Datatypes

CHAR Datatype

The CHAR datatype specifies a fixed-length character string. Oracle ensures that all values stored in a CHAR column have the length specified by size. If you insert a value that is shorter than the column length, then Oracle blank-pads the value to column length. If you try to insert a value that is too long for the column, then Oracle returns an error.

The default length for a CHAR column is 1 byte and the maximum allowed is 2000 bytes. A 1-byte string can be inserted into a CHAR(10) column, but the string is blank-padded to 10 bytes before it is stored.

When you create a table with a CHAR column, by default you supply the column length in bytes. The BYTE qualifier is the same as the default. If you use the CHAR qualifier, for example CHAR(10 CHAR), then you supply the column length in characters. A character is technically a code point of the database character set. Its size can range from 1 byte to 4 bytes, depending on the database character set. The BYTE and CHAR qualifiers override the semantics specified by the NLS_LENGTH_SEMANTICS parameter, which has a default of byte semantics. For performance reasons, Oracle recommends that you use the NLS_LENGTH_SEMANTICS parameter to set length semantics and that you use the BYTE and CHAR qualifiers only when necessary to override the parameter.

To ensure proper data conversion between databases with different character sets, you must ensure that CHAR data consists of well-formed strings. See Oracle Database Globalization Support Guide for more information on character set support.

NCHAR Datatype

The NCHAR datatype is a Unicode-only datatype. When you create a table with an NCHAR column, you define the column length in characters. You define the national character set when you create your database.

The maximum length of a column is determined by the national character set definition. Width specifications of character datatype NCHAR refer to the number of characters. The maximum column size allowed is 2000 bytes.

If you insert a value that is shorter than the column length, then Oracle blank-pads the value to column length. You cannot insert a CHAR value into an NCHAR column, nor can you insert an NCHAR value into a CHAR column.

The following example compares the translated_description column of the pm.product_descriptions table with a national character set string:

SELECT translated_description FROM product_descriptions
   WHERE translated_name = N'LCD Monitor 11/PM';

Please refer to Oracle Database Globalization Support Guide for information on Unicode datatype support.

NVARCHAR2 Datatype

The NVARCHAR2 datatype is a Unicode-only datatype. When you create a table with an NVARCHAR2 column, you supply the maximum number of characters it can hold. Oracle subsequently stores each value in the column exactly as you specify it, provided the value does not exceed the maximum length of the column.

The maximum length of the column is determined by the national character set definition. Width specifications of character datatype NVARCHAR2 refer to the number of characters. The maximum column size allowed is 4000 bytes. Please refer to Oracle Database Globalization Support Guide for information on Unicode datatype support.

VARCHAR2 Datatype

The VARCHAR2 datatype specifies a variable-length character string. When you create a VARCHAR2 column, you supply the maximum number of bytes or characters of data that it can hold. Oracle subsequently stores each value in the column exactly as you specify it, provided the value does not exceed the column's maximum length of the column. If you try to insert a value that exceeds the specified length, then Oracle returns an error.

You must specify a maximum length for a VARCHAR2 column. This maximum must be at least 1 byte, although the actual string stored is permitted to be a zero-length string (''). You can use the CHAR qualifier, for example VARCHAR2(10 CHAR), to give the maximum length in characters instead of bytes. A character is technically a code point of the database character set. CHAR and BYTE qualifiers override the setting of the NLS_LENGTH_SEMANTICS parameter, which has a default of bytes. For performance reasons, Oracle recommends that you use the NLS_LENGTH_SEMANTICS parameter to set length semantics and that you use the BYTE and CHAR qualifiers only when necessary to override the parameter. The maximum length of VARCHAR2 data is 4000 bytes. Oracle compares VARCHAR2 values using nonpadded comparison semantics.

To ensure proper data conversion between databases with different character sets, you must ensure that VARCHAR2 data consists of well-formed strings. See Oracle Database Globalization Support Guide for more information on character set support.

VARCHAR Datatype

Do not use the VARCHAR datatype. Use the VARCHAR2 datatype instead. Although the VARCHAR datatype is currently synonymous with VARCHAR2, the VARCHAR datatype is scheduled to be redefined as a separate datatype used for variable-length character strings compared with different comparison semantics.

Numeric Datatypes

The Oracle Database numeric datatypes store positive and negative fixed and floating-point numbers, zero, infinity, and values that are the undefined result of an operation (that is, is "not a number" or NAN). For information on specifying numeric datatypes as literals, please refer to "Numeric Literals".

NUMBER Datatype

The NUMBER datatype stores zero as well as positive and negative fixed numbers with absolute values from 1.0 x 10^-130 to (but not including) 1.0 x 10¹²⁶. If you specify an arithmetic expression whose value has an absolute value greater than or equal to 1.0 x 10¹²⁶, then Oracle returns an error. Each NUMBER value requires from 1 to 22 bytes.

Specify a fixed-point number using the following form:

NUMBER(p,s)

where:

• p is the precision, or the total number of significant decimal digits, where the most significant digit is the left-most nonzero digit, and the least significant digit is the right-most known digit. Oracle guarantees the portability of numbers with precision of up to 20 base-100 digits, which is equivalent to 39 or 40 decimal digits depending on the position of the decimal point.

• s is the scale, or the number of digits from the decimal point to the least significant digit. The scale can range from -84 to 127.

  • Positive scale is the number of significant digits to the right of the decimal point to and including the least significant digit.

  • Negative scale is the number of significant digits to the left of the decimal point, to but not including the least significant digit. For negative scale the least significant digit is on the left side of the decimal point, because the actual data is rounded to the specified number of places to the left of the decimal point. For example, a specification of (10,-2) means to round to hundreds.

Scale can be greater than precision, most commonly when e notation is used. When scale is greater than precision, the precision specifies the maximum number of significant digits to the right of the decimal point. For example, a column defined as NUMBER(4,5) requires a zero for the first digit after the decimal point and rounds all values past the fifth digit after the decimal point.

