embulk-parser-poi_excel 0.1.4 → 0.1.11

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

@@ -0,0 +1,52 @@
+package org.embulk.parser.poi_excel.visitor;
+
+import org.apache.poi.ss.usermodel.Cell;
+import org.apache.poi.ss.usermodel.CellType;
+import org.embulk.parser.poi_excel.bean.PoiExcelColumnBean;
+import org.embulk.parser.poi_excel.visitor.embulk.CellVisitor;
+import org.embulk.spi.Column;
+import org.embulk.spi.PageBuilder;
+import org.embulk.spi.type.StringType;
+
+public class PoiExcelCellTypeVisitor {
+	protected final PoiExcelVisitorValue visitorValue;
+	protected final PageBuilder pageBuilder;
+
+	public PoiExcelCellTypeVisitor(PoiExcelVisitorValue visitorValue) {
+		this.visitorValue = visitorValue;
+		this.pageBuilder = visitorValue.getPageBuilder();
+	}
+
+	public void visit(PoiExcelColumnBean bean, Cell cell, CellType cellType, CellVisitor visitor) {
+		assert cell != null;
+
+		Column column = bean.getColumn();
+		if (column.getType() instanceof StringType) {
+			String type = cellType.name();
+			visitor.visitCellValueString(column, cell, type);
+			return;
+		}
+
+		int code = getCode(cellType);
+		visitor.visitCellValueNumeric(column, cell, code);
+	}
+
+	private static int getCode(CellType cellType) {
+		switch (cellType) {
+		case NUMERIC:
+			return 0;
+		case STRING:
+			return 1;
+		case FORMULA:
+			return 2;
+		case BLANK:
+			return 3;
+		case BOOLEAN:
+			return 4;
+		case ERROR:
+			return 5;
+		default:
+			return -1;
+		}
+	}
+}

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

@@ -4,6 +4,7 @@ import java.text.MessageFormat;
 import java.util.List;
 
 import org.apache.poi.ss.usermodel.Cell;
+import org.apache.poi.ss.usermodel.CellType;
 import org.apache.poi.ss.usermodel.CellValue;
 import org.apache.poi.ss.usermodel.CreationHelper;
 import org.apache.poi.ss.usermodel.FormulaError;
@@ -16,7 +17,9 @@ import org.embulk.parser.poi_excel.PoiExcelColumnValueType;
 import org.embulk.parser.poi_excel.PoiExcelParserPlugin.FormulaReplaceTask;
 import org.embulk.parser.poi_excel.bean.PoiExcelColumnBean;
 import org.embulk.parser.poi_excel.bean.PoiExcelColumnBean.ErrorStrategy;
+import org.embulk.parser.poi_excel.bean.PoiExcelColumnBean.FormulaHandling;
 import org.embulk.parser.poi_excel.visitor.embulk.CellVisitor;
+import org.embulk.parser.poi_excel.visitor.util.MergedRegionFinder;
 import org.embulk.spi.Column;
 import org.embulk.spi.Exec;
 import org.embulk.spi.PageBuilder;
@@ -38,15 +41,15 @@ public class PoiExcelCellValueVisitor {
 
 		Column column = bean.getColumn();
 
