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 2016/12/20 21:32:52 UTC
[1/4] tika git commit: [TIKA-2189] fix for Default value mismatch for
"enableImageProcessing" in TesseractOCRConfig.properties and
TesseractOCRConfig.java
Repository: tika
Updated Branches:
refs/heads/master 2dbd65122 -> ae44b9e50
[TIKA-2189] fix for Default value mismatch for "enableImageProcessing" in TesseractOCRConfig.properties and TesseractOCRConfig.java
Project: http://git-wip-us.apache.org/repos/asf/tika/repo
Commit: http://git-wip-us.apache.org/repos/asf/tika/commit/40401e51
Tree: http://git-wip-us.apache.org/repos/asf/tika/tree/40401e51
Diff: http://git-wip-us.apache.org/repos/asf/tika/diff/40401e51
Branch: refs/heads/master
Commit: 40401e51b8a135634f54c8c437dafc10378059be
Parents: 99b5924
Author: dasbipulkumar <ku...@gmail.com>
Authored: Fri Dec 2 18:49:08 2016 +0530
Committer: dasbipulkumar <ku...@gmail.com>
Committed: Fri Dec 2 18:51:47 2016 +0530
----------------------------------------------------------------------
.../org/apache/tika/parser/ocr/TesseractOCRConfig.properties | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/tika/blob/40401e51/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties
----------------------------------------------------------------------
diff --git a/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties b/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties
index 2380282..ddcd473 100644
--- a/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties
+++ b/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties
@@ -24,7 +24,7 @@ outputType=txt
# properties for image processing
# to enable processing, set enableImageProcessing to 1
-enableImageProcessing=1
+enableImageProcessing=0
ImageMagickPath=
density=300
depth=4
[3/4] tika git commit: clean up tabs
Posted by ta...@apache.org.
clean up tabs
Project: http://git-wip-us.apache.org/repos/asf/tika/repo
Commit: http://git-wip-us.apache.org/repos/asf/tika/commit/c83f87b2
Tree: http://git-wip-us.apache.org/repos/asf/tika/tree/c83f87b2
Diff: http://git-wip-us.apache.org/repos/asf/tika/diff/c83f87b2
Branch: refs/heads/master
Commit: c83f87b2acc6a8fb69dc37991d43542367154c1a
Parents: 8e12ebe
Author: tballison <ta...@mitre.org>
Authored: Tue Dec 20 15:16:35 2016 -0500
Committer: tballison <ta...@mitre.org>
Committed: Tue Dec 20 15:16:35 2016 -0500
----------------------------------------------------------------------
.../tika/parser/ocr/TesseractOCRConfig.java | 808 ++++++++++---------
1 file changed, 415 insertions(+), 393 deletions(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/tika/blob/c83f87b2/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java
----------------------------------------------------------------------
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java b/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java
index 7d6cd3f..a08b0e2 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java
@@ -25,169 +25,172 @@ import java.util.Properties;
/**
* Configuration for TesseractOCRParser.
- *
+ * <p>
* This allows to enable TesseractOCRParser and set its parameters:
* <p>
* TesseractOCRConfig config = new TesseractOCRConfig();<br>
* config.setTesseractPath(tesseractFolder);<br>
* parseContext.set(TesseractOCRConfig.class, config);<br>
* </p>
- *
+ * <p>
* Parameters can also be set by either editing the existing TesseractOCRConfig.properties file in,
* tika-parser/src/main/resources/org/apache/tika/parser/ocr, or overriding it by creating your own
* and placing it in the package org/apache/tika/parser/ocr on the classpath.
- *
*/
-public class TesseractOCRConfig implements Serializable{
+public class TesseractOCRConfig implements Serializable {
- private static final long serialVersionUID = -4861942486845757891L;
+ private static final long serialVersionUID = -4861942486845757891L;
- public enum OUTPUT_TYPE {
- TXT,
- HOCR
- }
+ public enum OUTPUT_TYPE {
+ TXT,
+ HOCR
+ }
- // Path to tesseract installation folder, if not on system path.
- private String tesseractPath = "";
+ // Path to tesseract installation folder, if not on system path.
+ private String tesseractPath = "";
// Path to the 'tessdata' folder, which contains language files and config files.
private String tessdataPath = "";
- // Language dictionary to be used.
- private String language = "eng";
-
- // Tesseract page segmentation mode.
- private String pageSegMode = "1";
-
- // Minimum file size to submit file to ocr.
- private int minFileSizeToOcr = 0;
-
- // Maximum file size to submit file to ocr.
- private int maxFileSizeToOcr = Integer.MAX_VALUE;
-
- // Maximum time (seconds) to wait for the ocring process termination
- private int timeout = 120;
-
- // The format of the ocr'ed output to be returned, txt or hocr.
- private OUTPUT_TYPE outputType = OUTPUT_TYPE.TXT;
-
- // enable image processing (optional)
- private int enableImageProcessing = 0;
-
- // Path to ImageMagick program, if not on system path.