It is good practice to specify the scale and precision of a fixed-point number column for extra integrity checking on input. Specifying scale and precision does not force all values to a fixed length. If a value exceeds the precision, then Oracle returns an error. If a value exceeds the scale, then Oracle rounds it.

Specify an integer using the following form:

NUMBER(p)

This represents a fixed-point number with precision p and scale 0 and is equivalent to NUMBER(p,0).

Specify a floating-point number using the following form:

NUMBER 

The absence of precision and scale designators specifies the maximum range and precision for an Oracle number.

Table 2-2 show how Oracle stores data using different precisions and scales.

Floating-Point Numbers

Floating-point numbers can have a decimal point anywhere from the first to the last digit or can have no decimal point at all. An exponent may optionally be used following the number to increase the range (for example, 1.777 e^-20). A scale value is not applicable to floating-point numbers, because the number of digits that can appear after the decimal point is not restricted.

Binary floating-point numbers differ from NUMBER in the way the values are stored internally by Oracle Database. Values are stored using decimal precision for NUMBER. All literals that are within the range and precision supported by NUMBER are stored exactly as NUMBER. Literals are stored exactly because literals are expressed using decimal precision (the digits 0 through 9). Binary floating-point numbers are stored using binary precision (the digits 0 and 1). Such a storage scheme cannot represent all values using decimal precision exactly. Frequently, the error that occurs when converting a value from decimal to binary precision is undone when the value is converted back from binary to decimal precision. The literal 0.1 is such an example.

Oracle Database provides two numeric datatypes exclusively for floating-point numbers:

FLOAT
FLOAT(n)

Numeric Precedence

Numeric precedence determines, for operations that support numeric datatypes, the datatype Oracle uses if the arguments to the operation have different datatypes. BINARY_DOUBLE has the highest numeric precedence, followed by BINARY_FLOAT, and finally by NUMBER. Therefore, in any operation on multiple numeric values:

• If any of the operands is BINARY_DOUBLE, then Oracle attempts to convert all the operands implicitly to BINARY_DOUBLE before performing the operation.

• If none of the operands is BINARY_DOUBLE but any of the operands is BINARY_FLOAT, then Oracle attempts to convert all the operands implicitly to BINARY_FLOAT before performing the operation.

• Otherwise, Oracle attempts to convert all the operands to NUMBER before performing the operation.

If any implicit conversion is needed and fails, then the operation fails. Table 2-10, "Implicit Type Conversion Matrix" for more information on implicit conversion.

In the context of other datatypes, numeric datatypes have lower precedence than the datetime/interval datatypes and higher precedence than character and all other datatypes.

LONG Datatype

LONG columns store variable-length character strings containing up to 2 gigabytes -1, or 2³¹-1 bytes. LONG columns have many of the characteristics of VARCHAR2 columns. You can use LONG columns to store long text strings. The length of LONG values may be limited by the memory available on your computer. LONG literals are formed as described for "Text Literals".

Do not create tables with LONG columns. Use LOB columns (CLOB, NCLOB, BLOB) instead. LONG columns are supported only for backward compatibility.

Oracle also recommends that you convert existing LONG columns to LOB columns. LOB columns are subject to far fewer restrictions than LONG columns. Further, LOB functionality is enhanced in every release, whereas LONG functionality has been static for several releases. See the modify_col_properties clause of ALTER TABLE and TO_LOB for more information on converting LONG columns to LOB.

You can reference LONG columns in SQL statements in these places:

• SELECT lists

• SET clauses of UPDATE statements

• VALUES clauses of INSERT statements

The use of LONG values is subject to these restrictions:

• A table can contain only one LONG column.

• You cannot create an object type with a LONG attribute.

• LONG columns cannot appear in WHERE clauses or in integrity constraints (except that they can appear in NULL and NOT NULL constraints).

• LONG columns cannot be indexed.

• LONG data cannot be specified in regular expressions.

• A stored function cannot return a LONG value.

• You can declare a variable or argument of a PL/SQL program unit using the LONG datatype. However, you cannot then call the program unit from SQL.

• Within a single SQL statement, all LONG columns, updated tables, and locked tables must be located on the same database.

• LONG and LONG RAW columns cannot be used in distributed SQL statements and cannot be replicated.

• If a table has both LONG and LOB columns, then you cannot bind more than 4000 bytes of data to both the LONG and LOB columns in the same SQL statement. However, you can bind more than 4000 bytes of data to either the LONG or the LOB column.

In addition, LONG columns cannot appear in these parts of SQL statements:

• GROUP BY clauses, ORDER BY clauses, or CONNECT BY clauses or with the DISTINCT operator in SELECT statements

• The UNIQUE operator of a SELECT statement

• The column list of a CREATE CLUSTER statement

• The CLUSTER clause of a CREATE MATERIALIZED VIEW statement

• SQL built-in functions, expressions, or conditions

• SELECT lists of queries containing GROUP BY clauses

• SELECT lists of subqueries or queries combined by the UNION, INTERSECT, or MINUS set operators

• SELECT lists of CREATE TABLE ... AS SELECT statements

• ALTER TABLE ... MOVE statements

• SELECT lists in subqueries in INSERT statements

Triggers can use the LONG datatype in the following manner:

• A SQL statement within a trigger can insert data into a LONG column.

• If data from a LONG column can be converted to a constrained datatype (such as CHAR and VARCHAR2), then a LONG column can be referenced in a SQL statement within a trigger.

• Variables in triggers cannot be declared using the LONG datatype.

• :NEW and :OLD cannot be used with LONG columns.

You can use Oracle Call Interface functions to retrieve a portion of a LONG value from the database.

Datetime and Interval Datatypes

The datetime datatypes are DATE, TIMESTAMP, TIMESTAMP WITH TIME ZONE, and TIMESTAMP WITH LOCAL TIME ZONE. Values of datetime datatypes are sometimes called datetimes. The interval datatypes are INTERVAL YEAR TO MONTH and INTERVAL DAY TO SECOND. Values of interval datatypes are sometimes called intervals. For information on expressing datetime and interval values as literals, please refer to "Datetime Literals" and "Interval Literals".

