1CREATE TYPE(7) PostgreSQL 12.2 Documentation CREATE TYPE(7)
2
6 CREATE_TYPE - define a new data type
7
9 CREATE TYPE name AS
10 ( [ attribute_name data_type [ COLLATE collation ] [, ... ] ] )
11 CREATE TYPE name AS ENUM
13 ( [ 'label' [, ... ] ] )
14 CREATE TYPE name AS RANGE (
16 SUBTYPE = subtype
17 [ , SUBTYPE_OPCLASS = subtype_operator_class ]
18 [ , COLLATION = collation ]
19 [ , CANONICAL = canonical_function ]
20 [ , SUBTYPE_DIFF = subtype_diff_function ]
21 )
22 CREATE TYPE name (
24 INPUT = input_function,
25 OUTPUT = output_function
26 [ , RECEIVE = receive_function ]
27 [ , SEND = send_function ]
28 [ , TYPMOD_IN = type_modifier_input_function ]
29 [ , TYPMOD_OUT = type_modifier_output_function ]
30 [ , ANALYZE = analyze_function ]
31 [ , INTERNALLENGTH = { internallength | VARIABLE } ]
32 [ , PASSEDBYVALUE ]
33 [ , ALIGNMENT = alignment ]
34 [ , STORAGE = storage ]
35 [ , LIKE = like_type ]
36 [ , CATEGORY = category ]
37 [ , PREFERRED = preferred ]
38 [ , DEFAULT = default ]
39 [ , ELEMENT = element ]
40 [ , DELIMITER = delimiter ]
41 [ , COLLATABLE = collatable ]
42 )
43 CREATE TYPE name
45
47 CREATE TYPE registers a new data type for use in the current database.
48 The user who defines a type becomes its owner.
49 If a schema name is given then the type is created in the specified
51 schema. Otherwise it is created in the current schema. The type name
52 must be distinct from the name of any existing type or domain in the
53 same schema. (Because tables have associated data types, the type name
54 must also be distinct from the name of any existing table in the same
55 schema.)
56 There are five forms of CREATE TYPE, as shown in the syntax synopsis
58 above. They respectively create a composite type, an enum type, a range
59 type, a base type, or a shell type. The first four of these are
60 discussed in turn below. A shell type is simply a placeholder for a
61 type to be defined later; it is created by issuing CREATE TYPE with no
62 parameters except for the type name. Shell types are needed as forward
63 references when creating range types and base types, as discussed in
64 those sections.
65 Composite Types
67 The first form of CREATE TYPE creates a composite type. The composite
68 type is specified by a list of attribute names and data types. An
69 attribute's collation can be specified too, if its data type is
70 collatable. A composite type is essentially the same as the row type of
71 a table, but using CREATE TYPE avoids the need to create an actual
72 table when all that is wanted is to define a type. A stand-alone
73 composite type is useful, for example, as the argument or return type
74 of a function.
75 To be able to create a composite type, you must have USAGE privilege on
77 all attribute types.
78 Enumerated Types
80 The second form of CREATE TYPE creates an enumerated (enum) type, as
81 described in Section 8.7. Enum types take a list of quoted labels, each
82 of which must be less than NAMEDATALEN bytes long (64 bytes in a
83 standard PostgreSQL build). (It is possible to create an enumerated
84 type with zero labels, but such a type cannot be used to hold values
85 before at least one label is added using ALTER TYPE (ALTER_TYPE(7)).)
86 Range Types
88 The third form of CREATE TYPE creates a new range type, as described in
89 Section 8.17.
90 The range type's subtype can be any type with an associated b-tree
92 operator class (to determine the ordering of values for the range
93 type). Normally the subtype's default b-tree operator class is used to
94 determine ordering; to use a non-default operator class, specify its
95 name with subtype_opclass. If the subtype is collatable, and you want
96 to use a non-default collation in the range's ordering, specify the
97 desired collation with the collation option.
98 The optional canonical function must take one argument of the range
100 type being defined, and return a value of the same type. This is used
101 to convert range values to a canonical form, when applicable. See
102 Section 8.17.8 for more information. Creating a canonical function is a
103 bit tricky, since it must be defined before the range type can be
104 declared. To do this, you must first create a shell type, which is a
105 placeholder type that has no properties except a name and an owner.
106 This is done by issuing the command CREATE TYPE name, with no
107 additional parameters. Then the function can be declared using the
108 shell type as argument and result, and finally the range type can be
109 declared using the same name. This automatically replaces the shell
110 type entry with a valid range type.
