Subqueries in SELECT clause - mysql

Question

I need to SELECT from multiple tables to get a result table like the following:
+--------+-------+-------------------+----------------------+
| itemID | level | studyPhraseString | meaningPhraseStrings |
+--------+-------+-------------------+----------------------+
| 1 | 4 | la maison | house |
+--------+-------+-------------------+----------------------+
| 2 | 3 | voler | to fly,to steal |
+--------+-------+-------------------+----------------------+
Note: studyPhraseString and meaningPhraseStrings should be concatenated strings made up of values from the word table.
My tables:
item
CREATE TABLE `item` (
`itemID` bigint(11) unsigned NOT NULL AUTO_INCREMENT,
`groupID` bigint(11) unsigned NOT NULL,
`studyLang` varchar(5) NOT NULL,
`meaningLang` varchar(5) NOT NULL,
`studyPhraseID` bigint(11) unsigned NOT NULL,
PRIMARY KEY (`itemID`)
) ENGINE=InnoDB AUTO_INCREMENT=14 DEFAULT CHARSET=latin1;
meaning
CREATE TABLE `meaning` (
`meaningID` bigint(11) unsigned NOT NULL AUTO_INCREMENT,
`itemID` bigint(11) unsigned NOT NULL,
`meaningPhraseID` bigint(11) unsigned NOT NULL,
`meaningIndex` int(11) unsigned NOT NULL,
PRIMARY KEY (`meaningID`)
) ENGINE=InnoDB AUTO_INCREMENT=16 DEFAULT CHARSET=latin1;
phrase
CREATE TABLE `phrase` (
`phraseID` bigint(11) unsigned NOT NULL AUTO_INCREMENT,
`phraseLang` varchar(5) NOT NULL DEFAULT '',
PRIMARY KEY (`phraseID`)
) ENGINE=InnoDB AUTO_INCREMENT=29 DEFAULT CHARSET=latin1;
phrase_word
CREATE TABLE `phrase_word` (
`phrase_wordID` bigint(11) unsigned NOT NULL AUTO_INCREMENT,
`phraseID` bigint(11) unsigned NOT NULL,
`wordID` bigint(11) unsigned NOT NULL,
`wordIndex` int(11) unsigned NOT NULL,
PRIMARY KEY (`phrase_wordID`)
) ENGINE=InnoDB AUTO_INCREMENT=29 DEFAULT CHARSET=latin1;
status
CREATE TABLE `status` (
`statusID` bigint(11) unsigned NOT NULL AUTO_INCREMENT,
`itemID` bigint(11) unsigned NOT NULL,
`level` tinyint(1) unsigned NOT NULL DEFAULT '0',
`nextReviewTime` int(11) unsigned NOT NULL DEFAULT '0',
PRIMARY KEY (`statusID`)
) ENGINE=InnoDB AUTO_INCREMENT=14 DEFAULT CHARSET=latin1;
word
CREATE TABLE `word` (
`wordID` bigint(11) unsigned NOT NULL AUTO_INCREMENT,
`string` varchar(64) NOT NULL DEFAULT '',
PRIMARY KEY (`wordID`)
) ENGINE=InnoDB AUTO_INCREMENT=9 DEFAULT CHARSET=latin1;
I have written the following SELECT statement:
SELECT item.itemID, status.level,
(SELECT GROUP_CONCAT(word.string ORDER BY phrase_word.wordIndex SEPARATOR ' ')
FROM word INNER JOIN phrase_word ON word.wordID=phrase_word.wordID
INNER JOIN item AS subItem ON phrase_word.phraseID=subItem.studyPhraseID
WHERE subItem.itemID=item.itemID
GROUP BY subItem.itemID
) AS studyPhraseString
FROM item INNER JOIN status ON item.itemID=status.itemID
WHERE item.groupID=5
ORDER BY status.statusID DESC
This works but does not include the meaningPhraseString. I can’t figure out how to concat the words into phrases AND concat the phrases into one string separated by ,
I have tried nested GROUP_CONCAT clauses with no success (subquery returns more than 1 row):
The question:
How should this statement be written to include meaningPhraseStrings? Thanks in advance.
