//sumatrapdf code
int extractText(miniexp_t item, Arraylist list, fz_bbox * target) {
miniexp_t type = miniexp_car(item);
if (!miniexp_symbolp(type))
return 0;
item = miniexp_cdr(item);
if (!miniexp_numberp(miniexp_car(item))) return 0;
int x0 = miniexp_to_int(miniexp_car(item)); item = miniexp_cdr(item);
if (!miniexp_numberp(miniexp_car(item))) return 0;
int y0 = miniexp_to_int(miniexp_car(item)); item = miniexp_cdr(item);
if (!miniexp_numberp(miniexp_car(item))) return 0;
int x1 = miniexp_to_int(miniexp_car(item)); item = miniexp_cdr(item);
if (!miniexp_numberp(miniexp_car(item))) return 0;
int y1 = miniexp_to_int(miniexp_car(item)); item = miniexp_cdr(item);
//RectI rect = RectI::FromXY(x0, y0, x1, y1);
fz_bbox rect = {x0 , y0 , x1 , y1};
miniexp_t str = miniexp_car(item);
if (miniexp_stringp(str) && !miniexp_cdr(item)) {
fz_bbox inters = fz_intersect_bbox(rect, *target);
//LOGI("Start text extraction: rectangle=[%d,%d,%d,%d] %s", rect.x0, rect.y0, rect.x1, rect.y1, content);
if (!fz_is_empty_bbox(inters)) {
const char *content = miniexp_to_str(str);
while (*content) {
arraylist_add(list, *content++);
}
// if (value) {
// size_t len = str::Len(value);
// // TODO: split the rectangle into individual parts per glyph
// for (size_t i = 0; i < len; i++)
// coords.Append(RectI(rect.x, rect.y, rect.dx, rect.dy));
// extracted.AppendAndFree(value);
// }
if (miniexp_symbol("word") == type) {
arraylist_add(list, ' ');
//coords.Append(RectI(rect.x + rect.dx, rect.y, 2, rect.dy));
}
else if (miniexp_symbol("char") != type) {
arraylist_add(list, '\n');
// extracted.Append(lineSep);
// for (size_t i = 0; i < str::Len(lineSep); i++)
// coords.Append(RectI());
}
}
item = miniexp_cdr(item);
}
while (miniexp_consp(str)) {
extractText(str, list, target);
item = miniexp_cdr(item);
str = miniexp_car(item);
}
return !item;
}
Pvoid_t html_tokenize_into_features(const char * html, Pvoid_t features) {
xmlSubstituteEntitiesDefault(0);
htmlDocPtr doc = htmlParseDoc(BAD_CAST html, "UTF-8");
if (doc) {
Buffer *buf = extractText(doc);
features = tokenize_text(buf->buf, buf->length, features);
features = tokenize_uris(doc, features);
free_buffer(buf);
xmlFreeDoc(doc);
}
return features;
}
JNIEXPORT jstring JNICALL
Java_universe_constellation_orion_viewer_djvu_DjvuDocument_getText(JNIEnv *env, jobject thiz, int pageNumber,
int startX, int startY, int width, int height)
{
LOGI("==================Start Text Extraction==============");
miniexp_t pagetext;
while ((pagetext=ddjvu_document_get_pagetext(doc,pageNumber,0))==miniexp_dummy) {
//handle_ddjvu_messages(ctx, TRUE);
}
if (miniexp_nil == pagetext) {
return NULL;
}
//
ddjvu_status_t status;
ddjvu_pageinfo_t info;
while ((status = ddjvu_document_get_pageinfo(doc, pageNumber, &info)) < DDJVU_JOB_OK) {
//nothing
}
LOGI("Extraction rectangle=[%d,%d,%d,%d]", startX, startY, width, height);
Arraylist values = arraylist_create();
int w = info.width;
int h = info.height;
fz_bbox target = {startX, h - startY - height, startX + width, h - startY};
LOGI("Extraction irectangle=[%d,%d,%d,%d]", target.x0, target.y0, target.x1, target.y1);
