Merge pull request #1097 from nextcloud/dev/303-polymophism-in-filters

Dev/303 polymophism in filters

.github/workflows/dependabot-approve-merge.yml

@@ -12,17 +12,21 @@ on:
 
 jobs:
   auto-approve-merge:
-    if: github.actor == 'dependabot[bot]'
     runs-on: ubuntu-latest
 
     steps:
       # Github actions bot approve
       - uses: hmarr/auto-approve-action@v2
+        if: github.actor == 'dependabot[bot]'
         with:
           github-token: ${{ secrets.GITHUB_TOKEN }}
 
       # Nextcloud bot approve and merge request
       - uses: ahmadnassri/action-dependabot-auto-merge@v2
+        if: github.actor == 'dependabot[bot]'
         with:
           target: minor
           github-token: ${{ secrets.COOKBOOK_BOT_TOKEN }}
+
+      - run: "echo 'Skipping PR auto-merge as it is no dependabot PR'"
+        if: github.actor != 'dependabot[bot]'

CHANGELOG.md

@@ -1,5 +1,9 @@
 ## [Unreleased]
 
+### Changed
+- Parsing of JSON recipe objects in a cascade of filters
+  [#1097](https://github.com/nextcloud/cookbook/pull/1097) @christianlupus
+
 ### Fixed
 - Prevent slow loading of recipes due to iteration over all files
   [#1072](https://github.com/nextcloud/cookbook/pull/1072) @christianlupus

lib/Exception/InvalidDurationException.php

@@ -0,0 +1,9 @@
+<?php
+
+namespace OCA\Cookbook\Exception;
+
+class InvalidDurationException extends \Exception {
+	public function __construct($message = null, $code = null, $previous = null) {
+		parent::__construct($message, $code, $previous);
+	}
+}

lib/Exception/InvalidRecipeException.php

@@ -0,0 +1,9 @@
+<?php
+
+namespace OCA\Cookbook\Exception;
+
+class InvalidRecipeException extends \Exception {
+	public function __construct($message = null, $code = null, $previous = null) {
+		parent::__construct($message, $code, $previous);
+	}
+}

lib/Helper/Filter/AbstractJSONFilter.php

@@ -0,0 +1,36 @@
+<?php
+
+namespace OCA\Cookbook\Helper\Filter;
+
+use OCA\Cookbook\Exception\InvalidRecipeException;
+use OCP\Files\File;
+
+/**
+ * An abstract filter on a recipe JSON.
+ *
+ * A filter should have a single purpose that is serves and implement this interface
+ */
+abstract class AbstractJSONFilter {
+	/**
+	 * Filter the given recipe according to the filter class specification.
+	 *
+	 * This function can make changes to the recipe array to carry out the needed changes.
+	 * In order to signal if the JSON file needs updating, the return value must be true if and only if the recipe was changed.
+	 *
+	 * @param array $json The recipe data as array
+	 * @return bool true, if and only if the recipe was changed
+	 * @throws InvalidRecipeException if the recipe was not correctly filtered
+	 */
+	abstract public function apply(array &$json): bool;
+
+	protected function setJSONValue(array &$json, string $key, $value): bool {
+		if (!array_key_exists($key, $json)) {
+			$json[$key] = $value;
+			return true;
+		} elseif ($json[$key] !== $value) {
+			$json[$key] = $value;
+			return true;
+		}
+		return false;
+	}
+}

