restore gpt seo copy pipeline and stabilize seo tkd output

2026-04-16 22:45:00 +08:00
parent 09b2137fa6
commit 2a905bbf0f
18 changed files with 3587 additions and 85 deletions
--- a/code/app/common/helper/SeoCopyFallbackBuilder.php
+++ b/code/app/common/helper/SeoCopyFallbackBuilder.php
@@ -26,8 +26,10 @@ class SeoCopyFallbackBuilder
    {
        $strSiteName = self::siteName($arrFacts);
        $strScopes = self::joinItems((array)($arrFacts['content_scopes'] ?? []), 3, '影视内容');
+        $strLanding = $strSiteName . '首页当前覆盖' . $strScopes . '等常见内容入口，适合先看首屏推荐，再按分类、榜单或搜索继续缩小范围。';

        return [
+            'search_landing' => $strLanding,
            'intro_text' => $strSiteName . '首页当前覆盖' . $strScopes . '等常见内容入口，建议先看首屏内容卡片，再继续按分类、榜单或搜索缩小范围。',
            'intro_meta' => '首页更适合先发现值得点开的内容，说明性提示只做辅助，不建议压住首屏主体内容。',
            'guide_cards' => [
@@ -94,8 +96,10 @@ class SeoCopyFallbackBuilder
    private static function buildRankIndex(array $arrFacts): array
    {
        $strPeriods = self::joinItems((array)($arrFacts['rank_periods'] ?? []), 3, '常见榜单周期');
+        $strLanding = '当前榜单首页汇总了' . $strPeriods . '等常见周期入口，适合先确定周期，再继续进入具体榜单列表页。';

        return [
+            'search_landing' => $strLanding,
            'intro_text' => '当前榜单首页汇总了' . $strPeriods . '等常见周期入口，适合先确定要看的榜单范围再继续下钻。',
            'intro_meta' => '如果你更关心具体周期，可以直接进入对应榜单列表页，再继续看详情和播放路径。',
            'guide_cards' => [
@@ -111,8 +115,10 @@ class SeoCopyFallbackBuilder
        $strPeriod = self::fallback((string)($arrFacts['rank_period_name'] ?? ''), '当前周期');
        $strScope = self::fallback((string)($arrFacts['rank_scope_name'] ?? ''), '当前范围');
        $intVisible = (int)($arrFacts['page_stats']['visible'] ?? 0);
+        $strLanding = '当前榜单列表页展示的是' . $strScope . '下的' . $strPeriod . '结果，当前页可继续浏览的条目约为' . max(1, $intVisible) . '条。';

        return [
+            'search_landing' => $strLanding,
            'intro_text' => '当前榜单列表页展示的是' . $strScope . '下的' . $strPeriod . '结果，当前页可继续浏览的条目约为' . max(1, $intVisible) . '条。',
            'intro_meta' => '如果你想横向对比结果，可以切到其他周期榜单，或者直接进入详情页继续确认内容。',
            'guide_cards' => [
--- a/code/app/common/helper/SeoCopyStore.php
+++ b/code/app/common/helper/SeoCopyStore.php
@@ -24,7 +24,93 @@ class SeoCopyStore
            return trim($strTemplate);
        }

