parse

type DiagnosticInfo struct {
	ReadabilityOK    bool
	ReadabilityTitle string
	ContentLength    int
	StructureTags    map[string]int // 标签名 → 数量

	// 各阶段快照
	StagePreCleanLen int    // 预清理后 HTML 长度
	StageAnchoredLen int    // 结构锚定后 HTML 长度
	StageExtractSrc  string // 提取来源: "readability" | "heuristic"
	StageExtractLen  int    // 提取后内容 HTML 长度
	StageConvertLen  int    // markdown 转换后长度
	StageFinalLen    int    // 最终 markdown 长度
}

type HTMLResult struct {
	Title    string
	Markdown string
}

type Options struct {
	MaxInlineLines   int
	MaxInlineChars   int
	MaxContentLength int // 内容超过此长度也写文件，默认 100000
	FileOutputDir    string
	Store            StoreWriter
}

type Output struct {
	Mode       string // "inline" | "saved_to_file"
	Title      string
	Markdown   string // Mode=inline 时有值
	FilePath   string // Mode=saved_to_file 时有值
	TotalLines int
	TotalChars int
	Summary    string
	AgentHint  string
	StatusCode int
	ContType   string
}

type PDFOptions struct {
	MaxPages  int
	Heuristic bool
}

type PDFParseError struct {
	Source string
	Reason string
}

type PDFParseResult struct {
	SourcePath string
	Title      string
	PageCount  int
	Method     string
	Markdown   string
}

type StoreWriter interface {
	WriteMarkdown(content string, title string, sourceURL string) (filePath string, totalLines int, totalChars int, summary string, err error)
	WriteMarkdownFromReader(r io.Reader, title string, sourceURL string) (filePath string, totalLines int, totalChars int, summary string, err error)
}