ABBYY FlexiCapture Engine - Read Text From PDF Or Image File

1. 1. using FCEngine;
   2. using System;
   3. using System.Collections.Generic;
   4. using System.ComponentModel;
   5. using System.Data;
   6. using System.Drawing;
   7. using System.IO;
   8. using System.Runtime.InteropServices;
   9. using System.Text;
   10. using System.Windows.Forms;
   11. namespace AflToDocDiff {
   12. public partial class Form1: Form {
   13. IEngine engine = null;
   14. IFlexiCaptureProcessor processor;
   15. IDocumentDefinition definition;
   16. IDocumentDefinition definition1;
   17. StringBuilder sb;
   18. private Dictionary < string, Dictionary < string, FieldParams >> fieldToType = new Dictionary < string, Dictionary < string, FieldParams >> ();
   19. ////private string[] imageFiles;
   20. ////private int currentImageFileIndex = 0;
   21. ////private Image processedImage;
   22. public Form1() {
   23. InitializeComponent();
   24. }
   25. private IEngine loadEngine() {
   26. IEngine engine;
   27. int hresult = InitializeEngine(FceConfig.GetDeveloperSN(), out engine);
   28. Marshal.ThrowExceptionForHR(hresult);
   29. return engine;
   30. }
   31. private void unloadEngine(ref IEngine engine) {
   32. engine = null;
   33. int hResult = DeinitializeEngine();
   34. Marshal.ThrowExceptionForHR(hResult);
   35. }
   36. [DllImport(FceConfig.DllPath, CharSet = CharSet.Unicode), PreserveSig]
   37. internal static extern int InitializeEngine(String devSN, out IEngine engine);
   38. [DllImport(FceConfig.DllPath, CharSet = CharSet.Unicode), PreserveSig]
   39. internal static extern int DeinitializeEngine();
   40. class Gdi32 {
   41. [DllImport("gdi32.dll", EntryPoint = "DeleteObject")]
   42. internal static extern IntPtr DeleteObject(IntPtr hDc);
   43. }
   44. private void CreateDocumentDef() {}
   45. // IDocumentDefinition[] newDocumentDefinitions;
   46. //string[] templateNames;
   47. private void button1_Click(object sender, EventArgs e) {
   48. try {
   49. engine = loadEngine();
   50. //definition1 = engine.CreateDocumentDefinitionFromAFL(txtAflFilePath.Text, "English");
   51. //tuneRecognitionParams("1", definition1);
   52. //((ICustomStorage)definition1).SaveToFile(@"D:\\AflToDocDiff\\AflToDocDiff\\FcdotDoc\\Test1.fcdot");
   53. string extension = Path.GetExtension(txtAflFilePath.Text);
   54. if (processor == null) {
   55. ICustomStorage storage = null;
   56. switch (extension) {
   57. case ".xfd":
   58. definition = engine.CreateDocumentDefinitionFromXFD(txtAflFilePath.Text, "English");
   59. storage = (ICustomStorage) definition;
   60. break;
   61. case ".afl":
   62. definition = engine.CreateDocumentDefinitionFromAFL(txtAflFilePath.Text, "English");
   63. storage = (ICustomStorage) definition;
   64. break;
   65. case ".fcdot":
   66. definition = engine.CreateDocumentDefinition();
   67. storage = (ICustomStorage) definition;
   68. storage.LoadFromFile(txtAflFilePath.Text);
   69. break;
   70. default:
   71. throw new NotImplementedException();
   72. }
   73. processor = engine.CreateFlexiCaptureProcessor();
   74. processor.AddDocumentDefinition(definition);
   75. } else {
   76. processor.ResetProcessing();
   77. processor = null;
   78. }
   79. var imageTools = engine.CreateImageProcessingTools();
   80. var file = imageTools.OpenImageFile(txtFormPath.Text);
   81. int pageCount = file.PagesCount;
   82. for (int i = 0; i < pageCount; i++) {
   83. processor.AddImage(file.OpenImagePage(i));
   84. }
   85. IDocument document = processor.RecognizeNextDocument();
   86. if (document != null && document.DocumentDefinition != null) {
   87. sb = new StringBuilder();
   88. buildDocumentView(document);
   89. // File.WriteAllText(@"C:\Users\GT-PC-15\Desktop\AFL\Result\RF" + DateTime.Now.ToString("yyyyMMddTHHmmss") + ".txt", sb.ToString());
   90. }
   91. string desc = definition.Description;
   92. } catch (Exception ex) {
