embulk-parser-poi_excel 0.1.4 → 0.1.10

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_cellAddress.java

@@ -0,0 +1,69 @@
+package org.embulk.parser.poi_excel;
+
+import static org.hamcrest.CoreMatchers.is;
+import static org.hamcrest.MatcherAssert.assertThat;
+
+import java.net.URL;
+import java.text.ParseException;
+import java.util.List;
+
+import org.embulk.parser.EmbulkPluginTester;
+import org.embulk.parser.EmbulkTestOutputPlugin.OutputRecord;
+import org.embulk.parser.EmbulkTestParserConfig;
+import org.junit.experimental.theories.DataPoints;
+import org.junit.experimental.theories.Theories;
+import org.junit.experimental.theories.Theory;
+import org.junit.runner.RunWith;
+
+@RunWith(Theories.class)
+public class TestPoiExcelParserPlugin_cellAddress {
+
+	@DataPoints
+	public static String[] FILES = { "test1.xls", "test2.xlsx" };
+
+	@Theory
+	public void testCellAddress(String excelFile) throws ParseException {
+		try (EmbulkPluginTester tester = new EmbulkPluginTester()) {
+			tester.addParserPlugin(PoiExcelParserPlugin.TYPE, PoiExcelParserPlugin.class);
+
+			EmbulkTestParserConfig parser = tester.newParserConfig(PoiExcelParserPlugin.TYPE);
+			parser.set("sheet", "test1");
+			parser.set("skip_header_lines", 1);
+			parser.addColumn("text", "string").set("column_number", "D");
+			parser.addColumn("fix_value", "string").set("cell_address", "B1").set("value", "cell_value");
+			parser.addColumn("fix_sheet", "string").set("cell_address", "B1").set("value", "sheet_name");
+			parser.addColumn("fix_row", "long").set("cell_address", "B1").set("value", "row_number");
+			parser.addColumn("fix_col", "long").set("cell_address", "B1").set("value", "column_number");
+			parser.addColumn("other_sheet_value", "string").set("cell_address", "style!B5").set("value", "cell_value");
+			parser.addColumn("other_sheet_name", "string").set("cell_address", "style!B5").set("value", "sheet_name");
+			parser.addColumn("other_sheet_row", "long").set("cell_address", "style!B5").set("value", "row_number");
+			parser.addColumn("other_sheet_col", "string").set("cell_address", "style!B5").set("value", "column_number");
+
+			URL inFile = getClass().getResource(excelFile);
+			List<OutputRecord> result = tester.runParser(inFile, parser);
+
+			assertThat(result.size(), is(7));
+			check1(result, 0, "abc");
+			check1(result, 1, "def");
+			check1(result, 2, "456");
+			check1(result, 3, "abc");
+			check1(result, 4, "abc");
+			check1(result, 5, "true");
+			check1(result, 6, null);
+		}
+	}
+
+	private void check1(List<OutputRecord> result, int index, String text) {
+		OutputRecord record = result.get(index);
+		// System.out.println(record);
+		assertThat(record.getAsString("text"), is(text));
+		assertThat(record.getAsString("fix_value"), is("long"));
+		assertThat(record.getAsString("fix_sheet"), is("test1"));
+		assertThat(record.getAsLong("fix_row"), is(1L));
+		assertThat(record.getAsLong("fix_col"), is(2L));
+		assertThat(record.getAsString("other_sheet_value"), is("bottom"));
+		assertThat(record.getAsString("other_sheet_name"), is("style"));
+		assertThat(record.getAsLong("other_sheet_row"), is(5L));
+		assertThat(record.getAsString("other_sheet_col"), is("B"));
+	}
+}

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_cellComment.java

@@ -2,7 +2,7 @@ package org.embulk.parser.poi_excel;
 
 import static org.hamcrest.CoreMatchers.is;
 import static org.hamcrest.CoreMatchers.nullValue;
-import static org.junit.Assert.assertThat;
+import static org.hamcrest.MatcherAssert.assertThat;
 import static org.junit.Assert.fail;
 
 import java.net.URL;

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_cellError.java

@@ -2,7 +2,7 @@ package org.embulk.parser.poi_excel;
 
 import static org.hamcrest.CoreMatchers.is;
 import static org.hamcrest.CoreMatchers.nullValue;
-import static org.junit.Assert.assertThat;
+import static org.hamcrest.MatcherAssert.assertThat;
 import static org.junit.Assert.fail;
 
 import java.net.URL;

