.gitignore

@@ -5,4 +5,3 @@ env.json
 env.yml
 env.yaml
 .log.meta.json
-/test_res/

CHANGELOG.md

@@ -1,35 +1,5 @@
 # CHANGELOG
 
-## v1.5.1 (2026-06-08)
-- **JS 对齐 & 智能文档**:
-    - 将所有注册到 `jsmod` 的方法名统一为 PascalCase。
-    - **可选参数优化**: 将 `Open` 的 `password` 和 `Save` 的 `filename` 改为指针类型。配合最新的 `go/js` 引擎，生成的 `.d.ts` 定义将正确提示为可选参数 `?`，极大优化了 AI 编码体验。
-
-## v1.1.0 (2026-05-17)
-- **PDF 语义重构**: 引入全局语义分析引擎。
-    - **无缝流**: 彻底移除分页干扰（移除 `---` 和 `Page X` 标记），实现跨页内容自然合并。
-    - **智能标题层级**: 自动锁定文档总标题，智能识别并合并跨行长标题，避免封面期标题碎片化。
-    - **自动平衡**: 根据文档内容自动提升章节标题层级，确保 Markdown 目录结构平衡。
-    - **鲁棒表格识别**: 调优列间距算法并增加长句过滤，大幅降低 PDF 误判表格的概率。
-- **转换能力全面对齐**:
-    - **PPTX**: 每一页幻灯片标题现在统一识别为 `#` 一级标题，优化页面视觉分层。
-    - **DOCX**: 重构 XML 解析，完美支持 Word 原生表格 (`w:tbl`) 转换为 Markdown 表格。
-    - **XLSX**: 提升 Sheet 名称为 `#` 标题。
-- **工程化增强**: 新增 `test_res/testmd.sh` 自动化验证脚本，覆盖 4 种主流办公格式的 Markdown 转换质量。
-- **接口一致性**: 强制所有格式 Markdown 输出从一级标题 (`#`) 开始。
-
-## v1.0.9 (2026-05-17)
-- **新特性**: 内置功能完整的命令行工具 `doc` (`cmd/doc`)。
-- **功能增强**: `doc` 支持创建、打开、查看预览（Markdown）、转换为 JSON、查看元数据（Inspect）以及数据注入（Excel）。
-- **文档优化**: README 增加 `doc` 命令行工具的安装与使用指南。
-
-## v1.0.8 (2026-05-15)
-- **基础设施对齐**: 同步更新 `go/cast` 和 `go/file` 至最新版本。
-- **功能修复**: 修复 Excel 单元格解析在某些边界情况下的偏差。
-
-## v1.0.7 (2026-05-14)
-- **依赖同步**: 对齐基础设施版本。
-
 ## v1.0.6 (2026-05-13)
 - **新特性**: 支持 `.csv` 格式，支持对象数组与 Markdown 表格转换。
 - **新特性**: 支持 `.md` 和 `.txt` 格式，统一纳入 `Document` 接口管理。

README.md

@@ -72,37 +72,6 @@ mdStr := g.ToMarkdown() // 包含 Mermaid graph TD 的渲染内容
 - `ToMarkdown() string`
 - `Save(filename ...string) error`
 
-## 命令行工具 (doc)
-
-`document` 包内置了一个强大的命令行工具 `doc`，位于 `cmd/doc` 目录下。
-
-### 安装
-使用 `go install` 安装，生成的二进制文件名即为 `doc`：
-```bash
-go install apigo.cc/go/document/cmd/doc@latest
-```
-
-### 常用命令
-```bash
-# 1. 预览 Excel/Word/PDF 内容 (默认输出 Markdown)
-doc report.xlsx
-
-# 2. 将文档转为结构化 JSON (适合 RAG 或自动化脚本)
-doc manual.docx --json
-
-# 3. 提取 PDF 内容并保存为 Markdown 文件
-doc paper.pdf -o paper.md
-
-# 4. 向已有的 Excel 注入数据 (支持追加或覆盖)
-doc --data '[{"id":1,"name":"Alice"}]' -o test.xlsx test.xlsx
-
-# 5. 查看文档元数据 (如工作表名、PDF 页数等)
-doc paper.pdf --inspect
-```
-
-### 帮助信息
-运行 `doc --help` 查看完整参数说明。
-
 ### Graph 专用 (关系型文档)
 - `AddNode(n *Node)`
 - `OpenGraph(filename string) (*Graph, error)`

