fix(转换能力): 修复PDF转换回归
build.yaml / build (push) Successful in 4m39s

This commit is contained in:
xiaoqidun committed 2026-10-21 10:35:24 +08:00
1 parent 580592bbf8
commit b4d566019d
11 files changed
+793 -171

No files matched your search

+95 -69
View File
@@ -18,7 +18,6 @@ import (
"bytes"
"context"
"encoding/xml"
"errors"
"fmt"
"math"
"strconv"
@@ -36,12 +35,15 @@ type pdfDestinationFontBackend struct {
// 入参: ctx 取消上下文
// 返回: error 动作、编码或取消错误
func (p *pdfImporter) documentOpenActions(ctx context.Context) error {
raw, err := p.pdfCatalogObject("OpenAction")
if err != nil {
return err
}
opened, err := p.reader.ReadOpenAction(ctx)
if err != nil {
if p.warning == nil || !errors.Is(err, pdfgo.ErrInvalidDestination) {
if err := p.recoverPDFObject(ctx, "OpenAction", raw, err); err != nil {
return err
}
p.warning(pdfgo.Diagnostic{Message: fmt.Sprintf("invalid PDF initial destination ignored: %v", err)})
}
var object pdfgo.Object
if opened.Action != nil {
@@ -52,7 +54,7 @@ func (p *pdfImporter) documentOpenActions(ctx context.Context) error {
object = pdfgo.Dictionary{"S": pdfgo.Name("GoTo"), "D": array}
}
currentPage := 0
actions, err := p.eventActions(ctx, object, "DO", pdfgo.Annotation{}, &currentPage)
actions, err := p.eventActions(ctx, object, "DO", pdfgo.Annotation{}, &currentPage, raw)
if err != nil {
return err
}
@@ -61,11 +63,18 @@ func (p *pdfImporter) documentOpenActions(ctx context.Context) error {
}
triggers, err := p.reader.ReadDocumentTriggers(ctx)
if err != nil {
return err
object, readErr := p.pdfCatalogObject("AA")
if readErr != nil {
return readErr
}
return p.recoverPDFObject(ctx, "Document.AA", object, err)
}
for _, trigger := range triggers {
if _, err := p.reader.ReadJavaScriptAction(ctx, trigger.Action); err != nil {
return err
if err := p.recoverPDFObject(ctx, "Document.AA."+string(trigger.Event), trigger.Action, err); err != nil {
return err
}
continue
}
if err := p.unmappedTrigger(ctx, "document", trigger); err != nil {
return err
@@ -81,7 +90,7 @@ func (p *pdfImporter) pageOpenActions(ctx context.Context, page *pdfgo.Page) err
p.pageActionPage = p.page
triggers, err := p.reader.ReadPageTriggers(ctx, page)
if err != nil {
return err
return p.recoverPDFObject(ctx, "Page.AA", page.Dictionary["AA"], err)
}
for _, trigger := range triggers {
if trigger.Event != "O" {
@@ -102,32 +111,59 @@ func (p *pdfImporter) pageOpenActions(ctx context.Context, page *pdfgo.Page) err
}
// eventActions 按原顺序转换动作链,保留事件类型与链内跳转页状态
// 入参: ctx 取消上下文, object 动作对象, event OFD事件, source 触发注解,零值表示非图元动作, currentPage 当前页面
// 入参: ctx 取消上下文, object 动作对象, event OFD事件, source 触发注解,零值表示非图元动作, currentPage 当前页面, original 可选原始来源
// 返回: []Action 转换动作, error 动作或取消错误
func (p *pdfImporter) eventActions(ctx context.Context, object pdfgo.Object, event string, source pdfgo.Annotation, currentPage *int) ([]Action, error) {
func (p *pdfImporter) eventActions(ctx context.Context, object pdfgo.Object, event string, source pdfgo.Annotation, currentPage *int, original ...pdfgo.Object) ([]Action, error) {
raw := object
if len(original) != 0 {
raw = original[0]
}
retained, warned := false, false
warning := p.warning
if warning != nil {
p.warning = func(diagnostic pdfgo.Diagnostic) {
warned = true
warning(diagnostic)
}
defer func() { p.warning = warning }()
}
retain := func(cause error) error {
if cause != nil && (p.warning == nil || !pdfAuxiliaryError(cause)) {
return cause
}
if retained {
if cause != nil {
p.warning(pdfgo.Diagnostic{Message: "PDF Action." + event + " retained as metadata: " + cause.Error()})
}
return ctx.Err()
}
var err error
if cause != nil {
err = p.recoverPDFObject(ctx, "Action."+event, raw, cause)
