WordSplit/WindowsFormsApp1/WordAnalyze.cs

using Word = Microsoft.Office.Interop.Word;
using System.Diagnostics;
using Logger = Log.Log;
using System.Text.RegularExpressions;
using System.Collections;

namespace WordAnalyze
{
    public class Analyze
    {
        private string[] bigQuestionNum = { "一、", "二、", "三、", "四、", "五、", "六、", "七、", "八、", "九、" };
        private ArrayList smallQuestionNum = new ArrayList();
        private ArrayList smallQuestionNum2 = new ArrayList();
        private Word.ApplicationClass app;
        public Analyze()
        {
            //  开始app应用
            app = new Word.ApplicationClass();

            // 初始化小题目的编号
            for (int i = 1; i < 200; i++)
            {
                string num = i.ToString();
                smallQuestionNum.Add(num + ".");
                smallQuestionNum2.Add(num + "．");
            }
        }

        public void Close()
        {
            if (app != null)
            {
                app.Quit();
            }
        }

        public static string splitChar = "---";
        public string path = "";

        public void SetPath(string path)
        {
            this.path = path;
        }

        public string AnalyzeFile(string fileName)
        {
            if (!ValidFileName(fileName))
            {
                Logger.D("AnalyzeFile with invalid file name {0}", fileName);
                return "无效的文件名";       
            }

            object fn = fileName;

            //Word.ApplicationClass app = new Word.ApplicationClass(); // 打开word应用，只会打开一个，即单例模式
            Word.Document doc = null; // 源word文件对象
            try
            {
                doc = app.Documents.Open(ref fn);

                doc.Paragraphs.Add();

                Word.Paragraphs garapraph = doc.Paragraphs; // 源文件的内容
                //bool insert = false; //判断是否有插入数据到新文档中，如果有则保存新word，否则直接关掉新word
                //int fileIndex = 1; // 切割后word的文件序号，例子: test1.doc, test2.doc

                Word.Document newDoc = null;
                newDoc = app.Documents.Add();
                // 遍历word，通过 --- 三个横线分割
                return dot(doc, fileName);
                
            }
            catch (System.Exception e)
            {
                Logger.E("analyze with file name({0}) error ->({1})", fileName, e.Message.ToString());
                return e.Message.ToString();
            }
        }


        // 判断文件后缀名是否是.doc 或者.docx
        public bool ValidFileName(string fileName)
        {
            if (fileName == null || fileName.Length < 1)
            {
                return false;
            }
            if (!fileName.EndsWith(".doc") && !fileName.EndsWith(".docx"))
            {
                return false;
            }
            return true;
        }

        // 用设定的目录当做子word的目录，默认为选择文件的目录
        public string Rename(string filename, int index)
        {
            var filePathIndex = filename.LastIndexOf('\\');
            string name = "";
            string tempPath = "";
            if (-1 != filePathIndex)
            {
                name = filename.Substring(filePathIndex);
                if (name.Length > 0)
                {
                    name = name.Substring(1);
                }
                tempPath = filename.Substring(0, filePathIndex);
            }

            if ("" == path)
            {
                path = tempPath;
            }

            string newFilename = name;
            newFilename = path + "\\" + name;
            int lastIndex = newFilename.LastIndexOf('.');
            string newName = "";
            if (-1 != lastIndex)
            {
                newName = string.Format("{0}{1}{2}", newFilename.Substring(0, lastIndex), index, newFilename.Substring(lastIndex));
            }
            else
            {
                newName = string.Format("{0}{1}", newFilename, index);
            }
            return newName;
        }

        public string dot(Word.Document doc, string filename)
        {
            Word.Paragraphs paragraphs = doc.Paragraphs;
            Word.Document newDoc = app.Documents.Add();
            bool insert = false;
            bool bigMark = false;
            bool firstBig = false;
            int fileIndex = 1;
            try
            {
                for (int i = 1; i < paragraphs.Count; i++)  
                {
                    string rangeText = paragraphs[i].Range.Text.ToString();
                    if (rangeText.Trim().Length < 1)
                    {
                        continue;
                    }
                    string listNum = paragraphs[i].Range.ListFormat.ListString;
                    Logger.D("dot get index:{0}, message: {1}", i, rangeText);
                    if (checkTextStartWithBig(rangeText))
                    {
                        Logger.D("dot get index: {0}, firstBig: {1}", i, firstBig);
                        if (firstBig)
                        {
                            object file = Rename(filename, fileIndex++);
                            newDoc.SaveAs2(file);
                            newDoc.Close();
                            newDoc = app.Documents.Add();
                        }
                        paragraphs[i].Range.Select();
                        app.Selection.Copy();
                        app.Documents[1].Activate();
                        app.Selection.Paste();
                        insert = false;
                        bigMark = true;
                        continue;
                    }
                    if (checkTextStartWithNum(listNum) || checkTextStartWithNum(rangeText))
                    {
                        Logger.D("dot get index: {0}, insert: {1}, bigMark: {2}, condition: {3}, condition2:{4}", i, insert, bigMark, (!bigMark && insert), (!bigMark) && insert);
                        if (!bigMark && insert)
                        {
                            object file = Rename(filename, fileIndex++);
                            newDoc.SaveAs2(file);
                            newDoc.Close();
                            newDoc = app.Documents.Add();
                        }
                        paragraphs[i].Range.Select();
                        app.Selection.Copy();
                        app.Documents[1].Activate();
                        app.Selection.Paste();
                        insert = true;
                        bigMark = false;
                        firstBig = true;
                        continue;
                    }
                    paragraphs[i].Range.Select();
                    app.Selection.Copy();
                    app.Documents[1].Activate();
                    app.Selection.Paste();
                    insert = true;
                }
                if (insert)
                {
                    object file = Rename(filename, fileIndex++);
                    newDoc.SaveAs2(file);
                }
                newDoc.Close();
                insert = false;
            } catch(System.Exception e)
            {
                Logger.E("dot with filename({0}) error ->({1})", filename, e.Message.ToString());
                return e.Message.ToString();
            }
            finally
            {
                if(doc!= null)
                {
                    doc.Undo();
                    doc.Close();
                }
            }
            Logger.D("dot with filename({0}) success", filename);
            return "";
        }

        /// <summary>
        ///     判断内容是否是以大写的题目号开头，如一、或二、
        /// </summary>
        /// <param name="text"></param>
        /// <returns></returns>
        public bool checkTextStartWithBig(string text)
        {
            string textEscapeSpace = Regex.Replace(text, @"\s", "");
            textEscapeSpace = textEscapeSpace.Trim();
            for (int i = 0; i < bigQuestionNum.Length; i++)
            {
                if (textEscapeSpace.StartsWith(bigQuestionNum[i]))
                {
                    return true;
                }
            }
            return false;
        }
        
        /// <summary>
        ///     判断内容是否有题目号开头，如1.或2.或3.
        /// </summary>
        /// <param name="text"></param>
        /// <returns></returns>
        public bool checkTextStartWithNum(string text)
        {
            if (text.Length < 1)
            {
                return false;
            }
            string textEscapeSpace = Regex.Replace(text, @"\s", "");
            textEscapeSpace = textEscapeSpace.Trim();
            foreach(var arg in smallQuestionNum)
            {
                string num = arg.ToString();
                if (textEscapeSpace.StartsWith(num))
                {
                    return true;
                }
            }
            foreach(var arg in smallQuestionNum2)
            {
                string num = arg.ToString();
                if (textEscapeSpace.StartsWith(num))
                {
                    return true;
                }
            }
            return false;
        }
    }
}