Examples with RawIteratingFolderReader ambit2.core.io.RawIteratingFolderReader used on opensource projects

Example 1 with RawIteratingFolderReader

use of ambit2.core.io.RawIteratingFolderReader in project ambit-mirror by ideaconsult.

the class I5ParserTest method testi5dFolder.

@Test
public void testi5dFolder() throws Exception {
    URL url = getClass().getClassLoader().getResource("ambit2/core/data/i5z/RefSub_030913110311");
    File dir = new File(url.getFile());
    File[] files = dir.listFiles(new FileFilter() {

        @Override
        public boolean accept(File pathname) {
            return pathname.getName().endsWith("i5d");
        }
    });
    Assert.assertEquals(20, files.length);
    RawIteratingFolderReader reader = new RawIteratingFolderReader(files);
    int count = 0;
    int foundInChI = 0;
    int foundCas = 0;
    int foundName = 0;
    while (reader.hasNext()) {
        IStructureRecord record = reader.nextRecord();
        count++;
        if (record.getContent() != null && "INC".equals(record.getFormat())) {
            foundInChI++;
            System.out.println(record.getContent());
        }
        for (Property p : record.getRecordProperties()) {
            foundCas += record.getRecordProperty(p).equals("59-87-0") ? 1 : 0;
            foundName += record.getRecordProperty(p).equals("5-nitro-2-furaldehyde semicarbazone") ? 1 : 0;
            System.out.println(p.getName() + " = " + record.getRecordProperty(p));
        }
    // Assert.assertNotNull(record.getSmiles());
    // Assert.assertNotNull(record.getInchi());
    // Assert.assertNotNull(record.getProperty(Property.getI5UUIDInstance()));
    }
    reader.close();
    Assert.assertEquals(10, count);
    Assert.assertEquals(10, foundInChI);
}

Example 2 with RawIteratingFolderReader

use of ambit2.core.io.RawIteratingFolderReader in project ambit-mirror by ideaconsult.

the class I5ParserTest method testi5z.

@Test
public void testi5z() throws Exception {
    URL url = getClass().getClassLoader().getResource("ambit2/core/data/i5z/RefSub_030913110311.i5z");
    IIteratingChemObjectReader ireader = FileInputState.getReader(new File(url.getFile()));
    ireader.setErrorHandler(new IChemObjectReaderErrorHandler() {

        @Override
        public void handleError(String message, int row, int colStart, int colEnd, Exception exception) {
        }

        @Override
        public void handleError(String message, int row, int colStart, int colEnd) {
        }

        @Override
        public void handleError(String message, Exception exception) {
            exception.printStackTrace();
        }

        @Override
        public void handleError(String message) {
        }
    });
    int count = 0;
    int foundCas = 0;
    int foundName = 0;
    RawIteratingFolderReader reader = (RawIteratingFolderReader) ireader;
    while (reader.hasNext()) {
        IStructureRecord record = reader.nextRecord();
        /*
			for (Property p : record.getProperties())
				System.out.println(String.format("%s [%s] = %s",p.getName(),p.getReference().getTitle(),record.getProperty(p)));
			System.out.println();
			*/
        count++;
        for (Property p : record.getRecordProperties()) {
            foundCas += record.getRecordProperty(p).equals("59-87-0") ? 1 : 0;
            foundName += record.getRecordProperty(p).equals("5-nitro-2-furaldehyde semicarbazone") ? 1 : 0;
            System.out.println(p.getName() + " = " + record.getRecordProperty(p));
        }
    // Assert.assertNotNull(record.getSmiles());
    // Assert.assertNotNull(record.getInchi());
    // Assert.assertNotNull(record.getProperty(Property.getI5UUIDInstance()));
    }
    reader.close();
    Assert.assertEquals(10, count);
    Assert.assertEquals(1, foundCas);
    Assert.assertEquals(1, foundName);
}

Example 3 with RawIteratingFolderReader

use of ambit2.core.io.RawIteratingFolderReader in project ambit-mirror by ideaconsult.

the class RawIteratingFolderReaderTest method test.