lib/Helper/Filter/JSON/CleanCategoryFilter.php

@@ -0,0 +1,44 @@
+<?php
+
+namespace OCA\Cookbook\Helper\Filter\JSON;
+
+use OCA\Cookbook\Helper\Filter\AbstractJSONFilter;
+use OCA\Cookbook\Helper\TextCleanupHelper;
+
+/**
+ * Clean the category of a recipe.
+ *
+ * A recipe must have at most one category.
+ * This category must be a string.
+ * Recipes without category are assigned the empty string for the category.
+ */
+class CleanCategoryFilter extends AbstractJSONFilter {
+	/** @var TextCleanupHelper */
+	private $textCleaner;
+
+	public function __construct(TextCleanupHelper $cleanupHelper) {
+		$this->textCleaner = $cleanupHelper;
+	}
+
+	public function apply(array &$json): bool {
+		if (! isset($json['recipeCategory'])) {
+			$json['recipeCategory'] = '';
+			return true;
+		}
+
+		$cache = $json['recipeCategory'];
+
+		if (is_array($json['recipeCategory'])) {
+			reset($json['recipeCategory']);
+			$json['recipeCategory'] = current($json['recipeCategory']);
+		}
+
+		if (!is_string($json['recipeCategory'])) {
+			$json['recipeCategory'] = '';
+		}
+
+		$json['recipeCategory'] = $this->textCleaner->cleanUp($json['recipeCategory'], true, true);
+
+		return $cache !== $json['recipeCategory'];
+	}
+}

lib/Helper/Filter/JSON/ExtractImageUrlFilter.php

@@ -0,0 +1,104 @@
+<?php
+
+namespace OCA\Cookbook\Helper\Filter\JSON;
+
+use OCA\Cookbook\Helper\Filter\AbstractJSONFilter;
+use OCP\IL10N;
+use OCP\ILogger;
+
+/**
+ * Select the best image URL.
+ *
+ * This filter makes a few sanity checks on the recipe image URL.
+ * It tries to guess the image URL with the highest resolution heuristically.
+ *
+ * It ensures that the image property is no object but a pure URL.
+ */
+class ExtractImageUrlFilter extends AbstractJSONFilter {
+	/** @var IL10N */
+	private $l;
+
+	/** @var ILogger */
+	private $logger;
+
+	public function __construct(IL10N $l, ILogger $logger) {
+		$this->l = $l;
+		$this->logger = $logger;
+	}
+
+	public function apply(array &$json): bool {
+		if (!isset($json['image']) || !$json['image']) {
+			$json['image'] = '';
+			return true;
+		}
+
+		if (is_string($json['image'])) {
+			// We do not change anything here
+			return false;
+		}
+
+		if (!is_array($json['image'])) {
+			// It is neither a plain string nor a JSON object
+			$this->logger->info($this->l->t('The given image for the recipe %s cannot be parsed. Aborting and skipping it.', [$json['name']]));
+			$json['image'] = '';
+			return true;
+		}
+
+		// We have an array of images or an object
+		if (isset($json['image']['url'])) {
+			// We have a single object. Use it.
+			$json['image'] = $json['image']['url'];
+			return true;
+		}
+
+		// We have an array of images. Extract the URLs
+		$images = array_map(function ($x) {
+			if (is_string($x)) {
+				return $x;
+			}
+			if (is_array($x) && isset($x['url']) && $x['url'] && is_string($x['url'])) {
+				return $x['url'];
+			}
+			return null;
+		}, $json['image']);
+		$images = array_filter($images);
+
+		if (count($images) === 1) {
+			reset($images);
+			$json['image'] = current($images);
+			return true;
+		}
+
+		if (count($images) === 0) {
+			$this->logger->info($this->l->t('No valid recipe was left after heuristics of recipe  %s.', [$json['name']]));
+			$json['image'] = '';
+			return true;
+		}
+
+		$maxSum = -1;
+		$changed = false;
+
+		foreach ($images as $img) {
+			$regex = preg_match_all('/\d+/', $img, $matches);
+
+			if ($regex !== false && $regex > 0) {
+				$sum = array_sum(array_map(fn ($x) => ((int) $x), $matches[0]));
+
+				if ($sum > $maxSum) {
+					$json['image'] = $img;
+					$maxSum = $sum;
+					$changed = true;
+				}
+			}
+		}
+
+		if (!$changed) {
+			$this->logger->info($this->l->t('Heuristics failed for image extraction of recipe %s.', [$json['name']]));
+
+			reset($images);
+			$json['image'] = current($images);
+		}
+
+		return true;
+	}
+}

