Ignore empty lines for CSVs with multiple fields

Author: dehesa

sources/imperative/reader/Reader.swift

@@ -107,31 +107,39 @@ extension CSVReader {
         case .failed(let e): throw e
         }
 
-        let result: [String]?
-        do {
-            result = try self._parseLine(rowIndex: self.count.rows)
-        } catch let error {
-            self.status = .failed(error as! CSVError<CSVReader>)
-            throw error
-        }
-        
-        guard let numFields = result?.count else {
-            self.status = .finished
-            return nil
-        }
-        
-        if self.count.rows > 0 {
-            guard self.count.fields == numFields else {
-                let error = Error._invalidFieldCount(rowIndex: self.count.rows+1, parsed: numFields, expected: self.count.fields)
-                self.status = .failed(error)
+        loop: while true {
+            let result: [String]?
+            do {
+                result = try self._parseLine(rowIndex: self.count.rows)
+            } catch let error {
+                self.status = .failed(error as! CSVError<CSVReader>)
                 throw error
             }
-        } else {
-            self.count.fields = numFields
+            // If no fields were parsed, the EOF has been reached.
+            guard let fields = result else {
+                self.status = .finished
+                return nil
+            }
+            
+            let numFields = fields.count
+            // If a single empty field is received, a white line has been parsed. Ignore empty lines for CSV files were several fields are expected.
+            if numFields == 1, fields.first!.isEmpty, self.count.rows != 1 {
+                continue loop
+            }
+            
+            if self.count.rows > 0 {
+                guard self.count.fields == numFields else {
+                    let error = Error._invalidFieldCount(rowIndex: self.count.rows+1, parsed: numFields, expected: self.count.fields)
+                    self.status = .failed(error)
+                    throw error
+                }
+            } else {
+                self.count.fields = numFields
+            }
+
+            self.count.rows += 1
+            return result
         }
-        
-        self.count.rows += 1
-        return result
     }
 }
 

tests/declarative/DecodingBadInputTests.swift

@@ -17,8 +17,8 @@ extension DecodingBadInputTests {
 }
 
 extension DecodingBadInputTests {
-    /// Tests bad quoting resulting in too many fields in a particular row
-    func testBadQuoting() {
+    /// Tests bad input, in which a row in not escaped resulting in too many fields in a particular row
+    func testBadEscaping() {
         let input = """
             x,y
             A,A A
@@ -39,4 +39,16 @@ extension DecodingBadInputTests {
         let decoder = CSVDecoder { $0.headerStrategy = .firstLine }
         XCTAssertThrowsError(try decoder.decode([_Row].self, from: input))
     }
+    
+    /// Tests a valid CSV file with an extra new line delimeter at the end of the file.
+    func testExtraNewLine() throws {
+        let input = """
+            x,y
+            A,AA
+            B,BB
+            \n
+            """
+        let decoder = CSVDecoder { $0.headerStrategy = .firstLine }
+        XCTAssertNoThrow(try decoder.decode([_Row].self, from: input))
+    }
 }

tests/imperative/ReaderBadInputTests.swift

@@ -0,0 +1,54 @@
+import XCTest
+import CodableCSV
+
+/// Check support for handling bad input
+final class ReaderBadInputTests: XCTestCase {
+    override func setUp() {
+        self.continueAfterFailure = false
+    }
+}
+
+extension ReaderBadInputTests {
+    /// Representation of a CSV row containing a couple of strings.
+    private struct _Row: Codable, Equatable {
+        var x: String
+        var y: String
+    }
+}
+
+extension ReaderBadInputTests {
+    /// Tests bad input, in which a row in not escaped resulting in too many fields in a particular row
+    func testBadEscaping() throws {
+        let input = """
+            x,y
+            A,A A
+            C,C, C
+            D,D D
+            """
+        XCTAssertThrowsError(try CSVReader.decode(input: input) { $0.headerStrategy = .firstLine })
+    }
+
+    /// Tests a CSV with a header with three fields (one of them being empty) and subsequent rows with two fields.
+    func testIllFormedHeader() {
+        let input = """
+            x,y,
+            A,A A
+            B,"B, B"
+            """
+        XCTAssertThrowsError(try CSVReader.decode(input: input) { $0.headerStrategy = .firstLine })
+    }
+    
+    /// Tests a valid CSV file with an extra new line delimeter at the end of the file.
+    func testExtraNewLine() throws {
+        let input = """
+            x,y
+            A,AA
+            B,BB
+            \n
+            """
+        let reader = try CSVReader(input: input) { $0.headerStrategy = .firstLine }
+        XCTAssertNotNil(try reader.readRow())
+        XCTAssertNotNil(try reader.readRow())
+        XCTAssertNil(try reader.readRow())
+    }
+}