You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@tika.apache.org by ta...@apache.org on 2022/02/10 21:56:53 UTC
[tika] branch branch_1x updated: TIKA-3678 -- upgrade to POI 5.2.0
This is an automated email from the ASF dual-hosted git repository.
tallison pushed a commit to branch branch_1x
in repository https://gitbox.apache.org/repos/asf/tika.git
The following commit(s) were added to refs/heads/branch_1x by this push:
new 0f17c4c TIKA-3678 -- upgrade to POI 5.2.0
0f17c4c is described below
commit 0f17c4c5052d2eb878631e317591cfbc799fd9c4
Author: tallison <ta...@apache.org>
AuthorDate: Thu Feb 10 16:56:37 2022 -0500
TIKA-3678 -- upgrade to POI 5.2.0
---
CHANGES.txt | 6 +
tika-bundle/pom.xml | 45 +++++++-
tika-eval/pom.xml | 5 -
tika-eval/src/main/resources/log4j2.xml | 5 +
tika-parent/pom.xml | 2 +-
.../tika/parser/microsoft/OutlookExtractor.java | 7 +-
.../tika/parser/microsoft/SummaryExtractor.java | 1 -
.../microsoft/ooxml/OOXMLExtractorFactory.java | 34 ++++--
.../microsoft/ooxml/OOXMLTikaBodyPartHandler.java | 16 ++-
.../ooxml/OOXMLWordAndPowerPointTextHandler.java | 124 +++++++++------------
.../ooxml/SXSLFPowerPointExtractorDecorator.java | 4 +-
.../ooxml/XSLFPowerPointExtractorDecorator.java | 19 +---
.../ooxml/XSSFExcelExtractorDecorator.java | 5 +-
.../microsoft/ooxml/xps/XPSTextExtractor.java | 29 ++++-
.../xslf/XSLFEventBasedPowerPointExtractor.java | 61 +++++-----
.../ooxml/xwpf/XWPFEventBasedWordExtractor.java | 118 +++++++++++---------
.../apache/tika/server/resource/TikaResource.java | 2 +-
17 files changed, 280 insertions(+), 203 deletions(-)
diff --git a/CHANGES.txt b/CHANGES.txt
index 384011e..7cab47d 100644
--- a/CHANGES.txt
+++ b/CHANGES.txt
@@ -1,3 +1,9 @@
+Release 1.28.2 - ???
+
+ * Upgrade to Apache POI 5.2.0. This is the first upgrade to POI
+ 5.x and represents a major refactoring. Users will experience
+ significantly more logging (TIKA-3678).
+
Release 1.28.1 - 2/8/2022
* Security upgrades: xerces, log4j2, junrar and protobuf
diff --git a/tika-bundle/pom.xml b/tika-bundle/pom.xml
index 617d1f1..c850756 100644
--- a/tika-bundle/pom.xml
+++ b/tika-bundle/pom.xml
@@ -57,6 +57,11 @@
<artifactId>tika-parsers</artifactId>
<version>${project.version}</version>
</dependency>
+ <dependency>
+ <groupId>org.apache.logging.log4j</groupId>
+ <artifactId>log4j-api</artifactId>
+ <version>${log4j2.version}</version>
+ </dependency>
<!-- Test dependencies -->
<dependency>
@@ -201,9 +206,12 @@
bcmail-jdk15on|
bcprov-jdk15on|
bcpkix-jdk15on|
- poi|poi-scratchpad|
+ poi|
+ poi-scratchpad|
poi-ooxml|
- poi-ooxml-schemas|
+ poi-ooxml-lite|
+ log4j-core|
+ log4j-api|
commons-math3|
curvesapi|
xmlbeans|
@@ -275,6 +283,13 @@
com.dd.plist;resolution:=optional,
com.adobe.xmp.properties;resolution:=optional,
com.github.luben.zstd;resolution:=optional,
+ com.github.javaparser;resolution:=optional,
+ com.github.javaparser.ast;resolution:=optional,
+ com.github.javaparser.ast.body;resolution:=optional,
+ com.github.javaparser.ast.expr;resolution:=optional,
+ com.github.javaparser.ast.nodeTypes;resolution:=optional,
+ com.github.javaparser.ast.type;resolution:=optional,
+ com.github.javaparser.utils;resolution:=optional,
com.github.openjson;resolution:=optional,
com.github.jaiimageio.*;resolution:=optional,
com.google.appengine.api.*;resolution:=optional,
@@ -287,11 +302,16 @@
com.parso;resolution:=optional,
com.sleepycat.je;resolution:=optional,
com.sun.javadoc;resolution:=optional,
+ com.sun.org.apache.xalan.internal;resolution:=optional,
+ com.sun.org.apache.xml.internal.resolver;resolution:=optional,
+ com.sun.org.apache.xml.internal.resolver.tools;resolution:=optional,
+ com.sun.org.apache.xpath.internal.jaxp;resolution:=optional,
com.sun.xml.bind.marshaller;resolution:=optional,
com.sun.xml.internal.bind.marshaller;resolution:=optional,
com.sun.msv.datatype;resolution:=optional,
com.sun.msv.datatype.xsd;resolution:=optional,
com.sun.tools.javadoc;resolution:=optional,
+ de.rototor.pdfbox.graphics2d;resolution:=optional,
edu.mit.ll.mitie;resolution:=optional,
edu.stanford.nlp.*;resolution:=optional,
edu.wisc.ssec.mcidas;resolution:=optional,
@@ -321,19 +341,27 @@
opendap.dap.parser;resolution:=optional,
opennlp.maxent;resolution:=optional,
opennlp.tools.namefind;resolution:=optional,
- opennlp.tools.authorage;resolution:=optional,
+ opennlp.tools.authorage;resolution:=optional,
net.didion.jwnl;resolution:=optional,
net.sf.saxon;resolution:=optional,
net.sf.saxon.dom;resolution:=optional,
+ net.sf.saxon.lib;resolution:=optional,
+ net.sf.saxon.ma.map;resolution:=optional,
net.sf.saxon.om;resolution:=optional,
net.sf.saxon.query;resolution:=optional,
net.sf.saxon.sxpath;resolution:=optional,
+ net.sf.saxon.trans;resolution:=optional,
+ net.sf.saxon.tree.wrapper;resolution:=optional,
+ net.sf.saxon.type;resolution:=optional,
net.sf.saxon.value;resolution:=optional,
org.apache.batik.anim.dom;resolution:=optional,
org.apache.batik.bridge;resolution:=optional,
+ org.apache.batik.dom;resolution:=optional,
org.apache.batik.ext.awt;resolution:=optional,
org.apache.batik.ext.awt.image.renderable;resolution:=optional,
org.apache.batik.gvt;resolution:=optional,
+ org.apache.batik.parser;resolution:=optional,
+ org.apache.batik.svggen;resolution:=optional,
org.apache.batik.util;resolution:=optional,
org.apache.cxf.jaxrs.client;resolution:=optional,
org.apache.cxf.jaxrs.ext.multipart;resolution:=optional,
@@ -350,6 +378,14 @@
org.apache.commons.vfs2.util;resolution:=optional,
org.apache.crimson.jaxp;resolution:=optional,
org.apache.jcp.xml.dsig.internal.dom;resolution:=optional,
+ org.apache.logging.log4j;resolution:=optional,
+ org.apache.logging.log4j.util;resolution:=optional,
+ org.apache.logging.log4j.util.internal;resolution:=optional,
+ org.apache.logging.log4j.spi;resolution:=optional,
+ org.apache.maven.model;resolution:=optional,
+ org.apache.maven.plugin;resolution:=optional,
+ org.apache.maven.plugin.logging;resolution:=optional,
+ org.apache.maven.project;resolution:=optional,
org.apache.pdfbox.debugger;resolution:=optional,
org.apache.pdfbox.preflight.*;resolution:=optional,