Both datetimes and intervals are made up of fields. The values of these fields determine the value of the datatype. Table 2-4 lists the datetime fields and their possible values for datetimes and intervals.

To avoid unexpected results in your DML operations on datetime data, you can verify the database and session time zones by querying the built-in SQL functions DBTIMEZONE and SESSIONTIMEZONE. If the time zones have not been set manually, Oracle Database uses the operating system time zone by default. If the operating system time zone is not a valid Oracle time zone, then Oracle uses UTC as the default value.

Note: TIMEZONE_HOUR and TIMEZONE_MINUTE are specified together and interpreted as an entity in the format +|- hh:mm, with values ranging from -12:59 to +14:00. Please refer to Oracle Data Provider for .NET Developer's Guide for information on specifying time zone values for that API.

DATE Datatype

The DATE datatype stores date and time information. Although date and time information can be represented in both character and number datatypes, the DATE datatype has special associated properties. For each DATE value, Oracle stores the following information: century, year, month, date, hour, minute, and second.

You can specify a DATE value as a literal, or you can convert a character or numeric value to a date value with the TO_DATE function. For examples of expressing DATE values in both these ways, please refer to "Datetime Literals".

SELECT TO_DATE('2005', 'YYYY') FROM DUAL;

TO_DATE('
---------
01-MAY-05

SELECT TO_CHAR(TO_DATE('01-01-1997', 'MM-DD-YYYY'),'J')
    FROM DUAL;

TO_CHAR
--------
2450450

TIMESTAMP Datatype

The TIMESTAMP datatype is an extension of the DATE datatype. It stores the year, month, and day of the DATE datatype, plus hour, minute, and second values. This datatype is useful for storing precise time values. Specify the TIMESTAMP datatype as follows:

TIMESTAMP [(fractional_seconds_precision)] 

where fractional_seconds_precision optionally specifies the number of digits Oracle stores in the fractional part of the SECOND datetime field. When you create a column of this datatype, the value can be a number in the range 0 to 9. The default is 6.

TIMESTAMP WITH TIME ZONE Datatype

TIMESTAMP WITH TIME ZONE is a variant of TIMESTAMP that includes a time zone offset in its value. The time zone offset is the difference (in hours and minutes) between local time and UTC (Coordinated Universal Time—formerly Greenwich Mean Time). This datatype is useful for collecting and evaluating date information across geographic regions.

Specify the TIMESTAMP WITH TIME ZONE datatype as follows:

TIMESTAMP [(fractional_seconds_precision)] WITH TIME ZONE

where fractional_seconds_precision optionally specifies the number of digits Oracle stores in the fractional part of the SECOND datetime field. When you create a column of this datatype, the value can be a number in the range 0 to 9. The default is 6.

Oracle time zone data is derived from the public domain information available at ftp://elsie.nci.nih.gov/pub/. Oracle time zone data may not reflect the most recent data available at this site.

TIMESTAMP WITH LOCAL TIME ZONE Datatype

TIMESTAMP WITH LOCAL TIME ZONE is another variant of TIMESTAMP that includes a time zone offset in its value. It differs from TIMESTAMP WITH TIME ZONE in that data stored in the database is normalized to the database time zone, and the time zone offset is not stored as part of the column data. When a user retrieves the data, Oracle returns it in the user's local session time zone. The time zone offset is the difference (in hours and minutes) between local time and UTC (Coordinated Universal Time—formerly Greenwich Mean Time). This datatype is useful for displaying date information in the time zone of the client system in a two-tier application.

Specify the TIMESTAMP WITH LOCAL TIME ZONE datatype as follows:

TIMESTAMP [(fractional_seconds_precision)] WITH LOCAL TIME ZONE

where fractional_seconds_precision optionally specifies the number of digits Oracle stores in the fractional part of the SECOND datetime field. When you create a column of this datatype, the value can be a number in the range 0 to 9. The default is 6.

Oracle time zone data is derived from the public domain information available at ftp://elsie.nci.nih.gov/pub/. Oracle time zone data may not reflect the most recent data available at this site.

INTERVAL YEAR TO MONTH Datatype

INTERVAL YEAR TO MONTH stores a period of time using the YEAR and MONTH datetime fields. This datatype is useful for representing the difference between two datetime values when only the year and month values are significant.

Specify INTERVAL YEAR TO MONTH as follows:

INTERVAL YEAR [(year_precision)] TO MONTH

where year_precision is the number of digits in the YEAR datetime field. The default value of year_precision is 2.

You have a great deal of flexibility when specifying interval values as literals. Please refer to "Interval Literals" for detailed information on specify interval values as literals. Also see "Datetime and Interval Examples" for an example using intervals.

INTERVAL DAY TO SECOND Datatype

INTERVAL DAY TO SECOND stores a period of time in terms of days, hours, minutes, and seconds. This datatype is useful for representing the precise difference between two datetime values.

Specify this datatype as follows:

INTERVAL DAY [(day_precision)] 
   TO SECOND [(fractional_seconds_precision)]

where

• day_precision is the number of digits in the DAY datetime field. Accepted values are 0 to 9. The default is 2.

• fractional_seconds_precision is the number of digits in the fractional part of the SECOND datetime field. Accepted values are 0 to 9. The default is 6.

You have a great deal of flexibility when specifying interval values as literals. Please refer to "Interval Literals" for detailed information on specify interval values as literals. Also see "Datetime and Interval Examples" for an example using intervals.