- private String ImageMagickPath = "";
-
- // resolution of processed image (in dpi).
- private int density = 300;
-
- // number of bits in a color sample within a pixel.
- private int depth = 4;
-
- // colorspace of processed image.
- private String colorspace = "gray";
-
- // filter to be applied to the processed image.
- private String filter = "triangle";
-
- // factor by which image is to be scaled.
- private int resize = 900;
-
- /**
- * Default contructor.
- */
- public TesseractOCRConfig() {
- init(this.getClass().getResourceAsStream("TesseractOCRConfig.properties"));
- }
-
- /**
- * Loads properties from InputStream and then tries to close InputStream.
- * If there is an IOException, this silently swallows the exception
- * and goes back to the default.
- *
- * @param is
- */
- public TesseractOCRConfig(InputStream is) {
- init(is);
- }
-
- private void init(InputStream is) {
- if (is == null) {
- return;
- }
- Properties props = new Properties();
- try {
- props.load(is);
- } catch (IOException e) {
- } finally {
- if (is != null) {
- try {
- is.close();
- } catch (IOException e) {
- //swallow
- }
- }
- }
-
- // set parameters for Tesseract
- setTesseractPath(
- getProp(props, "tesseractPath", getTesseractPath()));
+ // Language dictionary to be used.
+ private String language = "eng";
+
+ // Tesseract page segmentation mode.
+ private String pageSegMode = "1";
+
+ // Minimum file size to submit file to ocr.
+ private int minFileSizeToOcr = 0;
+
+ // Maximum file size to submit file to ocr.
+ private int maxFileSizeToOcr = Integer.MAX_VALUE;
+
+ // Maximum time (seconds) to wait for the ocring process termination
+ private int timeout = 120;
+
+ // The format of the ocr'ed output to be returned, txt or hocr.
+ private OUTPUT_TYPE outputType = OUTPUT_TYPE.TXT;
+
+ // enable image processing (optional)
+ private int enableImageProcessing = 0;
+
+ // Path to ImageMagick program, if not on system path.
+ private String ImageMagickPath = "";
+
+ // resolution of processed image (in dpi).
+ private int density = 300;
+
+ // number of bits in a color sample within a pixel.
+ private int depth = 4;
+
+ // colorspace of processed image.
+ private String colorspace = "gray";
+
+ // filter to be applied to the processed image.
+ private String filter = "triangle";
+
+ // factor by which image is to be scaled.
+ private int resize = 900;
+
+ /**
+ * Default contructor.
+ */
+ public TesseractOCRConfig() {
+ init(this.getClass().getResourceAsStream("TesseractOCRConfig.properties"));
+ }
+
+ /**
+ * Loads properties from InputStream and then tries to close InputStream.
+ * If there is an IOException, this silently swallows the exception
+ * and goes back to the default.
+ *
+ * @param is
+ */
+ public TesseractOCRConfig(InputStream is) {
+ init(is);
+ }
+
+ private void init(InputStream is) {
+ if (is == null) {
+ return;
+ }
+ Properties props = new Properties();
+ try {
+ props.load(is);
+ } catch (IOException e) {
+ } finally {
+ if (is != null) {
+ try {
+ is.close();
+ } catch (IOException e) {
+ //swallow
+ }
+ }
+ }
+
+ // set parameters for Tesseract
+ setTesseractPath(
+ getProp(props, "tesseractPath", getTesseractPath()));
setTessdataPath(
getProp(props, "tessdataPath", getTessdataPath()));
- setLanguage(
- getProp(props, "language", getLanguage()));
- setPageSegMode(
- getProp(props, "pageSegMode", getPageSegMode()));
- setMinFileSizeToOcr(
- getProp(props, "minFileSizeToOcr", getMinFileSizeToOcr()));
- setMaxFileSizeToOcr(
- getProp(props, "maxFileSizeToOcr", getMaxFileSizeToOcr()));
- setTimeout(
+ setLanguage(
+ getProp(props, "language", getLanguage()));
+ setPageSegMode(
+ getProp(props, "pageSegMode", getPageSegMode()));
+ setMinFileSizeToOcr(
+ getProp(props, "minFileSizeToOcr", getMinFileSizeToOcr()));
+ setMaxFileSizeToOcr(
+ getProp(props, "maxFileSizeToOcr", getMaxFileSizeToOcr()));
+ setTimeout(
getProp(props, "timeout", getTimeout()));
- String outputTypeString = props.getProperty("outputType");
- if ("txt".equals(outputTypeString)) {
- setOutputType(OUTPUT_TYPE.TXT);
- } else if ("hocr".equals(outputTypeString)) {
- setOutputType(OUTPUT_TYPE.HOCR);
- }
-
- // set parameters for ImageMagick
- setEnableImageProcessing(
- getProp(props, "enableImageProcessing", isEnableImageProcessing()));
- setImageMagickPath(
- getProp(props, "ImageMagickPath", getImageMagickPath()));
- setDensity(
- getProp(props, "density", getDensity()));
- setDepth(
- getProp(props, "depth", getDepth()));
- setColorspace(
- getProp(props, "colorspace", getColorspace()));
- setFilter(
- getProp(props, "filter", getFilter()));
- setResize(
- getProp(props, "resize", getResize()));
-
- }
-
- /** @see #setTesseractPath(String tesseractPath)*/
- public String getTesseractPath() {
- return tesseractPath;
- }
-
- /**
- * Set the path to the Tesseract executable, needed if it is not on system path.
