Merge pull request #1478 from Alkarex/better_search

Better search

2 files changed, 192 insertions, 45 deletions

diff --git a/app/Models/EntryDAO.php b/app/Models/EntryDAO.php
index 61ec48d08..7e836097a 100644
--- a/app/Models/EntryDAO.php
+++ b/app/Models/EntryDAO.php
@@ -630,18 +630,6 @@ class FreshRSS_EntryDAO extends Minz_ModelPdo implements FreshRSS_Searchable {
 			$search .= 'AND ' . $alias . 'id >= ' . $date_min . '000000 ';
 		}
 		if ($filter) {
-			if ($filter->getIntitle()) {
-				$search .= 'AND ' . $alias . 'title LIKE ? ';
-				$values[] = "%{$filter->getIntitle()}%";
-			}
-			if ($filter->getInurl()) {
-				$search .= 'AND CONCAT(' . $alias . 'link, ' . $alias . 'guid) LIKE ? ';
-				$values[] = "%{$filter->getInurl()}%";
-			}
-			if ($filter->getAuthor()) {
-				$search .= 'AND ' . $alias . 'author LIKE ? ';
-				$values[] = "%{$filter->getAuthor()}%";
-			}
 			if ($filter->getMinDate()) {
 				$search .= 'AND ' . $alias . 'id >= ? ';
 				$values[] = "{$filter->getMinDate()}000000";
@@ -658,20 +646,69 @@ class FreshRSS_EntryDAO extends Minz_ModelPdo implements FreshRSS_Searchable {
 				$search .= 'AND ' . $alias . 'date <= ? ';
 				$values[] = $filter->getMaxPubdate();
 			}
+
+			if ($filter->getAuthor()) {
+				foreach ($filter->getAuthor() as $author) {
+					$search .= 'AND ' . $alias . 'author LIKE ? ';
+					$values[] = "%{$author}%";
+				}
+			}
+			if ($filter->getIntitle()) {
+				foreach ($filter->getIntitle() as $title) {
+					$search .= 'AND ' . $alias . 'title LIKE ? ';
+					$values[] = "%{$title}%";
+				}
+			}
 			if ($filter->getTags()) {
-				$tags = $filter->getTags();
-				foreach ($tags as $tag) {
+				foreach ($filter->getTags() as $tag) {
 					$search .= 'AND ' . $alias . 'tags LIKE ? ';
 					$values[] = "%{$tag}%";
 				}
 			}
+			if ($filter->getInurl()) {
+				foreach ($filter->getInurl() as $url) {
+					$search .= 'AND CONCAT(' . $alias . 'link, ' . $alias . 'guid) LIKE ? ';
+					$values[] = "%{$url}%";
+				}
+			}
+
+			if ($filter->getNotAuthor()) {
+				foreach ($filter->getNotAuthor() as $author) {
+					$search .= 'AND (NOT ' . $alias . 'author LIKE ?) ';
+					$values[] = "%{$author}%";
+				}
+			}
+			if ($filter->getNotIntitle()) {
+				foreach ($filter->getNotIntitle() as $title) {
+					$search .= 'AND (NOT ' . $alias . 'title LIKE ?) ';
+					$values[] = "%{$title}%";
+				}
+			}
+			if ($filter->getNotTags()) {
+				foreach ($filter->getNotTags() as $tag) {
+					$search .= 'AND (NOT ' . $alias . 'tags LIKE ?) ';
+					$values[] = "%{$tag}%";
+				}
+			}
+			if ($filter->getNotInurl()) {
+				foreach ($filter->getNotInurl() as $url) {
+					$search .= 'AND (NOT CONCAT(' . $alias . 'link, ' . $alias . 'guid) LIKE ?) ';
+					$values[] = "%{$url}%";
+				}
+			}
+
 			if ($filter->getSearch()) {
-				$search_values = $filter->getSearch();
-				foreach ($search_values as $search_value) {
+				foreach ($filter->getSearch() as $search_value) {
 					$search .= 'AND ' . $this->sqlconcat($alias . 'title', $this->isCompressed() ? 'UNCOMPRESS(' . $alias . 'content_bin)' : '' . $alias . 'content') . ' LIKE ? ';
 					$values[] = "%{$search_value}%";
 				}
 			}
+			if ($filter->getNotSearch()) {
+				foreach ($filter->getNotSearch() as $search_value) {
+					$search .= 'AND (NOT ' . $this->sqlconcat($alias . 'title', $this->isCompressed() ? 'UNCOMPRESS(' . $alias . 'content_bin)' : '' . $alias . 'content') . ' LIKE ?) ';
+					$values[] = "%{$search_value}%";
+				}
+			}
 		}
 		return array($values, $search);
 	}
diff --git a/app/Models/Search.php b/app/Models/Search.php
index 575a9a2cb..5cc7f8e8d 100644
--- a/app/Models/Search.php
+++ b/app/Models/Search.php
@@ -23,18 +23,35 @@ class FreshRSS_Search {
 	private $tags;
 	private $search;
 
