SQLite How to Join to an extra Table - sqlite

Question

In this SQLite example, I am selecting rows from tables a, b, and c using the common column 'aa'.
SELECT
a.aa,
a.ab,
a.ac,
b.ba,
b.bb,
b.bc,
c.ca,
c.cb,
c.cc
FROM
a
INNER JOIN b ON b.ba = a.aa
INNER JOIN c ON c.ca = a.aa
WHERE a.ab = 'blahblah'
This works OK. Now, I need to add an extra table and an extra JOIN. Table 'd' has a column 'd.dc' that is common with table 'c' and its column 'c.cc'.
When the correct row is selected in 'c', I want to be able to read the data in 'd.dd'.
SELECT
a.aa,
a.ab,
a.ac,
b.ba,
b.bb,
b.bc,
c.ca,
c.cb,
c.cc,
d.dc,
d.dd
FROM
a
INNER JOIN b ON b.ba = a.aa
INNER JOIN c ON c.ca = a.aa
INNER JOIN d ON d.dc = c.cc
WHERE a.ab = 'blahblah'
This does not work OK. Please can you tell me how to correct it?
I have also tried
FOREIGN KEY (cc) REFERENCES d(dc)
in the table 'c' definition, but it makes no apparent difference.
Here are my table definitions:
CREATE TABLE `a` ( `aa` TEXT PRIMARY KEY UNIQUE, `ab` TEXT, `ac` TEXT );
CREATE TABLE `b` ( `ba` TEXT PRIMARY KEY UNIQUE, `bb` TEXT, `bc` TEXT );
CREATE TABLE `c` ( `ca` TEXT PRIMARY KEY UNIQUE, `cb` TEXT, `cc` TEXT, FOREIGN KEY (cc) REFERENCES d(dc) );
CREATE TABLE `d` ( `dc` TEXT PRIMARY KEY UNIQUE, `dd` TEXT );
The strange results that I got were rather hard to describe, but one thing I noticed was that the only few rows returned were where c.cc were all the same value, whereas in fact there should have many more rows, and c.cc should have had a variety of values.

Answer 1

Perhaps the following will explain how to JOIN using the SQL (but this is really just tailored to meet the rules):-
DROP TABLE IF EXISTS a;
DROP TABLE IF EXISTS b;
DROP TABLE IF EXISTS c;
DROP TABLE IF EXISTS d;
CREATE TABLE `a` ( `aa` TEXT PRIMARY KEY UNIQUE, `ab` TEXT, `ac` TEXT );
CREATE TABLE `b` ( `ba` TEXT PRIMARY KEY UNIQUE, `bb` TEXT, `bc` TEXT );
CREATE TABLE `c` ( `ca` TEXT PRIMARY KEY UNIQUE, `cb` TEXT, `cc` TEXT, FOREIGN KEY (cc) REFERENCES d(dc) );
CREATE TABLE `d` ( `dc` TEXT PRIMARY KEY UNIQUE, `dd` TEXT );
INSERT INTO d VALUES('blah_c_cc_001','blahblah_d_dd'); -- MUST BE INSERTED BEFORE C else FK CONFLICT
INSERT INTO a VALUES('blah_a_aa_001','blahblah','blahblah_a_ac');
INSERT INTO b VALUES('blah_a_aa_001','blahblah_b_bb','blahblah_b_bc');
INSERT INTO c VALUES('blah_a_aa_001','blahblah_c_cb','blah_c_cc_001');
INSERT INTO d VALUES('blah_c_cc_002','blahblah_d_dd'); -- MUST BE INSERTED BEFORE C else FK CONFLICT
INSERT INTO a VALUES('blah_a_aa_002','blahblah','blahblah_a_ac');
INSERT INTO b VALUES('blah_a_aa_002','blahblah_b_bb','blahblah_b_bc');
INSERT INTO c VALUES('blah_a_aa_002','blahblah_c_cb','blah_c_cc_002');
INSERT INTO d VALUES('blah_c_cc_004','blahblah_d_dd');
INSERT INTO a VALUES('blah_a_aa_003','blahblah','blahblah_a_ac');
INSERT INTO b VALUES('blah_a_aa_003','blahblah_b_bb','blahblah_b_bc');
INSERT INTO c VALUES('blah_a_aa_003','blahblah_c_cb','blah_c_cc_004');
SELECT *
FROM
a
INNER JOIN b ON b.ba = a.aa
INNER JOIN c ON c.ca = a.aa
INNER JOIN d ON d.dc = c.cc
WHERE a.ab = 'blahblah'
;
- * used for brevity
This results in :-
Basically the rule is that for a INNER (normal/simple) JOIN there must be matched rows, so in your query (the following applies)
TABLE b must have a value in column ba that matches the aa column in table a as well as the matching row in table a having a value of blahblah in column ab
AND
Table c must have a value in column ca that matches the aa column in table a as well as the matching row in table a having a value of blahblah
AND
Table d must have a value in column dc that matches the cc column in table c AND that the matching row in cc is a row that matches a row in table a that has a value of blahblah in column ab
The FOREIGN KEY has no impact on the SELECT query, other than it restricting the insertion of a row in table c in that the cc column being inserted must match the value of column dc in one of the rows in table d.

Answer 2

OK, I found my problem. The original code was OK, but the problem was in the external data that I was using. It wasn't matching because I had not bothered to TRIM it before loading it to my database.
My strange results occurred because if Table 'd' did not have a value in column dc that matches the cc column in table c, then no row at all was returned. I had assumed that if Table 'd' did not have a value in column dc that matches the cc column in table c, then the SELECT d.dd would just return an empty result, whereas in fact it blocked all output.
Message to Self: Always TRIM your data. Also, when monitoring results, always surround with quote-marks, so you can see the extra whitespace.