Modified the GuessDelimiter function (#687)

mholt · Jun 29, 2019 · 792641e · 792641e
1 parent 54f8aec
commit 792641e
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 17 deletions.
diff --git a/papaparse.js b/papaparse.js
@@ -1261,14 +1261,12 @@ License: MIT
 			return _results;
 		}
 
-		function guessDelimiter(input, newline, skipEmptyLines, comments, delimitersToGuess)
-		{
-			var bestDelim, bestDelta, fieldCountPrevRow;
+		function guessDelimiter(input, newline, skipEmptyLines, comments, delimitersToGuess) {
+			var bestDelim, bestDelta, fieldCountPrevRow, maxFieldCount;
 
 			delimitersToGuess = delimitersToGuess || [',', '\t', '|', ';', Papa.RECORD_SEP, Papa.UNIT_SEP];
 
-			for (var i = 0; i < delimitersToGuess.length; i++)
-			{
+			for (var i = 0; i < delimitersToGuess.length; i++) {
 				var delim = delimitersToGuess[i];
 				var delta = 0, avgFieldCount = 0, emptyLinesCount = 0;
 				fieldCountPrevRow = undefined;
@@ -1280,23 +1278,19 @@ License: MIT
 					preview: 10
 				}).parse(input);
 
-				for (var j = 0; j < preview.data.length; j++)
-				{
-					if (skipEmptyLines && testEmptyLine(preview.data[j]))
-					{
+				for (var j = 0; j < preview.data.length; j++) {
+					if (skipEmptyLines && testEmptyLine(preview.data[j])) {
 						emptyLinesCount++;
 						continue;
 					}
 					var fieldCount = preview.data[j].length;
 					avgFieldCount += fieldCount;
 
-					if (typeof fieldCountPrevRow === 'undefined')
-					{
-						fieldCountPrevRow = 0;
+					if (typeof fieldCountPrevRow === 'undefined') {
+						fieldCountPrevRow = fieldCount;
 						continue;
 					}
-					else if (fieldCount > 1)
-					{
+					else if (fieldCount > 0) {
 						delta += Math.abs(fieldCount - fieldCountPrevRow);
 						fieldCountPrevRow = fieldCount;
 					}
@@ -1305,11 +1299,11 @@ License: MIT
 				if (preview.data.length > 0)
 					avgFieldCount /= (preview.data.length - emptyLinesCount);
 
-				if ((typeof bestDelta === 'undefined' || delta > bestDelta)
-					&& avgFieldCount > 1.99)
-				{
+				if ((typeof bestDelta === 'undefined' || delta <= bestDelta)
+					&& (typeof maxFieldCount === 'undefined' || avgFieldCount > maxFieldCount) && avgFieldCount > 1.99) {
 					bestDelta = delta;
 					bestDelim = delim;
+					maxFieldCount = avgFieldCount;
 				}
 			}
 

diff --git a/tests/test-cases.js b/tests/test-cases.js
@@ -1203,6 +1203,16 @@ var PARSE_TESTS = [
 			errors: []
 		}
 	},
+	{
+		description: "Pipe delimiter is guessed correctly choose avgFildCount max one",
+		notes: "Guessing the delimiter should work choose the min delta one and the max one",
+		config: {},
+		input: 'a,b,c\na,b,c|d|e|f',
+		expected: {
+			data: [['a', 'b', 'c'], ['a','b','c|d|e|f']],
+			errors: []
+		}
+	},
 	{
 		description: "Pipe delimiter is guessed correctly when first field are enclosed in quotes and contain delimiter characters",
 		notes: "Guessing the delimiter should work if the first field is enclosed in quotes, but others are not",