From db657b0149cc6a29eeb62bc8ccc96a06f505a7bf Mon Sep 17 00:00:00 2001
From: Michael <heluecht@pirati.ca>
Date: Fri, 17 Apr 2020 06:35:20 +0000
Subject: [PATCH] We now store the tags in two separate tables

---
 include/items.php                      | 45 ++++++++++++++--------
 src/Model/Tag.php                      | 52 ++++++++++++++++++++++++--
 src/Protocol/ActivityPub/Processor.php | 43 +++++++++------------
 src/Protocol/DFRN.php                  | 21 +----------
 src/Protocol/Diaspora.php              | 15 +-------
 src/Protocol/Feed.php                  | 11 ++----
 src/Protocol/OStatus.php               |  7 +---
 static/dbstructure.config.php          | 25 +++++++++----
 8 files changed, 121 insertions(+), 98 deletions(-)

diff --git a/include/items.php b/include/items.php
index 6068be4b94..2d73308582 100644
--- a/include/items.php
+++ b/include/items.php
@@ -141,31 +141,44 @@ function query_page_info($url, $photo = "", $keywords = false, $keyword_blacklis
 	return $data;
 }
 
-function add_page_keywords($url, $photo = "", $keywords = false, $keyword_blacklist = "", $return_array = false)
+function add_page_keywords($url, $photo = "", $keywords = false, $keyword_blacklist = "")
 {
 	$data = query_page_info($url, $photo, $keywords, $keyword_blacklist);
+	if (empty($data['keywords']) || !is_array($data['keywords'])) {
+		return '';
+	}
 
 	$tags = "";
-	$taglist = [];
-	if (isset($data["keywords"]) && count($data["keywords"])) {
-		foreach ($data["keywords"] as $keyword) {
-			$hashtag = str_replace([" ", "+", "/", ".", "#", "'"],
-				["", "", "", "", "", ""], $keyword);
+	foreach ($data["keywords"] as $keyword) {
+		$hashtag = str_replace([" ", "+", "/", ".", "#", "'"],
+			["", "", "", "", "", ""], $keyword);
 
-			if ($tags != "") {
-				$tags .= ", ";
-			}
-
-			$tags .= "#[url=" . DI::baseUrl() . "/search?tag=" . $hashtag . "]" . $hashtag . "[/url]";
-			$taglist[] = $hashtag;
+		if ($tags != "") {
+			$tags .= ", ";
 		}
+
+		$tags .= "#[url=" . DI::baseUrl() . "/search?tag=" . $hashtag . "]" . $hashtag . "[/url]";
 	}
 
-	if ($return_array) {
-		return $taglist;
-	} else {
-		return $tags;
+	return $tags;
+}
+
+function get_page_keywords($url, $photo = "", $keywords = false, $keyword_blacklist = "")
+{
+	$data = query_page_info($url, $photo, $keywords, $keyword_blacklist);
+	if (empty($data['keywords']) || !is_array($data['keywords'])) {
+		return [];
 	}
+
+	$taglist = [];
+	foreach ($data['keywords'] as $keyword) {
+		$hashtag = str_replace([" ", "+", "/", ".", "#", "'"],
+			["", "", "", "", "", ""], $keyword);
+
+		$taglist[] = $hashtag;
+	}
+
+	return $taglist;
 }
 
 function add_page_info($url, $no_photos = false, $photo = "", $keywords = false, $keyword_blacklist = "")
diff --git a/src/Model/Tag.php b/src/Model/Tag.php
index 7578372a9b..43885e493a 100644
--- a/src/Model/Tag.php
+++ b/src/Model/Tag.php
@@ -54,6 +54,54 @@ class Tag
 		self::EXCLUSIVE_MENTION => '!',
 	];
 
+	public static function store(int $uriid, int $type, string $name, string $url = '')
+	{
+		$name = trim($name, "\x00..\x20\xFF#!@");
+		if (empty($name)) {
+			return;
+		}
+
+		$fields = ['name' => substr($name, 0, 64), 'type' => $type];
+
+		if (!empty($url) && ($url != $name)) {
+			$fields['url'] = strtolower($url);
+		}
+
+		$tag = DBA::selectFirst('tag', ['id'], $fields);
+		if (!DBA::isResult($tag)) {
+			DBA::insert('tag', $fields, true);
+			$tagid = DBA::lastInsertId();
+		} else {
+			$tagid = $tag['id'];
+		}
+
+		if (empty($tagid)) {
+			Logger::error('No tag id created', $fields);
+			return;
+		}
+
+		DBA::insert('post-tag', ['uri-id' => $uriid, 'tid' => $tagid], true);
+
+		Logger::info('Stored tag/mention', ['uri-id' => $uriid, 'tag-id' => $tagid, 'tag' => $fields]);
+	}
+
+	public static function storeByHash(int $uriid, string $hash, string $name, string $url = '')
+	{
+		if ($hash == self::TAG_CHARACTER[self::MENTION]) {
+			$type = self::MENTION;
+		} elseif ($hash == self::TAG_CHARACTER[self::EXCLUSIVE_MENTION]) {
+			$type = self::EXCLUSIVE_MENTION;
+		} elseif ($hash == self::TAG_CHARACTER[self::IMPLICIT_MENTION]) {
+			$type = self::IMPLICIT_MENTION;
+		} elseif ($hash == self::TAG_CHARACTER[self::HASHTAG]) {
+			$type = self::HASHTAG;
+		} else {
+			return;
+		}
+
+		self::store($uriid, $type, $name, $url);
+	}
+
 	/**
 	 * Store tags from the body
 	 *
@@ -73,9 +121,7 @@ class Tag
 				continue;
 			}
 
