从 PDF 页面获取文本时，iTextSharp 异常“堆栈为空”

Question

我正在尝试遍历 PDF 上的每一页以查找特定关键字。除了 one

之外，代码在其他 PDF 上也能正常工作

我的代码

Using oReader As New pdf.PdfReader(pdfFilename)

    For pCurrent = oReader.NumberOfPages To 1 Step -1
        Dim strategy As pdf.parser.ITextExtractionStrategy = New pdf.parser.SimpleTextExtractionStrategy()
        Dim pageText As String = pdf.parser.PdfTextExtractor.GetTextFromPage(oReader, pCurrent, strategy)

        '
        'search for keywords
        '
        'FindVOI

    Next 'proceed next page

End Using

这是导致此异常的代码的 sn-p，

Dim pageText As String = pdf.parser.PdfTextExtractor.GetTextFromPage(oReader, pCurrent, strategy)

在此 PDF 的第 98 页上显示异常 Stack empty，有什么问题吗？

完全例外：

Exception thrown: 'System.InvalidOperationException' in System.dll
System.Transactions Critical: 0 : <TraceRecord xmlns="http://schemas.microsoft.com/2004/10/E2ETraceEvent/TraceRecord" Severity="Critical"><TraceIdentifier>http://msdn.microsoft.com/TraceCodes/System/ActivityTracing/2004/07/Reliability/Exception/Unhandled</TraceIdentifier><Description>Unhandled exception</Description><AppDomain>VipMonitorService.vshost.exe</AppDomain><Exception><ExceptionType>System.InvalidOperationException, mscorlib, Version=4.0.0.0, Culture=neutral, PublicKeyToken=b77a5c561934e089</ExceptionType><Message>Stack empty.</Message><StackTrace>   at System.ThrowHelper.ThrowInvalidOperationException(ExceptionResource resource)
   at System.Collections.Generic.Stack`1.Pop()
   at iTextSharp.text.pdf.parser.PdfContentStreamProcessor.EndMarkedContentC.Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List`1 operands)
   at iTextSharp.text.pdf.parser.PdfContentStreamProcessor.InvokeOperator(PdfLiteral oper, List`1 operands)
   at iTextSharp.text.pdf.parser.PdfContentStreamProcessor.ProcessContent(Byte[] contentBytes, PdfDictionary resources)
   at iTextSharp.text.pdf.parser.PdfReaderContentParser.ProcessContent[E](Int32 pageNumber, E renderListener, IDictionary`2 additionalContentOperators)
   at iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(PdfReader reader, Int32 pageNumber, ITextExtractionStrategy strategy)
   at WatcherApp.VipMonitorService.PDFHelper.FindVOI(List`1 voiList, String pdfFilename, Boolean searchFromLast, Int32 searchNumberOfPagesInPercent) in \\Mac\Dropbox\git\Personal\WatcherApp\VipMonitorService\PDFHelper.vb:line 59
   at WatcherApp.VipMonitorService.Controller.ProcessAnnualReport(Announcement a) in \\Mac\Dropbox\git\Personal\WatcherApp\VipMonitorService\Controller.vb:line 456
   at WatcherApp.VipMonitorService.Controller.ProcessARInQueueThread() in \\Mac\Dropbox\git\Personal\WatcherApp\VipMonitorService\Controller.vb:line 362
   at WatcherApp.VipMonitorService.Controller._Lambda$__40-0() in \\Mac\Dropbox\git\Personal\WatcherApp\VipMonitorService\Controller.vb:line 339
   at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state, Boolean preserveSyncCtx)
   at System.Threading.ExecutionContext.Run(ExecutionContext executionContext, ContextCallback callback, Object state, Boolean preserveSyncCtx)
   at System.Threading.ExecutionContext.Run(ExecutionContext executionContext, ContextCallback callback, Object state)
   at System.Threading.ThreadHelper.ThreadStart()</StackTrace><ExceptionString>System.InvalidOperationException: Stack empty.
   at System.ThrowHelper.ThrowInvalidOperationException(ExceptionResource resource)
   at System.Collections.Generic.Stack`1.Pop()
   at iTextSharp.text.pdf.parser.PdfContentStreamProcessor.EndMarkedContentC.Invoke(PdfContentStreamProcessor processor, PdfLiteral oper, List`1 operands)
   at iTextSharp.text.pdf.parser.PdfContentStreamProcessor.InvokeOperator(PdfLiteral oper, List`1 operands)
   at iTextSharp.text.pdf.parser.PdfContentStreamProcessor.ProcessContent(Byte[] contentBytes, PdfDictionary resources)
   at iTextSharp.text.pdf.parser.PdfReaderContentParser.ProcessContent[E](Int32 pageNumber, E renderListener, IDictionary`2 additionalContentOperators)
   at iTextSharp.text.pdf.parser.PdfTextExtractor.GetTextFromPage(PdfReader reader, Int32 pageNumber, ITextExtractionStrategy strategy)
   at WatcherApp.VipMonitorService.PDFHelper.FindVOI(List`1 voiList, String pdfFilename, Boolean searchFromLast, Int32 searchNumberOfPagesInPercent) in \\Mac\Dropbox\git\Personal\WatcherApp\VipMonitorService\PDFHelper.vb:line 59
   at WatcherApp.VipMonitorService.Controller.ProcessAnnualReport(Announcement a) in \\Mac\Dropbox\git\Personal\WatcherApp\VipMonitorService\Controller.vb:line 456
   at WatcherApp.VipMonitorService.Controller.ProcessARInQueueThread() in \\Mac\Dropbox\git\Personal\WatcherApp\VipMonitorService\Controller.vb:line 362
   at WatcherApp.VipMonitorService.Controller._Lambda$__40-0() in \\Mac\Dropbox\git\Personal\WatcherApp\VipMonitorService\Controller.vb:line 339
   at System.Threading.ExecutionContext.RunInternal(ExecutionContext executionContext, ContextCallback callback, Object state, Boolean preserveSyncCtx)
   at System.Threading.ExecutionContext.Run(ExecutionContext executionContext, ContextCallback callback, Object state, Boolean preserveSyncCtx)
   at System.Threading.ExecutionContext.Run(ExecutionContext executionContext, ContextCallback callback, Object state)
   at System.Threading.ThreadHelper.ThreadStart()</ExceptionString></Exception></TraceRecord>

