Cyrene/backend/pkg/plugins/markdown/plugin.go

package markdown

import (
	"context"
	"fmt"
	"regexp"
	"strings"

	"git.yeij.top/AskaEth/Cyrene/pkg/plugins/sdk"
)

type MarkdownPlugin struct{ sdk.BasePlugin }

func (p *MarkdownPlugin) Metadata() sdk.PluginMetadata {
	return sdk.PluginMetadata{
		Name: "markdown", DisplayName: "Markdown Processor", Version: "1.0.0",
		Description: "Markdown processing: to HTML, extract text/links/code, generate TOC",
		Category: "format", Author: sdk.PluginAuthor{Name: "Cyrene Team"},
	}
}

func (p *MarkdownPlugin) Tools() []sdk.Tool { return []sdk.Tool{&MarkdownTool{}} }

type MarkdownTool struct{ sdk.BaseTool }

func (t *MarkdownTool) Definition() sdk.ToolDefinition {
	return sdk.ToolDefinition{
		ID: "markdown", Name: "markdown", DisplayName: "Markdown Processor",
		Description: "Markdown processing. Convert to HTML, extract plain text, extract links/code blocks, generate TOC.",
		Category: "format", Complexity: sdk.ComplexitySimple,
		Parameters: map[string]interface{}{
			"type": "object",
			"properties": map[string]interface{}{
				"action":   map[string]interface{}{"type": "string", "enum": []string{"to_html", "to_text", "extract_links", "extract_code", "table_of_contents"}},
				"markdown": map[string]interface{}{"type": "string"},
			},
			"required": []string{"action", "markdown"},
		},
	}
}

func (t *MarkdownTool) Validate(args map[string]interface{}) error {
	for _, k := range []string{"action", "markdown"} {
		if _, ok := args[k]; !ok {
			return fmt.Errorf("missing required parameter: %s", k)
		}
	}
	return nil
}

func (t *MarkdownTool) Execute(_ context.Context, args map[string]interface{}) (*sdk.ToolResult, error) {
	action, _ := args["action"].(string)
	md, _ := args["markdown"].(string)

	switch action {
	case "to_html":
		return &sdk.ToolResult{ToolName: "markdown", Success: true, Output: mdToHTML(md)}, nil

	case "to_text":
		text := md
		reCode := regexp.MustCompile("(?s)```.*?```")
		text = reCode.ReplaceAllString(text, "")
		text = regexp.MustCompile(`\*\*([^*]+)\*\*`).ReplaceAllString(text, "$1")
		text = regexp.MustCompile(`\*([^*]+)\*`).ReplaceAllString(text, "$1")
		text = regexp.MustCompile(`~~([^~]+)~~`).ReplaceAllString(text, "$1")
		text = regexp.MustCompile(`^#{1,6}\s+`).ReplaceAllString(text, "")
		text = regexp.MustCompile(`^[*-]\s+`).ReplaceAllString(text, "- ")
		text = regexp.MustCompile(`^>\s+`).ReplaceAllString(text, "")
		text = regexp.MustCompile(`\n{3,}`).ReplaceAllString(text, "\n\n")
		return &sdk.ToolResult{ToolName: "markdown", Success: true, Output: strings.TrimSpace(text)}, nil

	case "extract_links":
		re := regexp.MustCompile(`\[([^\]]+)\]\(([^)]+)\)`)
		matches := re.FindAllStringSubmatch(md, -1)
		if len(matches) == 0 {
			return &sdk.ToolResult{ToolName: "markdown", Success: true, Output: "No links found"}, nil
		}
		var out strings.Builder
		for i, m := range matches {
			out.WriteString(fmt.Sprintf("%d. %s -> %s\n", i+1, m[1], m[2]))
		}
		return &sdk.ToolResult{ToolName: "markdown", Success: true, Output: out.String()}, nil

	case "extract_code":
		re := regexp.MustCompile("(?s)```(\\w*)\n?(.*?)```")
		matches := re.FindAllStringSubmatch(md, -1)
		if len(matches) == 0 {
			return &sdk.ToolResult{ToolName: "markdown", Success: true, Output: "No code blocks found"}, nil
		}
		var out strings.Builder
		for i, m := range matches {
			lang := m[1]
			if lang == "" {
				lang = "text"
			}
			code := m[2]
			if len([]rune(code)) > 500 {
				code = string([]rune(code)[:500]) + "..."
			}
			out.WriteString(fmt.Sprintf("--- Block %d (%s) ---\n%s\n\n", i+1, lang, code))
		}
		return &sdk.ToolResult{ToolName: "markdown", Success: true, Output: out.String()}, nil

