Adopt WebDriverAbstract as a solution for active (JavaScript) websites (

#3971) * first working version --------- Co-authored-by: Dag <[email protected]>
RSS-Bridge · Feb 10, 2024 · 8e8028b · 8e8028b
1 parent ff7840d
commit 8e8028b
Show file tree

Hide file tree

Showing 7 changed files with 473 additions and 1 deletion.
diff --git a/bridges/GULPProjekteBridge.php b/bridges/GULPProjekteBridge.php
@@ -0,0 +1,164 @@
+<?php
+
+use Facebook\WebDriver\Exception\NoSuchElementException;
+use Facebook\WebDriver\Remote\RemoteWebElement;
+use Facebook\WebDriver\WebDriverBy;
+use Facebook\WebDriver\WebDriverExpectedCondition;
+
+class GULPProjekteBridge extends WebDriverAbstract
+{
+    const NAME = 'GULP Projekte';
+    const URI = 'https://www.gulp.de/gulp2/g/projekte';
+    const DESCRIPTION = 'Projektsuche';
+    const MAINTAINER = 'hleskien';
+
+    const MAXITEMS = 60;
+
+    /**
+     * Adds accept language german to the Chrome Options.
+     *
+     * @return Facebook\WebDriver\Chrome\ChromeOptions
+     */
+    protected function getBrowserOptions()
+    {
+        $chromeOptions = parent::getBrowserOptions();
+        $chromeOptions->addArguments(['--accept-lang=de']);
+        return $chromeOptions;
+    }
+
+    /**
+     * @throws Facebook\WebDriver\Exception\NoSuchElementException
+     * @throws Facebook\WebDriver\Exception\TimeoutException
+     */
+    protected function clickAwayCookieBanner()
+    {
+        $this->getDriver()->wait()->until(WebDriverExpectedCondition::visibilityOfElementLocated(WebDriverBy::id('onetrust-reject-all-handler')));
+        $buttonRejectCookies = $this->getDriver()->findElement(WebDriverBy::id('onetrust-reject-all-handler'));
+        $buttonRejectCookies->click();
+        $this->getDriver()->wait()->until(WebDriverExpectedCondition::invisibilityOfElementLocated(WebDriverBy::id('onetrust-reject-all-handler')));
+    }
+
+    /**
+     * @throws Facebook\WebDriver\Exception\NoSuchElementException
+     * @throws Facebook\WebDriver\Exception\TimeoutException
+     */
+    protected function clickNextPage()
+    {
+        $nextPage = $this->getDriver()->findElement(WebDriverBy::xpath('//app-linkable-paginator//li[@id="next-page"]/a'));
+        $href = $nextPage->getAttribute('href');
+        $nextPage->click();
+        $this->getDriver()->wait()->until(WebDriverExpectedCondition::not(
+            WebDriverExpectedCondition::presenceOfElementLocated(
+                WebDriverBy::xpath('//app-linkable-paginator//li[@id="next-page"]/a[@href="' . $href . '"]')
+            )
+        ));
+    }
+
+    /**
+     * Returns the uri of the 'Projektanbieter' logo or false if there is
+     * no logo present in the item.
+     *
+     * @return string | false
+     */
+    protected function getLogo(RemoteWebElement $item)
+    {
+        try {
+            $logo = $item->findElement(WebDriverBy::tagName('img'))->getAttribute('src');
+            if (str_starts_with($logo, 'http')) {
+                // different domain
+                return $logo;
+            } else {
+                // relative path
+                $remove = substr(self::URI, strrpos(self::URI, '/') + 1);
+                return substr(self::URI, 0, -strlen($remove)) . $logo;
+            }
+        } catch (NoSuchElementException $e) {
+            return false;
+        }
+    }
+
+    /**
+     * Converts a string like "vor einigen Minuten" into a reasonable timestamp.
+     * Long and complicated, but we don't want to be more specific than
