Examples with Format org.talend.dataprep.schema.Format used on opensource projects

Example 11 with Format

use of org.talend.dataprep.schema.Format in project data-prep by Talend.

the class XlsDetectorTest method should_detect_new_xls_format.

@Test
public void should_detect_new_xls_format() throws Exception {
    try (InputStream inputStream = this.getClass().getResourceAsStream("test_new.xlsx")) {
        Format actual = xlsDetector.detect(inputStream);
        assertTrue(actual.getFormatFamily() instanceof XlsFormatFamily);
        assertTrue(StringUtils.equals("UTF-8", actual.getEncoding()));
    }
}

Example 12 with Format

use of org.talend.dataprep.schema.Format in project data-prep by Talend.

the class XlsSerializerTest method read_xls_TDP_143.

@Test
public void read_xls_TDP_143() throws Exception {
    String fileName = "state_table.xls";
    Format format;
    try (InputStream inputStream = this.getClass().getResourceAsStream(fileName)) {
        format = formatDetector.detect(inputStream);
        Assert.assertNotNull(format);
        Assert.assertTrue(format.getFormatFamily() instanceof XlsFormatFamily);
        Assert.assertEquals(XlsFormatFamily.MEDIA_TYPE, format.getFormatFamily().getMediaType());
    }
    try (InputStream inputStream = this.getClass().getResourceAsStream(fileName)) {
        List<ColumnMetadata> columnsMetadata = xlsSchemaParser.parse(getRequest(inputStream, "#852")).getSheetContents().get(0).getColumnMetadatas();
        logger.debug("columnsMetadata: {}", columnsMetadata);
        Assertions.assertThat(columnsMetadata).isNotNull().isNotEmpty().hasSize(17);
    }
}

Example 13 with Format

use of org.talend.dataprep.schema.Format in project data-prep by Talend.

the class XlsSerializerTest method read_xls_file_then_serialize.

@Test
public void read_xls_file_then_serialize() throws Exception {
    String fileName = "test.xls";
    Format format;
    DataSetMetadata dataSetMetadata = metadataBuilder.metadata().id("beer").build();
    assertFormat(fileName);
    try (InputStream inputStream = this.getClass().getResourceAsStream(fileName)) {
        List<ColumnMetadata> columnMetadatas = // 
        xlsSchemaParser.parse(// 
        getRequest(inputStream, "#123")).getSheetContents().get(0).getColumnMetadatas();
        dataSetMetadata.getRowMetadata().setColumns(columnMetadatas);
    }
    List<Map<String, String>> values = getValuesFromFile(fileName, dataSetMetadata);
    logger.debug("values: {}", values);
    // expected*
    // {country=Australie, note=10.0, beer name =Little Creatures, quality=Awesome}
    // {country=France , note=, beer name =Heinekein, quality=crappy}
    // {country=Australie, note=6.0, beer name =Foo, quality=10.0}
    // {country=France , note=2.0, beer name =Bar, quality=crappy}
    Assertions.assertThat(values).isNotEmpty().hasSize(4);
    // 
    Assertions.assertThat(values.get(0)).contains(// 
    entry("0000", "Little Creatures"), // 
    entry("0001", "Australie"), // 
    entry("0002", "Awesome"), // 
    entry("0003", "10"));
    // 
    Assertions.assertThat(values.get(1)).contains(// 
    entry("0000", "Heinekein"), // 
    entry("0001", "France"), // 
    entry("0002", "crappy"), // 
    entry("0003", ""));
    // 
    Assertions.assertThat(values.get(2)).contains(// 
    entry("0000", "Foo"), // 
    entry("0001", "Australie"), // 
    entry("0002", "10"), entry("0003", "6"));
    // 
    Assertions.assertThat(values.get(3)).contains(// 
    entry("0000", "Bar"), // 
    entry("0001", "France"), // 
    entry("0002", "crappy"), entry("0003", "2"));
}

Example 14 with Format

use of org.talend.dataprep.schema.Format in project data-prep by Talend.

the class XlsSerializerTest method testGeneralNumberFormat_TDP_222.

/**
 * <p>
 * See <a href="https://jira.talendforge.org/browse/TDP-222">https://jira.talendforge.org/browse/TDP-222</a>.
 * </p>
 * <p>
 * XlsSerializer should follow the data format as set in the Excel file. This test ensures XlsSerializer follows the
 * data format as defined and don't directly use {@link Cell#getNumericCellValue()}.
 * </p>
 */
@Test
public void testGeneralNumberFormat_TDP_222() throws Exception {
    final DataSetMetadata metadata = metadataBuilder.metadata().id("1234").row(column().name("id").id(0).type(Type.INTEGER), column().name("value1").id(1).type(Type.INTEGER)).build();
    Format format = assertFormat("excel_numbers.xls");
    // Test number serialization in XLS type guess
    InputStream input = this.getClass().getResourceAsStream("excel_numbers.xls");
    final String result = IOUtils.toString(format.getFormatFamily().getSerializer().serialize(input, metadata, -1), UTF_8);
    final String expected = "[{\"0000\":\"1\",\"0001\":\"123\"},{\"0000\":\"2\",\"0001\":\"123.1\"},{\"0000\":\"3\",\"0001\":\"209.9\"}]";
    assertThat(result, sameJSONAs(expected));
}