make xssf streaming code more extensible
git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1836795 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
parent
ad4f5ea6d2
commit
d2d9202772
@ -90,7 +90,7 @@ public class XSSFReader {
|
|||||||
// strict OOXML likely not fully supported, see #57699
|
// strict OOXML likely not fully supported, see #57699
|
||||||
// this code is similar to POIXMLDocumentPart.getPartFromOPCPackage(), but I could not combine it
|
// this code is similar to POIXMLDocumentPart.getPartFromOPCPackage(), but I could not combine it
|
||||||
// easily due to different return values
|
// easily due to different return values
|
||||||
if(coreDocRelationship == null) {
|
if (coreDocRelationship == null) {
|
||||||
if (this.pkg.getRelationshipsByType(
|
if (this.pkg.getRelationshipsByType(
|
||||||
PackageRelationshipTypes.STRICT_CORE_DOCUMENT).getRelationship(0) != null) {
|
PackageRelationshipTypes.STRICT_CORE_DOCUMENT).getRelationship(0) != null) {
|
||||||
throw new POIXMLException("Strict OOXML isn't currently supported, please see bug #57699");
|
throw new POIXMLException("Strict OOXML isn't currently supported, please see bug #57699");
|
||||||
@ -110,7 +110,7 @@ public class XSSFReader {
|
|||||||
* shared strings.
|
* shared strings.
|
||||||
*/
|
*/
|
||||||
public SharedStringsTable getSharedStringsTable() throws IOException, InvalidFormatException {
|
public SharedStringsTable getSharedStringsTable() throws IOException, InvalidFormatException {
|
||||||
ArrayList<PackagePart> parts = pkg.getPartsByContentType( XSSFRelation.SHARED_STRINGS.getContentType());
|
ArrayList<PackagePart> parts = pkg.getPartsByContentType(XSSFRelation.SHARED_STRINGS.getContentType());
|
||||||
return parts.size() == 0 ? null : new SharedStringsTable(parts.get(0));
|
return parts.size() == 0 ? null : new SharedStringsTable(parts.get(0));
|
||||||
}
|
}
|
||||||
|
|
||||||
@ -119,13 +119,13 @@ public class XSSFReader {
|
|||||||
* returns a handy object for working with cell styles
|
* returns a handy object for working with cell styles
|
||||||
*/
|
*/
|
||||||
public StylesTable getStylesTable() throws IOException, InvalidFormatException {
|
public StylesTable getStylesTable() throws IOException, InvalidFormatException {
|
||||||
ArrayList<PackagePart> parts = pkg.getPartsByContentType( XSSFRelation.STYLES.getContentType());
|
ArrayList<PackagePart> parts = pkg.getPartsByContentType(XSSFRelation.STYLES.getContentType());
|
||||||
if(parts.size() == 0) return null;
|
if (parts.size() == 0) return null;
|
||||||
|
|
||||||
// Create the Styles Table, and associate the Themes if present
|
// Create the Styles Table, and associate the Themes if present
|
||||||
StylesTable styles = new StylesTable(parts.get(0));
|
StylesTable styles = new StylesTable(parts.get(0));
|
||||||
parts = pkg.getPartsByContentType( XSSFRelation.THEME.getContentType());
|
parts = pkg.getPartsByContentType(XSSFRelation.THEME.getContentType());
|
||||||
if(parts.size() != 0) {
|
if (parts.size() != 0) {
|
||||||
styles.setTheme(new ThemesTable(parts.get(0)));
|
styles.setTheme(new ThemesTable(parts.get(0)));
|
||||||
}
|
}
|
||||||
return styles;
|
return styles;
|
||||||
@ -168,17 +168,18 @@ public class XSSFReader {
|
|||||||
/**
|
/**
|
||||||
* Returns an InputStream to read the contents of the
|
* Returns an InputStream to read the contents of the
|
||||||
* specified Sheet.
|
* specified Sheet.
