embulk-parser-poi_excel 0.1.7 → 0.1.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 3ab71cb044d2238a08719c80b1092f50cce9b84e
-  data.tar.gz: c02f1e64148806369e8aca0e11a86daba48f134f
+  metadata.gz: '058cae76dc2aff62982491ddd1d29f0bc707d030'
+  data.tar.gz: 4b7652888f66faa0b3b3653cda9c3ba4b9d49887
 SHA512:
-  metadata.gz: 0966c93aba6475e4c77a1a987be10a0413025bb96f7925b9f41413802ad9217c1366b06b1378369e7ee9dcad8b70ca843532cac70d606c08aa0553a9195f108e
-  data.tar.gz: 06a1f63b2ba99e9f88410e72d1a29b5b3eb3b221394674a1568b9d4ed32a3292a95e93ed6639c2f26b6be513ee0f15d0199e96ee7b35ff4ba9082947f48908c5
+  metadata.gz: b700c88e34a05edd16d1d7444053540fae8f8ef7d596c709505f070ca4d3d0582e9c69da6cc253b97a595cbe5d3aad60f1a1bbb75b90110e573575ba32c4d752
+  data.tar.gz: a36bc92ebd495a908f98c53cb92ebd3c84c294df3cde9d2ea926f0c452c9a6d2fcce7646f9f906a6ed25b21fb14de1a2ecc6ea5ee6e8aeec10c8b46346efd070

data/README.md

@@ -44,13 +44,15 @@ if omit **column_number** when **value** is `cell_value`, specified next column.
 * **type**: Embulk column type. (string, required)
 * **value**: value type. see below. (string, defualt: `cell_value`)
 * **column_number**: Excel column number. see below. (string, default: next column)
+* **cell_address**: Excel cell address such as `A1`, `Sheet1!B3`. only one of `column_number`, `cell_address` can be specified. (string, not required)
 * **numeric_format**: format of numeric(double) to string such as `%4.2f`. (default: Java's Double.toString())
 * **attribute_name**: use with value `cell_style`, `cell_font`, etc. see below. (list of string)
 * **on_cell_error**: processing method of Cell error. see below. (string, default: `constant`)
-* **formula_handling** : processing method of formula. see below. (`evaluate` or `cashed_value`. default: `evaluate`)
+* **formula_handling**: processing method of formula. see below. (`evaluate` or `cashed_value`. default: `evaluate`)
 * **on_evaluate_error**: processing method of evaluate formula error. see below. (string, default: `exception`)
-* **formula_replace** : replace formula before evaluate. see below.
+* **formula_replace**: replace formula before evaluate. see below.
 * **on_convert_error**: processing method of convert error. see below. (string, default: `exception`)
+* **search_merged_cell**: search merged cell when cell is BLANK. one of `none`, `linear_search`, `tree_search`. (string, default: `tree_search`)
 
 ### value
 
@@ -60,7 +62,7 @@ if omit **column_number** when **value** is `cell_value`, specified next column.
 * `cell_font`: all cell font attributes. returned json string. see **attribute_name**. (**type** required `string`)
 * `cell_comment`: all cell comment attributes. returned json string. see **attribute_name**. (**type** required `string`)
 * `cell_type`: cell type. returned Cell.getCellType() of POI.
-* `cell_cached_type`: cell cached formula result type. returned Cell.getCachedFormulaResultType() of POI when CellType==FORMULA, otherwise same as cell_type(returned Cell.getCellType()).
+* `cell_cached_type`: cell cached formula result type. returned Cell.getCachedFormulaResultType() of POI when CellType==FORMULA, otherwise same as `cell_type` (returned Cell.getCellType()).
 * `sheet_name`: sheet name.
 * `row_number`: row number(1 origin).
 * `column_number`: column number(1 origin).
@@ -126,7 +128,7 @@ Processing method of Cell error (`#DIV/0!`, `#REF!`, etc).
 ```
 
 * `constant`: set null. (default)
-* `constant.`*value*: set value.
+* `constant.`*value*: set specified value.
 * `error_code`: set error code.
 * `exception`: throw exception.
 
@@ -186,7 +188,7 @@ Processing method of convert error. ex) Excel boolean to Embulk timestamp
 
 ### sheet_options
 
-Options of indivisual sheet.
+Options of individual sheet.
 