PS: I'd like this to be a single query
I have tried the following but it fails. Why? It has two levels of correlated queries.
SELECT
item.itemID,
status.level,
(
SELECT GROUP_CONCAT(word.string ORDER BY phrase_word.wordIndex SEPARATOR ' ')
FROM word INNER JOIN phrase_word ON word.wordID=phrase_word.wordID
INNER JOIN item AS subItem ON phrase_word.phraseID=subItem.studyPhraseID
WHERE subItem.itemID=item.itemID
GROUP BY subItem.itemID
) AS studyPhraseString,
(
SELECT GROUP_CONCAT(meaningPhraseString SEPARATOR '.')
FROM (
(
SELECT GROUP_CONCAT(word.string ORDER BY phrase_word.wordIndex SEPARATOR ' ') AS meaningPhraseString
FROM word INNER JOIN phrase_word ON word.wordID=phrase_word.wordID
INNER JOIN meaning ON phrase_word.phraseID=meaning.meaningPhraseID
INNER JOIN item AS subItem ON meaning.itemID=subItem.itemID
WHERE subItem.itemID=item.itemID #This fails
GROUP BY meaning.meaningPhraseID
) AS meaningPhraseString
)
) AS meaningPhraseStrings
FROM item INNER JOIN status ON item.itemID=status.itemID
WHERE item.groupID=5
ORDER BY item.itemID DESC
Sample data:
INSERT INTO `status` VALUES (22,22,0,0),(23,23,0,0),(24,25,0,0),(25,24,0,0),(26,26,0,0);
INSERT INTO `item` VALUES (22,5,'fr','en',49),(23,5,'fr','en',48),(24,5,'fr','en',56),(25,5,'fr','en',50),(26,5,'fr','en',57);
INSERT INTO `meaning` VALUES (27,22,51,0),(28,23,52,0),(29,23,54,1),(30,24,59,0),(31,24,61,1),(32,25,53,0),(33,25,55,1),(34,26,58,0),(35,26,60,1);
INSERT INTO `phrase` VALUES (48,'fr'),(49,'fr'),(50,'fr'),(51,'en'),(52,'en'),(53,'en'),(54,'en'),(55,'en'),(56,'fr'),(57,'fr'),(58,'en'),(59,'en'),(60,'en'),(61,'en');
INSERT INTO `word` VALUES (46,'l\'autobus'),(47,'bus'),(48,'pourquoi'),(49,'comment'),(50,'why'),(51,'ça'),(52,'va?'),(53,'voler'),(54,'incroyable'),(55,'how'),(56,'is'),(57,'to'),(58,'are'),(59,'incredible'),(60,'that?'),(61,'you?'),(62,'fly'),(63,'amazing'),(64,'hi'),(65,'steal');
INSERT INTO `phrase_word` VALUES (86,49,46,0),(87,51,47,0),(88,48,48,0),(89,50,49,0),(90,52,50,0),(91,54,50,0),(92,50,51,1),(93,50,52,2),(94,57,53,0),(95,53,55,0),(96,56,54,0),(97,54,56,1),(98,53,58,1),(99,58,57,0),(100,59,59,0),(101,54,60,2),(102,53,61,2),(103,58,62,1),(104,61,63,0),(105,60,57,0),(106,55,64,0),(107,60,65,1);
Final answer:
SELECT i.itemID,
s.level,
sp.studyPhraseString,
GROUP_CONCAT(mp.meaningPhraseString
SEPARATOR ', ') AS meaningPhraseStrings
FROM item AS i
JOIN meaning AS m ON i.itemID = m.itemID
JOIN status AS s ON i.itemID = s.itemID
JOIN (
SELECT subItem.studyPhraseID,
GROUP_CONCAT(word.string
ORDER BY phrase_word.wordIndex
SEPARATOR ' ') AS studyPhraseString
FROM word
JOIN phrase_word
ON word.wordID=phrase_word.wordID
JOIN item AS subItem
ON phrase_word.phraseID=subItem.studyPhraseID
GROUP BY subItem.studyPhraseID
) AS sp ON i.studyPhraseID = sp.studyPhraseID
JOIN (
SELECT meaning.meaningPhraseID,
GROUP_CONCAT(word.string
ORDER BY phrase_word.wordIndex
SEPARATOR ' ') AS meaningPhraseString
FROM word
JOIN phrase_word ON word.wordID=phrase_word.wordID
JOIN meaning ON phrase_word.phraseID=meaning.meaningPhraseID
JOIN item AS subItem ON meaning.itemID=subItem.itemID
GROUP BY meaning.meaningPhraseID
) AS mp ON m.meaningPhraseID = mp.meaningPhraseID
GROUP BY i.itemID, s.level, sp.studyPhraseString
ORDER BY i.itemID, s.level, sp.studyPhraseString

Answer 1

Your question seems to be this:
how to concat the words into phrases AND concat the phrases into one string
Let's break it down. You need to join together five tables. Three of them are physical tables, namely item, meaning, and status. From those tables you get the references to the result set items you need called itemID and level, and you get the relationship between items and their meanings.