+ String outputTypeString = props.getProperty("outputType");
+ if ("txt".equals(outputTypeString)) {
+ setOutputType(OUTPUT_TYPE.TXT);
+ } else if ("hocr".equals(outputTypeString)) {
+ setOutputType(OUTPUT_TYPE.HOCR);
+ }
+
+ // set parameters for ImageMagick
+ setEnableImageProcessing(
+ getProp(props, "enableImageProcessing", isEnableImageProcessing()));
+ setImageMagickPath(
+ getProp(props, "ImageMagickPath", getImageMagickPath()));
+ setDensity(
+ getProp(props, "density", getDensity()));
+ setDepth(
+ getProp(props, "depth", getDepth()));
+ setColorspace(
+ getProp(props, "colorspace", getColorspace()));
+ setFilter(
+ getProp(props, "filter", getFilter()));
+ setResize(
+ getProp(props, "resize", getResize()));
+
+ }
+
+ /**
+ * @see #setTesseractPath(String tesseractPath)
+ */
+ public String getTesseractPath() {
+ return tesseractPath;
+ }
+
+ /**
+ * Set the path to the Tesseract executable, needed if it is not on system path.
* <p>
* Note that if you set this value, it is highly recommended that you also
* set the path to the 'tessdata' folder using {@link #setTessdataPath}.
* </p>
- */
- public void setTesseractPath(String tesseractPath) {
- if(!tesseractPath.isEmpty() && !tesseractPath.endsWith(File.separator))
- tesseractPath += File.separator;
+ */
+ public void setTesseractPath(String tesseractPath) {
+ if (!tesseractPath.isEmpty() && !tesseractPath.endsWith(File.separator))
+ tesseractPath += File.separator;
- this.tesseractPath = tesseractPath;
- }
+ this.tesseractPath = tesseractPath;
+ }
- /** @see #setTessdataPath(String tessdataPath) */
+ /**
+ * @see #setTessdataPath(String tessdataPath)
+ */
public String getTessdataPath() {
return tessdataPath;
}
@@ -198,265 +201,284 @@ public class TesseractOCRConfig implements Serializable{
* (such as when Tesseract is built from source), it may be located elsewhere.
*/
public void setTessdataPath(String tessdataPath) {
- if(!tessdataPath.isEmpty() && !tessdataPath.endsWith(File.separator))
+ if (!tessdataPath.isEmpty() && !tessdataPath.endsWith(File.separator))
tessdataPath += File.separator;
this.tessdataPath = tessdataPath;
}
- /** @see #setLanguage(String language)*/
- public String getLanguage() {
- return language;
- }
-
- /**
- * Set tesseract language dictionary to be used. Default is "eng".
- * Multiple languages may be specified, separated by plus characters.
- */
- public void setLanguage(String language) {
- if (!language.matches("([A-Za-z](\\+?))*")) {
- throw new IllegalArgumentException("Invalid language code");
- }
- this.language = language;
- }
-
- /** @see #setPageSegMode(String pageSegMode)*/
- public String getPageSegMode() {
- return pageSegMode;
- }
-
- /**
- * Set tesseract page segmentation mode.
- * Default is 1 = Automatic page segmentation with OSD (Orientation and Script Detection)
- */
- public void setPageSegMode(String pageSegMode) {
- if (!pageSegMode.matches("[1-9]|10")) {
- throw new IllegalArgumentException("Invalid language code");
- }
- this.pageSegMode = pageSegMode;
- }
-
- /** @see #setMinFileSizeToOcr(int minFileSizeToOcr)*/
- public int getMinFileSizeToOcr() {
- return minFileSizeToOcr;
- }
-
- /**
- * Set minimum file size to submit file to ocr.
- * Default is 0.
- */
- public void setMinFileSizeToOcr(int minFileSizeToOcr) {
- this.minFileSizeToOcr = minFileSizeToOcr;
- }
-
- /** @see #setMaxFileSizeToOcr(int maxFileSizeToOcr)*/
- public int getMaxFileSizeToOcr() {
- return maxFileSizeToOcr;
- }
-
- /**
- * Set maximum file size to submit file to ocr.
- * Default is Integer.MAX_VALUE.