} else if p.warning != nil {
err = p.preservePDFObject(ctx, "Action."+event, raw)
}
retained = err == nil
return err
}
var actions []Action
var conversionErr error
err := p.reader.WalkActions(ctx, object, func(info pdfgo.ActionInfo) error {
annotation := source
annotation.Dictionary = pdfgo.Dictionary{"A": info.Dictionary}
converted, err := p.linkActions(annotation, p.warning == nil, currentPage)
warned = false
warnings := len(p.report.Warnings)
converted, err := p.convertLinkActions(annotation, p.warning == nil, currentPage)
if err != nil {
if errors.Is(err, errPDFMediaRead) {
conversionErr = retain(err)
*currentPage = -1
return conversionErr
}
if p.warning != nil && (converted == nil || warned || len(p.report.Warnings) != warnings) {
if err := retain(nil); err != nil {
conversionErr = err
return err
}
var unsupported *pdfgo.UnsupportedError
if event != "CLICK" && p.warning != nil && errors.Is(err, pdfgo.ErrInvalidAction) {
p.warning(pdfgo.Diagnostic{Message: fmt.Sprintf("invalid PDF %s action ignored: %v", event, err)})
return nil
}
if event != "CLICK" && p.warning != nil && errors.Is(err, pdfgo.ErrInvalidDestination) {
p.warning(pdfgo.Diagnostic{Message: fmt.Sprintf("invalid PDF %s destination ignored: %v", event, err)})
return nil
}
if event != "CLICK" && p.warning != nil && errors.As(err, &unsupported) {
p.warning(pdfgo.Diagnostic{Message: fmt.Sprintf("PDF %s action not transferred to OFD: %s", event, unsupported.Feature)})
return nil
}
return err
}
for _, action := range converted {
action.Event = event
@@ -135,8 +171,14 @@ func (p *pdfImporter) eventActions(ctx context.Context, object pdfgo.Object, eve
}
return nil
})
if conversionErr != nil {
return nil, conversionErr
}
if err != nil {
return nil, err
if err := retain(err); err != nil {
return nil, err
}
*currentPage = -1
}
return actions, ctx.Err()
}
@@ -146,14 +188,10 @@ func (p *pdfImporter) eventActions(ctx context.Context, object pdfgo.Object, eve
// 返回: error 动作链、取消或严格模式错误
func (p *pdfImporter) unmappedTrigger(ctx context.Context, owner string, trigger pdfgo.ActionTrigger) error {
if err := p.reader.WalkActions(ctx, trigger.Action, func(pdfgo.ActionInfo) error { return nil }); err != nil {
return err
return p.recoverPDFObject(ctx, owner+".AA."+string(trigger.Event), trigger.Action, err)
}
feature := owner + " trigger " + string(trigger.Event)
if p.warning == nil {
return &pdfgo.UnsupportedError{Feature: feature}
}
p.warning(pdfgo.Diagnostic{Message: "PDF " + feature + " not transferred to OFD"})
return nil
return p.recoverPDFObject(ctx, owner+".AA."+string(trigger.Event), trigger.Action, &pdfgo.UnsupportedError{Feature: feature})
}
// annotationTriggerActions 按事件转换注解动作,页面打开动作独立于注解外观
@@ -165,7 +203,7 @@ func (p *pdfImporter) annotationTriggerActions(ctx context.Context, annotation p
}
triggers, err := p.reader.ReadAnnotationTriggers(ctx, annotation)
if err != nil {
return nil, err
return nil, p.recoverPDFObject(ctx, "Annotation.AA", annotation.Dictionary["AA"], err)
}
if len(triggers) == 0 {
return nil, ctx.Err()
@@ -193,7 +231,10 @@ func (p *pdfImporter) annotationTriggerActions(ctx context.Context, annotation p
switch trigger.Event {
case "U", "PO":
default:
return nil, &pdfgo.UnsupportedError{Feature: "annotation trigger " + string(trigger.Event)}
if err := p.unmappedTrigger(ctx, "annotation", trigger); err != nil {
return nil, err
}
continue
}
currentPage := p.page
if event == "PO" {
@@ -258,7 +299,8 @@ func (p *pdfImporter) destinationBounds(page *pdfgo.Page) (Box, error) {
local := pdfImporter{
resources: p.resources,
ctx: p.ctx, reader: p.reader, editor: editor, renderer: p.renderer,
