feat(migrations): Harden task audit tracking and optimize article templates

- Added migration to harden task audit tracking by modifying audit_logs and related tables. - Introduced operator_id to several tables for better tracking of actions. - Updated article_templates with new prompt templates for various article types, enhancing content generation. - Created prompt_rules and schedule_tasks tables to manage content generation rules and scheduling. - Added foreign key constraints to articles for better data integrity.
2026-04-02 00:31:28 +08:00
parent de30497f59
commit b31d8d0096
101 changed files with 16671 additions and 721 deletions
@@ -5,7 +5,8 @@ import (
 	"fmt"
 	"regexp"
 	"strings"
-	"unicode/utf8"
+
+	"github.com/geo-platform/tenant-api/internal/shared/contentstats"
 )

 var promptVariablePattern = regexp.MustCompile(`\{\{\s*([a-zA-Z0-9_]+)\s*\}\}`)
@@ -13,7 +14,7 @@ var promptVariablePattern = regexp.MustCompile(`\{\{\s*([a-zA-Z0-9_]+)\s*\}\}`)
 func buildGenerationPrompt(templateName string, promptTemplate *string, params map[string]interface{}) string {
 	basePrompt := strings.TrimSpace(renderPromptTemplate(promptTemplate, params))
 	if basePrompt == "" {
-		basePrompt = fmt.Sprintf("Write a complete article in markdown for the template %q.", templateName)
+		basePrompt = fmt.Sprintf("你是一名专业内容编辑，请基于已提供的信息，为模板「%s」写出一篇完整的 Markdown 文章。", templateName)
 	}

 	var sections []string
@@ -21,20 +22,108 @@ func buildGenerationPrompt(templateName string, promptTemplate *string, params m

 	contextBlock := buildPromptContext(params)
 	if contextBlock != "" {
-		sections = append(sections, "Request context:\n"+contextBlock)
+		sections = append(sections, "当前上下文:\n"+contextBlock)
 	}

 	sections = append(sections, strings.Join([]string{
-		"Output requirements:",
-		"- Return only article markdown, no surrounding explanation.",
-		"- Use the provided title when available.",
-		"- Follow the provided outline sections and key points when available.",
-		"- Keep structure clear with headings, short paragraphs, and concrete details.",
+		"写作总要求:",
+		"- 仅返回文章 Markdown 正文，不要附带额外说明、提示语或代码块。",
+		"- 输出语言与 locale 一致：zh-CN 使用简体中文，en-US 使用自然、专业的英语。",
+		"- 如提供了 title，使用该标题作为文章主标题，并围绕它展开，不要另起一个无关标题。",
+		"- 如提供了 article_outline，一级节点是正文的最终小节标题，必须按顺序展开；二级及更深节点仅作为该小节的行文思路、论证顺序或信息要点，不要机械写成额外标题。",
+		"- 除主标题和一级小节标题外，默认不要把大纲子节点直接写成 Markdown 标题；子节点内容应自然融入段落、列表或过渡句中。",
+		"- 如提供了 key_points，正文中必须覆盖这些重点，不要遗漏。",
+		"- 每个核心段落都要有明确判断、原因解释、适用场景或对比维度，避免空话、套话和重复表述。",
+		"- 信息不足时，不要编造具体事实、价格、数据、测试结果、用户评价或机构结论；可以用稳妥表述说明判断边界。",
+		"- 保持客观、克制、非广告化，不使用夸张宣传语，如“顶级”“颠覆性”“完美”等。",
+		"- 使用清晰的小节标题和短段落；仅在确实有助于理解时使用列表。",
+		"- 结尾应给出清晰结论、适合对象、不适合对象，或下一步建议，帮助读者完成判断。",
 	}, "\n"))

+	if lengthGuidance := buildGenerationLengthGuidance(params); lengthGuidance != "" {
+		sections = append(sections, "篇幅要求:\n"+lengthGuidance)
+	}
+
 	return strings.Join(sections, "\n\n")
 }

