org.docx4j.convert.in.xhtml.TableHelper Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of docx4j-ImportXHTML Show documentation
Show all versions of docx4j-ImportXHTML Show documentation
docx4j-ImportXHTML converts XHTML to OpenXML WordML (docx) using docx4j
package org.docx4j.convert.in.xhtml;
import java.math.BigInteger;
import java.util.ArrayList;
import java.util.Collections;
import java.util.LinkedList;
import org.docx4j.UnitsOfMeasurement;
import org.docx4j.XmlUtils;
import org.docx4j.convert.in.xhtml.XHTMLImporterImpl.TableProperties;
import org.docx4j.jaxb.Context;
import org.docx4j.model.properties.table.tr.TrHeight;
import org.docx4j.wml.CTBorder;
import org.docx4j.wml.CTHeight;
import org.docx4j.wml.CTShd;
import org.docx4j.wml.CTTblLayoutType;
import org.docx4j.wml.CTTblPrBase.TblStyle;
import org.docx4j.wml.CTVerticalJc;
import org.docx4j.wml.ContentAccessor;
import org.docx4j.wml.P;
import org.docx4j.wml.STBorder;
import org.docx4j.wml.STTblLayoutType;
import org.docx4j.wml.STVerticalJc;
import org.docx4j.wml.Style;
import org.docx4j.wml.Tbl;
import org.docx4j.wml.TblBorders;
import org.docx4j.wml.TblGrid;
import org.docx4j.wml.TblGridCol;
import org.docx4j.wml.TblPr;
import org.docx4j.wml.TblWidth;
import org.docx4j.wml.Tc;
import org.docx4j.wml.TcMar;
import org.docx4j.wml.TcPr;
import org.docx4j.wml.TcPrInner;
import org.docx4j.wml.TcPrInner.GridSpan;
import org.docx4j.wml.TcPrInner.VMerge;
import org.docx4j.wml.Tr;
import org.docx4j.wml.TrPr;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.w3c.dom.Element;
import com.openhtmltopdf.css.constants.CSSName;
import com.openhtmltopdf.css.constants.IdentValue;
import com.openhtmltopdf.css.parser.FSColor;
import com.openhtmltopdf.css.parser.FSRGBColor;
import com.openhtmltopdf.css.style.CalculatedStyle;
import com.openhtmltopdf.css.style.FSDerivedValue;
import com.openhtmltopdf.css.style.derived.LengthValue;
import com.openhtmltopdf.newtable.TableBox;
import com.openhtmltopdf.newtable.TableCellBox;
import com.openhtmltopdf.newtable.TableSectionBox;
import com.openhtmltopdf.render.Box;
public class TableHelper {
private static final String TABLE = "table";
public static Logger log = LoggerFactory.getLogger(TableHelper.class);
private XHTMLImporterImpl importer;
protected TableHelper(XHTMLImporterImpl importer) {
this.importer=importer;
}
/**
* @param cssTable
* @param tbl
* @param tableProperties
*/
protected void setupTblPr(TableBox cssTable, Tbl tbl, TableProperties tableProperties) {
Element e = cssTable.getElement();
TblPr tblPr = Context.getWmlObjectFactory().createTblPr();
tbl.setTblPr(tblPr);
String cssClass = null;
if (e.getAttribute("class")!=null) {
cssClass=e.getAttribute("class").trim();
setTableStyle(tblPr, cssClass);
}
// table borders
TblBorders borders = Context.getWmlObjectFactory().createTblBorders();
borders.setTop( copyBorderStyle(cssTable, "top", true) );
borders.setBottom( copyBorderStyle(cssTable, "bottom", true) );
borders.setLeft( copyBorderStyle(cssTable, "left", true) );
borders.setRight( copyBorderStyle(cssTable, "right", true) );
borders.setInsideH( createBorderStyle(STBorder.NONE, null, null) );
borders.setInsideV( createBorderStyle(STBorder.NONE, null, null) );
tblPr.setTblBorders(borders);
TblWidth spacingWidth = Context.getWmlObjectFactory().createTblWidth();
if(cssTable.getStyle().isCollapseBorders()) {
spacingWidth.setW(BigInteger.ZERO);
spacingWidth.setType(TblWidth.TYPE_AUTO);
} else {
int cssSpacing = cssTable.getStyle().getBorderHSpacing(importer.getRenderer().getLayoutContext());
spacingWidth.setW( BigInteger.valueOf(cssSpacing / 2) ); // appears twice thicker, probably taken from both sides
spacingWidth.setType(TblWidth.TYPE_DXA);
}
tblPr.setTblCellSpacing(spacingWidth);
// Table indent.