+     * the information we have available.
+     *
+     * @throws Exception If the DateInterval can't be parsed.
+     */
+    protected function getTimestamp(string $timeAgo): int
+    {
+        $dateTime = new DateTime();
+        $dateArray = explode(' ', $dateTime->format('Y m d H i s'));
+        $quantityStr = explode(' ', $timeAgo)[1];
+        // convert possible word into a number
+        if (in_array($quantityStr, ['einem', 'einer', 'einigen'])) {
+            $quantity = 1;
+        } else {
+            $quantity = intval($quantityStr);
+        }
+        // subtract time ago + inferior units for lower precision
+        if (str_contains($timeAgo, 'Sekunde')) {
+            $interval = new DateInterval('PT' . $quantity . 'S');
+        } elseif (str_contains($timeAgo, 'Minute')) {
+            $interval = new DateInterval('PT' . $quantity . 'M' . $dateArray[5] . 'S');
+        } elseif (str_contains($timeAgo, 'Stunde')) {
+            $interval = new DateInterval('PT' . $quantity . 'H' . $dateArray[4] . 'M' . $dateArray[5] . 'S');
+        } elseif (str_contains($timeAgo, 'Tag')) {
+            $interval = new DateInterval('P' . $quantity . 'DT' . $dateArray[3] . 'H' . $dateArray[4] . 'M' . $dateArray[5] . 'S');
+        } else {
+            throw new UnexpectedValueException($timeAgo);
+        }
+        $dateTime = $dateTime->sub($interval);
+        return $dateTime->getTimestamp();
+    }
+
+    /**
+     * The main loop which clicks through search result pages and puts
+     * the content into the $items array.
+     *
+     * @throws Facebook\WebDriver\Exception\NoSuchElementException
+     * @throws Facebook\WebDriver\Exception\TimeoutException
+     */
+    public function collectData()
+    {
+        parent::collectData();
+
+        try {
+            $this->clickAwayCookieBanner();
+            $this->setIcon($this->getDriver()->findElement(WebDriverBy::xpath('//link[@rel="shortcut icon"]'))->getAttribute('href'));
+
+            while (true) {
+                $items = $this->getDriver()->findElements(WebDriverBy::tagName('app-project-view'));
+                foreach ($items as $item) {
+                    $feedItem = new FeedItem();
+
+                    $heading = $item->findElement(WebDriverBy::xpath('.//app-heading-tag/h1/a'));
+                    $feedItem->setTitle($heading->getText());
+                    $feedItem->setURI('https://www.gulp.de' . $heading->getAttribute('href'));
+                    $info = $item->findElement(WebDriverBy::tagName('app-icon-info-list'));
+                    if ($logo = $this->getLogo($item)) {
+                        $feedItem->setEnclosures([$logo]);
+                    }
+                    if (str_contains($info->getText(), 'Projektanbieter:')) {
+                        $feedItem->setAuthor($info->findElement(WebDriverBy::xpath('.//li/span[2]/span'))->getText());
+                    } else {
+                        // mostly "Direkt vom Auftraggeber" or "GULP Agentur"
+                        $feedItem->setAuthor($item->findElement(WebDriverBy::tagName('b'))->getText());
+                    }
+                    $feedItem->setContent($item->findElement(WebDriverBy::xpath('.//p[@class="description"]'))->getText());
+                    $timeAgo = $item->findElement(WebDriverBy::xpath('.//small[contains(@class, "time-ago")]'))->getText();
+                    $feedItem->setTimestamp($this->getTimestamp($timeAgo));
+
+                    $this->items[] = $feedItem;