lib/Helper/Filter/JSON/FixDescriptionFilter.php

@@ -0,0 +1,53 @@
+<?php
+
+namespace OCA\Cookbook\Helper\Filter\JSON;
+
+use OCP\IL10N;
+use OCP\ILogger;
+use OCA\Cookbook\Helper\Filter\AbstractJSONFilter;
+use OCA\Cookbook\Helper\TextCleanupHelper;
+
+/**
+ * Fix the description of a recipe.
+ *
+ * The description must be present in all cases.
+ * If is is not present, this filter will insert the empty string instead.
+ *
+ * Apart from that the description is cleaned from malicious chars.
+ */
+class FixDescriptionFilter extends AbstractJSONFilter {
+	private const NUTRITION = 'description';
+
+	/** @var IL10N */
+	private $l;
+
+	/** @var ILogger */
+	private $logger;
+
+	/** @var TextCleanupHelper */
+	private $textCleaner;
+
+	public function __construct(
+		IL10N $l,
+		ILogger $logger,
+		TextCleanupHelper $textCleanupHelper
+	) {
+		$this->l = $l;
+		$this->logger = $logger;
+		$this->textCleaner = $textCleanupHelper;
+	}
+
+	public function apply(array &$json): bool {
+		if (!isset($json[self::NUTRITION])) {
+			$json[self::NUTRITION] = '';
+			return true;
+		}
+
+		$description = $this->textCleaner->cleanUp($json[self::NUTRITION], false);
+		$description = trim($description);
+
+		$changed = $description !== $json[self::NUTRITION];
+		$json[self::NUTRITION] = $description;
+		return $changed;
+	}
+}

lib/Helper/Filter/JSON/FixDurationsFilter.php

@@ -0,0 +1,69 @@
+<?php
+
+namespace OCA\Cookbook\Helper\Filter\JSON;
+
+use OCA\Cookbook\Exception\InvalidDurationException;
+use OCP\IL10N;
+use OCP\ILogger;
+use OCA\Cookbook\Helper\Filter\AbstractJSONFilter;
+use OCA\Cookbook\Helper\ISO8601DurationHelper;
+
+/**
+ * Fix the durations of a recipe.
+ *
+ * The filter works on the prepTime, cookTime and totalTime entries, individually.
+ *
+ * If a time is not present or not able to read, it is set to null.
+ * The time is parsed and reformatted as ISO 8601 duration.
+ */
+class FixDurationsFilter extends AbstractJSONFilter {
+	private const DURATIONS = 'durations';
+
+	/** @var IL10N */
+	private $l;
+
+	/** @var ILogger */
+	private $logger;
+
+	/** @var ISO8601DurationHelper */
+	private $iso8601DurationHelper;
+
+	public function __construct(
+		IL10N $l,
+		ILogger $logger,
+		ISO8601DurationHelper $isoHelper
+	) {
+		$this->l = $l;
+		$this->logger = $logger;
+		$this->iso8601DurationHelper = $isoHelper;
+	}
+
+	public function apply(array &$json): bool {
+		$changed = false;
+
+		$this->fixDate($json, 'prepTime', $changed);
+		$this->fixDate($json, 'cookTime', $changed);
+		$this->fixDate($json, 'totalTime', $changed);
+
+		return $changed;
+	}
+
+	private function fixDate(array &$json, string $type, bool &$changed): void {
+		if (!isset($json[$type])) {
+			$json[$type] = null;
+			$changed = true;
+			return;
+		}
+
+		$orig = $json[$type];
+		try {
+			$json[$type] = $this->iso8601DurationHelper->parseDuration($json[$type]);
+		} catch (InvalidDurationException $ex) {
+			$json[$type] = null;
+		}
+
+		if ($orig !== $json[$type]) {
+			$changed = true;
+		}
+	}
+}