UnicodeWrite

1//
2// Copyright (c) 2001-2021 by PDFTron Systems Inc. All Rights Reserved.
3//
4
5using System;
6using System.IO;
7using System.Text;
8using pdftron;
9using pdftron.Common;
10using pdftron.Filters;
11using pdftron.SDF;
12using pdftron.PDF;
13
14using NUnit.Framework;
15
16namespace MiscellaneousSamples
17{
18	/// <summary>
19	/// This example illustrates how to create Unicode text and how to embed composite fonts.
20	/// </summary>
21	[TestFixture]
22	public class UnicodeWriteTest
23	{
24		
25		// Note: This demo assumes that 'arialuni.ttf' is present in '/Samples/TestFiles' 
26		// directory. Arial Unicode MS is about 24MB in size and it comes together with Windows and 
27		// MS Office.
28		// 
29		// For more information about Arial Unicode MS, please consult the following Microsoft Knowledge 
30		// Base Article: WD2002: General Information About the Arial Unicode MS Font
31		//  http://support.microsoft.com/support/kb/articles/q287/2/47.asp
32		//
33		// For more information consult: 
34		//    http://office.microsoft.com/search/results.aspx?Scope=DC&Query=font&CTT=6&Origin=EC010331121033
35		//    http://www.microsoft.com/downloads/details.aspx?FamilyID=1F0303AE-F055-41DA-A086-A65F22CB5593
36		// 
37		// In case you don't have access to Arial Unicode MS you can use cyberbit.ttf 
38		// (ftp://ftp.netscape.com/pub/communicator/extras/fonts/windows/) instead.
39		//
40		[Test]
41		public static void Sample()
42		{
43
44			// Relative path to the folder containing test files.
45			const string input_path =  "TestFiles/";
46
47			try	
48			{
49				using (PDFDoc doc = new PDFDoc())
50				{
51					using (ElementBuilder eb = new ElementBuilder())
52					{
53						using (ElementWriter writer = new ElementWriter())
54						{
55							// Start a new page ------------------------------------
56							Page page = doc.PageCreate(new Rect(0, 0, 612, 794));
57
58							writer.Begin(page); // begin writing to this page
59
60                            Font fnt = null;
61							try
62							{
63								// Full font embedding
64								fnt = Font.Create(doc, "Arial", "");
65							}
66							catch (PDFNetException e)
67							{
68								Console.WriteLine(e.Message);
69								Assert.True(false);
70							}
71
72							if (fnt == null) {
73								try
74								{
75									fnt = Font.CreateCIDTrueTypeFont(doc, Utils.GetAssetTempFile(input_path + "ARIALUNI.TTF"), true, true);
76								}
77								catch (PDFNetException e)
78								{
79									Console.WriteLine(e.Message);
80									Assert.True(false);
81								}
82							}
83
84							if (fnt == null)
85							{
86								try
87								{
88									fnt = Font.CreateCIDTrueTypeFont(doc, "C:/Windows/Fonts/ARIALUNI.TTF", true, true);
89								}
90								catch (PDFNetException e)
91								{
92									Console.WriteLine(e.Message);
93									Assert.True(false);
94								}
95							}
96
97                            if (fnt == null)
98                            {
99                                Console.WriteLine("Note: using system font substitution for unshaped unicode text");
100                                fnt = Font.Create(doc, "Helvetica", "");
101                            }
102                            else
103                            {
104                                Console.WriteLine("Note: using Arial Unicode for unshaped unicode text");
105                            }
106
107							Element element = eb.CreateTextBegin(fnt, 1);
108							element.SetTextMatrix(10, 0, 0, 10, 50, 600);
109							element.GetGState().SetLeading(2);		 // Set the spacing between lines
110							writer.WriteElement(element);
111
112							// Hello World!!!
