Joakim Brännström (joakim.brannstrom@gmx.com)

This module contains an AST representation of a database query. The intention is to encode the structure in the types so errors in the SQL statement are detected at compile time.

Grammar

The grammar is expressed in PEG form using the same terminology as the dub package Pegged. It has not been verified to be correct so may contain errors.

This is simplified to only contain those parts of the grammar that is needed to pretty print constructed SQL queries. It is not intended to be a grammar that can correctly parse SQL. Thus it contains "blobs of data" that the grammar do not care about. The important part is to keep the structure. Where to insert "blobs".

A secondary purpose is to give some compile time guarantees of the constructed SQL queries. If it compiles it is reasonably assured to be correct.

When not necessary for pretty printing as a valid SQL blanks and such are skipped in the grammar definition.

1 SQL         <- blank* Query (spacing / eoi)
2 
3 Query       <- Select
4 
5 # --- SELECT START ---
6 Select      <- "SELECT" :blank+ ResultColumns :blank+ From? Where? GroupBy? (Window :blank+)? OrderBy? Values? Limit?
7 
8 ResultColumns       <- ResultColumn ("," ResultColumn)*
9 ResultColumn        <- Star / ResultColumnExpr
10 ResultColumnExpr    <- Query / Blob
11 
12 From    <- :blank+ "FROM" :blank+ (TableOrSubQueries / Blob)
13 Where   <- :blank+ "WHERE" :blank+ WhereExpr*
14 GroupBy <- :blank+ "GROUP BY" :blank+ Expr ("," Expr)* (:blank+ "HAVING" Expr)?
15 Window  <- Blob
16 
17 WhereExpr   <- Expr (:blank+ WhereOp :blank+ Expr)?
18 WhereOp     <- "AND" / "OR"
19 
20 OrderBy         <- :blank+ "ORDER BY" :blank+ OrderingTerm ("," OrderingTerm)
21 OrderingTerm    <- Expr :blank+ OrderingTermSort?
22 OrderingTermSort <- "ASC" / "DESC" / ""
23 
24 Limit <- "LIMIT" :blank+ Expr :blank+ (("OFFSET" :blank+ Expr) / ("," Expr))?
25 
26 TableOrSubQueries       <- TableOrQuery ("," TableOrSubQuery)*
27 TableOrSubQuery         <- TableOrSubQuerySelect / ("(" TableOrSubQueries ")") / (TableRef Blob*) / Blob
28 TableOrSubQuerySelect   <- "(" Select ")" TableAlias?
29 
30 # --- SELECT END ---
31 
32 # --- INSERT START ---
33 Insert          <- InsertOpt :blank+ "INTO" :blank+ TableRef TableAlias? InsertColumns? InsertValues
34 InsertOpt       <- "INSERT" / "REPLACE" / "INSERT OR REPLACE" / "INSERT OR ROLLBACK" / "INSERT OR ABORT" / "INSERT OR FAIL" / "INSERT OR IGNORE"
35 InsertColumns    <- :blank+ "(" ColumnName ("," ColumnName)* ")"
36 InsertValues    <- :blank+ (Values / Select / "DEFAULT VALUES")
37 ColumnName      <- identifier
38 
39 # --- INSERT END ---
40 
41 # --- DELETE START ---
42 Delete          <- "DELETE FROM" :blank+ TableRef Where?
43 # --- DELETE END ---
44 
45 # Reference an existing table
46 TableRef            <- SchemaName? TableName TableAlias?
47 
48 Values  <- :blank+ "VALUES" "(" Value ")" ("(" Value ")")*
49 Value   <- Expr ("," Expr)*
50 
51 TableAlias  <- :blank+ "AS" :blank+ identifier
52 
53 Expr        <- Blob
54 # Not representable in the grammar because it can be anything without a formal
55 # terminator. Its purpose is to be an injection point of user data.
56 Blob        <- ""
57 
58 SchemaName  <- identifier "."
59 TableName   <- identifier
60 Star        <- "*"

Grammar Encoding

* SumType is used when possible. * None is the first member of a SumType when the node is optional. * Literals are used as is. * All nodes have a toString.