Tags split improvement (#2023)

* Tags split improvement

Some feeds use a single category with comma-separated tags.
Better handling of tags containing a space

* Handle spaces in searches with +

Can now search in tags containing spaces

* Fix searches with spaces for title and author

3 files changed, 47 insertions, 25 deletions

diff --git a/app/Models/Entry.php b/app/Models/Entry.php
index ccbad5724..48a0b1bed 100644
--- a/app/Models/Entry.php
+++ b/app/Models/Entry.php
@@ -32,7 +32,7 @@ class FreshRSS_Entry extends Minz_Model {
 		$this->_isFavorite($is_favorite);
 		$this->_feedId($feedId);
 		$tags = mb_strcut($tags, 0, 1023, 'UTF-8');
-		$this->_tags(preg_split('/[\s#]/', $tags));
+		$this->_tags($tags);
 		$this->_guid($guid);
 	}
 
@@ -86,9 +86,9 @@ class FreshRSS_Entry extends Minz_Model {
 			return $this->feedId;
 		}
 	}
-	public function tags($inString = false) {
-		if ($inString) {
-			return empty($this->tags) ? '' : '#' . implode(' #', $this->tags);
+	public function tags($asString = false) {
+		if ($asString) {
+			return $this->tags == '' ? '' : '#' . implode(' #', $this->tags);
 		} else {
 			return $this->tags;
 		}
@@ -162,15 +162,8 @@ class FreshRSS_Entry extends Minz_Model {
 	public function _tags($value) {
 		$this->hash = null;
 		if (!is_array($value)) {
-			$value = array($value);
+			$value = preg_split('/\s*[#,]\s*/', $value, -1, PREG_SPLIT_NO_EMPTY);
 		}
-
-		foreach ($value as $key => $t) {
-			if (!$t) {
-				unset($value[$key]);
-			}
-		}
-
 		$this->tags = $value;
 	}
 
diff --git a/app/Models/Feed.php b/app/Models/Feed.php
index ed381a867..cc96cde44 100644
--- a/app/Models/Feed.php
+++ b/app/Models/Feed.php
@@ -345,13 +345,21 @@ class FreshRSS_Feed extends Minz_Model {
 			$link = $item->get_permalink();
 			$date = @strtotime($item->get_date());
 
-			// gestion des tags (catégorie == tag)
-			$tags_tmp = $item->get_categories();
+			//Tag processing (tag == category)
+			$categories = $item->get_categories();
 			$tags = array();
-			if ($tags_tmp !== null) {
-				foreach ($tags_tmp as $tag) {
-					$tags[] = html_only_entity_decode($tag->get_label());
+			if (is_array($categories)) {
+				foreach ($categories as $category) {
+					$text = html_only_entity_decode($category->get_label());
+					//Some feeds use a single category with comma-separated tags
+					$labels = explode(',', $text);
+					if (is_array($labels)) {
+						foreach ($labels as $label) {
+							$tags[] = trim($label);
+						}
+					}
 				}
+				$tags = array_unique($tags);
 			}
 
 			$content = html_only_entity_decode($item->get_content());
diff --git a/app/Models/Search.php b/app/Models/Search.php
index 5cc7f8e8d..c338d63a1 100644
--- a/app/Models/Search.php
+++ b/app/Models/Search.php
@@ -40,7 +40,7 @@ class FreshRSS_Search {
 		$input = $this->parseNotIntitleSearch($input);
 		$input = $this->parseNotAuthorSearch($input);
 		$input = $this->parseNotInurlSearch($input);
-		$input = $this->parseNotTagsSeach($input);
+		$input = $this->parseNotTagsSearch($input);
 
 		$input = $this->parsePubdateSearch($input);
 		$input = $this->parseDateSearch($input);
@@ -48,7 +48,7 @@ class FreshRSS_Search {
 		$input = $this->parseIntitleSearch($input);
 		$input = $this->parseAuthorSearch($input);
 		$input = $this->parseInurlSearch($input);
-		$input = $this->parseTagsSeach($input);
+		$input = $this->parseTagsSearch($input);
 
 		$input = $this->parseNotSearch($input);
 		$input = $this->parseSearch($input);
@@ -117,6 +117,17 @@ class FreshRSS_Search {
 		return is_array($anArray) ? array_filter($anArray, function($value) { return $value !== ''; }) : array();
 	}
 
