PDFToText with ITextSharp--Extract text from PDF in C# (100% .NET)(推荐）

using System;
using System.IO;
using iTextSharp.text.pdf;

namespace PdfToText
{
    /**//// <summary>
    /// Parses a PDF file and extracts the text from it.
    /// </summary>
    public class PDFParser 
    {
        /**//// BT = Beginning of a text object operator 
        /// ET = End of a text object operator
        /// Td move to the start of next line
        ///  5 Ts = superscript
        /// -5 Ts = subscript

        Fields#region Fields

        _numberOfCharsToKeep#region _numberOfCharsToKeep
        /**//// <summary>
        /// The number of characters to keep, when extracting text.
        /// </summary>
        private static int _numberOfCharsToKeep = 15;
        #endregion

        #endregion

        ExtractText#region ExtractText
        /**//// <summary>
        /// Extracts a text from a PDF file.
        /// </summary>
        /// <param name="inFileName">the full path to the pdf file.</param>
        /// <param name="outFileName">the output file name.</param>
        /// <returns>the extracted text</returns>
        public bool ExtractText(string inFileName, string outFileName)
        {
            StreamWriter outFile = null;
            try
            {
                // Create a reader for the given PDF file
                PdfReader reader = new PdfReader(inFileName);
                //outFile = File.CreateText(outFileName);
                outFile = new StreamWriter(outFileName, false, System.Text.Encoding.UTF8);
                
                Console.Write("Processing: ");
                
                int     totalLen    = 68;
                float   charUnit    = ((float)totalLen) / (float)reader.NumberOfPages;
                int     totalWritten= 0;
                float   curUnit     = 0;

                for (int page = 1; page <= reader.NumberOfPages; page++)
                {                    
                    outFile.Write(ExtractTextFromPDFBytes(reader.GetPageContent(page)) + " ");
                    
                    // Write the progress.
                    if (charUnit >= 1.0f)
                    {
                        for (int i = 0; i < (int)charUnit; i++)
                        {
                            Console.Write("#");
                            totalWritten++;
                        }
                    }
                    else
                    {
                        curUnit += charUnit;
                        if (curUnit >= 1.0f)
                        {
                            for (int i = 0; i < (int)curUnit; i++)
                            {
                                Console.Write("#");
                                totalWritten++;
                            }
                            curUnit = 0;
                        }
                        
                    }
                }

                if (totalWritten < totalLen)
                {
                    for (int i = 0; i < (totalLen - totalWritten); i++)
                    {
                        Console.Write("#");
                    }
                }
                return true;
            }
            catch
            {
                return false;
            }
            finally
            {
                if (outFile != null) outFile.Close();
            }
        }
        #endregion

        ExtractTextFromPDFBytes#region ExtractTextFromPDFBytes
        /**//// <summary>
        /// This method processes an uncompressed Adobe (text) object 
        /// and extracts text.
        /// </summary>
        /// <param name="input">uncompressed</param>
        /// <returns></returns>
        private string ExtractTextFromPDFBytes(byte[] input)
        {
            if (input == null || input.Length == 0) return "";

            try
            {
                string resultString = "";

                // Flag showing if we are we currently inside a text object
                bool inTextObject = false;

                // Flag showing if the next character is literal 
                // e.g. '\\' to get a '\' character or '\(' to get '('
                bool nextLiteral = false;

                // () Bracket nesting level. Text appears inside ()
                int bracketDepth = 0;

                // Keep previous chars to get extract numbers etc.:
                char[] previousCharacters = new char[_numberOfCharsToKeep];
                for (int j = 0; j < _numberOfCharsToKeep; j++) previousCharacters[j] = ' ';


                for (int i = 0; i < input.Length; i++)
                {
                    char c = (char)input[i];

                    if (inTextObject)
                    {
                        // Position the text
                        if (bracketDepth == 0)
                        {
                            if (CheckToken(new string[] { "TD", "Td" }, previousCharacters))
                            {
                                resultString += "\n\r";
                            }
                            else
                            {
                                if (CheckToken(new string[] {"'", "T*", "\""}, previousCharacters))
                                {
                                    resultString += "\n";
                                }
                                else
                                {
                                    if (CheckToken(new string[] { "Tj" }, previousCharacters))
                                    {
                                        resultString += " ";
                                    }
                                }
                            }
                        }

                        // End of a text object, also go to a new line.
                        if (bracketDepth == 0 && 
                            CheckToken( new string[]{"ET"}, previousCharacters))
                        {

                            inTextObject = false;
                            resultString += " ";
                        }
                        else
                        {
                            // Start outputting text
                            if ((c == '(') && (bracketDepth == 0) && (!nextLiteral))
                            {
                                bracketDepth = 1;
                            }
                            else
                            {
                                // Stop outputting text
                                if ((c == ')') && (bracketDepth == 1) && (!nextLiteral))
                                {
                                    bracketDepth = 0;
                                }
                                else
                                {
                                    // Just a normal text character:
                                    if (bracketDepth == 1)
                                    {
                                        // Only print out next character no matter what. 
                                        // Do not interpret.
                                        if (c == '\\' && !nextLiteral)
                                        {
                                            nextLiteral = true;
                                        }
                                        else
                                        {
                                            if (((c >= ' ') && (c <= '~')) ||
                                                ((c >= 128) && (c < 255)))
                                            {
                                                resultString += c.ToString();
                                            }

                                            nextLiteral = false;
                                        }
                                    }
                                }
                            }
                        }
                    }

