Online documentation for DictionaryFilter
Generated 2006-03-17
Component allowing the specification of rules used to filter
dictionary objects.
Description
Many components produce complex messages called dictionaries.
Examples include RSSClient and EmailReceiver. This component
can be used to filter such messages. For example, an email message
might be filtered so that only emails from a certain
person are accepted. Details of filtering are specified using a set of
rules.
If you are interested in filtering
emails, then you should look at EmailFilter, a version of
this component that
provides simple functionality to specify filters
using safe lists and blocked lists.
Usage
If your filter is configured with a number of rules, then
any dictionary/message placed onto property
inputMessages will be filtered using these rules.
Dictionaries that pass this filtering stage will be
placed onto property acceptedMessages, and
those that fail will be placed onto property
rejectedMessages.
Configuration
Configuration using default rule types
This component is currently configured to support
the specification of filters using the following
types of
rule:
- <field name> contains <text>
- <field name> doesn't contain <text>
- <field name> is <text>
- <field name> isn't <text>
- <field name> starts with <text>
- <field name> ends with <text>
To configure the filter, you provide instances of
these rule types to properties
allRules and anyRules. For example,
if you wanted a filter allowing any dictionary through
which has a from field containing sre@cs.nott.ac.uk
but not
stefan_egglestone@hotmail.com, you would set
allRules to
{from contains sre@cs.nott.ac.uk,from doesn't contain stefan_egglestone@hotmail.com}
Configuration with user-defined rule types
Hopefully, the default rule types will be sufficient for
the specification of most filters. But, they
can be replaced, with other rule types using different
keywords. Note that this is not an easy process to
perform, requiring knowledge of regular expressions.
If you want to do it, you'll need to modify the following
properties:
- configRegularExpressions - provide regular expressions
that can be used to match your keywords
- configKeywords - provide your new keywords here
- configNegationDefinitions - provide boolean values
indicating whether keyword involves negation or not - eg
keyword "is" doesn't, but keyword "isn't" does.
- configTemplateRegularExpressions - provide
regular expressions that match a rule constructed using
your keywords. Such regular expressions should include
the string text, which will be replaced
by the keyword to generate a valid regular expression, eg
(.)*text(.)*
In each of these properties, you should place an array
of values, with each element in the array corresponding
to one keyword that you are defining.