org.apache.sis;resolution:=optional,
@@ -449,6 +485,9 @@
org.slf4j.helpers;resolution:=optional,
org.sqlite;resolution:=optional,
org.w3c.dom;resolution:=optional,
+ org.w3c.dom.ranges;resolution:=optional,
+ org.w3c.dom.svg;resolution:=optional,
+ org.w3c.dom.traversal;resolution:=optional,
org.relaxng.datatype;resolution:=optional,
org.xml.sax;resolution:=optional,
org.xml.sax.ext;resolution:=optional,
diff --git a/tika-eval/pom.xml b/tika-eval/pom.xml
index bd09d5f..9aef970 100644
--- a/tika-eval/pom.xml
+++ b/tika-eval/pom.xml
@@ -127,11 +127,6 @@
</dependency>
<dependency>
<groupId>org.apache.poi</groupId>
- <artifactId>poi-ooxml-schemas</artifactId>
- <version>${poi.version}</version>
- </dependency>
- <dependency>
- <groupId>org.apache.poi</groupId>
<artifactId>poi-scratchpad</artifactId>
<version>${poi.version}</version>
</dependency>
diff --git a/tika-eval/src/main/resources/log4j2.xml b/tika-eval/src/main/resources/log4j2.xml
index c88e66e..a4bfda9 100644
--- a/tika-eval/src/main/resources/log4j2.xml
+++ b/tika-eval/src/main/resources/log4j2.xml
@@ -28,5 +28,10 @@
<Root level="info">
<AppenderRef ref="Console"/>
</Root>
+ <!-- turn this off, we know richtext isn't supported -->
+ <Logger name="org.apache.poi.xssf.streaming.SXSSFCreationHelper" level="error">
+ <AppenderRef ref="Console"/>
+ </Logger>
+
</Loggers>
</Configuration>
\ No newline at end of file
diff --git a/tika-parent/pom.xml b/tika-parent/pom.xml
index ba5b5c3..30ab125 100644
--- a/tika-parent/pom.xml
+++ b/tika-parent/pom.xml
@@ -272,7 +272,7 @@
<maven.shade.version>3.2.4</maven.shade.version>
<rat.version>0.13</rat.version>
<!-- NOTE: sync tukaani version with commons-compress in tika-parsers -->
- <poi.version>4.1.2</poi.version>
+ <poi.version>5.2.0</poi.version>
<commons.compress.version>1.21</commons.compress.version>
<commons.io.version>2.11.0</commons.io.version>
<commons.lang3.version>3.12.0</commons.lang3.version>
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/OutlookExtractor.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/OutlookExtractor.java
index c2e27d6..cf6f51d 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/OutlookExtractor.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/OutlookExtractor.java
@@ -700,8 +700,11 @@ public class OutlookExtractor extends AbstractPOIFSExtractor {
for (RecipientChunks chunks : recipientChunks) {
Recipient r = new Recipient();
- r.displayName = (chunks.recipientDisplayNameChunk != null) ? chunks.recipientDisplayNameChunk.toString() : null;
- r.name = (chunks.recipientNameChunk != null) ? chunks.recipientNameChunk.toString() : null;
+ r.displayName = (chunks.getRecipientDisplayNameChunk() != null) ?
+ chunks.getRecipientDisplayNameChunk().toString() : null;
+ r.name = (chunks.getRecipientNameChunk() != null) ?
+ chunks.getRecipientNameChunk().toString()
+ : null;
r.emailAddress = chunks.getRecipientEmailAddress();
List<PropertyValue> vals = chunks.getProperties().get(MAPIProperty.RECIPIENT_TYPE);
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/SummaryExtractor.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/SummaryExtractor.java
index ba98c0e..30c472d 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/SummaryExtractor.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/SummaryExtractor.java
@@ -24,7 +24,6 @@ import java.util.Set;
import org.apache.poi.hpsf.CustomProperties;
import org.apache.poi.hpsf.DocumentSummaryInformation;
-import org.apache.poi.hpsf.MarkUnsupportedException;
import org.apache.poi.hpsf.NoPropertySetStreamException;
import org.apache.poi.hpsf.PropertySet;
import org.apache.poi.hpsf.SummaryInformation;
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLExtractorFactory.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLExtractorFactory.java
index 63dd79f..34aeae1 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLExtractorFactory.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLExtractorFactory.java
@@ -24,8 +24,9 @@ import java.util.Locale;
import org.apache.commons.compress.archivers.zip.UnsupportedZipFeatureException;
import org.apache.commons.io.input.CloseShieldInputStream;
+import org.apache.poi.extractor.ExtractorFactory;
import org.apache.poi.ooxml.POIXMLDocument;
-import org.apache.poi.ooxml.extractor.ExtractorFactory;
+import org.apache.poi.ooxml.extractor.POIXMLExtractorFactory;
import org.apache.poi.ooxml.extractor.POIXMLTextExtractor;
import org.apache.poi.openxml4j.exceptions.InvalidFormatException;
import org.apache.poi.openxml4j.exceptions.InvalidOperationException;
@@ -35,7 +36,7 @@ import org.apache.poi.openxml4j.opc.PackageAccess;
import org.apache.poi.openxml4j.opc.PackagePart;
import org.apache.poi.openxml4j.opc.PackageRelationshipCollection;
import org.apache.poi.util.LocaleUtil;
-import org.apache.poi.xslf.extractor.XSLFPowerPointExtractor;
+import org.apache.poi.xslf.extractor.XSLFExtractor;
import org.apache.poi.xslf.usermodel.XMLSlideShow;
import org.apache.poi.xslf.usermodel.XSLFRelation;
import org.apache.poi.xslf.usermodel.XSLFSlideShow;
@@ -75,13 +76,24 @@ public class OOXMLExtractorFactory {
private static final Logger LOG = LoggerFactory.getLogger(OOXMLExtractorFactory.class);
private static final int MAX_BUFFER_LENGTH = 1000000;
+ private static POIXMLExtractorFactory EXTRACTOR_FACTORY = new POIXMLExtractorFactory();
+
+ //TODO find what happened to SUPPORTED_TYPES
+ private static XSLFRelation[] XSLF_RELATIONS = new XSLFRelation[] {
+ XSLFRelation.MAIN, XSLFRelation.MACRO, XSLFRelation.MACRO_TEMPLATE,
+ XSLFRelation.PRESENTATIONML,
+ XSLFRelation.PRESENTATIONML_TEMPLATE, XSLFRelation.PRESENTATION_MACRO
+ };
+
+ static {
+ ExtractorFactory.setAllThreadsPreferEventExtractors(true);
+ }
public static void parse(
InputStream stream, ContentHandler baseHandler,
Metadata metadata, ParseContext context)
throws IOException, SAXException, TikaException {
Locale locale = context.get(Locale.class, LocaleUtil.getUserLocale());
- ExtractorFactory.setThreadPrefersEventExtractors(true);
//if there's a problem opening the zip file;
//create a tmp file, and copy what you can read of it.