cmd/doc/main.go

@@ -1,159 +0,0 @@
-package main
-
-import (
-	"flag"
-	"fmt"
-	"os"
-	"strings"
-
-	"apigo.cc/go/cast"
-	"apigo.cc/go/document"
-)
-
-var (
-	jsonOut    = flag.Bool("json", false, "以 JSON 格式输出文档内容")
-	mdOut      = flag.Bool("md", false, "以 Markdown 格式输出文档内容 (默认模式)")
-	savePath   = flag.String("o", "", "保存结果到指定文件路径 (如: output.xlsx, content.md)")
-	createType = flag.String("create", "", "创建新文档，支持类型: xlsx, csv, graph, md")
-	password   = flag.String("password", "", "访问加密文档所需的密码 (主要针对 Excel)")
-	sheetName  = flag.String("sheet", "", "操作 Excel 时指定的工作表名称或索引 (0, 1...)")
-	dataStr    = flag.String("data", "", "注入数据的 JSON 字符串 (支持对象数组或单个对象)")
-	inspect    = flag.Bool("inspect", false, "只查看文档元数据 (如类型、页数、工作表列表等)")
-	version    = flag.Bool("v", false, "显示版本信息")
-)
-
-const docVersion = "1.0.0"
-
-func main() {
-	flag.Usage = func() {
-		fmt.Fprintf(os.Stderr, "🗂️  Document CLI (doc) - 极简办公文档处理工具 v%s\n\n", docVersion)
-		fmt.Fprintf(os.Stderr, "用法:\n")
-		fmt.Fprintf(os.Stderr, "  doc [flags] [file]          # 处理已有文件\n")
-		fmt.Fprintf(os.Stderr, "  doc --create [type] [flags]  # 创建新文档\n\n")
-
-		fmt.Fprintf(os.Stderr, "常见示例:\n")
-		fmt.Fprintf(os.Stderr, "  doc report.xlsx             # 预览 Excel 内容 (Markdown 表格)\n")
-		fmt.Fprintf(os.Stderr, "  doc manual.docx --json      # 提取 Word 内容为结构化 JSON\n")
-		fmt.Fprintf(os.Stderr, "  doc paper.pdf -o text.md    # 提取 PDF 文字并存为 Markdown\n")
-		fmt.Fprintf(os.Stderr, "  doc --create xlsx -o n.xlsx # 创建空白 Excel\n")
-		fmt.Fprintf(os.Stderr, "  doc test.xlsx --data '[{\"ID\":1}]' -o test.xlsx  # 向 Excel 追加数据\n\n")
-
-		fmt.Fprintf(os.Stderr, "参数详解:\n")
-		flag.PrintDefaults()
-		
-		fmt.Fprintf(os.Stderr, "\n支持的格式:\n")
-		fmt.Fprintf(os.Stderr, "  Excel (.xlsx), Word (.docx), PDF (.pdf), PPT (.pptx), CSV (.csv), Graph (.graph), Markdown (.md)\n")
-	}
-
-	flag.Parse()
-
-	if *version {
-		fmt.Printf("doc version %s\n", docVersion)
-		return
-	}
-
-	args := flag.Args()
-	var doc document.Document
-	var err error
-
-	// 1. 获取文档实例
-	if *createType != "" {
-		doc, err = document.Create(*createType)
-		if err != nil {
-			fail("创建文档失败: %v", err)
-		}
-	} else if len(args) > 0 {
-		filename := args[0]
-		if *password != "" {
-			doc, err = document.Open(filename, *password)
-		} else {
-			doc, err = document.Open(filename)
-		}
-		if err != nil {
-			fail("无法打开文件 '%s': %v", filename, err)
-		}
-	} else {
-		flag.Usage()
-		return
-	}
-
-	// 2. 数据注入逻辑
-	if *dataStr != "" {
-		applyData(doc, *dataStr, *sheetName)
-	}
-
-	// 3. 执行核心操作
-	if *inspect {
-		runInspect(doc)
-		return
-	}
-
-	if *savePath != "" {
-		if err := doc.Save(*savePath); err != nil {
-			fail("保存失败: %v", err)
-		}
-		fmt.Printf("✨ 成功保存至: %s\n", *savePath)
-	} else {
-		outputContent(doc, *jsonOut)
-	}
-}
-
-func applyData(doc document.Document, dataStr, sheet string) {
-	var data []map[string]any
-	if err := cast.UnmarshalJSON(dataStr, &data); err != nil {
-		var single map[string]any
-		if err2 := cast.UnmarshalJSON(dataStr, &single); err2 == nil {
-			data = []map[string]any{single}
-		} else {
-			fail("数据格式无效，请提供有效的 JSON 对象或数组: %v", err)
-		}
-	}
-
-	switch d := doc.(type) {
-	case *document.Excel:
-		if err := d.SetData(sheet, data, "A1", ""); err != nil {
-			fail("写入 Excel 失败: %v", err)
-		}
-	case *document.Graph:
-		fmt.Println("⚠️  提示: Graph 类型目前主要通过 API 操作，暂不支持通过 CLI 批量 SetData。")
-	default:
-		fmt.Printf("⚠️  警告: 当前文档类型 (%T) 不支持数据注入操作。\n", d)
-	}
-}
-
-func runInspect(doc document.Document) {
-	fmt.Printf("🔍 文档详情:\n")
-	fmt.Printf("  类型: %T\n", doc)
-	
-	switch d := doc.(type) {
-	case *document.Excel:
-		fmt.Printf("  工作表: %s\n", strings.Join(d.Sheets(), ", "))
-	case *document.PDF:
-		if pages, ok := d.Metadata["pages"]; ok {
-			fmt.Printf("  总页数: %v\n", pages)
-		}
-		for k, v := range d.Metadata {
-			if k != "pages" {
-				fmt.Printf("  %s: %v\n", k, v)
-			}
-		}
-	}
-}
-
-func outputContent(doc document.Document, asJSON bool) {
-	if asJSON {
-		fmt.Println(doc.ToJSON())
-	} else {
-		content := doc.ToMarkdown()
-		if content == "" {
-			fmt.Println("(文档内容为空)")
-		} else {
-			fmt.Println(content)
-		}
-	}
-}
-
-func fail(format string, a ...any) {
-	fmt.Fprintf(os.Stderr, "❌ 错误: "+format+"\n", a...)
-	os.Exit(1)
-}

docx.go

@@ -45,7 +45,7 @@ func OpenDocx(filename string) (*Docx, error) {
 	return d, nil
 }
 