-			$fields = ['uri-id' => $uriid, 'name' => substr($tag, 1, 64), 'type' => self::HASHTAG];
-			DBA::insert('tag', $fields, true);
-			Logger::info('Stored tag', ['uriid' => $uriid, 'tag' => $tag, 'fields' => $fields]);
+			self::storeByHash($uriid, '#', $tag);
 		}
 	}
 }
diff --git a/src/Protocol/ActivityPub/Processor.php b/src/Protocol/ActivityPub/Processor.php
index 023100dc72..1b605c30ab 100644
--- a/src/Protocol/ActivityPub/Processor.php
+++ b/src/Protocol/ActivityPub/Processor.php
@@ -34,6 +34,7 @@ use Friendica\Model\Event;
 use Friendica\Model\Item;
 use Friendica\Model\ItemURI;
 use Friendica\Model\Mail;
+use Friendica\Model\Tag;
 use Friendica\Model\Term;
 use Friendica\Model\User;
 use Friendica\Protocol\Activity;
@@ -585,53 +586,43 @@ class Processor
 	private static function storeTags(int $uriid, array $tags = null)
 	{
 		// Make sure to delete all existing tags (can happen when called via the update functionality)
-		DBA::delete('tag', ['uri-id' => $uriid]);
+		DBA::delete('post-tag', ['uri-id' => $uriid]);
 
 		foreach ($tags as $tag) {
 			if (empty($tag['name']) || empty($tag['type']) || !in_array($tag['type'], ['Mention', 'Hashtag'])) {
 				continue;
 			}
 
-			$fields = ['uri-id' => $uriid, 'name' => $tag['name']];
+			$hash = substr($tag['name'], 0, 1);
 
 			if ($tag['type'] == 'Mention') {
-				$fields['type'] = Term::MENTION;
-
-				if (substr($fields['name'], 0, 1) == Term::TAG_CHARACTER[Term::MENTION]) {
-					$fields['name'] = substr($fields['name'], 1);
-				} elseif (substr($fields['name'], 0, 1) == Term::TAG_CHARACTER[Term::EXCLUSIVE_MENTION]) {
-					$fields['type'] = Term::EXCLUSIVE_MENTION;
-					$fields['name'] = substr($fields['name'], 1);
-				} elseif (substr($fields['name'], 0, 1) == Term::TAG_CHARACTER[Term::IMPLICIT_MENTION]) {
-					$fields['type'] = Term::IMPLICIT_MENTION;
-					$fields['name'] = substr($fields['name'], 1);
+				if (in_array($hash, [Tag::TAG_CHARACTER[Tag::MENTION],
+					Tag::TAG_CHARACTER[Tag::EXCLUSIVE_MENTION],
+					Tag::TAG_CHARACTER[Tag::IMPLICIT_MENTION]])) {
+					$tag['name'] = substr($tag['name'], 1);
+				} else {
+					$hash = '#';
 				}
+
 				if (!empty($tag['href'])) {
 					$apcontact = APContact::getByURL($tag['href']);
 					if (!empty($apcontact['name']) || !empty($apcontact['nick'])) {
-						$fields['name'] = $apcontact['name'] ?: $apcontact['nick'];
+						$tag['name'] = $apcontact['name'] ?: $apcontact['nick'];
 					}
 				}
 			} elseif ($tag['type'] == 'Hashtag') {
-				$fields['type'] = Term::HASHTAG;
-				if (substr($fields['name'], 0, 1) == Term::TAG_CHARACTER[Term::HASHTAG]) {
-					$fields['name'] = substr($fields['name'], 1);
+				if (substr($tag['name'], 0, 1) == Term::TAG_CHARACTER[Term::HASHTAG]) {
+					$tag['name'] = substr($tag['name'], 1);
+				} else {
+					$hash = '@';
 				}
 			}
 
