feat(编辑搜索): 完善附件编辑与跨页搜索

2026-10-30 01:27:26 +08:00 · 2026-09-27 22:31:06 +08:00
1 parent 8d91752943
commit 31e67b2721
@@ -9,7 +9,7 @@ require (
 	github.com/tdewolff/canvas v0.0.0-20260923214215-09804640d00c
 	github.com/tdewolff/font v0.0.0-20260913163313-54f98bb59ee6
 	github.com/xiaoqidun/jbig2 v1.1.1
-	github.com/xiaoqidun/pdfgo v0.0.0-20260927125911-04f5476a9051
+	github.com/xiaoqidun/pdfgo v0.0.0-20260927142759-12ed20917f60
 	golang.org/x/crypto v0.57.0
 	golang.org/x/image v0.46.0
 	golang.org/x/text v0.42.0
@@ -62,8 +62,8 @@ github.com/tdewolff/test v1.0.12 h1:7F21DqIajswxuche0geHdrUZRCWE4oko4b7bcmkkrxk=
 github.com/tdewolff/test v1.0.12/go.mod h1:XPuWBzvdUzhCuxWO1ojpXsyzsA5bFoS3tO/Q3kFuTG8=
 github.com/xiaoqidun/jbig2 v1.1.1 h1:SjliS9ZO0e9HX5oYUfBuXrfFsXhWricrSz+XY4zORDI=
 github.com/xiaoqidun/jbig2 v1.1.1/go.mod h1:8QonVWtENmh2kc6l+leqeGl7psb79DWlHDkCpdChjfQ=
-github.com/xiaoqidun/pdfgo v0.0.0-20260927125911-04f5476a9051 h1:VhoxfD1yFJMBcgThFZCo480VVBEU5dB8qAZXWCUSeqI=
-github.com/xiaoqidun/pdfgo v0.0.0-20260927125911-04f5476a9051/go.mod h1:4Pf0JO0BEf1uzRY+uZjlMjN/OPn3PYa0YMi6cqPPTLQ=
+github.com/xiaoqidun/pdfgo v0.0.0-20260927142759-12ed20917f60 h1:0JC6RK3DMb11RIVoaSSlprHo9oAZb+YyfUThEsTs0Bk=
+github.com/xiaoqidun/pdfgo v0.0.0-20260927142759-12ed20917f60/go.mod h1:4Pf0JO0BEf1uzRY+uZjlMjN/OPn3PYa0YMi6cqPPTLQ=
 github.com/xyproto/randomstring v1.0.5 h1:YtlWPoRdgMu3NZtP45drfy1GKoojuR7hmRcnhZqKjWU=
 github.com/xyproto/randomstring v1.0.5/go.mod h1:rgmS5DeNXLivK7YprL0pY+lTuhNQW3iGxZ18UQApw/E=
 github.com/yuin/goldmark v1.8.5 h1:r6N5afV5qj/5S4UTch8agZHJ8UxNCMwX7WjkkJam2NA=
@@ -20,14 +20,25 @@ import (
 	"fmt"
 	"path"
 	"strings"
+	"time"
+	"unicode/utf8"
 )

