1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
|
_**Date:** 25th Sept. 2019_
_**Author:** Grégory Mantelet_
# Identifiers' rules in ADQL-Lib
## Definitions
These rules apply to any kind of identifier (e.g. table, column).
To simplify explanations, we will consider here that an identifier is composed of 3 pieces of information:
- adqlName _- string_
- adqlCaseSensitive _- boolean_
- dbName _- string_
`adqlName` and `dbName` are never stored in their qualified (e.g. the table prefix inside a full column name) or their delimited (e.g. `"aTable"` is the delimited form of `aTable`) form ; surrounded double quotes (and escaped double quotes: `""`) and prefixes are used/checked at initialization but always discarded for storage.
If `dbName` is not specified, it is the same as `adqlName`.
`adqlCaseSensitive` is set to `true` only if `adqlName` was delimited at initialization.
*Rules about how to build this 3-tuple depend on the origin of the identifier (e.g. `TAP_SCHEMA`, subquery, CTE) ; these rules are detailed below.*
Let's now see how to write ADQL and SQL queries with a such 3-tuple...
## Identifiers in ADQL
In ADQL queries, an identifier MUST be delimited only in the following cases:
* if a ADQL/SQL reserved keyword
```sql
-- declared table identifier: adqlName=`distance`
SELECT ... FROM distance -- INCORRECT because `distance` is a reserved keyword
SELECT ... FROM "distance" -- CORRECT
```
* if not a regular ADQL identifier
```sql
-- declared table identifier: adqlName=`2do`
SELECT ... FROM 2do -- INCORRECT because `2do` is starting with a digit
SELECT ... FROM "2do" -- CORRECT
```
* if ambiguous with another identifier of the same type
```sql
-- declared column identifiers: adqlName=`id` in `table1` and adqlName=`id` in `table2`
SELECT id FROM table1, table2 -- INCORRECT because the column `id` exists in both tables
SELECT table1.id FROM table1, table2 -- CORRECT
```
In any other case, the identifier _MAY_ be delimited, but if not, you are free to write it in upper-/lower-/mixed-case.
If the identifier is declared in a *CASE-SENSITIVE* way, it MUST be respected when delimited in the ADQL query.
If the identifier is *CASE-INSENSITIVE*, its delimited ADQL version MUST be all in lower-case.
Then, the following ADQL queries are perfectly allowed:
```sql
-- declared table: adqlName=`aTable`, adqlCaseSensitive=`false`
SELECT ... FROM atable -- OK
SELECT ... FROM ATABLE -- OK
SELECT ... FROM "atable" -- OK (because lower-case if not declared CS)
SELECT ... FROM "aTable" -- INCORRECT
-- declared table: adqlName=`aTable`, adqlCaseSensitive=`true`
SELECT ... FROM atable -- OK
SELECT ... FROM ATABLE -- OK
SELECT ... FROM "atable" -- INCORRECT
SELECT ... FROM "aTable" -- OK
```
## SQL translation
_In this part, we will consider PostgreSQL as SQL target._
The `dbName` of an identifier is _always_ considered as _case-sensitive_. So, it will _always_ be written delimited in SQL queries.
**Examples of SQL queries:**
```sql
-- declared table: adqlName=`aTable`, dbName=-
-- ADQL: SELECT ... FROM atable
SELECT ... FROM "aTable"
-- declare table: adqlName=`aTable`, dbName=`DBTable`
-- ADQL: SELECT ... FROM atable
SELECT ... FROM "DBTable"
```
## Automatic column aliases in SQL
To ensure having the expected labels in SQL query results, aliases are automatically added (if none is specified) to all items of the `SELECT` clause.
As `dbName`s, these default aliases are considered as case sensitive.
They are built using the `adqlName` of the aliased identifiers. If this name is _not case-sensitive_, the alias will be written in lower-case. But if _case-sensitive_, it is written exactly the same.