|
||||||
|
*
|
||||||
* @param relId The relationId of the sheet, from a r:id on the workbook
|
* @param relId The relationId of the sheet, from a r:id on the workbook
|
||||||
*/
|
*/
|
||||||
public InputStream getSheet(String relId) throws IOException, InvalidFormatException {
|
public InputStream getSheet(String relId) throws IOException, InvalidFormatException {
|
||||||
PackageRelationship rel = workbookPart.getRelationship(relId);
|
PackageRelationship rel = workbookPart.getRelationship(relId);
|
||||||
if(rel == null) {
|
if (rel == null) {
|
||||||
throw new IllegalArgumentException("No Sheet found with r:id " + relId);
|
throw new IllegalArgumentException("No Sheet found with r:id " + relId);
|
||||||
}
|
}
|
||||||
|
|
||||||
PackagePartName relName = PackagingURIHelper.createPartName(rel.getTargetURI());
|
PackagePartName relName = PackagingURIHelper.createPartName(rel.getTargetURI());
|
||||||
PackagePart sheet = pkg.getPart(relName);
|
PackagePart sheet = pkg.getPart(relName);
|
||||||
if(sheet == null) {
|
if (sheet == null) {
|
||||||
throw new IllegalArgumentException("No data found for Sheet with r:id " + relId);
|
throw new IllegalArgumentException("No data found for Sheet with r:id " + relId);
|
||||||
}
|
}
|
||||||
return sheet.getInputStream();
|
return sheet.getInputStream();
|
||||||
@ -232,7 +233,7 @@ public class XSSFReader {
|
|||||||
sheetMap = new HashMap<>();
|
sheetMap = new HashMap<>();
|
||||||
OPCPackage pkg = wb.getPackage();
|
OPCPackage pkg = wb.getPackage();
|
||||||
Set<String> worksheetRels = getSheetRelationships();
|
Set<String> worksheetRels = getSheetRelationships();
|
||||||
for(PackageRelationship rel : wb.getRelationships()){
|
for (PackageRelationship rel : wb.getRelationships()) {
|
||||||
String relType = rel.getRelationshipType();
|
String relType = rel.getRelationshipType();
|
||||||
if (worksheetRels.contains(relType)) {
|
if (worksheetRels.contains(relType)) {
|
||||||
PackagePartName relName = PackagingURIHelper.createPartName(rel.getTargetURI());
|
PackagePartName relName = PackagingURIHelper.createPartName(rel.getTargetURI());
|
||||||
@ -242,7 +243,7 @@ public class XSSFReader {
|
|||||||
//step 2. Read array of CTSheet elements, wrap it in a LinkedList
|
//step 2. Read array of CTSheet elements, wrap it in a LinkedList
|
||||||
//and construct an iterator
|
//and construct an iterator
|
||||||
sheetIterator = createSheetIteratorFromWB(wb);
|
sheetIterator = createSheetIteratorFromWB(wb);
|
||||||
} catch (InvalidFormatException e){
|
} catch (InvalidFormatException e) {
|
||||||
throw new POIXMLException(e);
|
throw new POIXMLException(e);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
@ -311,7 +312,7 @@ public class XSSFReader {
|
|||||||
try {
|
try {
|
||||||
PackagePart sheetPkg = sheetMap.get(sheetId);
|
PackagePart sheetPkg = sheetMap.get(sheetId);
|
||||||
return sheetPkg.getInputStream();
|
return sheetPkg.getInputStream();
|
||||||
} catch(IOException e) {
|
} catch (IOException e) {
|
||||||
throw new POIXMLException(e);
|
throw new POIXMLException(e);
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
@ -336,15 +337,14 @@ public class XSSFReader {
|
|||||||
try {
|
try {
|
||||||
PackageRelationshipCollection commentsList =
|
PackageRelationshipCollection commentsList =
|
||||||
sheetPkg.getRelationshipsByType(XSSFRelation.SHEET_COMMENTS.getRelation());
|
sheetPkg.getRelationshipsByType(XSSFRelation.SHEET_COMMENTS.getRelation());
|
||||||
if(commentsList.size() > 0) {
|
if (commentsList.size() > 0) {
|
||||||
PackageRelationship comments = commentsList.getRelationship(0);
|
PackageRelationship comments = commentsList.getRelationship(0);
|
||||||
PackagePartName commentsName = PackagingURIHelper.createPartName(comments.getTargetURI());
|
PackagePartName commentsName = PackagingURIHelper.createPartName(comments.getTargetURI());
|
||||||
PackagePart commentsPart = sheetPkg.getPackage().getPart(commentsName);
|
PackagePart commentsPart = sheetPkg.getPackage().getPart(commentsName);
|
||||||
return new CommentsTable(commentsPart);
|
return new CommentsTable(commentsPart);
|
||||||
}
|
}
|
||||||
} catch (InvalidFormatException e) {
|
} catch (InvalidFormatException|IOException e) {
|
||||||
return null;
|
LOGGER.log(POILogger.WARN, e);
|
||||||
} catch (IOException e) {
|
|
||||||
return null;
|
return null;
|
||||||
}
|
}
|
||||||
return null;
|
return null;
|
||||||
@ -356,27 +356,24 @@ public class XSSFReader {
|
|||||||
*/
|
*/
|
||||||
public List<XSSFShape> getShapes() {
|
public List<XSSFShape> getShapes() {
|
||||||
PackagePart sheetPkg = getSheetPart();
|
PackagePart sheetPkg = getSheetPart();
|
||||||
List<XSSFShape> shapes= new LinkedList<>();
|
List<XSSFShape> shapes = new LinkedList<>();
|
||||||
// Do we have a comments relationship? (Only ever one if so)
|
// Do we have a comments relationship? (Only ever one if so)