+// Attachments 获取当前编辑状态的附件列表，返回值与编辑器相互独立
+// 返回: []Attachment 附件信息, error 结构错误
+func (e *Editor) Attachments() ([]Attachment, error) {
+	if e.source == nil {
+		return nil, nil
+	}
+	return e.source.reader.Attachments()
+}
+
 // AddAttachment 添加文档附件，保留原始数据，一次操作计入一条撤销记录
 // 入参: name 显示文件名, data 文件数据
 // 返回: string 附件标识, error 文件名或文档结构错误
 func (e *Editor) AddAttachment(name string, data []byte) (string, error) {
-	if strings.TrimSpace(name) == "" {
-		return "", fmt.Errorf("attachment name is required")
+	if err := validateAttachmentName(name); err != nil {
+		return "", err
 	}
 	if err := e.prepareSourceIDs(); err != nil {
 		return "", err
@@ -103,3 +114,131 @@ func (e *Editor) AddAttachment(name string, data []byte) (string, error) {
 	}
 	return id, nil
 }
+
+// RenameAttachment 修改附件显示名称，保留格式、内容及扩展字段，相同名称不产生撤销记录
+// 入参: id 附件标识, name 显示名称
+// 返回: error 名称或附件结构错误
+func (e *Editor) RenameAttachment(id, name string) error {
+	if err := validateAttachmentName(name); err != nil {
+		return err
+	}
+	return e.editAttachment(id, func(data []byte, node *editorXML, parts map[string][]byte) ([]byte, error) {
+		if node.attr("Name") == name {
+			return data, nil
+		}
+		return editorXMLAttribute(data, node, "Name", name)
+	})
+}
+
+// ReplaceAttachment 替换附件内容，使用独立资源避免影响共享引用，保留名称和格式
+// 入参: id 附件标识, data 文件数据，调用后可复用
+// 返回: error 附件结构或提交错误
+func (e *Editor) ReplaceAttachment(id string, data []byte) error {
+	return e.editAttachment(id, func(fragment []byte, node *editorXML, parts map[string][]byte) ([]byte, error) {
+		if err := e.prepareSourceIDs(); err != nil {
+			return nil, err
+		}
+		file := e.packageName("Attachments/Attachment_" + e.nextID())
+		var location bytes.Buffer
+		if err := xml.EscapeText(&location, []byte("/"+file)); err != nil {
+			return nil, err
+		}
+		if child := node.child("FileLoc"); child != nil {
+			fragment = editorPatchXML(fragment, []editorXMLPatch{editorXMLContent(fragment, child, location.Bytes())})
+		} else if node.open == node.end {
+			fragment = editorPatchXML(fragment, []editorXMLPatch{editorXMLContent(fragment, node, editorXMLText("FileLoc", "/"+file))})
+		} else {
+			fragment = editorPatchXML(fragment, []editorXMLPatch{{node.open, node.open, editorXMLText("FileLoc", "/"+file)}})
+		}
+		for _, attr := range []struct{ name, value string }{
+			{"Size", ofdNumber(float64(len(data)) / 1024)}, {"ModDate", time.Now().UTC().Format(time.RFC3339)},
+		} {
+			root, err := parseEditorXML(fragment)
+			if err != nil {
+				return nil, err
+			}
+			fragment, err = editorXMLAttribute(fragment, root, attr.name, attr.value)
+			if err != nil {
+				return nil, err
+			}
+		}
+		parts[file] = bytes.Clone(data)
+		return fragment, nil
+	})
+}
+
+// editAttachment 对单个附件片段执行原子编辑，兼容独立附件文件及内联声明
+// 入参: id 附件标识, change 片段及资源修改函数
+// 返回: error 结构、标识或提交错误
+func (e *Editor) editAttachment(id string, change func([]byte, *editorXML, map[string][]byte) ([]byte, error)) error {
+	base := e.source
+	if base == nil {
+		return fmt.Errorf("attachment not found: %s", id)
+	}
+	name := cleanPackagePath(base.reader.OFD.DocBody[0].DocRoot)
+	external := strings.TrimSpace(base.document.Attachments.Path) != ""
+	if external {
+		name = base.reader.ResPath(base.document.Attachments.Path)
+	}
+	data, err := base.reader.readFile(name)
+	if err != nil {
+		return err
+	}
+	root, err := parseEditorXML(data)
+	if err != nil {
+		return err
+	}
+	if !external {
+		root = root.child("Attachments")
+	}
+	var found *editorXML
+	if root != nil {
+		for _, child := range root.children {
+			if child.name.Local == "Attachment" && child.attr("ID") == id {
+				if found != nil {
+					return fmt.Errorf("duplicate attachment ID: %s", id)
+				}
+				found = child
+			}
+		}
+	}
+	if found == nil {
+		return fmt.Errorf("attachment not found: %s", id)
+	}
+	fragment, err := editorXMLStandalone(data[found.start:found.end], found)
+	if err != nil {
+		return err
+	}
+	node, err := parseEditorXML(fragment)
+	if err != nil {
+		return err
+	}
+	parts := make(map[string][]byte)
+	updated, err := change(fragment, node, parts)
+	if err != nil {
+		return err
+	}
+	if bytes.Equal(updated, fragment) && len(parts) == 0 {
+		return nil
+	}
+	parts[name] = editorPatchXML(data, []editorXMLPatch{{found.start, found.end, updated}})
+	return e.commitAnnotationParts(base, parts)
+}
+
+// validateAttachmentName 校验非空附件名及XML可无损表示的Unicode字符
+// 入参: name 显示名称
+// 返回: error 名称错误
+func validateAttachmentName(name string) error {
+	if strings.TrimSpace(name) == "" {
+		return fmt.Errorf("attachment name is required")
+	}
+	if !utf8.ValidString(name) {
+		return fmt.Errorf("attachment name is not valid UTF-8")
+	}
+	for _, char := range name {
+		if char < 0x20 && char != '\t' && char != '\n' && char != '\r' || char == 0xfffe || char == 0xffff {
+			return fmt.Errorf("attachment name contains an invalid XML character")
+		}
+	}
+	return nil
+}
@@ -21,6 +21,7 @@ import (
 	"fmt"
 	"io"
 	"path"
+	"slices"
 	"strings"
 )

