Add search operator intext: (#7228)

* Add search operator intext:
fix https://github.com/FreshRSS/FreshRSS/issues/6188
https://github.com/FreshRSS/FreshRSS/discussions/7220

* Add example to doc

3 files changed, 138 insertions, 1 deletions

diff --git a/app/Models/Entry.php b/app/Models/Entry.php
index f86948122..fe0cf7429 100644
--- a/app/Models/Entry.php
+++ b/app/Models/Entry.php
@@ -673,6 +673,26 @@ HTML;
 						$ok &= preg_match($title, $this->title) === 0;
 					}
 				}
+				if ($ok && $filter->getIntext() !== null) {
+					foreach ($filter->getIntext() as $content) {
+						$ok &= stripos($this->content, $content) !== false;
+					}
+				}
+				if ($ok && $filter->getIntextRegex() !== null) {
+					foreach ($filter->getIntextRegex() as $content) {
+						$ok &= preg_match($content, $this->content) === 1;
+					}
+				}
+				if ($ok && $filter->getNotIntext() !== null) {
+					foreach ($filter->getNotIntext() as $content) {
+						$ok &= stripos($this->content, $content) === false;
+					}
+				}
+				if ($ok && $filter->getNotIntextRegex() !== null) {
+					foreach ($filter->getNotIntextRegex() as $content) {
+						$ok &= preg_match($content, $this->content) === 0;
+					}
+				}
 				if ($ok && $filter->getTags() !== null) {
 					foreach ($filter->getTags() as $tag2) {
 						$found = false;
diff --git a/app/Models/EntryDAO.php b/app/Models/EntryDAO.php
index af229df54..a234dce91 100644
--- a/app/Models/EntryDAO.php
+++ b/app/Models/EntryDAO.php
@@ -981,6 +981,30 @@ SQL;
 					$sub_search .= 'AND ' . static::sqlRegex($alias . 'title', $title, $values) . ' ';
 				}
 			}
+			if ($filter->getIntext() !== null) {
+				if (static::isCompressed()) {	// MySQL-only
+					foreach ($filter->getIntext() as $content) {
+						$sub_search .= "AND UNCOMPRESS({$alias}content_bin) LIKE ? ";
+						$values[] = "%{$content}%";
+					}
+				} else {
+					foreach ($filter->getIntext() as $content) {
+						$sub_search .= 'AND ' . $alias . 'content LIKE ? ';
+						$values[] = "%{$content}%";
+					}
+				}
+			}
+			if ($filter->getIntextRegex() !== null) {
+				if (static::isCompressed()) {	// MySQL-only
+					foreach ($filter->getIntextRegex() as $content) {
+						$sub_search .= 'AND ' . static::sqlRegex("UNCOMPRESS({$alias}content_bin)", $content, $values) . ') ';
+					}
+				} else {
+					foreach ($filter->getIntextRegex() as $content) {
+						$sub_search .= 'AND ' . static::sqlRegex($alias . 'content', $content, $values) . ' ';
+					}
+				}
+			}
 			if ($filter->getTags() !== null) {
 				foreach ($filter->getTags() as $tag) {
 					$sub_search .= 'AND ' . static::sqlConcat('TRIM(' . $alias . 'tags) ', " ' #'") . ' LIKE ? ';
@@ -1026,6 +1050,30 @@ SQL;
 					$sub_search .= 'AND NOT ' . static::sqlRegex($alias . 'title', $title, $values) . ' ';
 				}
 			}
+			if ($filter->getNotIntext() !== null) {
+				if (static::isCompressed()) {	// MySQL-only
+					foreach ($filter->getNotIntext() as $content) {
+						$sub_search .= "AND UNCOMPRESS({$alias}content_bin) NOT LIKE ? ";
+						$values[] = "%{$content}%";
+					}
+				} else {
+					foreach ($filter->getNotIntext() as $content) {
+						$sub_search .= 'AND ' . $alias . 'content NOT LIKE ? ';
+						$values[] = "%{$content}%";
+					}
+				}
+			}
+			if ($filter->getNotIntextRegex() !== null) {
+				if (static::isCompressed()) {	// MySQL-only
+					foreach ($filter->getNotIntextRegex() as $content) {
+						$sub_search .= 'AND NOT ' . static::sqlRegex("UNCOMPRESS({$alias}content_bin)", $content, $values) . ') ';
+					}
+				} else {
+					foreach ($filter->getNotIntextRegex() as $content) {
+						$sub_search .= 'AND NOT ' . static::sqlRegex($alias . 'content', $content, $values) . ' ';
+					}
+				}
+			}
 			if ($filter->getNotTags() !== null) {
 				foreach ($filter->getNotTags() as $tag) {
 					$sub_search .= 'AND ' . static::sqlConcat('TRIM(' . $alias . 'tags) ', " ' #'") . ' NOT LIKE ? ';
diff --git a/app/Models/Search.php b/app/Models/Search.php
index 5b88b1f3b..d425fcee8 100644
--- a/app/Models/Search.php
+++ b/app/Models/Search.php
@@ -29,6 +29,10 @@ class FreshRSS_Search implements \Stringable {
 	private ?array $intitle = null;
 	/** @var list<string>|null */
 	private ?array $intitle_regex = null;
+	/** @var list<string>|null */
+	private ?array $intext = null;
+	/** @var list<string>|null */
+	private ?array $intext_regex = null;
 	/** @var int|false|null */
 	private $min_date = null;
 	/** @var int|false|null */
@@ -66,6 +70,10 @@ class FreshRSS_Search implements \Stringable {
 	private ?array $not_intitle = null;
 	/** @var list<string>|null */
 	private ?array $not_intitle_regex = null;
+	/** @var list<string>|null */
+	private ?array $not_intext = null;
+	/** @var list<string>|null */
+	private ?array $not_intext_regex = null;
 	/** @var int|false|null */
 	private $not_min_date = null;
 	/** @var int|false|null */
@@ -106,6 +114,7 @@ class FreshRSS_Search implements \Stringable {
 		$input = $this->parseNotDateSearch($input);
 