+	private static function decodeSpaces($value) {
+		if (is_array($value)) {
+			for ($i = count($value) - 1; $i >= 0; $i--) {
+				$value[$i] = self::decodeSpaces($value[$i]);
+			}
+		} else {
+			$value = trim(str_replace('+', ' ', $value));
+		}
+		return $value;
+	}
+
 	/**
 	 * Parse the search string to find intitle keyword and the search related
 	 * to it.
@@ -130,11 +141,12 @@ class FreshRSS_Search {
 			$this->intitle = $matches['search'];
 			$input = str_replace($matches[0], '', $input);
 		}
-		if (preg_match_all('/\bintitle:(?P<search>\w*)/', $input, $matches)) {
+		if (preg_match_all('/\bintitle:(?P<search>[\w+]*)/', $input, $matches)) {
 			$this->intitle = array_merge($this->intitle ? $this->intitle : array(), $matches['search']);
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->intitle = self::removeEmptyValues($this->intitle);
+		$this->intitle = self::decodeSpaces($this->intitle);
 		return $input;
 	}
 
@@ -143,11 +155,12 @@ class FreshRSS_Search {
 			$this->not_intitle = $matches['search'];
 			$input = str_replace($matches[0], '', $input);
 		}
-		if (preg_match_all('/[!-]intitle:(?P<search>\w*)/', $input, $matches)) {
+		if (preg_match_all('/[!-]intitle:(?P<search>[\w+]*)/', $input, $matches)) {
 			$this->not_intitle = array_merge($this->not_intitle ? $this->not_intitle : array(), $matches['search']);
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->not_intitle = self::removeEmptyValues($this->not_intitle);
+		$this->not_intitle = self::decodeSpaces($this->not_intitle);
 		return $input;
 	}
 
@@ -166,11 +179,12 @@ class FreshRSS_Search {
 			$this->author = $matches['search'];
 			$input = str_replace($matches[0], '', $input);
 		}
-		if (preg_match_all('/\bauthor:(?P<search>\w*)/', $input, $matches)) {
+		if (preg_match_all('/\bauthor:(?P<search>[\w+]*)/', $input, $matches)) {
 			$this->author = array_merge($this->author ? $this->author : array(), $matches['search']);
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->author = self::removeEmptyValues($this->author);
+		$this->author = self::decodeSpaces($this->author);
 		return $input;
 	}
 
@@ -179,11 +193,12 @@ class FreshRSS_Search {
 			$this->not_author = $matches['search'];
 			$input = str_replace($matches[0], '', $input);
 		}
-		if (preg_match_all('/[!-]author:(?P<search>\w*)/', $input, $matches)) {
+		if (preg_match_all('/[!-]author:(?P<search>[\w+]*)/', $input, $matches)) {
 			$this->not_author = array_merge($this->not_author ? $this->not_author : array(), $matches['search']);
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->not_author = self::removeEmptyValues($this->not_author);
+		$this->not_author = self::decodeSpaces($this->not_author);
 		return $input;
 	}
 
@@ -201,6 +216,7 @@ class FreshRSS_Search {
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->inurl = self::removeEmptyValues($this->inurl);
+		$this->inurl = self::decodeSpaces($this->inurl);
 		return $input;
 	}
 
@@ -210,6 +226,7 @@ class FreshRSS_Search {
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->not_inurl = self::removeEmptyValues($this->not_inurl);
+		$this->not_inurl = self::decodeSpaces($this->not_inurl);
 		return $input;
 	}
 
@@ -259,21 +276,23 @@ class FreshRSS_Search {
 	 * @param string $input
 	 * @return string
 	 */
-	private function parseTagsSeach($input) {
+	private function parseTagsSearch($input) {
 		if (preg_match_all('/#(?P<search>[^\s]+)/', $input, $matches)) {
 			$this->tags = $matches['search'];
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->tags = self::removeEmptyValues($this->tags);
+		$this->tags = self::decodeSpaces($this->tags);
 		return $input;
 	}
 
-	private function parseNotTagsSeach($input) {
+	private function parseNotTagsSearch($input) {
 		if (preg_match_all('/[!-]#(?P<search>[^\s]+)/', $input, $matches)) {
 			$this->not_tags = $matches['search'];
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->not_tags = self::removeEmptyValues($this->not_tags);
+		$this->not_tags = self::decodeSpaces($this->not_tags);
 		return $input;
 	}
 
@@ -303,6 +322,7 @@ class FreshRSS_Search {
 		} else {
 			$this->search = explode(' ', $input);
 		}
+		$this->search = self::decodeSpaces($this->search);
 	}
 
 	private function parseNotSearch($input) {
@@ -322,6 +342,7 @@ class FreshRSS_Search {
 			$input = str_replace($matches[0], '', $input);
 		}
 		$this->not_search = self::removeEmptyValues($this->not_search);
+		$this->not_search = self::decodeSpaces($this->not_search);
 		return $input;
 	}