Refactor, and make 2022 scrapable

sportclimbing · May 10, 2023 · a2e727a · a2e727a
1 parent f2b5f02
commit a2e727a
Show file tree

Hide file tree

Showing 13 changed files with 1,018 additions and 31 deletions.
diff --git a/app/config/services.yml b/app/config/services.yml
@@ -1,6 +1,7 @@
 parameters:
   ifsc_youtube_channel_id: 'UC2MGuhIaOP6YLpUx106kTQw'
   calendar_product_identifier: '-//ifsc/ical//2.0/EN'
+  site_event_url: 'https://ifsc.stream/#/season/%%d/event/%%d'
 
 services:
   nicoSWD\IfscCalendar\Application\Command\:
@@ -84,6 +85,7 @@ services:
     class: nicoSWD\IfscCalendar\Domain\Event\IFSCEventsScraper
     arguments:
       - '@nicoSWD\IfscCalendar\Infrastructure\HttpClient\HttpGuzzleClient'
+      - '@nicoSWD\IfscCalendar\Domain\Event\IFSCEventFactory'
       - '@nicoSWD\IfscCalendar\Domain\Event\Helpers\DOMHelper'
       - '@nicoSWD\IfscCalendar\Domain\Event\Helpers\Normalizer'
 
@@ -93,6 +95,11 @@ services:
   nicoSWD\IfscCalendar\Domain\Event\Helpers\Normalizer:
     class: nicoSWD\IfscCalendar\Domain\Event\Helpers\Normalizer
 
+  nicoSWD\IfscCalendar\Domain\Event\IFSCEventFactory:
+    class: nicoSWD\IfscCalendar\Domain\Event\IFSCEventFactory
+    arguments:
+      - '%site_event_url%'
+
   # HTTP Client
   nicoSWD\IfscCalendar\Infrastructure\HttpClient\HttpGuzzleClient:
     class: nicoSWD\IfscCalendar\Infrastructure\HttpClient\HttpGuzzleClient

diff --git a/...main/Event/IFSCEventsScraperException.php → ...Exceptions/IFSCEventsScraperException.php b/...main/Event/IFSCEventsScraperException.php → ...Exceptions/IFSCEventsScraperException.php
@@ -5,7 +5,7 @@
  * @link     https://github.com/nicoSWD
  * @author   Nicolas Oelgart <[email protected]>
  */
-namespace nicoSWD\IfscCalendar\Domain\Event;
+namespace nicoSWD\IfscCalendar\Domain\Event\Exceptions;
 
 use Exception;
 

diff --git a/src/Domain/Event/Exceptions/InvalidURLException.php b/src/Domain/Event/Exceptions/InvalidURLException.php
@@ -0,0 +1,14 @@
+<?php declare(strict_types=1);
+
+/**
+ * @license  http://opensource.org/licenses/mit-license.php MIT
+ * @link     https://github.com/nicoSWD
+ * @author   Nicolas Oelgart <[email protected]>
+ */
+namespace nicoSWD\IfscCalendar\Domain\Event\Exceptions;
+
+use Exception;
+
+final class InvalidURLException extends Exception
+{
+}
diff --git a/src/Domain/Event/Helpers/Normalizer.php b/src/Domain/Event/Helpers/Normalizer.php
@@ -11,7 +11,7 @@
 {
     public function cupName(string $league): string
     {
-        return ucwords(strtolower($league));
+        return ucwords(strtolower(trim($league)));
     }
 
     public function normalizeTime(string $time): string
@@ -20,6 +20,9 @@ public function normalizeTime(string $time): string
             // We don't know the exact time yet. We'll set it to 8:00 for now
             // as it will automatically update once IFSC sets it
             $time = '8:00';
+        } else {
+            // Convert 12-hour format to 24-hour
+            $time = date('H:i', strtotime($time));
         }
 
         return $time;
@@ -32,6 +35,15 @@ public function nonEmptyLines(string $matches): array
 
     public function removeNonAsciiCharacters(string $text): string
     {
-        return preg_replace('~[^\w\s\'\r\n:,-./?=]+~', ' ', $text);
+        // This fixes a parsing issue for season 2022
+        // This is fun
+        $text = preg_replace('~\n\s{10,}~', ' ', $text);
+
+        return preg_replace('~[^\w\s\r\n\':,-./?=&]+~', ' ', $text);
+    }
+
+    public function firstUrl(string $urls): string
+    {
+        return preg_split('~\s+~', $urls, flags: PREG_SPLIT_NO_EMPTY)[0] ?? '';
     }
 }
diff --git a/src/Domain/Event/IFSCEvent.php b/src/Domain/Event/IFSCEvent.php
@@ -11,20 +11,16 @@
 
