embulk-parser-poi_excel 0.1.0

data/src/main/java/org/embulk/parser/poi_excel/visitor/PoiExcelColumnVisitor.java

@@ -0,0 +1,146 @@
+package org.embulk.parser.poi_excel.visitor;
+
+import java.text.MessageFormat;
+import java.util.List;
+
+import org.apache.poi.ss.usermodel.Cell;
+import org.apache.poi.ss.usermodel.Row;
+import org.apache.poi.ss.util.CellReference;
+import org.embulk.parser.poi_excel.PoiExcelColumnValueType;
+import org.embulk.parser.poi_excel.PoiExcelParserPlugin.ColumnOptionTask;
+import org.embulk.parser.poi_excel.PoiExcelParserPlugin.PluginTask;
+import org.embulk.parser.poi_excel.visitor.embulk.CellVisitor;
+import org.embulk.spi.Column;
+import org.embulk.spi.ColumnVisitor;
+import org.embulk.spi.PageBuilder;
+
+public class PoiExcelColumnVisitor implements ColumnVisitor {
+
+	protected final PoiExcelVisitorValue visitorValue;
+	protected final PageBuilder pageBuilder;
+	protected final PoiExcelVisitorFactory factory;
+
+	protected Row currentRow;
+
+	public PoiExcelColumnVisitor(PoiExcelVisitorValue visitorValue) {
+		this.visitorValue = visitorValue;
+		this.pageBuilder = visitorValue.getPageBuilder();
+		this.factory = visitorValue.getVisitorFactory();
+
+		initializeColumnOptions();
+	}
+
+	protected void initializeColumnOptions() {
+		PluginTask task = visitorValue.getPluginTask();
+		List<ColumnOptionTask> options = visitorValue.getColumnOptions();
+		new PoiExcelColumnIndex().initializeColumnIndex(task, options);
+	}
+
+	public void setRow(Row row) {
+		this.currentRow = row;
+	}
+
+	@Override
+	public final void booleanColumn(Column column) {
+		visitCell0(column, factory.getBooleanCellVisitor());
+	}
+
+	@Override
+	public final void longColumn(Column column) {
+		visitCell0(column, factory.getLongCellVisitor());
+	}
+
+	@Override
+	public final void doubleColumn(Column column) {
+		visitCell0(column, factory.getDoubleCellVisitor());
+	}
+
+	@Override
+	public final void stringColumn(Column column) {
+		visitCell0(column, factory.getStringCellVisitor());
+	}
+
+	@Override
+	public final void timestampColumn(Column column) {
+		visitCell0(column, factory.getTimestampCellVisitor());
+	}
+
+	protected final void visitCell0(Column column, CellVisitor visitor) {
+		try {
+			visitCell(column, visitor);
+		} catch (Exception e) {
+			String sheetName = visitorValue.getSheet().getSheetName();
+			String ref = new CellReference(currentRow.getRowNum(), visitorValue.getColumnOption(column)
+					.getColumnIndex()).formatAsString();
+			throw new RuntimeException(MessageFormat.format("error at {0} cell={1}!{2}. {3}", column, sheetName, ref,
+					e.getMessage()), e);
+		}
+	}
+
+	protected void visitCell(Column column, CellVisitor visitor) {
+		ColumnOptionTask option = visitorValue.getColumnOption(column);
+		PoiExcelColumnValueType valueType = option.getValueTypeEnum();
+
+		switch (valueType) {
+		case SHEET_NAME:
+			visitor.visitSheetName(column);
+			return;
+		case ROW_NUMBER:
+			visitor.visitRowNumber(column, currentRow.getRowNum() + 1);
+			return;
+		case COLUMN_NUMBER:
+			visitor.visitColumnNumber(column, option.getColumnIndex() + 1);
+			return;
+		default:
+			break;
+		}
+
+		assert valueType.useCell();
+		Cell cell = currentRow.getCell(option.getColumnIndex());
+		if (cell == null) {
+			visitCellNull(column);
+			return;
+		}
+		switch (valueType) {
+		case CELL_VALUE:
+		case CELL_FORMULA:
+			visitCellValue(column, option, cell, visitor);
+			return;
+		case CELL_STYLE:
+			visitCellStyle(column, option, cell, visitor);
+			return;
+		case CELL_FONT:
+			visitCellFont(column, option, cell, visitor);
+			return;
+		case CELL_COMMENT:
+			visitCellComment(column, option, cell, visitor);
+			return;
+		default:
+			throw new UnsupportedOperationException(MessageFormat.format("unsupported value_type={0}", valueType));
+		}
+	}
+
+	protected void visitCellNull(Column column) {
+		pageBuilder.setNull(column);
+	}
+
+	private void visitCellValue(Column column, ColumnOptionTask option, Cell cell, CellVisitor visitor) {
+		PoiExcelCellVisitor delegator = factory.getPoiExcelCellVisitor();
+		delegator.visitCellValue(column, option, cell, visitor);
+	}
+
+	private void visitCellStyle(Column column, ColumnOptionTask option, Cell cell, CellVisitor visitor) {
+		PoiExcelCellStyleVisitor delegator = factory.getPoiExcelCellStyleVisitor();
+		delegator.visit(column, option, cell, visitor);
+	}
+
+	private void visitCellFont(Column column, ColumnOptionTask option, Cell cell, CellVisitor visitor) {
+		PoiExcelCellFontVisitor delegator = factory.getPoiExcelCellFontVisitor();
+		delegator.visit(column, option, cell, visitor);
+	}
+
+	private void visitCellComment(Column column, ColumnOptionTask option, Cell cell, CellVisitor visitor) {
+		PoiExcelCellCommentVisitor delegator = factory.getPoiExcelCellCommentVisitor();
+		delegator.visit(column, option, cell, visitor);
+	}
+}