Datetime/Interval Arithmetic

You can perform a number of arithmetic operations on date (DATE), timestamp (TIMESTAMP, TIMESTAMP WITH TIME ZONE, and TIMESTAMP WITH LOCAL TIME ZONE) and interval (INTERVAL DAY TO SECOND and INTERVAL YEAR TO MONTH) data. Oracle calculates the results based on the following rules:

• You can use NUMBER constants in arithmetic operations on date and timestamp values, but not interval values. Oracle internally converts timestamp values to date values and interprets NUMBER constants in arithmetic datetime and interval expressions as numbers of days. For example, SYSDATE + 1 is tomorrow. SYSDATE - 7 is one week ago. SYSDATE + (10/1440) is ten minutes from now. Subtracting the hire_date column of the sample table employees from SYSDATE returns the number of days since each employee was hired. You cannot multiply or divide date or timestamp values.

• Oracle implicitly converts BINARY_FLOAT and BINARY_DOUBLE operands to NUMBER.

• Each DATE value contains a time component, and the result of many date operations include a fraction. This fraction means a portion of one day. For example, 1.5 days is 36 hours. These fractions are also returned by Oracle built-in functions for common operations on DATE data. For example, the MONTHS_BETWEEN function returns the number of months between two dates. The fractional portion of the result represents that portion of a 31-day month.

• If one operand is a DATE value or a numeric value (neither of which contains time zone or fractional seconds components), then:

  • Oracle implicitly converts the other operand to DATE data. (The exception is multiplication of a numeric value times an interval, which returns an interval.)

  • If the other operand has a time zone value, then Oracle uses the session time zone in the returned value.

  • If the other operand has a fractional seconds value, then the fractional seconds value is lost.

• When you pass a timestamp, interval, or numeric value to a built-in function that was designed only for the DATE datatype, Oracle implicitly converts the non-DATE value to a DATE value. Please refer to "Datetime Functions" for information on which functions cause implicit conversion to DATE.

• Oracle performs all timestamp arithmetic in UTC time. For TIMESTAMP WITH LOCAL TIME ZONE, Oracle converts the datetime value from the database time zone to UTC and converts back to the database time zone after performing the arithmetic. For TIMESTAMP WITH TIME ZONE, the datetime value is always in UTC, so no conversion is necessary.

Table 2-5 is a matrix of datetime arithmetic operations. Dashes represent operations that are not supported.

Examples You can add an interval value expression to a start time. Consider the sample table oe.orders with a column order_date. The following statement adds 30 days to the value of the order_date column:

SELECT order_id, order_date + INTERVAL '30' DAY FROM orders;

Support for Daylight Saving Times

Oracle Database automatically determines, for any given time zone region, whether daylight saving is in effect and returns local time values accordingly. The datetime value is sufficient for Oracle to determine whether daylight saving time is in effect for a given region in all cases except boundary cases. A boundary case occurs during the period when daylight saving goes into or comes out of effect. For example, in the US-Pacific region, when daylight saving goes into effect, the time changes from 2:00 a.m. to 3:00 a.m. The one hour interval between 2 and 3 a.m. does not exist. When daylight saving goes out of effect, the time changes from 2:00 a.m. back to 1:00 a.m., and the one-hour interval between 1 and 2 a.m. is repeated.

To resolve these boundary cases, Oracle uses the TZR and TZD format elements, as described in Table 2-19. TZR represents the time zone region in datetime input strings. Examples are 'Australia/North', 'UTC', and 'Singapore'. TZD represents an abbreviated form of the time zone region with daylight saving information. Examples are 'PST' for US/Pacific standard time and 'PDT' for US/Pacific daylight time. To see a listing of valid values for the TZR and TZD format elements, query the TZNAME and TZABBREV columns of the V$TIMEZONE_NAMES dynamic performance view.

Timezone region names are needed by the daylight saving feature. The region names are stored in two time zone files. The default time zone file is the complete (larger) file containing all time zones. The other time zone file is a small file containing only the most common time zones to maximize performance. If your time zone is in the small file, and you want to maximize performance, then you must provide a path to the small file by way of the ORA_TZFILE environment variable. Please refer to Oracle Database Administrator's Guide for more information about setting the ORA_TZFILE environment variable. For a complete listing of the timezone region names in both files, please refer to Oracle Database Globalization Support Guide.

Oracle time zone data is derived from the public domain information available at ftp://elsie.nci.nih.gov/pub/. Oracle time zone data may not reflect the most recent data available at this site.

Datetime and Interval Examples

The following example shows how to declare some datetime and interval datatypes.

CREATE TABLE time_table (
   start_time      TIMESTAMP,
   duration_1      INTERVAL DAY (6) TO SECOND (5),
   duration_2      INTERVAL YEAR TO MONTH);

The start_time column is of type TIMESTAMP. The implicit fractional seconds precision of TIMESTAMP is 6.

The duration_1 column is of type INTERVAL DAY TO SECOND. The maximum number of digits in field DAY is 6 and the maximum number of digits in the fractional second is 5. The maximum number of digits in all other datetime fields is 2.

The duration_2 column is of type INTERVAL YEAR TO MONTH. The maximum number of digits of the value in each field (YEAR and MONTH) is 2.

Interval datatypes do not have format models. Therefore, to adjust their presentation, you must combine character functions such as EXTRACT and concatenate the components. For example, the following examples query the hr.employees and oe.orders tables, respectively, and change interval output from the form "yy-mm" to "yy years mm months" and from "dd-hh" to "dddd days hh hours":

SELECT last_name, EXTRACT(YEAR FROM (SYSDATE - hire_date) YEAR TO MONTH )
   || ' years '
   || EXTRACT(MONTH FROM (SYSDATE - hire_date) YEAR TO MONTH )
   || ' months'  "Interval"
FROM employees ;

LAST_NAME                 Interval
------------------------- --------------------
King                      17 years 11 months
Kochhar                   15 years 8 months
De Haan                   12 years 4 months
Hunold                    15 years 4 months
Ernst                     14 years 0 months
Austin                    7 years 11 months
Pataballa                 7 years 3 months
Lorentz                   6 years 3 months
Greenberg                 10 years 9 months
. . .