                    // Store the recent characters for 
                    // when we have to go back for a checking
                    for (int j = 0; j < _numberOfCharsToKeep - 1; j++)
                    {
                        previousCharacters[j] = previousCharacters[j + 1];
                    }
                    previousCharacters[_numberOfCharsToKeep - 1] = c;

                    // Start of a text object
                    if (!inTextObject && CheckToken(new string[]{"BT"}, previousCharacters))
                    {
                        inTextObject = true;
                    }
                }
                return resultString;
            }
            catch
            {
                return "";
            }
        }
        #endregion

        CheckToken#region CheckToken
        /**//// <summary>
        /// Check if a certain 2 character token just came along (e.g. BT)
        /// </summary>
        /// <param name="search">the searched token</param>
        /// <param name="recent">the recent character array</param>
        /// <returns></returns>
        private bool CheckToken(string[] tokens, char[] recent)
        {
            foreach(string token in tokens)
            {
                if ((recent[_numberOfCharsToKeep - 3] == token[0]) &&
                    (recent[_numberOfCharsToKeep - 2] == token[1]) &&
                    ((recent[_numberOfCharsToKeep - 1] == ' ') ||
                    (recent[_numberOfCharsToKeep - 1] == 0x0d) ||
                    (recent[_numberOfCharsToKeep - 1] == 0x0a)) &&
                    ((recent[_numberOfCharsToKeep - 4] == ' ') ||
                    (recent[_numberOfCharsToKeep - 4] == 0x0d) ||
                    (recent[_numberOfCharsToKeep - 4] == 0x0a))
                    )
                {
                    return true;
                }
            }
            return false;
        }
        #endregion
    }
}

usage:

using System;
using System.Text;
using System.IO;

namespace PdfToText
{
    /**//// <summary>
    /// The main entry point to the program.
    /// </summary>
    class Program
    {
        static void Main(string[] args)
        {
            try
            {
                if (args.Length < 1)
                {
                    DisplayUsage();
                    return;
                }

                string file = args[0];
                if (!File.Exists(file))
                {
                    file = Path.GetFullPath(file);
                    if (!File.Exists(file))
                    {
                        Console.WriteLine("Please give in the path to the PDF file.");
                    }
                }

                PDFParser pdfParser = new PDFParser();
                pdfParser.ExtractText(file, Path.GetFileNameWithoutExtension(file)+".txt");
            }
            catch (Exception exc)
            {
                Console.WriteLine(exc);
            }
        }

        static void DisplayUsage()
        {
            Console.WriteLine();
            Console.WriteLine("Usage:\tpdftotext FILE");
            Console.WriteLine();
            Console.WriteLine("\tFILE\t the path to the PDF file, it may be relative or absolute.");
            Console.WriteLine();
        }
    }
}

问题，不支持中文，没有布局，仅仅是把每页的所以文字抽取出来，如果想真正实现PDFtoTxt，仍然有好多路要走，但毕竟是个好的开始。

from http://www.codeproject.com/useritems/PDFToText.asp

posted on 2006-06-16 07:26 RubyPDF 阅读(3302) 评论(8)  编辑 收藏 所属分类: iTextSharp(iText#)

发表评论

  回复  引用  查看    

2006-06-16 08:14 | 自適應軟件......

暈,這是個開源的,早就Release出來了.不過,有點遺憾的是,WritePDF的時候,換行的時候,不能判斷行首位字母的時候,自動換行!

另外,把文字從PDF抽出來的時候,怎麼不支持中文呢?不解.還有,既然你已經把文字抽出來了,為甚麼不能實現PDFtoTxt呢,你直接寫到Txt,或者Word不就行了?

  回复  引用  查看    

2006-06-16 09:17 | HardRock

这个不是软件，只是一段应用代码，而且也不是我写的，至于为什么不支持中文，这个就不想多说了，至少目前我还没有能力实现它。
关于你说的首位字母的问题，这个有人已经实现，但不愿意公布方法，我也没有研究过，一是能力问题，另外一个是因为写PDF不是我的研究重点。
PDFToText涉及很多问题的，有兴趣你可以看看XPDF的一个工具就比较清楚了。

  回复  引用  查看    

2006-07-20 11:46 | zwg51666 [未注册用户]

谢谢HardRock ,那多页tif怎么处理,我现在每次只能得到第一页,谢谢

  回复  引用  查看    

2006-07-20 11:55 | HardRock

@zwg51666
你在搞什么？再把一个问题发几遍或者到处乱发，我删除了!

  回复  引用  查看    

2006-07-20 12:01 | zwg51666 [未注册用户]

你好,你看下我的代码,为什么,他还是裁图了,我想得到缩放的效果,
if (tif1.ScaledWidth>760)
{
float tempin=(760/tif1.ScaledWidth);
tempin*=100;
this.textBox1.Text=tempin.ToString();
tif1.ScalePercent(tempin);

}
谢谢

  回复  引用  查看    

2006-07-20 18:13 | zwg51666 [未注册用户]

我的问题解决了,谢

  回复  引用  查看    

2007-03-07 16:40 | minghong [未注册用户]

用xpdf（http://www.foolabs.com/xpdf/about.html）中的pdftotext，加上enc就可以抽出中文了，例如
pdftotext -layout -enc UTF-8 test.pdf

  回复  引用  查看    

2007-10-11 05:06 | Ottoniel [未注册用户]

Esta bien, pero no entiendo nada :D