+func buildGenerationLengthGuidance(params map[string]interface{}) string {
+	sectionCount := estimateTopLevelSectionCount(params)
+	if sectionCount <= 0 {
+		sectionCount = 4
+	}
+
+	locale := strings.TrimSpace(stringValue(params["locale"]))
+	if locale == "en-US" {
+		minWords := 900
+		targetWords := 320 + sectionCount*130
+		if targetWords > minWords {
+			minWords = targetWords
+		}
+		if minWords > 1600 {
+			minWords = 1600
+		}
+		maxWords := minWords + 400
+		if maxWords > 2100 {
+			maxWords = 2100
+		}
+		return strings.Join([]string{
+			fmt.Sprintf("- 建议全文控制在 %d-%d English words，优先信息密度，不要为了拉长篇幅重复表达。", minWords, maxWords),
+			"- Intro and conclusion should stay concise. Most top-level sections only need 1-3 tight paragraphs; expand only the most important sections.",
+			"- If one paragraph can make the point clearly, do not split it into multiple repetitive paragraphs.",
+		}, "\n")
+	}
+
+	minChars := 1100
+	targetChars := 320 + sectionCount*180
+	if targetChars > minChars {
+		minChars = targetChars
+	}
+	if minChars > 1800 {
+		minChars = 1800
+	}
+	maxChars := minChars + 500
+	if maxChars > 2400 {
+		maxChars = 2400
+	}
+
+	return strings.Join([]string{
+		fmt.Sprintf("- 建议全文控制在 %d-%d 字左右，优先信息密度，不要为了凑字数重复表达。", minChars, maxChars),
+		"- 引言和结论保持简洁，大多数一级章节写 1-3 个自然段即可；只有最关键的章节需要更充分展开。",
+		"- 能用一段说清的内容，不要拆成多段同义反复；主体部分重点写判断、原因、场景和建议。",
+	}, "\n")
+}
+
+func estimateTopLevelSectionCount(params map[string]interface{}) int {
+	if params == nil {
+		return 0
+	}
+
+	switch value := params["article_outline"].(type) {
+	case []interface{}:
+		if len(value) > 0 {
+			return len(value)
+		}
+	case []map[string]interface{}:
+		if len(value) > 0 {
+			return len(value)
+		}
+	}
+
+	switch value := params["outline_sections"].(type) {
+	case []interface{}:
+		if len(value) > 0 {
+			return len(value)
+		}
+	case []string:
+		if len(value) > 0 {
+			return len(value)
+		}
+	}
+
+	return 0
+}
+
 func renderPromptTemplate(promptTemplate *string, params map[string]interface{}) string {
 	if promptTemplate == nil {
 		return ""
@@ -66,12 +155,15 @@ func buildPromptContext(params map[string]interface{}) string {
 		"product_name",
 		"subject",
 		"brand_name",
+		"primary_keyword",
 		"brand",
 		"category",
 		"count",
 		"depth",
+		"article_outline",
 		"outline_sections",
 		"key_points",
+		"review_intro_hook",
 		"keywords",
 		"competitors",
 	}
@@ -83,7 +175,10 @@ func buildPromptContext(params map[string]interface{}) string {
 		if value == nil {
 			return
 		}
-		formatted := strings.TrimSpace(formatPromptValue(value))
+		if key == "outline_sections" && hasStructuredOutline(params["article_outline"]) {
+			return
+		}
+		formatted := strings.TrimSpace(formatPromptContextValue(key, value))
 		if formatted == "" {
 			return
 		}
@@ -105,6 +200,21 @@ func buildPromptContext(params map[string]interface{}) string {
 	return strings.Join(lines, "\n")
 }

+func formatPromptContextValue(key string, value interface{}) string {
+	switch key {
+	case "keywords":
+		return formatKeywordList(value, 6)
+	case "competitors":
+		return formatCompetitorList(value, 6)
+	case "outline_sections":
+		return formatSectionList(value, 10)
+	case "article_outline":
+		return formatOutlineValue(value)
+	default:
+		return formatPromptValue(value)
+	}
+}
+
 func formatPromptValue(value interface{}) string {
 	switch v := value.(type) {
 	case nil:
@@ -114,6 +224,12 @@ func formatPromptValue(value interface{}) string {
 	case []string:
 		return strings.Join(v, ", ")
 	case []interface{}:
+		if hasOutlineItems(v) {
+			return formatOutlineItems(v, 0)
+		}
+		if hasNamedItems(v) {
+			return formatNamedItems(v, 6)
+		}
 		parts := make([]string, 0, len(v))
 		for _, item := range v {
 			formatted := strings.TrimSpace(formatPromptValue(item))
@@ -131,6 +247,156 @@ func formatPromptValue(value interface{}) string {
 	}
 }

