void fileWriter(MappingRaw* map, const int numRows, const string_t& fileName)
{
std::map<wchar_t, wchar_t> uniqnessCheck;
std::stringstream ss;
for (int i = 0; i < numRows; ++i) {
ss << toUtf8(wstring_t(1, map[i].from))
<< "|" << toUtf8(wstring_t(1, map[i].to));
//<< "|" << map[i].desc;
ss << std::endl;
auto it = uniqnessCheck.insert(std::make_pair(map[i].from, map[i].to));
if ( !it.second ) {
wchar_t ch = map[i].from;
wchar_t to = it.first->second;
std::stringstream ss;
ss << "Character INTCODE(" << (int) ch << "), HEX (0x"
<< Poco::NumberFormatter::formatHex((int)ch, 4) << "), CHAR ("
<< toUtf8(wstring_t(1, ch)) << ")"
<< " already mapped to INTCODE(" << (int)to << "), HEX (0x"
<< Poco::NumberFormatter::formatHex((int)to, 4) << "), CHAR ("
<< toUtf8(wstring_t(1, to)) << ")";
throw std::logic_error(ss.str().c_str());
}
}
writeFileAsBinary(fileName, ss.str());
}
void Converter::convertSingleDocQuick( const string_t& fileName )
{
tDocumentsSp docs = word()->getDocuments();
tDocumentSp doc = docs->open(toUtf16(getInputAbsPath(fileName)));
if (!doc) {
logError(logger(), "Error while opening document: " + fileName);
return;
}
tCharMappingSp cm;
string_t fontName, newFontName;
wstring_t text, textUnicode, docAsText;
int c = 0;
tSelectionSp s = word()->getSelection();
int pos = 0;
int totalCharsQty = s->getStoryLength();
do {
s->setStart(pos);
s->setEnd(pos + 1);
s->selectCurrentFont();
fontName = s->getFont()->getName();
if ( canSkipFont(fontName) ) {
//s->getFont()->haveCommonAttributes();
pos = s->getEnd();
docAsText += s->getText();
std::cout << "\r" << percentageStr(pos, totalCharsQty - 1);
continue;
}
text = s->getText();
if ( fontName.empty() ) {
saveSelection(s);
fontName = makeGuess(s);
restoreSelection(s);
/// if after all we have empty font name, log about that event
/// and go forward
if (fontName.empty()) {
logError(logger(), "EMPTY FONT NAME: Investigate");
pos = s->getEnd();
docAsText += text;
std::cout << "\r" << percentageStr(pos, totalCharsQty - 1);
continue;
}
}
/// use mapping
textUnicode.clear();
cm = getCM(fontName);
if (cm) {
bool spacingOnly = cm->doConversion(text, textUnicode, fontName);
newFontName = getFontSubstitution(cm, fontName);
//tFontSp fontDup = s->getFont()->duplicate();
s->setText(textUnicode);
//s->getFont()->haveCommonAttributes();
s->getFont()->setName(newFontName);
//s->setFont(fontDup);
}
/// extract text from the document as well
docAsText += textUnicode;
pos = s->getEnd();
std::cout << "\r" << percentageStr(pos, totalCharsQty - 1);
} while ( pos < totalCharsQty - 1 );
/// -------------------------------------------///
/// now save result in the appropriate folder ///
string_t outputDir = getOutputAbsPath(fileName);
Poco::File(outputDir).createDirectories();
Poco::Path p(fileName);
doc->saveAs( outputDir + p.getBaseName() + " QUICK." + p.getExtension() );
doc->close();
if ( config_->getBool("app.saveAlsoAsUTF8", false) )
writeFileAsBinary( outputDir + p.getBaseName() + " UTF8 QUICK.txt", toUtf8(docAsText));
}
void Converter::convertSingleDocPrecise( const string_t& fileName )
{
#ifdef SECURITY_ENABLED
if ( !security_.getKey().updateCounters(0) ) {
logContent(security_.getKey());
return;
}
#endif
tDocumentsSp docs = word()->getDocuments();
tDocumentSp doc = docs->open(toUtf16(getInputAbsPath(fileName)));
if (!doc) {
logError(logger(), "Error while opening document: " + fileName);
return;
}
/// -------------------------------------------///
usedFonts_.clear();
wstring_t docAsText;
// tParagraphsSp paragraphs = doc->getParagraphs();
// int count = paragraphs->getCount();
// tSentencesSp sentences = doc->getSentences();
// int sentCount = sentences->getCount();
// for (int i = 1; i <= count; ++i) {
// tParagraphSp p = paragraphs->getItem(i);
// docAsText += processRangePrecise(p->getRange(), false);
// std::cout << "\r" << percentageStr(i, count);
// }
tRangeSp r = doc->getContent();
int64 totalBytes = r->getStoryLength();
docAsText += processRangePreciseVer2(r, true);
std::cout << std::endl;
/// footnotes
logInfo(logger(), "Processing [footnotes]: ");
tFootnotesSp footnots = doc->getFootnotes();
int notesCount = footnots->getCount();
for (int i = 1; i <= notesCount; ++i) {
tNoteSp note = footnots->getItem(i);
tRangeSp r = note->getRange();
processRangePreciseVer2(r, false);
std::cout << "\r" << percentageStr(i, notesCount);
}
if (notesCount > 0)
std::cout << std::endl;
tSectionsSp sections = doc->getSections();
int sectionsCount = sections->getCount();
for (int i = 1; i <= sectionsCount; ++i) {
tSectionSp section(new Section(sections->getItem(i)));
tHeadersFootersSp hfs = section->getHeaders();
if (hfs) {
logInfo(logger(), "Processing [headers]: ");
tHeaderFooterSp hf( new HeaderFooter(hfs->getItem(1)) );
tRangeSp r = hf->getRange();
processRangePreciseVer2(r, false);
}
hfs = section->getFooters();
if (hfs) {
logInfo(logger(), "Processing [footers]: ");
tHeaderFooterSp hf( new HeaderFooter(hfs->getItem(1)) );
tRangeSp r = hf->getRange();
processRangePreciseVer2(r, false);
}
}
#ifdef SECURITY_ENABLED
security_.getKey().updateCounters(totalBytes);
#endif
logUsedFonts(fileName, usedFonts_);
usedFonts_.clear();
/// now save result in the appropriate folder
string_t outputDir = getOutputAbsPath(fileName);
Poco::File(outputDir).createDirectories();
Poco::Path p(fileName);
logInfo(logger(), "Saving document...");
doc->saveAs( outputDir + p.getBaseName() + " UNICODE." + p.getExtension() );
doc->close();
logInfo(logger(), "Save was successful.");
if ( wantUtf8Text_ )
writeFileAsBinary( outputDir + p.getBaseName() + " UTF8.txt", toUtf8(docAsText));
#ifdef SECURITY_ENABLED
logContent(security_.getKey());
#endif
}