data/src/main/java/org/embulk/parser/poi_excel/visitor/PoiExcelVisitorFactory.java

@@ -0,0 +1,171 @@
+package org.embulk.parser.poi_excel.visitor;
+
+import org.embulk.parser.poi_excel.visitor.embulk.BooleanCellVisitor;
+import org.embulk.parser.poi_excel.visitor.embulk.DoubleCellVisitor;
+import org.embulk.parser.poi_excel.visitor.embulk.LongCellVisitor;
+import org.embulk.parser.poi_excel.visitor.embulk.StringCellVisitor;
+import org.embulk.parser.poi_excel.visitor.embulk.TimestampCellVisitor;
+
+public class PoiExcelVisitorFactory {
+
+	protected final PoiExcelVisitorValue visitorValue;
+
+	public PoiExcelVisitorFactory(PoiExcelVisitorValue visitorValue) {
+		this.visitorValue = visitorValue;
+		visitorValue.setVisitorFactory(this);
+	}
+
+	// visitor root (Embulk ColumnVisitor)
+	private PoiExcelColumnVisitor poiExcelColumnVisitor;
+
+	public final PoiExcelColumnVisitor getPoiExcelColumnVisitor() {
+		if (poiExcelColumnVisitor == null) {
+			poiExcelColumnVisitor = newPoiExcelColumnVisitor();
+		}
+		return poiExcelColumnVisitor;
+	}
+
+	protected PoiExcelColumnVisitor newPoiExcelColumnVisitor() {
+		return new PoiExcelColumnVisitor(visitorValue);
+	}
+
+	// Embulk boolean
+	private BooleanCellVisitor booleanCellVisitor;
+
+	public final BooleanCellVisitor getBooleanCellVisitor() {
+		if (booleanCellVisitor == null) {
+			booleanCellVisitor = newBooleanCellVisitor();
+		}
+		return booleanCellVisitor;
+	}
+
+	protected BooleanCellVisitor newBooleanCellVisitor() {
+		return new BooleanCellVisitor(visitorValue);
+	}
+
+	// Embulk long
+	private LongCellVisitor longCellVisitor;
+
+	public final LongCellVisitor getLongCellVisitor() {
+		if (longCellVisitor == null) {
+			longCellVisitor = newLongCellVisitor();
+		}
+		return longCellVisitor;
+	}
+
+	protected LongCellVisitor newLongCellVisitor() {
+		return new LongCellVisitor(visitorValue);
+	}
+
+	// Embulk double
+	private DoubleCellVisitor doubleCellVisitor;
+
+	public final DoubleCellVisitor getDoubleCellVisitor() {
+		if (doubleCellVisitor == null) {
+			doubleCellVisitor = newDoubleCellVisitor();
+		}
+		return doubleCellVisitor;
+	}
+
+	protected DoubleCellVisitor newDoubleCellVisitor() {
+		return new DoubleCellVisitor(visitorValue);
+	}
+
+	// Embulk string
+	private StringCellVisitor stringCellVisitor;
+
+	public final StringCellVisitor getStringCellVisitor() {
+		if (stringCellVisitor == null) {
+			stringCellVisitor = newStringCellVisitor();
+		}
+		return stringCellVisitor;
+	}
+
+	protected StringCellVisitor newStringCellVisitor() {
