embulk-parser-poi_excel 0.1.4 → 0.1.10

data/src/main/java/org/embulk/parser/poi_excel/bean/PoiExcelColumnIndex.java

@@ -8,6 +8,7 @@ import java.util.Map;
 import org.apache.poi.ss.util.CellReference;
 import org.embulk.parser.poi_excel.PoiExcelColumnValueType;
 import org.embulk.parser.poi_excel.PoiExcelParserPlugin.PluginTask;
+import org.embulk.parser.poi_excel.bean.util.PoiExcelCellAddress;
 import org.embulk.spi.Column;
 import org.embulk.spi.Exec;
 import org.embulk.spi.Schema;
@@ -36,43 +37,10 @@ public class PoiExcelColumnIndex {
 				}
 				bean.setColumnIndex(index);
 				indexMap.put(column.getName(), index);
-				if (log.isInfoEnabled()) {
-					String c = CellReference.convertNumToColString(index);
-					switch (valueType) {
-					default:
-						String suffix = bean.getValueTypeSuffix();
-						if (suffix != null) {
-							log.info("column.name={} <- cell_column={}, value_type={}, value=[{}]", column.getName(),
-									c, valueType, suffix);
-						} else {
-							log.info("column.name={} <- cell_column={}, value_type={}, value={}", column.getName(), c,
-									valueType, suffix);
-						}
-						break;
-					case CELL_VALUE:
-					case CELL_FORMULA:
-					case COLUMN_NUMBER:
-						log.info("column.name={} <- cell_column={}, value_type={}", column.getName(), c, valueType);
-						break;
-					}
-				}
-			} else {
-				if (log.isInfoEnabled()) {
-					switch (valueType) {
-					default:
-						String suffix = bean.getValueTypeSuffix();
-						if (suffix != null) {
-							log.info("column.name={} <- value_type={}, value=[{}]", column.getName(), valueType, suffix);
-						} else {
-							log.info("column.name={} <- value_type={}, value={}", column.getName(), valueType, suffix);
-						}
-						break;
-					case SHEET_NAME:
-					case ROW_NUMBER:
-						log.info("column.name={} <- value_type={}", column.getName(), valueType);
-						break;
-					}
-				}
+			}
+
+			if (log.isInfoEnabled()) {
+				logColumn(column, bean, valueType, index);
 			}
 		}
 	}
@@ -80,6 +48,15 @@ public class PoiExcelColumnIndex {
 	protected int resolveColumnIndex(Column column, PoiExcelColumnBean bean, int index,
 			PoiExcelColumnValueType valueType) {
 		Optional<String> numberOption = bean.getColumnNumber();
+		PoiExcelCellAddress cellAddress = bean.getCellAddress();
+
+		if (cellAddress != null) {
+			if (numberOption.isPresent()) {
+				throw new RuntimeException("only one of column_number, cell_address can be specified");
+			}
+			return index;
+		}
+
 		if (numberOption.isPresent()) {
 			String columnNumber = numberOption.get();
 			if (columnNumber.length() >= 1) {
@@ -190,4 +167,64 @@ public class PoiExcelColumnIndex {
 		}
 		return index;
 	}
+
+	protected void logColumn(Column column, PoiExcelColumnBean bean, PoiExcelColumnValueType valueType, int index) {
+		PoiExcelCellAddress cellAddress = bean.getCellAddress();
+
+		String cname, cvalue;
+		if (cellAddress != null) {
+			cname = "cell_address";
+			cvalue = cellAddress.getString();
+		} else {
+			cname = "cell_column";
+			cvalue = CellReference.convertNumToColString(index);
+		}
+
+		switch (valueType) {
+		default:
+		case CELL_VALUE:
+		case CELL_FORMULA:
+		case CELL_TYPE:
+		case CELL_CACHED_TYPE:
+		case COLUMN_NUMBER:
+			log.info("column.name={} <- {}={}, value_type={}", column.getName(), cname, cvalue, valueType);
+			break;
+		case CELL_STYLE:
+		case CELL_FONT:
+		case CELL_COMMENT:
+			String suffix = bean.getValueTypeSuffix();
+			if (suffix != null) {
+				log.info("column.name={} <- {}={}, value_type={}, value=[{}]", column.getName(), cname, cvalue,
+						valueType, suffix);
+			} else {
+				log.info("column.name={} <- {}={}, value_type={}, value={}", column.getName(), cname, cvalue,
+						valueType, suffix);
+			}
+			break;
+
+		case SHEET_NAME:
+			if (cellAddress != null && cellAddress.getSheetName() != null) {
+				log.info("column.name={} <- {}={}, value_type={}", column.getName(), cname, cvalue, valueType);
+			} else {
+				log.info("column.name={} <- value_type={}", column.getName(), valueType);
+			}
+			break;
+		case ROW_NUMBER:
+			if (cellAddress != null) {
+				log.info("column.name={} <- {}={}, value_type={}", column.getName(), cname, cvalue, valueType);
+			} else {
+				log.info("column.name={} <- value_type={}", column.getName(), valueType);
+			}
+			break;
+
+		case CONSTANT:
+			String value = bean.getValueTypeSuffix();
+			if (value != null) {
+				log.info("column.name={} <- value_type={}, value=[{}]", column.getName(), valueType, value);
+			} else {
+				log.info("column.name={} <- value_type={}, value={}", column.getName(), valueType, value);
+			}
+			break;
+		}
+	}
 }