111 The optional subtype_diff function must take two values of the subtype
113 type as argument, and return a double precision value representing the
114 difference between the two given values. While this is optional,
115 providing it allows much greater efficiency of GiST indexes on columns
116 of the range type. See Section 8.17.8 for more information.
117 Base Types
119 The fourth form of CREATE TYPE creates a new base type (scalar type).
120 To create a new base type, you must be a superuser. (This restriction
121 is made because an erroneous type definition could confuse or even
122 crash the server.)
123 The parameters can appear in any order, not only that illustrated
125 above, and most are optional. You must register two or more functions
126 (using CREATE FUNCTION) before defining the type. The support functions
127 input_function and output_function are required, while the functions
128 receive_function, send_function, type_modifier_input_function,
129 type_modifier_output_function and analyze_function are optional.
130 Generally these functions have to be coded in C or another low-level
131 language.
132 The input_function converts the type's external textual representation
134 to the internal representation used by the operators and functions
135 defined for the type. output_function performs the reverse
136 transformation. The input function can be declared as taking one
137 argument of type cstring, or as taking three arguments of types
138 cstring, oid, integer. The first argument is the input text as a C
139 string, the second argument is the type's own OID (except for array
140 types, which instead receive their element type's OID), and the third
141 is the typmod of the destination column, if known (-1 will be passed if
142 not). The input function must return a value of the data type itself.
143 Usually, an input function should be declared STRICT; if it is not, it
144 will be called with a NULL first parameter when reading a NULL input
145 value. The function must still return NULL in this case, unless it
146 raises an error. (This case is mainly meant to support domain input
147 functions, which might need to reject NULL inputs.) The output function
148 must be declared as taking one argument of the new data type. The
149 output function must return type cstring. Output functions are not
150 invoked for NULL values.
151 The optional receive_function converts the type's external binary
153 representation to the internal representation. If this function is not
154 supplied, the type cannot participate in binary input. The binary
155 representation should be chosen to be cheap to convert to internal
156 form, while being reasonably portable. (For example, the standard
157 integer data types use network byte order as the external binary
158 representation, while the internal representation is in the machine's
159 native byte order.) The receive function should perform adequate
160 checking to ensure that the value is valid. The receive function can be
161 declared as taking one argument of type internal, or as taking three
162 arguments of types internal, oid, integer. The first argument is a
163 pointer to a StringInfo buffer holding the received byte string; the
164 optional arguments are the same as for the text input function. The
165 receive function must return a value of the data type itself. Usually,
166 a receive function should be declared STRICT; if it is not, it will be
167 called with a NULL first parameter when reading a NULL input value. The
168 function must still return NULL in this case, unless it raises an
169 error. (This case is mainly meant to support domain receive functions,
170 which might need to reject NULL inputs.) Similarly, the optional
171 send_function converts from the internal representation to the external
172 binary representation. If this function is not supplied, the type
173 cannot participate in binary output. The send function must be declared
174 as taking one argument of the new data type. The send function must
175 return type bytea. Send functions are not invoked for NULL values.
176 You should at this point be wondering how the input and output
178 functions can be declared to have results or arguments of the new type,
179 when they have to be created before the new type can be created. The
180 answer is that the type should first be defined as a shell type, which
181 is a placeholder type that has no properties except a name and an
182 owner. This is done by issuing the command CREATE TYPE name, with no
183 additional parameters. Then the C I/O functions can be defined
184 referencing the shell type. Finally, CREATE TYPE with a full definition
185 replaces the shell entry with a complete, valid type definition, after
186 which the new type can be used normally.
187 The optional type_modifier_input_function and
189 type_modifier_output_function are needed if the type supports
190 modifiers, that is optional constraints attached to a type declaration,
191 such as char(5) or numeric(30,2). PostgreSQL allows user-defined types
192 to take one or more simple constants or identifiers as modifiers.
193 However, this information must be capable of being packed into a single
194 non-negative integer value for storage in the system catalogs. The
195 type_modifier_input_function is passed the declared modifier(s) in the
196 form of a cstring array. It must check the values for validity
197 (throwing an error if they are wrong), and if they are correct, return
198 a single non-negative integer value that will be stored as the column
199 “typmod”. Type modifiers will be rejected if the type does not have a
200 type_modifier_input_function. The type_modifier_output_function
201 converts the internal integer typmod value back to the correct form for
202 user display. It must return a cstring value that is the exact string
203 to append to the type name; for example numeric's function might return
204 (30,2). It is allowed to omit the type_modifier_output_function, in
205 which case the default display format is just the stored typmod integer
206 value enclosed in parentheses.