@@ -190,7 +202,7 @@ public class OOXMLExtractorFactory {
}
if (poiExtractor == null) {
- poiExtractor = (POIXMLTextExtractor) ExtractorFactory.createExtractor(pkg);
+ poiExtractor = EXTRACTOR_FACTORY.create(pkg);
}
POIXMLDocument document = poiExtractor.getDocument();
@@ -215,8 +227,8 @@ public class OOXMLExtractorFactory {
"The extractor returned was a " + poiExtractor
);
} else if (document instanceof XMLSlideShow) {
- extractor = new XSLFPowerPointExtractorDecorator(
- context, (org.apache.poi.xslf.extractor.XSLFPowerPointExtractor) poiExtractor);
+ extractor = new XSLFPowerPointExtractorDecorator(metadata, context,
+ (org.apache.poi.xslf.extractor.XSLFExtractor) poiExtractor);
} else if (document instanceof XWPFDocument) {
extractor = new XWPFWordExtractorDecorator( metadata,
context, (XWPFWordExtractor) poiExtractor);
@@ -304,15 +316,13 @@ public class OOXMLExtractorFactory {
}
String targetContentType = corePart.getContentType();
- XSLFRelation[] xslfRelations = org.apache.poi.xslf.extractor.XSLFPowerPointExtractor.SUPPORTED_TYPES;
-
- for (int i = 0; i < xslfRelations.length; i++) {
- XSLFRelation xslfRelation = xslfRelations[i];
+ for (int i = 0; i < XSLF_RELATIONS.length; i++) {
+ XSLFRelation xslfRelation = XSLF_RELATIONS[i];
if (xslfRelation.getContentType().equals(targetContentType)) {
if (eventBased) {
return new XSLFEventBasedPowerPointExtractor(pkg);
} else {
- return new XSLFPowerPointExtractor(new XSLFSlideShow(pkg));
+ return new XSLFExtractor(new XMLSlideShow(pkg));
}
}
}
@@ -321,7 +331,7 @@ public class OOXMLExtractorFactory {
if (eventBased) {
return new XSLFEventBasedPowerPointExtractor(pkg);
} else {
- return new XSLFPowerPointExtractor(new XSLFSlideShow(pkg));
+ return new XSLFExtractor(new XMLSlideShow(pkg));
}
}
return null;
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLTikaBodyPartHandler.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLTikaBodyPartHandler.java
index 6778280..f336435 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLTikaBodyPartHandler.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLTikaBodyPartHandler.java
@@ -33,9 +33,9 @@ import org.apache.tika.sax.XHTMLContentHandler;
public class OOXMLTikaBodyPartHandler
implements OOXMLWordAndPowerPointTextHandler.XWPFBodyContentsHandler {
- private final static String P = "p";
+ private static final String P = "p";
- private final static char[] NEWLINE = new char[]{'\n'};
+ private static final char[] NEWLINE = new char[]{'\n'};
private final XHTMLContentHandler xhtml;
private final XWPFListManager listManager;
@@ -83,6 +83,7 @@ public class OOXMLTikaBodyPartHandler
@Override
public void run(RunProperties runProperties, String contents) throws SAXException {
+
// True if we are currently in the named style tag:
if (runProperties.isBold() != isBold) {
if (isStrikeThrough) {
@@ -92,7 +93,6 @@ public class OOXMLTikaBodyPartHandler
if (isUnderline) {
xhtml.endElement("u");
isUnderline = false;
- ;
}
if (isItalics) {
xhtml.endElement("i");
@@ -214,6 +214,7 @@ public class OOXMLTikaBodyPartHandler
} else if (tableCellDepth == 0) {
xhtml.characters(NEWLINE, 0, 1);
}
+
if (tableCellDepth > 0) {
pWithinCell++;
}
@@ -222,14 +223,18 @@ public class OOXMLTikaBodyPartHandler
@Override
public void startTable() throws SAXException {
+
xhtml.startElement("table");
tableDepth++;
+
}
@Override
public void endTable() throws SAXException {
+
xhtml.endElement("table");
tableDepth--;
+
}
@Override
@@ -278,7 +283,7 @@ public class OOXMLTikaBodyPartHandler
}
@Override
- public boolean getIncludeDeletedText() {
+ public boolean isIncludeDeletedText() {
return includeDeletedText;
}
@@ -301,7 +306,7 @@ public class OOXMLTikaBodyPartHandler
}
@Override
- public boolean getIncludeMoveFromText() {
+ public boolean isIncludeMoveFromText() {
return includeMoveFromText;
}
@@ -331,6 +336,7 @@ public class OOXMLTikaBodyPartHandler
xhtml.startElement("img", attr);
xhtml.endElement("img");
+
}
@Override
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLWordAndPowerPointTextHandler.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLWordAndPowerPointTextHandler.java
index 2cd4e31..77d0887 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLWordAndPowerPointTextHandler.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/OOXMLWordAndPowerPointTextHandler.java
@@ -22,43 +22,36 @@ import java.util.Date;
import java.util.Map;
import org.apache.poi.xwpf.usermodel.UnderlinePatterns;
-import org.apache.tika.parser.microsoft.OfficeParserConfig;
-import org.apache.tika.utils.DateUtils;
import org.xml.sax.Attributes;
import org.xml.sax.SAXException;
import org.xml.sax.helpers.DefaultHandler;
+import org.apache.tika.utils.DateUtils;
+
/**
* This class is intended to handle anything that might contain IBodyElements:
* main document, headers, footers, notes, slides, etc.
- *
+ * <p>
* <p/>
- *
+ * <p>
* This class does not generally check for namespaces, and it can be applied
* to PPTX and DOCX for text extraction.
- *
+ * <p>
* <p/>
* This can be used to scrape content from charts. It currently ignores
* formula (<c:f/>) elements
- *
+ * <p>
* <p/>
* This does not work with .xlsx or .vsdx.
- *
+ * <p>
* TODO: move this into POI?