data/src/main/java/org/embulk/parser/poi_excel/bean/util/PoiExcelCellAddress.java

@@ -0,0 +1,50 @@
+package org.embulk.parser.poi_excel.bean.util;
+
+import java.text.MessageFormat;
+
+import org.apache.poi.ss.usermodel.Cell;
+import org.apache.poi.ss.usermodel.Row;
+import org.apache.poi.ss.usermodel.Sheet;
+import org.apache.poi.ss.usermodel.Workbook;
+import org.apache.poi.ss.util.CellReference;
+
+public class PoiExcelCellAddress {
+	private final CellReference cellReference;
+
+	public PoiExcelCellAddress(String cellAddress) {
+		this.cellReference = new CellReference(cellAddress);
+	}
+
+	public String getSheetName() {
+		return cellReference.getSheetName();
+	}
+
+	public Sheet getSheet(Row currentRow) {
+		String sheetName = getSheetName();
+		if (sheetName != null) {
+			Workbook book = currentRow.getSheet().getWorkbook();
+			Sheet sheet = book.getSheet(sheetName);
+			if (sheet == null) {
+				throw new RuntimeException(MessageFormat.format("not found sheet. sheetName={0}", sheetName));
+			}
+			return sheet;
+		} else {
+			return currentRow.getSheet();
+		}
+	}
+
+	public Cell getCell(Row currentRow) {
+		Sheet sheet = getSheet(currentRow);
+
+		Row row = sheet.getRow(cellReference.getRow());
+		if (row == null) {
+			return null;
+		}
+
+		return row.getCell(cellReference.getCol());
+	}
+
+	public String getString() {
+		return cellReference.formatAsString();
+	}
+}

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

@@ -100,12 +100,6 @@ public class PoiExcelCellFontVisitor extends AbstractPoiExcelCellAttributeVisito
 				return (long) font.getIndex();
 			}
 		});
-		map.put("boldweight", new AttributeSupplier<Font>() {
-			@Override
-			public Object get(Column column, Cell cell, Font font) {
-				return (long) font.getBoldweight();
-			}
-		});
 		map.put("bold", new AttributeSupplier<Font>() {
 			@Override
 			public Object get(Column column, Cell cell, Font font) {