+                }
+
+                if (count($this->items) < self::MAXITEMS) {
+                    $this->clickNextPage();
+                } else {
+                    break;
+                }
+            }
+        } finally {
+            $this->cleanUp();
+        }
+    }
+}
diff --git a/bridges/ScalableCapitalBlogBridge.php b/bridges/ScalableCapitalBlogBridge.php
@@ -0,0 +1,73 @@
+<?php
+
+use Facebook\WebDriver\WebDriverBy;
+use Facebook\WebDriver\WebDriverExpectedCondition;
+
+class ScalableCapitalBlogBridge extends WebDriverAbstract
+{
+    const NAME = 'Scalable Capital Blog';
+    const URI = 'https://de.scalable.capital/blog';
+    const DESCRIPTION = 'Alle Artikel';
+    const MAINTAINER = 'hleskien';
+
+    /**
+     * Adds accept language german to the Chrome Options.
+     *
+     * @return Facebook\WebDriver\Chrome\ChromeOptions
+     */
+    protected function getBrowserOptions()
+    {
+        $chromeOptions = parent::getBrowserOptions();
+        $chromeOptions->addArguments(['--accept-lang=de']);
+        return $chromeOptions;
+    }
+
+    /**
+     * Puts the content of the first page into the $items array.
+     *
+     * @throws Facebook\WebDriver\Exception\NoSuchElementException
+     * @throws Facebook\WebDriver\Exception\TimeoutException
+     */
+    public function collectData()
+    {
+        parent::collectData();
+
+        try {
+            // wait until last item is loaded
+            $this->getDriver()->wait()->until(WebDriverExpectedCondition::visibilityOfElementLocated(
+                WebDriverBy::xpath('//div[contains(@class, "articles")]//div[@class="items"]//div[contains(@class, "item")][15]')
+            ));
+            $this->setIcon($this->getDriver()->findElement(WebDriverBy::xpath('//link[@rel="shortcut icon"]'))->getAttribute('href'));
+
+            $items = $this->getDriver()->findElements(WebDriverBy::xpath('//div[contains(@class, "articles")]//div[@class="items"]//div[contains(@class, "item")]'));
+            foreach ($items as $item) {
+                $feedItem = new FeedItem();
+
+                $feedItem->setEnclosures(['https://de.scalable.capital' . $item->findElement(WebDriverBy::tagName('img'))->getAttribute('src')]);
+                $heading = $item->findElement(WebDriverBy::tagName('a'));
+                $feedItem->setTitle($heading->getText());
+                $feedItem->setURI('https://de.scalable.capital' . $heading->getAttribute('href'));
+                $feedItem->setContent($item->findElement(WebDriverBy::xpath('.//div[@class="summary"]'))->getText());
+                $date = $item->findElement(WebDriverBy::xpath('.//div[@class="published-date"]'))->getText();
+                $feedItem->setTimestamp($this->formatItemTimestamp($date));
+                $feedItem->setAuthor($item->findElement(WebDriverBy::xpath('.//div[@class="author"]'))->getText());
+
+                $this->items[] = $feedItem;
+            }
+        } finally {
+            $this->cleanUp();
+        }
+    }
+
+    /**
+     * Converts the given date (dd.mm.yyyy) into a timestamp.
+     *
+     * @param $value string
+     * @return int
+     */
+    protected function formatItemTimestamp($value)
+    {
+        $formatter = new IntlDateFormatter('de', IntlDateFormatter::LONG, IntlDateFormatter::NONE);
+        return $formatter->parse($value);
+    }
+}
diff --git a/config.default.ini.php b/config.default.ini.php
@@ -99,6 +99,16 @@
 ; false = disabled (default)
 by_bridge = false
 
