embulk-parser-poi_excel 0.1.4 → 0.1.11

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

@@ -11,6 +11,9 @@ import org.embulk.parser.poi_excel.PoiExcelParserPlugin.ColumnCommonOptionTask;
 import org.embulk.parser.poi_excel.PoiExcelParserPlugin.ColumnOptionTask;
 import org.embulk.parser.poi_excel.PoiExcelParserPlugin.FormulaReplaceTask;
 import org.embulk.parser.poi_excel.bean.PoiExcelColumnBean.ErrorStrategy.Strategy;
+import org.embulk.parser.poi_excel.bean.util.PoiExcelCellAddress;
+import org.embulk.parser.poi_excel.bean.util.SearchMergedCell;
+import org.embulk.parser.poi_excel.visitor.util.MergedRegionFinder;
 import org.embulk.spi.Column;
 
 import com.google.common.base.Optional;
@@ -107,6 +110,25 @@ public class PoiExcelColumnBean {
 		return Optional.absent();
 	}
 
+	private Optional<PoiExcelCellAddress> cellAddress;
+
+	public PoiExcelCellAddress getCellAddress() {
+		if (cellAddress == null) {
+			this.cellAddress = initializeCellAddress();
+		}
+		return cellAddress.orNull();
+	}
+
+	protected Optional<PoiExcelCellAddress> initializeCellAddress() {
+		for (ColumnOptionTask task : columnTaskList) {
+			Optional<String> option = task.getCellAddress();
+			if (option.isPresent()) {
+				return Optional.of(new PoiExcelCellAddress(option.get()));
+			}
+		}
+		return Optional.absent();
+	}
+
 	protected abstract class CacheValue<T> {
 		private T value;
 
@@ -231,23 +253,105 @@ public class PoiExcelColumnBean {
 		return attributeName.get();
 	}
 
-	private CacheValue<Boolean> searchMergedCell = new CacheValue<Boolean>() {
+	private CacheValue<String> numericFormat = new CacheValue<String>() {
 
 		@Override
-		protected Optional<Boolean> getTaskValue(ColumnCommonOptionTask task) {
-			return task.getSearchMergedCell();
+		protected Optional<String> getTaskValue(ColumnCommonOptionTask task) {
+			return task.getNumericFormat();
 		}
 
 		@Override
-		protected Boolean getDefaultValue() {
-			return true;
+		protected String getDefaultValue() {
+			return "";
 		}
 	};
 
-	public boolean getSearchMergedCell() {
+	public String getNumericFormat() {
+		return numericFormat.get();
+	}
+
+	private CacheValue<SearchMergedCell> searchMergedCell = new CacheValue<SearchMergedCell>() {
+
+		@Override
+		protected Optional<SearchMergedCell> getTaskValue(ColumnCommonOptionTask task) {
+			Optional<String> option = task.getSearchMergedCell();
+			String value = option.or("null").trim();
+			switch (value.toLowerCase()) {
+			case "null":
+				return Optional.absent();
+			case "true": // compatibility ver 0.1.7
+				return Optional.of(getDefaultValue());
+			case "false": // compatibility ver 0.1.7
+				return Optional.of(SearchMergedCell.NONE);
+			default:
+				break;
+			}
+			try {
+				return Optional.of(SearchMergedCell.valueOf(value.toUpperCase()));
+			} catch (Exception e) {
+				List<String> list = new ArrayList<>();
+				for (SearchMergedCell s : SearchMergedCell.values()) {
+					list.add(s.name().toLowerCase());
+				}
+				throw new ConfigException(MessageFormat.format("illegal search_merged_cell={0}. expected={1}", value,
+						list), e);
+			}
+		}
+
+		@Override
+		protected SearchMergedCell getDefaultValue() {
+			return SearchMergedCell.HASH_SEARCH;
+		}
+	};
+
+	public SearchMergedCell getSearchMergedCell() {
 		return searchMergedCell.get();
 	}
 