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

@@ -41,16 +41,16 @@ public class PoiExcelCellStyleVisitor extends AbstractPoiExcelCellAttributeVisit
 		map.put("alignment", new AttributeSupplier<CellStyle>() {
 			@Override
 			public Object get(Column column, Cell cell, CellStyle style) {
-				return (long) style.getAlignment();
+				return (long) style.getAlignmentEnum().getCode();
 			}
 		});
 		map.put("border", new AttributeSupplier<CellStyle>() {
 			@Override
 			public Object get(Column column, Cell cell, CellStyle style) {
-				int n0 = style.getBorderTop();
-				int n1 = style.getBorderBottom();
-				int n2 = style.getBorderLeft();
-				int n3 = style.getBorderRight();
+				int n0 = style.getBorderTopEnum().getCode();
+				int n1 = style.getBorderBottomEnum().getCode();
+				int n2 = style.getBorderLeftEnum().getCode();
+				int n3 = style.getBorderRightEnum().getCode();
 				if (column.getType() instanceof StringType) {
 					return String.format("%02x%02x%02x%02x", n0, n1, n2, n3);
 				}
@@ -60,25 +60,25 @@ public class PoiExcelCellStyleVisitor extends AbstractPoiExcelCellAttributeVisit
 		map.put("border_bottom", new AttributeSupplier<CellStyle>() {
 			@Override
 			public Object get(Column column, Cell cell, CellStyle style) {
-				return (long) style.getBorderBottom();
+				return (long) style.getBorderBottomEnum().getCode();
 			}
 		});
 		map.put("border_left", new AttributeSupplier<CellStyle>() {
 			@Override
 			public Object get(Column column, Cell cell, CellStyle style) {
-				return (long) style.getBorderLeft();
+				return (long) style.getBorderLeftEnum().getCode();
 			}
 		});
 		map.put("border_right", new AttributeSupplier<CellStyle>() {
 			@Override
 			public Object get(Column column, Cell cell, CellStyle style) {
-				return (long) style.getBorderRight();
+				return (long) style.getBorderRightEnum().getCode();
 			}
 		});
 		map.put("border_top", new AttributeSupplier<CellStyle>() {
 			@Override
 			public Object get(Column column, Cell cell, CellStyle style) {
-				return (long) style.getBorderTop();
+				return (long) style.getBorderTopEnum().getCode();
 			}
 		});
 		map.put("border_bottom_color", new AttributeSupplier<CellStyle>() {
@@ -154,7 +154,7 @@ public class PoiExcelCellStyleVisitor extends AbstractPoiExcelCellAttributeVisit
 		map.put("fill_pattern", new AttributeSupplier<CellStyle>() {
 			@Override
 			public Object get(Column column, Cell cell, CellStyle style) {
-				return (long) style.getFillPattern();
+				return (long) style.getFillPatternEnum().getCode();
 			}
 		});
 		map.put("font_index", new AttributeSupplier<CellStyle>() {
@@ -190,7 +190,7 @@ public class PoiExcelCellStyleVisitor extends AbstractPoiExcelCellAttributeVisit
 		map.put("vertical_alignment", new AttributeSupplier<CellStyle>() {
 			@Override
 			public Object get(Column column, Cell cell, CellStyle style) {
-				return (long) style.getVerticalAlignment();
+				return (long) style.getVerticalAlignmentEnum().getCode();
 			}
 		});
 		map.put("wrap_text", new AttributeSupplier<CellStyle>() {

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 +17,10 @@ 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.bean.PoiExcelColumnBean.SearchMergedCell;
 import org.embulk.parser.poi_excel.visitor.embulk.CellVisitor;
+import org.embulk.parser.poi_excel.visitor.util.MergedRegionMap;
 import org.embulk.spi.Column;
 import org.embulk.spi.Exec;
 import org.embulk.spi.PageBuilder;
@@ -38,15 +42,15 @@ 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 +58,13 @@ 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:
@@ -69,46 +73,111 @@ 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;
+		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;
+			}
+		}
+
+		visitor.visitCellValueBlank(column, cell);
+	}
+
+	protected CellRangeAddress findRegion(PoiExcelColumnBean bean, Cell cell) {
+		SearchMergedCell search = bean.getSearchMergedCell();
+		switch (search) {
+		case NONE:
+			return null;
+		case LINEAR_SEARCH:
+			return findRegionLinearSearch(bean, cell);
+		default:
+			return findRegionTreeSearch(bean, cell);
 		}
+	}
 
+	protected CellRangeAddress findRegionLinearSearch(PoiExcelColumnBean bean, Cell cell) {
+		Sheet sheet = cell.getSheet();
 		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;
-				}
-
-				visitCellValue(bean, firstCell, visitor);
-				return;
+			CellRangeAddress region = sheet.getMergedRegion(i);
+			if (region.isInRange(r, c)) {
+				return region;
 			}
 		}
 
-		visitor.visitCellValueBlank(column, cell);
+		return null;
+	}
+
+	private final MergedRegionMap mergedRegionMap = new MergedRegionMap();
+
+	protected CellRangeAddress findRegionTreeSearch(PoiExcelColumnBean bean, Cell cell) {
+		Sheet sheet = cell.getSheet();
+		int r = cell.getRowIndex();
+		int c = cell.getColumnIndex();
+
+		return mergedRegionMap.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 +228,24 @@ 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));
 		}