- */
- public void setMaxFileSizeToOcr(int maxFileSizeToOcr) {
- this.maxFileSizeToOcr = maxFileSizeToOcr;
- }
-
- /**
- * Set maximum time (seconds) to wait for the ocring process to terminate.
- * Default value is 120s.
- */
- public void setTimeout(int timeout) {
- this.timeout = timeout;
- }
-
- /** @see #setTimeout(int timeout)
- * @return timeout value for Tesseract */
- public int getTimeout() {
- return timeout;
- }
-
- /**
- * Set output type from ocr process. Default is "txt", but can be "hocr".
- * Default value is 120s.
- */
- public void setOutputType(OUTPUT_TYPE outputType) {
- this.outputType = outputType;
- }
-
- /** @see #setOutputType(OUTPUT_TYPE outputType) */
- public OUTPUT_TYPE getOutputType() {
- return outputType;
- }
-
- /** @see #setEnableImageProcessing(int)
- * @return image processing is enabled or not */
- public int isEnableImageProcessing() {
- return enableImageProcessing;
- }
-
- /**
- * Set the value to true if processing is to be enabled.
- * Default value is false.
- */
- public void setEnableImageProcessing(int enableImageProcessing) {
- this.enableImageProcessing = enableImageProcessing;
- }
-
- /**
- * @return the density
- */
- public int getDensity() {
- return density;
- }
-
- /**
- * @param density the density to set. Valid range of values is 150-1200.
- * Default value is 300.
- */
- public void setDensity(int density) {
- if(density < 150 || density > 1200) {
- throw new IllegalArgumentException("Invalid density value. Valid range of values is 150-1200.");
- }
- this.density = density;
- }
-
- /**
- * @return the depth
- */
- public int getDepth() {
- return depth;
- }
-
- /**
- * @param depth the depth to set. Valid values are 2, 4, 8, 16, 32, 64, 256, 4096.
- * Default value is 4.
- */
- public void setDepth(int depth) {
- int[] allowedValues = {2, 4, 8, 16, 32, 64, 256, 4096};
- for (int i = 0; i < allowedValues.length; i++) {
- if(depth == allowedValues[i]) {
- this.depth = depth;
- return;
- }
- }
- throw new IllegalArgumentException("Invalid depth value. Valid values are 2, 4, 8, 16, 32, 64, 256, 4096.");
- }
-
- /**
- * @return the colorspace
- */
- public String getColorspace() {
- return colorspace;
- }
-
- /**
- * @param colorspace the colorspace to set
- * Deafult value is gray.
- */
- public void setColorspace(String colorspace) {
- if(!colorspace.equals(null)) {
- this.colorspace = colorspace;
- } else {
- throw new IllegalArgumentException("Colorspace value cannot be null.");
- }
- }
-
- /**
- * @return the filter
- */
- public String getFilter() {
- return filter;
- }
-
- /**
- * @param filter the filter to set. Valid values are point, hermite, cubic, box, gaussian, catrom, triangle, quadratic and mitchell.
- * Default value is triangle.
- */
- public void setFilter(String filter) {
- if(filter.equals(null)) {
- throw new IllegalArgumentException("Filter value cannot be null. Valid values are point, hermite, "
- + "cubic, box, gaussian, catrom, triangle, quadratic and mitchell.");
- }
-
- String[] allowedFilters = {"Point", "Hermite", "Cubic", "Box", "Gaussian", "Catrom", "Triangle", "Quadratic", "Mitchell"};
- for (int i = 0; i < allowedFilters.length; i++) {
- if(filter.equalsIgnoreCase(allowedFilters[i])) {
- this.filter = filter;
- return;
- }
- }
- throw new IllegalArgumentException("Invalid filter value. Valid values are point, hermite, "
- + "cubic, box, gaussian, catrom, triangle, quadratic and mitchell.");
- }
-
- /**
- * @return the resize
- */
- public int getResize() {
- return resize;
- }
-
- /**
- * @param resize the resize to set. Valid range of values is 100-900.
- * Default value is 900.
- */
- public void setResize(int resize) {
- for(int i=1;i<10;i++) {
- if(resize == i*100) {
- this.resize = resize;
- return;
- }
- }
- throw new IllegalArgumentException("Invalid resize value. Valid range of values is 100-900.");
- }
-
- /** @see #setImageMagickPath(String ImageMagickPath)
- * @return path to ImageMagick file. */
- public String getImageMagickPath() {
-
- return ImageMagickPath;
- }
-
- /**
- * Set the path to the ImageMagick executable, needed if it is not on system path.
- * @param ImageMagickPath to ImageMagick file.
- */
- public void setImageMagickPath(String ImageMagickPath) {
- if(!ImageMagickPath.isEmpty() && !ImageMagickPath.endsWith(File.separator))
- ImageMagickPath += File.separator;
-
- this.ImageMagickPath = ImageMagickPath;
- }
-
- /**
- * Get property from the properties file passed in.
- * @param properties properties file to read from.
- * @param property the property to fetch.
