/**
* Determines if a csv sample has a header row - not 100% accurate by any means
* It basically looks at each row in each column. If all but the first column are similar,
* it likely has a header. The way we determine this is first by type, then by length
* Other possible methods I could use to determine whether the first row is a header is I
* could look to see if all but the first CONTAIN certain characters or something - think about this
*/
public function hasHeader($data)
{
$reader = new Csv_Reader_String($data, $this->detect($data));
list($has_headers, $checked, $types, $lengths, $total_lines, $headers) = array(0, 0, array(), array(), $reader->count(), $reader->getRow());
if ($total_lines <= 2) {
// please try again with a a larger file :)
return false;
}
$total_columns = count($headers);
foreach (range(0, $total_columns - 1) as $key => $col) {
$types[$col] = null;
}
// loop through each remaining rows
while ($row = $reader->current()) {
// no need to check more than 20 lines
if ($checked > 20) {
break;
}
$checked++;
$line = $reader->key();
// loop through row and grab type for each column
foreach ($row as $col => $val) {
$types[$col][] = $this->getType($val);
$lengths[$col][] = strlen($val);
}
$reader->next();
}
// now take a vote and if more than a certain threshold have a likely header, we'll return that we think it has a header
foreach ($types as $key => $column) {
$unique = array_unique($column);
if (count($unique) == 1) {
// if all are of the same type
if ($unique[0] == $this->getType($headers[$key])) {
// all rows type matched header type, so try length now
$unique = array_unique($lengths[$key]);
if (count($unique) == 1) {
if ($unique[0] == strlen($headers[$key])) {
$has_headers--;
} else {
$has_headers++;
}
}
//printf ("%s is the same as %s<br>", $unique[0], $this->getType($headers[$key]));
} else {
$has_headers++;
}
}
}
return $has_headers > 0;
}
public function test_Reader_String()
{
$sample = "";
for ($i = 0; $i < 10; $i++) {
$sample .= "this,is,some,test,data,{$i}\r\n";
}
$reader = new Csv_Reader_String($sample);
$this->assertEqual($reader->count(), 10);
}