 final readonly class IFSCEvent
 {
-    private const EVENT_INFO_URL = 'https://ifsc.stream/#/event/%d';
-
-    public string $siteUrl;
-
     public function __construct(
         public string $name,
         public int $id,
         public string $description,
         public string $streamUrl,
+        public string $siteUrl,
         public string $poster,
         public DateTimeImmutable $startTime,
         public DateTimeImmutable $endTime,
     ) {
-        $this->siteUrl = sprintf(self::EVENT_INFO_URL, $id);
     }
 
     public function updateStreamUrl(string $streamUrl): self
@@ -34,6 +30,7 @@ public function updateStreamUrl(string $streamUrl): self
             id: $this->id,
             description: $this->description,
             streamUrl: $streamUrl,
+            siteUrl: $this->siteUrl,
             poster: $this->poster,
             startTime: $this->startTime,
             endTime: $this->endTime,

diff --git a/src/Domain/Event/IFSCEventFactory.php b/src/Domain/Event/IFSCEventFactory.php
@@ -0,0 +1,39 @@
+<?php declare(strict_types=1);
+
+/**
+ * @license  http://opensource.org/licenses/mit-license.php MIT
+ * @link     https://github.com/nicoSWD
+ * @author   Nicolas Oelgart <[email protected]>
+ */
+namespace nicoSWD\IfscCalendar\Domain\Event;
+
+use DateTimeImmutable;
+
+final readonly class IFSCEventFactory
+{
+    public function __construct(
+        private string $siteUrl,
+    ) {
+    }
+
+    public function create(
+        string $name,
+        int $id,
+        string $description,
+        string $streamUrl,
+        string $poster,
+        DateTimeImmutable $startTime,
+        DateTimeImmutable $endTime,
+    ): IFSCEvent {
+        return new IFSCEvent(
+            name: $name,
+            id: $id,
+            description: $description,
+            streamUrl: $streamUrl,
+            siteUrl: sprintf($this->siteUrl, $startTime->format('Y'), $id),
+            poster: $poster,
+            startTime: $startTime,
+            endTime: $endTime,
+        );
+    }
+}
diff --git a/src/Domain/Event/IFSCEventsScraper.php b/src/Domain/Event/IFSCEventsScraper.php
@@ -12,6 +12,8 @@
 use DateTimeZone;
 use DOMElement;
 use DOMXPath;
+use nicoSWD\IfscCalendar\Domain\Event\Exceptions\IFSCEventsScraperException;
+use nicoSWD\IfscCalendar\Domain\Event\Exceptions\InvalidURLException;
 use nicoSWD\IfscCalendar\Domain\Event\Helpers\DOMHelper;
 use nicoSWD\IfscCalendar\Domain\Event\Helpers\Normalizer;
 use nicoSWD\IfscCalendar\Domain\HttpClient\HttpClientInterface;
@@ -22,12 +24,17 @@
 
     public function __construct(
         private HttpClientInterface $client,
+        private IFSCEventFactory $eventFactory,
         private DOMHelper $domHelper,
         private Normalizer $normalizer,
     ) {
     }
 
-    /** @throws IFSCEventsScraperException */
+    /**
+     * @throws IFSCEventsScraperException
+     * @throws InvalidURLException
+     * @return IFSCEvent[]
+     */
     public function fetchEventsForLeague(int $season, int $eventId, string $timezone, string $eventName): array
     {
         $xpath = $this->getXPathForEventsWithId($eventId);
@@ -47,9 +54,9 @@ public function fetchEventsForLeague(int $season, int $eventId, string $timezone
                     $schedules[] = IFSCSchedule::create(
                         day: (int) $matches['day'][$key],
                         month: Month::fromName($matches['month'][$key]),
-                        time: $this->normalizer->normalizeTime($eventTime),
+                        time: $eventTime,
                         season: $season,
-                        cupName: $this->normalizer->cupName($cupName),
+                        cupName: $cupName,
                         streamUrl: $streamUrl,
                     );
                 }
@@ -63,7 +70,7 @@ public function fetchEventsForLeague(int $season, int $eventId, string $timezone
             $startDateTime = $this->getStartDateTime($schedule, $timezone);
             $endDateTime = $this->getEndDateTime($startDateTime);
 
-            $events[] = new IFSCEvent(
+            $events[] = $this->eventFactory->create(
                 name: $schedule->cupName,
                 id: $eventId,
                 description: $eventName,
@@ -116,21 +123,24 @@ private function buildDateRegex(): string
             ~xsi";
     }
 