	case "table_of_contents":
		re := regexp.MustCompile(`(?m)^(#{1,6})\s+(.+)$`)
		matches := re.FindAllStringSubmatch(md, -1)
		if len(matches) == 0 {
			return &sdk.ToolResult{ToolName: "markdown", Success: true, Output: "No headings found"}, nil
		}
		var out strings.Builder
		for _, m := range matches {
			depth := len(m[1])
			indent := strings.Repeat("  ", depth-1)
			out.WriteString(fmt.Sprintf("%s- %s\n", indent, m[2]))
		}
		return &sdk.ToolResult{ToolName: "markdown", Success: true, Output: out.String()}, nil
	}
	return &sdk.ToolResult{ToolName: "markdown", Success: false, Error: "unknown action: " + action}, nil
}

func mdToHTML(md string) string {
	// Save code blocks
	type placeholder struct {
		orig     string
		content  string
		language string
	}
	blocks := []*placeholder{}
	reCode := regexp.MustCompile("(?s)```(\\w*)\n?(.*?)```")
	md = reCode.ReplaceAllStringFunc(md, func(s string) string {
		m := reCode.FindStringSubmatch(s)
		b := &placeholder{orig: fmt.Sprintf("\x00CODE%d\x00", len(blocks)), language: m[1], content: escapeHTML(m[2])}
		blocks = append(blocks, b)
		return b.orig
	})

	// Inline elements
	md = regexp.MustCompile("`([^`]+)`").ReplaceAllString(md, "<code>$1</code>")
	md = regexp.MustCompile(`!\[([^\]]*)\]\(([^)]+)\)`).ReplaceAllString(md, `<img src="$2" alt="$1">`)
	md = regexp.MustCompile(`\[([^\]]+)\]\(([^)]+)\)`).ReplaceAllString(md, `<a href="$2">$1</a>`)
	md = regexp.MustCompile(`\*\*([^*]+)\*\*`).ReplaceAllString(md, `<strong>$1</strong>`)
	md = regexp.MustCompile(`\*([^*]+)\*`).ReplaceAllString(md, `<em>$1</em>`)
	md = regexp.MustCompile(`~~([^~]+)~~`).ReplaceAllString(md, `<del>$1</del>`)
	md = regexp.MustCompile(`(?m)^#{6}\s+(.+)$`).ReplaceAllString(md, `<h6>$1</h6>`)
	md = regexp.MustCompile(`(?m)^#{5}\s+(.+)$`).ReplaceAllString(md, `<h5>$1</h5>`)
	md = regexp.MustCompile(`(?m)^#{4}\s+(.+)$`).ReplaceAllString(md, `<h4>$1</h4>`)
	md = regexp.MustCompile(`(?m)^#{3}\s+(.+)$`).ReplaceAllString(md, `<h3>$1</h3>`)
	md = regexp.MustCompile(`(?m)^#{2}\s+(.+)$`).ReplaceAllString(md, `<h2>$1</h2>`)
	md = regexp.MustCompile(`(?m)^#{1}\s+(.+)$`).ReplaceAllString(md, `<h1>$1</h1>`)
	md = regexp.MustCompile(`(?m)^---\s*$`).ReplaceAllString(md, `<hr>`)
	md = regexp.MustCompile(`(?m)^>\s+(.+)$`).ReplaceAllString(md, `<blockquote>$1</blockquote>`)

	// Restore code blocks
	for _, b := range blocks {
		langAttr := ""
		if b.language != "" {
			langAttr = " class=\"language-" + b.language + "\""
		}
		md = strings.Replace(md, b.orig, "<pre><code"+langAttr+">"+b.content+"</code></pre>", 1)
	}

	// Paragraphs
	lines := strings.Split(md, "\n")
	var out strings.Builder
	for _, line := range lines {
		trimmed := strings.TrimSpace(line)
		if trimmed == "" {
			continue
		}
		if strings.HasPrefix(trimmed, "<") {
			out.WriteString(trimmed + "\n")
		} else {
			out.WriteString("<p>" + trimmed + "</p>\n")
		}
	}
	return out.String()
}

func escapeHTML(s string) string {
	s = strings.ReplaceAll(s, "&", "&amp;")
	s = strings.ReplaceAll(s, "<", "&lt;")
	s = strings.ReplaceAll(s, ">", "&gt;")
	return s
}