Section:

PDF Data Extraction - Images, Text, Paths - C++ Sample Code

Sample code for using Apryse SDK to extract text, paths, and images from a PDF. The sample also shows how to do color conversion, image normalization, and process changes in the graphics state. Sample code provided in Python, C++, C#, Java, Node.js (JavaScript), PHP, Ruby and VB.

Learn more about our full PDF Data Extraction SDK Capabilities.

To start your free trial, get stated with Server SDK.

1//---------------------------------------------------------------------------------------
2// Copyright (c) 2001-2024 by Apryse Software Inc. All Rights Reserved.
3// Consult legal.txt regarding legal and license information.
4//---------------------------------------------------------------------------------------
5
6#include <PDF/PDFNet.h>
7#include <PDF/PDFDoc.h>
8#include <PDF/ElementReader.h>
9#include <PDF/Element.h>
10#include <PDF/Font.h>
11#include <Filters/FilterReader.h>
12#include <PDF/Image/Image2RGB.h>
13
14#include <iostream>
15#include <assert.h>
16#include "../../LicenseKey/CPP/LicenseKey.h"
17
18using namespace std;
19
20using namespace pdftron;
21using namespace PDF;
22using namespace SDF;
23using namespace Common;
24using namespace Filters; 
25
26char m_buf[4000];
27
28void ProcessElements(ElementReader& reader);
29
30void ProcessPath(ElementReader& reader, Element path)
31{
32	if (path.IsClippingPath())
33	{
34		cout << "This is a clipping path" << endl;
35	}
36
37	PathData d = path.GetPathData();
38
39	const UChar* opr = &d.GetOperators().front();
40	const UChar *opr_itr = opr, *opr_end = opr + d.GetOperators().size();
41	const double* data = &d.GetPoints().front();
42	const double *data_itr = data, *data_end = data + d.GetPoints().size();
43
44	double x1, y1, x2, y2, x3, y3;
45
46	// Use path.GetCTM() if you are interested in CTM (current transformation matrix).
47
48	cout << " Path Data Points := \"";
49	for (; opr_itr<opr_end; ++opr_itr)
50	{
51		switch(*opr_itr)
52		{
53		case PathData::e_moveto:
54			x1 = *data_itr; ++data_itr;
55			y1 = *data_itr; ++data_itr;
56			sprintf(m_buf, "M%.0f %.0f", x1, y1);
57			cout << m_buf;
58			break;
59		case PathData::e_lineto:
60			x1 = *data_itr; ++data_itr;
61			y1 = *data_itr; ++data_itr;
62			sprintf(m_buf, " L%.0f %.0f", x1, y1);
63			cout << m_buf;
64			break;
65		case PathData::e_cubicto:
66			x1 = *data_itr; ++data_itr;
67			y1 = *data_itr; ++data_itr;
68			x2 = *data_itr; ++data_itr;
69			y2 = *data_itr; ++data_itr;
70			x3 = *data_itr; ++data_itr;
71			y3 = *data_itr; ++data_itr;
72			sprintf(m_buf, " C%.0f %.0f %.0f %.0f %.0f %.0f", x1, y1, x2, y2, x3, y3);
73			cout << m_buf;
74			break;
75		case PathData::e_rect:
76			{
77				x1 = *data_itr; ++data_itr;
78				y1 = *data_itr; ++data_itr;
79				double w = *data_itr; ++data_itr;
80				double h = *data_itr; ++data_itr;
81				x2 = x1 + w;
82				y2 = y1;
83				x3 = x2;
84				y3 = y1 + h;
85				double x4 = x1; 
86				double y4 = y3;
87				sprintf(m_buf, "M%.0f %.0f L%.0f %.0f L%.0f %.0f L%.0f %.0f Z", 
88					x1, y1, x2, y2, x3, y3, x4, y4);
89				cout << m_buf;
90			}
91			break;
92		case PathData::e_closepath:
93			cout << " Close Path" << endl;
94			break;
95		default: 
96			assert(false);
97			break;
98		}	
99	}
100
101	cout << "\" ";
102
103	GState gs = path.GetGState();
104
105	// Set Path State 0 (stroke, fill, fill-rule) -----------------------------------
106	if (path.IsStroked()) 
107	{
108		cout << "Stroke path" << endl; 
109
110		if (gs.GetStrokeColorSpace().GetType() == ColorSpace::e_pattern)
111		{
112			cout << "Path has associated pattern" << endl; 
113		}
114		else
115		{
116			// Get stroke color (you can use PDFNet color conversion facilities)
117			// ColorPt rgb;
118			// gs.GetStrokeColorSpace().Convert2RGB(gs.GetStrokeColor(), rgb);
119		}
120	}
121	else 
122	{
123		// Do not stroke path
124	}
125
126	if (path.IsFilled())
127	{
128		cout << "Fill path" << endl; 
129
130		if (gs.GetFillColorSpace().GetType() == ColorSpace::e_pattern)