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_cellFont.java

@@ -2,7 +2,7 @@ package org.embulk.parser.poi_excel;
 
 import static org.hamcrest.CoreMatchers.is;
 import static org.hamcrest.CoreMatchers.nullValue;
-import static org.junit.Assert.assertThat;
+import static org.hamcrest.MatcherAssert.assertThat;
 import static org.junit.Assert.fail;
 
 import java.net.URL;

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_cellStyle.java

@@ -1,7 +1,7 @@
 package org.embulk.parser.poi_excel;
 
 import static org.hamcrest.CoreMatchers.is;
-import static org.junit.Assert.assertThat;
+import static org.hamcrest.MatcherAssert.assertThat;
 import static org.junit.Assert.fail;
 
 import java.net.URL;
@@ -9,7 +9,7 @@ import java.text.ParseException;
 import java.util.Arrays;
 import java.util.List;
 
-import org.apache.poi.ss.usermodel.CellStyle;
+import org.apache.poi.ss.usermodel.BorderStyle;
 import org.embulk.parser.EmbulkPluginTester;
 import org.embulk.parser.EmbulkTestOutputPlugin.OutputRecord;
 import org.embulk.parser.EmbulkTestParserConfig;
@@ -44,11 +44,11 @@ public class TestPoiExcelParserPlugin_cellStyle {
 			List<OutputRecord> result = tester.runParser(inFile, parser);
 
 			assertThat(result.size(), is(5));
-			check1(result, 0, "red", 255, 0, 0, "top", CellStyle.BORDER_THIN, 0, 0, 0);
+			check1(result, 0, "red", 255, 0, 0, "top", BorderStyle.THIN.getCode(), 0, 0, 0);
 			check1(result, 1, "green", 0, 128, 0, null, 0, 0, 0, 0);
-			check1(result, 2, "blue", 0, 0, 255, "left", 0, 0, CellStyle.BORDER_THIN, 0);
-			check1(result, 3, "white", 255, 255, 255, "right", 0, 0, 0, CellStyle.BORDER_THIN);
-			check1(result, 4, "black", 0, 0, 0, "bottom", 0, CellStyle.BORDER_MEDIUM, 0, 0);
+			check1(result, 2, "blue", 0, 0, 255, "left", 0, 0, BorderStyle.THIN.getCode(), 0);
+			check1(result, 3, "white", 255, 255, 255, "right", 0, 0, 0, BorderStyle.THIN.getCode());
+			check1(result, 4, "black", 0, 0, 0, "bottom", 0, BorderStyle.MEDIUM.getCode(), 0, 0);
 		}
 	}
 
@@ -81,11 +81,11 @@ public class TestPoiExcelParserPlugin_cellStyle {
 			List<OutputRecord> result = tester.runParser(inFile, parser);
 
 			assertThat(result.size(), is(5));
-			check2(result, 0, "red", 255, 0, 0, "top", CellStyle.BORDER_THIN, 0, 0, 0);
+			check2(result, 0, "red", 255, 0, 0, "top", BorderStyle.THIN.getCode(), 0, 0, 0);
 			check2(result, 1, "green", 0, 128, 0, null, 0, 0, 0, 0);
-			check2(result, 2, "blue", 0, 0, 255, "left", 0, 0, CellStyle.BORDER_THIN, 0);
-			check2(result, 3, "white", 255, 255, 255, "right", 0, 0, 0, CellStyle.BORDER_THIN);
-			check2(result, 4, "black", 0, 0, 0, "bottom", 0, CellStyle.BORDER_MEDIUM, 0, 0);
+			check2(result, 2, "blue", 0, 0, 255, "left", 0, 0, BorderStyle.THIN.getCode(), 0);
+			check2(result, 3, "white", 255, 255, 255, "right", 0, 0, 0, BorderStyle.THIN.getCode());
+			check2(result, 4, "black", 0, 0, 0, "bottom", 0, BorderStyle.MEDIUM.getCode(), 0, 0);
 		}
 	}
 