+		return new StringCellVisitor(visitorValue);
+	}
+
+	// Embulk timestamp
+	private TimestampCellVisitor timestampCellVisitor;
+
+	public final TimestampCellVisitor getTimestampCellVisitor() {
+		if (timestampCellVisitor == null) {
+			timestampCellVisitor = newTimestampCellVisitor();
+		}
+		return timestampCellVisitor;
+	}
+
+	protected TimestampCellVisitor newTimestampCellVisitor() {
+		return new TimestampCellVisitor(visitorValue);
+	}
+
+	// cell value/formula
+	private PoiExcelCellVisitor poiExcelCellVisitor;
+
+	public final PoiExcelCellVisitor getPoiExcelCellVisitor() {
+		if (poiExcelCellVisitor == null) {
+			poiExcelCellVisitor = newPoiExcelCellVisitor();
+		}
+		return poiExcelCellVisitor;
+	}
+
+	protected PoiExcelCellVisitor newPoiExcelCellVisitor() {
+		return new PoiExcelCellVisitor(visitorValue);
+	}
+
+	// cell style
+	private PoiExcelCellStyleVisitor poiExcelCellStyleVisitor;
+
+	public final PoiExcelCellStyleVisitor getPoiExcelCellStyleVisitor() {
+		if (poiExcelCellStyleVisitor == null) {
+			poiExcelCellStyleVisitor = newPoiExcelCellStyleVisitor();
+		}
+		return poiExcelCellStyleVisitor;
+	}
+
+	protected PoiExcelCellStyleVisitor newPoiExcelCellStyleVisitor() {
+		return new PoiExcelCellStyleVisitor(visitorValue);
+	}
+
+	// cell font
+	private PoiExcelCellFontVisitor poiExcelCellFontVisitor;
+
+	public final PoiExcelCellFontVisitor getPoiExcelCellFontVisitor() {
+		if (poiExcelCellFontVisitor == null) {
+			poiExcelCellFontVisitor = newPoiExcelCellFontVisitor();
+		}
+		return poiExcelCellFontVisitor;
+	}
+
+	protected PoiExcelCellFontVisitor newPoiExcelCellFontVisitor() {
+		return new PoiExcelCellFontVisitor(visitorValue);
+	}
+
+	// cell comment
+	private PoiExcelCellCommentVisitor PoiExcelCellCommentVisitor;
+
+	public final PoiExcelCellCommentVisitor getPoiExcelCellCommentVisitor() {
+		if (PoiExcelCellCommentVisitor == null) {
+			PoiExcelCellCommentVisitor = newPoiExcelCellCommentVisitor();
+		}
+		return PoiExcelCellCommentVisitor;
+	}
+
+	protected PoiExcelCellCommentVisitor newPoiExcelCellCommentVisitor() {
+		return new PoiExcelCellCommentVisitor(visitorValue);
+	}
+
+	// color
+	private PoiExcelColorVisitor poiExcelColorVisitor;
+
+	public final PoiExcelColorVisitor getPoiExcelColorVisitor() {
+		if (poiExcelColorVisitor == null) {
+			poiExcelColorVisitor = newPoiExcelColorVisitor();
+		}
+		return poiExcelColorVisitor;
+	}
+
+	protected PoiExcelColorVisitor newPoiExcelColorVisitor() {
+		return new PoiExcelColorVisitor(visitorValue);
+	}
+}