+	private MergedRegionFinder mergedRegionFinder;
+
+	public MergedRegionFinder getMergedRegionFinder() {
+		if (mergedRegionFinder == null) {
+			this.mergedRegionFinder = getSearchMergedCell().getMergedRegionFinder();
+		}
+		return mergedRegionFinder;
+	}
+
+	public enum FormulaHandling {
+		EVALUATE, CASHED_VALUE
+	}
+
+	private CacheValue<FormulaHandling> formulaHandling = new CacheValue<FormulaHandling>() {
+
+		@Override
+		protected Optional<FormulaHandling> getTaskValue(ColumnCommonOptionTask task) {
+			Optional<String> option = task.getFormulaHandling();
+			String value = option.or("null");
+			if ("null".equalsIgnoreCase(value)) {
+				return Optional.absent();
+			}
+			try {
+				return Optional.of(FormulaHandling.valueOf(value.trim().toUpperCase()));
+			} catch (Exception e) {
+				List<String> list = new ArrayList<>();
+				for (FormulaHandling s : FormulaHandling.values()) {
+					list.add(s.name().toLowerCase());
+				}
+				throw new ConfigException(MessageFormat.format("illegal formula_handling={0}. expected={1}", value,
+						list), e);
+			}
+		}
+
+		@Override
+		protected FormulaHandling getDefaultValue() {
+			return FormulaHandling.EVALUATE;
+		}
+	};
+
+	public FormulaHandling getFormulaHandling() {
+		return formulaHandling.get();
+	}
+
 	private CacheValue<List<FormulaReplaceTask>> formulaReplace = new CacheValue<List<FormulaReplaceTask>>() {
 
 		@Override

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/bean/util/SearchMergedCell.java

@@ -0,0 +1,71 @@
+package org.embulk.parser.poi_excel.bean.util;
+
+import java.util.HashMap;
+import java.util.Map;
+import java.util.TreeMap;
+
+import org.apache.poi.ss.util.CellRangeAddress;
+import org.embulk.parser.poi_excel.visitor.util.MergedRegionFinder;
+import org.embulk.parser.poi_excel.visitor.util.MergedRegionList;
+import org.embulk.parser.poi_excel.visitor.util.MergedRegionMap;
+import org.embulk.parser.poi_excel.visitor.util.MergedRegionNothing;
+
+public enum SearchMergedCell {
+	NONE {
+		@Override
+		public MergedRegionFinder createMergedRegionFinder() {
+			return new MergedRegionNothing();
+		}
+	},
+	LINEAR_SEARCH {
+		@Override
+		public MergedRegionFinder createMergedRegionFinder() {
+			return new MergedRegionList();
+		}
+	},
+	TREE_SEARCH {
+		@Override
+		public MergedRegionFinder createMergedRegionFinder() {
+			return new MergedRegionMap() {
+
+				@Override
+				protected Map<Integer, Map<Integer, CellRangeAddress>> newRowMap() {
+					return new TreeMap<>();
+				}
+
+				@Override
+				protected Map<Integer, CellRangeAddress> newColumnMap() {
+					return new TreeMap<>();
+				}
+			};
+		}
+	},
+	HASH_SEARCH {
+		@Override
+		public MergedRegionFinder createMergedRegionFinder() {
+			return new MergedRegionMap() {
+
+				@Override
+				protected Map<Integer, Map<Integer, CellRangeAddress>> newRowMap() {
+					return new HashMap<>();
+				}
+
+				@Override
+				protected Map<Integer, CellRangeAddress> newColumnMap() {
+					return new HashMap<>();
+				}
+			};
+		}
+	};
+
+	private MergedRegionFinder mergedRegionFinder;
+
+	public MergedRegionFinder getMergedRegionFinder() {
+		if (mergedRegionFinder == null) {
+			this.mergedRegionFinder = createMergedRegionFinder();
+		}
+		return mergedRegionFinder;
+	}
+
+	protected abstract MergedRegionFinder createMergedRegionFinder();
+}

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>() {