131		{		
132			cout << "Path has associated pattern" << endl; 
133		}
134		else
135		{
136			// ColorPt rgb;
137			// gs.GetFillColorSpace().Convert2RGB(gs.GetFillColor(), rgb);
138		}        
139	}
140	else 
141	{
142		// Do not fill path
143	}
144
145	// Process any changes in graphics state  ---------------------------------
146
147	GSChangesIterator gs_itr = reader.GetChangesIterator();
148	for (; gs_itr.HasNext(); gs_itr.Next()) 
149	{
150		switch(gs_itr.Current())
151		{
152		case GState::e_transform :
153			// Get transform matrix for this element. Unlike path.GetCTM() 
154			// that return full transformation matrix gs.GetTransform() return 
155			// only the transformation matrix that was installed for this element.
156			//
157			// gs.GetTransform();
158			break;
159		case GState::e_line_width :
160			// gs.GetLineWidth();
161			break;
162		case GState::e_line_cap :
163			// gs.GetLineCap();
164			break;
165		case GState::e_line_join :
166			// gs.GetLineJoin();
167			break;
168		case GState::e_flatness :	
169			break;
170		case GState::e_miter_limit :
171			// gs.GetMiterLimit();
172			break;
173		case GState::e_dash_pattern :
174			{
175				// std::vector<double> dashes;
176				// gs.GetDashes(dashes);
177				// gs.GetPhase()
178			}
179			break;
180		case GState::e_fill_color:
181			{
182				if ( gs.GetFillColorSpace().GetType() == ColorSpace::e_pattern &&
183					gs.GetFillPattern().GetType() != PatternColor::e_shading )
184				{	
185					//process the pattern data
186					reader.PatternBegin(true);
187					ProcessElements(reader);
188					reader.End();
189				}
190			}
191			break;
192		}
193	}
194	reader.ClearChangeList();
195}
196
197void ProcessText(ElementReader& page_reader) 
198{
199	// Begin text element
200	cout << "Begin Text Block:" << endl;
201
202	Element element; 
203	while ((element = page_reader.Next()) != 0) 
204	{
205		switch (element.GetType())
206		{
207		case Element::e_text_end: 
208			// Finish the text block
209			cout << "End Text Block." << endl;
210			return;
211
212		case Element::e_text:
213			{
214				GState gs =  element.GetGState();
215
216				ColorSpace cs_fill = gs.GetFillColorSpace();
217				ColorPt fill = gs.GetFillColor();
218
219				ColorPt out;
220				cs_fill.Convert2RGB(fill, out);
221
222
223				ColorSpace cs_stroke = gs.GetStrokeColorSpace();
224				ColorPt stroke = gs.GetStrokeColor();
225
226				Font font = gs.GetFont();
227
228				cout << "Font Name: " << font.GetName() << endl;
229				// font.IsFixedWidth();
230				// font.IsSerif();
231				// font.IsSymbolic();
232				// font.IsItalic();
233				// ... 
234
235				// double font_size = gs.GetFontSize();
236				// double word_spacing = gs.GetWordSpacing();
237				// double char_spacing = gs.GetCharSpacing();
238				// const UString* txt = element.GetTextString();
239
240				if ( font.GetType() == Font::e_Type3 )
241				{
242					//type 3 font, process its data
243					for (CharIterator itr = element.GetCharIterator(); itr.HasNext(); itr.Next()) 
244					{
245						page_reader.Type3FontBegin(itr.Current());
246						ProcessElements(page_reader);
247						page_reader.End();
248					}
249				}
250
251				else
252				{	
253					Matrix2D text_mtx = element.GetTextMatrix();
254					double x, y;
255					unsigned int char_code;
256
257					for (CharIterator itr = element.GetCharIterator(); itr.HasNext(); itr.Next()) 
258					{
259						cout << "Character code: ";
260						char_code = itr.Current().char_code;
261						if (char_code>=32 || char_code<=127)
262						{ 
263							// Print if in ASCII range...
264							cout << char(char_code);
265						}
266
267						x = itr.Current().x;		// character positioning information
268						y = itr.Current().y;
269
270						// Use element.GetCTM() if you are interested in the CTM 
271						// (current transformation matrix).
272						Matrix2D ctm = element.GetCTM();
273
274						// To get the exact character positioning information you need to 
275						// concatenate current text matrix with CTM and then multiply 