+	private $not_intitle;
+	private $not_inurl;
+	private $not_author;
+	private $not_tags;
+	private $not_search;
+
 	public function __construct($input) {
-		if (strcmp($input, '') == 0) {
+		if ($input == '') {
 			return;
 		}
 		$this->raw_input = $input;
+
+		$input = preg_replace('/:&quot;(.*?)&quot;/', ':"\1"', $input);
+
+		$input = $this->parseNotIntitleSearch($input);
+		$input = $this->parseNotAuthorSearch($input);
+		$input = $this->parseNotInurlSearch($input);
+		$input = $this->parseNotTagsSeach($input);
+
+		$input = $this->parsePubdateSearch($input);
+		$input = $this->parseDateSearch($input);
+
 		$input = $this->parseIntitleSearch($input);
 		$input = $this->parseAuthorSearch($input);
 		$input = $this->parseInurlSearch($input);
-		$input = $this->parsePubdateSearch($input);
-		$input = $this->parseDateSearch($input);
 		$input = $this->parseTagsSeach($input);
-		$this->parseSearch($input);
+
+		$input = $this->parseNotSearch($input);
+		$input = $this->parseSearch($input);
 	}
 
 	public function __toString() {
@@ -48,6 +65,9 @@ class FreshRSS_Search {
 	public function getIntitle() {
 		return $this->intitle;
 	}
+	public function getNotIntitle() {
+		return $this->not_intitle;
+	}
 
 	public function getMinDate() {
 		return $this->min_date;
@@ -68,18 +88,34 @@ class FreshRSS_Search {
 	public function getInurl() {
 		return $this->inurl;
 	}
+	public function getNotInurl() {
+		return $this->not_inurl;
+	}
 
 	public function getAuthor() {
 		return $this->author;
 	}
+	public function getNotAuthor() {
+		return $this->not_author;
+	}
 
 	public function getTags() {
 		return $this->tags;
 	}
+	public function getNotTags() {
+		return $this->not_tags;
+	}
 
 	public function getSearch() {
 		return $this->search;
 	}
+	public function getNotSearch() {
+		return $this->not_search;
+	}
+
+	private static function removeEmptyValues($anArray) {
+		return is_array($anArray) ? array_filter($anArray, function($value) { return $value !== ''; }) : array();
+	}
 
 	/**
 	 * Parse the search string to find intitle keyword and the search related
@@ -90,14 +126,28 @@ class FreshRSS_Search {
 	 * @return string
 	 */
 	private function parseIntitleSearch($input) {
-		if (preg_match('/intitle:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
+		if (preg_match_all('/\bintitle:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
 			$this->intitle = $matches['search'];
-			return str_replace($matches[0], '', $input);
+			$input = str_replace($matches[0], '', $input);
 		}
-		if (preg_match('/intitle:(?P<search>\w*)/', $input, $matches)) {
-			$this->intitle = $matches['search'];
-			return str_replace($matches[0], '', $input);
+		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);
+		return $input;
+	}
+
+	private function parseNotIntitleSearch($input) {
+		if (preg_match_all('/[!-]intitle:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
+			$this->not_intitle = $matches['search'];
+			$input = str_replace($matches[0], '', $input);
+		}
+		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);
 		return $input;
 	}
 
@@ -112,30 +162,54 @@ class FreshRSS_Search {
 	 * @return string
 	 */
 	private function parseAuthorSearch($input) {
-		if (preg_match('/author:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
+		if (preg_match_all('/\bauthor:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
 			$this->author = $matches['search'];
-			return str_replace($matches[0], '', $input);
+			$input = str_replace($matches[0], '', $input);
 		}
-		if (preg_match('/author:(?P<search>\w*)/', $input, $matches)) {
-			$this->author = $matches['search'];
-			return str_replace($matches[0], '', $input);
+		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);
+		return $input;
+	}
+
+	private function parseNotAuthorSearch($input) {
+		if (preg_match_all('/[!-]author:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
+			$this->not_author = $matches['search'];
+			$input = str_replace($matches[0], '', $input);
+		}
+		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);
 		return $input;
 	}
 
