tech.tablesaw.api.Table#create

Source File: AnalyticQueryTest.java From tablesaw with Apache License 2.0

6 votes

@Test
public void executeInPlaceNumbering() {
  Table table = Table.create("table", StringColumn.create("col1", new String[] {}));

  AnalyticQuery.numberingQuery()
      .from(table)
      .partitionBy()
      .orderBy("col1")
      .rowNumber()
      .as("rowNumber")
      .rank()
      .as("rank")
      .denseRank()
      .as("denseRank")
      .executeInPlace();

  assertEquals(ImmutableList.of("col1", "rowNumber", "rank", "denseRank"), table.columnNames());
}

Source File: TableTransposeTest.java From tablesaw with Apache License 2.0

6 votes

@Test
void transposeUseFirstColumnForHeadings() {
  Table testTable =
      Table.create(
          TABLE_NAME,
          StringColumn.create("label", "row1", "row2", "row3"),
          DoubleColumn.create("value1", 1.0, 1.1, 1.2),
          DoubleColumn.create("value2", 2.0, 2.1, 2.2));
  Table result = testTable.transpose(false, true);

  assertTableEquals(
      Table.create(
          TABLE_NAME,
          DoubleColumn.create("row1", 1.0, 2.0),
          DoubleColumn.create("row2", 1.1, 2.1),
          DoubleColumn.create("row3", 1.2, 2.2)),
      result);
}

Source File: AnalyticQueryEngineTest.java From tablesaw with Apache License 2.0

6 votes

@Test
public void countWithStrings() {
  Table table =
      Table.create(
          "table",
          StringColumn.create("col1", new String[] {"A", "B", null, "C", "C", "C", "D"}));

  AnalyticQuery query =
      AnalyticQuery.quickQuery()
          .from(table)
          .rowsBetween()
          .unboundedPreceding()
          .andCurrentRow()
          .count("col1")
          .as("count")
          .build();

  AnalyticQueryEngine queryEngine = AnalyticQueryEngine.create(query);
  Table result = queryEngine.execute();

  assertEquals(ImmutableList.of(1, 2, 2, 3, 4, 5, 6), result.intColumn("count").asList());
}

Source File: AnalyticQueryTest.java From tablesaw with Apache License 2.0

6 votes

@Test
public void executeInPlaceNumbering() {
  Table table = Table.create("table", StringColumn.create("col1", new String[] {}));

  AnalyticQuery.numberingQuery()
      .from(table)
      .partitionBy()
      .orderBy("col1")
      .rowNumber()
      .as("rowNumber")
      .rank()
      .as("rank")
      .denseRank()
      .as("denseRank")
      .executeInPlace();

  assertEquals(ImmutableList.of("col1", "rowNumber", "rank", "denseRank"), table.columnNames());
}

Source File: Summarizer.java From tablesaw with Apache License 2.0

6 votes

/**
 * Returns an object capable of summarizing the given columns in the given sourceTable, by
 * applying the given functions
 */
public Summarizer(
    Table sourceTable,
    Column<?> column1,
    Column<?> column2,
    Column<?> column3,
    Column<?> column4,
    AggregateFunction<?, ?>... functions) {
  Preconditions.checkArgument(!sourceTable.isEmpty(), "The table to summarize is empty.");
  Table tempTable = Table.create(sourceTable.name());
  tempTable.addColumns(column1);
  tempTable.addColumns(column2);
  tempTable.addColumns(column3);
  tempTable.addColumns(column4);
  this.temp = tempTable;
  this.original = sourceTable;
  summarizedColumns.add(column1.name());
  summarizedColumns.add(column2.name());
  summarizedColumns.add(column3.name());
  summarizedColumns.add(column4.name());
  this.reductions = functions;
}

Source File: AnalyticQueryTest.java From tablesaw with Apache License 2.0

6 votes

@Test
public void executeInPlaceWithDuplicateColumnsThrows() {
  Table table = Table.create("myTable", DoubleColumn.create("col1", new Double[] {}));

  Throwable thrown =
      assertThrows(
          IllegalArgumentException.class,
          () ->
              AnalyticQuery.query()
                  .from(table)
                  .partitionBy()
                  .orderBy("col1")
                  .rowsBetween()
                  .unboundedPreceding()
                  .andUnBoundedFollowing()
                  .sum("col1")
                  .as("col1")
                  .executeInPlace());

  assertTrue(thrown.getMessage().contains("Cannot add column with duplicate name"));
}

