|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
require_once __DIR__ . '/../vendor/autoload.php'; |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* Demonstration how some data from internet can be downloaded and kept in sync. |
|
7
|
|
|
* |
|
8
|
|
|
* Demo downloads country codes and their ISO-3166-2 codes. |
|
9
|
|
|
* The source is a wiki page https://en.wikipedia.org/wiki/ISO_3166-2 |
|
10
|
|
|
* It contains the list of countries already sorted by code. |
|
11
|
|
|
*/ |
|
12
|
|
|
|
|
13
|
|
|
$pdo = new PDO('sqlite:demo.sqlite'); |
|
14
|
|
|
$pdo->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION); |
|
15
|
|
|
$pdo->query('CREATE TABLE IF NOT EXISTS countries (code TEXT NOT NULL, name TEXT NOT NULL)'); |
|
16
|
|
|
|
|
17
|
|
|
/** |
|
18
|
|
|
* Generator returns country names from internet. |
|
19
|
|
|
* Countries are ordered by code and then by name. |
|
20
|
|
|
* The returned string is in format '<code><name>' ie the two first characters are the country code. |
|
21
|
|
|
*/ |
|
22
|
|
|
function source(): Generator |
|
23
|
|
|
{ |
|
24
|
|
|
$ch = curl_init(); |
|
25
|
|
|
try { |
|
26
|
|
|
|
|
27
|
|
|
curl_setopt($ch, CURLOPT_URL, 'https://en.wikipedia.org/wiki/ISO_3166-2'); |
|
28
|
|
|
curl_setopt($ch, CURLOPT_HEADER, 0); |
|
29
|
|
|
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1); |
|
30
|
|
|
|
|
31
|
|
|
$content = curl_exec($ch); |
|
32
|
|
|
if ($content === false) { |
|
33
|
|
|
throw new Exception('Could not download country names and codes from wiki page. ' . curl_error($ch)); |
|
34
|
|
|
} |
|
35
|
|
|
} finally { |
|
36
|
|
|
curl_close($ch); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
|
|
40
|
|
|
$dom = new DOMDocument(); |
|
41
|
|
|
|
|
42
|
|
|
$use_errors = libxml_use_internal_errors(true); |
|
43
|
|
|
try { |
|
44
|
|
|
if ($dom->loadHTML($content, LIBXML_NOWARNING) === false) { |
|
45
|
|
|
throw new Exception('Source data loading failed. Could not parse the wiki page. ' . libxml_get_last_error()); |
|
|
|
|
|
|
46
|
|
|
} |
|
47
|
|
|
} finally { |
|
48
|
|
|
libxml_clear_errors(); |
|
49
|
|
|
libxml_use_internal_errors($use_errors); |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
$xpath = new DOMXPath($dom); |
|
53
|
|
|
|
|
54
|
|
|
// we know the first one is the table we are interested in it. |
|
55
|
|
|
$tables = $xpath->query('//table'); |
|
56
|
|
|
$table = $tables->item(0); |
|
57
|
|
|
$rows = $xpath->query("./tbody/tr/td/a[starts-with(@href, '/wiki/ISO_3166-2:')]/parent::td/parent::tr", $table); |
|
58
|
|
|
|
|
59
|
|
|
if ($rows === false or $rows->length === 0) { // something went wrong. Terminate in order to prevent deleting current target. |
|
60
|
|
|
throw new \Exception('Source data loading failed. Could not find companies data in wiki page content.'); |
|
61
|
|
|
} |
|
62
|
|
|
|
|
63
|
|
|
foreach ($rows as $row) { |
|
64
|
|
|
$columns = $row->getElementsByTagName('td'); |
|
65
|
|
|
$code = trim($columns[0]->nodeValue); |
|
66
|
|
|
$name = trim($columns[1]->nodeValue); |
|
67
|
|
|
|
|
68
|
|
|
yield $code . $name; |
|
69
|
|
|
} |
|
70
|
|
|
} |
|
71
|
|
|
|
|
72
|
|
|
/** |
|
73
|
|
|
* Generator returning country names from target database. |
|
74
|
|
|
* Countries are ordered by code and then by name. |
|
75
|
|
|
* The returned string is in format '<code><name>' ie the two first characters are the country code. |
|
76
|
|
|
*/ |
|
77
|
|
|
function target($pdo): Generator |
|
78
|
|
|
{ |
|
79
|
|
|
$sql = 'SELECT code, name FROM countries ORDER BY code ASC, name ASC'; |
|
80
|
|
|
foreach ($pdo->query($sql) as $row) { |
|
81
|
|
|
yield $row['code'] . $row['name']; |
|
82
|
|
|
} |
|
83
|
|
|
} |
|
84
|
|
|
|
|
85
|
|
|
$sthAdd = $pdo->prepare('INSERT INTO countries (code, name) VALUES (?, ?)'); |
|
86
|
|
|
$sthRemove = $pdo->prepare('DELETE FROM countries WHERE code=?'); |
|
87
|
|
|
|
|
88
|
|
|
$counts = [ |
|
89
|
|
|
'added' => 0, |
|
90
|
|
|
'removed' => 0, |
|
91
|
|
|
]; |
|
92
|
|
|
|
|
93
|
|
|
$synchronization = new \Raigu\OrderedListsSynchronization\Synchronization(); |
|
94
|
|
|
$synchronization( |
|
95
|
|
|
source(), |
|
96
|
|
|
target($pdo), |
|
97
|
|
|
function ($element) use ($sthAdd, &$counts) { |
|
98
|
|
|
$code = substr($element, 0, 2); |
|
99
|
|
|
$name = substr($element, 2); |
|
100
|
|
|
echo "+ {$code}" . PHP_EOL; |
|
101
|
|
|
$counts['added'] += 1; |
|
102
|
|
|
$sthAdd->execute([$code, $name]); |
|
103
|
|
|
}, |
|
104
|
|
|
function ($element) use ($sthRemove, &$counts) { |
|
105
|
|
|
$code = substr($element, 0, 2); |
|
106
|
|
|
echo "- {$code}" . PHP_EOL; |
|
107
|
|
|
$counts['removed'] += 1; |
|
108
|
|
|
$sthRemove->execute([$code]); |
|
109
|
|
|
} |
|
110
|
|
|
); |
|
111
|
|
|
|
|
112
|
|
|
echo "Added: {$counts['added']}" . PHP_EOL; |
|
113
|
|
|
echo "Removed: {$counts['removed']}" . PHP_EOL; |
|
114
|
|
|
|
raigu /
ordered-lists-sync
Loading history...