@@ -130,11 +130,11 @@ public class TestPoiExcelParserPlugin_cellStyle {
 			List<OutputRecord> result = tester.runParser(inFile, parser);
 
 			assertThat(result.size(), is(5));
-			check2(result, 0, "red", 255, 0, 0, "top", CellStyle.BORDER_THIN, 0, 0, 0);
+			check2(result, 0, "red", 255, 0, 0, "top", BorderStyle.THIN.getCode(), 0, 0, 0);
 			check2(result, 1, "green", 0, 128, 0, null, 0, 0, 0, 0);
-			check2(result, 2, "blue", 0, 0, 255, "left", 0, 0, CellStyle.BORDER_THIN, 0);
-			check2(result, 3, "white", 255, 255, 255, "right", 0, 0, 0, CellStyle.BORDER_THIN);
-			check2(result, 4, "black", 0, 0, 0, "bottom", 0, CellStyle.BORDER_MEDIUM, 0, 0);
+			check2(result, 2, "blue", 0, 0, 255, "left", 0, 0, BorderStyle.THIN.getCode(), 0);
+			check2(result, 3, "white", 255, 255, 255, "right", 0, 0, 0, BorderStyle.THIN.getCode());
+			check2(result, 4, "black", 0, 0, 0, "bottom", 0, BorderStyle.MEDIUM.getCode(), 0, 0);
 		}
 	}
 }

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_cellType.java

@@ -0,0 +1,79 @@
+package org.embulk.parser.poi_excel;
+
+import static org.hamcrest.CoreMatchers.is;
+import static org.hamcrest.MatcherAssert.assertThat;
+
+import java.net.URL;
+import java.text.ParseException;
+import java.util.Arrays;
+import java.util.List;
+
+import org.apache.poi.ss.usermodel.CellType;
+import org.embulk.parser.EmbulkPluginTester;
+import org.embulk.parser.EmbulkTestOutputPlugin.OutputRecord;
+import org.embulk.parser.EmbulkTestParserConfig;
+import org.junit.experimental.theories.DataPoints;
+import org.junit.experimental.theories.Theories;
+import org.junit.experimental.theories.Theory;
+import org.junit.runner.RunWith;
+
+@RunWith(Theories.class)
+public class TestPoiExcelParserPlugin_cellType {
+
+	@DataPoints
+	public static String[] FILES = { "test1.xls", "test2.xlsx" };
+
+	@Theory
+	public void testCellType(String excelFile) throws ParseException {
+		try (EmbulkPluginTester tester = new EmbulkPluginTester()) {
+			tester.addParserPlugin(PoiExcelParserPlugin.TYPE, PoiExcelParserPlugin.class);
+
+			EmbulkTestParserConfig parser = tester.newParserConfig(PoiExcelParserPlugin.TYPE);
+			parser.set("sheets", Arrays.asList("test1"));
+			parser.set("skip_header_lines", 3);
+			parser.addColumn("long", "long").set("column_number", "A").set("value", "cell_type");
+			parser.addColumn("string", "string").set("column_number", "A").set("value", "cell_type");
+
+			URL inFile = getClass().getResource(excelFile);
+			List<OutputRecord> result = tester.runParser(inFile, parser);
+
+			assertThat(result.size(), is(5));
+			check1(result, 0, CellType.NUMERIC, "NUMERIC");
+			check1(result, 1, CellType.STRING, "STRING");
+			check1(result, 2, CellType.FORMULA, "FORMULA");
+			check1(result, 3, CellType.BOOLEAN, "BOOLEAN");
+			check1(result, 4, CellType.FORMULA, "FORMULA");
+		}
+	}
+
+	@Theory
+	public void testCellCachedType(String excelFile) throws ParseException {
+		try (EmbulkPluginTester tester = new EmbulkPluginTester()) {
+			tester.addParserPlugin(PoiExcelParserPlugin.TYPE, PoiExcelParserPlugin.class);
+
+			EmbulkTestParserConfig parser = tester.newParserConfig(PoiExcelParserPlugin.TYPE);
+			parser.set("sheets", Arrays.asList("test1"));
+			parser.set("skip_header_lines", 3);
+			parser.addColumn("long", "long").set("column_number", "A").set("value", "cell_cached_type");
+			parser.addColumn("string", "string").set("column_number", "A").set("value", "cell_cached_type");
+
+			URL inFile = getClass().getResource(excelFile);
+			List<OutputRecord> result = tester.runParser(inFile, parser);
+
+			assertThat(result.size(), is(5));
+			check1(result, 0, CellType.NUMERIC, "NUMERIC");
+			check1(result, 1, CellType.STRING, "STRING");
+			check1(result, 2, CellType.BOOLEAN, "BOOLEAN");
+			check1(result, 3, CellType.BOOLEAN, "BOOLEAN");
+			check1(result, 4, CellType.ERROR, "ERROR");
+		}
+	}
+
+	@SuppressWarnings("deprecation")
+	private void check1(List<OutputRecord> result, int index, CellType cellType, String s) throws ParseException {
+		OutputRecord r = result.get(index);
+		// System.out.println(r);
+		assertThat(r.getAsLong("long"), is((long) cellType.getCode()));
+		assertThat(r.getAsString("string"), is(s));
+	}
+}

