| 1 |
<?php |
| 2 |
|
| 3 |
namespace ImportWP\Common\Importer\File; |
| 4 |
|
| 5 |
use ImportWP\Common\Importer\FileInterface; |
| 6 |
|
| 7 |
class JSONFile extends AbstractIndexedFile implements FileInterface |
| 8 |
{ |
| 9 |
private $base_path; |
| 10 |
|
| 11 |
private $chunk; |
| 12 |
private $chunk_size = 8192; |
| 13 |
private $chunk_offset = 0; |
| 14 |
|
| 15 |
private $record_counter = 0; |
| 16 |
private $record_start_index = 0; |
| 17 |
|
| 18 |
public function setRecordPath($path) |
| 19 |
{ |
| 20 |
$this->base_path = "\"{$path}\""; |
| 21 |
} |
| 22 |
|
| 23 |
/** |
| 24 |
* Generate record file positions |
| 25 |
* |
| 26 |
* Loop through each record and save each position |
| 27 |
*/ |
| 28 |
protected function generateIndex() |
| 29 |
{ |
| 30 |
$this->record_counter = 0; |
| 31 |
$this->record_start_index = 0; |
| 32 |
|
| 33 |
rewind($this->getFileHandle()); |
| 34 |
while (!feof($this->getFileHandle())) { |
| 35 |
|
| 36 |
$this->chunk .= $this->getChunk(); |
| 37 |
$this->processChunk(); |
| 38 |
} |
| 39 |
} |
| 40 |
|
| 41 |
/** |
| 42 |
* Read chunk from file |
| 43 |
* |
| 44 |
* @return bool|string |
| 45 |
*/ |
| 46 |
private function getChunk() |
| 47 |
{ |
| 48 |
return fread($this->getFileHandle(), $this->chunk_size); |
| 49 |
} |
| 50 |
|
| 51 |
private function processChunk() |
| 52 |
{ |
| 53 |
|
| 54 |
$regex_parts = [ |
| 55 |
'"[^"\\\\]*(?:\\\\.[^"\\\\]*)*"', // skip escaped \" |
| 56 |
"'[^'\\\\]*(?:\\\\.[^'\\\\]*)*'", // skip escaped \' |
| 57 |
'([^,"\'{}\[\]:\s]+)', // values not in quotes or double quotes |
| 58 |
'({)', |
| 59 |
'(})', |
| 60 |
'(\[)', |
| 61 |
'(\])', |
| 62 |
'(:)', |
| 63 |
'(,)', |
| 64 |
]; |
| 65 |
|
| 66 |
$open_tags = ['{', '[']; |
| 67 |
$close_tags = ['}', ']']; |
| 68 |
|
| 69 |
$regex = '/' . implode('|', $regex_parts) . '/s'; |
| 70 |
$track_depth = false; |
| 71 |
$track_depth_init = false; |
| 72 |
$depth = 0; |
| 73 |
$depth_to_capture = 1; |
| 74 |
|
| 75 |
while (preg_match($regex, $this->chunk, $matches, PREG_OFFSET_CAPTURE) !== 0) { |
| 76 |
list($captured, $offset) = $matches[0]; |
| 77 |
|
| 78 |
$this->record_offset = $offset; |
| 79 |
$depth_changed = false; |
| 80 |
|
| 81 |
if (true === $track_depth) { |
| 82 |
|
| 83 |
if (in_array($captured, $open_tags)) { |
| 84 |
$depth++; |
| 85 |
$depth_changed = true; |
| 86 |
} elseif (in_array($captured, $close_tags)) { |
| 87 |
$depth--; |
| 88 |
$depth_changed = true; |
| 89 |
} |
| 90 |
|
| 91 |
if ($depth_changed) { |
| 92 |
if ($depth === 0) { |
| 93 |
$this->record_start_index = $this->chunk_offset + $this->record_offset; |
| 94 |
} elseif ($depth < -1) { |
| 95 |
$track_depth = false; |
| 96 |
} elseif ($depth === -1) { |
| 97 |
$this->setIndex( |
| 98 |
$this->record_counter, |
| 99 |
$this->record_start_index, |
| 100 |
$this->chunk_offset + $this->record_offset + strlen($captured) |
| 101 |
); |
| 102 |
$this->record_counter++; |
| 103 |
} |
| 104 |
} |
| 105 |
} elseif ($track_depth_init && $depth_to_capture >= 0) { |
| 106 |
if (in_array($captured, $open_tags)) { |
| 107 |
$depth_to_capture--; |
| 108 |
if ($depth_to_capture < 0) { |
| 109 |
$track_depth = true; |
| 110 |
$this->record_start_index = $this->chunk_offset + $this->record_offset; |
| 111 |
} |
| 112 |
} |
| 113 |
} |
| 114 |
|
| 115 |
|
| 116 |
if ($captured === $this->base_path) { |
| 117 |
$track_depth_init = true; |
| 118 |
} |
| 119 |
|
| 120 |
$string_offset = $offset + strlen($captured); |
| 121 |
$this->chunk_offset += $string_offset; |
| 122 |
$this->chunk = substr($this->chunk, $string_offset); |
| 123 |
} |
| 124 |
} |
| 125 |
} |
| 126 |
|
