feat: add monitoring marked articles and retention updates

2026-06-17 12:48:41 +08:00
parent 9ed857e159
commit 31c4dd9358
40 changed files with 2373 additions and 488 deletions
@@ -279,6 +279,8 @@ type monitoringAliasResolution struct {
 	ArticleID            *int64
 	PublishRecordID      *int64
 	NormalizedURLKey     string
+	SourceType           string
+	SourceID             *int64
 	ResolutionStatus     string
 	ResolutionConfidence string
 }
@@ -298,6 +300,9 @@ type monitoringAliasCandidate struct {
 	MatchKeys         []string
 	ArticleID         *int64
 	PublishRecordID   *int64
+	SourceType        string
+	SourceID          *int64
+	SourcePriority    int
 	RegistrableDomain string
 	SiteKey           string
 	Host              string
@@ -316,6 +321,8 @@ type monitoringCitationFact struct {
 	SiteKey              string
 	ArticleID            *int64
 	PublishRecordID      *int64
+	ContentSourceType    *string
+	ContentSourceID      *int64
 	ResolutionStatus     string
 	ResolutionConfidence string
 }
@@ -2119,7 +2126,7 @@ func (s *MonitoringCallbackService) processTaskResult(ctx context.Context, task
 				return nil, err
 			}
 			citationSourceCount++
-			if fact.ArticleID != nil {
+			if fact.ContentSourceType != nil && fact.ContentSourceID != nil {
 				contentCitationCount++
 			}
 		}
@@ -2591,23 +2598,61 @@ func (s *MonitoringCallbackService) loadAliasResolutions(ctx context.Context, tx
 	}

 	rows, err := tx.Query(ctx, `
+		WITH alias_sources AS (
+			SELECT
+				normalized_url,
+				article_id,
+				publish_record_id,
+					'published_article' AS source_type,
+					article_id AS source_id,
+					1 AS source_priority,
+					registrable_domain,
+				site_key,
+				host,
+				last_path_segment,
+				article_title_snapshot,
+				updated_at,
+				id
+			FROM monitoring_article_url_aliases
+			WHERE tenant_id = $1
+			  AND confidence = 'high'
+			UNION ALL
+			SELECT
+				normalized_url,
+				NULL AS article_id,
+				NULL AS publish_record_id,
+					'manual_mark' AS source_type,
+					id AS source_id,
+					2 AS source_priority,
+					registrable_domain,
+				site_key,
+				host,
+				last_path_segment,
+				article_title AS article_title_snapshot,
+				updated_at,
+				id
+			FROM monitoring_user_marked_articles
+			WHERE tenant_id = $1
+			  AND expires_at > NOW()
+		)
 		SELECT
 			normalized_url,
 			article_id,
 			publish_record_id,
-			registrable_domain,
+				source_type,
+				source_id,
+				source_priority,
+				registrable_domain,
 			site_key,
 			host,
 			last_path_segment,
 			article_title_snapshot
-		FROM monitoring_article_url_aliases
-		WHERE tenant_id = $1
-		  AND (
-			normalized_url = ANY($2::text[])
-			OR NULLIF(LOWER(TRIM(registrable_domain)), '') = ANY($3::text[])
-			OR NULLIF(LOWER(TRIM(site_key)), '') = ANY($3::text[])
-			OR NULLIF(LOWER(TRIM(host)), '') = ANY($3::text[])
-		  )
+		FROM alias_sources
+		WHERE normalized_url = ANY($2::text[])
+		   OR NULLIF(LOWER(TRIM(registrable_domain)), '') = ANY($3::text[])
+		   OR NULLIF(LOWER(TRIM(site_key)), '') = ANY($3::text[])
+		   OR NULLIF(LOWER(TRIM(host)), '') = ANY($3::text[])
+		ORDER BY updated_at DESC, id DESC
 	`, tenantID, normalizedURLs, domainKeys)
 	if err != nil {
 		return nil, response.ErrInternal(50041, "alias_lookup_failed", "failed to resolve article aliases")
@@ -2619,6 +2664,7 @@ func (s *MonitoringCallbackService) loadAliasResolutions(ctx context.Context, tx
 		var candidate monitoringAliasCandidate
 		var articleID sql.NullInt64
 		var publishRecordID sql.NullInt64
+		var sourceID sql.NullInt64
 		var registrableDomain sql.NullString
 		var siteKey sql.NullString
 		var host sql.NullString
@@ -2628,6 +2674,9 @@ func (s *MonitoringCallbackService) loadAliasResolutions(ctx context.Context, tx
 			&candidate.NormalizedURL,
 			&articleID,
 			&publishRecordID,
+			&candidate.SourceType,
+			&sourceID,
+			&candidate.SourcePriority,
 			&registrableDomain,
 			&siteKey,
 			&host,
@@ -2638,6 +2687,7 @@ func (s *MonitoringCallbackService) loadAliasResolutions(ctx context.Context, tx
 		}
 		candidate.ArticleID = nullableInt64Value(articleID)
 		candidate.PublishRecordID = nullableInt64Value(publishRecordID)
+		candidate.SourceID = nullableInt64Value(sourceID)
 		candidate.RegistrableDomain = strings.ToLower(strings.TrimSpace(nullableStringText(registrableDomain)))
 		candidate.SiteKey = strings.ToLower(strings.TrimSpace(nullableStringText(siteKey)))
 		candidate.Host = strings.ToLower(strings.TrimSpace(nullableStringText(host)))
@@ -2665,8 +2715,14 @@ func (s *MonitoringCallbackService) loadAliasResolutions(ctx context.Context, tx
 				best = candidate
 				continue
 			}
-			if score == bestScore && !sameInt64Pointer(candidate.ArticleID, best.ArticleID) {
-				bestAmbiguous = true
+			if score == bestScore && !sameMonitoringAliasCandidateSource(candidate, best) {
+				if best.SourcePriority == 0 || candidate.SourcePriority == best.SourcePriority {
+					bestAmbiguous = true
+				}
+				if best.SourcePriority == 0 || (candidate.SourcePriority > 0 && candidate.SourcePriority < best.SourcePriority) {
+					bestAmbiguous = false
+					best = candidate
+				}
 			}
 		}