data/src/main/java/org/embulk/parser/poi_excel/visitor/PoiExcelVisitorValue.java

@@ -0,0 +1,63 @@
+package org.embulk.parser.poi_excel.visitor;
+
+import java.util.ArrayList;
+import java.util.List;
+
+import org.apache.poi.ss.usermodel.Sheet;
+import org.embulk.parser.poi_excel.PoiExcelParserPlugin.ColumnOptionTask;
+import org.embulk.parser.poi_excel.PoiExcelParserPlugin.PluginTask;
+import org.embulk.spi.Column;
+import org.embulk.spi.ColumnConfig;
+import org.embulk.spi.PageBuilder;
+import org.embulk.spi.SchemaConfig;
+
+public class PoiExcelVisitorValue {
+	private final PluginTask task;
+	private final Sheet sheet;
+	private final PageBuilder pageBuilder;
+	private PoiExcelVisitorFactory factory;
+
+	private List<ColumnOptionTask> columnOptions;
+
+	public PoiExcelVisitorValue(PluginTask task, Sheet sheet, PageBuilder pageBuilder) {
+		this.task = task;
+		this.sheet = sheet;
+		this.pageBuilder = pageBuilder;
+	}
+
+	public PluginTask getPluginTask() {
+		return task;
+	}
+
+	public Sheet getSheet() {
+		return sheet;
+	}
+
+	public PageBuilder getPageBuilder() {
+		return pageBuilder;
+	}
+
+	public void setVisitorFactory(PoiExcelVisitorFactory factory) {
+		this.factory = factory;
+	}
+
+	public PoiExcelVisitorFactory getVisitorFactory() {
+		return factory;
+	}
+
+	public ColumnOptionTask getColumnOption(Column column) {
+		return getColumnOptions().get(column.getIndex());
+	}
+
+	public List<ColumnOptionTask> getColumnOptions() {
+		if (columnOptions == null) {
+			SchemaConfig schemaConfig = task.getColumns();
+			columnOptions = new ArrayList<>(schemaConfig.getColumnCount());
+			for (ColumnConfig c : schemaConfig.getColumns()) {
+				ColumnOptionTask option = c.getOption().loadConfig(ColumnOptionTask.class);
+				columnOptions.add(option);
+			}
+		}
+		return columnOptions;
+	}
+}

data/src/main/java/org/embulk/parser/poi_excel/visitor/embulk/BooleanCellVisitor.java

@@ -0,0 +1,54 @@
+package org.embulk.parser.poi_excel.visitor.embulk;
+
+import org.apache.poi.ss.usermodel.Sheet;
+import org.embulk.parser.poi_excel.visitor.PoiExcelVisitorValue;
+import org.embulk.spi.Column;
+
+public class BooleanCellVisitor extends CellVisitor {
+
+	public BooleanCellVisitor(PoiExcelVisitorValue visitorValue) {
+		super(visitorValue);
+	}
+
+	@Override
+	public void visitCellValueNumeric(Column column, Object source, double value) {
+		pageBuilder.setBoolean(column, value != 0d);
+	}
+
+	@Override
+	public void visitCellValueString(Column column, Object source, String value) {
+		pageBuilder.setBoolean(column, Boolean.parseBoolean(value));
+	}
+
+	@Override
+	public void visitCellValueBoolean(Column column, Object source, boolean value) {
+		pageBuilder.setBoolean(column, value);
+	}
+
+	@Override
+	public void visitCellValueError(Column column, Object source, int code) {
+		pageBuilder.setNull(column);
+	}
+
+	@Override
+	public void visitValueLong(Column column, Object source, long value) {
+		pageBuilder.setBoolean(column, value != 0);
+	}
+
+	@Override
+	public void visitSheetName(Column column) {
+		Sheet sheet = visitorValue.getSheet();
+		int index = sheet.getWorkbook().getSheetIndex(sheet);
+		pageBuilder.setBoolean(column, index != 0);
+	}
+
+	@Override
+	public void visitRowNumber(Column column, int index1) {
+		pageBuilder.setBoolean(column, index1 != 0);
+	}
+
+	@Override
+	public void visitColumnNumber(Column column, int index1) {
+		pageBuilder.setBoolean(column, index1 != 0);
+	}
+}

