Load/store DataFrames from CSV #6

* expanding CSV test suite * supporting commons-csv formats
dflib · Feb 18, 2019 · 5e8e500 · 5e8e500
1 parent b4e027f
commit 5e8e500
Show file tree

Hide file tree

Showing 4 changed files with 42 additions and 0 deletions.
diff --git a/dflib-csv/src/main/java/com/nhl/dflib/csv/CsvLoader.java b/dflib-csv/src/main/java/com/nhl/dflib/csv/CsvLoader.java
@@ -81,6 +81,19 @@ public CsvLoader columnType(String column, ValueMapper<String, ?> typeConverter)
         return this;
     }
 
+    /**
+     * Optionally sets the style or format of the imported CSV. CSVFormat comes from "commons-csv" library and
+     * contains a number of predefined formats, such as CSVFormat.MYSQL, etc. It also allows to customize the format
+     * further, by defining custom delimiters, line separators, etc.
+     *
+     * @param format a format object defined in commons-csv library
+     * @return this loader instance
+     */
+    public CsvLoader format(CSVFormat format) {
+        this.format = format;
+        return this;
+    }
+
     public DataFrame fromFile(String filePath) {
         try (Reader r = readerFromFilePath(filePath)) {
             return fromReader(r);

diff --git a/dflib-csv/src/test/java/com/nhl/dflib/csv/CsvLoaderTest.java b/dflib-csv/src/test/java/com/nhl/dflib/csv/CsvLoaderTest.java
@@ -3,6 +3,7 @@
 import com.nhl.dflib.DFAsserts;
 import com.nhl.dflib.DataFrame;
 import com.nhl.dflib.map.ValueMapper;
+import org.apache.commons.csv.CSVFormat;
 import org.junit.Test;
 
 import java.io.StringReader;
@@ -32,6 +33,26 @@ public void testFromFile() {
                 .expectRow(1, "4", "5", "6");
     }
 
+    @Test
+    public void testFromFile_DefaultFormat_Excel() {
+        DataFrame df = new CsvLoader().fromFile(csvPath("from_excel.csv"));
+        new DFAsserts(df, "A", "b", "C")
+                .expectHeight(2)
+                .expectRow(0, "commas,quotes\"'", "-85.7", "3")
+                .expectRow(1, "with, commas", "5.50001", "6");
+    }
+
+    @Test
+    public void testFromFile_MySQLFormat() {
+        DataFrame df = new CsvLoader().format(CSVFormat.MYSQL).fromFile(csvPath("from_mysql.csv"));
+        new DFAsserts(df, "1", "3365430", " xxxx")
+                .expectHeight(4)
+                .expectRow(0, "2", "2289959", "yyyy")
+                .expectRow(1, "3", "3995478", "zzzzz")
+                .expectRow(2, "4", "4112467", "nnn")
+                .expectRow(3, "5", "1474364", "eee");
+    }
+
     @Test
     public void testFromFile_Columns() {
         DataFrame df = new CsvLoader().columns("X", "Y", "Z").fromFile(csvPath("f1.csv"));

diff --git a/dflib-csv/src/test/resources/com/nhl/dflib/csv/from_excel.csv b/dflib-csv/src/test/resources/com/nhl/dflib/csv/from_excel.csv
@@ -0,0 +1,3 @@
+A,b,C
+"commas,quotes""'",-85.7,3
+"with, commas",5.50001,6
diff --git a/dflib-csv/src/test/resources/com/nhl/dflib/csv/from_mysql.csv b/dflib-csv/src/test/resources/com/nhl/dflib/csv/from_mysql.csv
@@ -0,0 +1,5 @@
+1	3365430	 xxxx
+2	2289959	yyyy
+3	3995478	zzzzz
+4	4112467	nnn
+5	1474364	eee