SELECT order_id,
   EXTRACT(DAY FROM (SYSDATE - order_date) DAY TO SECOND )
   || ' days '
   || EXTRACT(HOUR FROM (SYSDATE - order_date) DAY TO SECOND )
   || ' hours' "Interval"
FROM orders;

  ORDER_ID Interval
---------- --------------------
      2458 2095 days 18 hours
      2397 2000 days 17 hours
      2454 2048 days 16 hours
      2354 1762 days 16 hours
      2358 1950 days 15 hours
      2381 1823 days 13 hours
      2440 2080 days 12 hours
      2357 2680 days 11 hours
      2394 1917 days 10 hours
      2435 2078 days 10 hours
. . .

RAW and LONG RAW Datatypes

The RAW and LONG RAW datatypes store data that is not to be interpreted (that is, not explicitly converted when moving data between different systems) by Oracle Database. These datatypes are intended for binary data or byte strings. For example, you can use LONG RAW to store graphics, sound, documents, or arrays of binary data, for which the interpretation is dependent on the use.

Oracle strongly recommends that you convert LONG RAW columns to binary LOB (BLOB) columns. LOB columns are subject to far fewer restrictions than LONG columns. See TO_LOB for more information.

RAW is a variable-length datatype like VARCHAR2, except that Oracle Net (which connects user sessions to the instance) and the Import and Export utilities do not perform character conversion when transmitting RAW or LONG RAW data. In contrast, Oracle Net and Import/Export automatically convert CHAR, VARCHAR2, and LONG data from the database character set to the user session character set (which you can set with the NLS_LANGUAGE parameter of the ALTER SESSION statement), if the two character sets are different.

When Oracle automatically converts RAW or LONG RAW data to and from CHAR data, the binary data is represented in hexadecimal form, with one hexadecimal character representing every four bits of RAW data. For example, one byte of RAW data with bits 11001011 is displayed and entered as CB.

Large Object (LOB) Datatypes

The built-in LOB datatypes BLOB, CLOB, and NCLOB (stored internally) and BFILE (stored externally) can store large and unstructured data such as text, image, video, and spatial data. The size of BLOB, CLOB, and NCLOB data can be up to (4 gigabytes -1) * (the value of the CHUNK parameter of LOB storage). If the tablespaces in your database are of standard block size, and if you have used the default value of the CHUNK parameter of LOB storage when creating a LOB column, then this is equivalent to (4 gigabytes - 1) * (database block size). BFILE data can be up to 2³²-1 bytes, although your operating system may impose restrictions on this maximum.

When creating a table, you can optionally specify different tablespace and storage characteristics for LOB columns or LOB object attributes from those specified for the table.

LOB columns contain LOB locators that can refer to in-line (in the database) or out-of-line (outside the database) LOB values. Selecting a LOB from a table actually returns the LOB locator and not the entire LOB value. The DBMS_LOB package and Oracle Call Interface (OCI) operations on LOBs are performed through these locators.

LOBs are similar to LONG and LONG RAW types, but differ in the following ways:

• LOBs can be attributes of an object type (user-defined datatype).

• The LOB locator is stored in the table column, either with or without the actual LOB value. BLOB, NCLOB, and CLOB values can be stored in separate tablespaces. BFILE data is stored in an external file on the server.

• When you access a LOB column, the locator is returned.

• A LOB can be up to (4 gigabytes - 1)*(database block size) in size. BFILE data can be up to 2³²-1 bytes, although your operating system may impose restrictions on this maximum.

  Preceding corrected; thomas.chang, 8/26/04.

• LOBs permit efficient, random, piece-wise access to and manipulation of data.

• You can define more than one LOB column in a table.

• With the exception of NCLOB, you can define one or more LOB attributes in an object.

• You can declare LOB bind variables.

• You can select LOB columns and LOB attributes.

• You can insert a new row or update an existing row that contains one or more LOB columns or an object with one or more LOB attributes. In update operations, you can set the internal LOB value to NULL, empty, or replace the entire LOB with data. You can set the BFILE to NULL or make it point to a different file.

• You can update a LOB row-column intersection or a LOB attribute with another LOB row-column intersection or LOB attribute.

• You can delete a row containing a LOB column or LOB attribute and thereby also delete the LOB value. For BFILEs, the actual operating system file is not deleted.

You can access and populate rows of an in-line LOB column (a LOB column stored in the database) or a LOB attribute (an attribute of an object type column stored in the database) simply by issuing an INSERT or UPDATE statement.

Restrictions on LOB Columns  LOB columns are subject to the following restrictions:

• You cannot specify a LOB as a primary key column.

• Oracle Database has limited support for remote LOBs. You can perform create table as select and insert/update as select of LOB columns.

  For example:

  CREATE TABLE t AS SELECT * FROM table1@remote_site;
  INSERT INTO t SELECT * FROM table1@remote_site;
  UPDATE t SET lobcol = (SELECT lobcol FROM table1@remote_site);
  INSERT INTO table1@remote_site SELECT * from local_table;
  UPDATE table1@remote_site SET lobcol = (SELECT lobcol FROM local_table);
  DELETE FROM table1@remote_site <WHERE clause involving non_lob_columns>
  
  

  In statements structured like the preceding examples, only standalone LOB columns are allowed in the select list.

  SQL functions and DBMS_LOB APIs are not supported for use with remote LOB columns. For example, the following statement is supported:

  CREATE TABLE AS SELECT clob_col FROM tab@dbs2; 
  
  

  However, the following statement is not supported:

  CREATE TABLE AS SELECT DBMS_LOB.SUBSTR(clob_col) from tab@dbs2; 
  
  

  In addition, you can insert a character or binary buffer into a remote CLOB or BLOB, and select a remote CLOB or BLOB into a character or binary buffer. For example (in PL/SQL):

  SELECT clobcol1, type1.blobattr INTO varchar_buf1, raw_buf2 FROM
     table1@remote_site;
  INSERT INTO table1@remotesite (clobcol1, type1.blobattr) VALUES varchar_buf1,
     raw_buf2;
  
  

  This is the only supported syntax involving LOBs in remote tables. No other usage is supported.