@@ -2677,6 +2733,8 @@ func (s *MonitoringCallbackService) loadAliasResolutions(ctx context.Context, tx
 			ArticleID:            cloneInt64(best.ArticleID),
 			PublishRecordID:      cloneInt64(best.PublishRecordID),
 			NormalizedURLKey:     best.NormalizedURL,
+			SourceType:           strings.TrimSpace(best.SourceType),
+			SourceID:             cloneInt64(best.SourceID),
 			ResolutionStatus:     "resolved",
 			ResolutionConfidence: "high",
 		}
@@ -2745,19 +2803,28 @@ func sameInt64Pointer(left, right *int64) bool {
 	return *left == *right
 }

+func sameMonitoringAliasCandidateSource(left, right monitoringAliasCandidate) bool {
+	leftType := strings.TrimSpace(left.SourceType)
+	rightType := strings.TrimSpace(right.SourceType)
+	if leftType == "" && rightType == "" {
+		return sameInt64Pointer(left.ArticleID, right.ArticleID)
+	}
+	return leftType == rightType && sameInt64Pointer(left.SourceID, right.SourceID)
+}
+
 func (s *MonitoringCallbackService) insertCitationFact(ctx context.Context, tx pgx.Tx, task *monitoringCollectTask, runID int64, fact monitoringCitationFact) error {
 	if _, err := tx.Exec(ctx, `
 		INSERT INTO monitoring_citation_facts (
 			tenant_id, run_id, brand_id, ai_platform_id, business_date,
 			cited_url, cited_title, normalized_url, host, registrable_domain,
 			subdomain, suffix, site_key, article_id, publish_record_id,
-			resolution_status, resolution_confidence
+			content_source_type, content_source_id, resolution_status, resolution_confidence
 		)
 		VALUES (
 			$1, $2, $3, $4, $5::date,
 			$6, $7, $8, $9, $10,
 			$11, $12, $13, $14, $15,
-			$16, $17
+			$16, $17, $18, $19
 		)
 	`,
 		task.TenantID,
@@ -2775,6 +2842,8 @@ func (s *MonitoringCallbackService) insertCitationFact(ctx context.Context, tx p
 		fact.SiteKey,
 		nullableInt64(fact.ArticleID),
 		nullableInt64(fact.PublishRecordID),
+		nullableString(fact.ContentSourceType),
+		nullableInt64(fact.ContentSourceID),
 		fact.ResolutionStatus,
 		fact.ResolutionConfidence,
 	); err != nil {
@@ -2926,10 +2995,10 @@ func (s *MonitoringCallbackService) loadExistingRunSummary(ctx context.Context,
 	var citationSourceCount int
 	var contentCitationCount int
 	if err := s.monitoringPool.QueryRow(ctx, `
-		SELECT
-			COUNT(*)::int AS citation_source_count,
-			COUNT(*) FILTER (WHERE article_id IS NOT NULL AND $3::text <> 'qwen')::int AS content_citation_count
-		FROM monitoring_citation_facts
+			SELECT
+				COUNT(*)::int AS citation_source_count,
+				COUNT(*) FILTER (WHERE content_source_type IS NOT NULL AND content_source_id IS NOT NULL AND $3::text <> 'qwen')::int AS content_citation_count
+			FROM monitoring_citation_facts
 		WHERE tenant_id = $1
 		  AND run_id = $2
 	`, task.TenantID, runID.Int64, normalizeMonitoringPlatformID(task.AIPlatformID)).Scan(&citationSourceCount, &contentCitationCount); err != nil {
@@ -3173,6 +3242,8 @@ func buildMonitoringCitationFact(input MonitoringSourceItem, aliasMap map[string

 	var articleID *int64
 	var publishRecordID *int64
+	var contentSourceType *string
+	var contentSourceID *int64
 	var aliasResolution *monitoringAliasResolution
 	if resolveContentCitations {
 		articleID = cloneInt64(input.ArticleID)
@@ -3186,6 +3257,10 @@ func buildMonitoringCitationFact(input MonitoringSourceItem, aliasMap map[string
 			if publishRecordID == nil {
 				publishRecordID = cloneInt64(alias.PublishRecordID)
 			}
+			if strings.TrimSpace(alias.SourceType) != "" && alias.SourceID != nil {
+				contentSourceType = optionalString(strings.TrimSpace(alias.SourceType))
+				contentSourceID = cloneInt64(alias.SourceID)
+			}
 		}
 	}

@@ -3217,6 +3292,8 @@ func buildMonitoringCitationFact(input MonitoringSourceItem, aliasMap map[string
 		SiteKey:              siteKey,
 		ArticleID:            articleID,
 		PublishRecordID:      publishRecordID,
+		ContentSourceType:    contentSourceType,
+		ContentSourceID:      contentSourceID,
 		ResolutionStatus:     resolutionStatus,
 		ResolutionConfidence: resolutionConfidence,
 	}, true