+    /** @throws IFSCEventsScraperException */
     private function parseEventDetails(string $line): array
     {
-        $parts = preg_split('~(\s{2,}|\s\W+\s)~', $line, flags: PREG_SPLIT_NO_EMPTY);
-
-        if (count($parts) >= 3) {
-            [$time, $eventName, $streamUrl] = $parts;
-        } else {
-            [$time, $eventName] = $parts;
+        $regex = '~^
+            (?<time>(\d{1,2}:\d{1,2}(?:\s+(?:AM|PM))?|TBC|TBD))\s+
+            (?<name>[\w\'\-&\s]+)
+            (?<url>\s*(http[^\s]+\s*)*)
+            $~x';
+
+        if (!preg_match($regex, trim($line), $match)) {
+            throw new IFSCEventsScraperException("No event found in line: {$line}");
         }
 
-        return [
-            $time,
-            $eventName,
-            $streamUrl ?? '',
-        ];
+        $startTime = $this->normalizer->normalizeTime($match['time']);
+        $eventName = $this->normalizer->cupName($match['name']);
+        $streamUrl = $this->normalizer->firstUrl($match['url'] ?? '');
+
+        return [$startTime, $eventName, $streamUrl];
     }
 
     private function buildLeagueUri(int $id): string

diff --git a/src/Domain/Event/IFSCSchedule.php b/src/Domain/Event/IFSCSchedule.php
@@ -7,8 +7,11 @@
  */
 namespace nicoSWD\IfscCalendar\Domain\Event;
 
+use nicoSWD\IfscCalendar\Domain\Event\Exceptions\InvalidURLException;
+
 final readonly class IFSCSchedule
 {
+    /** @throws InvalidURLException */
     private function __construct(
         public int $day,
         public Month $month,
@@ -17,8 +20,10 @@ private function __construct(
         public string $cupName,
         public string $streamUrl,
     ) {
+        $this->assertValidUrl($streamUrl);
     }
 
+    /** @throws InvalidURLException */
     public static function create(
         int $day,
         Month $month,
@@ -36,4 +41,12 @@ public static function create(
             $streamUrl,
         );
     }
+
+    /** @throws InvalidURLException */
+    public function assertValidUrl(string $streamUrl): void
+    {
+        if (!empty($streamUrl) && !filter_var($streamUrl, FILTER_VALIDATE_URL)) {
+            throw new InvalidURLException("Invalid URL: {$streamUrl}");
+        }
+    }
 }
diff --git a/src/Domain/Event/Month.php b/src/Domain/Event/Month.php
@@ -27,6 +27,8 @@ enum Month: int
     /** @throws InvalidArgumentException */
     public static function fromName(string $name): self
     {
+        $name = strtoupper($name);
+
         foreach (self::cases() as $case) {
             if ($case->name === $name) {
                 return $case;

diff --git a/src/Infrastructure/Events/IFSCGuzzleEventsFetcher.php b/src/Infrastructure/Events/IFSCGuzzleEventsFetcher.php
@@ -8,9 +8,11 @@
 namespace nicoSWD\IfscCalendar\Infrastructure\Events;
 
 use GuzzleHttp\Client;
+use GuzzleHttp\Exception\GuzzleException;
+use JsonException;
+use nicoSWD\IfscCalendar\Domain\Event\Exceptions\IFSCEventsScraperException;
 use nicoSWD\IfscCalendar\Domain\Event\IFSCEventFetcherInterface;
 use nicoSWD\IfscCalendar\Domain\Event\IFSCEventsScraper;
-use nicoSWD\IfscCalendar\Domain\Event\IFSCEventsScraperException;
 use nicoSWD\IfscCalendar\Domain\League\IFSCLeague;
 
 final readonly class IFSCGuzzleEventsFetcher implements IFSCEventFetcherInterface
@@ -29,13 +31,7 @@ public function __construct(
      */
     public function fetchEventsForLeague(int $season, IFSCLeague $league): array
     {
-        $response = $this->client->get($this->buildLeagueUri($league->id))->getBody()->getContents();
-        $response = @json_decode($response);
-
-        if (json_last_error()) {
-            throw new \Exception(json_last_error_msg());
-        }
-
+        $response = $this->fetchHtmlForLeague($league);
         $events = [];
 
         foreach ($response->events as $event) {
@@ -62,4 +58,18 @@ public function buildLeagueUri(int $leagueId): string
     {
         return sprintf(self::IFSC_LEAGUE_API_ENDPOINT, $leagueId);
     }
+
+    /** @throws IFSCEventsScraperException */
+    public function fetchHtmlForLeague(IFSCLeague $league): object
+    {
+        try {
+            $response = $this->client->get($this->buildLeagueUri($league->id))->getBody()->getContents();
+
+            return @json_decode($response, flags: JSON_THROW_ON_ERROR);
+        } catch (GuzzleException $e) {
+            throw new IFSCEventsScraperException("Unable to retrieve HTML: {$e->getMessage()}");
+        } catch (JsonException $e) {
+            throw new IFSCEventsScraperException("Unable to parse JSON: {$e->getMessage()}");
+        }
+    }
 }