-		int cellType = cell.getCellType();
+		CellType cellType = cell.getCellTypeEnum();
 		switch (cellType) {
-		case Cell.CELL_TYPE_NUMERIC:
+		case NUMERIC:
 			visitor.visitCellValueNumeric(column, cell, cell.getNumericCellValue());
 			return;
-		case Cell.CELL_TYPE_STRING:
+		case STRING:
 			visitor.visitCellValueString(column, cell, cell.getStringCellValue());
 			return;
-		case Cell.CELL_TYPE_FORMULA:
+		case FORMULA:
 			PoiExcelColumnValueType valueType = bean.getValueType();
 			if (valueType == PoiExcelColumnValueType.CELL_FORMULA) {
 				visitor.visitCellFormula(column, cell);
@@ -54,13 +57,13 @@ public class PoiExcelCellValueVisitor {
 				visitCellValueFormula(bean, cell, visitor);
 			}
 			return;
-		case Cell.CELL_TYPE_BLANK:
+		case BLANK:
 			visitCellValueBlank(bean, cell, visitor);
 			return;
-		case Cell.CELL_TYPE_BOOLEAN:
+		case BOOLEAN:
 			visitor.visitCellValueBoolean(column, cell, cell.getBooleanCellValue());
 			return;
-		case Cell.CELL_TYPE_ERROR:
+		case ERROR:
 			visitCellValueError(bean, cell, cell.getErrorCellValue(), visitor);
 			return;
 		default:
@@ -69,35 +72,24 @@ public class PoiExcelCellValueVisitor {
 	}
 
 	protected void visitCellValueBlank(PoiExcelColumnBean bean, Cell cell, CellVisitor visitor) {
-		assert cell.getCellType() == Cell.CELL_TYPE_BLANK;
+		assert cell.getCellTypeEnum() == CellType.BLANK;
 
 		Column column = bean.getColumn();
 
-		boolean search = bean.getSearchMergedCell();
-		if (!search) {
-			visitor.visitCellValueBlank(column, cell);
-			return;
-		}
-
-		int r = cell.getRowIndex();
-		int c = cell.getColumnIndex();
-
-		Sheet sheet = cell.getSheet();
-		int size = sheet.getNumMergedRegions();
-		for (int i = 0; i < size; i++) {
-			CellRangeAddress range = visitorValue.getSheet().getMergedRegion(i);
-			if (range.isInRange(r, c)) {
-				Row firstRow = sheet.getRow(range.getFirstRow());
-				if (firstRow == null) {
-					visitCellNull(column);
-					return;
-				}
-				Cell firstCell = firstRow.getCell(range.getFirstColumn());
-				if (firstCell == null) {
-					visitCellNull(column);
-					return;
-				}
+		CellRangeAddress region = findRegion(bean, cell);
+		if (region != null) {
+			Row firstRow = cell.getSheet().getRow(region.getFirstRow());
+			if (firstRow == null) {
+				visitCellNull(column);
+				return;
+			}
+			Cell firstCell = firstRow.getCell(region.getFirstColumn());
+			if (firstCell == null) {
+				visitCellNull(column);
+				return;
+			}
 
+			if (firstCell.getRowIndex() != cell.getRowIndex() || firstCell.getColumnIndex() != cell.getColumnIndex()) {
 				visitCellValue(bean, firstCell, visitor);
 				return;
 			}
@@ -106,9 +98,56 @@ public class PoiExcelCellValueVisitor {
 		visitor.visitCellValueBlank(column, cell);
 	}
 
+	protected CellRangeAddress findRegion(PoiExcelColumnBean bean, Cell cell) {
+		Sheet sheet = cell.getSheet();
+		int r = cell.getRowIndex();
+		int c = cell.getColumnIndex();
+
+		MergedRegionFinder finder = bean.getMergedRegionFinder();
+		return finder.get(sheet, r, c);
+	}
+
 	protected void visitCellValueFormula(PoiExcelColumnBean bean, Cell cell, CellVisitor visitor) {
-		assert cell.getCellType() == Cell.CELL_TYPE_FORMULA;
+		assert cell.getCellTypeEnum() == CellType.FORMULA;
+
+		FormulaHandling handling = bean.getFormulaHandling();
+		switch (handling) {
+		case CASHED_VALUE:
+			visitCellValueFormulaCashedValue(bean, cell, visitor);
+			break;
+		default:
+			visitCellValueFormulaEvaluate(bean, cell, visitor);
+			break;
+		}
+	}
+
+	protected void visitCellValueFormulaCashedValue(PoiExcelColumnBean bean, Cell cell, CellVisitor visitor) {
+		Column column = bean.getColumn();
+
+		CellType cellType = cell.getCachedFormulaResultTypeEnum();
+		switch (cellType) {
+		case NUMERIC:
+			visitor.visitCellValueNumeric(column, cell, cell.getNumericCellValue());
+			return;
+		case STRING:
+			visitor.visitCellValueString(column, cell, cell.getStringCellValue());
+			return;
+		case BLANK:
+			visitCellValueBlank(bean, cell, visitor);
+			return;
+		case BOOLEAN:
+			visitor.visitCellValueBoolean(column, cell, cell.getBooleanCellValue());
+			return;
+		case ERROR:
+			visitCellValueError(bean, cell, cell.getErrorCellValue(), visitor);
+			return;
+		case FORMULA:
+		default:
+			throw new IllegalStateException(MessageFormat.format("unsupported POI cellType={0}", cellType));
+		}
+	}
 