- * @param defaultMissing default parameter to use.
- * @return the value.
- */
- private int getProp(Properties properties, String property, int defaultMissing) {
- String p = properties.getProperty(property);
- if (p == null || p.isEmpty()){
- return defaultMissing;
- }
- try {
- return Integer.parseInt(p);
- } catch (Throwable ex) {
- throw new RuntimeException(String.format(Locale.ROOT, "Cannot parse TesseractOCRConfig variable %s, invalid integer value",
- property), ex);
- }
- }
-
- /**
- * Get property from the properties file passed in.
- * @param properties properties file to read from.
- * @param property the property to fetch.
- * @param defaultMissing default parameter to use.
- * @return the value.
- */
- private String getProp(Properties properties, String property, String defaultMissing) {
- return properties.getProperty(property, defaultMissing);
- }
+ /**
+ * @see #setLanguage(String language)
+ */
+ public String getLanguage() {
+ return language;
+ }
+
+ /**
+ * Set tesseract language dictionary to be used. Default is "eng".
+ * Multiple languages may be specified, separated by plus characters.
+ */
+ public void setLanguage(String language) {
+ if (!language.matches("([A-Za-z](\\+?))*")) {
+ throw new IllegalArgumentException("Invalid language code");
+ }
+ this.language = language;
+ }
+
+ /**
+ * @see #setPageSegMode(String pageSegMode)
+ */
+ public String getPageSegMode() {
+ return pageSegMode;
+ }
+
+ /**
+ * Set tesseract page segmentation mode.
+ * Default is 1 = Automatic page segmentation with OSD (Orientation and Script Detection)
+ */
+ public void setPageSegMode(String pageSegMode) {
+ if (!pageSegMode.matches("[1-9]|10")) {
+ throw new IllegalArgumentException("Invalid language code");
+ }
+ this.pageSegMode = pageSegMode;
+ }
+
+ /**
+ * @see #setMinFileSizeToOcr(int minFileSizeToOcr)
+ */
+ public int getMinFileSizeToOcr() {
+ return minFileSizeToOcr;
+ }
+
+ /**
+ * Set minimum file size to submit file to ocr.
+ * Default is 0.
+ */
+ public void setMinFileSizeToOcr(int minFileSizeToOcr) {
+ this.minFileSizeToOcr = minFileSizeToOcr;
+ }
+
+ /**
+ * @see #setMaxFileSizeToOcr(int maxFileSizeToOcr)
+ */
+ public int getMaxFileSizeToOcr() {
+ return maxFileSizeToOcr;
+ }
+
+ /**
+ * Set maximum file size to submit file to ocr.
+ * Default is Integer.MAX_VALUE.
+ */
+ public void setMaxFileSizeToOcr(int maxFileSizeToOcr) {
+ this.maxFileSizeToOcr = maxFileSizeToOcr;
+ }
+
+ /**
+ * Set maximum time (seconds) to wait for the ocring process to terminate.
+ * Default value is 120s.
+ */
+ public void setTimeout(int timeout) {
+ this.timeout = timeout;
+ }
+
+ /**
+ * @return timeout value for Tesseract
+ * @see #setTimeout(int timeout)
+ */
+ public int getTimeout() {
+ return timeout;
+ }
+
+ /**
+ * Set output type from ocr process. Default is "txt", but can be "hocr".
+ * Default value is 120s.
+ */
+ public void setOutputType(OUTPUT_TYPE outputType) {
+ this.outputType = outputType;
+ }
+
+ /**
+ * @see #setOutputType(OUTPUT_TYPE outputType)
+ */
+ public OUTPUT_TYPE getOutputType() {
+ return outputType;
+ }
+
+ /**
+ * @return image processing is enabled or not
+ * @see #setEnableImageProcessing(int)
+ */
+ public int isEnableImageProcessing() {
+ return enableImageProcessing;
+ }
+
+ /**
+ * Set the value to true if processing is to be enabled.
+ * Default value is false.
+ */
+ public void setEnableImageProcessing(int enableImageProcessing) {
+ this.enableImageProcessing = enableImageProcessing;
+ }
+
+ /**
+ * @return the density
+ */
+ public int getDensity() {
+ return density;
+ }
+
+ /**
+ * @param density the density to set. Valid range of values is 150-1200.
+ * Default value is 300.
+ */
+ public void setDensity(int density) {
+ if (density < 150 || density > 1200) {
+ throw new IllegalArgumentException("Invalid density value. Valid range of values is 150-1200.");
+ }
+ this.density = density;
+ }
+
+ /**
+ * @return the depth
+ */
+ public int getDepth() {
+ return depth;
+ }
+
+ /**
+ * @param depth the depth to set. Valid values are 2, 4, 8, 16, 32, 64, 256, 4096.
+ * Default value is 4.