207 The optional analyze_function performs type-specific statistics
209 collection for columns of the data type. By default, ANALYZE will
210 attempt to gather statistics using the type's “equals” and “less-than”
211 operators, if there is a default b-tree operator class for the type.
212 For non-scalar types this behavior is likely to be unsuitable, so it
213 can be overridden by specifying a custom analysis function. The
214 analysis function must be declared to take a single argument of type
215 internal, and return a boolean result. The detailed API for analysis
216 functions appears in src/include/commands/vacuum.h.
217 While the details of the new type's internal representation are only
219 known to the I/O functions and other functions you create to work with
220 the type, there are several properties of the internal representation
221 that must be declared to PostgreSQL. Foremost of these is
222 internallength. Base data types can be fixed-length, in which case
223 internallength is a positive integer, or variable-length, indicated by
224 setting internallength to VARIABLE. (Internally, this is represented by
225 setting typlen to -1.) The internal representation of all
226 variable-length types must start with a 4-byte integer giving the total
227 length of this value of the type. (Note that the length field is often
228 encoded, as described in Section 68.2; it's unwise to access it
229 directly.)
230 The optional flag PASSEDBYVALUE indicates that values of this data type
232 are passed by value, rather than by reference. Types passed by value
233 must be fixed-length, and their internal representation cannot be
234 larger than the size of the Datum type (4 bytes on some machines, 8
235 bytes on others).
236 The alignment parameter specifies the storage alignment required for
238 the data type. The allowed values equate to alignment on 1, 2, 4, or 8
239 byte boundaries. Note that variable-length types must have an alignment
240 of at least 4, since they necessarily contain an int4 as their first
241 component.
242 The storage parameter allows selection of storage strategies for
244 variable-length data types. (Only plain is allowed for fixed-length
245 types.) plain specifies that data of the type will always be stored
246 in-line and not compressed. extended specifies that the system will
247 first try to compress a long data value, and will move the value out of
248 the main table row if it's still too long. external allows the value
249 to be moved out of the main table, but the system will not try to
250 compress it. main allows compression, but discourages moving the value
251 out of the main table. (Data items with this storage strategy might
252 still be moved out of the main table if there is no other way to make a
253 row fit, but they will be kept in the main table preferentially over
254 extended and external items.)
255 All storage values other than plain imply that the functions of the
257 data type can handle values that have been toasted, as described in
258 Section 68.2 and Section 37.13.1. The specific other value given merely
259 determines the default TOAST storage strategy for columns of a
260 toastable data type; users can pick other strategies for individual
261 columns using ALTER TABLE SET STORAGE.
262 The like_type parameter provides an alternative method for specifying
264 the basic representation properties of a data type: copy them from some
265 existing type. The values of internallength, passedbyvalue, alignment,
266 and storage are copied from the named type. (It is possible, though
267 usually undesirable, to override some of these values by specifying
268 them along with the LIKE clause.) Specifying representation this way is
269 especially useful when the low-level implementation of the new type
270 “piggybacks” on an existing type in some fashion.
271 The category and preferred parameters can be used to help control which
273 implicit cast will be applied in ambiguous situations. Each data type
274 belongs to a category named by a single ASCII character, and each type
275 is either “preferred” or not within its category. The parser will
276 prefer casting to preferred types (but only from other types within the
277 same category) when this rule is helpful in resolving overloaded
278 functions or operators. For more details see Chapter 10. For types that
279 have no implicit casts to or from any other types, it is sufficient to
280 leave these settings at the defaults. However, for a group of related
281 types that have implicit casts, it is often helpful to mark them all as
282 belonging to a category and select one or two of the “most general”
283 types as being preferred within the category. The category parameter is
284 especially useful when adding a user-defined type to an existing
285 built-in category, such as the numeric or string types. However, it is
286 also possible to create new entirely-user-defined type categories.
287 Select any ASCII character other than an upper-case letter to name such
288 a category.
289 A default value can be specified, in case a user wants columns of the
291 data type to default to something other than the null value. Specify
292 the default with the DEFAULT key word. (Such a default can be
293 overridden by an explicit DEFAULT clause attached to a particular
294 column.)
295 To indicate that a type is an array, specify the type of the array
297 elements using the ELEMENT key word. For example, to define an array of
298 4-byte integers (int4), specify ELEMENT = int4. More details about
299 array types appear below.
300 To indicate the delimiter to be used between values in the external
302 representation of arrays of this type, delimiter can be set to a
303 specific character. The default delimiter is the comma (,). Note that
304 the delimiter is associated with the array element type, not the array
305 type itself.