+[webdriver]
+
+; Sets the url of the webdriver or selenium server
+selenium_server_url = "http://localhost:4444"
+
+; Sets whether the browser should run in headless mode (no visible ui)
+; true = enabled
+; false = disabled (default)
+headless = false
+
 [authentication]
 
 ; HTTP basic authentication

diff --git a/docs/05_Bridge_API/04_WebDriverAbstract.md b/docs/05_Bridge_API/04_WebDriverAbstract.md
@@ -0,0 +1,83 @@
+`WebDriverAbstract` extends [`BridgeAbstract`](./02_BridgeAbstract.md) and adds functionality for generating feeds
+from active websites that use XMLHttpRequest (XHR) to load content and / or JavaScript to
+modify content.
+It highly depends on the php-webdriver library which offers Selenium WebDriver bindings for PHP.
+
+- https://github.com/php-webdriver/php-webdriver (Project Repository)
+- https://php-webdriver.github.io/php-webdriver/latest/ (API)
+
+Please note that this class is intended as a solution for websites _that cannot be covered
+by the other classes_. The WebDriver starts a browser and is therefore very resource-intensive.
+
+# Configuration
+
+You need a running WebDriver to use bridges that depend on `WebDriverAbstract`.
+The easiest way is to start the Selenium server from the project of the same name:
+```
+docker run -d -p 4444:4444 --shm-size="2g" docker.io/selenium/standalone-chrome:latest
+```
+
+- https://github.com/SeleniumHQ/docker-selenium
+
+With these parameters only one browser window can be started at a time.
+On a multi-user site, Selenium Grid should be used
+and the number of sessions should be adjusted to the number of processor cores.
+
+Finally, the `config.ini.php` file must be adjusted so that the WebDriver
+can find the Selenium server:
+```
+[webdriver]
+
+selenium_server_url = "http://localhost:4444"
+```
+
+# Development
+
+While you are programming a new bridge, it is easier to start a local WebDriver because then you can see what is happening and where the errors are. I've also had good experience recording the process with a screen video to find any timing problems.
+
+```
+chromedriver --port=4444
+```
+
+- https://chromedriver.chromium.org/
+
+If you start rss-bridge from a container, then Chrome driver is only accessible
+if you call it with the `--allowed-ips` option so that it binds to all network interfaces.
+
+```
+chromedriver --port=4444 --allowed-ips=192.168.1.42
+```
+
+The **most important rule** is that after an event such as loading the web page
+or pressing a button, you often have to explicitly wait for the desired elements to appear.
+
+A simple example is the bridge `ScalableCapitalBlogBridge.php`.
+A more complex and relatively complete example is the bridge `GULPProjekteBridge.php`.
+
+# Template
+
+Use this template to create your own bridge.
+
+```PHP
+<?php
+
+class MyBridge extends WebDriverAbstract
+{
+    const NAME = 'My Bridge';
+    const URI = 'https://www.example.org';
+    const DESCRIPTION = 'Further description';
+    const MAINTAINER = 'your name';
+
+    public function collectData()
+    {
+        parent::collectData();
+
+        try {
+            // TODO
+        } finally {
+            $this->cleanUp();
+        }
+    }
+}
+
+```
diff --git a/docs/05_Bridge_API/04_XPathAbstract.md → docs/05_Bridge_API/05_XPathAbstract.md b/docs/05_Bridge_API/04_XPathAbstract.md → docs/05_Bridge_API/05_XPathAbstract.md
diff --git a/docs/05_Bridge_API/index.md b/docs/05_Bridge_API/index.md
@@ -8,6 +8,7 @@ Base class | Description
 -----------|------------
 [`BridgeAbstract`](./02_BridgeAbstract.md) | This class is intended for standard _Bridges_ that need to filter HTML pages for content.
 [`FeedExpander`](./03_FeedExpander.md) | Expand/modify existing feed urls
-[`XPathAbstract`](./04_XPathAbstract.md) | This class is meant as an alternative base class for bridge implementations. It offers preliminary functionality for generating feeds based on _XPath expressions_.
+[`WebDriverAbstract`](./04_WebDriverAbstract) |
+[`XPathAbstract`](./05_XPathAbstract) | This class is meant as an alternative base class for bridge implementations. It offers preliminary functionality for generating feeds based on _XPath expressions_.
 
 For more information about how to create a new _Bridge_, read [How to create a new Bridge?](./01_How_to_create_a_new_bridge.md)