1 IN-Operator Implementation Notes
2 ================================
6 An IN operator has one of the following formats:
12 The "x" is referred to as the LHS (left-hand side). The list or subquery
13 on the right is called the RHS (right-hand side). If the RHS is a list
14 it must be a non-empty list. But if the RHS is a subquery, it can be an
17 The LHS can be a scalar (a single quantity) or a vector (a list of
18 two or or more values) or a subquery that returns one or more columns.
19 We use the term "vector" to mean an actually list of values or a
20 subquery that returns two or more columns. An isolated value or
21 a subquery that returns a single columns is called a scalar.
23 The RHS can be a subquery that returns a single column, a subquery
24 that returns two or more columns, or a list of scalars. It is not
25 currently support for the RHS to be a list of vectors.
27 The number of columns for LHS must match the number of columns for
28 the RHS. If the RHS is a list of values, then the LHS must be a
29 scalar. If the RHS is a subquery returning N columns, then the LHS
30 must be a vector of size N.
32 NULL values can occur in either or both of the LHS and RHS.
33 If the LHS contains only
34 NULL values then we say that it is a "total-NULL". If the LHS contains
35 some NULL values and some non-NULL values, then it is a "partial-NULL".
36 For a scalar, there is no difference between a partial-NULL and a total-NULL.
37 The RHS is a partial-NULL if any row contains a NULL value. The RHS is
38 a total-NULL if it contains one or more rows that contain only NULL values.
39 The LHS is called "non-NULL" if it contains no NULL values. The RHS is
40 called "non-NULL" if it contains no NULL values in any row.
42 The result of an IN operator is one of TRUE, FALSE, or NULL. A NULL result
43 means that it cannot be determined if the LHS is contained in the RHS due
44 to the presence of NULL values. In some contexts (for example, when the IN
45 operator occurs in a WHERE clause)
46 the system only needs a binary result: TRUE or NOT-TRUE. One can also
47 to define a binary result of FALSE and NOT-FALSE, but
48 it turns out that no extra optimizations are possible in that case, so if
49 the FALSE/NOT-FALSE binary is needed, we have to compute the three-state
50 TRUE/FALSE/NULL result and then combine the TRUE and NULL values into
53 A "NOT IN" operator is computed by first computing the equivalent IN
54 operator, then interchanging the TRUE and FALSE results.
56 ## Simple Full-Scan Algorithm
58 The following algorithm always compute the correct answer. However, this
59 algorithm is suboptimal, especially if there are many rows on the RHS.
61 1. Set the null-flag to false
62 2. For each row in the RHS:
64 <li> Compare the LHS against the RHS
65 <li> If the LHS exactly matches the RHS, immediately return TRUE
66 <li> If the comparison result is NULL, set the null-flag to true
68 3. If the null-flag is true, return NULL.
71 ## Optimized Algorithm
73 The following procedure computes the same answer as the simple full-scan
74 algorithm, though it does so with less work in the common case. This
75 is the algorithm that is implemented in SQLite.
77 1. If the RHS is a constant list of length 1 or 2, then rewrite the
78 IN operator as a simple expression. Implement
86 This is the INDEX_NOOP optimization and is only undertaken if the
87 IN operator is used for membership testing. If the IN operator is
88 driving a loop, then skip this step entirely.
90 2. Check the LHS to see if it is a partial-NULL and if it is, jump
93 3. Do a binary search of the RHS using the LHS as a probe. If
94 an exact match is found, return TRUE.
96 4. If the RHS is non-NULL then return FALSE.
98 5. If we do not need to distinguish between FALSE and NULL,
101 6. For each row in the RHS, compare that row against the LHS and
102 if the result is NULL, immediately return NULL. In the case
103 of a scalar IN operator, we only need to look at the very first
104 row the RHS because for a scalar RHS, all NULLs will always come
105 first. If the RHS is empty, this step is a no-op.