data/src/main/java/org/embulk/parser/poi_excel/visitor/embulk/CellVisitor.java

@@ -0,0 +1,41 @@
+package org.embulk.parser.poi_excel.visitor.embulk;
+
+import org.apache.poi.ss.usermodel.Cell;
+import org.embulk.parser.poi_excel.visitor.PoiExcelVisitorValue;
+import org.embulk.spi.Column;
+import org.embulk.spi.PageBuilder;
+
+public abstract class CellVisitor {
+
+	protected final PoiExcelVisitorValue visitorValue;
+	protected final PageBuilder pageBuilder;
+
+	public CellVisitor(PoiExcelVisitorValue visitorValue) {
+		this.visitorValue = visitorValue;
+		this.pageBuilder = visitorValue.getPageBuilder();
+	}
+
+	public abstract void visitCellValueNumeric(Column column, Object source, double value);
+
+	public abstract void visitCellValueString(Column column, Object source, String value);
+
+	public void visitCellValueBlank(Column column, Object source) {
+		pageBuilder.setNull(column);
+	}
+
+	public abstract void visitCellValueBoolean(Column column, Object source, boolean value);
+
+	public abstract void visitCellValueError(Column column, Object source, int code);
+
+	public void visitCellFormula(Column column, Cell cell) {
+		pageBuilder.setString(column, cell.getCellFormula());
+	}
+
+	public abstract void visitValueLong(Column column, Object source, long value);
+
+	public abstract void visitSheetName(Column column);
+
+	public abstract void visitRowNumber(Column column, int index1);
+
+	public abstract void visitColumnNumber(Column column, int index1);
+}

data/src/main/java/org/embulk/parser/poi_excel/visitor/embulk/DoubleCellVisitor.java

@@ -0,0 +1,54 @@
+package org.embulk.parser.poi_excel.visitor.embulk;
+
+import org.apache.poi.ss.usermodel.Sheet;
+import org.embulk.parser.poi_excel.visitor.PoiExcelVisitorValue;
+import org.embulk.spi.Column;
+
+public class DoubleCellVisitor extends CellVisitor {
+
+	public DoubleCellVisitor(PoiExcelVisitorValue visitorValue) {
+		super(visitorValue);
+	}
+
+	@Override
+	public void visitCellValueNumeric(Column column, Object source, double value) {
+		pageBuilder.setDouble(column, value);
+	}
+
+	@Override
+	public void visitCellValueString(Column column, Object source, String value) {
+		pageBuilder.setDouble(column, Double.parseDouble(value));
+	}
+
+	@Override
+	public void visitCellValueBoolean(Column column, Object source, boolean value) {
+		pageBuilder.setDouble(column, value ? 1 : 0);
+	}
+
+	@Override
+	public void visitCellValueError(Column column, Object source, int code) {
+		pageBuilder.setDouble(column, code);
+	}
+
+	@Override
+	public void visitValueLong(Column column, Object source, long value) {
+		pageBuilder.setDouble(column, value);
+	}
+
+	@Override
+	public void visitSheetName(Column column) {
+		Sheet sheet = visitorValue.getSheet();
+		int index = sheet.getWorkbook().getSheetIndex(sheet);
+		pageBuilder.setDouble(column, index);
+	}
+
+	@Override
+	public void visitRowNumber(Column column, int index1) {
+		pageBuilder.setDouble(column, index1);
+	}
+
+	@Override
+	public void visitColumnNumber(Column column, int index1) {
+		pageBuilder.setDouble(column, index1);
+	}
+}

data/src/main/java/org/embulk/parser/poi_excel/visitor/embulk/LongCellVisitor.java