- *
*/
public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
- public enum EditType {
- NONE,
- INSERT,
- DELETE,
- MOVE_TO,
- MOVE_FROM
- }
-
+ public final static String W_NS =
+ "http://schemas.openxmlformats.org/wordprocessingml/2006/main";
private final static String R = "r";
private final static String FLD = "fld";
private final static String RPR = "rPr";
@@ -92,21 +85,18 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
private final static String RUBY = "ruby"; //phonetic section
private final static String RT = "rt"; //phonetic run
private static final String VAL = "val";
-
-
- public final static String W_NS = "http://schemas.openxmlformats.org/wordprocessingml/2006/main";
- private final static String MC_NS = "http://schemas.openxmlformats.org/markup-compatibility/2006";
+ private final static String MC_NS =
+ "http://schemas.openxmlformats.org/markup-compatibility/2006";
private final static String O_NS = "urn:schemas-microsoft-com:office:office";
private final static String PIC_NS = "http://schemas.openxmlformats.org/drawingml/2006/picture";
- private final static String DRAWING_MAIN_NS = "http://schemas.openxmlformats.org/drawingml/2006/main";
+ private final static String DRAWING_MAIN_NS =
+ "http://schemas.openxmlformats.org/drawingml/2006/main";
private final static String V_NS = "urn:schemas-microsoft-com:vml";
private final static String C_NS = "http://schemas.openxmlformats.org/drawingml/2006/chart";
-
- private final static String OFFICE_DOC_RELATIONSHIP_NS = "http://schemas.openxmlformats.org/officeDocument/2006/relationships";
-
+ private final static String OFFICE_DOC_RELATIONSHIP_NS =
+ "http://schemas.openxmlformats.org/officeDocument/2006/relationships";
private final static char[] TAB_CHAR = new char[]{'\t'};
private final static char NEWLINE = '\n';
-
private final static String BOOKMARK_START = "bookmarkStart";
private final static String BOOKMARK_END = "bookmarkEnd";
private final static String FOOTNOTE_REFERENCE = "footnoteReference";
@@ -117,24 +107,25 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
private final static String MOVE_TO = "moveTo";
private final static String ENDNOTE_REFERENCE = "endnoteReference";
private static final String TEXTBOX = "textbox";
-
-
private final XWPFBodyContentsHandler bodyContentsHandler;
-
private final Map<String, String> linkedRelationships;
-
- private boolean inR = false;//in run or in field. TODO: convert this to an integer because you can have a run within a run
+ private final RunProperties currRunProperties = new RunProperties();
+ private final ParagraphProperties currPProperties = new ParagraphProperties();
+ private final boolean includeTextBox;
+ private final boolean concatenatePhoneticRuns;
+ private final StringBuilder runBuffer = new StringBuilder();
+ private final StringBuilder rubyBuffer = new StringBuilder();
+ private boolean inR = false;
+ //in run or in field. TODO: convert this to an integer because you can have a run within a run
private boolean inT = false;
private boolean inRPr = false;
private boolean inNumPr = false;
private boolean inRt = false;
-
private boolean inPic = false;
private boolean inPict = false;
private String picDescription = null;
private String picRId = null;
private String picFilename = null;
-
//mechanism used to determine when to
//signal the start of the p, and still
//handle p with pPr and those without
@@ -143,28 +134,18 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
//pPr can happen multiple times within a p
//<p><pPr/><r><t>text</t></r><pPr></p>
private boolean pStarted = false;
-
//alternate content can be embedded in itself.
//need to track depth.
//if in alternate, choose fallback, maybe make this configurable?
private int inACChoiceDepth = 0;
private int inACFallbackDepth = 0;
-
- private final RunProperties currRunProperties = new RunProperties();
- private final ParagraphProperties currPProperties = new ParagraphProperties();
- private final boolean includeTextBox;
- private final boolean concatenatePhoneticRuns;
- private final StringBuilder runBuffer = new StringBuilder();
- private final StringBuilder rubyBuffer = new StringBuilder();//buffers rt in ruby sections (see 17.3.3.25)
-
-
private boolean inDelText = false;
+ //buffers rt in ruby sections (see 17.3.3.25)
private boolean inHlinkClick = false;
private boolean inTextBox = false;
private boolean inV = false; //in c:v in chart file
-
- private OOXMLWordAndPowerPointTextHandler.EditType editType = OOXMLWordAndPowerPointTextHandler.EditType.NONE;
-
+ private OOXMLWordAndPowerPointTextHandler.EditType editType =
+ OOXMLWordAndPowerPointTextHandler.EditType.NONE;
private DateUtils dateUtils = new DateUtils();
public OOXMLWordAndPowerPointTextHandler(XWPFBodyContentsHandler bodyContentsHandler,
@@ -172,16 +153,15 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
this(bodyContentsHandler, hyperlinks, true, true);
}
-
public OOXMLWordAndPowerPointTextHandler(XWPFBodyContentsHandler bodyContentsHandler,
- Map<String, String> hyperlinks, boolean includeTextBox, boolean concatenatePhoneticRuns) {
+ Map<String, String> hyperlinks, boolean includeTextBox,
+ boolean concatenatePhoneticRuns) {
this.bodyContentsHandler = bodyContentsHandler;
this.linkedRelationships = hyperlinks;
this.includeTextBox = includeTextBox;
this.concatenatePhoneticRuns = concatenatePhoneticRuns;
}
-
@Override
public void startDocument() throws SAXException {
}
@@ -199,10 +179,11 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
}
@Override
- public void startElement(String uri, String localName, String qName, Attributes atts) throws SAXException {
+ public void startElement(String uri, String localName, String qName, Attributes atts)
+ throws SAXException {
//TODO: checkBox, textBox, sym, headerReference, footerReference, commentRangeEnd
- if (lastStartElementWasP && ! PPR.equals(localName)) {
+ if (lastStartElementWasP && !PPR.equals(localName)) {
bodyContentsHandler.startParagraph(currPProperties);
}
@@ -220,7 +201,7 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
return;
}
- if (! includeTextBox && localName.equals(TEXTBOX)) {
+ if (!includeTextBox && localName.equals(TEXTBOX)) {
inTextBox = true;
return;
}
@@ -238,7 +219,7 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
} else if (P.equals(localName)) {
lastStartElementWasP = true;
} else if (B.equals(localName)) { //TODO: add bCs
- if(inR && inRPr) {
+ if (inR && inRPr) {
currRunProperties.setBold(true);
}
} else if (TC.equals(localName)) {
@@ -271,7 +252,7 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
if (inNumPr) {
currPProperties.setNumId(getIntVal(atts));
}
- } else if(BR.equals(localName)) {
+ } else if (BR.equals(localName)) {
runBuffer.append(NEWLINE);
} else if (BOOKMARK_START.equals(localName)) {
String name = atts.getValue(W_NS, "name");
@@ -301,7 +282,7 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
bodyContentsHandler.hyperlinkStart(hyperlink);
inHlinkClick = true;
}
- } else if(TBL.equals(localName)) {
+ } else if (TBL.equals(localName)) {
bodyContentsHandler.startTable();
} else if (BLIP.equals(localName)) { //check for DRAWING_NS
picRId = atts.getValue(OFFICE_DOC_RELATIONSHIP_NS, "embed");
@@ -326,7 +307,7 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
startEditedSection(EditType.MOVE_TO, atts);
} else if (MOVE_FROM.equals(localName)) {
startEditedSection(editType.MOVE_FROM, atts);
- } else if (OLE_OBJECT.equals(localName)){ //check for O_NS?
