CSVReader/Parser: Added option to trim unquoted strings

src/main/scala/com/github/tototoshi/csv/CSVFormat.scala

@@ -29,4 +29,6 @@ trait CSVFormat extends Serializable {
 
   val treatEmptyLineAsNil: Boolean
 
+  val trimUnquoted: Boolean
+
 }

src/main/scala/com/github/tototoshi/csv/CSVParser.scala

@@ -37,7 +37,7 @@ object CSVParser {
    * res1: Option[List[String]] = Some(List(a, b, c))
    * }}}
    */
-  def parse(input: String, escapeChar: Char, delimiter: Char, quoteChar: Char): Option[List[String]] = {
+  def parse(input: String, escapeChar: Char, delimiter: Char, quoteChar: Char, trimUnquoted: Boolean = false): Option[List[String]] = {
     val buf: Array[Char] = input.toCharArray
     var fields: Vector[String] = Vector()
     var field = new StringBuilder
@@ -148,13 +148,13 @@ object CSVParser {
               }
             }
             case `delimiter` => {
-              fields :+= field.toString
+              fields :+= (if (trimUnquoted) field.toString().trim else field.toString())
               field = new StringBuilder
               state = Delimiter
               pos += 1
             }
             case '\n' | '\u2028' | '\u2029' | '\u0085' => {
-              fields :+= field.toString
+              fields :+= (if (trimUnquoted) field.toString().trim else field.toString())
               field = new StringBuilder
               state = End
               pos += 1
@@ -163,7 +163,7 @@ object CSVParser {
               if (pos + 1 < buflen && buf(1) == '\n') {
                 pos += 1
               }
-              fields :+= field.toString
+              fields :+= (if (trimUnquoted) field.toString().trim else field.toString())
               field = new StringBuilder
               state = End
               pos += 1
@@ -300,7 +300,7 @@ object CSVParser {
 class CSVParser(format: CSVFormat) extends Serializable {
 
   def parseLine(input: String): Option[List[String]] = {
-    val parsedResult = CSVParser.parse(input, format.escapeChar, format.delimiter, format.quoteChar)
+    val parsedResult = CSVParser.parse(input, format.escapeChar, format.delimiter, format.quoteChar, format.trimUnquoted)
     if (parsedResult == Some(List("")) && format.treatEmptyLineAsNil) Some(Nil)
     else parsedResult
   }

src/main/scala/com/github/tototoshi/csv/Formats.scala

@@ -29,6 +29,8 @@ trait DefaultCSVFormat extends CSVFormat {
 
   val treatEmptyLineAsNil: Boolean = false
 
+  val trimUnquoted: Boolean = false
+
 }
 
 trait TSVFormat extends CSVFormat {
@@ -45,5 +47,7 @@ trait TSVFormat extends CSVFormat {
 
   val treatEmptyLineAsNil: Boolean = false
 
+  val trimUnquoted: Boolean = false
+
 }
 

src/test/resources/trim-unquoted.csv

@@ -0,0 +1,2 @@
+a, b, c
+"a"," b","c "

src/test/scala/com/github/tototoshi/csv/CSVReaderSpec.scala

@@ -10,6 +10,16 @@ class CSVReaderSpec extends FunSpec with Matchers with Using {
 
   }
 
+  private def read(fileName: String)(implicit format: CSVFormat): List[String] = {
+    var res: List[String] = Nil
+    using(CSVReader.open(fileName)(format)) { reader =>
+      reader foreach { fields =>
+        res = res ++ fields
+      }
+    }
+    res
+  }
+
   describe("CSVReader") {
 
     it("should be constructed with java.io.File") {
@@ -23,13 +33,7 @@ class CSVReaderSpec extends FunSpec with Matchers with Using {
     }
 
     it("should be constructed with filename") {
-      var res: List[String] = Nil
-      using(CSVReader.open("src/test/resources/simple.csv")) { reader =>
-        reader foreach { fields =>
-          res = res ++ fields
-        }
-      }
-      res.mkString should be("abcdef")
+      read("src/test/resources/simple.csv").mkString should be("abcdef")
     }
 
     it("should be constructed with CSVFormat") {
@@ -83,13 +87,7 @@ class CSVReaderSpec extends FunSpec with Matchers with Using {
     }
 
     it("read simple CSV from file") {
-      var res: List[String] = Nil
-      using(CSVReader.open(new FileReader("src/test/resources/simple.csv"))) { reader =>
-        reader foreach { fields =>
-          res = res ++ fields
-        }
-      }
-      res.mkString should be("abcdef")
+      read("src/test/resources/simple.csv").mkString should be("abcdef")
     }
 
     it("read simple CSV string") {
@@ -104,56 +102,35 @@ class CSVReaderSpec extends FunSpec with Matchers with Using {
     }
 
     it("issue #22") {
-      var res: List[String] = Nil
-      using(CSVReader.open("src/test/resources/issue22.csv")) { reader =>
-        reader foreach { fields =>
-          res = res ++ fields
-        }
-      }
+      read("src/test/resources/issue22.csv")
     }
 
     it("issue #32") {
-      var res: List[String] = Nil
-      using(CSVReader.open("src/test/resources/issue32.csv")(new DefaultCSVFormat {
+      read("src/test/resources/issue32.csv")(new DefaultCSVFormat {
         override val escapeChar: Char = '\\'
-      })) { reader =>
-        reader foreach { fields =>
-          res = res ++ fields
-        }
-      }
+      })
     }
 
     it("should read csv file whose escape char is backslash") {
-      var res: List[String] = Nil
       implicit val format = new DefaultCSVFormat {
         override val escapeChar: Char = '\\'
       }
-      using(CSVReader.open("src/test/resources/backslash-escape.csv")(format)) { reader =>
-        reader foreach { fields =>
-          res = res ++ fields
-        }
-      }
+      val res: List[String] = read("src/test/resources/backslash-escape.csv")(format)
       res should be(List("field1", "field2", "field3 says, \"escaped with backslash\""))
     }
 
     it("read simple CSV file with empty quoted fields") {
-      var res: List[String] = Nil
-      using(CSVReader.open("src/test/resources/issue30.csv")) { reader =>
-        reader foreach { fields =>
-          res = res ++ fields
-        }
-      }
-      res.mkString(",") should be("h1,h2,h3,a1,,a3,b1,b2,b3")
+      read("src/test/resources/issue30.csv").mkString(",") should be("h1,h2,h3,a1,,a3,b1,b2,b3")
     }
 
     it("should read a file starting with BOM") {
-      var res: List[String] = Nil
-      using(CSVReader.open("src/test/resources/bom.csv")) { reader =>
-        reader foreach { fields =>
-          res = res ++ fields
-        }
-      }
-      res should be(List("a", "b", "c"))
+      read("src/test/resources/bom.csv") should be(List("a", "b", "c"))
+    }
+
+    it("should trim unquoted fields when wanted") {
+      read("src/test/resources/trim-unquoted.csv")(new DefaultCSVFormat {
+        override val trimUnquoted: Boolean = true
+      }) should be(List("a", "b", "c", "a", " b", "c "))
     }
 
     it("should be throw exception against malformed input") {