276						// relative positioning coordinates with the resulting matrix.
277						Matrix2D mtx = ctm * text_mtx;
278						mtx.Mult(x, y);
279
280						// Get glyph path...
281						//vector<UChar> oprs;
282						//vector<double> glyph_data;
283						//font.GetGlyphPath(char_code, oprs, glyph_data, false, 0);
284					}
285				}
286
287				cout << endl;
288			}
289			break;
290		}
291	}
292}
293
294void ProcessImage(Element image)  
295{
296	bool image_mask = image.IsImageMask();
297	bool interpolate = image.IsImageInterpolate();
298	int width = image.GetImageWidth();
299	int height = image.GetImageHeight();
300	int out_data_sz = width * height * 3;
301
302	cout << "Image:" 
303		<< " width=\"" << width << "\""
304		<< " height=\"" << height << "\"" << endl;
305
306	// Matrix2D& mtx = image->GetCTM(); // image matrix (page positioning info)
307
308	// You can use GetImageData to read the raw (decoded) image data
309	//image->GetBitsPerComponent();	
310	//image->GetImageData();	// get raw image data
311	// .... or use Image2RGB filter that converts every image to RGB format,
312	// This should save you time since you don't need to deal with color conversions, 
313	// image up-sampling, decoding etc.
314
315	Image2RGB img_conv(image);	// Extract and convert image to RGB 8-bpc format
316	FilterReader reader(img_conv);
317
318	// A buffer used to keep image data.
319	std::vector<UChar> image_data_out; 
320	image_data_out.resize(out_data_sz);
321
322	reader.Read(&image_data_out.front(), out_data_sz);
323	// &image_data_out.front() contains RGB image data.
324
325	// Note that you don't need to read a whole image at a time. Alternatively
326	// you can read a chuck at a time by repeatedly calling reader.Read(buf, buf_sz) 
327	// until the function returns 0. 
328}
329
330void ProcessElements(ElementReader& reader) 
331{
332	Element element;
333	while ((element = reader.Next()) != 0) 	// Read page contents
334	{
335		switch (element.GetType())
336		{
337		case Element::e_path:						// Process path data...
338			{
339				ProcessPath(reader, element);
340			}
341			break; 
342		case Element::e_text_begin: 				// Process text block...
343			{
344				ProcessText(reader);
345			}
346			break;
347		case Element::e_form:						// Process form XObjects
348			{
349				reader.FormBegin(); 
350				ProcessElements(reader);
351				reader.End();
352			}
353			break; 
354		case Element::e_image:						// Process Images
355			{
356				ProcessImage(element);
357			}	
358			break; 
359		}
360	}
361}
362
363int main(int argc, char *argv[])
364{
365	int ret = 0;
366	PDFNet::Initialize(LicenseKey);
367
368	// Relative path to the folder containing test files.
369	string input_path =  "../../TestFiles/";
370	// string output_path = "../../TestFiles/Output/";
371
372
373	try	// Extract text data from all pages in the document
374	{
375		cout << "-------------------------------------------------" << endl;
376		cout << "Extract page element information from all " << endl;
377		cout << "pages in the document." << endl;
378
379		PDFDoc doc((input_path + "newsletter.pdf").c_str());
380		doc.InitSecurityHandler();
381
382		int pgnum = doc.GetPageCount();
383		PageIterator page_begin = doc.GetPageIterator();
384
385		ElementReader page_reader;
386
387		PageIterator itr;
388		for (itr = page_begin; itr.HasNext(); itr.Next())		//  Read every page
389		{				
390			cout << "Page " << itr.Current().GetIndex() << "----------------------------------------" << endl;
391			page_reader.Begin(itr.Current());
392			ProcessElements(page_reader);
393			page_reader.End();
394		}
395
396		cout << "Done." << endl;
397	}
398	catch(Exception& e)
399	{
400		cout << e << endl;
401		ret = 1;
402	}
403	catch(...)
404	{
405		cout << "Unknown Exception" << endl;
406		ret = 1;
407	}
408
409	PDFNet::Terminate();
410	return ret;
411}

Did you find this helpful?

Trial setup questions?

Ask experts on Discord

Need other help?

Contact Support

Pricing or product questions?

Contact Sales