**Examples of SQL queries:**
```sql
-- declared table: adqlName=`aTable`, dbName=`DBTable`
-- declared columns in `aTable`:
-- * adqlName=`ColCS`, adqlCaseSensitive=`false`, dbName=`dbCol1`
-- * adqlName=`ColCI`, adqlCaseSensitive=`true`, dbName=`dbCol2`
-- ADQL: SELECT colcs, colci FROM atable
SELECT "dbCol1" AS "colcs", "dbCol2" AS "ColCI" FROM "DBTable"
```
## Schemas/Tables/Columns declared in `TAP_SCHEMA`
* `adqlName` = `TAP_SCHEMA.(schemas.schema_name|tables.table_name|columns.column_name)`
* `dbName` = `TAP_SCHEMA.(schemas|tables|columns).dbName` or if NULL `adqlName`
**Examples with `TAP_SCHEMA.tables`:**
| In TAP_SCHEMA.tables | In ADQL-Lib |
| ---------------------------------------------- | ------------------------------------------------------------ |
| table_name=`aTable`, dbName=_null_ | adqlName=`aTable`, adqlCaseSensitive=`false`, dbName=_null_ |
| table_name=`schema.aTable`, dbName=_null_ | adqlName=`aTable`, adqlCaseSensitive=`false`, dbName=_null_ |
| table_name=`"aTable"`, dbName=_null_ | adqlName=`aTable`, adqlCaseSensitive=`true`, dbName=_null_ |
| table_name=`schema."aTable"`, dbName=_null_ | adqlName=`aTable`, adqlCaseSensitive=`true`, dbName=_null_ |
| table_name=`aTable`, dbName=`DBTable` | adqlName=`aTable`, adqlCaseSensitive=`false`, dbName=`DBTable` |
| table_name=`aTable`, dbName=`"DBTable"` | adqlName=`aTable`, adqlCaseSensitive=`false`, dbName=`DBTable` |
| table_name=`aTable`, dbName=`schema.DBTable` | adqlName=`aTable`, adqlCaseSensitive=`false`, dbName=`schema.DBTable` |
| table_name=`aTable`, dbName=`schema."DBTable"` | adqlName=`aTable`, adqlCaseSensitive=`false`, dbName=`schema."DBTable"` |
## Tables from subqueries (i.e. `FROM` and `WITH`)
_Reminder: in ADQL, a subquery declared as table in the `FROM` clause and a CTE declared in the `WITH` clause MUST always be aliased/named._
* `adqlName` = _subquery's alias/CTE's name_
* `dbName` = `adqlName`
If the alias/name is *delimited* (i.e. case sensitive), `adqlCaseSensitive` will be set to `true` and surrounding double quotes are removed from `adqlName` .
If the alias/name is *not delimited*, `adqlName` is set to the alias put into lower-case, and `adqlCaseSensitive` is `false`.
**Examples:**
```sql
--
-- Subqueries in FROM:
--
SELECT ... FROM (SELECT * FROM atable) AS t1
SELECT ... FROM (SELECT * FROM atable) AS T1
-- => adqlName=`t1`, adqlCaseSensitive=`false`, dbName=`t1`
SELECT ... FROM (SELECT * FROM atable) AS "T1"
-- => adqlName=`T1`, adqlCaseSensitive=`true`, dbName=`T1`
--
-- CTEs in WITH:
--
WITH t1 AS (SELECT * FROM atable) SELECT ... FROM t1
WITH T1 AS (SELECT * FROM atable) SELECT ... FROM t1
-- => adqlName=`t1`, adqlCaseSensitive=`false`, dbName=`t1`
WITH "T1" AS (SELECT * FROM atable) SELECT ... FROM t1
-- => adqlName=`T1`, adqlCaseSensitive=`true`, dbName=`T1`
```
## Columns of a (sub)query
* If _NOT aliased_:
* `adqlName` = _original's `adqlName`_
* `adqlCaseSensitive` = _original's `adqlCaseSensitive`_
* `dbName` = _original's `dbName`_
* If _aliased_:
* `adqlName` = _alias in lower-case if not delimited, exact same alias otherwise_
* `adqlCaseSensitive` = `true` _if alias is delimited_, `false` _otherwise_
* `dbName` = `adqlName`
**Examples:**
```sql
-- declared column: adqlName=`aColumn`, adqlCaseSensitive=`false`, dbName=`DBCol`
SELECT acolumn FROM atable
SELECT AColumn FROM atable
-- => the declared column
-- declared column: adqlName=`aColumn`, adqlCaseSensitive=`true`, dbName=`DBCol`
SELECT acolumn FROM atable
SELECT AColumn FROM atable
-- => the declared column
-- declared column: adqlName=`aColumn`, adqlCaseSensitive=`true`, dbName=`DBCol`
SELECT acolumn AS myColumn FROM atable
-- => adqlName=`mycolumn`, adqlCaseSensitive=`false`, dbName=`mycolumn`
SELECT acolumn AS "myColumn" FROM atable
-- => adqlName=`myColumn`, adqlCaseSensitive=`true`, dbName=`myColumn`
```
_**Note:** The new representation of an aliased column has different ADQL and DB names, but the other metadata (e.g. datatype, UCD, ...) of the original column are copied as such._
## Duplicated output columns
_The term 'output columns' means here the columns written in the output format (e.g. VOTable). They are not the columns represented as a 3-tuple in this document._
**TODO**
|
