Trying to make the link between Csniper and CQP

[simonmeoni]

I try to install CSniper everything until the CQP installation
I download cwb and I install it with the command : ./install-cwb.sh
I modify the properties of CSniper, I add this line : engine.cqp.executable=path/to/cqp/file
I start tomcat and I reached the website succesfully but I cannot select the cqp engine. I have no log of this, I think this is the mistake of me. Anyone have an idea ?
Thanks in advance

[dodinh]

This sounds like tomcat could not find the csniper configuration file.
Are you sure there is no log? Usually, these should be at /var/log/tomcat6/catalina.out (Debian-based).

[simonmeoni]

I see the log and there are no errors about cqp. I think that my configuration is detected because the configuration of db is working (otherwise the csniper server crash i suppose).
I join my docker file and my propertie file if it is help.
PS : I use tomcat 8 :s, I am trying to use tomcat 6 and I come back to you
csniper.tar.gz

[simonmeoni]

I modify my docker file in order to use tomcat 6 and I have this warning during deployment phase :
csniper_1 | 2017-10-09 08:12:09 WARN [main] (PropertiesLoaderSupport:198) - Could not load properties from class path resource [META-INF/csniper.properties]: class path resource [META-INF/csniper.properties] cannot be opened because it does not exist
I have the same problem

(I join my configuration on this comment)
csniper.tar.gz

[reckart]

That is just a warning that you should be able to safely ignore. CSniper tries to load the properties from the classpath in addition of looking for them in its home directory.

[simonmeoni]

Ok thanks Richard. Can you tell me more about the configuration about cwb and cqp ? Where I can find a test corpus in the csniper format ?

[reckart]

We do not have a pre-converted corpus at the moment (would be a good idea), but here is a description which describes the conversion process and lists several corpora that have been tested.

https://dkpro.github.io/dkpro-csniper/documentation/conversion/

[reckart]

It has been a while...

Looking at the code, it looks like you get the choice for a given engine only if there is actually a corpus present that was prepared for this engine.

dkpro-csniper/csniper-webapp/src/main/java/de/tudarmstadt/ukp/csniper/webapp/evaluation/CorpusServiceImpl.java

Lines 115 to 125 in ae36b3a

	public List<SearchEngine> listEngines(String aCorpusId)
	{
	File corpusPath = new File(repositoryPath, aCorpusId);
	List<SearchEngine> filteredEngines = new ArrayList<SearchEngine>();
	for (SearchEngine engine : engines) {
	if (new File(corpusPath, engine.getName()).exists()) {
	filteredEngines.add(engine);
	}
	}
	return filteredEngines;
	}

[simonmeoni]

Ok thank you, I will prepare a corpus !

[simonmeoni]

I tried to prepare data using this sample : http://www.coli.uni-saarland.de/projects/sfb378/negra-corpus/corpus-sample.export.

This is my script :

import de.tudarmstadt.ukp.dkpro.core.api.resources.CompressionMethod;
import de.tudarmstadt.ukp.dkpro.core.io.bincas.SerializedCasWriter;
import de.tudarmstadt.ukp.dkpro.core.io.imscwb.ImsCwbWriter;
import de.tudarmstadt.ukp.dkpro.core.io.negra.NegraExportReader;
import org.apache.uima.UIMAException;
import org.apache.uima.fit.pipeline.SimplePipeline;
import java.io.IOException;

import static org.apache.uima.fit.factory.AnalysisEngineFactory.createPrimitiveDescription;
import static org.apache.uima.fit.factory.CollectionReaderFactory.createDescription;

public class Converter {
    public static void main(String[] args) throws UIMAException, IOException {
        // Collection ID
        String id = args[0];

        // Source file (e.g. tuebadz-5.0.anaphora.export.bz2)
        String source = args[1];

        // Target folder
        String target = args[2];

        SimplePipeline.runPipeline(
                createDescription(NegraExportReader.class,
                        NegraExportReader.PARAM_SOURCE_LOCATION, source,
                        NegraExportReader.PARAM_COLLECTION_ID, id,
                        NegraExportReader.PARAM_LANGUAGE, "de",
                        NegraExportReader.PARAM_ENCODING, "ISO-8859-15",
                        NegraExportReader.PARAM_READ_PENN_TREE, true),

                createPrimitiveDescription(SerializedCasWriter.class,
                        SerializedCasWriter.PARAM_TARGET_LOCATION, target + "/bin",
                        SerializedCasWriter.PARAM_USE_DOCUMENT_ID, true,
                        SerializedCasWriter.PARAM_COMPRESSION, CompressionMethod.XZ),

                createPrimitiveDescription(ImsCwbWriter.class,
                        ImsCwbWriter.PARAM_TARGET_ENCODING, "UTF-8",
                        ImsCwbWriter.PARAM_TARGET_LOCATION, target + "/cqp/",
                        ImsCwbWriter.PARAM_WRITE_TEXT_TAG, true,
                        ImsCwbWriter.PARAM_WRITE_DOCUMENT_TAG, true,
                        ImsCwbWriter.PARAM_WRITE_OFFSETS, true,
                        ImsCwbWriter.PARAM_WRITE_LEMMA, true,
                        ImsCwbWriter.PARAM_WRITE_DOC_ID, false));
    }
}

I used gradle to resolve dependencies. But I have some problem, I obtain two files after excute my jar :
SAMPLE -- cqp (uima CONLL files (i dont know))
'-- bin -- 1.ser.xz (binary files)

(I attach to my post the output tree
SAMPLE.tar.gz
)
csniper logs give an undefined error for the corpus that I have generated.

Can you tell me more about the input format ?

[dodinh]

What ImsCwbWriter produces here is the input format used for the cwb tools [1].
Your cqp file has to be converted using the tools detailed in the link.

The easier alternative (which is not documented on the csniper website, only in the code... mea culpa) is to specify the directory in which the cwb-tools (cwb-make, cwb-encode) reside, with ImsCwbWriter.PARAM_CQP_HOME. Running the pipeline then automatically creates the encoded cqp files, which have to be moved into the correct directories as outlined at the end of [2].

[1] http://cwb.sourceforge.net/files/CWB_Encoding_Tutorial/node3.html
[2] https://dkpro.github.io/dkpro-csniper/documentation/conversion/#example-conversion