admin/SEONexus
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

老模板优化

Browse Source
This commit is contained in:
Your Name
2026-04-17 21:09:06 +08:00
parent 4dc78e3366
commit c8fb6956ef
60 changed files with 9113 additions and 146 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
code/app/common/helper/DomainSpiderCrawlLogHelper.php
View File

@@ -15,6 +15,9 @@ class DomainSpiderCrawlLogHelper
$arrItem['host'] = strtolower(trim((string)($arrItem['host'] ?? '')));
$arrItem['path'] = trim((string)($arrItem['path'] ?? ''));
$arrItem['page_type'] = DomainSpiderCrawlLogSchemaHelper::normalizePageType((string)($arrItem['page_type'] ?? 'other'));
if ($arrItem['page_type'] === 'other') {
$arrItem['page_type'] = DomainSpiderCrawlLogSchemaHelper::inferPageTypeFromPath((string)$arrItem['path'], 'other');
}
$arrItem['bot_name'] = DomainSpiderCrawlLogSchemaHelper::normalizeBotName((string)($arrItem['bot_name'] ?? 'other_bot'));
$arrItem['status'] = (int)($arrItem['status'] ?? 0);
$arrItem['count'] = max(0, (int)($arrItem['count'] ?? 0));
@@ -26,6 +29,9 @@ class DomainSpiderCrawlLogHelper
$arrItem['host'] = strtolower(trim((string)($arrItem['host'] ?? '')));
$arrItem['url'] = trim((string)($arrItem['url'] ?? ''));
$arrItem['page_type'] = DomainSpiderCrawlLogSchemaHelper::normalizePageType((string)($arrItem['page_type'] ?? 'other'));
if ($arrItem['page_type'] === 'other') {
$arrItem['page_type'] = DomainSpiderCrawlLogSchemaHelper::inferPageTypeFromPath((string)$arrItem['url'], 'other');
}
$arrItem['bot_name'] = DomainSpiderCrawlLogSchemaHelper::normalizeBotName((string)($arrItem['bot_name'] ?? 'other_bot'));
$arrItem['status'] = (int)($arrItem['status'] ?? 0);
$arrItem['cache_status'] = self::normalizeCacheStatus((string)($arrItem['cache_status'] ?? ''));

54
code/app/common/helper/DomainSpiderCrawlLogSchemaHelper.php
View File

@@ -202,6 +202,60 @@ class DomainSpiderCrawlLogSchemaHelper
return in_array($strPageType, self::buildSchema()['page_type_enum'], true) ? $strPageType : 'other';
}
public static function inferPageTypeFromPath(string $strPath, string $strFallback = 'other'): string
{
$strPath = strtolower(trim($strPath));
if ($strPath === '') {
return self::normalizePageType($strFallback);
}
$strPath = parse_url($strPath, PHP_URL_PATH) ?: $strPath;
$strPath = '/' . ltrim($strPath, '/');
$strNormalized = trim($strPath, '/');
if ($strNormalized === '' || $strNormalized === 'index.php') {
return 'home';
}
if ($strNormalized === 'robots.txt') {
return 'robots';
}
if (str_contains($strNormalized, 'sitemap')) {
return 'sitemap';
}
if (str_starts_with($strNormalized, 'search') || str_starts_with($strNormalized, 'get-index')) {
return 'search';
}
if (
str_contains($strNormalized, 'videotype')
|| str_contains($strNormalized, 'vodtype')
|| str_contains($strNormalized, 'fenlei')
) {
return 'category';
}
if (
str_starts_with($strNormalized, 'voddetail/')
|| str_starts_with($strNormalized, 'video-info/')
|| str_starts_with($strNormalized, 'neirong-')
) {
return 'detail';
}
if (
str_starts_with($strNormalized, 'vodplay/')
|| str_starts_with($strNormalized, 'video-play/')
|| str_starts_with($strNormalized, 'bf-')
) {
return 'play';
}
return self::normalizePageType($strFallback);
}
public static function normalizeBotName(string $strBotName): string
{
$strValue = strtolower(trim($strBotName));