This is an old revision of the document!
The CREATE TABLE statement declares a table in an Omnidex Environment File. Tables that are declared in an environment can then be indexed and retrieved using SQL statements. An Omnidex table can point to an underlying Oracle, SQL Server, MySQL or ODBC table or view. An Omnidex table can also point to one or more raw data files that match the column layout of the table.
The basic CREATE TABLE statement records information about the underlying table or raw data files in the Omnidex Environment Catalog, but it does not actually create an underlying relational table or raw data file. Normally, the relational table or view file is created using the relational database's tools. In the case of raw data files, the files are usually received from a data provider, and application or an ETL tool. The CREATE TABLE statement simply stores meta-data about the table in the Omnidex Environment File. Omnidex does support an exception to this rule for the creation of temporary tables.
The CREATE TABLE statement declares the characteristics of the table. At a minimum, each table has a name and instructions for accessing the physical data. Omnidex supports configuration settings for each table, including settings to for delimited files, data caches and index maintenance. Omnidex also provides rollup tables which pre-aggregate another table in the database and automatically optimize queries against the table.
The CREATE TABLE statement specifies the columns for a table. At a minimum, each column has a name and a datatype; however, it can also have other characteristics as well. Omnidex allows a column to have a different name than the one used in the underlying relational database. Omnidex allows the declaration of column usages that describe the purpose of the column. Omnidex also allows expression-based columns that are the result of a SQL expression, even if it does not exist in the underlying database or data file.
The CREATE TABLE statement specifies the constraints for a table. The most common constraints are primary and foreign constraints, used to declare parent-child relationships between tables. Omnidex allows constraints to be declared on both relational tables and raw data files, allowing either to be described using a full relational model. Omnidex allows these constraints to be declared, but it does not verify or enforce these constraints.
Like most databases, Omnidex allows indexes to be declared using the CREATE INDEX statement. For convenience, Omnidex also allows indexes to be declared using the CREATE TABLE statement. In most applications, many (if not all) of the columns are indexed with Omnidex, so it is convenient to be simply declare each index alongside the column declaration.
CREATE [ < GLOBAL | LOCAL > ] TEMPORARY ] TABLE table [ NODE node] [ OPTIONS "options" ] [ PHYSICAL " ] [ AUTOFILTER "criteria" ] [ DATA_CACHING < cache_size | DYNAMIC | NONE > ] [ PARTITION_BY "criteria" ] [ INDEX_MAINTENANCE <API | DBMS> ] ( column_specification, [ constraint_specification, ] [ index_specification, ] [ index_group_specification ] ) [ AS statement ] [ IN filename ] [ WITH options ]
column_name column_datatype [ PHYSICAL "physical_name" ] [ USAGE usage ] [ FORMAT format ] [ inline_constraint_specification ] [ inline_index_specification ] [ AS expression ],
/* Omnidex Index Types */ [< OMNIDEX | QUICKTEXT | FULLTEXT | CUSTOM | NATIVE> [ INDEX ] /* column options */ [DEFAULT < literal | niladic-function | NULL >] [USAGE usage] [FORMAT format] [ , ] /* separator between column and constraint definitions */ /* Inline Constraint Specifications - native indexes and Omnidex indexes */ [CONSTRAINT constraint] <[NOT] NULL | UNIQUE [KEY] | DISTINCT [KEY] | PRIMARY [KEY] | [FOREIGN [KEY]] REFERENCES table(column[, column …]) [PREJOIN]>] /* Inline Omnidex custom index options */ [KEYWORDING] [PROXIMITY] [<CASE_INSENSITIVE | CASE_SENSITIVE>][STANDALONE] [BITMAP][EXCLUDED_WORDS][PHONETIC] [PREJOIN table][<RECORD_SPECIFIC | RECORD_COMPLEX>]] /* Inline Omnidex pseudo or derived columns */ [AS “select_item”] /* Separately defined Constraints and Omnidex Indexes */ [[,] [CONSTRAINT constraint] < UNIQUE [KEY] (column_spec [, column_spec …]) | DISTINCT [KEY] (column_spec [, column_spec …]) | PRIMARY [KEY] (column_spec [, column_spec …]) | FOREIGN [KEY] (column_spec [, column_spec …]) REFERENCES table(column_spec [, column_spec …] [PREJOIN] > ] [[,] <OMNIDEX | QUICKTEXT | FULLTEXT | CUSTOM | NATIVE> [INDEX] index (<column | substring> [, <column | substring>…]) [KEYWORDING] [PROXIMITY] [ <CASE_INSENSITIVE | CASE_SENSITIVE> ] [STANDALONE][BITMAP][EXCLUDED_WORDS][PHONETIC] [PREJOIN table][<RECORD_SPECIFIC | RECORD_COMPLEX > ] ] /* place the column in a previously declared Omnidex index group */ [[,] INDEX GROUP [(owner)]group (index, index [, index ...]) )] [AS “sql_statement”] /* Specifics the physical Omnidex Environment file. This has to be the last clause */ [< IN “env_filespec” | ON [INSTANCE] instance >] [WITH options]
CREATE ENVIRONMENT and CREATE DATABASE have to be specified before specifying CREATE TABLE.
The IN “filespec.xml” clause has to be specified after the column definitions.
Tables must be declared in order of their constraints.
Parent tables should be declared before children tables.
