private void menuSQLTest_Click(object sender, EventArgs e)
{
SQLUtils sqlUtils = SQLUtils.getInstance();
conventionRead = null;
try
{
sqlUtils.makeConnect();
if (sqlUtils.isConnected)
{
MessageBox.Show("连接成功");
this.toolStripStatusLabel1.Text = "连接成功";
}
}
catch (Exception err)
{
MessageBox.Show("连接失败");
this.toolStripStatusLabel1.Text = "连接失败" + err.Message;
}
}
// /Uploads/imagesrc/neihechuanbo/num5/2006
private void btnWordRead_Click(object sender, EventArgs e)
{
if (this.tbHtmlPath.Text != string.Empty && this.tbParentGuid.Text != string.Empty &&
this.tbParentDepth.Text != string.Empty && this.tbParentIDfolder.Text != string.Empty &&
this.tbParentTitleCnFolder.Text != string.Empty && this.tbFilesPath.Text != string.Empty)
{
try
{
conventionRead = new ConventionRead();
conventionRead.imageFilePath = tbFilesPath.Text;
ConventionRow rootNode = new ConventionRow(new Guid(this.tbParentGuid.Text), int.Parse(this.tbParentDepth.Text),
this.tbParentIDfolder.Text, this.tbParentTitleCnFolder.Text);
conventionRead.htmlPath = tbHtmlPath.Text;
if (rdbtTitle1Bold.Checked)
{
if (this.tbTitle1Xpath.Text != string.Empty && tbTitle2Xpath.Text.Trim() != String.Empty)
{
conventionRead.title1_select = tbTitle1Xpath.Text;
conventionRead.title2_select = tbTitle2Xpath.Text;
conventionRead.method = ReadMethod.TITLE_CLASS;
}
}
else if (rdbtTitleHTag.Checked)
{
if (tbTitle1TagName.Text.Trim() != string.Empty && tbTitle2TagName.Text.Trim() != string.Empty)
{
conventionRead.title1_select = tbTitle1TagName.Text;
conventionRead.title2_select = tbTitle2TagName.Text;
conventionRead.method = ReadMethod.TITLE_TAG;
}
}
else
{
if (tbTitle1SpanStyle.Text.Trim() != String.Empty && tbTitle2SpanStyle.Text.Trim() != String.Empty)
{
conventionRead.title1_select = tbTitle1SpanStyle.Text;
conventionRead.title2_select = tbTitle2SpanStyle.Text;
conventionRead.method = ReadMethod.TITLE_SPANSTYLE;
}
}
info = conventionRead.ReadHtml(rootNode);
this.toolStripStatusLabel1.Text = "Html识别成功：一级目录有" + info.title1s.Count +
"个,二级目录共有" + info.title2s.Count + "个" + "一级标题直接内容有" + info.title1ContentsNum + "个,图片识别结果 " + info.picResult;
this.tbTitle1Guids.Text = "";
for (int i = 0; i < info.title1Guids.Count; i++)
{
this.tbTitle1Guids.Text += info.title1s[i] + "\r\n";// " : "+info.title1Guids[i]+"\r\n";
}
this.tbTitle1Guids.Text += "\r\n\r\n";
for (int i = 0; i < info.titles.Count; i++)
{
this.tbTitle1Guids.Text += info.titles[i] + "\r\n";
}
}
catch (Exception err)
{
MessageBox.Show(err.Message);
this.toolStripStatusLabel1.Text = err.Message;
}
}
else
{
MessageBox.Show("请输入信息!");
}
#region 废弃代码
// ConventionRow rootNode=new ConventionRow(new Guid("1b506d0f-8956-46d3-a023-78d24e300ed0"),2,ConventionOptions.CATEGORY.IS_CATEGORY);
// conventionRead.ReadCatalogue(rootNode);
// Word.Application app = new Word.Application();
// Word.Document doc = null;
// object unknow = Type.Missing;
// app.Visible = false;
// string str = @"D:\work\WordRead\test.docx";
// object file = str;
// doc = app.Documents.Open(ref file,
// ref unknow, ref unknow, ref unknow, ref unknow,
// ref unknow, ref unknow, ref unknow, ref unknow,
// ref unknow, ref unknow, ref unknow, ref unknow,
// ref unknow, ref unknow, ref unknow);
// string temp;
// //int paraCount = doc.Paragraphs.Count;
// //for (int i = 1; i < paraCount + 1; i++)
// //{
// // temp = doc.Paragraphs[i].Range.Text.Trim();
// // Console.WriteLine(temp);
// //}
// doc.ActiveWindow.Selection.WholeStory();
// doc.ActiveWindow.Selection.Copy();
// IDataObject data = Clipboard.GetDataObject();
// temp = data.GetData(DataFormats.Text).ToString();
// //回车换行使用了\r\n 和 \n
// string pattern_title1 = @"第\d{1,}章 {1,2}[\w ]+\r";//查找一级标题
// string pattern_title2 = @"(?<=\r\n|\r\n\s{1,})\d{1,}\s{1,}\w[^，。]+?\r";//查找二级标题
// string pattern_zhengwen = @"(?<=(?<=\r\n|\r\n\s{1,})\d{1,}\s{1,}\w[^，。]+?\r)" +
// @"[\s\S]+?(?=((?<=\r\n|\r\n\s{1,})\d{1,}\s{1,}\w[^，。]+?\r)|" +
// @"第\d{1,}章 {1,2}[\w ]+\r\n|(?<=\n|\n\s+)附录[\w\W]+?(?=\r\n)|$)";//查找正文
// string pattern_fulu = @"(?<=\n|\n\s+)附录[\w\W]+?(?=\r\n)"; //查找附录
// /*******缩进*********/
// string pattern_suojin1 = @"(?<=\n|^)[ \t\s]*(?=\d+[\.．]\d+)";//匹配1.1、1.1.1缩进，替换顶行无缩进
// string pattern_suojin2 = @"(?<=\n)[ \t\s]*(?=[（(]\d+[）)])";//查找正文中的（1）替换为2空格
// string pattern_suojin3 = @"(?<=\n)[ \t\s]*(?=[①②③④⑤⑥⑦⑧⑨⑩])";//匹配 ①缩进，替换为6空格
// string pattern_suojin4 = @"(?<=\n)[ \t\s]*(?=[（(][a-z]+[）)])";//匹配（a）缩进，替换为8空格
// Regex.Replace(temp, pattern_suojin1, "");
// Regex.Replace(temp, pattern_suojin2, " ");
// Regex.Replace(temp, pattern_suojin3, " ");
// Regex.Replace(temp, pattern_suojin4, " ");
// /****************/
// //FileStream mytxt = new FileStream(@"D:\work\WordRead\testResult.txt",
// // FileMode.Open, FileAccess.Read, FileShare.ReadWrite);
// MatchCollection mymatches = Regex.Matches(temp, pattern_zhengwen);
// foreach (Match match in mymatches)
// {
// File.AppendAllText(@"D:\work\WordRead\testResult.txt", match.Value);
// File.AppendAllText(@"D:\work\WordRead\testResult.txt", "\r\n\r\n\r\n\r\n完成一段正文。\r\n\r\n\r\n\r\n");
// }
// Console.WriteLine(mymatches.Count);
// Console.WriteLine("\nFinished");
// doc.Close();
#endregion
}
