Excluded data parsing to a data parser class that handles document and returns info class or array of those.

This commit is contained in:
kplaczek
2022-05-25 19:58:15 +02:00
parent 0e3c8852ce
commit 7d70ad40cb
5 changed files with 85 additions and 42 deletions

View File

@@ -4,6 +4,7 @@ namespace Techtube\Bookinfo\Api;
abstract class AbstractBookInfo abstract class AbstractBookInfo
{ {
public string $url;
public string $author; public string $author;
public string $isbn; public string $isbn;
public string $description; public string $description;
@@ -11,10 +12,8 @@ abstract class AbstractBookInfo
public string $category; public string $category;
public string $cover_url; public string $cover_url;
public int $pages; public int $pages;
public string $series; public string $cycle;
public int $volume; public int $volume;
public string $language; public string $language;
public string $datePublished; public string $datePublished;
abstract public function parse();
} }

View File

@@ -7,15 +7,28 @@ use Techtube\Bookinfo\Api\AbstractBookInfo;
class BookFinder class BookFinder
{ {
private static $searchUrl = 'https://lubimyczytac.pl/szukaj/ksiazki?phrase=';
public function search(string $searchText): \Exception private DataParser $parser;
public function __construct()
{ {
return new \Exception('need to implement this'); $this->parser = new DataParser();
}
public function search(string $phrase): array
{
return $this->parser->searchPage(new Document($this->getSearchUrl($phrase), true));
} }
public function byUrl($url): AbstractBookInfo public function byUrl($url): AbstractBookInfo
{ {
return (new BookInfo(new Document($url, true)))->parse(); return $this->parser->singlePage(new Document($url, true));
}
public function getSearchUrl(string $phrase): string
{
return self::$searchUrl . $phrase;
} }
} }

View File

@@ -1,36 +0,0 @@
<?php
namespace Techtube\Bookinfo;
use DiDom\Document;
use Techtube\Bookinfo\Api\AbstractBookInfo;
class BookInfo extends AbstractBookInfo
{
private Document $document;
public function __construct($document)
{
$this->document = $document;
}
public function parse(): self
{
$jsonInfo = json_decode($this->document->first('script[type="application/ld+json"]')->text());
$this->author = $this->document->first('meta[property="books:author"]')->getAttribute('content');
$this->isbn = $this->document->first('meta[property="books:isbn"]')->getAttribute('content');
$this->description = $this->document->first('meta[property="og:description"]')->getAttribute('content');
$this->title = trim($this->document->first('h1.book__title')->text());
$this->category = trim($this->document->first('.book__category')->text());
$this->cover_url = $this->document->first('meta[property="og:image"]')->getAttribute('content');
$this->pages = (int)$this->document->first('span.book__pages')?->text();
if (preg_match('#(.*) \(tom (\d*)\)#ism', trim($this->document->first('a[href*="/cykl/"]')?->text()), $series)) {
$this->series = $series[1];
$this->volume = $series[2];
}
$this->language = $jsonInfo->inLanguage ?? null;
$this->datePublished = $jsonInfo->datePublished ?? null;
return $this;
}
}

57
src/DataParser.php Normal file
View File

@@ -0,0 +1,57 @@
<?php
namespace Techtube\Bookinfo;
use DiDom\Document;
use DiDom\Exceptions\InvalidSelectorException;
use Techtube\Bookinfo\Api\AbstractBookInfo;
class DataParser
{
/**
* @param Document $document
* @return AbstractBookInfo
* @throws InvalidSelectorException
*/
public function singlePage(Document $document): AbstractBookInfo
{
$info = new Info();
$jsonInfo = json_decode($document->first('script[type="application/ld+json"]')->text());
$info->author = $document->first('meta[property="books:author"]')->getAttribute('content');
$info->isbn = $document->first('meta[property="books:isbn"]')->getAttribute('content');
$info->description = $document->first('meta[property="og:description"]')->getAttribute('content');
$info->title = trim($document->first('h1.book__title')->text());
$info->category = trim($document->first('.book__category')->text());
$info->cover_url = $document->first('meta[property="og:image"]')->getAttribute('content');
$info->pages = (int)$document->first('span.book__pages')?->text();
if (preg_match('#(.*) \(tom (\d*)\)#ism', trim($document->first('a[href*="/cykl/"]')?->text()), $series)) {
$info->cycle = $series[1];
$info->volume = $series[2];
}
$info->language = $jsonInfo->inLanguage ?? null;
$info->datePublished = $jsonInfo->datePublished ?? null;
return $info;
}
/**
* @param Document $document
* @return AbstractBookInfo[]
* @throws InvalidSelectorException
*/
public function searchPage(Document $document): array
{
$books = $document->find('#search .authorAllBooks__single');
$booksInfo = [];
foreach ($books as $book) {
$bookInfo = new Info();
$bookInfo->title = trim($book->first('.authorAllBooks__singleTextTitle')->text());
$bookInfo->author = trim($book->first('.authorAllBooks__singleTextAuthor')->text());
$bookInfo->url = $book->first('button[data-book-url]')->getAttribute('data-book-url');
$bookInfo->cover_url = $book->first('.img-fluid')->getAttribute('src');
$booksInfo[] = $bookInfo;
}
return $booksInfo;
}
}

10
src/Info.php Normal file
View File

@@ -0,0 +1,10 @@
<?php
namespace Techtube\Bookinfo;
use Techtube\Bookinfo\Api\AbstractBookInfo;
class Info extends AbstractBookInfo
{
}