@Test
public void test() throws Exception {
    File[] files = new File[] { new File("src/test/resources/ambit2/core/data/mdl/12027-77-9.sdf"), new File("src/test/resources/ambit2/core/data/mdl/12040-13-0.sdf"), new File("src/test/resources/ambit2/core/data/mdl/12042-37-4.sdf"), new File("src/test/resources/ambit2/core/data/mdl/12401-47-7.mol"), new File("src/test/resources/ambit2/core/data/mdl/12401-47-7.sdf"), new File("src/test/resources/ambit2/core/data/mdl/polymer.mol") };
    int count = 0;
    RawIteratingFolderReader reader = new RawIteratingFolderReader(files);
    while (reader.hasNext()) {
        logger.info(reader.next().toString());
        count++;
    }
    reader.close();
    Assert.assertEquals(6, count);
}

Example 4 with RawIteratingFolderReader

use of ambit2.core.io.RawIteratingFolderReader in project ambit-mirror by ideaconsult.

the class RepositoryWriterTest method testWriteMultipleFiles_i5d.

@Test
public void testWriteMultipleFiles_i5d() throws Exception {
    setUpDatabaseFromResource("ambit2/db/processors/test/empty-datasets.xml");
    IDatabaseConnection c = getConnection();
    ITable chemicals = c.createQueryTable("EXPECTED", "SELECT * FROM chemicals");
    Assert.assertEquals(0, chemicals.getRowCount());
    ITable strucs = c.createQueryTable("EXPECTED", "SELECT * FROM structure");
    Assert.assertEquals(0, strucs.getRowCount());
    ITable srcdataset = c.createQueryTable("EXPECTED", "SELECT * FROM src_dataset");
    Assert.assertEquals(0, srcdataset.getRowCount());
    ITable struc_src = c.createQueryTable("EXPECTED", "SELECT * FROM struc_dataset");
    Assert.assertEquals(0, struc_src.getRowCount());
    ITable property = c.createQueryTable("EXPECTED", "SELECT * FROM properties");
    Assert.assertEquals(0, property.getRowCount());
    ITable property_values = c.createQueryTable("EXPECTED", "SELECT * FROM property_values");
    Assert.assertEquals(0, property_values.getRowCount());
    File dir = new File("src/test/resources/ambit2/db/processors/i5d");
    FilenameFilter filter = new FilenameFilter() {

        public boolean accept(File dir, String name) {
            return !name.startsWith(".");
        }
    };
    File[] files = dir.listFiles(filter);
    Assert.assertEquals(3, files.length);
    RawIteratingFolderReader reader = new RawIteratingFolderReader(files);
    write(reader, c.getConnection(), new EINECSKey());
    reader.close();
    c.close();
    c = getConnection();
    chemicals = c.createQueryTable("EXPECTED", "SELECT * FROM chemicals");
    Assert.assertEquals(3, chemicals.getRowCount());
    // there are two empty file without $$$$ sign, which are skipped
    strucs = c.createQueryTable("EXPECTED", "SELECT * FROM structure");
    Assert.assertEquals(3, strucs.getRowCount());
    srcdataset = c.createQueryTable("EXPECTED", "SELECT * FROM src_dataset where name='TEST INPUT'");
    Assert.assertEquals(1, srcdataset.getRowCount());
    struc_src = c.createQueryTable("EXPECTED", "SELECT * FROM struc_dataset");
    Assert.assertEquals(3, struc_src.getRowCount());
    property = c.createQueryTable("EXPECTED", "SELECT * FROM properties join catalog_references using(idreference) where name='Names' and title in ('IUCLID5 SYNONYM#2','IUCLID5')");
    Assert.assertEquals(2, property.getRowCount());
    property = c.createQueryTable("EXPECTED", "SELECT * FROM properties join catalog_references using(idreference) order by name");
    // Assert.assertEquals(34,property.getRowCount());
    Assert.assertEquals(7, property.getRowCount());
    property_values = c.createQueryTable("EXPECTED", "SELECT * FROM property_values");
    Assert.assertEquals(19, property_values.getRowCount());
    srcdataset = c.createQueryTable("EXPECTED", "SELECT * FROM src_dataset join template_def using(idtemplate) where name='TEST INPUT'");
    Assert.assertEquals(7, srcdataset.getRowCount());
    ITable p_cas = c.createQueryTable("EXPECTED", "SELECT idchemical,idstructure,value FROM property_values join property_string using(idvalue_string) join properties using(idproperty) where name='CasRN'");
    Assert.assertEquals(3, p_cas.getRowCount());
    ITable p_ec = c.createQueryTable("EXPECTED", "SELECT idchemical,idstructure,value FROM property_values join property_string using(idvalue_string) join properties using(idproperty) where name='EC'");
    Assert.assertEquals(3, p_ec.getRowCount());
    ITable p_uuid = c.createQueryTable("EXPECTED", "SELECT idchemical,idstructure,value FROM property_values join property_string using(idvalue_string) join properties using(idproperty) where name='I5UUID'");
    Assert.assertEquals(3, p_uuid.getRowCount());
    c.close();
}