Most CREATE TABLE statements will typically look like the following example where Omnidex indexing is specified after the column_name and column_type.
CREATE TABLE mytable PHYSICAL "mydir/mytable.dat" ( user_id INTEGER OMNIDEX, region CHARACTER(2) OMNIDEX, full_name CHARACTER(30) QUICKTEXT, address CHARACTER(60) QUICKTEXT, city CHARACTER(30) QUICKTEXT, state CHARACTER(2) OMNIDEX );
create table "HOUSEHOLDS" physical "dat\households*.dat" ( "HOUSEHOLD" character(12), "ADDRESS" character(50), "CITY" character(28), "STATE" character(2), "ZIP" character(5), "COUNTRY" character(2), constraint HOUSEHOLD_HOUSEHOLD_PK primay ("HOUSEHOLD"), constraint HOUSEHOLD_STATE_fk FOREIGN ("STATE") references "states", constraint HOUSEHOLDS_COUNTRY_FK foreign ("COUNTRY") references "COUNTRIES" ) in "simple.xml";
Here is the same CREATE TABLE statement as above but with the Omnidex Index type specified for each column.
create table "HOUSEHOLDS" physical "dat\households*.dat" ( "HOUSEHOLD" character(12) omnidex, "ADDRESS" character(50) quicktext, "CITY" character(28) quicktext, "STATE" character(2) omnidex, "ZIP" character(5) omnidex, "COUNTRY" character(2) omnidex, constraint HOUSEHOLD_HOUSEHOLD_PK primay ("HOUSEHOLD"), constraint HOUSEHOLD_STATE_fk FOREIGN ("STATE") references "states", constraint HOUSEHOLDS_COUNTRY_FK foreign ("COUNTRY") references "COUNTRIES" ) in "simple.xml";
Here is the same CREATE TABLE statement as above but with the Omnidex Index type specified for each column.
Additionally, it creates an Omnidex Composite index and combines the STATE and CITY columns into a single index.
create table "HOUSEHOLDS" physical "dat\households*.dat" ( "HOUSEHOLD" character(12) omnidex, "ADDRESS" character(50) quicktext, "CITY" character(28) quicktext, "STATE" character(2) omnidex, "ZIP" character(5) omnidex, "COUNTRY" character(2) omnidex, constraint HOUSEHOLD_HOUSEHOLD_PK primay ("HOUSEHOLD"), constraint HOUSEHOLD_STATE_fk FOREIGN ("STATE") references "states", constraint HOUSEHOLDS_COUNTRY_FK foreign ("COUNTRY") references "COUNTRIES" omnidex index "HSHD_STATE_CITY" ("STATE","CITY") ) in "simple.xml";
There are three types of Omnidex indexes called Omnidex, QuickText, and FullText.
The following table shows the key types of retrieval properties of these Omnidex index types.
Omnidex | QuickText | FullText | Custom | |
---|---|---|---|---|
Criteria | Yes | Yes | Yes | |
Counts | Yes | Yes | Yes | |
Sum, Average, Min, Max | Yes | No | No | |
Table Joins | Yes | No | No | |
Group By | Yes | No | No | |
Order By | Yes | No | No | |
Geographic Searches | Yes | No | No | |
Textual Searches | No | Yes | Yes | |
Proximity Searches | No | No | Yes | |
Relevancy Scoring | No | No | Yes | |
Indexing Overhead | Low | Low | High |
In addition to the three basic Omnidex Index Types: Omnidex, QuickText, and FullText, there is a Custom Index Type that can be used to specify advanced indexing options.
Number | Data Type | Length |
---|---|---|
CHAR(ACTER) | # of characters | |
C STRING | # of characters | |
VARCHAR | # of characters | |
CLOB | # of characters | |
NCHAR | # of characters * 2 | |
NC STRING | # of characters | |
NVARCHAR | # of characters | |
NCLOB | (# of characters | |
OMNIDEX VARCHAR | # of characters | |
OMNIDEX CLOB | # of characters | |
[SIGNED] TINYINT | 1 | |
UNSIGNED TINYINT | 1 | |
[SIGNED] SMALLINT | 2 | |
UNSIGNED SMALLINT | 2 | |
[SIGNED] INTEGER | 1, 2, 4, 8 | |
UNSIGNED INTEGER | 1, 2, 4, 8 | |
[SIGNED] BIGINT | 8 | |
UNSIGNED BIGINT | 8 | |
FLOAT | 4, 8 | |
DOUBLE | 8 | |
DATE | 10 | |
ODBC DATE | 6 | |
DB2 DATE | 6 | |
INFORMIX DATE | 4 | |
ASCII DATE | 6, 8 | 6, 8 (optional, default 8) |
OMNIDEX DATE | 1 - 4 | 2 - 8 (optional, default 8) |
TIME | 11 | |
ODBC TIME | 6 | |
DB2 | TIME 6 | |
OMNIDEX TIME | 1 - 4 | 2 - 8 (optional, default 8) |
DATETIME | 22 | |
ORACLE DATETIME | 7 | |
ODBC DATETIME | 6 | |
DB2 DATETIME | 16 | |
INFORMIX DATETIME | 24 | |
C DATETIME | 4 | |
OMNIDEX DATETIME | 1 - 8 | 2 - 16 (optional, default 16) |
BLOB | # of bytes | |
OMNIDEX BLOB | # of bytes |