-			if (empty($fields['name'])) {
+			if (empty($tag['name'])) {
 				continue;
-			} else {
-				$fields['name'] = substr($fields['name'], 0, 64);
 			}
 			
-			if (!empty($tag['href'] && ($tag['href'] != $tag['name']))) {
-				$fields['url'] = $tag['href'];
-			}
-
-			DBA::insert('tag', $fields, true);
-
-			Logger::info('Stored tag/mention', ['uriid' => $uriid, 'tag' => $tag, 'fields' => $fields]);
+			Tag::storeByHash($uriid, $hash, $tag['name'], $tag['href']);
 		}
 	}
 
diff --git a/src/Protocol/DFRN.php b/src/Protocol/DFRN.php
index 9ab8bc70c9..32067613f1 100644
--- a/src/Protocol/DFRN.php
+++ b/src/Protocol/DFRN.php
@@ -2464,26 +2464,7 @@ class DFRN
 
 						$item["tag"] .= $termhash . "[url=" . $termurl . "]" . $term . "[/url]";
 
-						// Store the hashtag/mention
-						$fields = ['uri-id' => $item['uri-id'], 'name' => substr($term, 0, 64)];
-
-						if ($termhash == Term::TAG_CHARACTER[Term::MENTION]) {
-							$fields['type'] = Term::MENTION;
-						} elseif ($termhash == Term::TAG_CHARACTER[Term::EXCLUSIVE_MENTION]) {
-							$fields['type'] = Term::EXCLUSIVE_MENTION;
-						} elseif ($termhash == Term::TAG_CHARACTER[Term::IMPLICIT_MENTION]) {
-							$fields['type'] = Term::IMPLICIT_MENTION;
-						} elseif ($termhash == Term::TAG_CHARACTER[Term::HASHTAG]) {
-							$fields['type'] = Term::HASHTAG;
-						}
-
-						if (!empty($termurl)) {
-							$fields['url'] = $termurl;
-						}
-
-						DBA::insert('tag', $fields, true);
-
-						Logger::info('Stored tag/mention', ['uri-id' => $item['uri-id'], 'tag' => $term, 'url' => $termurl, 'hash' => $termhash, 'fields' => $fields]);
+						Tag::storeByHash($item['uri-id'], $termhash, $term, $termurl);
 					}
 				}
 			}
diff --git a/src/Protocol/Diaspora.php b/src/Protocol/Diaspora.php
index 5580234c64..20a2be4cc1 100644
--- a/src/Protocol/Diaspora.php
+++ b/src/Protocol/Diaspora.php
@@ -1840,20 +1840,7 @@ class Diaspora
 				continue;
 			}
 
-			$fields = ['uri-id' => $uriid, 'name' => substr($person['name'] ?: $person['nick'], 0, 64), 'url' => $person['url']];
-
-			if ($match[1] == Term::TAG_CHARACTER[Term::MENTION]) {
-				$fields['type'] = Term::MENTION;
-			} elseif ($match[1] == Term::TAG_CHARACTER[Term::EXCLUSIVE_MENTION]) {
-				$fields['type'] = Term::EXCLUSIVE_MENTION;
-			} elseif ($match[1] == Term::TAG_CHARACTER[Term::IMPLICIT_MENTION]) {
-				$fields['type'] = Term::IMPLICIT_MENTION;
-			} else {
-				continue;
-			}
-
-			DBA::insert('tag', $fields, true);
-			Logger::info('Stored mention', ['uriid' => $uriid, 'match' => $match, 'fields' => $fields]);
+			Tag::storeByHash($uriid, $match[1], $person['name'] ?: $person['nick'], $person['url']);
 		}
 	}
 