-// extractMarkdown 尝试从 docx 的 XML 中提取带标题和表格的 Markdown。
+// extractMarkdown 尝试从 docx 的 XML 中提取带标题的 Markdown。
 func (d *Docx) extractMarkdown(filename string) (string, error) {
 	r, err := zip.OpenReader(filename)
 	if err != nil {
@@ -69,10 +69,6 @@ func (d *Docx) extractMarkdown(filename string) (string, error) {
 	var sb strings.Builder
 	var inT bool
 	var currentStyle string
-	var inTable bool
-	var tableRows [][]string
-	var currentRow []string
-	var cellText strings.Builder
 
 	for {
 		t, err := decoder.Token()
@@ -85,56 +81,29 @@ func (d *Docx) extractMarkdown(filename string) (string, error) {
 
 		switch se := t.(type) {
 		case xml.StartElement:
-			switch se.Name.Local {
+			if se.Name.Local == "p" {
-			case "p":
 				currentStyle = ""
-				cellText.Reset()
+			} else if se.Name.Local == "pStyle" {
-			case "pStyle":
 				for _, attr := range se.Attr {
 					if attr.Name.Local == "val" {
 						currentStyle = attr.Value
 					}
 				}
-			case "t":
+			} else if se.Name.Local == "t" {
 				inT = true
-			case "tbl":
-				inTable = true
-				tableRows = nil
-			case "tr":
-				currentRow = nil
-			case "tc":
-				cellText.Reset()
 			}
 		case xml.EndElement:
-			switch se.Name.Local {
+			if se.Name.Local == "p" {
-			case "p":
-				if inTable {
-					// Paragraph inside table cell is handled by cellEnd
-				} else {
-					sb.WriteString("\n")
-				}
-			case "t":
-				inT = false
-			case "tc":
-				currentRow = append(currentRow, strings.TrimSpace(cellText.String()))
-				cellText.Reset()
-			case "tr":
-				tableRows = append(tableRows, currentRow)
-			case "tbl":
-				inTable = false
-				sb.WriteString(renderMarkdownTable(tableRows))
 				sb.WriteString("\n")
+			} else if se.Name.Local == "t" {
+				inT = false
 			}
 		case xml.CharData:
-			text := string(se)
+			if inT {
-			if inTable {
+				text := string(se)
-				cellText.WriteString(text)
+				if strings.Contains(strings.ToLower(currentStyle), "heading") {
-			} else if inT {
-				if strings.Contains(strings.ToLower(currentStyle), "heading") || 
-				   strings.Contains(strings.ToLower(currentStyle), "title") ||
-				   strings.Contains(strings.ToLower(currentStyle), "subject") {
 					level := "1"
-					if strings.Contains(strings.ToLower(currentStyle), "heading") && len(currentStyle) > 7 {
+					if len(currentStyle) > 7 {
 						level = currentStyle[7:]
 					}
 					l := cast.To[int](level)
@@ -153,30 +122,6 @@ func (d *Docx) extractMarkdown(filename string) (string, error) {
 	return strings.TrimSpace(sb.String()), nil
 }
 
-func renderMarkdownTable(rows [][]string) string {
-	if len(rows) == 0 {
-		return ""
-	}
-	var sb strings.Builder
-	sb.WriteString("\n")
-	for i, row := range rows {
-		sb.WriteString("| ")
-		for _, col := range row {
-			sb.WriteString(strings.ReplaceAll(col, "|", "\\|"))
-			sb.WriteString(" | ")
-		}
-		sb.WriteString("\n")
-		if i == 0 {
-			sb.WriteString("|")
-			for range row {
-				sb.WriteString(" --- |")
-			}
-			sb.WriteString("\n")
-		}
-	}
-	return sb.String()
-}
-
 // ToJSON 返回包含元数据和内容的 JSON 字符串。
 func (d *Docx) ToJSON() string {
 	res, _ := cast.ToJSON(map[string]any{

excel.go

@@ -87,7 +87,7 @@ func (xls *Excel) ToMarkdown() string {
 		}
 
 		if len(sheets) > 1 {
-			sb.WriteString("# Sheet: " + sheetName + "\n\n")
+			sb.WriteString("## Sheet: " + sheetName + "\n\n")
 		}
 
