From 139d85d87435d0ff11d24a2070c5846f8b73a53e Mon Sep 17 00:00:00 2001
From: Adrien Crivelli <adrien.crivelli@gmail.com>
Date: Thu, 28 Dec 2017 12:22:01 +0900
Subject: [PATCH] Better auto-detection of CSV separators

Closes #305
---
 CHANGELOG.md                                 | 12 ++++-
 src/PhpSpreadsheet/Reader/Csv.php            |  6 +++
 tests/PhpSpreadsheetTests/Reader/CsvTest.php | 53 ++++++++++++++++++--
 tests/data/Reader/CSV/enclosure.csv          |  4 ++
 4 files changed, 68 insertions(+), 7 deletions(-)
 create mode 100644 tests/data/Reader/CSV/enclosure.csv

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 4a7e5e03..28563460 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -5,12 +5,20 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](http://keepachangelog.com/)
 and this project adheres to [Semantic Versioning](http://semver.org/).
 
+## [Unreleased]
+
+### Added
+
+### Fixed
+
+- Better auto-detection of CSV separators - [#305](https://github.com/PHPOffice/PhpSpreadsheet/issues/305)
+
 ## [1.0.0] - 2017-12-25
 
 ### Added
 
-- Support to write merged cells in ODS format [#287](https://github.com/PHPOffice/PhpSpreadsheet/issues/287)
-- Able to set the `topLeftCell` in freeze panes [#261](https://github.com/PHPOffice/PhpSpreadsheet/pull/261) 
+- Support to write merged cells in ODS format - [#287](https://github.com/PHPOffice/PhpSpreadsheet/issues/287)
+- Able to set the `topLeftCell` in freeze panes - [#261](https://github.com/PHPOffice/PhpSpreadsheet/pull/261) 
 - Support `DateTimeImmutable` as cell value
 - Support migration of prefixed classes
 
diff --git a/src/PhpSpreadsheet/Reader/Csv.php b/src/PhpSpreadsheet/Reader/Csv.php
index 954b9e8d..2f13e368 100644
--- a/src/PhpSpreadsheet/Reader/Csv.php
+++ b/src/PhpSpreadsheet/Reader/Csv.php
@@ -155,6 +155,10 @@ class Csv extends BaseReader
         // Count how many times each of the potential delimiters appears in each line
         $numberLines = 0;
         while (($line = fgets($this->fileHandle)) !== false && (++$numberLines < 1000)) {
+            // Drop everything that is enclosed to avoid counting false positives in enclosures
+            $enclosure = preg_quote($this->enclosure, '/');
+            $line = preg_replace('/(' . $enclosure . '.*' . $enclosure . ')/U', '', $line);
+
             $countLine = [];
             for ($i = strlen($line) - 1; $i >= 0; --$i) {
                 $char = $line[$i];
@@ -223,6 +227,8 @@ class Csv extends BaseReader
      * @param string $pFilename
      *
      * @throws Exception
+     *
+     * @return array
      */
     public function listWorksheetInfo($pFilename)
     {
diff --git a/tests/PhpSpreadsheetTests/Reader/CsvTest.php b/tests/PhpSpreadsheetTests/Reader/CsvTest.php
index b1811777..003478fa 100644
--- a/tests/PhpSpreadsheetTests/Reader/CsvTest.php
+++ b/tests/PhpSpreadsheetTests/Reader/CsvTest.php
@@ -7,17 +7,60 @@ use PHPUnit\Framework\TestCase;
 
 class CsvTest extends TestCase
 {
-    public function testDelimiterDetection()
+    /**
+     * @dataProvider providerDelimiterDetection
+     *
+     * @param string $filename
+     * @param string $expectedDelimiter
+     * @param string $cell
+     * @param float|int|string $expectedValue
+     */
+    public function testDelimiterDetection($filename, $expectedDelimiter, $cell, $expectedValue)
     {
         $reader = new Csv();
         self::assertNull($reader->getDelimiter());
 
-        $filename = __DIR__ . '/../../data/Reader/CSV/semicolon_separated.csv';
         $spreadsheet = $reader->load($filename);
 
-        self::assertSame(';', $reader->getDelimiter(), 'should be able to infer the delimiter');
+        self::assertSame($expectedDelimiter, $reader->getDelimiter(), 'should be able to infer the delimiter');
 
-        $actual = $spreadsheet->getActiveSheet()->getCell('C2')->getValue();
-        self::assertSame('25,5', $actual, 'should be able to retrieve values with commas');
+        $actual = $spreadsheet->getActiveSheet()->getCell($cell)->getValue();
+        self::assertSame($expectedValue, $actual, 'should be able to retrieve correct value');
+    }
+
+    public function providerDelimiterDetection()
+    {
+        return [
+            [
+                __DIR__ . '/../../data/Reader/CSV/enclosure.csv',
+                ',',
+                'C4',
+                'username2',
+            ],
+            [
+                __DIR__ . '/../../data/Reader/CSV/semicolon_separated.csv',
+                ';',
+                'C2',
+                '25,5',
+            ],
+            [
+                __DIR__ . '/../../data/Reader/HTML/csv_with_angle_bracket.csv',
+                ',',
+                'B1',
+                'Number of items with weight <= 50kg',
+            ],
+            [
+                __DIR__ . '/../../../samples/Reader/sampleData/example1.csv',
+                ',',
+                'I4',
+                '100%',
+            ],
+            [
+                __DIR__ . '/../../../samples/Reader/sampleData/example2.csv',
+                ',',
+                'D8',
+                -58.373161,
+            ],
+        ];
     }
 }
diff --git a/tests/data/Reader/CSV/enclosure.csv b/tests/data/Reader/CSV/enclosure.csv
new file mode 100644
index 00000000..a768c63b
--- /dev/null
+++ b/tests/data/Reader/CSV/enclosure.csv
@@ -0,0 +1,4 @@
+"Date/Time","task","username"
+"12/17/2017 14:35","viewpage","username1",
+"12/17/2017 04:00","editpage","username5",
+"11/29/2017 08:20","deletepage","username2",,