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_columnNumber.java

@@ -2,7 +2,7 @@ package org.embulk.parser.poi_excel;
 
 import static org.hamcrest.CoreMatchers.is;
 import static org.hamcrest.CoreMatchers.nullValue;
-import static org.junit.Assert.assertThat;
+import static org.hamcrest.MatcherAssert.assertThat;
 
 import java.net.URL;
 import java.text.ParseException;

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_constant.java

@@ -2,7 +2,7 @@ package org.embulk.parser.poi_excel;
 
 import static org.hamcrest.CoreMatchers.is;
 import static org.hamcrest.CoreMatchers.nullValue;
-import static org.junit.Assert.assertThat;
+import static org.hamcrest.MatcherAssert.assertThat;
 
 import java.net.URL;
 import java.text.ParseException;

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_convertError.java

@@ -2,7 +2,7 @@ package org.embulk.parser.poi_excel;
 
 import static org.hamcrest.CoreMatchers.is;
 import static org.hamcrest.CoreMatchers.nullValue;
-import static org.junit.Assert.assertThat;
+import static org.hamcrest.MatcherAssert.assertThat;
 import static org.junit.Assert.fail;
 
 import java.net.URL;

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_formula.java

@@ -0,0 +1,90 @@
+package org.embulk.parser.poi_excel;
+
+import static org.hamcrest.CoreMatchers.is;
+import static org.hamcrest.MatcherAssert.assertThat;
+
+import java.net.URL;
+import java.text.ParseException;
+import java.util.Arrays;
+import java.util.List;
+
+import org.embulk.config.ConfigSource;
+import org.embulk.parser.EmbulkPluginTester;
+import org.embulk.parser.EmbulkTestOutputPlugin.OutputRecord;
+import org.embulk.parser.EmbulkTestParserConfig;
+import org.junit.experimental.theories.DataPoints;
+import org.junit.experimental.theories.Theories;
+import org.junit.experimental.theories.Theory;
+import org.junit.runner.RunWith;
+
+@RunWith(Theories.class)
+public class TestPoiExcelParserPlugin_formula {
+
+	@DataPoints
+	public static String[] FILES = { "test1.xls", "test2.xlsx" };
+
+	@Theory
+	public void testForumlaHandlingCashedValue(String excelFile) throws ParseException {
+		try (EmbulkPluginTester tester = new EmbulkPluginTester()) {
+			tester.addParserPlugin(PoiExcelParserPlugin.TYPE, PoiExcelParserPlugin.class);
+
+			EmbulkTestParserConfig parser = tester.newParserConfig(PoiExcelParserPlugin.TYPE);
+			parser.set("sheet", "formula_replace");
+
+			parser.addColumn("text", "string").set("formula_handling", "cashed_value");
+
+			URL inFile = getClass().getResource(excelFile);
+			List<OutputRecord> result = tester.runParser(inFile, parser);
+
+			assertThat(result.size(), is(2));
+			assertThat(result.get(0).getAsString("text"), is("boolean"));
+			assertThat(result.get(1).getAsString("text"), is("test2-b1"));
+		}
+	}
+
+	@Theory
+	public void testForumlaHandlingEvaluate(String excelFile) throws ParseException {
+		try (EmbulkPluginTester tester = new EmbulkPluginTester()) {
+			tester.addParserPlugin(PoiExcelParserPlugin.TYPE, PoiExcelParserPlugin.class);
+
+			EmbulkTestParserConfig parser = tester.newParserConfig(PoiExcelParserPlugin.TYPE);
+			parser.set("sheet", "formula_replace");
+
+			parser.addColumn("text", "string").set("formula_handling", "evaluate");
+
+			URL inFile = getClass().getResource(excelFile);
+			List<OutputRecord> result = tester.runParser(inFile, parser);
+
+			assertThat(result.size(), is(2));
+			assertThat(result.get(0).getAsString("text"), is("boolean"));
+			assertThat(result.get(1).getAsString("text"), is("test2-b1"));
+		}
+	}
+
+	@Theory
+	public void testForumlaReplace(String excelFile) throws ParseException {
+		try (EmbulkPluginTester tester = new EmbulkPluginTester()) {
+			tester.addParserPlugin(PoiExcelParserPlugin.TYPE, PoiExcelParserPlugin.class);
+
+			EmbulkTestParserConfig parser = tester.newParserConfig(PoiExcelParserPlugin.TYPE);
+			parser.set("sheet", "formula_replace");
+
+			ConfigSource replace0 = tester.newConfigSource();
+			replace0.set("regex", "test1");
+			replace0.set("to", "merged_cell");
+			ConfigSource replace1 = tester.newConfigSource();
+			replace1.set("regex", "B1");
+			replace1.set("to", "B${row}");
+			parser.set("formula_replace", Arrays.asList(replace0, replace1));
+
+			parser.addColumn("text", "string");
+
+			URL inFile = getClass().getResource(excelFile);
+			List<OutputRecord> result = tester.runParser(inFile, parser);
+
+			assertThat(result.size(), is(2));
+			assertThat(result.get(0).getAsString("text"), is("test3-a1"));
+			assertThat(result.get(1).getAsString("text"), is("test2-b2"));
+		}
+	}
+}