306 If the optional Boolean parameter collatable is true, column
308 definitions and expressions of the type may carry collation information
309 through use of the COLLATE clause. It is up to the implementations of
310 the functions operating on the type to actually make use of the
311 collation information; this does not happen automatically merely by
312 marking the type collatable.
313 Array Types
315 Whenever a user-defined type is created, PostgreSQL automatically
316 creates an associated array type, whose name consists of the element
317 type's name prepended with an underscore, and truncated if necessary to
318 keep it less than NAMEDATALEN bytes long. (If the name so generated
319 collides with an existing type name, the process is repeated until a
320 non-colliding name is found.) This implicitly-created array type is
321 variable length and uses the built-in input and output functions
322 array_in and array_out. The array type tracks any changes in its
323 element type's owner or schema, and is dropped if the element type is.
324 You might reasonably ask why there is an ELEMENT option, if the system
326 makes the correct array type automatically. The only case where it's
327 useful to use ELEMENT is when you are making a fixed-length type that
328 happens to be internally an array of a number of identical things, and
329 you want to allow these things to be accessed directly by subscripting,
330 in addition to whatever operations you plan to provide for the type as
331 a whole. For example, type point is represented as just two
332 floating-point numbers, which can be accessed using point[0] and
333 point[1]. Note that this facility only works for fixed-length types
334 whose internal form is exactly a sequence of identical fixed-length
335 fields. A subscriptable variable-length type must have the generalized
336 internal representation used by array_in and array_out. For historical
337 reasons (i.e., this is clearly wrong but it's far too late to change
338 it), subscripting of fixed-length array types starts from zero, rather
339 than from one as for variable-length arrays.
340
342 name
343 The name (optionally schema-qualified) of a type to be created.
344 attribute_name
346 The name of an attribute (column) for the composite type.
347 data_type
349 The name of an existing data type to become a column of the
350 composite type.
351 collation
353 The name of an existing collation to be associated with a column of
354 a composite type, or with a range type.
355 label
357 A string literal representing the textual label associated with one
358 value of an enum type.
359 subtype
361 The name of the element type that the range type will represent
362 ranges of.
363 subtype_operator_class
365 The name of a b-tree operator class for the subtype.
366 canonical_function
368 The name of the canonicalization function for the range type.
369 subtype_diff_function
371 The name of a difference function for the subtype.
372 input_function
374 The name of a function that converts data from the type's external
375 textual form to its internal form.
376 output_function
378 The name of a function that converts data from the type's internal
379 form to its external textual form.
380 receive_function
382 The name of a function that converts data from the type's external
383 binary form to its internal form.
384 send_function
386 The name of a function that converts data from the type's internal
387 form to its external binary form.
388 type_modifier_input_function
390 The name of a function that converts an array of modifier(s) for
391 the type into internal form.
392 type_modifier_output_function
394 The name of a function that converts the internal form of the
395 type's modifier(s) to external textual form.
396 analyze_function
398 The name of a function that performs statistical analysis for the
399 data type.
400 internallength
402 A numeric constant that specifies the length in bytes of the new
403 type's internal representation. The default assumption is that it
404 is variable-length.
405 alignment
407 The storage alignment requirement of the data type. If specified,
408 it must be char, int2, int4, or double; the default is int4.
409 storage
411 The storage strategy for the data type. If specified, must be
412 plain, external, extended, or main; the default is plain.
413 like_type
415 The name of an existing data type that the new type will have the
416 same representation as. The values of internallength,
417 passedbyvalue, alignment, and storage are copied from that type,
418 unless overridden by explicit specification elsewhere in this
419 CREATE TYPE command.
420 category
422 The category code (a single ASCII character) for this type. The
423 default is 'U' for “user-defined type”. Other standard category
424 codes can be found in Table 51.64. You may also choose other ASCII
425 characters in order to create custom categories.
426 preferred
428 True if this type is a preferred type within its type category,
429 else false. The default is false. Be very careful about creating a
430 new preferred type within an existing type category, as this could
431 cause surprising changes in behavior.
432 default
434 The default value for the data type. If this is omitted, the
435 default is null.
436 element
438 The type being created is an array; this specifies the type of the
439 array elements.
440 delimiter
442 The delimiter character to be used between values in arrays made of
443 this type.
444 collatable
446 True if this type's operations can use collation information. The
447 default is false.
448
450 Because there are no restrictions on use of a data type once it's been
451 created, creating a base type or range type is tantamount to granting
452 public execute permission on the functions mentioned in the type
453 definition. This is usually not an issue for the sorts of functions
454 that are useful in a type definition. But you might want to think twice
455 before designing a type in a way that would require “secret”
456 information to be used while converting it to or from external form.