113							string hello = "Hello World!";
114							writer.WriteElement(eb.CreateUnicodeTextRun(hello));
115							writer.WriteElement(eb.CreateTextNewLine());
116
117							// Latin
118							char[] latin = {   
119								'a', 'A', 'b', 'B', 'c', 'C', 'd', 'D', '\x45', '\x0046', '\x00C0', 
120								'\x00C1', '\x00C2', '\x0143', '\x0144', '\x0145', '\x0152', '1', '2' // etc.
121							};
122							writer.WriteElement(eb.CreateUnicodeTextRun(new string(latin)));
123							writer.WriteElement(eb.CreateTextNewLine());
124
125							// Greek
126							char[] greek = {   
127								(char)0x039E, (char)0x039F, (char)0x03A0, (char)0x03A1, (char)0x03A3,
128								(char)0x03A6, (char)0x03A8, (char)0x03A9  // etc.
129							};
130							writer.WriteElement(eb.CreateUnicodeTextRun(new string(greek)));
131							writer.WriteElement(eb.CreateTextNewLine());
132
133							// Cyrillic
134							char[] cyrillic = {   
135								(char)0x0409, (char)0x040A, (char)0x040B, (char)0x040C, (char)0x040E, (char)0x040F, (char)0x0410, (char)0x0411,
136								(char)0x0412, (char)0x0413, (char)0x0414, (char)0x0415, (char)0x0416, (char)0x0417, (char)0x0418, (char)0x0419 // etc.
137							};
138							writer.WriteElement(eb.CreateUnicodeTextRun(new string(cyrillic)));
139							writer.WriteElement(eb.CreateTextNewLine());
140
141							// Hebrew
142							char[] hebrew = {
143								(char)0x05D0, (char)0x05D1, (char)0x05D3, (char)0x05D3, (char)0x05D4, (char)0x05D5, (char)0x05D6, (char)0x05D7, (char)0x05D8, 
144								(char)0x05D9, (char)0x05DA, (char)0x05DB, (char)0x05DC, (char)0x05DD, (char)0x05DE, (char)0x05DF, (char)0x05E0, (char)0x05E1 // etc. 
145							};
146							writer.WriteElement(eb.CreateUnicodeTextRun(new string(hebrew)));
147							writer.WriteElement(eb.CreateTextNewLine());
148
149							// Arabic
150							char[] arabic = {
151								(char)0x0624, (char)0x0625, (char)0x0626, (char)0x0627, (char)0x0628, (char)0x0629, (char)0x062A, (char)0x062B, (char)0x062C, 
152								(char)0x062D, (char)0x062E, (char)0x062F, (char)0x0630, (char)0x0631, (char)0x0632, (char)0x0633, (char)0x0634, (char)0x0635 // etc. 
153							};
154							writer.WriteElement(eb.CreateUnicodeTextRun(new string(arabic)));
155							writer.WriteElement(eb.CreateTextNewLine());
156
157							// Thai 
158							char[] thai = {
159								(char)0x0E01, (char)0x0E02, (char)0x0E03, (char)0x0E04, (char)0x0E05, (char)0x0E06, (char)0x0E07, (char)0x0E08, (char)0x0E09, 
160								(char)0x0E0A, (char)0x0E0B, (char)0x0E0C, (char)0x0E0D, (char)0x0E0E, (char)0x0E0F, (char)0x0E10, (char)0x0E11, (char)0x0E12 // etc. 
161							};
162							writer.WriteElement(eb.CreateUnicodeTextRun(new string(thai)));
163							writer.WriteElement(eb.CreateTextNewLine());
164
165							// Hiragana - Japanese 
166							char[] hiragana = {
167								(char)0x3041, (char)0x3042, (char)0x3043, (char)0x3044, (char)0x3045, (char)0x3046, (char)0x3047, (char)0x3048, (char)0x3049, 
168								(char)0x304A, (char)0x304B, (char)0x304C, (char)0x304D, (char)0x304E, (char)0x304F, (char)0x3051, (char)0x3051, (char)0x3052 // etc. 