-        return trim(strtr($strTemplate, $arrReplace));
+        foreach ($arrTokens as $strKey => $mValue) {
+            $strKey = trim((string)$strKey);
+            if ($strKey === '' || trim((string)$mValue) !== '') {
+                continue;
+            }
+
+            $strTemplate = self::stripEmptyTokenArtifacts($strTemplate, $strKey);
+        }
+
+        return self::normalizeInterpolatedText(strtr($strTemplate, $arrReplace));
+    }
+
+    private static function stripEmptyTokenArtifacts(string $strTemplate, string $strKey): string
+    {
+        $strToken = preg_quote('{' . $strKey . '}', '/');
+        $arrLabelMap = [
+            'video_alias' => ['别名'],
+            'year' => ['年份'],
+            'area_name' => ['地区'],
+            'lang_name' => ['语言'],
+            'actor_names' => ['演员'],
+            'director_names' => ['导演'],
+            'remarks' => ['备注'],
+            'play_line' => ['播放线路', '线路'],
+            'episode_name' => ['集数', '期数', '剧集'],
+        ];
+
+        foreach ((array)($arrLabelMap[$strKey] ?? []) as $strLabel) {
+            $strLabel = preg_quote($strLabel, '/');
+            $arrLabelPatterns = [
+                '/' . $strLabel . '\s*' . $strToken . '\s*[、，,]/u',
+                '/[、，,]\s*' . $strLabel . '\s*' . $strToken . '/u',
+                '/' . $strLabel . '\s*' . $strToken . '/u',
+            ];
+
+            foreach ($arrLabelPatterns as $strPattern) {
+                $strTemplate = preg_replace($strPattern, '', $strTemplate) ?? $strTemplate;
+            }
+        }
+
+        $arrPatterns = [
+            // 删掉前置连词 + 空 token，例如 “和{remarks}”
+            '/\s*(?:和|及|与|并|或)\s*' . $strToken . '/u',
+            // 删掉空 token + 后置连词，例如 “{remarks}和”
+            '/' . $strToken . '\s*(?:和|及|与|并|或)\s*/u',
+            // 删掉与空 token 紧邻的顿号/逗号
+            '/[、，,]\s*' . $strToken . '/u',
+            '/' . $strToken . '\s*[、，,]/u',
+            // 默认兜底：直接删掉空 token
+            '/' . $strToken . '/u',
+        ];
+
+        foreach ($arrPatterns as $strPattern) {
+            $strTemplate = preg_replace($strPattern, '', $strTemplate) ?? $strTemplate;
+        }
+
+        return $strTemplate;
+    }
+
+    private static function normalizeInterpolatedText(string $strValue): string
+    {
+        $strValue = preg_replace('/\{[a-z0-9_]+\}/iu', '', $strValue) ?? $strValue;
+        $strValue = str_replace(['高清高清播放', '高清高清'], ['高清播放', '高清'], $strValue);
+        $strValue = preg_replace('/\s{2,}/u', ' ', $strValue) ?? $strValue;
+        $strValue = preg_replace('/\s*[_|｜]+\s*/u', ' - ', $strValue) ?? $strValue;
+        $strValue = preg_replace('/\s*-\s*-\s*/u', ' - ', $strValue) ?? $strValue;
+        $strValue = preg_replace('/(会把|把)([^，。；！？、\\s]{1,24})和([^，。；！？、\\s]{1,24})(等资料|这类线索)/u', '$1$2、$3$4', $strValue) ?? $strValue;
+        $strValue = preg_replace('/(年份|地区|语言|演员|导演|备注|别名|线路|集数)(?=[A-Za-z0-9])/u', '$1 ', $strValue) ?? $strValue;
+        $strValue = preg_replace('/(地区|语言|演员|导演|备注|别名)([\p{Han}A-Za-z])/u', '$1 $2', $strValue) ?? $strValue;
+        $strValue = preg_replace('/地区\s*([^，。；！？、]+?)和语言\s*([^，。；！？、]+)/u', '地区 $1，语言 $2', $strValue) ?? $strValue;
+        $strValue = preg_replace('/(会把)([^，。；！？、\\s]{1,24})、([^，。；！？、\\s]{1,24})(这类线索)/u', '$1$2和$3$4', $strValue) ?? $strValue;
+        $strValue = preg_replace('/(会把)([^，。；！？、\\s]{1,24})、([^，。；！？、\\s]{1,24})(等资料)/u', '$1$2、$3$4', $strValue) ?? $strValue;
+        $strValue = preg_replace('/补看(\d{4}[^、，。；！？]*?)、([\p{Han}A-Za-z]{1,20})、([\p{Han}A-Za-z]{1,20})与([\p{Han}A-Za-z·、,，]{1,40})/u', '补看 $1、$2、$3，以及$4', $strValue) ?? $strValue;
+        $strValue = preg_replace('/(别名|年份|地区|语言|演员|导演|备注)\s+与/u', '$1与', $strValue) ?? $strValue;
+        $strValue = preg_replace('/本页，建议/u', '本页建议', $strValue) ?? $strValue;
+        $strValue = preg_replace('/本页，适合/u', '本页适合', $strValue) ?? $strValue;
+        $strValue = preg_replace('/本页，方便/u', '本页方便', $strValue) ?? $strValue;
+        $strValue = preg_replace('/([、，,])(?:\s*\\1)+/u', '$1', $strValue) ?? $strValue;
+        $strValue = preg_replace('/([（(【\[])\s*[、，,\-]+\s*/u', '$1', $strValue) ?? $strValue;
+        $strValue = preg_replace('/\s*[、，,\-]+\s*([）)】\]])/u', '$1', $strValue) ?? $strValue;
+        $strValue = preg_replace('/(会把|把|将|用|看)、/u', '$1', $strValue) ?? $strValue;
+        $strValue = preg_replace('/(\d{4})-\d{2}-\d{2}(?:\s+\d{2}:\d{2}:\d{2})?/u', '$1', $strValue) ?? $strValue;
+        $strValue = preg_replace('/\s*([，。；！？、])/u', '$1', $strValue) ?? $strValue;
+        $strValue = preg_replace('/([（(【\[])\s+/u', '$1', $strValue) ?? $strValue;
+        $strValue = preg_replace('/\s+([）)】\]])/u', '$1', $strValue) ?? $strValue;
+
+        return trim($strValue);
    }