 		for i, row := range rows {

go.mod

@@ -3,18 +3,17 @@ module apigo.cc/go/document
 go 1.25.0
 
 require (
-	apigo.cc/go/cast v1.5.0
+	apigo.cc/go/cast v1.3.2
-	apigo.cc/go/file v1.5.0
+	apigo.cc/go/file v1.3.1
-	apigo.cc/go/jsmod v1.5.0
 	github.com/dslipak/pdf v0.0.2
 	github.com/xuri/excelize/v2 v2.10.1
 	github.com/young2j/oxmltotext v1.0.3
 )
 
 require (
-	apigo.cc/go/encoding v1.5.0 // indirect
+	apigo.cc/go/encoding v1.3.0 // indirect
-	apigo.cc/go/rand v1.5.0 // indirect
+	apigo.cc/go/rand v1.3.0 // indirect
-	apigo.cc/go/safe v1.5.0 // indirect
+	apigo.cc/go/safe v1.3.0 // indirect
 	github.com/andybalholm/brotli v1.0.5 // indirect
 	github.com/dgrr/quickxml v0.0.0-20201022091424-4977de546d6c // indirect
 	github.com/klauspost/compress v1.17.0 // indirect
@@ -28,10 +27,10 @@ require (
 	github.com/xuri/nfp v0.0.2-0.20250530014748-2ddeb826f9a9 // indirect
 	go.uber.org/multierr v1.10.0 // indirect
 	go.uber.org/zap v1.26.0 // indirect
-	golang.org/x/crypto v0.52.0 // indirect
+	golang.org/x/crypto v0.51.0 // indirect
 	golang.org/x/image v0.40.0 // indirect
 	golang.org/x/net v0.54.0 // indirect
-	golang.org/x/sys v0.45.0 // indirect
+	golang.org/x/sys v0.44.0 // indirect
 	golang.org/x/text v0.37.0 // indirect
 	gopkg.in/yaml.v3 v3.0.1 // indirect
 )

