HtmlDocument::HtmlDocument(const QString docfilename, bool bypasscontroll)
: DObject() {
bypass_validate = bypasscontroll;
this->SetBaseFileDoc(docfilename);
QFileInfo infofromfile(docfilename);
HTMLSTREAM.clear();
bool is_loading = false;
bool debugmodus = (DEBUGMODUS == 1) ? true : false;
console(QString("Init talking this class. %1..").arg(this->name()), 0);
bool doc_continue = true;
console(QString("OpenFile: %1").arg(docfilename));
this->SetStyle("docx2html.xsl");
KZip::Stream *unzip = new KZip::Stream(docfilename);
if (debugmodus) {
unzip->explode_todir(this->get_cachedir(), 1);
}
if (unzip->canread() && docx_validator(unzip->filelist())) {
console(QString("Zip is open OK docx validate ok."));
corefile = unzip->listData();
unzip->~Stream();
unzip = NULL;
QByteArray xmldataword = docitem("word/document.xml");
const QString rootxmlfile = QString("%1index.xml").arg(this->get_cachedir());
const QString indexhtmlfile = QString("%1index.html").arg(this->get_cachedir());
QFile::remove(rootxmlfile);
QFile::remove(indexhtmlfile);
QFile::remove(TEMPFILEXSLTRESULTS); /// if exist.. xslt2 temp file on cache up
if (xmldataword.size() > 0) {
doc_continue = Tools::_write_file(rootxmlfile, xmldataword);
}
///// contibue ok...
if (doc_continue) {
/// convert xslt now
console(QString("Init xslt 1 convert class DOCXML::XsltVersion1."));
QStringList xsltparam;
DOCXML::Xslt xsltOne(xsltparam);
if (xsltOne.TransformXSLT_1(this->getStyleFile(), rootxmlfile)) {
HTMLSTREAM = xsltOne.streamUtf8();
scanimage();
is_loading = Tools::_write_file(indexhtmlfile, HTMLSTREAM, "utf-8");
if (!is_loading) {
this->setError(QString("Unable to write on home cache.."));
} else {
lastreturnfile_html = indexhtmlfile;
QString success = QString("Valid document size: %1.").arg( SystemSecure::bytesToSize( HTMLSTREAM.size()));
console(success);
}
} else {
this->setError(QString("Register a xslt 1 error .. activate debug to find.."));
}
}
} else {
is_loading = false;
this->setError(QString("Document bad format unable to open as zip..."));
}
LASTERRORMSG = ""; /// error from image not open format ...
if (this->basicError().size() > 0) {
LASTERRORMSG = this->basicError().join(" - ");
}
}
void FoxGroundDocument::SetBaseFileDoc(const QString file) {
QFileInfo localfile(file);
if (!localfile.exists()) {
CurrentRunDoc = UNKNOW;
return;
}
document_file = file;
//// try on pole...
bool poleisrun = false;
POLE::Storage storage(file.toLocal8Bit());
if (storage.open()) {
QBuffer basestore;
if (Conversion::readStreamPole(storage, "/SummaryInformation", basestore)) {
LEInputStream *sis_baseSummaryInformation = new LEInputStream(&basestore);
poleisrun = true;
CurrentRunDoc = OFFICEMSO;
document_name = "DOC";
}
}
QByteArray x;
QFile *f = new QFile(file);
if (f->open(QIODevice::ReadOnly)) {
x = f->readAll();
}
f->close();
const QByteArray data = x;
QDomDocument document;
QString c;
int d, o;
bool xmlbased = false;
if (document.setContent(data, true, &c)) {
/// not xml based...