// cssTable.getLeftMBP() which is setLeftMBP((int) margin.left() + (int) border.left() + (int) padding.left());
// cssTable.getTx(); which is (int) margin.left() + (int) border.left() + (int) padding.left();
// But want just margin.left
log.debug("list depth:" + importer.getListHelper().getDepth());
if (importer.getListHelper().getDepth()>0 ) {
TblWidth tblInd = tblPr.getTblInd();
if (tblInd==null) {
tblInd = new TblWidth();
tblPr.setTblInd(tblInd);
}
tblInd.setType(TblWidth.TYPE_DXA);
// Handle using the same logic we use for indenting a paragraph
int totalPadding = importer.getListHelper().getAbsoluteIndent(cssTable);
int tableIndentContrib = tableIndentContrib(importer.getContentContextStack());
if (importer.getListHelper().peekListItemStateStack().isFirstChild) {
// totalPadding gives indent to the bullet;
log.debug("Table in list indent case 1: tblInd set for item itself");
tblInd.setW(BigInteger.valueOf(totalPadding-tableIndentContrib));
importer.getListHelper().peekListItemStateStack().isFirstChild=false;
} else {
// totalPadding gives indent to the bullet;
// we want to align this subsequent p with the preceding text;
// assume 360 twips
log.debug("Table in list indent case 2: tblInd set for follwing child");
tblInd.setW(BigInteger.valueOf(totalPadding+ListHelper.INDENT_AFTER-tableIndentContrib));
}
} else if (cssTable.getMargin(importer.getRenderer().getLayoutContext()) !=null
&& cssTable.getMargin(importer.getRenderer().getLayoutContext()).left()>0) {
log.debug("Calculating TblInd from margin.left: " + cssTable.getMargin(importer.getRenderer().getLayoutContext()).left() );
TblWidth tblInd = Context.getWmlObjectFactory().createTblWidth();
tblInd.setW( BigInteger.valueOf( Math.round(
cssTable.getMargin(importer.getRenderer().getLayoutContext()).left()
)));
tblInd.setType(TblWidth.TYPE_DXA);
tblPr.setTblInd(tblInd);
} else {
// Indent is zero. In this case, if the table has borders,
// adjust the indent to align the left border with the left edge of text outside the table
// See http://superuser.com/questions/126451/changing-the-placement-of-the-left-border-of-tables-in-word
CTBorder leftBorder = borders.getLeft();
if (leftBorder!=null
&& leftBorder.getVal()!=null
&& leftBorder.getVal()!=STBorder.NONE
&& leftBorder.getVal()!=STBorder.NIL) {
// set table indent to .08", ie 115 twip
//
//
// but for an AutoFit table (most common), it
// is the w:gridCol val which prob specifies the actual width
TblGrid tblGrid = Context.getWmlObjectFactory().createTblGrid();
tbl.setTblGrid(tblGrid);
int[] colPos = tableProperties.getColumnPos();
log.debug("setupTblGrid " + cssTable.numEffCols() + " " + colPos.length);
for (int i=1; i<=cssTable.numEffCols(); i++) {
TblGridCol tblGridCol = Context.getWmlObjectFactory().createTblGridCol();
tblGrid.getGridCol().add(tblGridCol);
log.debug("colpos=" + colPos[i]);
tblGridCol.setW( BigInteger.valueOf(colPos[i]-colPos[i-1]) );
}
}
protected void setupTrPr(com.openhtmltopdf.newtable.TableRowBox trBox, Tr tr) {
TrPr trPr = Context.getWmlObjectFactory().createTrPr();
tr.setTrPr(trPr);
/* Row height:-
*
* In HTML, you can only set height on a cell td, not tr, but Flying Saucer calculates this
* based on cell contents.
*
* Since there is no height property on a tr in CSS, this.importer.getCascadedProperties
* does not contain that property.
*
* But happily, Flying Saucer, sets trBox.getHeight()
*
* A web browser ignores td height 0 (tested in Chrome), or any value less than height of contents.
*
* But Flying Saucer, at least the version we modified, sets trBox.getHeight() based on
* td CSS values, not the greater of the specified height and the actual height of the contents.
*
* So in Word, we want to use STHeightRule.AT_LEAST (rather than EXACT) so contents don't
* get cut off.
*
* 2021 11 03: openhtmltopdf produces something like 1114 for a single line of text.
