TextSearch

1//
2// Copyright (c) 2001-2021 by PDFTron Systems Inc. All Rights Reserved.
3//
4
5using System;
6using pdftron;
7using pdftron.Common;
8using pdftron.Filters;
9using pdftron.SDF;
10using pdftron.PDF;
11
12
13using NUnit.Framework;
14
15namespace MiscellaneousSamples
16{
17	// This sample illustrates various text search capabilities of PDFNet.
18
19	[TestFixture]
20	public class TextSearchTest
21	{		
22		
23		[Test]
24		public static void Sample()
25		{
26
27			// Relative path to the folder containing test files.
28			const string input_path =  "TestFiles/";
29
30			// Sample code showing how to use high-level text extraction APIs.
31			try	
32			{
33				using (PDFDoc doc = new PDFDoc(Utils.GetAssetTempFile(input_path + "credit card numbers.pdf")))
34				{
35					doc.InitSecurityHandler();
36
37					Int32 page_num = 0;
38					String result_str = "", ambient_string = "";
39					Highlights hlts = new Highlights();
40
41					TextSearch txt_search = new TextSearch();
42					Int32 mode = (Int32)(TextSearch.SearchMode.e_whole_word | TextSearch.SearchMode.e_page_stop | TextSearch.SearchMode.e_highlight);
43					String pattern = "joHn sMiTh";
44
45					//call Begin() method to initialize the text search.
46					txt_search.Begin( doc, pattern, mode, -1, -1 );
47
48					int step = 0;
49			
50					//call Run() method iteratively to find all matching instances.
51					while ( true )
52					{
53						TextSearch.ResultCode code = txt_search.Run(ref page_num, ref result_str, ref ambient_string, hlts );
54
55						if ( code == TextSearch.ResultCode.e_found )
56						{
57							if ( step == 0 )
58							{	//step 0: found "John Smith"
59								//note that, here, 'ambient_string' and 'hlts' are not written to, 
60								//as 'e_ambient_string' and 'e_highlight' are not set.
61								Console.WriteLine(result_str + "'s credit card number is: ");
62
63								//now switch to using regular expressions to find John's credit card number
64								mode = txt_search.GetMode();
65								mode |= (Int32)(TextSearch.SearchMode.e_reg_expression | TextSearch.SearchMode.e_highlight);
66								txt_search.SetMode(mode);
67								pattern = "\\d{4}-\\d{4}-\\d{4}-\\d{4}"; //or "(\\d{4}-){3}\\d{4}"
68								txt_search.SetPattern(pattern);
69
70								++step;
71							}
72							else if ( step == 1 )
73							{
74								//step 1: found John's credit card number
75								//result_str.ConvertToAscii(char_buf, 32, true);
76								//cout << "  " << char_buf << endl;
77								Console.WriteLine("  " + result_str);
78
79								//note that, here, 'hlts' is written to, as 'e_highlight' has been set.
80								//output the highlight info of the credit card number
81								hlts.Begin(doc);
82								while (hlts.HasNext())
83								{
84									Console.WriteLine("The current highlight is from page: " + hlts.GetCurrentPageNumber());
85									hlts.Next();
86								}
87
88								//see if there is an AMEX card number
89								pattern = "\\d{4}-\\d{6}-\\d{5}";
90								txt_search.SetPattern(pattern);
91
92								++step;
93							}
94							else if ( step == 2 )
95							{
96								//found an AMEX card number
97								Console.WriteLine("\nThere is an AMEX card number:\n  " + result_str);
98
99								//change mode to find the owner of the credit card; supposedly, the owner's
100								//name proceeds the number
101								mode = txt_search.GetMode();
102								mode |= (Int32)(TextSearch.SearchMode.e_search_up);
103								txt_search.SetMode(mode);
104								pattern = "[A-z]++ [A-z]++";
105								txt_search.SetPattern(pattern);
106
107								++step;
108							}
109							else if ( step == 3 )
110							{
111								//found the owner's name of the AMEX card
112								Console.WriteLine("Is the owner's name:\n  " + result_str + "?");
113
114								//add a link annotation based on the location of the found instance
115								hlts.Begin(doc);
116								while (hlts.HasNext())
117								{
118									Page cur_page = doc.GetPage(hlts.GetCurrentPageNumber());
119									double[] quads = hlts.GetCurrentQuads();
120									int quad_count = quads.Length / 8;
121									for (int i = 0; i < quad_count; ++i)
122									{
123										//assume each quad is an axis-aligned rectangle
124										int offset = 8 * i;
125										double x1 = Math.Min(Math.Min(Math.Min(quads[offset + 0], quads[offset + 2]), quads[offset + 4]), quads[offset + 6]);
126										double x2 = Math.Max(Math.Max(Math.Max(quads[offset + 0], quads[offset + 2]), quads[offset + 4]), quads[offset + 6]);
127										double y1 = Math.Min(Math.Min(Math.Min(quads[offset + 1], quads[offset + 3]), quads[offset + 5]), quads[offset + 7]);
128										double y2 = Math.Max(Math.Max(Math.Max(quads[offset + 1], quads[offset + 3]), quads[offset + 5]), quads[offset + 7]);
129
130										pdftron.PDF.Annots.Link hyper_link = pdftron.PDF.Annots.Link.Create(doc, new Rect(x1, y1, x2, y2), pdftron.PDF.Action.CreateURI(doc, "http://www.pdftron.com"));
131										hyper_link.RefreshAppearance();
132										cur_page.AnnotPushBack(hyper_link);
133									}
134									hlts.Next();
135								}
136								doc.Save(Utils.CreateExternalFile("credit card numbers_linked.pdf"), SDFDoc.SaveOptions.e_linearized);
137
138								break;
139							}
140						}
141						else if ( code == TextSearch.ResultCode.e_page )
142						{
143							//you can update your UI here, if needed
144						}
145						else
146						{
147							break;
148						}
149					}
150				}
151			}
152
153			catch (PDFNetException e)
154			{
155				Console.WriteLine(e.Message);
156				Assert.True(false);
157			}
158		}
159	}
160}