Source File: SortTest.java From tablesaw with Apache License 2.0

5 votes

@Test
public void createSortInvalidPrefixColumnExists() {
  Table table = Table.create("t", DoubleColumn.create("col1"));
  Throwable thrown = assertThrows(IllegalStateException.class, () -> Sort.create(table, "<col1"));

  assertEquals("Column prefix: < is unknown.", thrown.getMessage());
}

Source File: DataFramePrinterTest.java From tablesaw with Apache License 2.0

5 votes

@Test
public void printWithSmallFloatNumber() {
  FloatColumn col = FloatColumn.create("testCol");
  col.append(0.000003f);
  Table table = Table.create("small float table", col);
  String out = table.print();
  assertTrue(out.contains("0.000003"));
}

Source File: ByteDictionaryMap.java From tablesaw with Apache License 2.0

5 votes

/** */
@Override
public Table countByCategory(String columnName) {
  Table t = Table.create("Column: " + columnName);
  StringColumn categories = StringColumn.create("Category");
  IntColumn counts = IntColumn.create("Count");
  // Now uses the keyToCount map
  for (Map.Entry<Byte, Integer> entry : keyToCount.byte2IntEntrySet()) {
    categories.append(getValueForKey(entry.getKey()));
    counts.append(entry.getValue());
  }
  t.addColumns(categories);
  t.addColumns(counts);
  return t;
}

Source File: DataFrameReaderTest.java From tablesaw with Apache License 2.0

5 votes

@Test
public void csv() throws IOException {
  Path path = mockFileHelper("/data/file.csv", ImmutableList.of("region", "canada", "us"));
  Table expected = Table.create(StringColumn.create("region", new String[] {"canada", "us"}));
  Table actual = Table.read().csv(Files.newInputStream(path));
  assertEquals(expected.columnNames(), actual.columnNames());
  assertEquals(expected.stringColumn(0).asList(), actual.stringColumn(0).asList());
}

Source File: GettingStarted.java From tablesaw with Apache License 2.0

5 votes

private void summarizing() {
  Table table = Table.create("table");

  try {
    // @@ summarize_basic
    Table summary = table.summarize("sales", mean, sum, min, max).by("province", "status");
    // @@ summarize_basic

    // @@ summarize_calculated_column
    summary = table.summarize("sales", mean, median)
        .by(table.dateColumn("sales date").dayOfWeek());
    // @@ summarize_calculated_column

  } catch (Exception e){}
}

Source File: LocalTimeFilterTest.java From tablesaw with Apache License 2.0

5 votes

@Test
public void testAfter() {
  Table t = Table.create("test");
  t.addColumns(column1);
  column1.appendCell("05:15:30");
  column1.appendCell("10:15:30");
  Table result = t.where(t.timeColumn("Game time").isAfter(LocalTime.of(7, 4, 2, 0)));
  assertEquals(1, result.rowCount());
}

Source File: TableSummaryTest.java From tablesaw with Apache License 2.0

5 votes

@Test
public void summaryMixedTypes() {
  Table testTable =
      Table.create(
          "Data",
          StringColumn.create("label", "yellow", "yellow", "green"),
          DoubleColumn.create("value1", 1.0, 1.1, 1.2),
          BooleanColumn.create("truthy", true, false, true),
          DateColumn.create(
              "dates",
              new LocalDate[] {
                LocalDate.of(2001, 1, 1), LocalDate.of(2002, 1, 1), LocalDate.of(2001, 1, 1)
              }));
  Table result = testTable.summary();
  assertEquals(
      "                                   Data                                    \n"
          + "  Summary   |  label   |         value1         |  truthy  |    dates     |\n"
          + "---------------------------------------------------------------------------\n"
          + "     Count  |       3  |                     3  |          |           3  |\n"
          + "    Unique  |       2  |                        |          |              |\n"
          + "       Top  |  yellow  |                        |          |              |\n"
          + " Top Freq.  |       2  |                        |          |              |\n"
          + "       sum  |          |                   3.3  |          |              |\n"
          + "      Mean  |          |                   1.1  |          |              |\n"
          + "       Min  |          |                     1  |          |              |\n"
          + "       Max  |          |                   1.2  |          |              |\n"
          + "     Range  |          |   0.19999999999999996  |          |              |\n"
          + "  Variance  |          |  0.009999999999999995  |          |              |\n"
          + "  Std. Dev  |          |   0.09999999999999998  |          |              |\n"
          + "     false  |          |                        |       1  |              |\n"
          + "      true  |          |                        |       2  |              |\n"
          + "   Missing  |          |                        |          |           0  |\n"
          + "  Earliest  |          |                        |          |  2001-01-01  |\n"
          + "    Latest  |          |                        |          |  2002-01-01  |",
      result.print());
}

Source File: Summarizer.java From tablesaw with Apache License 2.0

5 votes

/**
 * Returns an object capable of summarizing the given column in the given sourceTable, by applying
 * the given functions
 */
public Summarizer(Table sourceTable, Column<?> column, AggregateFunction<?, ?>... functions) {
  Table tempTable = Table.create(sourceTable.name());
  tempTable.addColumns(column);
  this.temp = tempTable;
  this.original = sourceTable;
  summarizedColumns.add(column.name());
  this.reductions = functions;
}