• Clusters cannot contain LOBs, either as key or nonkey columns.

• The following data structures are supported only as temporary instances. You cannot store these instances in database tables:

  • VARRAY of any LOB type

  • VARRAY of any type containing a LOB type, such as an object type with a LOB attribute.

• You cannot specify LOB columns in the ORDER BY clause of a query, or in the GROUP BY clause of a query or in an aggregate function.

• You cannot specify a LOB column in a SELECT... DISTINCT or SELECT... UNIQUE statement or in a join. However, you can specify a LOB attribute of an object type column in a SELECT... DISTINCT statement or in a query that uses the UNION or MINUS set operator if the column's object type has a MAP or ORDER function defined on it.

• You cannot specify LOB columns in ANALYZE... COMPUTE or ANALYZE... ESTIMATE statements.

• The first (INITIAL) extent of a LOB segment must contain at least three database blocks.

• When creating an UPDATE DML trigger, you cannot specify a LOB column in the UPDATE OF clause.

• You cannot specify a LOB column as part of an index key. However, you can specify a LOB column in the indextype specification of a domain index. In addition, Oracle Text lets you define an index on a CLOB column.

• In an INSERT... AS SELECT operation, you can bind up to 4000 bytes of data to LOB columns and attributes.

• If a table has both LONG and LOB columns, you cannot bind more than 4000 bytes of data to both the LONG and LOB columns in the same SQL statement. However, you can bind more than 4000 bytes of data to either the LONG or the LOB column.

BFILE Datatype

The BFILE datatype enables access to binary file LOBs that are stored in file systems outside Oracle Database. A BFILE column or attribute stores a BFILE locator, which serves as a pointer to a binary file on the server file system. The locator maintains the directory name and the filename.

You can change the filename and path of a BFILE without affecting the base table by using the BFILENAME function. Please refer to BFILENAME for more information on this built-in SQL function.

Correction in last sentence below; thomas.chang, 8/26/04.

Binary file LOBs do not participate in transactions and are not recoverable. Rather, the underlying operating system provides file integrity and durability. BFILE data can be up to 2³²-1 bytes, although your operating system may impose restrictions on this maximum.

The database administrator must ensure that the external file exists and that Oracle processes have operating system read permissions on the file.

The BFILE datatype enables read-only support of large binary files. You cannot modify or replicate such a file. Oracle provides APIs to access file data. The primary interfaces that you use to access file data are the DBMS_LOB package and the Oracle Call Interface (OCI).

BLOB Datatype

The BLOB datatype stores unstructured binary large objects. BLOB objects can be thought of as bitstreams with no character set semantics. BLOB objects can store binary data up to (4 gigabytes -1) * (the value of the CHUNK parameter of LOB storage). If the tablespaces in your database are of standard block size, and if you have used the default value of the CHUNK parameter of LOB storage when creating a LOB column, then this is equivalent to (4 gigabytes - 1) * (database block size).

BLOB objects have full transactional support. Changes made through SQL, the DBMS_LOB package, or the Oracle Call Interface (OCI) participate fully in the transaction. BLOB value manipulations can be committed and rolled back. However, you cannot save a BLOB locator in a PL/SQL or OCI variable in one transaction and then use it in another transaction or session.

CLOB Datatype

The CLOB datatype stores single-byte and multibyte character data. Both fixed-width and variable-width character sets are supported, and both use the database character set. CLOB objects can store up to (4 gigabytes -1) * (the value of the CHUNK parameter of LOB storage) of character data. If the tablespaces in your database are of standard block size, and if you have used the default value of the CHUNK parameter of LOB storage when creating a LOB column, then this is equivalent to (4 gigabytes - 1) * (database block size).

CLOB objects have full transactional support. Changes made through SQL, the DBMS_LOB package, or the Oracle Call Interface (OCI) participate fully in the transaction. CLOB value manipulations can be committed and rolled back. However, you cannot save a CLOB locator in a PL/SQL or OCI variable in one transaction and then use it in another transaction or session.

NCLOB Datatype

The NCLOB datatype stores Unicode data. Both fixed-width and variable-width character sets are supported, and both use the national character set. NCLOB objects can store up to (4 gigabytes -1) * (the value of the CHUNK parameter of LOB storage) of character text data. If the tablespaces in your database are of standard block size, and if you have used the default value of the CHUNK parameter of LOB storage when creating a LOB column, then this is equivalent to (4 gigabytes - 1) * (database block size)(4 gigabytes-1) * (database block size).

NCLOB objects have full transactional support. Changes made through SQL, the DBMS_LOB package, or the OCI participate fully in the transaction. NCLOB value manipulations can be committed and rolled back. However, you cannot save an NCLOB locator in a PL/SQL or OCI variable in one transaction and then use it in another transaction or session.

ROWID Datatype

Each row in the database has an address. You can examine a row address by querying the pseudocolumn ROWID. Values of this pseudocolumn are strings representing the address of each row. These strings have the datatype ROWID. You can also create tables and clusters that contain actual columns having the ROWID datatype. Oracle Database does not guarantee that the values of such columns are valid rowids. Please refer to Chapter 3, "Pseudocolumns" for more information on the ROWID pseudocolumn.

Restricted Rowids

Beginning with Oracle8, Oracle SQL incorporated an extended format for rowids to efficiently support partitioned tables and indexes and tablespace-relative data block addresses (DBAs) without ambiguity.

Character values representing rowids in Oracle7 and earlier releases are called restricted rowids. Their format is as follows:

block.row.file

where:

• block is a hexadecimal string identifying the data block of the datafile containing the row. The length of this string depends on your operating system.

• row is a four-digit hexadecimal string identifying the row in the data block. The first row of the block has a digit of 0.

• file is a hexadecimal string identifying the database file containing the row. The first datafile has the number 1. The length of this string depends on your operating system.