go.sum

@@ -1,15 +1,13 @@
-apigo.cc/go/cast v1.5.0 h1:UBGJtFQ8eJPMQXs37cUgqd7YQo1zI9opuSDBDmn2/pE=
+apigo.cc/go/cast v1.3.2 h1:hh9MWDSwh3T/kQdCHjFpjDwHrh2A05Q4wt1AAWs8NBI=
-apigo.cc/go/cast v1.5.0/go.mod h1:z2GW5p5WCZGEqVVIJUdhl232vRbLf2Qu4EDlEakX/D8=
+apigo.cc/go/cast v1.3.2/go.mod h1:lGlwImiOvHxG7buyMWhFzcdvQzmSaoKbmr7bcDfUpHk=
-apigo.cc/go/encoding v1.5.0 h1:EJNdRVDOMoI2DAvZwQNQTbYuqB/6zsEzvg7lS5pQI+I=
+apigo.cc/go/encoding v1.3.0 h1:8jqNHoZBR8vOU/BGsLFebfp1Txa1UxDRpd7YwzIFLJs=
-apigo.cc/go/encoding v1.5.0/go.mod h1:8++NfZj3hWig0qh2g7GQRw/4LpSvCYMWUZ+8J+x58cA=
+apigo.cc/go/encoding v1.3.0/go.mod h1:kT/uUJiuAOkZ4LzUWrUtk/I0iL1D8aatvD+59bDnHBo=
-apigo.cc/go/file v1.5.0 h1:Fh1NSDBqaxjuXYJ71yPHPXVJ8BFEv/AGS3l+jkLi5uw=
+apigo.cc/go/file v1.3.1 h1:qHgiJsn1K9DazWRrPoHVnXtp6hDGGsUpAE/4G1bFXqY=
-apigo.cc/go/file v1.5.0/go.mod h1:4YhOGgBINTpmmmgws3H8LAyXQQBGzBp44hYUoCS+kr0=
+apigo.cc/go/file v1.3.1/go.mod h1:pYHBlB/XwsrnWpEh7GIFpbiqobrExfiB+rEN8V2d2kY=
-apigo.cc/go/jsmod v1.5.0 h1:JgQtJNiJWy1NOP9AzE8NX5VXJkpO/x3GqLsCCSny5Ec=
+apigo.cc/go/rand v1.3.0 h1:k+UFAhMySwXf+dq8Om9TniZV6fm6gAE0evbrqMEdwQU=
-apigo.cc/go/jsmod v1.5.0/go.mod h1:bmyeZtOAP/j5am+YRnaiM89smysK24K7ebk0koFtsSw=
+apigo.cc/go/rand v1.3.0/go.mod h1:mZ/4Soa3bk+XvDaqPWJuUe1bfEi4eThBj1XmEAuYxsk=
-apigo.cc/go/rand v1.5.0 h1:1o8hh8fhdBuk1/h02IvugvamuT3dkWbVJrqEJVQKB2E=
+apigo.cc/go/safe v1.3.0 h1:uctdAUsphT9p60Tk4oS5xPCe0NoIdOHfsYv4PNS0Rok=
-apigo.cc/go/rand v1.5.0/go.mod h1:Lh98S2dm9UY0X+M+kNQQEKyXHG5pcCKSFPyXN0QCGdk=
+apigo.cc/go/safe v1.3.0/go.mod h1:tC9X14V+qh0BqIrVg4UkXbl+2pEN+lj2ZNI8IjDB6Fs=
-apigo.cc/go/safe v1.5.0 h1:W1NblmcU8cex1f9Y5z8mNLUJOzZTE1s6fszb3FbhGnk=
-apigo.cc/go/safe v1.5.0/go.mod h1:OfQ5d6COePSGEuPvMeOk6KagX2sezw7nvKh7exj9SeM=
 github.com/andybalholm/brotli v1.0.5 h1:8uQZIdzKmjc/iuPu7O2ioW48L81FgatrcpfFmiq/cCs=
 github.com/andybalholm/brotli v1.0.5/go.mod h1:fO7iG3H7G2nSZ7m0zPUDn85XEX2GTukHGRSepvi9Eig=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
@@ -58,12 +56,14 @@ go.uber.org/multierr v1.10.0 h1:S0h4aNzvfcFsC3dRF1jLoaov7oRaKqRGC/pUEJ2yvPQ=
 go.uber.org/multierr v1.10.0/go.mod h1:20+QtiLqy0Nd6FdQB9TLXag12DsQkrbs3htMFfDN80Y=
 go.uber.org/zap v1.26.0 h1:sI7k6L95XOKS281NhVKOFCUNIvv9e0w4BF8N3u+tCRo=
 go.uber.org/zap v1.26.0/go.mod h1:dtElttAiwGvoJ/vj4IwHBS/gXsEu/pZ50mUIRWuG0so=
-golang.org/x/crypto v0.52.0 h1:RMs7fP2rXdep0CftQlK8Uf+kibLm7qkCcradZWYz988=
+golang.org/x/crypto v0.51.0 h1:IBPXwPfKxY7cWQZ38ZCIRPI50YLeevDLlLnyC5wRGTI=
+golang.org/x/crypto v0.51.0/go.mod h1:8AdwkbraGNABw2kOX6YFPs3WM22XqI4EXEd8g+x7Oc8=
 golang.org/x/image v0.40.0 h1:Tw4GyDXMo+daZN1znreBRC3VayR1aLFUyUEOLUdW1a8=
 golang.org/x/image v0.40.0/go.mod h1:uIc348UZMSvS5Z65CVZ7iDPaNobNFEPeJ4kbqTOszmA=
 golang.org/x/net v0.54.0 h1:2zJIZAxAHV/OHCDTCOHAYehQzLfSXuf/5SoL/Dv6w/w=
 golang.org/x/net v0.54.0/go.mod h1:Sj4oj8jK6XmHpBZU/zWHw3BV3abl4Kvi+Ut7cQcY+cQ=
-golang.org/x/sys v0.45.0 h1:dO4czNzziLiiXplLQgBCEpCvXQ3dnkn0SdaZSYdQ+FY=
+golang.org/x/sys v0.44.0 h1:ildZl3J4uzeKP07r2F++Op7E9B29JRUy+a27EibtBTQ=
+golang.org/x/sys v0.44.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
 golang.org/x/text v0.37.0 h1:Cqjiwd9eSg8e0QAkyCaQTNHFIIzWtidPahFWR83rTrc=
 golang.org/x/text v0.37.0/go.mod h1:a5sjxXGs9hsn/AJVwuElvCAo9v8QYLzvavO5z2PiM38=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=

js_export.go

@@ -1,98 +0,0 @@
-package document
-
-import (
-	"context"
-
-	"apigo.cc/go/file"
-	"apigo.cc/go/jsmod"
-)
-
-func init() {
-	jsmod.Register("document", map[string]any{
-		"Open": func(ctx context.Context, filename string, password *string) (*jsDocument, error) {
-			p, err := file.VerifyPathForSafeMode(ctx, filename)
-			if err != nil {
-				return nil, err
-			}
-			var doc Document
-			if password != nil {
-				doc, err = Open(p, *password)
-			} else {
-				doc, err = Open(p)
-			}
-			if err != nil {
-				return nil, err
-			}
-			return &jsDocument{ctx: ctx, d: doc}, nil
-		},
-		"Create": func(ctx context.Context, ext string) (*jsDocument, error) {
-			doc, err := Create(ext)
-			if err != nil {
-				return nil, err
-			}
-			return &jsDocument{ctx: ctx, d: doc}, nil
-		},
-		"NewExcel": func(ctx context.Context) *jsDocument {
-			return &jsDocument{ctx: ctx, d: NewExcel()}
-		},
-		"NewGraph": func(ctx context.Context) *jsDocument {
-			return &jsDocument{ctx: ctx, d: NewGraph()}
-		},
-	})
-}
-
-type jsDocument struct {
-	ctx context.Context
-	d   Document
-}
-
-func (j *jsDocument) ToJSON() string     { return j.d.ToJSON() }
-func (j *jsDocument) ToMarkdown() string { return j.d.ToMarkdown() }
-
-func (j *jsDocument) Save(filename *string) error {
-	var targetPath string
-	if filename != nil && *filename != "" {
-		p, err := file.VerifyPathForSafeMode(j.ctx, *filename)
-		if err != nil {
-			return err
-		}
-		targetPath = p
-	}
-	return j.d.Save(targetPath)
-}
-
-// Excel 增强方法 (如果底层是 Excel)
-func (j *jsDocument) Get(sheetName string, start, end string) ([][]any, error) {
-	if x, ok := j.d.(*Excel); ok {
-		return x.Get(sheetName, start, end)
-	}
-	return nil, nil
-}
-
-func (j *jsDocument) GetData(sheetName string, start, end string) ([]map[string]any, error) {
-	if x, ok := j.d.(*Excel); ok {
-		return x.GetData(sheetName, start, end)
-	}
-	return nil, nil
-}
-
-func (j *jsDocument) Set(sheetName string, table [][]any, start, end string) error {
-	if x, ok := j.d.(*Excel); ok {
-		return x.Set(sheetName, table, start, end)
-	}
-	return nil
-}
-
-func (j *jsDocument) SetData(sheetName string, data []map[string]any, start, end string) error {
-	if x, ok := j.d.(*Excel); ok {
-		return x.SetData(sheetName, data, start, end)
-	}
-	return nil
-}
-
-func (j *jsDocument) Sheets() []string {
-	if x, ok := j.d.(*Excel); ok {
-		return x.Sheets()
-	}
-	return nil
-}