+ } else if (OLE_OBJECT.equals(localName)) { //check for O_NS?
String type = null;
String refId = null;
//TODO: clean this up and ...want to get ProgID?
@@ -335,14 +316,15 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
String attValue = atts.getValue(i);
if (attLocalName.equals("Type")) {
type = attValue;
- } else if (OFFICE_DOC_RELATIONSHIP_NS.equals(atts.getURI(i)) && attLocalName.equals("id")) {
+ } else if (OFFICE_DOC_RELATIONSHIP_NS.equals(atts.getURI(i)) &&
+ attLocalName.equals("id")) {
refId = attValue;
}
}
if ("Embed".equals(type)) {
bodyContentsHandler.embeddedOLERef(refId);
}
- } else if(CR.equals(localName)) {
+ } else if (CR.equals(localName)) {
runBuffer.append(NEWLINE);
} else if (ENDNOTE_REFERENCE.equals(localName)) {
String id = atts.getValue(W_NS, "id");
@@ -386,7 +368,6 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
return -1;
}
-
@Override
public void endElement(String uri, String localName, String qName) throws SAXException {
@@ -399,7 +380,7 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
return;
}
- if (! includeTextBox && localName.equals(TEXTBOX)) {
+ if (!includeTextBox && localName.equals(TEXTBOX)) {
inTextBox = false;
return;
}
@@ -438,8 +419,8 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
handleEndOfRun();
} else if (DEL_TEXT.equals(localName)) {
inDelText = false;
- } else if (INS.equals(localName) || DEL.equals(localName) ||
- MOVE_TO.equals(localName) || MOVE_FROM.equals(localName)) {
+ } else if (INS.equals(localName) || DEL.equals(localName) || MOVE_TO.equals(localName) ||
+ MOVE_FROM.equals(localName)) {
editType = EditType.NONE;
} else if (HYPERLINK.equals(localName)) {
bodyContentsHandler.hyperlinkEnd();
@@ -447,7 +428,6 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
handlePict();
} else if (V.equals(localName) && C_NS.equals(uri)) { // in value in a chart
inV = false;
- appendToBuffer(TAB_CHAR, 0, 1);
handleEndOfRun();
} else if (RT.equals(localName)) {
inRt = false;
@@ -492,22 +472,24 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
@Override
public void characters(char[] ch, int start, int length) throws SAXException {
+
if (inACChoiceDepth > 0) {
return;
- } else if (! includeTextBox && inTextBox) {
+ } else if (!includeTextBox && inTextBox) {
return;
}
if (editType.equals(EditType.MOVE_FROM) && inT) {
- if (bodyContentsHandler.getIncludeMoveFromText()) {
+ if (bodyContentsHandler.isIncludeMoveFromText()) {
appendToBuffer(ch, start, length);
}
} else if (inT) {
appendToBuffer(ch, start, length);
- } else if (bodyContentsHandler.getIncludeDeletedText() && editType.equals(EditType.DELETE)) {
+ } else if (bodyContentsHandler.isIncludeDeletedText() && editType.equals(EditType.DELETE)) {
appendToBuffer(ch, start, length);
} else if (inV) {
appendToBuffer(ch, start, length);
+ appendToBuffer(TAB_CHAR, 0, 1);
}
}
@@ -515,13 +497,13 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
public void ignorableWhitespace(char[] ch, int start, int length) throws SAXException {
if (inACChoiceDepth > 0) {
return;
- } else if (! includeTextBox && inTextBox) {
+ } else if (!includeTextBox && inTextBox) {
return;
}
if (inT) {
appendToBuffer(ch, start, length);
- } else if (bodyContentsHandler.getIncludeDeletedText() && inDelText) {
+ } else if (bodyContentsHandler.isIncludeDeletedText() && inDelText) {
appendToBuffer(ch, start, length);
}
}
@@ -534,6 +516,10 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
}
}
+ public enum EditType {
+ NONE, INSERT, DELETE, MOVE_TO, MOVE_FROM
+ }
+
public interface XWPFBodyContentsHandler {
void run(RunProperties runProperties, String contents) throws SAXException;
@@ -569,13 +555,13 @@ public class OOXMLWordAndPowerPointTextHandler extends DefaultHandler {
void endEditedSection() throws SAXException;
- boolean getIncludeDeletedText();
+ boolean isIncludeDeletedText() throws SAXException;
void footnoteReference(String id) throws SAXException;
void endnoteReference(String id) throws SAXException;
- boolean getIncludeMoveFromText();
+ boolean isIncludeMoveFromText() throws SAXException;
void embeddedOLERef(String refId) throws SAXException;
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/SXSLFPowerPointExtractorDecorator.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/SXSLFPowerPointExtractorDecorator.java
index ac6e278..5350f30 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/SXSLFPowerPointExtractorDecorator.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/SXSLFPowerPointExtractorDecorator.java
@@ -33,7 +33,7 @@ import org.apache.poi.openxml4j.opc.PackageRelationship;
import org.apache.poi.openxml4j.opc.PackageRelationshipCollection;
import org.apache.poi.openxml4j.opc.PackagingURIHelper;
import org.apache.poi.openxml4j.opc.TargetMode;
-import org.apache.poi.xslf.extractor.XSLFPowerPointExtractor;
+import org.apache.poi.xslf.extractor.XSLFExtractor;
import org.apache.poi.xslf.usermodel.XSLFRelation;
import org.apache.tika.exception.TikaException;
import org.apache.tika.metadata.Metadata;
@@ -95,7 +95,7 @@ public class SXSLFPowerPointExtractorDecorator extends AbstractOOXMLExtractor {
}
/**
- * @see XSLFPowerPointExtractor#getText()
+ * @see XSLFExtractor#getText()
*/
protected void buildXHTML(XHTMLContentHandler xhtml) throws SAXException, IOException {
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/XSLFPowerPointExtractorDecorator.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/XSLFPowerPointExtractorDecorator.java
index c63fcb3..6a496b4 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/XSLFPowerPointExtractorDecorator.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/XSLFPowerPointExtractorDecorator.java
@@ -30,13 +30,11 @@ import org.apache.poi.openxml4j.opc.PackageRelationship;
import org.apache.poi.openxml4j.opc.PackageRelationshipCollection;
import org.apache.poi.openxml4j.opc.PackagingURIHelper;
import org.apache.poi.openxml4j.opc.TargetMode;
-import org.apache.poi.sl.extractor.SlideShowExtractor;
import org.apache.poi.sl.usermodel.Placeholder;
-import org.apache.poi.xslf.extractor.XSLFPowerPointExtractor;
+import org.apache.poi.xslf.extractor.XSLFExtractor;
import org.apache.poi.xslf.usermodel.XMLSlideShow;
import org.apache.poi.xslf.usermodel.XSLFComment;
import org.apache.poi.xslf.usermodel.XSLFCommentAuthors;
-import org.apache.poi.xslf.usermodel.XSLFComments;
import org.apache.poi.xslf.usermodel.XSLFGraphicFrame;