169							};
170							writer.WriteElement(eb.CreateUnicodeTextRun(new string(hiragana)));
171							writer.WriteElement(eb.CreateTextNewLine());
172
173							// CJK Unified Ideographs
174							char[] cjk_uni = {
175								(char)0x5841, (char)0x5842, (char)0x5843, (char)0x5844, (char)0x5845, (char)0x5846, (char)0x5847, (char)0x5848, (char)0x5849, 
176								(char)0x584A, (char)0x584B, (char)0x584C, (char)0x584D, (char)0x584E, (char)0x584F, (char)0x5850, (char)0x5851, (char)0x5852 // etc. 
177							};
178							writer.WriteElement(eb.CreateUnicodeTextRun(new string(cjk_uni)));
179							writer.WriteElement(eb.CreateTextNewLine());
180
181				            // Simplified Chinese
182				            char[] chinese_simplified = {
183                              (char)0x4e16, (char)0x754c, (char)0x60a8,(char) 0x597d
184				            };
185							writer.WriteElement(eb.CreateUnicodeTextRun(new string(chinese_simplified)));
186							writer.WriteElement(eb.CreateTextNewLine());
187
188							// Finish the block of text
189							writer.WriteElement(eb.CreateTextEnd());		
190                            Console.WriteLine("Now using text shaping logic to place text");
191
192							// Create a font in indexed encoding mode 
193							// normally this would mean that we are required to provide glyph indices
194							// directly to CreateUnicodeTextRun, but instead, we will use the GetShapedText
195							// method to take care of this detail for us.
196							Font indexedFont = Font.CreateCIDTrueTypeFont(doc, Utils.GetAssetTempFile(input_path + "NotoSans_with_hindi.ttf"), true, true, Font.Encoding.e_Indices);
197							element = eb.CreateTextBegin(indexedFont, 10.0);
198							writer.WriteElement(element);
199
200							double linePos = 350.0;
201							double lineSpace = 20.0;
202
203							// Transform unicode text into an abstract collection of glyph indices and positioning info 
204							ShapedText shapedText = indexedFont.GetShapedText("Shaped Hindi Text:");
205
206							// transform the shaped text info into a PDF element and write it to the page
207							element = eb.CreateShapedTextRun(shapedText);
208							element.SetTextMatrix(1.5, 0, 0, 1.5, 50, linePos);
209							linePos -= lineSpace;
210							writer.WriteElement(element);
211
212							// read in unicode text lines from a file File. ReadAllLines(path, Encoding.UTF8)
213							String[] hindiTextLines = File.ReadAllLines(Utils.GetAssetTempFile(input_path + "hindi_sample_utf16le.txt"), Encoding.UTF8);
214
215							Console.WriteLine("Read in " + hindiTextLines.Length + " lines of Unicode text from file");
216							foreach (String textLine in hindiTextLines)
217							{
218								shapedText = indexedFont.GetShapedText(textLine);
219								element = eb.CreateShapedTextRun(shapedText);
220								element.SetTextMatrix(1.5, 0, 0, 1.5, 50, linePos);
221								linePos -= lineSpace;
222								writer.WriteElement(element);
223								Console.WriteLine("Wrote shaped line to page");
224							}
225
226							// Finish the shaped block of text
227							writer.WriteElement(eb.CreateTextEnd());
228
229							writer.End();  // save changes to the current page
230							doc.PagePushBack(page);
231							doc.Save(Utils.CreateExternalFile("unicodewrite.pdf"), SDFDoc.SaveOptions.e_remove_unused | SDFDoc.SaveOptions.e_hex_strings);
232							Console.WriteLine("Done. Result saved in unicodewrite.pdf...");
233						}
234					}
235				}
236			}
237			catch (PDFNetException e)
238			{
239				Console.WriteLine(e.Message);
240				Assert.True(false);
241			}
242		}
243	}
244}