| ... | ... |
@@ -1,10 +1,138 @@ |
| 1 | 1 |
<?php |
| 2 | 2 |
|
| 3 | 3 |
/** |
| 4 |
+ * Class representing restaurant |
|
| 5 |
+ */ |
|
| 6 |
+class Restaurant |
|
| 7 |
+{
|
|
| 8 |
+ |
|
| 9 |
+ private $url; |
|
| 10 |
+ private $name; |
|
| 11 |
+ private $xpaths = array(); |
|
| 12 |
+ |
|
| 13 |
+ public function getName() |
|
| 14 |
+ {
|
|
| 15 |
+ return $this->name; |
|
| 16 |
+ } |
|
| 17 |
+ |
|
| 18 |
+ public function getUrl() |
|
| 19 |
+ {
|
|
| 20 |
+ return $this->url; |
|
| 21 |
+ } |
|
| 22 |
+ |
|
| 23 |
+ public function getXpaths() |
|
| 24 |
+ {
|
|
| 25 |
+ return $this->xpaths; |
|
| 26 |
+ } |
|
| 27 |
+ |
|
| 28 |
+ /** |
|
| 29 |
+ * @param $name |
|
| 30 |
+ * @param $url |
|
| 31 |
+ * @param $xpaths |
|
| 32 |
+ */ |
|
| 33 |
+ function __construct($name, $url, $xpaths) |
|
| 34 |
+ {
|
|
| 35 |
+ |
|
| 36 |
+ if (empty($url) || empty($xpaths)) {
|
|
| 37 |
+ throw new InvalidArgumentException("All argument are required");
|
|
| 38 |
+ } |
|
| 39 |
+ |
|
| 40 |
+ $this->name = $name; |
|
| 41 |
+ $this->url = $url; |
|
| 42 |
+ |
|
| 43 |
+ if (is_string($xpaths)) {
|
|
| 44 |
+ $this->xpaths[] = $xpaths; |
|
| 45 |
+ } else if (is_array($xpaths)) {
|
|
| 46 |
+ $this->xpaths = $xpaths; |
|
| 47 |
+ } else {
|
|
| 48 |
+ throw new InvalidArgumentException("Xpaths must be string or array");
|
|
| 49 |
+ } |
|
| 50 |
+ } |
|
| 51 |
+ |
|
| 52 |
+} |
|
| 53 |
+ |
|
| 54 |
+ |
|
| 55 |
+class Parser {
|
|
| 56 |
+ |
|
| 57 |
+ private $restaurants = array(); |
|
| 58 |
+ |
|
| 59 |
+ function __construct($restaurants) |
|
| 60 |
+ {
|
|
| 61 |
+ |
|
| 62 |
+ if (empty($restaurants) || !is_array($restaurants)) {
|
|
| 63 |
+ throw new InvalidArgumentException("Array argument required");
|
|
| 64 |
+ } |
|
| 65 |
+ |
|
| 66 |
+ $this->restaurants = $restaurants; |
|
| 67 |
+ } |
|
| 68 |
+ |
|
| 69 |
+ public function parse() |
|
| 70 |
+ {
|
|
| 71 |
+ foreach ($this->restaurants as $restaurant) {
|
|
| 72 |
+ |
|
| 73 |
+ $source = file_get_contents($restaurant->getUrl()); |
|
| 74 |
+ |
|
| 75 |
+ if ($source === false) {
|
|
| 76 |
+ throw new RuntimeException("Can't read source address: " . $restaurant->getUrl());
|
|
| 77 |
+ } |
|
| 78 |
+ |
|
| 79 |
+ $dom = new DOMDocument(); |
|
| 80 |
+ @$dom->loadHTML($source); |
|
| 81 |
+ $xpath_obj = new DOMXPath($dom); |
|
| 82 |
+ |
|
| 83 |
+ echo("Restaurant: <a href='{$restaurant->getUrl()}'>" . $restaurant->getName() . "</a><br>\n");
|
|
| 84 |
+ |
|
| 85 |
+ foreach ($restaurant->getXpaths() as $name => $xpath) {
|
|
| 86 |
+ |
|
| 87 |
+ $found = $xpath_obj->query($xpath); |
|
| 88 |
+ |
|
| 89 |
+ if ($found === false || $found->length === 0) {
