• PQL is our language to create algebra expressions à Create a logical query model (quite similar to the relational algebra)
• To allow an easy extension, the PQL parser knows nothing about concrete operators
• The parser only knows about (abstract) logical operators and configurations
• Concrete operators are pluged into the parser
• For this:
  • An annotation model
  • An automatic loading mechanismen (no need to write services for operators)

Logical Operator Model

• Schema provides (meta-)information about the stream:
  • What stream type (tuple, key value, xml …)
  • Out of order
  • For tuple: set of attributes
  • Constraints
  • …

Annotation Model

• @LogicalOperator: State ILogicalOperator implementation as PQL operator
• maxInputPorts, minInputPorts: How many input must be bound, can be bound
• Name: Name of the operator in PQL
• Helper for user interface
  • Doc: A text describing the operator function
  • Url: Link to documentation
• Category: Kind of operator

• Typically, each logical operator has a set of properties (e.g. in SELECT the predicate)
• A large set of specialized parameter handler (e.g.PredicateParameter)
• Can (of course) be extended ;-)
• The PQL parser does some preprocessing and delegates parameter handling to the parameter classes

Different predefined parameter classes

PQL and Parameter

• Text inside of „“ or ‚‘ are directly parsed, e.g. StringParameter

• Numbers/Boolean are parsed as types, e.g. IntegerParameter

Usage in logical operator

Predicate Parameter

Only one handler for all predicate types

• If isList = true, then input will be interpreted as List, must be enclosed by „[" and „]“
• Create List of elements
• No new parameter handler is needed!
• Example Rename:
• 
• Example Schemacreation (here from AbstractAccessAO)
• 
• Example Attributereference (here from ProjectAO)
•