extractText(pagetext, values, &target);
arraylist_add(values, 0);
LOGI("Data: %s", arraylist_getData(values));
jstring result = (*env)->NewStringUTF(env, arraylist_getData(values));
arraylist_free(values);
return result;
}
bool CodeCompletionsExtractor::next()
{
const uint cxCodeCompleteResultCount = cxCodeCompleteResults->NumResults;
++cxCodeCompleteResultIndex;
if (cxCodeCompleteResultIndex < cxCodeCompleteResultCount) {
currentCxCodeCompleteResult = cxCodeCompleteResults->Results[cxCodeCompleteResultIndex];
currentCodeCompletion_ = CodeCompletion();
extractCompletionKind();
extractText();
extractPriority();
extractAvailability();
extractHasParameters();
extractCompletionChunks();
adaptPriority();
return true;
}
return false;
}
bool KDReports::XmlParser::processNode( const QDomNode& node, KDReports::ReportBuilder* builder, bool inHeader, bool inFooter )
{
// Loop over elements
for ( QDomElement element = node.firstChildElement() ; !element.isNull();
element = element.nextSiblingElement() ) {
if ( testForErrorAndFillErrorDetails() )
return false;
const QString name = element.tagName();
if( name == QLatin1String( "text" ) ) {
// Handle <text> element
KDReports::TextElement textElement;
QString id;
const QString text = extractText( element, &id, m_report->d->m_currentModel, m_report->d->m_currentRow );
textElement.setText( text );
textElement.setId( id );
const QColor bgColor = KDReports::XmlHelper::readBackground( element );
if( bgColor.isValid() )
textElement.setBackground( bgColor );
if( element.hasAttribute( QLatin1String( "pointsize" ) ) ) {
const int pointSize = element.attribute( QLatin1String( "pointsize" ) ).toInt();
if( pointSize )
textElement.setPointSize( pointSize );
}
if( element.hasAttribute( QLatin1String( "color" ) ) ) {
const QString name = element.attribute( QLatin1String( "color" ) );
textElement.setTextColor( QColor( name ) );
}
if( element.hasAttribute( QLatin1String( "font" ) ) ) {
textElement.setFontFamily( element.attribute( QLatin1String( "font" ) ) );
}
if( element.hasAttribute( QLatin1String( "bold" ) ) ) {
bool bold = false;
if( element.attribute( QLatin1String( "bold" ) ) == QLatin1String( "true" ) )
bold = true;
textElement.setBold( bold );
}
if( element.hasAttribute( QLatin1String( "italic" ) ) ) {
bool italic = false;
if( element.attribute( QLatin1String( "italic" ) ) == QLatin1String( "true" ) )
italic = true;
textElement.setItalic( italic );
}
if( element.hasAttribute( QLatin1String( "strikeout" ) ) ) {
bool strikeOut = false;
if( element.attribute( QLatin1String( "strikeout" ) ) == QLatin1String( "true" ) )
strikeOut = true;
textElement.setStrikeOut( strikeOut );
}
if( element.hasAttribute( QLatin1String( "underline" ) ) ) {
bool underline = false;
if( element.attribute( QLatin1String( "underline" ) ) == QLatin1String( "true" ) )
underline = true;
textElement.setUnderline( underline );
}
const QString oldId = textElement.id();
if ( m_xmlElementHandler && !m_xmlElementHandler->textElement( textElement, element ) )
continue;
if ( textElement.id() != oldId ) {
// The handler modified the text element's id, look up the value again.
