问题
I try to do OCR and output as PDF using Tess4J and the following code on Linux (Ubuntu 16 Xenial).
public void testOcr() throws Exception {
File imageFile = new File("/projects/de.conradt.core/tessdata/urkunde.jpg");
ITesseract instance = new Tesseract1(); // tried both Tesseract() and Tesseract1()
// File tessDataFolder = LoadLibs.extractTessResources("tessdata"); // Maven build bundles English data
// instance.setDatapath(tessDataFolder.getParent());
instance.setDatapath("/projects/de.conradt.core/tessdata");
instance.setLanguage("deu");
try {
String result = instance.doOCR(imageFile);
System.out.println(result);
} catch (TesseractException e) {
System.err.println(e.getMessage());
}
List<ITesseract.RenderedFormat> list = new ArrayList<ITesseract.RenderedFormat>();
list.add(ITesseract.RenderedFormat.PDF);
File pdfFile = new File("/projects/de.conradt.core/tessdata/urkunde.pdf");
instance.createDocuments(pdfFile.getAbsolutePath(), "/projects/de.conradt.core/tessdata/urkunde", list);
}
The last line
instance.createDocuments(pdfFile.getAbsolutePath(), "/projects/de.conradt.core/tessdata/urkunde", list);
throws an Exception:
11:03:12.651 [http-nio-8080-exec-1] ERROR net.sourceforge.tess4j.Tesseract - Invalid calling convention 63
java.lang.IllegalArgumentException: Invalid calling convention 63
at com.sun.jna.Native.createNativeCallback(Native Method)
at com.sun.jna.CallbackReference.<init>(CallbackReference.java:239)
at com.sun.jna.CallbackReference.getFunctionPointer(CallbackReference.java:413)
at com.sun.jna.CallbackReference.getFunctionPointer(CallbackReference.java:395)
at com.sun.jna.Function.convertArgument(Function.java:541)
at com.sun.jna.Function.invoke(Function.java:305)
at com.sun.jna.Library$Handler.invoke(Library.java:236)
at com.sun.proxy.$Proxy89.gsapi_set_stdio(Unknown Source)
at org.ghost4j.Ghostscript.initialize(Ghostscript.java:323)
at net.sourceforge.tess4j.util.PdfUtilities.convertPdf2Png(PdfUtilities.java:103)
at net.sourceforge.tess4j.util.PdfUtilities.convertPdf2Tiff(PdfUtilities.java:48)
at net.sourceforge.tess4j.Tesseract.createDocuments(Tesseract.java:535)
at net.sourceforge.tess4j.Tesseract.createDocuments(Tesseract.java:507)
at de.conradt.core.Example.testOcr(Example.java:62)
at de.conradt.core.Example.ocr(Example.java:35)
I found this to be a known (but supposedly closed) issue with Tess4J:
- https://github.com/nguyenq/tess4j/issues/35
- https://sourceforge.net/p/tess4j/discussion/1202294/thread/2a25344c/
- https://github.com/zippy1978/ghost4j/issues/44
but I checked my versions as well as the TESSDATA_PREFIX env variable. It's all set correctly as far as I can see.
Tesseract and Leptonica version:
$ /usr/bin/tesseract --version
tesseract 3.04.01
leptonica-1.73
libgif 5.1.2 : libjpeg 8d (libjpeg-turbo 1.4.2) : libpng 1.2.54 : libtiff 4.0.6 : zlib 1.2.8 : libwebp 0.4.4 : libopenjp2 2.1.0
Ghostscript version: (this is the latest version I get via apt-get
)
$ ghostscript -v
GPL Ghostscript 9.18 (2015-10-05)
Copyright (C) 2015 Artifex Software, Inc. All rights reserved.
Tess4j version:
3.2.1
and the TESSDATA_PREFIX (the config files etc. are under /projects/de.conradt.core/tessdata):
$ echo $TESSDATA_PREFIX
/projects/de.conradt.core
Looking at the Release log of Tess4j: http://tess4j.sourceforge.net/changelog.html, I should be using the correct version stack. Especially version 3.2 in the change log says:
Version 3.2 - 15 May 2016: Revert JNA to 4.1.0 due to "Invalid calling convention 63" errors invoking GhostScript via Ghost4J on Linux
so I thought I should be safe with 3.2.1.
Do I need to manually set anything about JNA? From my understanding, this had been fixed in 3.2.0 for Linux explicitly.
回答1:
Ok, I didn't explicitly reference JNA anywhere in my project pom, I thought this is all done in Tess4J 3.2.1 and its pom.xml. I added JNA 4.1.0 as a dependency in my own pom.xml now as well and this seems to solve the problem.
<dependency>
<groupId>net.java.dev.jna</groupId>
<artifactId>jna</artifactId>
<version>4.1.0</version>
</dependency>
来源:https://stackoverflow.com/questions/40361873/tess4j-invalid-calling-convention-63-despite-correct-versions