+	protected void visitCellValueFormulaEvaluate(PoiExcelColumnBean bean, Cell cell, CellVisitor visitor) {
 		Column column = bean.getColumn();
 
 		List<FormulaReplaceTask> list = bean.getFormulaReplace();
@@ -159,24 +198,24 @@ public class PoiExcelCellValueVisitor {
 			throw new RuntimeException(MessageFormat.format("evaluate error. formula={0}", cell.getCellFormula()), e);
 		}
 
-		int cellType = cellValue.getCellType();
+		CellType cellType = cellValue.getCellTypeEnum();
 		switch (cellType) {
-		case Cell.CELL_TYPE_NUMERIC:
+		case NUMERIC:
 			visitor.visitCellValueNumeric(column, cellValue, cellValue.getNumberValue());
 			return;
-		case Cell.CELL_TYPE_STRING:
+		case STRING:
 			visitor.visitCellValueString(column, cellValue, cellValue.getStringValue());
 			return;
-		case Cell.CELL_TYPE_BLANK:
+		case BLANK:
 			visitor.visitCellValueBlank(column, cellValue);
 			return;
-		case Cell.CELL_TYPE_BOOLEAN:
+		case BOOLEAN:
 			visitor.visitCellValueBoolean(column, cellValue, cellValue.getBooleanValue());
 			return;
-		case Cell.CELL_TYPE_ERROR:
+		case ERROR:
 			visitCellValueError(bean, cellValue, cellValue.getErrorValue(), visitor);
 			return;
-		case Cell.CELL_TYPE_FORMULA:
+		case FORMULA:
 		default:
 			throw new IllegalStateException(MessageFormat.format("unsupported POI cellType={0}", cellType));
 		}

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

@@ -40,7 +40,7 @@ public class PoiExcelClientAnchorVisitor extends AbstractPoiExcelCellAttributeVi
 		map.put("anchor_type", new AttributeSupplier<ClientAnchor>() {
 			@Override
 			public Object get(Column column, Cell cell, ClientAnchor anchor) {
-				return (long) anchor.getAnchorType();
+				return (long) anchor.getAnchorType().value;
 			}
 		});
 		map.put("col1", new AttributeSupplier<ClientAnchor>() {

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

@@ -3,16 +3,22 @@ package org.embulk.parser.poi_excel.visitor;
 import java.text.MessageFormat;
 
 import org.apache.poi.ss.usermodel.Cell;
+import org.apache.poi.ss.usermodel.CellType;
 import org.apache.poi.ss.usermodel.Row;
+import org.apache.poi.ss.usermodel.Sheet;
 import org.apache.poi.ss.util.CellReference;
 import org.embulk.parser.poi_excel.PoiExcelColumnValueType;
 import org.embulk.parser.poi_excel.bean.PoiExcelColumnBean;
+import org.embulk.parser.poi_excel.bean.util.PoiExcelCellAddress;
 import org.embulk.parser.poi_excel.visitor.embulk.CellVisitor;
 import org.embulk.spi.Column;
 import org.embulk.spi.ColumnVisitor;
+import org.embulk.spi.Exec;
 import org.embulk.spi.PageBuilder;
+import org.slf4j.Logger;
 