 	/**
 	 * Parse the search string to find inurl keyword and the search related
 	 * to it.
-	 * The search is the first word following the keyword except.
+	 * The search is the first word following the keyword.
 	 *
 	 * @param string $input
 	 * @return string
 	 */
 	private function parseInurlSearch($input) {
-		if (preg_match('/inurl:(?P<search>[^\s]*)/', $input, $matches)) {
+		if (preg_match_all('/\binurl:(?P<search>[^\s]*)/', $input, $matches)) {
 			$this->inurl = $matches['search'];
-			return str_replace($matches[0], '', $input);
+			$input = str_replace($matches[0], '', $input);
+		}
+		$this->inurl = self::removeEmptyValues($this->inurl);
+		return $input;
+	}
+
+	private function parseNotInurlSearch($input) {
+		if (preg_match_all('/[!-]inurl:(?P<search>[^\s]*)/', $input, $matches)) {
+			$this->not_inurl = $matches['search'];
+			$input = str_replace($matches[0], '', $input);
 		}
+		$this->not_inurl = self::removeEmptyValues($this->not_inurl);
 		return $input;
 	}
 
@@ -148,9 +222,12 @@ class FreshRSS_Search {
 	 * @return string
 	 */
 	private function parseDateSearch($input) {
-		if (preg_match('/date:(?P<search>[^\s]*)/', $input, $matches)) {
-			list($this->min_date, $this->max_date) = parseDateInterval($matches['search']);
-			return str_replace($matches[0], '', $input);
+		if (preg_match_all('/\bdate:(?P<search>[^\s]*)/', $input, $matches)) {
+			$input = str_replace($matches[0], '', $input);
+			$dates = self::removeEmptyValues($matches['search']);
+			if (!empty($dates[0])) {
+				list($this->min_date, $this->max_date) = parseDateInterval($dates[0]);
+			}
 		}
 		return $input;
 	}
@@ -164,9 +241,12 @@ class FreshRSS_Search {
 	 * @return string
 	 */
 	private function parsePubdateSearch($input) {
-		if (preg_match('/pubdate:(?P<search>[^\s]*)/', $input, $matches)) {
-			list($this->min_pubdate, $this->max_pubdate) = parseDateInterval($matches['search']);
-			return str_replace($matches[0], '', $input);
+		if (preg_match_all('/\bpubdate:(?P<search>[^\s]*)/', $input, $matches)) {
+			$input = str_replace($matches[0], '', $input);
+			$dates = self::removeEmptyValues($matches['search']);
+			if (!empty($dates[0])) {
+				list($this->min_pubdate, $this->max_pubdate) = parseDateInterval($dates[0]);
+			}
 		}
 		return $input;
 	}
@@ -182,8 +262,18 @@ class FreshRSS_Search {
 	private function parseTagsSeach($input) {
 		if (preg_match_all('/#(?P<search>[^\s]+)/', $input, $matches)) {
 			$this->tags = $matches['search'];
-			return str_replace($matches[0], '', $input);
+			$input = str_replace($matches[0], '', $input);
 		}
+		$this->tags = self::removeEmptyValues($this->tags);
+		return $input;
+	}
+
+	private function parseNotTagsSeach($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);
 		return $input;
 	}
 
@@ -196,16 +286,16 @@ class FreshRSS_Search {
 	 * @return string
 	 */
 	private function parseSearch($input) {
-		$input = $this->cleanSearch($input);
-		if (strcmp($input, '') == 0) {
+		$input = self::cleanSearch($input);
+		if ($input == '') {
 			return;
 		}
 		if (preg_match_all('/(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
 			$this->search = $matches['search'];
 			$input = str_replace($matches[0], '', $input);
 		}
-		$input = $this->cleanSearch($input);
-		if (strcmp($input, '') == 0) {
+		$input = self::cleanSearch($input);
+		if ($input == '') {
 			return;
 		}
 		if (is_array($this->search)) {
@@ -215,13 +305,33 @@ class FreshRSS_Search {
 		}
 	}
 
+	private function parseNotSearch($input) {
+		$input = self::cleanSearch($input);
+		if ($input == '') {
+			return;
+		}
+		if (preg_match_all('/[!-](?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
+			$this->not_search = $matches['search'];
+			$input = str_replace($matches[0], '', $input);
+		}
+		if ($input == '') {
+			return;
+		}
+		if (preg_match_all('/[!-](?P<search>[^\s]+)/', $input, $matches)) {
+			$this->not_search = array_merge(is_array($this->not_search) ? $this->not_search : array(), $matches['search']);
+			$input = str_replace($matches[0], '', $input);
+		}
+		$this->not_search = self::removeEmptyValues($this->not_search);
+		return $input;
+	}
+
 	/**
 	 * Remove all unnecessary spaces in the search
 	 *
 	 * @param string $input
 	 * @return string
 	 */
-	private function cleanSearch($input) {
+	private static function cleanSearch($input) {
 		$input = preg_replace('/\s+/', ' ', $input);
 		return trim($input);
 	}