457 Before PostgreSQL version 8.3, the name of a generated array type was
459 always exactly the element type's name with one underscore character
460 (_) prepended. (Type names were therefore restricted in length to one
461 less character than other names.) While this is still usually the case,
462 the array type name may vary from this in case of maximum-length names
463 or collisions with user type names that begin with underscore. Writing
464 code that depends on this convention is therefore deprecated. Instead,
465 use pg_type.typarray to locate the array type associated with a given
466 type.
467 It may be advisable to avoid using type and table names that begin with
469 underscore. While the server will change generated array type names to
470 avoid collisions with user-given names, there is still risk of
471 confusion, particularly with old client software that may assume that
472 type names beginning with underscores always represent arrays.
473 Before PostgreSQL version 8.2, the shell-type creation syntax CREATE
475 TYPE name did not exist. The way to create a new base type was to
476 create its input function first. In this approach, PostgreSQL will
477 first see the name of the new data type as the return type of the input
478 function. The shell type is implicitly created in this situation, and
479 then it can be referenced in the definitions of the remaining I/O
480 functions. This approach still works, but is deprecated and might be
481 disallowed in some future release. Also, to avoid accidentally
482 cluttering the catalogs with shell types as a result of simple typos in
483 function definitions, a shell type will only be made this way when the
484 input function is written in C.
485 In PostgreSQL versions before 7.3, it was customary to avoid creating a
487 shell type at all, by replacing the functions' forward references to
488 the type name with the placeholder pseudo-type opaque. The cstring
489 arguments and results also had to be declared as opaque before 7.3. To
490 support loading of old dump files, CREATE TYPE will accept I/O
491 functions declared using opaque, but it will issue a notice and change
492 the function declarations to use the correct types.
493
495 This example creates a composite type and uses it in a function
496 definition:
497 CREATE TYPE compfoo AS (f1 int, f2 text);
499 CREATE FUNCTION getfoo() RETURNS SETOF compfoo AS $$
501 SELECT fooid, fooname FROM foo
502 $$ LANGUAGE SQL;
503 This example creates an enumerated type and uses it in a table
505 definition:
506 CREATE TYPE bug_status AS ENUM ('new', 'open', 'closed');
508 CREATE TABLE bug (
510 id serial,
511 description text,
512 status bug_status
513 );
514 This example creates a range type:
516 CREATE TYPE float8_range AS RANGE (subtype = float8, subtype_diff = float8mi);
518 This example creates the base data type box and then uses the type in a
520 table definition:
521 CREATE TYPE box;
523 CREATE FUNCTION my_box_in_function(cstring) RETURNS box AS ... ;
525 CREATE FUNCTION my_box_out_function(box) RETURNS cstring AS ... ;
526 CREATE TYPE box (
528 INTERNALLENGTH = 16,
529 INPUT = my_box_in_function,
530 OUTPUT = my_box_out_function
531 );
532 CREATE TABLE myboxes (
534 id integer,
535 description box
536 );
537 If the internal structure of box were an array of four float4 elements,
539 we might instead use:
540 CREATE TYPE box (
542 INTERNALLENGTH = 16,
543 INPUT = my_box_in_function,
544 OUTPUT = my_box_out_function,
545 ELEMENT = float4
546 );
547 which would allow a box value's component numbers to be accessed by
549 subscripting. Otherwise the type behaves the same as before.
550 This example creates a large object type and uses it in a table
552 definition:
553 CREATE TYPE bigobj (
555 INPUT = lo_filein, OUTPUT = lo_fileout,
556 INTERNALLENGTH = VARIABLE
557 );
558 CREATE TABLE big_objs (
559 id integer,
560 obj bigobj
561 );
562 More examples, including suitable input and output functions, are in
564 Section 37.13.
565
567 The first form of the CREATE TYPE command, which creates a composite
568 type, conforms to the SQL standard. The other forms are PostgreSQL
569 extensions. The CREATE TYPE statement in the SQL standard also defines
570 other forms that are not implemented in PostgreSQL.
571 The ability to create a composite type with zero attributes is a
573 PostgreSQL-specific deviation from the standard (analogous to the same
574 case in CREATE TABLE).
575
577 ALTER TYPE (ALTER_TYPE(7)), CREATE DOMAIN (CREATE_DOMAIN(7)), CREATE
578 FUNCTION (CREATE_FUNCTION(7)), DROP TYPE (DROP_TYPE(7))
579PostgreSQL 12.2 2020 CREATE TYPE(7)