 ```yaml
   parser:
@@ -210,7 +212,7 @@ Options of indivisual sheet.
 ```
 
 **sheet_options** is map of sheet name.  
-Map values are **skip_header_lines**, **colums**.
+Map values are **skip_header_lines**, **columns**.
 
 **columns** is map of column name.  
 Map values are same **columns** in **parser** (excluding `name`, `type`).

data/build.gradle

@@ -1,6 +1,6 @@
 plugins {
     id "com.jfrog.bintray" version "1.1"
-    id "com.github.jruby-gradle.base" version "0.1.5"
+    id "com.github.jruby-gradle.base" version "1.5.0"
     id "java"
     id "eclipse"
 }
@@ -13,7 +13,7 @@ configurations {
     provided
 }
 
-version = "0.1.7"
+version = "0.1.10"
 
 sourceCompatibility = 1.7
 targetCompatibility = 1.7
@@ -23,8 +23,8 @@ dependencies {
     provided "org.embulk:embulk-core:0.7.5"
     compile  "org.embulk:embulk-standards:0.7.5"
     // compile "YOUR_JAR_DEPENDENCY_GROUP:YOUR_JAR_DEPENDENCY_MODULE:YOUR_JAR_DEPENDENCY_VERSION"
-    compile group: 'org.apache.poi', name : 'poi', version: '3.13'
-    compile(group: 'org.apache.poi', name : 'poi-ooxml', version: '3.13') {
+    compile group: 'org.apache.poi', name : 'poi', version: '3.17'
+    compile(group: 'org.apache.poi', name : 'poi-ooxml', version: '3.17') {
         exclude group: 'stax', module: 'stax-api'
     }
     testCompile "junit:junit:4.+"
@@ -38,14 +38,22 @@ task classpath(type: Copy, dependsOn: ["jar"]) {
 clean { delete "classpath" }
 
 task gem(type: JRubyExec, dependsOn: ["gemspec", "classpath"]) {
-    jrubyArgs "-rrubygems/gem_runner", "-eGem::GemRunner.new.run(ARGV)", "build"
-    script "${project.name}.gemspec"
+    jrubyArgs "-S"
+    script "gem"
+    scriptArgs "build", "${project.name}.gemspec"
     doLast { ant.move(file: "${project.name}-${project.version}.gem", todir: "pkg") }
 }
 
 task gemPush(type: JRubyExec, dependsOn: ["gem"]) {
-    jrubyArgs "-rrubygems/gem_runner", "-eGem::GemRunner.new.run(ARGV)", "push"
-    script "pkg/${project.name}-${project.version}.gem"
+    jrubyArgs "-S"
+    script "gem"
+    scriptArgs "push", "pkg/${project.name}-${project.version}.gem"
+}
+
+task gemUnpack(type: JRubyExec) {
+    jrubyArgs "-S"
+    script "gem"
+    scriptArgs "unpack", "pkg/${project.name}-${project.version}.gem"
 }
 
 task "package"(dependsOn: ["gemspec", "classpath"]) << {

data/src/main/java/org/embulk/parser/poi_excel/PoiExcelParserPlugin.java

@@ -96,6 +96,11 @@ public class PoiExcelParserPlugin implements ParserPlugin {
 		@ConfigDefault("null")
 		public Optional<String> getColumnNumber();
 
+		// A1,B2,... or Sheet1!A1
+		@Config("cell_address")
+		@ConfigDefault("null")
+		public Optional<String> getCellAddress();
+
 		// use when value_type=cell_style, cell_font, ...
 		@Config("attribute_name")
 		@ConfigDefault("null")
@@ -111,7 +116,7 @@ public class PoiExcelParserPlugin implements ParserPlugin {
 		// search merged cell if cellType=BLANK
 		@Config("search_merged_cell")
 		@ConfigDefault("null")
-		public Optional<Boolean> getSearchMergedCell();
+		public Optional<String> getSearchMergedCell();
 
 		@Config("formula_handling")
 		@ConfigDefault("null")

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

@@ -11,6 +11,7 @@ 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.spi.Column;
 
 import com.google.common.base.Optional;
@@ -107,6 +108,25 @@ 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;
 
@@ -248,20 +268,45 @@ public class PoiExcelColumnBean {
 		return numericFormat.get();
 	}
 
-	private CacheValue<Boolean> searchMergedCell = new CacheValue<Boolean>() {
+	public enum SearchMergedCell {
+		NONE, LINEAR_SEARCH, TREE_SEARCH
+	}
+
+	private CacheValue<SearchMergedCell> searchMergedCell = new CacheValue<SearchMergedCell>() {
 
 		@Override
-		protected Optional<Boolean> getTaskValue(ColumnCommonOptionTask task) {
-			return task.getSearchMergedCell();
+		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 Boolean getDefaultValue() {
-			return true;
+		protected SearchMergedCell getDefaultValue() {
+			return SearchMergedCell.TREE_SEARCH;
 		}
 	};
 
-	public boolean getSearchMergedCell() {
+	public SearchMergedCell getSearchMergedCell() {
 		return searchMergedCell.get();
 	}
 

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;
@@ -36,45 +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 CELL_TYPE:
-					case CELL_CACHED_TYPE:
-					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);
 			}
 		}
 	}
@@ -82,6 +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) {
@@ -192,4 +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

@@ -1,6 +1,7 @@
 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;
@@ -16,23 +17,36 @@ public class PoiExcelCellTypeVisitor {
 		this.pageBuilder = visitorValue.getPageBuilder();
 	}
 
-	private static final String[] CELL_TYPE_STRING = { "NUMERIC", "STRING", "FORMULA", "BLANK", "BOOLEAN", "ERROR" };
-
-	public void visit(PoiExcelColumnBean bean, Cell cell, int cellType, CellVisitor visitor) {
+	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;
-			if (0 <= cellType && cellType < CELL_TYPE_STRING.length) {
-				type = CELL_TYPE_STRING[cellType];
-			} else {
-				type = Integer.toString(cellType);
-			}
+			String type = cellType.name();
 			visitor.visitCellValueString(column, cell, type);
 			return;
 		}
 
-		visitor.visitCellValueNumeric(column, cell, cellType);
+		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;
+		}
 	}
 }