const QHash<QString, QString>::const_iterator it = m_textValues.constFind( textElement.id() );
if ( it != m_textValues.constEnd() ) {
textElement.setText( *it );
}
}
if ( !builder ) {
error( QObject::tr( "<text> is only supported in WordProcessing mode" ) );
} else {
if( element.hasAttribute( QLatin1String( "inline" ) ) ) {
builder->addInlineElementPublic( textElement );
} else {
Qt::AlignmentFlag alignment = Qt::AlignLeft;
if( element.hasAttribute( QLatin1String( "alignment" ) ) )
alignment = KDReports::XmlHelper::stringToAlignment( element.attribute( QLatin1String( "alignment" ) ) );
const QColor bgColor = KDReports::XmlHelper::readColor( element, "paragraph-background" );
builder->addBlockElementPublic( textElement, alignment, bgColor );
}
}
} else if( name == QLatin1String( "html" ) ) {
// Handle <html> element
KDReports::HtmlElement htmlElement;
QString id;
const QString text = extractText( element, &id );
htmlElement.setHtml( text );
htmlElement.setId( id );
QColor bgColor = KDReports::XmlHelper::readBackground( element );
if( bgColor.isValid() )
htmlElement.setBackground( bgColor );
const QString oldId = htmlElement.id();
if ( m_xmlElementHandler && !m_xmlElementHandler->htmlElement( htmlElement, element ) )
continue;
if ( htmlElement.id() != oldId ) {
// The handler modified the text element's id, look up the value again.
const QHash<QString, QString>::const_iterator it = m_textValues.constFind( htmlElement.id() );
if ( it != m_textValues.constEnd() ) {
htmlElement.setHtml( *it );
}
}
addElement( htmlElement, builder, element );
} else if ( name == QLatin1String( "tabs" ) ) {
if ( !builder ) {
error( QObject::tr( "<tabs> is only supported in WordProcessing mode" ) );
} else {
parseTabs( builder, element );
}
} else if ( name == QLatin1String( "paragraph-margins" ) ) {
if ( !builder ) {
error( QObject::tr( "<paragraph-margins> is only supported in WordProcessing mode" ) );
} else {
parseParagraphMargins( builder, element );
}
} else if( name == QLatin1String( "hr" ) ) {
// Handle <hr> element
KDReports::HtmlElement htmlElement;
htmlElement.setHtml( QString::fromLatin1("<hr>") );
if ( m_xmlElementHandler && !m_xmlElementHandler->htmlElement( htmlElement, element ) )
continue;
addElement( htmlElement, builder, element );
} else if( name == QLatin1String( "vspace" ) ) {
// Handle <vspace> element
if( !element.hasAttribute( QLatin1String( "size" ) ) )
continue; // size attribute is mandatory for VSpace
int size = element.attribute( QLatin1String( "size" ) ).toInt();
if ( !builder ) {
error( QObject::tr( "<vspace> is only supported in WordProcessing mode" ) );
} else {
if( builder != m_report->d->builder() ) {
error( QObject::tr("<vspace> not allowed in headers, footers, or table cells") );
return false;
}
}
XmlElementHandlerV2* v2 = dynamic_cast<XmlElementHandlerV2 *>(m_xmlElementHandler);
if (v2 && !v2->vspace(size, element))
continue;
m_report->addVerticalSpacing( size );
} else if( name == QLatin1String( "table" ) ) {
// Handle <table> element
const QString model = element.attribute( QLatin1String( "model" ) );
if ( model.isEmpty() ) {
if ( !builder ) {
error( QObject::tr( "<table> without a model is only supported in WordProcessing mode" ) );
