new feature: getting total data row count without parsing all data

This commit is contained in:
Susann Sgorzaly
2018-02-24 16:55:45 +01:00
parent ef44ea3989
commit c9cc9697ef
2 changed files with 72 additions and 0 deletions

View File

@@ -540,6 +540,41 @@ class Csv {
return $this->delimiter;
}
/**
* Get total number of rows in csv without parsing whole data.
*
* @return bool|int
*/
public function getTotalRowCount(){
if (empty($this->file_data)){
return false;
}
$this->_detect_and_remove_sep_row_from_data($this->file_data);
$pattern = sprintf('/("[^%s]*")|[^%s]*/i',$this->enclosure, $this->enclosure);
preg_match_all($pattern,$this->file_data, $matches);
foreach ($matches[0] as $match){
if (empty($match) || !preg_match("/{$this->enclosure}/", $match)){
continue;
}
$replace = str_replace(["\r", "\n"], '', $match);
$this->file_data = str_replace($match, $replace, $this->file_data);
}
$headingRow = $this->heading ? 1 : 0;
$count = substr_count($this->file_data, "\r")
+ substr_count($this->file_data, "\n")
- substr_count($this->file_data, "\r\n")
- $headingRow;
return $count;
}
// ==============================================
// ----- [ Core Functions ] ---------------------
// ==============================================

View File

@@ -198,4 +198,41 @@ class ParseTest extends TestCase
$this->assertArrayHasKey('column1', $csv->data[0], 'Data parsed incorrectly with enclosure ' . $enclosure);
$this->assertEquals('value1', $csv->data[0]['column1'], 'Data parsed incorrectly with enclosure ' . $enclosure);
}
public function countRowsProvider(){
return [
'auto-double-enclosure' => [
'auto-double-enclosure.csv',
2
],
'auto-single-enclosure' => [
'auto-single-enclosure.csv',
2
],
'UTF-8_sep_row' => [
'datatype.csv',
3
]
];
}
/**
* @dataProvider countRowsProvider
*
* @param string $file
* @param int $expectedRows
*/
public function testGetTotalRowCountFromFile($file, $expectedRows){
$this->csv->heading = true;
$this->csv->load_data(__DIR__ . '/fixtures/' . $file);
$this->assertEquals($expectedRows, $this->csv->getTotalRowCount());
}
public function testGetTotalRowCountMissingEndingLineBreak(){
$this->csv->heading = false;
$this->csv->enclosure = '"';
$sInput = "86545235689,a\r\n34365587654,b\r\n13469874576,\"c\r\nd\"";
$this->csv->load_data($sInput);
$this->assertEquals(3, $this->csv->getTotalRowCount());
}
}