pdf.go

@@ -2,7 +2,6 @@ package document
 
 import (
 	"fmt"
-	"sort"
 	"strings"
 
 	"apigo.cc/go/cast"
@@ -28,355 +27,42 @@ func OpenPDF(filename string) (*PDF, error) {
 	}
 	
 	f, err := pdf.Open(filename)
-	if err != nil {
+	if err == nil {
-		return nil, err
+		p.Metadata["pages"] = f.NumPage()
-	}
+		trailer := f.Trailer()
-
+		infoDict := trailer.Key("Info")
-	p.Metadata["pages"] = f.NumPage()
+		if !infoDict.IsNull() {
-	trailer := f.Trailer()
+			for _, field := range infoDict.Keys() {
-	infoDict := trailer.Key("Info")
+				val := infoDict.Key(field).Text()
-	if !infoDict.IsNull() {
+				if val != "" {
-		for _, field := range infoDict.Keys() {
+					p.Metadata[strings.ToLower(field)] = val
-			val := infoDict.Key(field).Text()
-			if val != "" {
-				p.Metadata[strings.ToLower(field)] = val
-			}
-		}
-	}
-
-	// 收集所有页面的 block
-	type Block struct {
-		Type     string
-		Level    int
-		Text     string
-		FontSize float64
-	}
-	var allBlocks []Block
-
-	for i := 1; i <= f.NumPage(); i++ {
-		page := f.Page(i)
-		if page.V.IsNull() {
-			continue
-		}
-
-		content := page.Content()
-		texts := content.Text
-		if len(texts) == 0 {
-			continue
-		}
-
-		// 1. 估算正文字体大小（众数）
-		fontSizes := make(map[int]int)
-		for _, t := range texts {
-			fontSizes[int(t.FontSize)]++
-		}
-		bodySize := 0
-		maxCount := 0
-		for size, count := range fontSizes {
-			if count > maxCount {
-				maxCount = count
-				bodySize = size
-			}
-		}
-
-		// 2. 按行分组（基于 Y 坐标）
-		type Line struct {
-			Y           float64
-			MaxFontSize float64
-			Text        string
-			Texts       []pdf.Text
-		}
-		var lines []Line
-		for _, t := range texts {
-			found := false
-			for i := range lines {
-				if t.Y > lines[i].Y-2 && t.Y < lines[i].Y+2 {
-					lines[i].Texts = append(lines[i].Texts, t)
-					found = true
-					break
 				}
 			}
-			if !found {
-				lines = append(lines, Line{Y: t.Y, Texts: []pdf.Text{t}})
-			}
 		}
 		
