &$dayPages, array &$ipDays, array &$ipPaths, array &$ipPathTs, array &$ipAgents): void
- {
+ private function parseFile(
+ array $file,
+ int $cutoff,
+ array &$pages,
+ array &$books,
+ array &$ips,
+ array &$dayPages,
+ array &$ipPaths,
+ array &$ipPathTs,
+ array &$ipAllPaths,
+ array &$ipAllDays,
+ array &$ipAgents
+ ): void {
if ($file['type'] === 'tgz') {
try {
$phar = new PharData($file['path']);
@@ -222,7 +258,7 @@ class AccessLogParser
continue;
}
foreach (explode("\n", $content) as $line) {
- $this->parseLine($line, $cutoff, $pages, $books, $ips, $dayPages, $ipDays, $ipPaths, $ipPathTs, $ipAgents);
+ $this->parseLine($line, $cutoff, $pages, $books, $ips, $dayPages, $ipPaths, $ipPathTs, $ipAllPaths, $ipAllDays, $ipAgents);
}
}
} catch (\Exception $e) {
@@ -235,7 +271,7 @@ class AccessLogParser
while (!gzeof($h)) {
$line = gzgets($h, 8192);
if ($line !== false) {
- $this->parseLine($line, $cutoff, $pages, $books, $ips, $dayPages, $ipDays, $ipPaths, $ipPathTs, $ipAgents);
+ $this->parseLine($line, $cutoff, $pages, $books, $ips, $dayPages, $ipPaths, $ipPathTs, $ipAllPaths, $ipAllDays, $ipAgents);
}
}
gzclose($h);
@@ -245,7 +281,7 @@ class AccessLogParser
return;
}
while (($line = fgets($h)) !== false) {
- $this->parseLine($line, $cutoff, $pages, $books, $ips, $dayPages, $ipDays, $ipPaths, $ipPathTs, $ipAgents);
+ $this->parseLine($line, $cutoff, $pages, $books, $ips, $dayPages, $ipPaths, $ipPathTs, $ipAllPaths, $ipAllDays, $ipAgents);
}
fclose($h);
}
diff --git a/templates/admin_stats.php b/templates/admin_stats.php
index 42ad2fa..36e33f9 100644
--- a/templates/admin_stats.php
+++ b/templates/admin_stats.php
@@ -1,11 +1,12 @@
@@ -21,12 +22,13 @@ $_ipData = $adminData['stats_ip_data'] ?? [];