|
|
| 90 |
+ continue; |
|
| 91 |
+ } |
|
| 92 |
+ |
|
| 93 |
+ echo "<b>$name</b>"; |
|
| 94 |
+ $exists = false; |
|
| 95 |
+ foreach ($found->item(0)->childNodes as $elem) {
|
|
| 96 |
+ |
|
| 97 |
+ $meal = trim($elem->nodeValue); |
|
| 98 |
+ if (!empty($meal)) {
|
|
| 99 |
+ echo("\t<br>$meal\n");
|
|
| 100 |
+ $exists = true; |
|
| 101 |
+ } |
|
| 102 |
+ } |
|
| 103 |
+ if (!$exists) {
|
|
| 104 |
+ echo "<br>{}";
|
|
| 105 |
+ } |
|
| 106 |
+ echo "<br>"; |
|
| 107 |
+ } |
|
| 108 |
+ echo "<br>"; |
|
| 109 |
+ } |
|
| 110 |
+ } |
|
| 111 |
+ |
|
| 112 |
+ |
|
| 113 |
+} |
|
| 114 |
+ |
|
| 115 |
+$twenties = new Restaurant("Twenties", "http://www.twenties.sk/", array("Polievky" => '//*[@id="article"]/div[2]/p[1]'));
|
|
| 116 |
+$pulitzer = new Restaurant("Pulitzer", "http://www.pulitzer.sk",
|
|
| 117 |
+ array("Polievky" => '//*[@id="soups"]',
|
|
| 118 |
+ "Hlavne jedla" => '//*[@id="meals"]')); |
|
| 119 |
+$slovak = new Restaurant("Slovak pub", "http://www.arcaderestaurant.sk/articles/public_menu/show-modules/id/24",
|
|
| 120 |
+ array("Salat" => '//*[@id="table2"]/tbody/tr[10]',
|
|
| 121 |
+ "Polievka" => '//*[@id="table2"]/tbody/tr[2]')); |
|
| 122 |
+ |
|
| 123 |
+ |
|
| 124 |
+$parser = new Parser(array($twenties, $pulitzer, $slovak)); |
|
| 125 |
+$parser->parse(); |
|
| 126 |
+ |
|
| 127 |
+ |
|
| 128 |
+ |
|
| 129 |
+/** |
|
| 4 | 130 |
* Lebo bez masa nie je den dnom. Maso je sucast jedla, preto chodim jest do |
| 5 | 131 |
* blizkych malych restauracii. |
| 6 | 132 |
*/ |
| 7 | 133 |
|
| 134 |
+/* |
|
| 135 |
+ |
|
| 8 | 136 |
$sites = [ |
| 9 | 137 |
'Pulitzer' => 'http://pulitzer.sk/', |
| 10 | 138 |
'Twenties' => 'http://www.twenties.sk/' |
| ... | ... |
@@ -44,7 +172,7 @@ foreach ($sites as $restaurant => $site) |
| 44 | 44 |
} |
| 45 | 45 |
|
| 46 | 46 |
$found_meal = FALSE; |
| 47 |
- foreach ($oh_my_god_here_are_saved_meals->item(0)->childNodes as $elem) |
|
| 47 |
+ foreach ($oh_my_god_here_are_saved_meals->item(0)->childNodes as $elem) |
|
| 48 | 48 |
{
|
| 49 | 49 |
$meal_meal_meal = trim($elem->nodeValue); |
| 50 | 50 |
if (! empty($meal_meal_meal)) |
| ... | ... |
@@ -58,7 +186,7 @@ foreach ($sites as $restaurant => $site) |
| 58 | 58 |
{
|
| 59 | 59 |
append_result("\tNevaria!!! Daj mi niekto gulomet, nech ich zabijem!\n");
|
| 60 | 60 |
} |
| 61 |
- else |
|
| 61 |
+ else |
|
| 62 | 62 |
{
|
| 63 | 63 |
append_result("\n");
|
| 64 | 64 |
} |
| ... | ... |
@@ -72,3 +200,5 @@ function append_result($text) |
| 72 | 72 |
global $output_file; |
| 73 | 73 |
file_put_contents($output_file, $text, FILE_APPEND | FILE_TEXT); |
| 74 | 74 |
} |
| 75 |
+ |
|
| 76 |
+*/ |