+func formatKeywordList(value interface{}, limit int) string {
+	items := extractStringList(value, limit)
+	return strings.Join(items, ", ")
+}
+
+func formatSectionList(value interface{}, limit int) string {
+	items := extractStringList(value, limit)
+	return strings.Join(items, " > ")
+}
+
+func formatCompetitorList(value interface{}, limit int) string {
+	switch items := value.(type) {
+	case []interface{}:
+		names := make([]string, 0, len(items))
+		for _, item := range items {
+			if data, ok := item.(map[string]interface{}); ok {
+				name := strings.TrimSpace(stringValue(data["name"]))
+				if name == "" {
+					name = strings.TrimSpace(stringValue(data["brand_name"]))
+				}
+				if name != "" {
+					names = append(names, name)
+				}
+			}
+			if len(names) >= limit {
+				break
+			}
+		}
+		return strings.Join(names, ", ")
+	default:
+		return formatPromptValue(value)
+	}
+}
+
+func formatOutlineValue(value interface{}) string {
+	switch items := value.(type) {
+	case []interface{}:
+		return formatOutlineItems(items, 0)
+	default:
+		return formatPromptValue(value)
+	}
+}
+
+func hasStructuredOutline(value interface{}) bool {
+	switch items := value.(type) {
+	case []interface{}:
+		return hasOutlineItems(items)
+	default:
+		return false
+	}
+}
+
+func hasOutlineItems(items []interface{}) bool {
+	for _, item := range items {
+		if data, ok := item.(map[string]interface{}); ok {
+			if strings.TrimSpace(stringValue(data["outline"])) != "" {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+func hasNamedItems(items []interface{}) bool {
+	for _, item := range items {
+		if data, ok := item.(map[string]interface{}); ok {
+			if strings.TrimSpace(stringValue(data["name"])) != "" || strings.TrimSpace(stringValue(data["brand_name"])) != "" {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+func formatNamedItems(items []interface{}, limit int) string {
+	names := make([]string, 0, len(items))
+	for _, item := range items {
+		data, ok := item.(map[string]interface{})
+		if !ok {
+			continue
+		}
+		name := strings.TrimSpace(stringValue(data["name"]))
+		if name == "" {
+			name = strings.TrimSpace(stringValue(data["brand_name"]))
+		}
+		if name == "" {
+			continue
+		}
+		names = append(names, name)
+		if len(names) >= limit {
+			break
+		}
+	}
+	return strings.Join(names, ", ")
+}
+
+func formatOutlineItems(items []interface{}, level int) string {
+	lines := make([]string, 0)
+	indent := strings.Repeat("  ", level)
+	for _, item := range items {
+		node, ok := item.(map[string]interface{})
+		if !ok {
+			continue
+		}
+		outline := strings.TrimSpace(stringValue(node["outline"]))
+		if outline == "" {
+			continue
+		}
+		lines = append(lines, fmt.Sprintf("%s- %s", indent, outline))
+		children, ok := node["children"].([]interface{})
+		if ok && len(children) > 0 {
+			childText := formatOutlineItems(children, level+1)
+			if childText != "" {
+				lines = append(lines, childText)
+			}
+		}
+	}
+	return strings.Join(lines, "\n")
+}
+
+func extractStringList(value interface{}, limit int) []string {
+	items := make([]string, 0)
+	appendItem := func(text string) {
+		text = strings.TrimSpace(text)
+		if text == "" {
+			return
+		}
+		items = append(items, text)
+	}
+
+	switch list := value.(type) {
+	case []string:
+		for _, item := range list {
+			appendItem(item)
+			if limit > 0 && len(items) >= limit {
+				break
+			}
+		}
+	case []interface{}:
+		for _, item := range list {
+			appendItem(formatPromptValue(item))
+			if limit > 0 && len(items) >= limit {
+				break
+			}
+		}
+	}
+
+	return items
+}
+
 func resolveArticleTitle(params map[string]interface{}, markdown string) string {
 	if title := strings.TrimSpace(stringValue(params["title"])); title != "" {
 		return title
@@ -155,17 +421,7 @@ func resolveArticleTitle(params map[string]interface{}, markdown string) string
 }

 func estimateWordCount(markdown string) int {
-	trimmed := strings.TrimSpace(markdown)
-	if trimmed == "" {
-		return 0
-	}
-
-	words := len(strings.Fields(trimmed))
-	if words > 0 {
-		return words
-	}
-
-	return utf8.RuneCountInString(trimmed)
+	return contentstats.CountWords(markdown)
 }

 func stringValue(value interface{}) string {