continue;
}
KDReports::TableElement tableElement;
const int headerRowCount = element.attribute( QLatin1String( "headerRowCount" ) ).toInt(); // default 0
tableElement.setHeaderRowCount( headerRowCount );
if ( element.hasAttribute( QLatin1String( "cellpadding" ) ) )
tableElement.setPadding( element.attribute( QLatin1String( "cellpadding" ) ).toInt() );
parseCommonTableAttributes( tableElement, element );
if ( m_xmlElementHandler && !m_xmlElementHandler->startTableElement( tableElement, element ) )
continue;
if ( !parseTableContents( tableElement, element, *builder, inHeader, inFooter ) )
return false;
if ( m_xmlElementHandler && !m_xmlElementHandler->endTableElement( tableElement, element ) )
continue;
addElement( tableElement, builder, element );
} else {
KDReports::AutoTableElement tableElement( model );
if( element.attribute( QLatin1String( "verticalHeaderVisible" ) ) == QLatin1String( "false" ) )
tableElement.setVerticalHeaderVisible(false);
if( element.attribute( QLatin1String( "horizontalHeaderVisible" ) ) == QLatin1String( "false" ) )
tableElement.setHorizontalHeaderVisible(false);
QColor headerBgColor = KDReports::XmlHelper::readColor( element, "header-background" );
if ( headerBgColor.isValid() )
tableElement.setHeaderBackground( headerBgColor );
parseCommonTableAttributes( tableElement, element );
if ( m_xmlElementHandler && !m_xmlElementHandler->autoTableElement( tableElement, element ) )
continue;
if ( m_report->reportMode() == Report::SpreadSheet ) {
m_report->mainTable()->setAutoTableElement( tableElement );
} else {
addElement( tableElement, builder, element );
}
}
} else if( name == QLatin1String( "chart" ) ) {
// Handle <chart> element
KDReports::ChartElement chartElement( element.attribute( QLatin1String( "model" ) ) );
QColor bgColor = KDReports::XmlHelper::readBackground( element );
if( bgColor.isValid() )
chartElement.setBackground( bgColor );
int width = 100, height = 100;
Unit unit = Millimeters;
if( element.hasAttribute( QLatin1String( "width" ) ) ) {
QString str = element.attribute( QLatin1String( "width" ) );
if ( str.endsWith( QLatin1Char('%') ) ) {
str.chop( 1 );
unit = Percent;
}
width = str.toInt();
}
if( element.hasAttribute( QLatin1String( "height" ) ) ) {
QString str = element.attribute( QLatin1String( "height" ) );
if ( str.endsWith( QLatin1Char('%') ) ) {
str.chop( 1 );
unit = Percent;
}
height = str.toInt();
}
chartElement.setSize( width, height, unit );
if ( m_xmlElementHandler && !m_xmlElementHandler->chartElement( chartElement, element ) )
continue;
addElement( chartElement, builder, element );
} else if( name == QLatin1String( "image" ) ) {
// Handle <image> element
QString id;
QImage image = extractImage( element, &id );
KDReports::ImageElement imageElement( image );
imageElement.setId(id);
if( element.hasAttribute( QLatin1String( "width" ) ) ) {
QString widthStr = element.attribute( QLatin1String( "width" ) );
if ( widthStr.endsWith( QLatin1Char('%') ) ) {
widthStr.truncate( widthStr.length() - 1 );
imageElement.setWidth( widthStr.toInt(), KDReports::Percent );
} else {
imageElement.setWidth( widthStr.toInt() );
}
} else if( element.hasAttribute( QLatin1String( "height" ) ) ) { // mutually exclusive with "width"!
