/**
* Parses replacements from html.
* If them cannot be find or they are in incorrect format returns null.
* @param SimpleHtmlDom $dom
* @return Replacements[]|null
*/
public function getReplacements($dom)
{
$itemPage = $dom->root()->find('div[class=item-page]', 0);
if ($itemPage === null) {
return null;
}
$h4 = $itemPage->find('h4', 0);
$table = $itemPage->find('table', 0);
if ($h4 === null || $table === null) {
return null;
}
//parse date
$date = ReplacementsParser::parseDate(trim($h4->text()));
if ($date === null) {
$this->errors[] = "incorrect date format: " . $h4->text();
return null;
}
//parse content
$rows = $table->find('tr');
/** @var Replacements[] $replacements */
$replacements = [];
/** @var Replacements|null $current */
$current = null;
foreach ($rows as $i => $row) {
$cells = $row->find('th, td');
if (count($cells) === 1) {
//class name, e.g. | 2a |
$current = new Replacements();
$current->date = $date;
$current->class = $cells[0]->text();
$current->value = [];
$replacements[] = $current;
} else {
if (count($cells) === 2) {
//replacement entry, e.g. | 1 | j. niemiecki, mgr T. Wajdzik |
if ($current === null) {
$this->errors[] = "row: {$i}, no class name occurred before replacement text";
continue;
//skip this row
}
$hourText = $cells[0]->text();
$text = $cells[1]->text();
if (!is_numeric($hourText)) {
$this->errors[] = "row: {$i}, invalid hour no.: '{$hourText}'";
continue;
//skip this row
}
$hour = intval($hourText);
$current->value[$hour] = $text;
}
}
}
return $replacements;
}
/**
* @param SimpleHtmlDom $dom
* @return null|string , a text of a next paragraph after phrase "szczęśliwy numerek"
* or null if it wasn't found
*/
private static function findText($dom)
{
$ps = $dom->root()->find("div.column_right div.custom p");
//find paragraph containing phrase "szczęśliwy numerek"
$pTitleIndex = -1;
foreach ($ps as $i => $p) {
if (strcasecmp($p->text(), "SZCZĘŚLIWY NUMEREK") === 0) {
$pTitleIndex = $i;
}
}
//get context of the next paragraph
if ($pTitleIndex !== -1 && isset($ps[$pTitleIndex + 1])) {
$p = $ps[$pTitleIndex + 1];
return $p->text();
}
return null;
}
public function testIncorrect()
{
$dom = SimpleHtmlDom::fromUrl(__DIR__ . '/incorrect.html');
$a = new LuckyNumberParser();
$result = $a->getLuckyNumber($dom);
$this->assertEquals(null, $result);
$this->assertGreaterThan(0, count($a->getErrors()));
}
public function run()
{
Database::init(Config::getInstance()->getDatabaseConfig());
$url = self::URL;
$dom = SimpleHtmlDom::fromUrl($url);
if ($dom === false) {
echo "cannot get {$url}\n";
return;
}
$this->updateLuckyNumbers($dom);
$this->updateReplacements($dom);
echo "done\n";
}