Extended Rowids

The extended ROWID datatype stored in a user column includes the data in the restricted rowid plus a data object number. The data object number is an identification number assigned to every database segment. You can retrieve the data object number from the data dictionary views USER_OBJECTS, DBA_OBJECTS, and ALL_OBJECTS. Objects that share the same segment (clustered tables in the same cluster, for example) have the same object number.

Extended rowids are stored as base 64 values that can contain the characters A-Z, a-z, 0-9, and the plus sign (+) and forward slash (/). Extended rowids are not available directly. You can use a supplied package, DBMS_ROWID, to interpret extended rowid contents. The package functions extract and provide information that would be available directly from a restricted rowid as well as information specific to extended rowids.

Compatibility and Migration

The restricted form of a rowid is still supported in this release for backward compatibility, but all tables return rowids in the extended format.

UROWID Datatype

Each row in a database has an address. However, the rows of some tables have addresses that are not physical or permanent or were not generated by Oracle Database. For example, the row addresses of index-organized tables are stored in index leaves, which can move. Rowids of foreign tables (such as DB2 tables accessed through a gateway) are not standard Oracle rowids.

Oracle uses universal rowids (urowids) to store the addresses of index-organized and foreign tables. Index-organized tables have logical urowids and foreign tables have foreign urowids. Both types of urowid are stored in the ROWID pseudocolumn (as are the physical rowids of heap-organized tables).

Oracle creates logical rowids based on the primary key of the table. The logical rowids do not change as long as the primary key does not change. The ROWID pseudocolumn of an index-organized table has a datatype of UROWID. You can access this pseudocolumn as you would the ROWID pseudocolumn of a heap-organized table (that is, using a SELECT ... ROWID statement). If you want to store the rowids of an index-organized table, then you can define a column of type UROWID for the table and retrieve the value of the ROWID pseudocolumn into that column.

Object Types

Object types are abstractions of the real-world entities, such as purchase orders, that application programs deal with. An object type is a schema object with three kinds of components:

• A name, which identifies the object type uniquely within that schema.

• Attributes, which are built-in types or other user-defined types. Attributes model the structure of the real-world entity.

• Methods, which are functions or procedures written in PL/SQL and stored in the database, or written in a language like C or Java and stored externally. Methods implement operations the application can perform on the real-world entity.

REF Datatypes

An object identifier (represented by the keyword OID) uniquely identifies an object and enables you to reference the object from other objects or from relational tables. A datatype category called REF represents such references. A REF datatype is a container for an object identifier. REF values are pointers to objects.

When a REF value points to a nonexistent object, the REF is said to be "dangling". A dangling REF is different from a null REF. To determine whether a REF is dangling or not, use the condition IS [NOT] DANGLING. For example, given object view oc_orders in the sample schema oe, the column customer_ref is of type REF to type customer_typ, which has an attribute cust_email:

SELECT o.customer_ref.cust_email
   FROM oc_orders o 
   WHERE o.customer_ref IS NOT DANGLING;

Varrays

An array is an ordered set of data elements. All elements of a given array are of the same datatype. Each element has an index, which is a number corresponding to the position of the element in the array.

The number of elements in an array is the size of the array. Oracle arrays are of variable size, which is why they are called varrays. You must specify a maximum size when you declare the varray.

When you declare a varray, it does not allocate space. It defines a type, which you can use as:

• The datatype of a column of a relational table

• An object type attribute

• A PL/SQL variable, parameter, or function return type

Oracle normally stores an array object either in line (that is, as part of the row data) or out of line (in a LOB), depending on its size. However, if you specify separate storage characteristics for a varray, then Oracle stores it out of line, regardless of its size. Please refer to the varray_col_properties of CREATE TABLE for more information about varray storage.

Nested Tables

A nested table type models an unordered set of elements. The elements may be built-in types or user-defined types. You can view a nested table as a single-column table or, if the nested table is an object type, as a multicolumn table, with a column for each attribute of the object type.

A nested table definition does not allocate space. It defines a type, which you can use to declare:

• The datatype of a column of a relational table

• An object type attribute

• A PL/SQL variable, parameter, or function return type

When a nested table appears as the type of a column in a relational table or as an attribute of the underlying object type of an object table, Oracle stores all of the nested table data in a single table, which it associates with the enclosing relational or object table.

ANYTYPE

This type can contain a type description of any named SQL type or unnamed transient type.

ANYDATA

This type contains an instance of a given type, with data, plus a description of the type. ANYDATA can be used as a table column datatype and lets you store heterogeneous values in a single column. The values can be of SQL built-in types as well as user-defined types.

ANYDATASET

This type contains a description of a given type plus a set of data instances of that type. ANYDATASET can be used as a procedure parameter datatype where such flexibility is needed. The values of the data instances can be of SQL built-in types as well as user-defined types.

XMLType

This Oracle-supplied type can be used to store and query XML data in the database. XMLType has member functions you can use to access, extract, and query the XML data using XPath expressions. XPath is another standard developed by the W3C committee to traverse XML documents. Oracle XMLType functions support many W3C XPath expressions. Oracle also provides a set of SQL functions and PL/SQL packages to create XMLType values from existing relational or object-relational data.

XMLType is a system-defined type, so you can use it as an argument of a function or as the datatype of a table or view column. You can also create tables and views of XMLType. When you create an XMLType column in a table, you can choose to store the XML data in a CLOB column or object relationally.

You can also register the schema (using the DBMS_XMLSCHEMA package) and create a table or column conforming to the registered schema. In this case Oracle stores the XML data in underlying object-relational columns by default, but you can specify storage in a CLOB column even for schema-based data.

Queries and DML on XMLType columns operate the same regardless of the storage mechanism.

URI Datatypes

Oracle supplies a family of URI types—URIType, DBURIType, XDBURIType, and HTTPURIType—which are related by an inheritance hierarchy. URIType is an object type and the others are subtypes of URIType. Since URIType is the supertype, you can create columns of this type and store DBURIType or HTTPURIType type instances in this column.

HTTPURIType You can use HTTPURIType to store URLs to external Web pages or to files. Oracle accesses these files using HTTP (Hypertext Transfer Protocol).