-		sort.Slice(lines, func(i, j int) bool {
+		var sb strings.Builder
-			return lines[i].Y > lines[j].Y
+		for i := 1; i <= f.NumPage(); i++ {
-		})
+			p_ := f.Page(i)
-
+			if p_.V.IsNull() {
-		for i := range lines {
-			sort.Slice(lines[i].Texts, func(m, n int) bool {
-				return lines[i].Texts[m].X < lines[i].Texts[n].X
-			})
-			var sb strings.Builder
-			maxFS := 0.0
-			for _, t := range lines[i].Texts {
-				sb.WriteString(t.S)
-				if t.FontSize > maxFS {
-					maxFS = t.FontSize
-				}
-			}
-			lines[i].Text = strings.TrimSpace(sb.String())
-			lines[i].MaxFontSize = maxFS
-		}
-
-		// 3. 语义块识别
-		for j := 0; j < len(lines); j++ {
-			line := lines[j]
-			if line.Text == "" {
 				continue
 			}
-
+			t := p_.Content().Text
-			isTableLine, cells := p.identifyTableLine(line)
+			if len(t) > 0 {
-			if isTableLine {
+				if i > 1 {
-				tableStr := "| " + strings.Join(cells, " | ") + " |"
+					sb.WriteString("\n\n")
-				allBlocks = append(allBlocks, Block{Type: "table", Text: tableStr})
-				continue
-			}
-
-			if line.MaxFontSize > float64(bodySize)+1 {
-				level := 1
-				if line.MaxFontSize < float64(bodySize)+4 {
-					level = 3
-				} else if line.MaxFontSize < float64(bodySize)+8 {
-					level = 2
 				}
-
+				sb.WriteString(fmt.Sprintf("<!-- Page %d -->\n", i))
-				fullText := line.Text
+				for _, text := range t {
-				lastY := line.Y
+					sb.WriteString(text.S)
-				for k := j + 1; k < len(lines); k++ {
-					if lines[k].Text != "" && MathAbs(lines[k].MaxFontSize-line.MaxFontSize) < 1.0 {
-						if MathAbs(lines[k].Y-lastY) < 25 {
-							fullText += " " + lines[k].Text
-							lastY = lines[k].Y
-							j = k
-						} else {
-							break
-						}
-					} else {
-						break
-					}
 				}
-
-				if j < 5 && i == 1 {
-					if title, ok := p.Metadata["title"].(string); ok && strings.Contains(title, fullText) {
-						continue
-					}
-				}
-
-				if len(allBlocks) > 0 && allBlocks[len(allBlocks)-1].Type == "heading" && allBlocks[len(allBlocks)-1].Level == level {
-					allBlocks[len(allBlocks)-1].Text += " " + fullText
-				} else {
-					allBlocks = append(allBlocks, Block{Type: "heading", Level: level, Text: fullText})
-				}
-			} else {
-				if (line.Y < 50 || line.Y > 800) && isPageNumber(line.Text) {
-					continue
-				}
-
-				fullText := line.Text
-				lastY := line.Y
-				for k := j + 1; k < len(lines); k++ {
-					if lines[k].Text != "" && lines[k].MaxFontSize <= float64(bodySize)+1 {
-						isT, _ := p.identifyTableLine(lines[k])
-						if isT {
-							break
-						}
-						if (lines[k].Y < 50 || lines[k].Y > 800) && isPageNumber(lines[k].Text) {
-							break
-						}
-						if MathAbs(lines[k].Y-lastY) > 25 {
-							break
-						}
-
-						fullText += lines[k].Text
-						lastY = lines[k].Y
-						j = k
-
-						if isPunctuation(lines[k].Text) {
-							break
-						}
-					} else {
-						break
-					}
-				}
-				allBlocks = append(allBlocks, Block{Type: "paragraph", Text: fullText})
 			}
 		}
+		p.Content = strings.TrimSpace(sb.String())
 	}
 	
-	// 4. 智能封面/标题逻辑 (全局)
-	h1Count := 0
-	for _, b := range allBlocks {
-		if b.Type == "heading" && b.Level == 1 {
-			h1Count++
-		}
-	}
-	shouldPromote := h1Count <= 1
-
-	contentStarted := false
-	hasMetadataTitle := false
-	if t, ok := p.Metadata["title"].(string); ok && t != "" {
-		hasMetadataTitle = true
-	}
-
-	firstHeadingProcessed := false
-
-	var sb strings.Builder
-	for _, b := range allBlocks {
-		if b.Type == "heading" {
-			level := b.Level
-			
-			if !contentStarted {
-				if !firstHeadingProcessed {
-					firstHeadingProcessed = true
-					// 如果有 Metadata Title，则 Metadata Title 充当了真正的第一级标题
-					// 我们把遇到的第一个大标题降级为正文文本（除非它已经是标准章节）
-					if hasMetadataTitle && !isStandardSection(b.Text) {
-						sb.WriteString("\n" + b.Text + "\n\n")
-						continue
-					} else {
-						// 否则作为文档的主标题
-						sb.WriteString("\n# " + b.Text + "\n\n")
-						continue
-					}
-				} else {
-					if isStandardSection(b.Text) {
-						contentStarted = true
-						sb.WriteString("\n# " + b.Text + "\n\n")
-						continue
-					}
-					// 否则作为封面副标题/文本
-					sb.WriteString(b.Text + "\n\n")
-					continue
-				}
-			}
-
-			if shouldPromote {
-				if level == 2 {
-					level = 1
-				} else if level == 3 {
-					level = 2
-				}
-			}
-			sb.WriteString("\n" + strings.Repeat("#", level) + " " + b.Text + "\n\n")
-		} else if b.Type == "paragraph" {
-			contentStarted = true
-			sb.WriteString(b.Text + "\n\n")
-		} else if b.Type == "table" {
-			contentStarted = true
-			sb.WriteString(b.Text + "\n")
-		}
-	}
-
-	p.Content = strings.TrimSpace(sb.String())
 	return p, nil
 }
 