matrix: matrix, pageBox: page.CropBox, pageWidth: width, pageHeight: height,
warning: p.warning,
matrix: matrix, pageBox: page.CropBox, pageWidth: width, pageHeight: height,
rasterDPI: p.rasterDPI, fontIDs: make(map[*pdfgo.Font]string),
resolveFile: p.resolveFile, resolveReference: p.resolveReference,
references: &codecpdf.ReferenceCache{},
@@ -320,17 +362,15 @@ func (p *pdfImporter) destinationBounds(page *pdfgo.Page) (Box, error) {
// 返回: error 严格模式转换错误或取消错误
func (p *pdfImporter) outlines(ctx context.Context) error {
count := 0
var conversionErr error
data, err := encodeOFDXMLContext(ctx, func(x *ofdXML) {
x.root("Outlines", nil)
depth := 0
err := p.reader.WalkOutlines(ctx, func(level int, item pdfgo.OutlineItem) error {
actions, err := p.outlineActions(ctx, item)
if err != nil {
if p.warning == nil || ctx.Err() != nil || errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) || errors.Is(err, errPDFMediaRead) {
return err
}
p.warning(pdfgo.Diagnostic{Message: fmt.Sprintf("PDF outline %q actions not transferred to OFD: %v", item.Title, err)})
actions = nil
conversionErr = err
return err
}
for depth > level {
x.end("OutlineElem")
@@ -354,15 +394,21 @@ func (p *pdfImporter) outlines(ctx context.Context) error {
}
x.end("Outlines")
})
if conversionErr != nil {
return conversionErr
}
if err != nil {
if ctx.Err() != nil {
return ctx.Err()
}
if p.warning == nil || errors.Is(err, context.Canceled) || errors.Is(err, context.DeadlineExceeded) || errors.Is(err, errPDFMediaRead) {
if p.warning == nil || !pdfAuxiliaryError(err) {
return err
}
p.warning(pdfgo.Diagnostic{Message: fmt.Sprintf("PDF outlines not transferred to OFD: %v", err)})
return nil
object, readErr := p.pdfCatalogObject("Outlines")
if readErr != nil {
return readErr
}
return p.recoverPDFObject(ctx, "Outlines", object, err)
}
if count > 0 {
p.editor.outlines = bytes.TrimPrefix(data, []byte(xml.Header))
@@ -374,42 +420,22 @@ func (p *pdfImporter) outlines(ctx context.Context) error {
// 入参: ctx 取消上下文, item PDF目录节点
// 返回: []Action OFD动作序列, error 动作或取消错误
func (p *pdfImporter) outlineActions(ctx context.Context, item pdfgo.OutlineItem) ([]Action, error) {
raw := item.Action
object, err := p.reader.Resolve(item.Action)
if err != nil {
return nil, err
return nil, p.recoverPDFObject(ctx, "Outline.A", item.Action, err)
}
dest, err := p.reader.Resolve(item.Destination)
if err != nil {
return nil, err
return nil, p.recoverPDFObject(ctx, "Outline.Dest", item.Destination, err)
}
if dest != nil {
if object != nil {
return nil, fmt.Errorf("PDF outline has both destination and action")
return nil, p.recoverPDFObject(ctx, "Outline", pdfgo.Dictionary{"A": item.Action, "Dest": item.Destination}, fmt.Errorf("PDF outline has both destination and action"))
}
object = pdfgo.Dictionary{"S": pdfgo.Name("GoTo"), "D": dest}
raw = item.Destination
}
currentPage := -1
var actions []Action
err = p.reader.WalkActions(ctx, object, func(info pdfgo.ActionInfo) error {
converted, err := p.linkActions(pdfgo.Annotation{Dictionary: pdfgo.Dictionary{"A": info.Dictionary}}, p.warning == nil, &currentPage)
if err != nil {
if errors.Is(err, errPDFMediaRead) {
return err
}
var unsupported *pdfgo.UnsupportedError
if p.warning != nil && (errors.As(err, &unsupported) || errors.Is(err, pdfgo.ErrInvalidDestination) || errors.Is(err, pdfgo.ErrInvalidAction)) {
p.warning(pdfgo.Diagnostic{Message: fmt.Sprintf("PDF outline action not transferred to OFD: %v", err)})
currentPage = -1
return nil
}
return err
}
if converted != nil {
actions = append(actions, converted...)
} else {
currentPage = -1
}
return nil
})
return actions, err
return p.eventActions(ctx, object, "CLICK", pdfgo.Annotation{}, &currentPage, raw)
}