The other two tables you need are virtual tables (that is, subqueries). One of these gives you your French language phrases, and the other gives you your English-language translations.
Let us create the two queries for the virtual tables. Let's put the words into phrases first. A query like this achieves that goal.
SELECT subItem.studyPhraseID,
GROUP_CONCAT(word.string
ORDER BY phrase_word.wordIndex
SEPARATOR ' ') AS studyPhraseString
FROM word
INNER JOIN phrase_word ON word.wordID=phrase_word.wordID
INNER JOIN item AS subItem ON phrase_word.phraseID=subItem.studyPhraseID
GROUP BY subItem.studyPhraseID
This gives you a resultset table of phrase ID numbers and the phrases. Here's a SQL fiddle for this one based on your samples. http://sqlfiddle.com/#!2/11ae2/9/0
Then, create a similar query giving you the meaningPhraseString values.
SELECT meaning.meaningPhraseID,
GROUP_CONCAT(word.string
ORDER BY phrase_word.wordIndex
SEPARATOR ' ') AS meaningPhraseString
FROM word
INNER JOIN phrase_word ON word.wordID=phrase_word.wordID
INNER JOIN meaning ON phrase_word.phraseID=meaning.meaningPhraseID
INNER JOIN item AS subItem ON meaning.itemID=subItem.itemID
GROUP BY meaning.meaningPhraseID
This gives a list of ids and meaning phrases. Here's the fiddle. http://sqlfiddle.com/#!2/11ae2/6/0
So, we're going to need a five-way join (three physical tables and two subqueries) to get to our final result set. In summary, it will look like this:
SELECT i.itemID,
s.level,
sp.studyPhraseString,
mp.meaningPhraseString
FROM item AS i
JOIN meaning AS m ON i.itemID = m.itemID
JOIN status AS s ON i.itemID = s.itemID
JOIN (
/* the studyPhrase subquery */
) AS sp ON i.studyPhraseID = sp.studyPhraseID
JOIN (
/* the meaningPhrase subquery */
) AS mp ON m.meaningPhraseID = mp.meaningPhraseID
The trick here is that you can use a query (or virtual table) and a physical table interchangeably. So when you need to summarize a bunch of tables, you create a query to do that and then paste it into JOIN (/*query*/) AS alias.
Finally, you need to create the comma-joined strings (e.g. to fly, to steal) by adding yet another GROUP_CONCAT() and GROUP BY to your query. The end result is then
SELECT i.itemID,
s.level,
sp.studyPhraseString,
GROUP_CONCAT(mp.meaningPhraseString
SEPARATOR ', ') AS meaningPhraseStrings
FROM item AS i
JOIN meaning AS m ON i.itemID = m.itemID
JOIN status AS s ON i.itemID = s.itemID
JOIN (
/* the studyPhrase subquery */
) AS sp ON i.studyPhraseID = sp.studyPhraseID
JOIN (
/* the meaningPhrase subquery */
) AS mp ON m.meaningPhraseID = mp.meaningPhraseID
GROUP BY i.itemID, s.level, sp.studyPhraseString
ORDER BY i.itemID, s.level, sp.studyPhraseString
And that is your query. http://sqlfiddle.com/#!2/11ae2/16/0 It definitely takes advantage of the Structured in Structured Query Language.