data/src/test/java/org/embulk/parser/poi_excel/TestPoiExcelParserPlugin_mergedCell.java

@@ -0,0 +1,89 @@
+package org.embulk.parser.poi_excel;
+
+import static org.hamcrest.CoreMatchers.is;
+import static org.hamcrest.MatcherAssert.assertThat;
+
+import java.net.URL;
+import java.text.ParseException;
+import java.util.List;
+
+import org.embulk.parser.EmbulkPluginTester;
+import org.embulk.parser.EmbulkTestOutputPlugin.OutputRecord;
+import org.embulk.parser.EmbulkTestParserConfig;
+import org.junit.experimental.theories.DataPoints;
+import org.junit.experimental.theories.Theories;
+import org.junit.experimental.theories.Theory;
+import org.junit.runner.RunWith;
+
+@RunWith(Theories.class)
+public class TestPoiExcelParserPlugin_mergedCell {
+
+	@DataPoints
+	public static String[] FILES = { "test1.xls", "test2.xlsx" };
+
+	@Theory
+	public void testSearchMergedCell_default(String excelFile) throws ParseException {
+		test(excelFile, null, true);
+	}
+
+	@Theory
+	public void testSearchMergedCell_true(String excelFile) throws ParseException {
+		// compatibility ver 0.1.7
+		test(excelFile, true, true);
+	}
+
+	@Theory
+	public void testSearchMergedCell_false(String excelFile) throws ParseException {
+		// compatibility ver 0.1.7
+		test(excelFile, false, false);
+	}
+
+	@Theory
+	public void testSearchMergedCell_none(String excelFile) throws ParseException {
+		test(excelFile, "none", false);
+	}
+
+	@Theory
+	public void testSearchMergedCell_linear(String excelFile) throws ParseException {
+		test(excelFile, "linear_search", true);
+	}
+
+	@Theory
+	public void testSearchMergedCell_tree(String excelFile) throws ParseException {
+		test(excelFile, "tree_search", true);
+	}
+
+	private void test(String excelFile, Object arg, boolean search) {
+		try (EmbulkPluginTester tester = new EmbulkPluginTester()) {
+			tester.addParserPlugin(PoiExcelParserPlugin.TYPE, PoiExcelParserPlugin.class);
+
+			EmbulkTestParserConfig parser = tester.newParserConfig(PoiExcelParserPlugin.TYPE);
+			parser.set("sheet", "merged_cell");
+			if (arg != null) {
+				parser.set("search_merged_cell", arg);
+			}
+			parser.addColumn("a", "string");
+			parser.addColumn("b", "string");
+
+			URL inFile = getClass().getResource(excelFile);
+			List<OutputRecord> result = tester.runParser(inFile, parser);
+
+			assertThat(result.size(), is(4));
+			if (search) {
+				check6(result, 0, "test3-a1", "test3-a1");
+			} else {
+				check6(result, 0, "test3-a1", null);
+			}
+			check6(result, 1, "data", "0");
+			check6(result, 2, null, null);
+			check6(result, 3, null, null);
+		}
+	}
+
+	private void check6(List<OutputRecord> result, int index, String a, String b) {
+		OutputRecord r = result.get(index);
+		// System.out.println(r);
+		assertThat(r.getAsString("a"), is(a));
+		assertThat(r.getAsString("b"), is(b));
+	}
+}