QString heightStr = element.attribute( QLatin1String( "height" ) );
if ( heightStr.endsWith( QLatin1Char('%') ) ) {
heightStr.truncate( heightStr.length() - 1 );
imageElement.setHeight( heightStr.toInt(), KDReports::Percent );
} else {
imageElement.setHeight( heightStr.toInt() );
}
} else if ( element.hasAttribute( QLatin1String( "fitToPage" ) ) ) {
imageElement.setFitToPage();
}
if ( m_xmlElementHandler && !m_xmlElementHandler->imageElement( imageElement, element ) )
continue;
addElement( imageElement, builder, element );
} else if( name == QLatin1String( "header" ) ) {
const KDReports::HeaderLocations loc = KDReports::XmlHelper::parseHeaderLocation( element.attribute( QLatin1String( "location" ) ) );
KDReports::Header& header = m_report->header( loc );
parseHeaderFooterAttribute( header, element );
if ( m_xmlElementHandler && !m_xmlElementHandler->startHeader( header, element ) )
continue;
if( !processNode( element, &header.d->m_builder, true, false ) )
return false;
if ( m_xmlElementHandler )
m_xmlElementHandler->endHeader( header, element );
} else if( name == QLatin1String( "footer" ) ) {
const KDReports::HeaderLocations loc = KDReports::XmlHelper::parseHeaderLocation( element.attribute( QLatin1String( "location" ) ) );
KDReports::Footer& footer = m_report->footer( loc );
parseHeaderFooterAttribute( footer, element );
if ( m_xmlElementHandler && !m_xmlElementHandler->startFooter( footer, element ) )
continue;
if( !processNode( element, &footer.d->m_builder, false, true ) )
return false;
if ( m_xmlElementHandler )
m_xmlElementHandler->endFooter( footer, element );
} else if( name == QLatin1String( "variable" ) ) {
if( !inHeader && !inFooter ) {
error( QObject::tr("<variable> tags only allowed in headers and footers") );
return false;
}
if( !element.hasAttribute( QLatin1String( "type" ) ) ) {
error( QObject::tr("<variable> tags must have a 'type' attribute") );
return false;
}
Q_ASSERT( builder );
if ( builder ) {
const QString type = element.attribute( QLatin1String( "type" ) );
KDReports::VariableType vt = KDReports::XmlHelper::stringToVariableType( type );
XmlElementHandlerV2* v2 = dynamic_cast<XmlElementHandlerV2 *>(m_xmlElementHandler);
if (v2 && !v2->variable(vt, element))
continue;
builder->addVariablePublic( vt );
}
} else if ( name == QLatin1String( "page-break" ) ) {
if ( m_xmlElementHandler && !m_xmlElementHandler->pageBreak( element ) )
continue;
m_report->addPageBreak();
} else if ( name == QLatin1String( "ifdef" ) ) {
if ( element.hasAttribute( QLatin1String( "id" ) ) ) {
const QString id = element.attribute( QLatin1String( "id" ) );
const bool skip = m_textValues.value( id ).isEmpty();
if( !skip ) {
if ( !processNode( element, builder, inHeader, inFooter ) )
return false;
}
}
} else if ( name == QLatin1String( "custom" ) ) {
if ( m_xmlElementHandler )
m_xmlElementHandler->customElement( element );
} else if ( name == QLatin1String( "hline" ) ) {
KDReports::HLineElement hLineElement;
if ( element.hasAttribute( QLatin1String( "thickness" ) ) ) {
const double thickness = element.attribute( QLatin1String( "thickness" ) ).toDouble();
hLineElement.setThickness( thickness );
}
if ( element.hasAttribute( QLatin1String( "color" ) ) ) {
const QColor color = KDReports::XmlHelper::readColor( element, "color" );
hLineElement.setColor( color );
}
if ( element.hasAttribute( QLatin1String( "margin" ) ) ) {
const int margin = element.attribute( QLatin1String( "margin" ) ).toInt();
hLineElement.setMargin( margin );
}
#ifdef KDREPORTS_ALLOW_BINARY_INCOMPATIBILITY
if ( m_xmlElementHandler && !m_xmlElementHandler->hLineElement( hLineElement, element ) )
continue;
#else
XmlElementHandlerV2* v2 = dynamic_cast<XmlElementHandlerV2 *>(m_xmlElementHandler);
if (v2 && !v2->hLineElement(hLineElement, element))
continue;
#endif
addElement( hLineElement, builder, element );
} else {
error( QObject::tr( "KDReports::Report::loadFromXML(): Unknown element %1" ).arg( name ) );
}
}
if ( testForErrorAndFillErrorDetails() )
return false;
return true;
}