Source File: CsvWriterTest.java From tablesaw with Apache License 2.0

5 votes

@Test
public void toWriterWithExtension() throws IOException {
  StringColumn colA = StringColumn.create("colA", ImmutableList.of("a", "b"));
  StringColumn colB = StringColumn.create("colB", ImmutableList.of("1", "2"));
  Table table = Table.create("testTable", colA, colB);
  StringWriter writer = new StringWriter();
  table.write().toWriter(writer, "csv");
  assertEquals("colA,colB\na,1\nb,2\n", writer.toString().replaceAll("\\r\\n", "\n"));
}

Source File: DataFramePrinterTest.java From tablesaw with Apache License 2.0

5 votes

@Test
public void printWithSmallFloatNumber() {
  FloatColumn col = FloatColumn.create("testCol");
  col.append(0.000003f);
  Table table = Table.create("small float table", col);
  String out = table.print();
  assertTrue(out.contains("0.000003"));
}

Source File: TableSliceGroup.java From tablesaw with Apache License 2.0

4 votes

public static Table summaryTableName(Table source) {
  return Table.create(source.name() + " summary");
}

Source File: DateFiltersTest.java From tablesaw with Apache License 2.0

4 votes

@Test
public void testColumnComparisons() {
  LocalDate dateTime = LocalDate.of(2015, 1, 25);
  DateColumn dateColumn = DateColumn.create("test");

  LocalDate beforeDate = dateTime.minusDays(1);
  LocalDate afterDate = dateTime.plusDays(1);

  dateColumn.append(beforeDate);
  dateColumn.append(dateTime);
  dateColumn.append(afterDate);

  DateColumn same = DateColumn.create("same");
  same.append(beforeDate);
  same.append(dateTime);
  same.append(afterDate);

  DateColumn before = DateColumn.create("before");
  before.append(beforeDate.minusDays(1));
  before.append(dateTime.minusDays(1));
  before.append(afterDate.minusDays(1));

  DateColumn after = DateColumn.create("after");
  after.append(beforeDate.plusDays(1));
  after.append(dateTime.plusDays(1));
  after.append(afterDate.plusDays(1));

  Table t = Table.create("test", dateColumn, same, before, after);

  assertTrue(dateColumn.isOnOrAfter(same).contains(0));
  assertTrue(t.dateColumn("test").isOnOrAfter(same).contains(0));
  assertTrue(t.dateColumn("test").isOnOrAfter(t.dateColumn("same")).contains(0));

  assertTrue(dateColumn.isOnOrBefore(same).contains(0));
  assertTrue(t.dateColumn("test").isOnOrBefore(same).contains(0));
  assertTrue(t.dateColumn("test").isOnOrBefore(t.dateColumn("same")).contains(0));

  assertTrue(dateColumn.isEqualTo(same).contains(0));
  assertTrue(t.dateColumn("test").isEqualTo(same).contains(0));
  assertTrue(t.dateColumn("test").isEqualTo(t.dateColumn("same")).contains(0));

  assertTrue(dateColumn.isBefore(after).contains(0));
  assertFalse(dateColumn.isOnOrAfter(after).contains(0));
  assertTrue(t.dateColumn("test").isBefore(after).contains(0));
  assertTrue(t.dateColumn("test").isBefore(t.dateColumn("after")).contains(0));

  assertTrue(dateColumn.isAfter(before).contains(0));
  assertFalse(dateColumn.isOnOrBefore(before).contains(0));
  assertTrue(t.dateColumn("test").isAfter(before).contains(0));
  assertTrue(t.dateColumn("test").isAfter(t.dateColumn("before")).contains(0));

  assertFalse(dateColumn.isNotEqualTo(same).contains(0));
  assertFalse(t.dateColumn("test").isNotEqualTo(same).contains(0));
  assertFalse(t.dateColumn("test").isNotEqualTo(t.dateColumn("same")).contains(0));

  assertTrue(dateColumn.isOnOrBefore(same).contains(0));
  assertTrue(dateColumn.isOnOrBefore(after).contains(0));
  assertFalse(dateColumn.isOnOrBefore(before).contains(0));
  assertTrue(dateColumn.isNotEqualTo(before).contains(0));

  assertTrue(dateColumn.isOnOrAfter(same).contains(1));
  assertTrue(dateColumn.isOnOrAfter(before).contains(2));
  assertFalse(dateColumn.isOnOrAfter(after).contains(2));
  assertTrue(dateColumn.isNotEqualTo(after).contains(0));

