Package org.corpus_tools.pepper.modules

Interface PepperImporter

  • All Superinterfaces:
    PepperModule
    All Known Implementing Classes:
    DoNothingImporter, PepperImporterImpl, SaltXMLImporter, TextImporter
    public interface PepperImporter
extends PepperModule

    A mapping task in the Pepper workflow is not a monolithic block. It consists of several smaller steps.

    • Declare the fingerprint of the module. This is part of the constructor.
    • Check readyness of the module.
    • Analyze whether the files in the passed corpus path is importable by this importer.
    • Import the corpus structure.
    • Import the document structure and create a mapper for each corpus and document.
    • clean-up
    The following describes the single steps in short. To get a more detailed explanation, take a look to the documentations found at http://u.hu-berlin.de/ saltnpepper.

    Declare the fingerprint

    Initialize the module and set the modules name, its description and the format description of data which are importable. This is part of the constructor: 
     public MyModule() {
        super("Name of the module");
        setSupplierContact(URI.createURI("Contact address of the module's supplier"));
        setSupplierHomepage(URI.createURI("homepage of the module"));
        setDesc("A short description of what is the intention of this module, for instance which formats are importable. ");
        this.addSupportedFormat("The name of a format which is importable e.g. txt",
                        "The version corresponding to the format name", null);
 }

    Check readyness of the module

    This method is invoked by the Pepper framework before the mapping process is started. This method must return true, otherwise, this Pepper module could not be used in a Pepper workflow. At this point problems which prevent the module from being used you can report all problems to the user, for instance a database connection could not be established. 
     public boolean isReadyToStart() {
        return (true);
 }

    Analyze data

    Depending on the formats you want to support with your importer the detection can be very different. In the simplest case, it only is necessary, to search through the files at the given location (or to recursively traverse through directories, in case the location points to a directory), and to read their header section. For instance some formats like the xml formats PAULA (see: http:// www.sfb632.uni-potsdam.de/en/paula.html ) or TEI (see: http://www.tei-c.org/Guidelines/P5/). The method should return a value between 0 and 1, where 0 means not importable and 1 means definitely importable. If null is returned, Pepper interprets this as unknown and will never suggest this module to the user. 
     public Double isImportable(URI corpusPath) {
        return null;
 }

    Import corpus structure

    The classes PepperImporterImpl and PepperExporterImpl provide an automatic mechanism to im- or export the corpus-structure. This mechanism is adaptable step by step, according to your specific purpose. Since many formats do not care about the corpus-structure and they only encode the document-structure, the corpus-structure is simultaneous to the file structure of a corpus. Pepper's default mapping maps the root-folder to a root-corpus (SCorpus object). A sub-folder then corresponds to a sub-corpus (SCorpus object). The relation between super- and sub-corpus, is represented as a SCorpusRelation object. Following the assumption, that files contain the document-structure, there is one SDocument corresponding to each file in a sub-folder. The SCorpus and the SDocument objects are linked with a SCorpusDocumentRelation.
    For keeping the correspondance between the corpus-structure and the file structure, both the im- and the exporter make use of a map, which can be accessed via getIdentifier2ResourceTable().
    To adapt the behavior, you can set the file endings in the constructor as follows: 
     this.getDocumentEndings().add("file ending");
    You can also add the value PepperModule.ENDING_LEAF_FOLDER to import not files but leaf folders as SDocument objects. Another possibility is to add the value PepperModule.ENDING_ALL_FILES to import all files no matter their ending.

    Import the document structure

    In the method PepperModule.createPepperMapper(Identifier) a PepperMapper object needs to be initialized and returned. The PepperMapper is the major part major part doing the mapping. It provides the methods PepperMapper.mapSCorpus() to handle the mapping of a single SCorpus object and PepperMapper.mapSDocument() to handle a single SDocument object. Both methods are invoked by the Pepper framework. To set the PepperMapper.getResourceURI(), which offers the mapper the file or folder of the current SCorpus or SDocument object, this filed needs to be set in the PepperModule.createPepperMapper(Identifier) method. The following snippet shows a dummy of that method: 
     public PepperMapper createPepperMapper(Identifier sElementId) {
        PepperMapper mapper = new PepperMapperImpl() {
                @Override
                public DOCUMENT_STATUS mapSCorpus() {
                        // handling the mapping of a single corpus
 
                        // accessing the current file or folder
                        getResourceURI();
 
                        // returning, that the corpus was mapped successfully
                        return (DOCUMENT_STATUS.COMPLETED);
                }
 
                @Override
                public DOCUMENT_STATUS mapSDocument() {
                        // handling the mapping of a single document
 
                        // accessing the current file or folder
                        getResourceURI();
 
                        // returning, that the document was mapped successfully
                        return (DOCUMENT_STATUS.COMPLETED);
                }
        };
        // pass current file or folder to mapper. When using
        // PepperImporter.importCorpusStructure or
        // PepperExporter.exportCorpusStructure, the mapping between file or
        // folder
        // and SCorpus or SDocument was stored here
        mapper.setResourceURI(getIdentifier2ResourceTable().get(sElementId));
        return (mapper);
 }

    clean-up

    Sometimes it might be necessary to clean up after the module did the job. For instance when writing an im- or an exporter it might be necessary to close file streams, a db connection etc. Therefore, after the processing is done, the Pepper framework calls the method described in the following snippet: 
     public void end() {
        super.end();
        // do some clean up like closing of streams etc.
 }