 public class PoiExcelColumnVisitor implements ColumnVisitor {
+	private final Logger log = Exec.getLogger(getClass());
 
 	protected final PoiExcelVisitorValue visitorValue;
 	protected final PageBuilder pageBuilder;
@@ -56,6 +62,9 @@ public class PoiExcelColumnVisitor implements ColumnVisitor {
 	}
 
 	protected final void visitCell0(Column column, CellVisitor visitor) {
+		if (log.isTraceEnabled()) {
+			log.trace("{} start", column);
+		}
 		try {
 			visitCell(column, visitor);
 		} catch (Exception e) {
@@ -65,21 +74,52 @@ public class PoiExcelColumnVisitor implements ColumnVisitor {
 			throw new RuntimeException(MessageFormat.format("error at {0} cell={1}!{2}. {3}", column, sheetName, ref,
 					e.getMessage()), e);
 		}
+		if (log.isTraceEnabled()) {
+			log.trace("{} end", column);
+		}
 	}
 
 	protected void visitCell(Column column, CellVisitor visitor) {
 		PoiExcelColumnBean bean = visitorValue.getColumnBean(column);
 		PoiExcelColumnValueType valueType = bean.getValueType();
+		PoiExcelCellAddress cellAddress = bean.getCellAddress();
 
 		switch (valueType) {
 		case SHEET_NAME:
-			visitor.visitSheetName(column);
+			if (cellAddress != null) {
+				Sheet sheet = cellAddress.getSheet(currentRow);
+				visitor.visitSheetName(column, sheet);
+			} else {
+				visitor.visitSheetName(column);
+			}
 			return;
 		case ROW_NUMBER:
-			visitor.visitRowNumber(column, currentRow.getRowNum() + 1);
+			int rowIndex;
+			if (cellAddress != null) {
+				Cell cell = cellAddress.getCell(currentRow);
+				if (cell == null) {
+					visitCellNull(column);
+					return;
+				}
+				rowIndex = cell.getRowIndex();
+			} else {
+				rowIndex = currentRow.getRowNum();
+			}
+			visitor.visitRowNumber(column, rowIndex + 1);
 			return;
 		case COLUMN_NUMBER:
-			visitor.visitColumnNumber(column, bean.getColumnIndex() + 1);
+			int columnIndex;
+			if (cellAddress != null) {
+				Cell cell = cellAddress.getCell(currentRow);
+				if (cell == null) {
+					visitCellNull(column);
+					return;
+				}
+				columnIndex = cell.getColumnIndex();
+			} else {
+				columnIndex = bean.getColumnIndex();
+			}
+			visitor.visitColumnNumber(column, columnIndex + 1);
 			return;
 		case CONSTANT:
 			visitCellConstant(column, bean.getValueTypeSuffix(), visitor);
@@ -89,7 +129,12 @@ public class PoiExcelColumnVisitor implements ColumnVisitor {
 		}
 
 		assert valueType.useCell();
-		Cell cell = currentRow.getCell(bean.getColumnIndex());
+		Cell cell;
+		if (cellAddress != null) {
+			cell = cellAddress.getCell(currentRow);
+		} else {
+			cell = currentRow.getCell(bean.getColumnIndex());
+		}
 		if (cell == null) {
 			visitCellNull(column);
 			return;
@@ -108,6 +153,16 @@ public class PoiExcelColumnVisitor implements ColumnVisitor {
 		case CELL_COMMENT:
 			visitCellComment(bean, cell, visitor);
 			return;
+		case CELL_TYPE:
+			visitCellType(bean, cell, cell.getCellTypeEnum(), visitor);
+			return;
+		case CELL_CACHED_TYPE:
+			if (cell.getCellTypeEnum() == CellType.FORMULA) {
+				visitCellType(bean, cell, cell.getCachedFormulaResultTypeEnum(), visitor);
+			} else {
+				visitCellType(bean, cell, cell.getCellTypeEnum(), visitor);
+			}
+			return;
 		default:
 			throw new UnsupportedOperationException(MessageFormat.format("unsupported value_type={0}", valueType));
 		}
@@ -144,4 +199,9 @@ public class PoiExcelColumnVisitor implements ColumnVisitor {
 		PoiExcelCellCommentVisitor delegator = factory.getPoiExcelCellCommentVisitor();
 		delegator.visit(bean, cell, visitor);
 	}
+
+	private void visitCellType(PoiExcelColumnBean bean, Cell cell, CellType cellType, CellVisitor visitor) {
+		PoiExcelCellTypeVisitor delegator = factory.getPoiExcelCellTypeVisitor();
+		delegator.visit(bean, cell, cellType, visitor);
+	}
 }