   93. string msg = ex.Message;
   94. } finally {
   95. unloadEngine(ref engine);
   96. processor = null;
   97. }
   98. }
   99. private void tuneRecognitionParams(string name, IDocumentDefinition definition) {
   100. if (fieldToType.ContainsKey(name)) {
   101. bool modified = false;
   102. var fields = definition.Sections[0].Fields;
   103. for (int i = 0; i < fields.Count; i++) {
   104. if (fieldToType[name].ContainsKey(fields[i].Name)) {
   105. var fieldParams = fieldToType[name][fields[i].Name];
   106. var textParams = fields[i].RecognitionParams.AsTextParams();
   107. if (fieldParams.Type == FieldValueTypeEnum.FVT_Text) {
   108. if (fieldParams.RegExp != null) {
   109. int pos = fieldParams.Params.IndexOf(' ');
   110. string letters = fieldParams.Params.Substring(0, pos);
   111. string regExp = fieldParams.Params.Substring(pos + 1);
   112. var newLanguage = textParams.CreateEmbeddedLanguage(LanguageTypeEnum.LT_Simple, null);
   113. newLanguage.AsSimpleLanguage().set_LetterSet(LanguageLetterSetEnum.LLS_Alphabet, letters);
   114. newLanguage.AsSimpleLanguage().RegularExpression = regExp;
   115. textParams.Language = newLanguage;
   116. }
   117. } else {
   118. var newLanguage = textParams.CreateEmbeddedLanguageByDataType(fieldParams.Type);
   119. textParams.Language = newLanguage;
   120. }
   121. textParams.TextType = fieldParams.TextType;
   122. textParams.CaseRecognitionMode = fieldParams.CaseType;
   123. modified = true;
   124. }
   125. }
   126. if (modified) {
   127. definition.Check();
   128. }
   129. }
   130. }
   131. private void button2_Click(object sender, EventArgs e) {
   132. DialogResult result = openFileDialog1.ShowDialog(); // Show the dialog.
   133. if (result == DialogResult.OK) // Test result.
   134. {
   135. txtAflFilePath.Text = openFileDialog1.FileName;
   136. }
   137. }
   138. private void button3_Click(object sender, EventArgs e) {
   139. DialogResult result = openFileDialog2.ShowDialog(); // Show the dialog.
   140. if (result == DialogResult.OK) // Test result.
   141. {
   142. txtFormPath.Text = openFileDialog2.FileName;
   143. }
   144. }
   145. private void buildDocumentView(IDocument document) {
   146. IField firstSection = document.Sections[0];
   147. addDocumentNodeChildren(firstSection.Children);
   148. }
   149. private void addDocumentNodeChildren(IFields children) {
   150. for (int i = 0; i < children.Count; i++) {
   151. addDocumentNode(children[i]);
   152. }
   153. }
   154. private void addDocumentNode(IField documentNode) {
   155. IFieldValue value = documentNode.Value;
   156. if (value == null) {
   157. // sb.AppendLine(string.Format("{0}:{1}", documentNode.Name, value.AsString));
   158. } else {
   159. string keyval = documentNode.Name + ": " + documentNode.Value;
   160. sb.AppendLine(string.Format("{0}:{1}", documentNode.Name, value.AsString));
   161. string FormName = Convert.ToString(value.AsString);
   162. if (FormName.Contains("125")) {
   163. lblFormName.Text = "Accord Form 125";
   164. string str = FormName.Substring(0, 8);
   165. str = str.Split('(', ')')[1];
   166. } else if (FormName.Contains("126")) {
   167. lblFormName.Text = "Accord Form 126";
   168. } else if (FormName.Contains("140")) {
   169. lblFormName.Text = "Accord Form 140";
   170. } else {
   171. lblFormName.Text = "Please upload valid form.";
   172. }
   173. }
   174. lblFormName.Text = sb.ToString();
   175. if (documentNode.Instances != null) {
   176. addDocumentNodeInstances(documentNode.Instances);
   177. } else if (documentNode.Children != null) {
   178. addDocumentNodeChildren(documentNode.Children);
   179. }
   180. }
   181. private void addDocumentNodeInstances(IFieldInstances instances) {
   182. for (int i = 0; i < instances.Count; i++) {
   183. if (instances[i].Children != null) {
   184. addDocumentNodeChildren(instances[i].Children);
   185. }
   186. }
   187. }
   188. }
   189. }

Output