@@ -0,0 +1,54 @@
+package org.embulk.parser.poi_excel.visitor.embulk;
+
+import org.apache.poi.ss.usermodel.Sheet;
+import org.embulk.parser.poi_excel.visitor.PoiExcelVisitorValue;
+import org.embulk.spi.Column;
+
+public class LongCellVisitor extends CellVisitor {
+
+	public LongCellVisitor(PoiExcelVisitorValue visitorValue) {
+		super(visitorValue);
+	}
+
+	@Override
+	public void visitCellValueNumeric(Column column, Object source, double value) {
+		pageBuilder.setLong(column, (long) value);
+	}
+
+	@Override
+	public void visitCellValueString(Column column, Object source, String value) {
+		pageBuilder.setLong(column, Long.parseLong(value));
+	}
+
+	@Override
+	public void visitCellValueBoolean(Column column, Object source, boolean value) {
+		pageBuilder.setLong(column, value ? 1 : 0);
+	}
+
+	@Override
+	public void visitCellValueError(Column column, Object source, int code) {
+		pageBuilder.setLong(column, code);
+	}
+
+	@Override
+	public void visitValueLong(Column column, Object source, long value) {
+		pageBuilder.setLong(column, value);
+	}
+
+	@Override
+	public void visitSheetName(Column column) {
+		Sheet sheet = visitorValue.getSheet();
+		int index = sheet.getWorkbook().getSheetIndex(sheet);
+		pageBuilder.setLong(column, index);
+	}
+
+	@Override
+	public void visitRowNumber(Column column, int index1) {
+		pageBuilder.setLong(column, index1);
+	}
+
+	@Override
+	public void visitColumnNumber(Column column, int index1) {
+		pageBuilder.setLong(column, index1);
+	}
+}

data/src/main/java/org/embulk/parser/poi_excel/visitor/embulk/StringCellVisitor.java

@@ -0,0 +1,63 @@
+package org.embulk.parser.poi_excel.visitor.embulk;
+
+import org.apache.poi.ss.usermodel.FormulaError;
+import org.apache.poi.ss.usermodel.Sheet;
+import org.apache.poi.ss.util.CellReference;
+import org.embulk.parser.poi_excel.visitor.PoiExcelVisitorValue;
+import org.embulk.spi.Column;
+
+public class StringCellVisitor extends CellVisitor {
+
+	public StringCellVisitor(PoiExcelVisitorValue visitorValue) {
+		super(visitorValue);
+	}
+
+	@Override
+	public void visitCellValueNumeric(Column column, Object source, double value) {
+		String s = Double.toString(value);
+		if (s.endsWith(".0")) {
+			s = s.substring(0, s.length() - 2);
+		}
+		pageBuilder.setString(column, s);
+	}
+
+	@Override
+	public void visitCellValueString(Column column, Object source, String value) {
+		pageBuilder.setString(column, value);
+	}
+
+	@Override
+	public void visitCellValueBoolean(Column column, Object source, boolean value) {
+		pageBuilder.setString(column, Boolean.toString(value));
+	}
+
+	@Override
+	public void visitCellValueError(Column column, Object source, int code) {
+		FormulaError error = FormulaError.forInt((byte) code);
+		String value = error.getString();
+		pageBuilder.setString(column, value);
+	}
+
+	@Override
+	public void visitValueLong(Column column, Object source, long value) {
+		String s = Long.toString(value);
+		pageBuilder.setString(column, s);
+	}
+
+	@Override
+	public void visitSheetName(Column column) {
+		Sheet sheet = visitorValue.getSheet();
+		pageBuilder.setString(column, sheet.getSheetName());
+	}
+
+	@Override
+	public void visitRowNumber(Column column, int index1) {
+		pageBuilder.setString(column, Integer.toString(index1));
+	}
+
+	@Override
+	public void visitColumnNumber(Column column, int index1) {
+		String value = CellReference.convertNumToColString(index1 - 1);
+		pageBuilder.setString(column, value);
+	}
+}