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

@@ -159,6 +159,20 @@ public class PoiExcelVisitorFactory {
 		return new PoiExcelCellCommentVisitor(visitorValue);
 	}
 
+	// cell type
+	private PoiExcelCellTypeVisitor poiExcelCellTypeVisitor;
+
+	public final PoiExcelCellTypeVisitor getPoiExcelCellTypeVisitor() {
+		if (poiExcelCellTypeVisitor == null) {
+			poiExcelCellTypeVisitor = newPoiExcelCellTypeVisitor();
+		}
+		return poiExcelCellTypeVisitor;
+	}
+
+	protected PoiExcelCellTypeVisitor newPoiExcelCellTypeVisitor() {
+		return new PoiExcelCellTypeVisitor(visitorValue);
+	}
+
 	// ClientAnchor
 	private PoiExcelClientAnchorVisitor poiExcelClientAnchorVisitor;
 

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

@@ -38,6 +38,11 @@ public class BooleanCellVisitor extends CellVisitor {
 	@Override
 	public void visitSheetName(Column column) {
 		Sheet sheet = visitorValue.getSheet();
+		visitSheetName(column, sheet);
+	}
+
+	@Override
+	public void visitSheetName(Column column, Sheet sheet) {
 		int index = sheet.getWorkbook().getSheetIndex(sheet);
 		pageBuilder.setBoolean(column, index != 0);
 	}

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

@@ -3,6 +3,7 @@ package org.embulk.parser.poi_excel.visitor.embulk;
 import java.text.MessageFormat;
 
 import org.apache.poi.ss.usermodel.Cell;
+import org.apache.poi.ss.usermodel.Sheet;
 import org.embulk.config.ConfigException;
 import org.embulk.parser.poi_excel.bean.PoiExcelColumnBean;
 import org.embulk.parser.poi_excel.bean.PoiExcelColumnBean.ErrorStrategy;
@@ -40,6 +41,8 @@ public abstract class CellVisitor {
 
 	public abstract void visitSheetName(Column column);
 
+	public abstract void visitSheetName(Column column, Sheet sheet);
+
 	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

@@ -45,6 +45,11 @@ public class DoubleCellVisitor extends CellVisitor {
 	@Override
 	public void visitSheetName(Column column) {
 		Sheet sheet = visitorValue.getSheet();
+		visitSheetName(column, sheet);
+	}
+
+	@Override
+	public void visitSheetName(Column column, Sheet sheet) {
 		int index = sheet.getWorkbook().getSheetIndex(sheet);
 		pageBuilder.setDouble(column, index);
 	}

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

@@ -45,6 +45,11 @@ public class LongCellVisitor extends CellVisitor {
 	@Override
 	public void visitSheetName(Column column) {
 		Sheet sheet = visitorValue.getSheet();
+		visitSheetName(column, sheet);
+	}
+
+	@Override
+	public void visitSheetName(Column column, Sheet sheet) {
 		int index = sheet.getWorkbook().getSheetIndex(sheet);
 		pageBuilder.setLong(column, index);
 	}