+ */
+ public void setDepth(int depth) {
+ int[] allowedValues = {2, 4, 8, 16, 32, 64, 256, 4096};
+ for (int i = 0; i < allowedValues.length; i++) {
+ if (depth == allowedValues[i]) {
+ this.depth = depth;
+ return;
+ }
+ }
+ throw new IllegalArgumentException("Invalid depth value. Valid values are 2, 4, 8, 16, 32, 64, 256, 4096.");
+ }
+
+ /**
+ * @return the colorspace
+ */
+ public String getColorspace() {
+ return colorspace;
+ }
+
+ /**
+ * @param colorspace the colorspace to set
+ * Deafult value is gray.
+ */
+ public void setColorspace(String colorspace) {
+ if (!colorspace.equals(null)) {
+ this.colorspace = colorspace;
+ } else {
+ throw new IllegalArgumentException("Colorspace value cannot be null.");
+ }
+ }
+
+ /**
+ * @return the filter
+ */
+ public String getFilter() {
+ return filter;
+ }
+
+ /**
+ * @param filter the filter to set. Valid values are point, hermite, cubic, box, gaussian, catrom, triangle, quadratic and mitchell.
+ * Default value is triangle.
+ */
+ public void setFilter(String filter) {
+ if (filter.equals(null)) {
+ throw new IllegalArgumentException("Filter value cannot be null. Valid values are point, hermite, "
+ + "cubic, box, gaussian, catrom, triangle, quadratic and mitchell.");
+ }
+
+ String[] allowedFilters = {"Point", "Hermite", "Cubic", "Box", "Gaussian", "Catrom", "Triangle", "Quadratic", "Mitchell"};
+ for (int i = 0; i < allowedFilters.length; i++) {
+ if (filter.equalsIgnoreCase(allowedFilters[i])) {
+ this.filter = filter;
+ return;
+ }
+ }
+ throw new IllegalArgumentException("Invalid filter value. Valid values are point, hermite, "
+ + "cubic, box, gaussian, catrom, triangle, quadratic and mitchell.");
+ }
+
+ /**
+ * @return the resize
+ */
+ public int getResize() {
+ return resize;
+ }
+
+ /**
+ * @param resize the resize to set. Valid range of values is 100-900.
+ * Default value is 900.
+ */
+ public void setResize(int resize) {
+ for (int i = 1; i < 10; i++) {
+ if (resize == i * 100) {
+ this.resize = resize;
+ return;
+ }
+ }
+ throw new IllegalArgumentException("Invalid resize value. Valid range of values is 100-900.");
+ }
+
+ /**
+ * @return path to ImageMagick file.
+ * @see #setImageMagickPath(String ImageMagickPath)
+ */
+ public String getImageMagickPath() {
+
+ return ImageMagickPath;
+ }
+
+ /**
+ * Set the path to the ImageMagick executable, needed if it is not on system path.
+ *
+ * @param ImageMagickPath to ImageMagick file.
+ */
+ public void setImageMagickPath(String ImageMagickPath) {
+ if (!ImageMagickPath.isEmpty() && !ImageMagickPath.endsWith(File.separator))
+ ImageMagickPath += File.separator;
+
+ this.ImageMagickPath = ImageMagickPath;
+ }
+
+ /**
+ * Get property from the properties file passed in.
+ *
+ * @param properties properties file to read from.
+ * @param property the property to fetch.
+ * @param defaultMissing default parameter to use.
+ * @return the value.
+ */
+ private int getProp(Properties properties, String property, int defaultMissing) {
+ String p = properties.getProperty(property);
+ if (p == null || p.isEmpty()) {
+ return defaultMissing;
+ }
+ try {
+ return Integer.parseInt(p);
+ } catch (Throwable ex) {
+ throw new RuntimeException(String.format(Locale.ROOT, "Cannot parse TesseractOCRConfig variable %s, invalid integer value",
+ property), ex);
+ }
+ }
+
+ /**
+ * Get property from the properties file passed in.
+ *
+ * @param properties properties file to read from.
+ * @param property the property to fetch.
+ * @param defaultMissing default parameter to use.
+ * @return the value.
+ */
+ private String getProp(Properties properties, String property, String defaultMissing) {
+ return properties.getProperty(property, defaultMissing);
+ }
}
[4/4] tika git commit: TIKA-2190 -- add configurability for preserve
interword spacing
Posted by ta...@apache.org.
