Sleds/pdf-bookmarker/PDFBookmarker/PDFBookmarker.cpp

#include "PDFBookmarker.h"

using namespace sequencelogic;

PDFBookmarker::PDFBookmarker()
{
}


PDFBookmarker::~PDFBookmarker()
{
}

void PDFBookmarker::bookmarkPDFs(std::string classificationJSON, std::string inFile, const std::string &outFilename)
{
	std::string outFile;
   if (outFilename  == "")
      outFile = getOutputFile(inFile);
   else
      outFile = outFilename;
	std::string pdfMarkFile = getMarkFileName(inFile);
	JSONObject classification;
	JSONArray docs;
	std::stringstream pdfMarkTxt;
	std::stringstream gsCmd;

	classification.LoadFromFile(classificationJSON);

	docs = classification.getJSONArray(DOCS_TAG);
	std::cout << "Found " << docs.getnumitems() << " docTypes in file" << std::endl;
	//get all the pages for each doctype
	for (int i =0; i < docs.getnumitems(); i++)
	{
		
		std::string docType;
		JSONObject &currentDoc = static_cast<JSONObject&>(docs[i]);
		std::vector<int> pageNums;
		docType = currentDoc.getJSONString(DOCTYPE_TAG.c_str());

		JSONArray pages = currentDoc.getJSONArray(PAGES_ARRAY_TAG.c_str());
		for (int j = 0; j < pages.getnumitems(); j++)
		{
			JSONObject &currentPage = static_cast<JSONObject&>(pages[j]);
			std::string pageFile;

			pageFile = currentPage.getJSONString(PAGE_FILE_TAG.c_str());
			pageNums.push_back(getPageNum(pageFile));
			
		}
		int pageNum = getSmallest(pageNums);

		std::cout << "   Found " << docType << " on page " << pageNum << std::endl;
		pdfMarkTxt << getBookmarkTxt(docType, pageNum);
	}
	std::string fileTxt = pdfMarkTxt.str();
	saveFile(pdfMarkFile, fileTxt);

	gsCmd << GHOSTSCRIPT_COMMAND << "\"" << inFile << "\" update_info \"" << pdfMarkFile << "\" output \"" << outFile << "\"";

	std::cout << "Bookmarking file " << inFile << "..." << std::endl;

   // It seems that 'pdftk' translates the '$' character. The string "$-1" in a filename becomes "hBc1"...
   std::string cmdStr = gsCmd.str();
   size_t pos = cmdStr.find('$');
   while ((pos = cmdStr.find('$', pos)) != std::string::npos)
   {
      cmdStr = cmdStr.replace(pos,1, "\\$");
      ++pos;
   }
   runCommand(cmdStr);
	std::cout << "   Saved." << std::endl;
	
	deleteFile(pdfMarkFile);
}

std::string PDFBookmarker::getOutputFile(std::string inFile)
{
	//ALWAYS a pdf file.
	std::stringstream result;

	result << inFile.substr(0, inFile.rfind('.')) << "_marked.pdf";

	return result.str();
}

int PDFBookmarker::getSmallest(std::vector<int> pageNums)
{
	int smallest = pageNums[0];
	for (size_t i = 1; i < pageNums.size(); i++)
	{
		if (pageNums[i] < smallest)
		smallest = pageNums[i];
	}
	return smallest;
}

std::string PDFBookmarker::getMarkFileName(std::string inFile)	
{
	std::stringstream result;
	result << inFile.substr(0, inFile.rfind('/') + 1) << PDFMARK_FILE;
	return result.str();
}

void PDFBookmarker::saveFile(std::string fileName, std::string &fileText)
{
	std::cout << "Saving file " << fileName <<std::endl;
	std::ofstream out;

	try
	{
		out.open(fileName, std::ofstream::out);
		out << fileText;
	}
	catch (std::exception ex)
	{
		std::cout << "Error opening/writing to file " << fileName << std::endl
					<< "   " <<ex.what() << std::endl << std::endl;
	}
		
	out.close();
}

void PDFBookmarker::deleteFile(std::string fileName)
{
	std::cout << "Deleting File " << fileName << "..." << std::endl;
	std::remove(fileName.c_str());
	std::cout << "   Done." << std::endl;
}

std::string PDFBookmarker::runCommand(std::string cmd)
{
   std::cout << "Running cmd:\n" << cmd << std::endl;
	std::string result;
	FILE *stream;
	const int MAX_BUFFER = 256;
	char buffer[MAX_BUFFER];
	stream = popen(cmd.c_str(), "r");

	if (stream)
	{
		while (!feof(stream))
		{
			if (fgets(buffer, MAX_BUFFER, stream) != NULL)
				result.append(buffer);
		}
		pclose(stream);
	}
	return result;
}

int PDFBookmarker::getPageNum(std::string imgFile)
{
	int result;

	std::string tmpRes = imgFile.substr(imgFile.find("page") + 4, imgFile.find('.'));
	result = atoi(tmpRes.c_str());

	return result;
}

std::string PDFBookmarker::getBookmarkTxt(std::string docType, int startPageNum)
{
	std::stringstream result;

	result << "BookmarkBegin\n" 
		   << "BookmarkTitle: " <<docType << "\n"
		   << "BookmarkLevel: 1\n" 
		   << "BookmarkPageNumber: " <<startPageNum <<"\n";

	return result.str();
}