import org.apache.poi.xslf.usermodel.XSLFGroupShape;
import org.apache.poi.xslf.usermodel.XSLFHyperlink;
@@ -73,23 +71,14 @@ public class XSLFPowerPointExtractorDecorator extends AbstractOOXMLExtractor {
private Metadata metadata;
- public XSLFPowerPointExtractorDecorator(Metadata metadata, ParseContext context, XSLFPowerPointExtractor extractor) {
+ public XSLFPowerPointExtractorDecorator(Metadata metadata, ParseContext context,
+ XSLFExtractor extractor) {
super(context, extractor);
this.metadata = metadata;
}
/**
- * use {@link XSLFPowerPointExtractorDecorator#XSLFPowerPointExtractorDecorator(Metadata, ParseContext, XSLFPowerPointExtractor)}
- * @param context
- * @param extractor
- */
- @Deprecated
- public XSLFPowerPointExtractorDecorator(ParseContext context, XSLFPowerPointExtractor extractor) {
- this(new Metadata(),context, extractor);
- }
-
- /**
- * @see org.apache.poi.xslf.extractor.XSLFPowerPointExtractor#getText()
+ * @see XSLFExtractor#getText()
*/
protected void buildXHTML(XHTMLContentHandler xhtml) throws SAXException, IOException {
XMLSlideShow slideShow = (XMLSlideShow) extractor.getDocument();
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/XSSFExcelExtractorDecorator.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/XSSFExcelExtractorDecorator.java
index 64ff7b1..f3b8de4 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/XSSFExcelExtractorDecorator.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/XSSFExcelExtractorDecorator.java
@@ -46,6 +46,7 @@ import org.apache.poi.xssf.eventusermodel.XSSFReader;
import org.apache.poi.xssf.eventusermodel.XSSFSheetXMLHandler;
import org.apache.poi.xssf.eventusermodel.XSSFSheetXMLHandler.SheetContentsHandler;
import org.apache.poi.xssf.extractor.XSSFEventBasedExcelExtractor;
+import org.apache.poi.xssf.model.Comments;
import org.apache.poi.xssf.model.CommentsTable;
import org.apache.poi.xssf.model.StylesTable;
import org.apache.poi.xssf.usermodel.XSSFComment;
@@ -161,7 +162,7 @@ public class XSSFExcelExtractorDecorator extends AbstractOOXMLExtractor {
addDrawingHyperLinks(sheetPart);
sheetParts.add(sheetPart);
- CommentsTable comments = iter.getSheetComments();
+ Comments comments = iter.getSheetComments();
// Start, and output the sheet name
xhtml.startElement("div");
@@ -346,7 +347,7 @@ public class XSSFExcelExtractorDecorator extends AbstractOOXMLExtractor {
public void processSheet(
SheetContentsHandler sheetContentsExtractor,
- CommentsTable comments,
+ Comments comments,
StylesTable styles,
ReadOnlySharedStringsTable strings,
InputStream sheetInputStream)
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xps/XPSTextExtractor.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xps/XPSTextExtractor.java
index 0212920..297290b 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xps/XPSTextExtractor.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xps/XPSTextExtractor.java
@@ -18,6 +18,9 @@
package org.apache.tika.parser.microsoft.ooxml.xps;
+import java.io.Closeable;
+import java.io.IOException;
+
import org.apache.poi.ooxml.POIXMLDocument;
import org.apache.poi.ooxml.POIXMLProperties;
import org.apache.poi.ooxml.extractor.POIXMLTextExtractor;
@@ -25,20 +28,17 @@ import org.apache.poi.openxml4j.exceptions.OpenXML4JException;
import org.apache.poi.openxml4j.opc.OPCPackage;
import org.apache.xmlbeans.XmlException;
-import java.io.IOException;
-
/**
* Currently, mostly a pass-through class to hold pkg and properties
* and keep the general framework similar to our other POI-integrated
* extractors.
*/
-public class XPSTextExtractor extends POIXMLTextExtractor {
+public class XPSTextExtractor implements POIXMLTextExtractor {
private final OPCPackage pkg;
private final POIXMLProperties properties;
public XPSTextExtractor(OPCPackage pkg) throws OpenXML4JException, XmlException, IOException {
- super((POIXMLDocument)null);
this.pkg = pkg;
this.properties = new POIXMLProperties(pkg);
@@ -53,6 +53,22 @@ public class XPSTextExtractor extends POIXMLTextExtractor {
public String getText() {
return null;
}
+
+ @Override
+ public void setCloseFilesystem(boolean b) {
+
+ }
+
+ @Override
+ public boolean isCloseFilesystem() {
+ return false;
+ }
+
+ @Override
+ public Closeable getFilesystem() {
+ return null;
+ }
+
public POIXMLProperties.CoreProperties getCoreProperties() {
return this.properties.getCoreProperties();
}
@@ -64,4 +80,9 @@ public class XPSTextExtractor extends POIXMLTextExtractor {
public POIXMLProperties.CustomProperties getCustomProperties() {
return this.properties.getCustomProperties();
}
+
+ @Override
+ public POIXMLDocument getDocument() {
+ return null;
+ }
}
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xslf/XSLFEventBasedPowerPointExtractor.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xslf/XSLFEventBasedPowerPointExtractor.java
index bd5615d..46ada51 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xslf/XSLFEventBasedPowerPointExtractor.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xslf/XSLFEventBasedPowerPointExtractor.java
@@ -17,6 +17,7 @@
package org.apache.tika.parser.microsoft.ooxml.xslf;
+import java.io.Closeable;
import java.io.IOException;
import java.util.Date;
@@ -25,41 +26,24 @@ import org.apache.poi.ooxml.POIXMLProperties;
import org.apache.poi.ooxml.extractor.POIXMLTextExtractor;
import org.apache.poi.openxml4j.exceptions.OpenXML4JException;
import org.apache.poi.openxml4j.opc.OPCPackage;
+import org.apache.xmlbeans.XmlException;
+
import org.apache.tika.parser.microsoft.ooxml.OOXMLWordAndPowerPointTextHandler;
import org.apache.tika.parser.microsoft.ooxml.ParagraphProperties;
import org.apache.tika.parser.microsoft.ooxml.RunProperties;
-import org.apache.tika.parser.microsoft.ooxml.xwpf.XWPFEventBasedWordExtractor;
-import org.apache.xmlbeans.XmlException;
-public class XSLFEventBasedPowerPointExtractor extends POIXMLTextExtractor {
+public class XSLFEventBasedPowerPointExtractor implements POIXMLTextExtractor {
private OPCPackage container;
private POIXMLProperties properties;
- public XSLFEventBasedPowerPointExtractor(String path) throws XmlException, OpenXML4JException, IOException {
- this(OPCPackage.open(path));
- }
-
- public XSLFEventBasedPowerPointExtractor(OPCPackage container) throws XmlException, OpenXML4JException, IOException {
- super((POIXMLDocument) null);
+ public XSLFEventBasedPowerPointExtractor(OPCPackage container)