|
||||||
try {
|
try {
|
||||||
PackageRelationshipCollection drawingsList = sheetPkg.getRelationshipsByType(XSSFRelation.DRAWINGS.getRelation());
|
PackageRelationshipCollection drawingsList = sheetPkg.getRelationshipsByType(XSSFRelation.DRAWINGS.getRelation());
|
||||||
for (int i = 0; i < drawingsList.size(); i++){
|
for (int i = 0; i < drawingsList.size(); i++) {
|
||||||
PackageRelationship drawings = drawingsList.getRelationship(i);
|
PackageRelationship drawings = drawingsList.getRelationship(i);
|
||||||
PackagePartName drawingsName = PackagingURIHelper.createPartName(drawings.getTargetURI());
|
PackagePartName drawingsName = PackagingURIHelper.createPartName(drawings.getTargetURI());
|
||||||
PackagePart drawingsPart = sheetPkg.getPackage().getPart(drawingsName);
|
PackagePart drawingsPart = sheetPkg.getPackage().getPart(drawingsName);
|
||||||
if (drawingsPart == null) {
|
if (drawingsPart == null) {
|
||||||
//parts can go missing; Excel ignores them silently -- TIKA-2134
|
//parts can go missing; Excel ignores them silently -- TIKA-2134
|
||||||
LOGGER.log(POILogger.WARN, "Missing drawing: "+drawingsName +". Skipping it.");
|
LOGGER.log(POILogger.WARN, "Missing drawing: " + drawingsName + ". Skipping it.");
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
XSSFDrawing drawing = new XSSFDrawing(drawingsPart);
|
XSSFDrawing drawing = new XSSFDrawing(drawingsPart);
|
||||||
shapes.addAll(drawing.getShapes());
|
shapes.addAll(drawing.getShapes());
|
||||||
}
|
}
|
||||||
} catch (XmlException e){
|
} catch (XmlException|InvalidFormatException|IOException e) {
|
||||||
return null;
|
LOGGER.log(POILogger.WARN, e);
|
||||||
} catch (InvalidFormatException e) {
|
|
||||||
return null;
|
|
||||||
} catch (IOException e) {
|
|
||||||
return null;
|
return null;
|
||||||
}
|
}
|
||||||
return shapes;
|
return shapes;
|
||||||
|
@ -58,16 +58,16 @@ public class XSSFEventBasedExcelExtractor extends POIXMLTextExtractor
|
|||||||
|
|
||||||
private static final POILogger LOGGER = POILogFactory.getLogger(XSSFEventBasedExcelExtractor.class);
|
private static final POILogger LOGGER = POILogFactory.getLogger(XSSFEventBasedExcelExtractor.class);
|
||||||
|
|
||||||
private OPCPackage container;
|
protected OPCPackage container;
|
||||||
private POIXMLProperties properties;
|
protected POIXMLProperties properties;
|
||||||
|
|
||||||
private Locale locale;
|
protected Locale locale;
|
||||||
private boolean includeTextBoxes = true;
|
protected boolean includeTextBoxes = true;
|
||||||
private boolean includeSheetNames = true;
|
protected boolean includeSheetNames = true;
|
||||||
private boolean includeCellComments;
|
protected boolean includeCellComments;
|
||||||
private boolean includeHeadersFooters = true;
|
protected boolean includeHeadersFooters = true;
|
||||||
private boolean formulasNotResults;
|
protected boolean formulasNotResults;
|
||||||
private boolean concatenatePhoneticRuns = true;
|
protected boolean concatenatePhoneticRuns = true;
|
||||||
|
|
||||||
public XSSFEventBasedExcelExtractor(String path) throws XmlException, OpenXML4JException, IOException {
|
public XSSFEventBasedExcelExtractor(String path) throws XmlException, OpenXML4JException, IOException {
|
||||||
this(OPCPackage.open(path));
|
this(OPCPackage.open(path));
|
||||||
@ -254,7 +254,7 @@ public class XSSFEventBasedExcelExtractor extends POIXMLTextExtractor
|
|||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
protected SharedStrings createSharedStringsTable(OPCPackage container, boolean concatenatePhoneticRuns)
|
protected SharedStrings createSharedStringsTable(XSSFReader xssfReader, OPCPackage container)
|
||||||
throws IOException, SAXException {
|
throws IOException, SAXException {
|
||||||
return new ReadOnlySharedStringsTable(container, concatenatePhoneticRuns);
|
return new ReadOnlySharedStringsTable(container, concatenatePhoneticRuns);
|
||||||
}
|
}
|
||||||
@ -264,8 +264,8 @@ public class XSSFEventBasedExcelExtractor extends POIXMLTextExtractor
|
|||||||
*/
|
*/
|
||||||
public String getText() {
|
public String getText() {
|
||||||
try {
|
try {
|
||||||
SharedStrings strings = createSharedStringsTable(container, concatenatePhoneticRuns);
|
|
||||||
XSSFReader xssfReader = new XSSFReader(container);
|
XSSFReader xssfReader = new XSSFReader(container);
|
||||||
|
SharedStrings strings = createSharedStringsTable(xssfReader, container);
|
||||||
StylesTable styles = xssfReader.getStylesTable();
|
StylesTable styles = xssfReader.getStylesTable();
|
||||||
XSSFReader.SheetIterator iter = (XSSFReader.SheetIterator) xssfReader.getSheetsData();
|
XSSFReader.SheetIterator iter = (XSSFReader.SheetIterator) xssfReader.getSheetsData();
|
||||||
StringBuilder text = new StringBuilder(64);
|
StringBuilder text = new StringBuilder(64);
|
||||||
|
Loading…
Reference in New Issue
Block a user