*
*/
int height = trBox.getHeight(); // 100px = 2000 so looks like 1/20 of a pixel!
if (height == 0) {
// do nothing; equivalent per spec to STHeightRule.AUTO
} else {
log.debug("trBox.getHeight(): " + height);
// Since we don't have a CSSValue we can use, do it manually
TrHeight thr = new TrHeight(); // uses STHeightRule.AT_LEAST
thr.set(trPr);
/* Per https://github.com/danfickle/openhtmltopdf/blob/4a475a91563f7f0c8c1d81029b1726e4ff112093/openhtmltopdf-pdfbox/src/main/java/com/openhtmltopdf/pdfboxout/PdfBoxFastOutputDevice.java#L76
// PDF points are defined as 1/72 inch.
// CSS pixels are defined as 1/96 inch.
// PDF text units are defined as 1/1000 of a PDF point.
// OpenHTMLtoPDF dots are defined as 1/20 of a CSS pixel.
// Therefore dots per point is 20 * 96/72 or about 26.66.
// Dividing by _dotsPerPoint will convert OpenHTMLtoPDF dots to PDF points.
* *
*/
int twip = UnitsOfMeasurement.pxToTwip(height/20);
((CTHeight)thr.getObject()).setVal(BigInteger.valueOf(twip));
}
}
protected void setupTcPr(TableCellBox tcb, Tc tc, TableProperties tableProperties) {
int effCol = tcb.getTable().colToEffCol(tcb.getCol());
// Do we need a vMerge tag with "restart" attribute?
// get cell below (only 1 section supported at present)
TcPr tcPr = Context.getWmlObjectFactory().createTcPr();
tc.setTcPr(tcPr);
if (tcb.getStyle().getRowSpan()> 1) {
VMerge vm = Context.getWmlObjectFactory().createTcPrInnerVMerge();
vm.setVal("restart");
tcPr.setVMerge(vm);
}
// eg adjCells = new ArrayList();
int numEffCols = tcb.getTable().numEffCols();
for ( int i = tcb.getCol(); i >= 0 && i < numEffCols; i += backwards ? -1 : 1 ) {
TableSectionBox tsb = tcb.getSection();
TableCellBox adjCell = tsb.cellAt(tcb.getRow(), i);
if ( adjCell == null ) {
// Check your table is OK
log.error("XHTML table import: Null adjCell for row " + tcb.getRow() + ", col " + tcb.getCol() + " at col " + i);
break;
}
if ( adjCell == tcb || adjCell == TableCellBox.SPANNING_CELL ) {
continue;
}
log.debug("Got adjCell, it is " + adjCell.getRow() + ", " + adjCell.getCol());
if ( adjCell.getRow() < tcb.getRow()
&& adjCell.getStyle()!=null
&& adjCell.getStyle().getRowSpan()>1 ) {
// eg tcb is r2,c1 & adjCell is r1,c0
adjCells.add(adjCell);
} else {
break;
}
}
if ( backwards && !adjCells.isEmpty() ) {
Collections.reverse(adjCells);
}
for (TableCellBox adjCell : adjCells) {
Tc dummy = Context.getWmlObjectFactory().createTc();
trContext.getContent().add(dummy);
TcPr tcPr = Context.getWmlObjectFactory().createTcPr();
dummy.setTcPr(tcPr);
VMerge vm = Context.getWmlObjectFactory().createTcPrInnerVMerge();
//vm.setVal("continue");
tcPr.setVMerge(vm);
int colspan = adjCell.getStyle().getColSpan();
if (colspan > 1) {
GridSpan gs = Context.getWmlObjectFactory().createTcPrInnerGridSpan();
gs.setVal( BigInteger.valueOf(colspan));
tcPr.setGridSpan(gs);
}
TcPrInner.TcBorders borders = copyCellBorderStyles(adjCell);
borders.setTop( createBorderStyle(STBorder.NIL, null, null) );
tcPr.setTcBorders(borders);
this.setCellWidthAuto(tcPr);
// Must have an empty w:p
dummy.getContent().add( new P() );
}
}
/**
* nested tables XHTML renderer seems to construct a tree: table/table
* instead of table/tr/td/table?
* TODO fix this upstream.
* TestCase is http://en.wikipedia.org/wiki/Office_Open_XML
*
* @param contentContext
* @param parent
* @return
*/
protected void nestedTableHierarchyFix(ContentAccessor contentContext,
Box parent) {
if (parent==null) return; // where importing a table fragment
if (parent instanceof TableBox
|| parent.getElement().getNodeName().equals("table") ) {
log.warn("table: Constructing missing w:tr/w:td..");
//if table was with caption move P (generated for caption) to nested column
P captionP = null;
// Iterator