+ throws XmlException, OpenXML4JException, IOException {
this.container = container;
this.properties = new POIXMLProperties(container);
}
-
- public static void main(String[] args) throws Exception {
- if (args.length < 1) {
- System.err.println("Use:");
- System.err.println(" XSLFEventBasedPowerPointExtractor <filename.pptx>");
- System.exit(1);
- }
-
- XWPFEventBasedWordExtractor extractor = new XWPFEventBasedWordExtractor(args[0]);
- System.out.println(extractor.getText());
- extractor.close();
- }
-
public OPCPackage getPackage() {
return this.container;
}
@@ -76,6 +60,11 @@ public class XSLFEventBasedPowerPointExtractor extends POIXMLTextExtractor {
return this.properties.getCustomProperties();
}
+ @Override
+ public POIXMLDocument getDocument() {
+ return null;
+ }
+
@Override
public String getText() {
@@ -83,9 +72,28 @@ public class XSLFEventBasedPowerPointExtractor extends POIXMLTextExtractor {
return "";
}
+ @Override
+ public void setCloseFilesystem(boolean b) {
+
+ }
+
+ @Override
+ public boolean isCloseFilesystem() {
+ return false;
+ }
+
+ @Override
+ public Closeable getFilesystem() {
+ return null;
+ }
+ @Override
+ public void close() throws IOException {
+ getPackage().revert();
+ }
- private class XSLFToTextContentHandler implements OOXMLWordAndPowerPointTextHandler.XWPFBodyContentsHandler {
+ private static class XSLFToTextContentHandler
+ implements OOXMLWordAndPowerPointTextHandler.XWPFBodyContentsHandler {
private final StringBuilder buffer;
public XSLFToTextContentHandler(StringBuilder buffer) {
@@ -158,7 +166,8 @@ public class XSLFEventBasedPowerPointExtractor extends POIXMLTextExtractor {
}
@Override
- public void startEditedSection(String editor, Date date, OOXMLWordAndPowerPointTextHandler.EditType editType) {
+ public void startEditedSection(String editor, Date date,
+ OOXMLWordAndPowerPointTextHandler.EditType editType) {
}
@@ -168,7 +177,7 @@ public class XSLFEventBasedPowerPointExtractor extends POIXMLTextExtractor {
}
@Override
- public boolean getIncludeDeletedText() {
+ public boolean isIncludeDeletedText() {
return false;
}
@@ -183,7 +192,7 @@ public class XSLFEventBasedPowerPointExtractor extends POIXMLTextExtractor {
}
@Override
- public boolean getIncludeMoveFromText() {
+ public boolean isIncludeMoveFromText() {
return false;
}
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xwpf/XWPFEventBasedWordExtractor.java b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xwpf/XWPFEventBasedWordExtractor.java
index e39dfd4..fd8f764 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xwpf/XWPFEventBasedWordExtractor.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/ooxml/xwpf/XWPFEventBasedWordExtractor.java
@@ -17,7 +17,7 @@
package org.apache.tika.parser.microsoft.ooxml.xwpf;
-import javax.xml.parsers.ParserConfigurationException;
+import java.io.Closeable;
import java.io.IOException;
import java.io.InputStream;
import java.util.Date;
@@ -29,7 +29,6 @@ import org.apache.commons.io.input.CloseShieldInputStream;
import org.apache.poi.ooxml.POIXMLDocument;
import org.apache.poi.ooxml.POIXMLProperties;
import org.apache.poi.ooxml.extractor.POIXMLTextExtractor;
-import org.apache.poi.ooxml.util.SAXHelper;
import org.apache.poi.openxml4j.exceptions.InvalidFormatException;
import org.apache.poi.openxml4j.exceptions.OpenXML4JException;
import org.apache.poi.openxml4j.opc.OPCPackage;
@@ -38,55 +37,39 @@ import org.apache.poi.openxml4j.opc.PackageRelationship;
import org.apache.poi.openxml4j.opc.PackageRelationshipCollection;
import org.apache.poi.xwpf.usermodel.XWPFNumbering;
import org.apache.poi.xwpf.usermodel.XWPFRelation;
+import org.apache.xmlbeans.XmlException;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
+import org.xml.sax.SAXException;
import org.apache.tika.exception.RuntimeSAXException;
+import org.apache.tika.exception.TikaException;
import org.apache.tika.exception.WriteLimitReachedException;
+import org.apache.tika.parser.ParseContext;
import org.apache.tika.parser.microsoft.ooxml.OOXMLWordAndPowerPointTextHandler;
import org.apache.tika.parser.microsoft.ooxml.ParagraphProperties;
import org.apache.tika.parser.microsoft.ooxml.RunProperties;
import org.apache.tika.parser.microsoft.ooxml.XWPFListManager;
-import org.apache.xmlbeans.XmlException;
-import org.slf4j.Logger;
-import org.slf4j.LoggerFactory;
-import org.xml.sax.InputSource;
-import org.xml.sax.SAXException;
-import org.xml.sax.XMLReader;
+import org.apache.tika.utils.XMLReaderUtils;
//TODO: move this into POI?
+
/**
* Experimental class that is based on POI's XSSFEventBasedExcelExtractor
- *
*/
-public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
+public class XWPFEventBasedWordExtractor implements POIXMLTextExtractor {
private static final Logger LOG = LoggerFactory.getLogger(XWPFEventBasedWordExtractor.class);
private OPCPackage container;
private POIXMLProperties properties;
- public XWPFEventBasedWordExtractor(String path) throws XmlException, OpenXML4JException, IOException {
- this(OPCPackage.open(path));
- }
-
- public XWPFEventBasedWordExtractor(OPCPackage container) throws XmlException, OpenXML4JException, IOException {
- super((POIXMLDocument) null);
+ public XWPFEventBasedWordExtractor(OPCPackage container)
+ throws XmlException, OpenXML4JException, IOException {
this.container = container;
this.properties = new POIXMLProperties(container);
}
-
- public static void main(String[] args) throws Exception {
- if (args.length < 1) {
- System.err.println("Use:");
- System.err.println(" XWPFEventBasedWordExtractor <filename.xlsx>");
- System.exit(1);
- }
-
- XWPFEventBasedWordExtractor extractor = new XWPFEventBasedWordExtractor(args[0]);
- System.out.println(extractor.getText());
- extractor.close();
- }
-
public OPCPackage getPackage() {
return this.container;
}
@@ -103,12 +86,18 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
return this.properties.getCustomProperties();
}
+ @Override
+ public POIXMLDocument getDocument() {
+ return null;
+ }
+
@Override
public String getText() {
StringBuilder sb = new StringBuilder();
//handle main document
- List<PackagePart> pps = container.getPartsByContentType(XWPFRelation.DOCUMENT.getContentType());
+ List<PackagePart> pps =
+ container.getPartsByContentType(XWPFRelation.DOCUMENT.getContentType());
if (pps != null) {
for (PackagePart pp : pps) {
//likely only one, but why not...