diff --git a/src/Protocol/Feed.php b/src/Protocol/Feed.php
index 8171348409..c03f959865 100644
--- a/src/Protocol/Feed.php
+++ b/src/Protocol/Feed.php
@@ -29,7 +29,7 @@ use Friendica\Core\Protocol;
 use Friendica\Database\DBA;
 use Friendica\DI;
 use Friendica\Model\Item;
-use Friendica\Model\Term;
+use Friendica\Model\Tag;
 use Friendica\Util\Network;
 use Friendica\Util\ParseUrl;
 use Friendica\Util\XML;
@@ -478,7 +478,7 @@ class Feed {
 				$item["title"] = "";
 				$item["body"] = $item["body"] . add_page_info($item["plink"], false, $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 				$item["tag"] = add_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
-				$taglist = add_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"], true);
+				$taglist = get_page_keywords($item["plink"], $preview, ($contact["fetch_further_information"] == 2), $contact["ffi_keyword_blacklist"]);
 				$item["object-type"] = Activity\ObjectType::BOOKMARK;
 				unset($item["attach"]);
 			} else {
@@ -492,7 +492,7 @@ class Feed {
 					} else {
 						// @todo $preview is never set in this case, is it intended? - @MrPetovan 2018-02-13
 						$item["tag"] = add_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"]);
-						$taglist = add_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"], true);
+						$taglist = get_page_keywords($item["plink"], $preview, true, $contact["ffi_keyword_blacklist"]);
 					}
 					$item["body"] .= "\n" . $item['tag'];
 				} else {
@@ -531,10 +531,7 @@ class Feed {
 				if (!empty($id) && !empty($taglist)) {
 					$feeditem = Item::selectFirst(['uri-id'], ['id' => $id]);
 					foreach ($taglist as $tag) {
-						$fields = ['uri-id' => $feeditem['uri-id'], 'name' => substr($tag, 0, 64), 'type' => Term::HASHTAG];
-						DBA::insert('tag', $fields, true);
-		
-						Logger::info('Stored tag', ['uri-id' => $feeditem['uri-id'], 'tag' => $tag, 'fields' => $fields]);
+						Tag::storeByHash($feeditem['uri-id'], '#', $tag);
 					}					
 				}
 			}
diff --git a/src/Protocol/OStatus.php b/src/Protocol/OStatus.php
index e155708aec..7dad68550b 100644
--- a/src/Protocol/OStatus.php
+++ b/src/Protocol/OStatus.php
@@ -36,7 +36,7 @@ use Friendica\Model\Conversation;
 use Friendica\Model\GContact;
 use Friendica\Model\Item;
 use Friendica\Model\ItemURI;
-use Friendica\Model\Term;
+use Friendica\Model\Tag;
 use Friendica\Model\User;
 use Friendica\Network\Probe;
 use Friendica\Util\DateTimeFormat;
@@ -665,10 +665,7 @@ class OStatus
 						$item['tag'] .= '#[url=' . DI::baseUrl() . '/search?tag=' . $term . ']' . $term . '[/url]';
 
 						// Store the hashtag
-						$fields = ['uri-id' => $item['uri-id'], 'name' => substr($term, 0, 64), 'type' => Term::HASHTAG];
-						DBA::insert('tag', $fields, true);
-
-						Logger::info('Stored tag', ['uri-id' => $item['uri-id'], 'tag' => $term, 'fields' => $fields]);
+						Tag::storeByHash($item['uri-id'], '#', $term);
 					}
 				}
 			}
diff --git a/static/dbstructure.config.php b/static/dbstructure.config.php
index 110842cbc0..204fca9bfe 100755
--- a/static/dbstructure.config.php
+++ b/static/dbstructure.config.php
@@ -1293,16 +1293,27 @@ return [
 		]
 	],
 	"tag" => [
-		"comment" => "item tags and mentions",
+		"comment" => "tags and mentions",
 		"fields" => [
-			"uri-id" => ["type" => "int unsigned", "not null" => "1", "primary" => "1", "relation" => ["item-uri" => "id"], "comment" => "Id of the item-uri table entry that contains the item uri"],
-			"type" => ["type" => "tinyint unsigned", "not null" => "1", "default" => "0", "primary" => "1", "comment" => ""],
-			"name" => ["type" => "varchar(64)", "not null" => "1", "default" => "", "primary" => "1", "comment" => ""],
-			"url" => ["type" => "varchar(255)", "not null" => "1", "default" => "", "comment" => ""]
+			"id" => ["type" => "int unsigned", "not null" => "1", "extra" => "auto_increment", "primary" => "1", "comment" => ""],
+			"type" => ["type" => "tinyint unsigned", "not null" => "1", "default" => "0", "comment" => ""],
+			"name" => ["type" => "varchar(64)", "not null" => "1", "default" => "", "comment" => ""],
+			"url" => ["type" => "varbinary(255)", "not null" => "1", "default" => "", "comment" => ""]
 		],
 		"indexes" => [
-			"PRIMARY" => ["uri-id", "type", "name"],
-			"type_name" => ["type", "name"]
+			"PRIMARY" => ["id"],			
+			"type_name_url" => ["UNIQUE", "type", "name", "url"]
+		]
+	],
+	"post-tag" => [
+		"comment" => "post relation to tags",
+		"fields" => [
+			"tid" => ["type" => "int unsigned", "not null" => "1", "relation" => ["tag" => "id"], "primary" => "1", "comment" => ""],
+			"uri-id" => ["type" => "int unsigned", "not null" => "1", "primary" => "1", "relation" => ["item-uri" => "id"], "comment" => "Id of the item-uri table entry that contains the item uri"],
+		],
+		"indexes" => [
+			"PRIMARY" => ["tid", "uri-id"],
+			"uri-id" => ["uri-id"]
 		]
 	],
 	"thread" => [