    public static function buildPageKey(array|string $mValue, string $strFallback = 'index'): string
--- a/code/app/common/helper/SeoWordsHelper.php
+++ b/code/app/common/helper/SeoWordsHelper.php
@@ -4,6 +4,31 @@ namespace app\common\helper;

 class SeoWordsHelper
 {
+    /**
+     * Resolve a stable forge keyword from video seo words.
+     */
+    public static function resolveForgeWord(array $video = [], int $intForgeId = 1): string
+    {
+        $intForgeId = max(1, $intForgeId);
+
+        $arrWords = array_values(array_filter(array_map(
+            static fn($mVal): string => trim((string)$mVal),
+            (array)($video['v_seo_words'] ?? [])
+        ), static fn(string $strVal): bool => $strVal !== ''));
+
+        if (!empty($arrWords)) {
+            $intIndex = ($intForgeId - 1) % count($arrWords);
+            return $arrWords[$intIndex];
+        }
+
+        $strName = trim((string)($video['v_name'] ?? ''));
+        if ($strName !== '') {
+            return $strName;
+        }
+
+        return trim((string)($video['v_name_en'] ?? ''));
+    }
+
    /**
     * 根据 $intRewrite 处理 SeoWords
     */
@@ -22,32 +47,39 @@ class SeoWordsHelper
            // 0：原序全部
            case 0:
                $arrSeoWords = $words;
+                break;

-                // 1：前 6
+            // 1：前 6
            case 1:
                $arrSeoWords =  array_slice($words, 0, 6);
+                break;

-                // 2：随机 8
+            // 2：随机 8
            case 2:
                shuffle($words);
                $arrSeoWords =  array_slice($words, 0, 8);
+                break;

-                // 3：插年份
+            // 3：插年份
            case 3:
                $arrSeoWords =  self::appendYear($words, $video);
+                break;

-                // 4：插“在线观看 / 免费 / 高清”
+            // 4：插“在线观看 / 免费 / 高清”
            case 4:
                $arrSeoWords =  self::appendSuffix($words);
+                break;

-                // 5：重排 + 去重
+            // 5：重排 + 去重
            case 5:
                shuffle($words);
                $arrSeoWords =  array_values(array_unique($words));
+                break;

-                // 兜底
+            // 兜底
            default:
                $arrSeoWords =  array_slice($words, 0, 6);
+                break;
        }

        $arrNewsSeoWords = [];
--- a/code/app/common/helper/UrlBuilder.php
+++ b/code/app/common/helper/UrlBuilder.php
@@ -83,7 +83,7 @@ class UrlBuilder
        return $this->replacePattern($pattern, [
            'strParentCategory' => $strParentCategory,
            'strCategory'       => $strCategory,
-            'intPage'           => max(1, (int)$intPage),
+            'intPage'           => $this->normalizePageValue($intPage),
        ]);
    }
    
@@ -226,4 +226,13 @@ class UrlBuilder
        }
        return $this->trimSlash($pattern);
    }
+
+    protected function normalizePageValue(int|string $page): string
+    {
+        if (is_string($page) && strpos($page, '{page}') !== false) {
+            return '{page}';
+        }
+
+        return (string)max(1, (int)$page);
+    }
 }