TIKA-2190 -- add configurability for preserve interword spacing
Project: http://git-wip-us.apache.org/repos/asf/tika/repo
Commit: http://git-wip-us.apache.org/repos/asf/tika/commit/ae44b9e5
Tree: http://git-wip-us.apache.org/repos/asf/tika/tree/ae44b9e5
Diff: http://git-wip-us.apache.org/repos/asf/tika/diff/ae44b9e5
Branch: refs/heads/master
Commit: ae44b9e507dbb11b9b9f5c57cf342b47966ffb66
Parents: c83f87b
Author: tballison <ta...@mitre.org>
Authored: Tue Dec 20 16:32:44 2016 -0500
Committer: tballison <ta...@mitre.org>
Committed: Tue Dec 20 16:32:44 2016 -0500
----------------------------------------------------------------------
CHANGES.txt | 3 ++
.../tika/parser/ocr/TesseractOCRConfig.java | 38 +++++++++++++++++++
.../tika/parser/ocr/TesseractOCRParser.java | 10 +++--
.../parser/ocr/TesseractOCRConfig.properties | 1 +
.../tika/parser/ocr/TesseractOCRParserTest.java | 22 +++++++++++
.../test-documents/testOCR_spacing.png | Bin 0 -> 7232 bytes
6 files changed, 71 insertions(+), 3 deletions(-)
----------------------------------------------------------------------
http://git-wip-us.apache.org/repos/asf/tika/blob/ae44b9e5/CHANGES.txt
----------------------------------------------------------------------
diff --git a/CHANGES.txt b/CHANGES.txt
index e215499..bd6a45e 100644
--- a/CHANGES.txt
+++ b/CHANGES.txt
@@ -1,5 +1,8 @@
Release 1.15 - ??
+ * Add configurability of "preserve-interword-spacing" to
+ TesseractOCRParser (TIKA-2190).
+
* Added experimental SAX parser for .pptx files. To select this parser,
set useSAXPptxExtractor(true) on OfficeParserConfig (TIKA-2210).
http://git-wip-us.apache.org/repos/asf/tika/blob/ae44b9e5/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java
----------------------------------------------------------------------
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java b/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java
index a08b0e2..27853d3 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRConfig.java
@@ -91,6 +91,10 @@ public class TesseractOCRConfig implements Serializable {
// factor by which image is to be scaled.
private int resize = 900;
+ // whether or not to preserve interword spacing
+ private boolean preserveInterwordSpacing = false;
+
+
/**
* Default contructor.
*/
@@ -148,6 +152,7 @@ public class TesseractOCRConfig implements Serializable {
} else if ("hocr".equals(outputTypeString)) {
setOutputType(OUTPUT_TYPE.HOCR);
}
+ setPreserveInterwordSpacing(getProp(props, "preserveInterwordSpacing", false));
// set parameters for ImageMagick
setEnableImageProcessing(
@@ -244,6 +249,22 @@ public class TesseractOCRConfig implements Serializable {
}
/**
+ * Whether or not to maintain interword spacing. Default is <code>false</code>.
+ *
+ * @param preserveInterwordSpacing
+ */
+ public void setPreserveInterwordSpacing(boolean preserveInterwordSpacing) {
+ this.preserveInterwordSpacing = preserveInterwordSpacing;
+ }
+
+ /**
+ *
+ * @return whether or not to maintain interword spacing.
+ */
+ public boolean getPreserveInterwordSpacing() {
+ return preserveInterwordSpacing;
+ }
+ /**
* @see #setMinFileSizeToOcr(int minFileSizeToOcr)
*/
public int getMinFileSizeToOcr() {
@@ -481,4 +502,21 @@ public class TesseractOCRConfig implements Serializable {
private String getProp(Properties properties, String property, String defaultMissing) {
return properties.getProperty(property, defaultMissing);
}
+
+ private boolean getProp(Properties properties, String property, boolean defaultMissing) {
+ String propVal = properties.getProperty(property);
+ if (propVal == null) {
+ return defaultMissing;
+ }
+ if (propVal.equalsIgnoreCase("true")) {
+ return true;
+ } else if (propVal.equalsIgnoreCase("false")) {
+ return false;
+ }
+
+ throw new RuntimeException(String.format(Locale.ROOT,
+ "Cannot parse TesseractOCRConfig variable %s, invalid boolean value: %s",
+ property, propVal));
+ }
+
}
http://git-wip-us.apache.org/repos/asf/tika/blob/ae44b9e5/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRParser.java
----------------------------------------------------------------------
diff --git a/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRParser.java b/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRParser.java
index ffbef1c..46a3f55 100644
--- a/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRParser.java
+++ b/tika-parsers/src/main/java/org/apache/tika/parser/ocr/TesseractOCRParser.java
@@ -417,9 +417,12 @@ public class TesseractOCRParser extends AbstractParser {
* if an input error occurred
*/
private void doOCR(File input, File output, TesseractOCRConfig config) throws IOException, TikaException {
- String[] cmd = { config.getTesseractPath() + getTesseractProg(), input.getPath(), output.getPath(), "-l",