  /*
          assertTrue(t.dateColumn("test")
                  .isOnOrAfter(t.dateColumn("same")).contains(0));
          assertTrue(t.dateColumn("test")
                  .isOnOrAfter(same).contains(0));

          assertFalse(t.dateColumn("test")
                  .isOnOrAfter(t.dateColumn("after")).contains(0));
          assertFalse(t.dateColumn("test")
                  .isOnOrAfter(after).contains(0));

          assertTrue(t.dateColumn("test")
                  .isOnOrBefore(t.dateColumn("same")).contains(0));
          assertTrue(t.dateColumn("test")
                  .isOnOrBefore(same).contains(0));

          assertTrue(t.dateColumn("test")
                  .isOnOrBefore(t.dateColumn("after")).contains(0));
          assertTrue(t.dateColumn("test")
                  .isOnOrBefore(after).contains(0));

          assertFalse(t.dateColumn("test")
                  .isOnOrBefore(t.dateColumn("before")).contains(0));
          assertFalse(t.dateColumn("test")
                  .isOnOrBefore(before).contains(0));
  */

  assertTrue(t.dateColumn("test").isNotEqualTo(t.dateColumn("before")).contains(0));
  assertTrue(t.dateColumn("test").isNotEqualTo(before).contains(0));
  assertFalse(t.dateColumn("test").isNotEqualTo(t.dateColumn("same")).contains(0));
  assertFalse(t.dateColumn("test").isNotEqualTo(same).contains(0));
}

Source File: PivotTable.java From tablesaw with Apache License 2.0

4 votes

public static Table pivot(
    Table table,
    CategoricalColumn<?> column1,
    CategoricalColumn<?> column2,
    NumericColumn<?> values,
    AggregateFunction<?, ?> aggregateFunction) {

  TableSliceGroup tsg = table.splitOn(column1);

  Table pivotTable = Table.create("Pivot: " + column1.name() + " x " + column2.name());
  pivotTable.addColumns(column1.type().create(column1.name()));

  List<String> valueColumnNames = getValueColumnNames(table, column2);

  for (String colName : valueColumnNames) {
    pivotTable.addColumns(DoubleColumn.create(colName));
  }

  int valueIndex = table.columnIndex(column2);
  int keyIndex = table.columnIndex(column1);

  String key;

  for (TableSlice slice : tsg.getSlices()) {
    key = String.valueOf(slice.get(0, keyIndex));
    pivotTable.column(0).appendCell(key);

    Map<String, Double> valueMap =
        getValueMap(column1, column2, values, valueIndex, slice, aggregateFunction);

    for (String columnName : valueColumnNames) {
      Double aDouble = valueMap.get(columnName);
      NumericColumn<?> pivotValueColumn = pivotTable.numberColumn(columnName);
      if (aDouble == null) {
        pivotValueColumn.appendMissing();
      } else {
        pivotValueColumn.appendObj(aDouble);
      }
    }
  }

  return pivotTable;
}

Source File: HtmlReader.java From tablesaw with Apache License 2.0

4 votes

@Override
public Table read(HtmlReadOptions options) throws IOException {
  Document doc;
  InputStream inputStream = options.source().inputStream();
  if (inputStream != null) {
    // Reader must support mark, so can't use InputStreamReader
    // Parse the InputStream directly
    doc = Jsoup.parse(inputStream, null, "");
  } else {
    doc = Parser.htmlParser().parseInput(options.source().createReader(null), "");
  }
  Elements tables = doc.select("table");
  int tableIndex = 0;
  if (tables.size() != 1) {
    if (options.tableIndex() != null) {
      if (options.tableIndex() >= 0 && options.tableIndex() < tables.size()) {
        tableIndex = options.tableIndex();
      } else {
        throw new IndexOutOfBoundsException(
            "Table index outside bounds. The URL has " + tables.size() + " tables");
      }
    } else {
      throw new IllegalStateException(
          tables.size()
              + " tables found. When more than one html table is present on the page you must specify the index of the table to read from.");
    }
  }
  Element htmlTable = tables.get(tableIndex);

  List<String[]> rows = new ArrayList<>();
  for (Element row : htmlTable.select("tr")) {
    Elements headerCells = row.getElementsByTag("th");
    Elements cells = row.getElementsByTag("td");
    String[] nextLine =
        Stream.concat(headerCells.stream(), cells.stream())
            .map(Element::text)
            .toArray(String[]::new);
    rows.add(nextLine);
  }

  Table table = Table.create(options.tableName());

  if (rows.isEmpty()) {
    return table;
  }

  List<String> columnNames = new ArrayList<>();
  if (options.header()) {
    String[] headerRow = rows.remove(0);
    for (int i = 0; i < headerRow.length; i++) {
      columnNames.add(headerRow[i]);
    }
  } else {
    for (int i = 0; i < rows.get(0).length; i++) {
      columnNames.add("C" + i);
    }
  }

  return TableBuildingUtils.build(columnNames, rows, options);
}

Java Code Examples for tech.tablesaw.api.Table#create()