 		$input = $this->parseNotIntitleSearch($input);
+		$input = $this->parseNotIntextSearch($input);
 		$input = $this->parseNotAuthorSearch($input);
 		$input = $this->parseNotInurlSearch($input);
 		$input = $this->parseNotTagsSearch($input);
@@ -119,6 +128,7 @@ class FreshRSS_Search implements \Stringable {
 		$input = $this->parseDateSearch($input);
 
 		$input = $this->parseIntitleSearch($input);
+		$input = $this->parseIntextSearch($input);
 		$input = $this->parseAuthorSearch($input);
 		$input = $this->parseInurlSearch($input);
 		$input = $this->parseTagsSearch($input);
@@ -189,6 +199,23 @@ class FreshRSS_Search implements \Stringable {
 		return $this->not_intitle_regex;
 	}
 
+	/** @return list<string>|null */
+	public function getIntext(): ?array {
+		return $this->intext;
+	}
+	/** @return list<string>|null */
+	public function getIntextRegex(): ?array {
+		return $this->intext_regex;
+	}
+	/** @return list<string>|null */
+	public function getNotIntext(): ?array {
+		return $this->not_intext;
+	}
+	/** @return list<string>|null */
+	public function getNotIntextRegex(): ?array {
+		return $this->not_intext_regex;
+	}
+
 	public function getMinDate(): ?int {
 		return $this->min_date ?: null;
 	}
@@ -494,7 +521,6 @@ class FreshRSS_Search implements \Stringable {
 
 	/**
 	 * Parse the search string to find intitle keyword and the search related to it.
-	 * The search is the first word following the keyword.
 	 */
 	private function parseIntitleSearch(string $input): string {
 		if (preg_match_all('#\\bintitle:(?P<search>/.*?(?<!\\\\)/[im]*)#', $input, $matches)) {
@@ -537,6 +563,49 @@ class FreshRSS_Search implements \Stringable {
 	}
 
 	/**
+	 * Parse the search string to find intext keyword and the search related to it.
+	 */
+	private function parseIntextSearch(string $input): string {
+		if (preg_match_all('#\\bintext:(?P<search>/.*?(?<!\\\\)/[im]*)#', $input, $matches)) {
+			$this->intext_regex = self::htmlspecialchars_decodes($matches['search']);
+			$input = str_replace($matches[0], '', $input);
+		}
+		if (preg_match_all('/\\bintext:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
+			$this->intext = $matches['search'];
+			$input = str_replace($matches[0], '', $input);
+		}
+		if (preg_match_all('/\\bintext:(?P<search>[^\s"]*)/', $input, $matches)) {
+			$this->intext = array_merge($this->intext ?? [], $matches['search']);
+			$input = str_replace($matches[0], '', $input);
+		}
+		$this->intext = self::removeEmptyValues($this->intext);
+		if (empty($this->intext)) {
+			$this->intext = null;
+		}
+		return $input;
+	}
+
+	private function parseNotIntextSearch(string $input): string {
+		if (preg_match_all('#(?<=[\\s(]|^)[!-]intext:(?P<search>/.*?(?<!\\\\)/[im]*)#', $input, $matches)) {
+			$this->not_intext_regex = self::htmlspecialchars_decodes($matches['search']);
+			$input = str_replace($matches[0], '', $input);
+		}
+		if (preg_match_all('/(?<=[\\s(]|^)[!-]intext:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $input, $matches)) {
+			$this->not_intext = $matches['search'];
+			$input = str_replace($matches[0], '', $input);
+		}
+		if (preg_match_all('/(?<=[\\s(]|^)[!-]intext:(?P<search>[^\s"]*)/', $input, $matches)) {
+			$this->not_intext = array_merge($this->not_intext ?? [], $matches['search']);
+			$input = str_replace($matches[0], '', $input);
+		}
+		$this->not_intext = self::removeEmptyValues($this->not_intext);
+		if (empty($this->not_intext)) {
+			$this->not_intext = null;
+		}
+		return $input;
+	}
+
+	/**
 	 * Parse the search string to find author keyword and the search related to it.
 	 * The search is the first word following the keyword except when using
 	 * a delimiter. Supported delimiters are single quote (') and double quotes (").