- config.getLanguage(), "-psm", config.getPageSegMode(), config.getOutputType().name().toLowerCase(Locale.US)};
+ String[] cmd = { config.getTesseractPath() + getTesseractProg(), input.getPath(), output.getPath(), "-l",
+ config.getLanguage(), "-psm", config.getPageSegMode(),
+ config.getOutputType().name().toLowerCase(Locale.US),
+ "-c",
+ (config.getPreserveInterwordSpacing())? "preserve_interword_spaces=1" : "preserve_interword_spaces=0"};
ProcessBuilder pb = new ProcessBuilder(cmd);
setEnv(config, pb);
final Process process = pb.start();
@@ -479,8 +482,9 @@ public class TesseractOCRParser extends AbstractParser {
try (Reader reader = new InputStreamReader(stream, UTF_8)) {
char[] buffer = new char[1024];
for (int n = reader.read(buffer); n != -1; n = reader.read(buffer)) {
- if (n > 0)
+ if (n > 0) {
xhtml.characters(buffer, 0, n);
+ }
}
}
xhtml.endElement("div");
http://git-wip-us.apache.org/repos/asf/tika/blob/ae44b9e5/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties
----------------------------------------------------------------------
diff --git a/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties b/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties
index ddcd473..3cc2bcd 100644
--- a/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties
+++ b/tika-parsers/src/main/resources/org/apache/tika/parser/ocr/TesseractOCRConfig.properties
@@ -21,6 +21,7 @@ maxFileSizeToOcr=2147483647
minFileSizeToOcr=0
timeout=120
outputType=txt
+preserveInterwordSpacing=false
# properties for image processing
# to enable processing, set enableImageProcessing to 1
http://git-wip-us.apache.org/repos/asf/tika/blob/ae44b9e5/tika-parsers/src/test/java/org/apache/tika/parser/ocr/TesseractOCRParserTest.java
----------------------------------------------------------------------
diff --git a/tika-parsers/src/test/java/org/apache/tika/parser/ocr/TesseractOCRParserTest.java b/tika-parsers/src/test/java/org/apache/tika/parser/ocr/TesseractOCRParserTest.java
index e0f89ac..4c0ab76 100644
--- a/tika-parsers/src/test/java/org/apache/tika/parser/ocr/TesseractOCRParserTest.java
+++ b/tika-parsers/src/test/java/org/apache/tika/parser/ocr/TesseractOCRParserTest.java
@@ -23,6 +23,8 @@ import static org.junit.Assume.assumeTrue;
import java.io.InputStream;
import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
import org.apache.tika.TikaTest;
import org.apache.tika.metadata.Metadata;
@@ -256,4 +258,24 @@ public class TesseractOCRParserTest extends TikaTest {
}
//TODO: add unit tests for jp2/jpx/ppm TIKA-2174
+
+ @Test
+ public void testInterwordSpacing() throws Exception {
+ assumeTrue(canRun());
+ //default
+ String xml = getXML("testOCR_spacing.png").xml;
+ assertContains("The quick", xml);
+
+ TesseractOCRConfig tesseractOCRConfigconfig = new TesseractOCRConfig();
+ tesseractOCRConfigconfig.setPreserveInterwordSpacing(true);
+ ParseContext parseContext = new ParseContext();
+ parseContext.set(TesseractOCRConfig.class, tesseractOCRConfigconfig);
+
+ //with preserve interwordspacing "on"
+ //allow some flexibility in case Tesseract is computing spaces
+ //somewhat differently in different versions/OS's, etc.
+ xml = getXML("testOCR_spacing.png", parseContext).xml;
+ Matcher m = Pattern.compile("The\\s{5,20}quick").matcher(xml);
+ assertTrue(m.find());
+ }
}
http://git-wip-us.apache.org/repos/asf/tika/blob/ae44b9e5/tika-parsers/src/test/resources/test-documents/testOCR_spacing.png
----------------------------------------------------------------------
diff --git a/tika-parsers/src/test/resources/test-documents/testOCR_spacing.png b/tika-parsers/src/test/resources/test-documents/testOCR_spacing.png
new file mode 100644
index 0000000..10a7e40
Binary files /dev/null and b/tika-parsers/src/test/resources/test-documents/testOCR_spacing.png differ
[2/4] tika git commit: Merge branch 'bug_TIKA-2189' of
https://github.com/dasbipulkumar/tika
Posted by ta...@apache.org.
Merge branch 'bug_TIKA-2189' of https://github.com/dasbipulkumar/tika
Project: http://git-wip-us.apache.org/repos/asf/tika/repo
Commit: http://git-wip-us.apache.org/repos/asf/tika/commit/8e12ebea
Tree: http://git-wip-us.apache.org/repos/asf/tika/tree/8e12ebea
Diff: http://git-wip-us.apache.org/repos/asf/tika/diff/8e12ebea
Branch: refs/heads/master
Commit: 8e12ebea4178048373c3c946eb159d6f68baa4e9
Parents: 2dbd651 40401e5
Author: tballison <ta...@mitre.org>
Authored: Tue Dec 20 15:04:01 2016 -0500
Committer: tballison <ta...@mitre.org>
Committed: Tue Dec 20 15:04:01 2016 -0500
----------------------------------------------------------------------
.../org/apache/tika/parser/ocr/TesseractOCRConfig.properties | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
----------------------------------------------------------------------