-func isStandardSection(s string) bool {
-	s = strings.TrimSpace(s)
-	// 常见的章节开头关键词
-	standards := []string{"引言", "摘要", "目录", "前言", "结论", "背景", "概述", "Introduction", "Abstract", "Conclusion", "Summary"}
-	for _, std := range standards {
-		if strings.Contains(s, std) {
-			return true
-		}
-	}
-	// 如果标题带有数字编号且较短，也认为是章节开始
-	if len([]rune(s)) < 20 {
-		if s[0] >= '0' && s[0] <= '9' {
-			return true
-		}
-		if strings.HasPrefix(s, "第") && (strings.Contains(s, "章") || strings.Contains(s, "节") || strings.Contains(s, "部分")) {
-			return true
-		}
-	}
-	return false
-}
-
-func isPageNumber(s string) bool {
-	s = strings.TrimSpace(s)
-	if s == "" {
-		return false
-	}
-	isNum := true
-	for _, r := range s {
-		if r < '0' || r > '9' {
-			isNum = false
-			break
-		}
-	}
-	if isNum {
-		return true
-	}
-	lower := strings.ToLower(s)
-	return strings.HasPrefix(lower, "page") || (strings.HasPrefix(s, "-") && strings.HasSuffix(s, "-"))
-}
-
-func isPunctuation(s string) bool {
-	if s == "" {
-		return false
-	}
-	runes := []rune(s)
-	last := runes[len(runes)-1]
-	return strings.ContainsRune("。.!！?？:：", last)
-}
-
-func (p *PDF) identifyTableLine(line struct {
-	Y           float64
-	MaxFontSize float64
-	Text        string
-	Texts       []pdf.Text
-}) (bool, []string) {
-	if len(line.Texts) < 3 {
-		return false, nil
-	}
-	var cells []string
-	var currentCell strings.Builder
-	lastX := -1.0
-	for _, t := range line.Texts {
-		if lastX != -1.0 && t.X-(lastX) > 40 {
-			content := strings.TrimSpace(currentCell.String())
-			if content != "" {
-				cells = append(cells, content)
-			}
-			currentCell.Reset()
-		}
-		currentCell.WriteString(t.S)
-		lastX = t.X + t.W
-	}
-	finalCell := strings.TrimSpace(currentCell.String())
-	if finalCell != "" {
-		cells = append(cells, finalCell)
-	}
-
-	if len(cells) >= 2 {
-		allSingleChar := true
-		for _, c := range cells {
-			r := []rune(c)
-			if len(r) > 1 {
-				allSingleChar = false
-			}
-			if len(r) > 40 {
-				return false, nil
-			}
-		}
-		if allSingleChar {
-			return false, nil
-		}
-		return true, cells
-	}
-	return false, nil
-}
-
-func MathAbs(v float64) float64 {
-	if v < 0 {
-		return -v
-	}
-	return v
-}
-
 // ToJSON 返回结构化 JSON。
 func (p *PDF) ToJSON() string {
 	res, _ := cast.ToJSON(map[string]any{
@@ -388,15 +74,7 @@ func (p *PDF) ToJSON() string {
 
 // ToMarkdown 返回 Markdown。
 func (p *PDF) ToMarkdown() string {
-	if p.Content == "" {
+	return p.Content
-		return ""
-	}
-	var sb strings.Builder
-	if title, ok := p.Metadata["title"]; ok && title != "" {
-		sb.WriteString("# " + cast.To[string](title) + "\n\n")
-	}
-	sb.WriteString(p.Content)
-	return sb.String()
 }
 
 // Save 保存（目前保存为提取后的文本）。

pptx.go

@@ -3,7 +3,6 @@ package document
 import (
 	"fmt"
 	"io"
-	"strings"
 
 	"apigo.cc/go/cast"
 	"apigo.cc/go/file"
@@ -47,41 +46,7 @@ func (p *Pptx) ToJSON() string {
 
 // ToMarkdown 返回 Markdown。
 func (p *Pptx) ToMarkdown() string {
-	if p.Content == "" {
+	return p.Content
-		return ""
-	}
-	
-	lines := strings.Split(p.Content, "\n")
-	var res []string
-	nextIsMainTitle := true
-	
-	for _, line := range lines {
-		trimmed := strings.TrimSpace(line)
-		if trimmed == "" {
-			continue
-		}
-		
-		// 检查幻灯片分隔符
-		if strings.Contains(line, "--------------------------------") {
-			res = append(res, "\n---")
-			nextIsMainTitle = true
-			continue
-		}
-		
-		// 启发式识别标题：行短、无句末标点
-		if len([]rune(trimmed)) < 50 && !strings.HasSuffix(trimmed, ".") && !strings.HasSuffix(trimmed, "。") && !strings.HasSuffix(trimmed, ":") && !strings.HasSuffix(trimmed, "：") {
-			if nextIsMainTitle {
-				res = append(res, "\n# "+trimmed)
-				nextIsMainTitle = false
-			} else {
-				res = append(res, "\n## "+trimmed)
-			}
-		} else {
-			res = append(res, trimmed)
-			// 如果该页已经有了主标题，后续的长文本不会重置标题状态
-		}
-	}
-	return strings.TrimSpace(strings.Join(res, "\n"))
 }
 
 // Save 保存文档（目前保存为提取后的文本）。