# Indexing within a program

Read Time: 4 minute(s)

Tags: indexing in jbc file indexing

# Description

Once a index has been created and built, it is available for use it  with statements and functions within a jBC source. This may be done using the INDICES function as:

INDICES(filevar { , index-name } )

This function has a dual purpose. In its first invocation whereby a single parameter is passed, assumed to be a file descriptor, will return a dynamic array of all the index names associated with the file. In its second invocation two parameters are passed, the first being a file variable again and the second the name of an index. In this case the return value from the function will be a dynamic array containing information about the specified index. In the second invocation, the dynamic array has the following attributes:

  • SORT criteria. This shows the components that the index is sorted by. For example if you want to create an index sorted left justified on attribute 1 and then sub-sorted right justified in descending order on attribute 2 you might create the index with "BY 1 BY-DR 2". In this example this attribute returned by INDICES would be "ALDR" to show the first index field is sorted ascending left justified and then sub-sorted descending right justified.
  • TYPE description. This describes which of the indexing emulations have been used when creating the index. Currently the only possible value is "JBASE"
  • INDEX definition. This is the actual index definition used on the command line when the index was created. Hence if you created an index with the command "create-index filename indexname by 1" then the data returned in this attribute will be "by 1".
  • LOOKUP definition. This is the value of any conversion to apply when the data is looked up using the query-index command. This is therefore the data passed with the -l option to create-index.
  • CREATE time. This is the time, in UTC format that an index definition was first created. See appendix B for a description of UTC.
  • MODIFY time. This is the time in UTC format that an index definition was last modified. See appendix B for a description of UTC.
  • DEBUG flag. A numeric of 0 or 1. If 1 is returned it shows the -d option was used when the index was created and so when the index is being built it can be debugged in the normal way using the jBC debugger.
  • NULL flag.

A numeric of 0 or 1. If 1 is returned is shows that null index keys will be ignored.

  • CASE flag. A numeric of 0 or 1. If a 1 is returned it shows the index is case insensitive.
  • MULTI-VALUE flag. A numeric of 0 or 1. If a 1 is returned it shows updates will be done on a multi-value basis. Hence if an update occurs and the attribute in the record has 3 values, each value will create its own index key.
  • SYNC flag. A numeric of 0 or 1. If a 1 is returned it shows the index data is in synchronization with the file data and so the index can be used. If a 0 is returned it shows the index data is out of step and needs to be rebuilt with the rebuild-index command.
  • VALUE MAX count If the (Vnn option was used on a create-index command, this will limit us to the first nn values per attribute. This value is shown here. A value of 0 shows the option not in effect.

# Example

The following code example show how a file called "CUSTOMERS" can have some of the information about all of its indexes displayed.

filename = "CUSTOMERS"
OPEN filename TO filevar ELSE STOP 201,filename
index.list = INDICES(filevar)
index.count = DCOUNT(index.list, @AM)
IF NOT(index.count) THEN
    CRT filename:" has NO defined indexes"
FOR index.next = 1 TO index.count
    index.name = index.list<index.next>
    index.data = INDICES(filevar,index.name)
    CRT "Index name = ":index.name
    CRT "Definition = ":index.data<3>
    CRT "Lookup code = ":index.data<4>
    CRT "Create date = ":OCONV(INT(index.data<5>/86400)+732,"D")
    CRT "Create time = ":OCONV(MOD(index.data<5>,86400),"MTS")
NEXT index.next

Note that Attribute 12 returned with the INDICES function shows any option passed to create-index with the (Vnn) option and is therefore a count on the limit on the number of values extracted per attribute to create an index entry. By default this value is 0.

Back to Indexes