xmlbased = true;
}
const QByteArray rtfsearch = x.mid(0, 5).simplified();
/// qDebug() << "rtfsearch :" << rtfsearch;
//// no office family .. doc
if (!poleisrun) {
if (xmlbased && data.indexOf("encoding=") != -1) {
if (data.indexOf("DOCTYPE html") > 0) {
CurrentRunDoc = HTML;
document_name = "HTML";
} else {
document_name = "XML";
CurrentRunDoc = XML;
}
} else if (rtfsearch == "{\\rtf") {
CurrentRunDoc = RTF;
document_name = "RTF";
} else if (data.mid(0, 4) == "%PDF") {
if (bin_pdftohtml.size() > 3) {
CurrentRunDoc = PDF;
document_name = "PDF";
} else {
CurrentRunDoc = UNKNOW;
}
} else if (data.mid(0, 2) == "PK") {
KZip::Stream *unzip = new KZip::Stream(file);
if (unzip->canread()) {
unzipfileliste = unzip->filelist();
corefile = unzip->listData();
if (docitem("word/document.xml").size() > 20) {
CurrentRunDoc = DOCX;
document_name = "DOCX";
} else if (docitem("content.xml").size() > 20) {
CurrentRunDoc = ODT;
document_name = "ODT";
} else {
CurrentRunDoc = UNKNOW;
}
}
unzip->~Stream();
unzip = NULL;
} else {
CurrentRunDoc = UNKNOW;
}
}
if (CurrentRunDoc == UNKNOW) {
document_name = "NULL";
}
}
DocxZipHandler::DocxZipHandler(const QString docfilename) : Document("DOCX"), is_loading(false) {
QFileInfo infofromfile(docfilename);
bool debugmodus = (DEBUGMODUS == 1)? true : false;
console(QString("Init talking this class. %1..").arg(this->name()), 0);
bool doc_continue = true;
console(QString("OpenFile: %1").arg(docfilename));
this->SetBaseFileDoc(docfilename);
this->SetStyle(__localDOCXSTYLE__);
KZip::Stream *unzip = new KZip::Stream(docfilename);
if (debugmodus) {
unzip->explode_todir(cache_dir, 1);
}
if (unzip->canread() && docx_validator(unzip->filelist())) {
console(QString("Zip is open OK doc valid ...."));
corefile = unzip->listData();
unzip->~Stream();
unzip = NULL;
QByteArray xmldataword = docitem("word/document.xml");
const QString rootxmlfile = QString("%1index.xml").arg(cache_dir);
const QString indexhtmlfile = QString("%1index.html").arg(cache_dir);
QFile::remove(rootxmlfile);
QFile::remove(indexhtmlfile);
if (xmldataword.size() > 0) {
doc_continue = Tools::_write_file(rootxmlfile, xmldataword);
}
///// contibue ok...
if (doc_continue && !bin_xsltproc.isEmpty()) {
/// convert xslt now
console(QString("Init xslt convert...%1.").arg(bin_xsltproc));
QStringList xsltparam;
xsltparam << "--param";
xsltparam << "convert_time";
xsltparam << QString("'%1'").arg(Tools::TimeNow());
xsltparam << "--output";
xsltparam << indexhtmlfile;
xsltparam << style_file;
xsltparam << rootxmlfile;
/// xslt param --param data_path2 "'e:/tmp/stu/xx'"
const QString errorcmd = execlocal(bin_xsltproc, xsltparam);
if (errorcmd.isEmpty()) {
console(QString("Xslt convert successful document..."));
} else {
console(QString("Xslt convert... say: %1").arg(errorcmd));
}
}
QByteArray prehtml = this->loadbinfile(indexhtmlfile);
HTMLSTREAM = QString::fromUtf8(prehtml.constData());
if (doc_continue && prehtml.size() > 0) {
scanimage(); /// insert image on html embedet
is_loading = Tools::_write_file(indexhtmlfile, HTMLSTREAM, "utf-8");
console(QString("Write last file: %1").arg(indexhtmlfile));
}
if (!doc_continue) {
is_loading = false;
console(QString("Error on Buffer...."), 3);
}
} else {
is_loading = false;
console(QString("Document bad format "), 3);
}
clean();
}