@@ -122,6 +111,8 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
}
//swallow this because we don't actually call it
LOG.warn("SAXException handling document part", e);
+ } catch (TikaException e) {
+ LOG.warn("ParseException handling document part", e);
}
}
}
@@ -141,6 +132,8 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
}
//swallow this because we don't actually call it
LOG.warn("SAXException handling glossary document part", e);
+ } catch (TikaException e) {
+ LOG.warn("ParseException handling document part", e);
}
}
}
@@ -148,8 +141,24 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
return sb.toString();
}
+ @Override
+ public void setCloseFilesystem(boolean b) {
+
+ }
+
+ @Override
+ public boolean isCloseFilesystem() {
+ return false;
+ }
+
+ @Override
+ public Closeable getFilesystem() {
+ return null;
+ }
- private void handleDocumentPart(PackagePart documentPart, StringBuilder sb) throws IOException, SAXException {
+
+ private void handleDocumentPart(PackagePart documentPart, StringBuilder sb)
+ throws IOException, SAXException, TikaException {
//load the numbering/list manager and styles from the main document part
XWPFNumbering numbering = loadNumbering(documentPart);
XWPFListManager xwpfListManager = new XWPFListManager(numbering);
@@ -157,7 +166,8 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
//headers
try {
- PackageRelationshipCollection headersPRC = documentPart.getRelationshipsByType(XWPFRelation.HEADER.getRelation());
+ PackageRelationshipCollection headersPRC =
+ documentPart.getRelationshipsByType(XWPFRelation.HEADER.getRelation());
if (headersPRC != null) {
for (int i = 0; i < headersPRC.size(); i++) {
PackagePart header = documentPart.getRelatedPart(headersPRC.getRelationship(i));
@@ -172,17 +182,15 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
handlePart(documentPart, xwpfListManager, sb);
//for now, just dump other components at end
- for (XWPFRelation rel : new XWPFRelation[]{
- XWPFRelation.FOOTNOTE,
- XWPFRelation.COMMENT,
- XWPFRelation.FOOTER,
- XWPFRelation.ENDNOTE
- }) {
+ for (XWPFRelation rel : new XWPFRelation[]{XWPFRelation.FOOTNOTE, XWPFRelation.COMMENT,
+ XWPFRelation.FOOTER, XWPFRelation.ENDNOTE}) {
try {
- PackageRelationshipCollection prc = documentPart.getRelationshipsByType(rel.getRelation());
+ PackageRelationshipCollection prc =
+ documentPart.getRelationshipsByType(rel.getRelation());
if (prc != null) {
for (int i = 0; i < prc.size(); i++) {
- PackagePart packagePart = documentPart.getRelatedPart(prc.getRelationship(i));
+ PackagePart packagePart =
+ documentPart.getRelatedPart(prc.getRelationship(i));
handlePart(packagePart, xwpfListManager, sb);
}
}
@@ -192,18 +200,14 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
}
}
- private void handlePart(PackagePart packagePart,
- XWPFListManager xwpfListManager, StringBuilder buffer) throws IOException, SAXException {
+ private void handlePart(PackagePart packagePart, XWPFListManager xwpfListManager,
+ StringBuilder buffer) throws IOException, SAXException, TikaException {
Map<String, String> hyperlinks = loadHyperlinkRelationships(packagePart);
try (InputStream stream = packagePart.getInputStream()) {
- XMLReader reader = SAXHelper.newXMLReader();
- reader.setContentHandler(new OOXMLWordAndPowerPointTextHandler(
- new XWPFToTextContentHandler(buffer), hyperlinks));
- reader.parse(new InputSource(new CloseShieldInputStream(stream)));
-
- } catch (ParserConfigurationException e) {
- LOG.warn("Can't configure XMLReader", e);
+ XMLReaderUtils.parseSAX(new CloseShieldInputStream(stream),
+ new OOXMLWordAndPowerPointTextHandler(new XWPFToTextContentHandler(buffer),
+ hyperlinks), new ParseContext());
}
}
@@ -211,7 +215,8 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
private Map<String, String> loadHyperlinkRelationships(PackagePart bodyPart) {
Map<String, String> hyperlinks = new HashMap<>();
try {
- PackageRelationshipCollection prc = bodyPart.getRelationshipsByType(XWPFRelation.HYPERLINK.getRelation());
+ PackageRelationshipCollection prc =
+ bodyPart.getRelationshipsByType(XWPFRelation.HYPERLINK.getRelation());
for (int i = 0; i < prc.size(); i++) {
PackageRelationship pr = prc.getRelationship(i);
if (pr == null) {
@@ -231,7 +236,8 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
private XWPFNumbering loadNumbering(PackagePart packagePart) throws IOException {
try {
- PackageRelationshipCollection numberingParts = packagePart.getRelationshipsByType(XWPFRelation.NUMBERING.getRelation());
+ PackageRelationshipCollection numberingParts =
+ packagePart.getRelationshipsByType(XWPFRelation.NUMBERING.getRelation());
if (numberingParts.size() > 0) {
PackageRelationship numberingRelationShip = numberingParts.getRelationship(0);
if (numberingRelationShip == null) {
@@ -249,7 +255,8 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
return null;
}
- private class XWPFToTextContentHandler implements OOXMLWordAndPowerPointTextHandler.XWPFBodyContentsHandler {
+ private static class XWPFToTextContentHandler
+ implements OOXMLWordAndPowerPointTextHandler.XWPFBodyContentsHandler {
private final StringBuilder buffer;
public XWPFToTextContentHandler(StringBuilder buffer) {
@@ -322,7 +329,8 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
}
@Override
- public void startEditedSection(String editor, Date date, OOXMLWordAndPowerPointTextHandler.EditType editType) {
+ public void startEditedSection(String editor, Date date,
+ OOXMLWordAndPowerPointTextHandler.EditType editType) {
}
@@ -332,7 +340,7 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
}
@Override
- public boolean getIncludeDeletedText() {
+ public boolean isIncludeDeletedText() {
return true;
}
@@ -347,7 +355,7 @@ public class XWPFEventBasedWordExtractor extends POIXMLTextExtractor {
}
@Override
- public boolean getIncludeMoveFromText() {
+ public boolean isIncludeMoveFromText() {
return false;
}
diff --git a/tika-server/src/main/java/org/apache/tika/server/resource/TikaResource.java b/tika-server/src/main/java/org/apache/tika/server/resource/TikaResource.java
index b326f7f..c77694d 100644
--- a/tika-server/src/main/java/org/apache/tika/server/resource/TikaResource.java
+++ b/tika-server/src/main/java/org/apache/tika/server/resource/TikaResource.java
@@ -22,7 +22,6 @@ import org.apache.commons.lang3.StringUtils;
import org.apache.cxf.attachment.ContentDisposition;
import org.apache.cxf.jaxrs.ext.multipart.Attachment;
import org.apache.cxf.jaxrs.impl.MetadataMap;
-import org.apache.poi.ooxml.extractor.ExtractorFactory;
import org.apache.tika.Tika;
import org.apache.tika.config.TikaConfig;
import org.apache.tika.detect.Detector;
@@ -55,6 +54,7 @@ import org.apache.tika.server.ServerStatus;
import org.apache.tika.server.TikaServerParseException;
import org.apache.tika.utils.ExceptionUtils;
+import org.apache.poi.extractor.ExtractorFactory;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import org.xml.sax.ContentHandler;