Answer 1

在此PDF 的第 98 页上显示异常堆栈为空，有什么想法吗？

堆栈跟踪显示 堆栈为空 发生在 iTextSharp.text.pdf.parser.PdfContentStreamProcessor.EndMarkedContentC.Invoke。因此，我们应该看看开始和结束标记的内容运算符：

标签 BMC 开始一个由平衡 EMC 运算符终止的标记内容序列。 tag 应该是一个名称对象，表明序列的作用或意义。

标签属性 BDC 以关联的属性列表开始标记内容序列，由平衡 EMC 运算符终止。 tag 应该是一个名称对象，指示序列的作用或意义。 properties 应该是包含属性列表的内联字典或在当前资源字典的 Properties 子字典中与之关联的名称对象（参见 14.6.2，“Property Lists ”)。

— 电磁兼容 结束由 BMC 或 BDC 运算符开始的标记内容序列。

（表 320 – 标记内容运算符，ISO 32000-1）

如果您查看相关页面上标记内容的 BDC/BMC 和 EMC 开始和结束，您会看到：

/Artifact <</O /Layout >>BDC
EMC 
/Artifact <</O /Layout >>BDC  
EMC  
/Artifact <</O /Layout >>BDC  
EMC 
/Artifact <</BBox [0 33.8887 407.4289 0 ]/O /Layout >>BDC  
EMC 
EMC
...

因此，有一个多余的 EMC 运算符，没有 BMC 或 BDC 运算符来结束其标记的内容。 p>

因此，此文档不是有效的 PDF；特别是它的标记内容结构被破坏了。

---

话虽如此，如果 iTextSharp 会在 Pop 之前检查堆栈并可选择抛出更明显的异常或忽略 EMC 运算符，这将是合适的。