XDBURIType You can use XDBURIType to expose documents in the XML database hierarchy as URIs that can be embedded in any URIType column in a table. The XDBURIType consists of a URL, which comprises the hierarchical name of the XML document to which it refers and an optional fragment representing the XPath syntax. The fragment is separated from the URL part by a pound sign (#). The following lines are examples of XDBURIType:

/home/oe/doc1.xml
/home/oe/doc1.xml#/orders/order_item

DBURIType DBURIType can be used to store DBURIRef values, which reference data inside the database. Storing DBURIRef values lets you reference data stored inside or outside the database and access the data consistently.

DBURIRef values use an XPath-like representation to reference data inside the database. If you imagine the database as an XML tree, then you would see the tables, rows, and columns as elements in the XML document. For example, the sample human resources user hr would see the following XML tree:

<HR> 
  <EMPLOYEES> 
    <ROW> 
      <EMPLOYEE_ID>205</EMPLOYEE_ID> 
      <LAST_NAME>Higgins</LAST_NAME> 
      <SALARY>12000</SALARY> 
      .. <!-- other columns --> 
    </ROW> 
    ... <!-- other rows --> 
  </EMPLOYEES> 
  <!-- other tables..--> 
</HR> 
<!-- other user schemas on which you have some privilege on..--> 

The DBURIRef is an XPath expression over this virtual XML document. So to reference the SALARY value in the EMPLOYEES table for the employee with employee number 205, we can write a DBURIRef as,

/HR/EMPLOYEES/ROW[EMPLOYEE_ID=205]/SALARY 

Using this model, you can reference data stored in CLOB columns or other columns and expose them as URLs to the external world.

URIFactory Package

Oracle also provides the URIFactory package, which can create and return instances of the various subtypes of the URITypes. The package analyzes the URL string, identifies the type of URL (HTTP, DBURI, and so on), and creates an instance of the subtype. To create a DBURI instance, the URL must start with the prefix /oradb. For example, URIFactory.getURI('/oradb/HR/EMPLOYEES') would create a DBURIType instance and URIFactory.getUri('/sys/schema') would create an XDBURIType instance.

SDO_GEOMETRY

The geometric description of a spatial object is stored in a single row, in a single column of object type SDO_GEOMETRY in a user-defined table. Any table that has a column of type SDO_GEOMETRY must have another column, or set of columns, that defines a unique primary key for that table. Tables of this sort are sometimes called geometry tables.

The SDO_GEOMETRY object type has the following definition:

CREATE TYPE SDO_GEOMETRY AS OBJECT (
   sgo_gtype        NUMBER, 
   sdo_srid         NUMBER,
   sdo_point        SDO_POINT_TYPE,
   sdo_elem_info    SDO_ELEM_INFO_ARRAY,
   sdo_ordinates    SDO_ORDINATE_ARRAY);

SDO_TOPO_GEOMETRY

This type describes a topology geometry, which is stored in a single row, in a single column of object type SDO_TOPO_GEOMETRY in a user-defined table.

The SDO_TOPO_GEOMETRY object type has the following definition:

CREATE TYPE SDO_TOPO_GEOMETRY AS OBJECT (
   tg_type          NUMBER, 
   tg_id            NUMBER,
   tg_layer_id      NUMBER,
   topology_id      NUMBER);

SDO_GEORASTER

In the GeoRaster object-relational model, a raster grid or image object is stored in a single row, in a single column of object type SDO_GEORASTER in a user-defined table. Tables of this sort are called GeoRaster tables.

The SDO_GEORASTER object type has the following definition:

CREATE TYPE SDO_GEORASTER AS OBJECT (
   rasterType      NUMBER,
   spatialExtent   SDO_GEOMETRY,
   rasterDataTable VARCHAR2(32),
   rasterID        NUMBER,
   metadata        XMLType);

ORDAudio

The ORDAUDIO object type supports the storage and management of audio data.

ORDImage

The ORDIMAGE object type supports the storage and management of image data.

ORDImageSignature

The ORDImageSignature object type supports a compact representation of the color, texture, and shape information of image data.

ORDVideo

The ORDVIDEO object type supports the storage and management of video data.

ORDDoc

The ORDDOC object type supports storage and management of any type of media data, including audio, image and video data. Use this type when you want all media to be stored in a single column.

The following datatypes provide compliance with the ISO-IEC 13249-5 Still Image standard, commonly referred to as SQL/MM StillImage.

SI_StillImage

The SI_StillImage object type represents digital images with inherent image characteristics such as height, width, and format.

SI_Color

The SI_Color object type encapsulates color values.

SI_AverageColor

The SI_AverageColor object type represents a feature that characterizes an image by its average color.

SI_ColorHistogram

The SI_ColorHistogram object type represents a feature that characterizes an image by the relative frequencies of the colors exhibited by samples of the raw image.

SI_PositionalColor

Given an image divided into n by m rectangles, the SI_PositionalColor object type represents the feature that characterizes an image by the n by m most significant colors of the rectangles.

SI_Texture

The SI_Texture object type represents a feature that characterizes an image by the size of repeating items (coarseness), brightness variations (contrast), and predominant direction (directionality).

SI_FeatureList

The SI_FeatureList object type is a list containing up to four of the image features represented by the preceding object types (SI_AverageColor, SI_ColorHistogram, SI_PositionalColor, and SI_Texture), where each feature is associated with a feature weight.

Expression

Expression Filter uses a virtual datatype called Expression to manage and evaluate conditional expressions as data in database tables. The Expression Filter creates a column of Expression datatype from a VARCHAR2 column by assigning an attribute set to the column. This assignment enables a data constraint that ensures the validity of expressions stored in the column.

You can define conditions using the EVALUATE operator on an Expression datatype to evaluate the expressions stored in a column for some data. If you are using Enterprise Edition, then you can also define an Expression Filter index on a column of Expression datatype to process queries using the EVALUATE operator.