@@ -600,7 +601,7 @@ func (r *Reader) Bookmarks() ([]Bookmark, error) {
 	return doc.Bookmarks.Bookmark, nil
 }

-// Attachments 获取附件列表
+// Attachments 获取独立的附件列表，修改返回值不影响阅读器缓存
 // 返回: []Attachment 附件列表, error 错误信息
 func (r *Reader) Attachments() ([]Attachment, error) {
 	doc, err := r.Doc()
@@ -624,7 +625,14 @@ func (r *Reader) Attachments() ([]Attachment, error) {
 		}
 		doc.Attachments.Attachment = attachments.Attachment
 	}
-	return doc.Attachments.Attachment, nil
+	attachments := slices.Clone(doc.Attachments.Attachment)
+	for i := range attachments {
+		if attachments[i].Size != nil {
+			size := *attachments[i].Size
+			attachments[i].Size = &size
+		}
+	}
+	return attachments, nil
 }

 // OpenAttachment 打开附件文件流，调用方负责关闭，完整读取后校验ZIP数据
@@ -15,6 +15,7 @@
 package ofdgo

 import (
+	"context"
 	"encoding/xml"
 	"fmt"
 	"io"
@@ -171,16 +172,42 @@ func (p *PageText) writeTo(writer io.Writer, separator bool) (int64, error) {
 // 返回: []TextMatch 匹配结果及上下文
 func (p *PageText) Search(query string) []TextMatch {
 	var matches []TextMatch
+	_ = p.SearchEach(context.Background(), query, func(match TextMatch) error {
+		matches = append(matches, match)
+		return nil
+	})
+	return matches
+}
+
+// SearchEach 逐项返回忽略大小写的字面匹配，不累积全部结果，在对象与匹配之间检查取消
+// 入参: ctx 取消上下文, query 搜索文字, visit 匹配访问函数，可返回错误停止
+// 返回: error 取消或访问错误
+func (p *PageText) SearchEach(ctx context.Context, query string, visit func(TextMatch) error) error {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
+	return p.searchEach(ctx, foldText(query), visit)
+}
+
+// searchEach 使用预折叠的查询执行搜索，供跨页搜索复用
+// 入参: ctx 取消上下文, query 已折叠查询, visit 匹配访问函数
+// 返回: error 取消或访问错误
+func (p *PageText) searchEach(ctx context.Context, query string, visit func(TextMatch) error) error {
 	if query == "" {
-		return matches
+		return ctx.Err()
 	}
-	query = foldText(query)
 	length := utf8.RuneCountInString(query)
 	for index, run := range p.Runs {
+		if err := ctx.Err(); err != nil {
+			return err
+		}
 		text := foldText(run.Text)
 		var runes []rune
 		codeOffset := 0
 		for offset := 0; offset < len(text); {
+			if err := ctx.Err(); err != nil {
+				return err
+			}
 			found := strings.Index(text[offset:], query)
 			if found < 0 {
 				break
@@ -203,12 +230,48 @@ func (p *PageText) Search(query string) []TextMatch {
 					match.Boxes = append(match.Boxes, box)
 				}
 			}
-			matches = append(matches, match)
+			if err := visit(match); err != nil {
+				return err
+			}
 			offset += len(query)
 			codeOffset = end
 		}
 	}
-	return matches
+	return ctx.Err()
+}
+
+// SearchText 逐页搜索原文并立即返回匹配，不保留整本文字或识别图像
+// 入参: ctx 取消上下文，页面提取之间及匹配时检查, query 搜索文字, visit 零基页码和匹配访问函数, indices 页码，省略则全部，按原页序去重
+// 返回: error 页面提取、取消或访问错误
+func (r *Renderer) SearchText(ctx context.Context, query string, visit func(int, TextMatch) error, indices ...int) error {
+	if err := ctx.Err(); err != nil {
+		return err
+	}
+	if query == "" {
+		return nil
+	}
+	count, err := r.Reader.PageCount()
+	if err != nil {
+		return err
+	}
+	indices, err = exportPageIndices(count, indices)
+	if err != nil {
+		return err
+	}
+	query = foldText(query)
+	for _, index := range indices {
+		if err := ctx.Err(); err != nil {
+			return err
+		}
+		text, err := r.PageTextByIndex(index)
+		if err != nil {
+			return fmt.Errorf("failed to search page %d: %w", index+1, err)
+		}
+		if err := text.searchEach(ctx, query, func(match TextMatch) error { return visit(index, match) }); err != nil {
+			return err
+		}
+	}
+	return ctx.Err()
 }

 // foldText 统一Unicode大小写等价字符，保留字符数量