Example 5 with RawIteratingFolderReader

use of ambit2.core.io.RawIteratingFolderReader in project ambit-mirror by ideaconsult.

the class RepositoryWriterTest method testWriteMultipleFiles.

@Test
public void testWriteMultipleFiles() throws Exception {
    setUpDatabaseFromResource("ambit2/db/processors/test/empty-datasets.xml");
    IDatabaseConnection c = getConnection();
    ITable chemicals = c.createQueryTable("EXPECTED", "SELECT * FROM chemicals");
    Assert.assertEquals(0, chemicals.getRowCount());
    ITable strucs = c.createQueryTable("EXPECTED", "SELECT * FROM structure");
    Assert.assertEquals(0, strucs.getRowCount());
    ITable srcdataset = c.createQueryTable("EXPECTED", "SELECT * FROM src_dataset");
    Assert.assertEquals(0, srcdataset.getRowCount());
    ITable struc_src = c.createQueryTable("EXPECTED", "SELECT * FROM struc_dataset");
    Assert.assertEquals(0, struc_src.getRowCount());
    ITable property = c.createQueryTable("EXPECTED", "SELECT * FROM properties");
    Assert.assertEquals(0, property.getRowCount());
    ITable property_values = c.createQueryTable("EXPECTED", "SELECT * FROM property_values");
    Assert.assertEquals(0, property_values.getRowCount());
    File dir = new File("src/test/resources/ambit2/db/processors/sdf");
    FilenameFilter filter = new FilenameFilter() {

        public boolean accept(File dir, String name) {
            return !name.startsWith(".");
        }
    };
    File[] files = dir.listFiles(filter);
    Assert.assertEquals(12, files.length);
    RawIteratingFolderReader reader = new RawIteratingFolderReader(files);
    write(reader, c.getConnection());
    reader.close();
    c.close();
    c = getConnection();
    chemicals = c.createQueryTable("EXPECTED", "SELECT * FROM chemicals");
    Assert.assertEquals(9, chemicals.getRowCount());
    // there are two empty file without $$$$ sign, which are skipped
    strucs = c.createQueryTable("EXPECTED", "SELECT * FROM structure");
    Assert.assertEquals(17, strucs.getRowCount());
    srcdataset = c.createQueryTable("EXPECTED", "SELECT * FROM src_dataset where name='TEST INPUT'");
    Assert.assertEquals(1, srcdataset.getRowCount());
    struc_src = c.createQueryTable("EXPECTED", "SELECT * FROM struc_dataset");
    Assert.assertEquals(17, struc_src.getRowCount());
    property = c.createQueryTable("EXPECTED", "SELECT * FROM properties");
    // Assert.assertEquals(34,property.getRowCount());
    Assert.assertEquals(214, property.getRowCount());
    property_values = c.createQueryTable("EXPECTED", "SELECT * FROM property_values");
    Assert.assertEquals(370, property_values.getRowCount());
    srcdataset = c.createQueryTable("EXPECTED", "SELECT * FROM src_dataset join template_def using(idtemplate) where name='TEST INPUT'");
    Assert.assertEquals(214, srcdataset.getRowCount());
    ITable p_cas = c.createQueryTable("EXPECTED", "SELECT idchemical,idstructure,name,value FROM structure join values_string using(idstructure) where name=\"CasRN\"");
    Assert.assertEquals(12, p_cas.getRowCount());
    c.close();
/**
 * Removing redundant properties insert ignore into property_values
 * select id,idproperty,idstructure,idvalue,idtype,user_name,status from
 * property_values where idstructure>3 on duplicate key update
 * idstructure=3 delete from property_values where idstructure>3
 *
 * insert ignore into struc_dataset select idstructure,id_srcdataset
 * from struc_dataset where idstructure>3 on duplicate key update
 * idstructure=3 delete from struc_dataset where idstructure>3
 */
}