    Author:
    Florian Zipser

    • Field Detail

      • NEGATIVE_FILE_EXTENSION_MARKER

        static final String NEGATIVE_FILE_EXTENSION_MARKER
        A character or character sequence to mark a file extension as not to be one of the imported ones.
        See Also:
        Constant Field Values

    • Method Detail

      • getSupportedFormats

        List<FormatDesc> getSupportedFormats()
        Returns a list of formats, which are importable by this PepperImporter object.
        Returns:

      • getCorpusDesc

        CorpusDesc getCorpusDesc()
        TODO docu
        Returns:

      • setCorpusDesc

        void setCorpusDesc​(CorpusDesc corpusDesc)
        TODO docu

      • getIdentifier2ResourceTable

        Map<org.corpus_tools.salt.graph.Identifier,​org.eclipse.emf.common.util.URI> getIdentifier2ResourceTable()
        Stores Identifier objects corresponding to either a SDocument or a SCorpus object, which has been created during the run of importCorpusStructure(SCorpusGraph). Corresponding to the Identifier object this table stores the resource from where the element shall be imported.
        For instance:
        corpus_1 /home/me/corpora/myCorpus
        corpus_2 /home/me/corpora/myCorpus/subcorpus
        doc_1 /home/me/corpora/myCorpus/subcorpus/document1.xml
        doc_2 /home/me/corpora/myCorpus/subcorpus/document2.xml

      • getDocumentEndings

        Collection<String> getDocumentEndings()
        Returns list containing all format endings for files, which are importable and could be mapped to SDocument or SDocumentGraph objects by this Pepper module.
        Returns:
        a collection of endings

      • getCorpusEndings

        Collection<String> getCorpusEndings()
        Returns a collection of all file endings for a SCorpus object. See {@inheritDoc #sCorpusEndings}. This list contains per default value "FOLDER". To remove the default value, call Collection.remove(Object) on getCorpusEndings(). To add endings to the collection, call Collection#add(Ending) and to remove endings from the collection, call Collection#remove(Ending).
        Returns:
        a collection of endings

      • getIgnoreEndings

        Collection<String> getIgnoreEndings()
        Returns a collection of filenames, not to be imported. {@inheritDoc #importIgnoreList} . To add endings to the collection, call Collection#add(Ending) and to remove endings from the collection, call Collection#remove(Ending).
        Returns:
        a collection of endings to be ignored

      • setTypeOfResource

        org.corpus_tools.salt.SALT_TYPE setTypeOfResource​(org.eclipse.emf.common.util.URI resource)
        This method is a callback and can be overridden by derived importers. This method is called via the import of the corpus-structure ( importCorpusStructure(SCorpusGraph)). During the traversal of the file-structure the method importCorpusStructure(SCorpusGraph) calls this method for each resource, to determine if the resource either represents a SCorpus, a SDocument object or shall be ignored.
        If this method is not overridden, the default behavior is:
        Parameters:
        resource - URI resource to be specified
        Returns:
        SALT_TYPE.SCORPUS if resource represents a SCorpus object, SALT_TYPE.SDOCUMENT if resource represents a SDocument object or null, if it shall be igrnored.

      • importCorpusStructure

        void importCorpusStructure​(org.corpus_tools.salt.common.SCorpusGraph corpusGraph)
                    throws PepperModuleException
        This method is called by Pepper at the start of a conversion process to create the corpus-structure. A corpus-structure consists of corpora (represented via the Salt element SCorpus), documents (represented represented via the Salt element SDocument) and a linking between corpora and a corpus and a document (represented via the Salt element SCorpusRelation and SCorpusDocumentRelation ). Each corpus corpus can contain 0..* subcorpus and 0..* documents, but a corpus cannot contain both document and corpus.
        For many cases the creation of the corpus-struccture can be done automatically, therefore, just adopt the two lists #gets
        This method creates the corpus-structure via a top down traversal in file structure. For each found file (real file and folder), the method setTypeOfResource(URI) is called to set the type of the resource. If the type is a SALT_TYPE.SDOCUMENT a SDocument object is created for the resource, if the type is a SALT_TYPE.SCORPUS a SCorpus object is created, if the type is null, the resource is ignored.
        Parameters:
        corpusGraph - an empty graph given by Pepper, which shall contains the corpus structure
        Throws:
        PepperModuleException

      • addSupportedFormat

        FormatDesc addSupportedFormat​(String formatName,
                              String formatVersion,
                              org.eclipse.emf.common.util.URI formatReference)
        {@inheritDoc PepperModuleDesc#addSupportedFormat(String, String, URI)}

      • isImportable

        Double isImportable​(org.eclipse.emf.common.util.URI corpusPath)
        This method is called by Pepper and returns if a corpus located at the given URI is importable by this importer. If yes, 1 must be returned, if no 0 must be returned. If it is not quite sure, if the given corpus is importable by this importer any value between 0 and 1 can be returned. If this method is not overridden, null is returned.
        Returns:
        1 if corpus is importable, 0 if corpus is not importable, 0 < X < 1, if no definitiv answer is possible, null if method is not overridden