CREXX Logical AST Specification

Page Status: This page is ready for review for Phase 0 (PoC). It may have errors and be changed based on feedback and implementation experience

The AST specification covers all phases of the project and all REXX levels, this means that a single language processor (compiler backend) will be able to be used across the project.

Obviously the initial specification will only handle Phase 0 scope and language levels A and B but the intention is that this will be extended (where possible without breaking existing code - "backwards compatible")

The AST notation is described here, essentially:

(root child1 child2 child3) - simple tree with 3 children
(root child1 (subroot subchild1 subshild2) child3) - a tree with a subtree

AST Node List

Node Type	Description	Terminal
ABS_POS	Absolute Pos for Parsing	No
ADDRESS	Address Instruction	No
ARG	Argument input for Parsing	Yes
ASSIGN	Assign Instruction	No
BY	BY part of REPEAT part of DO	No
CALL	Call Instruction	No
CONST_SYMBOL	Constant Symbol	Yes
DO	Do Instruction	No
ENVIRONMENT	Environment for Address	Yes
ERROR	Error Marker	No
FOR	FOR part of REPEAT part of DO	No
FUNCTION	Function	No
IF	If Instruction	No
INSTRUCTIONS	Instruction List	No
ITERATE	Iterate Instruction	No
LABEL	Label	Yes
LEAVE	Leave Instruction	No
NUMBER	Number	Yes
OP_ADD	Add/Subtraction Op	No
OP_AND	And Op	No
OP_COMPARE	Compare Op	No
OP_CONCAT	Concat Op	No
OP_MULT	Multiply/divide OP	No
OP_OR	Or Op	No
OP_POWER	Power Op	No
OP_PREFIX	Prefix Op (+, -, )	No
OP_SCONCAT	Concat with space Op	No
OPTIONS	Options for Parsing	No
PARSE	Parse Instruction	No
PATTERN	Pattern for Parsing	No
PROCEDURE	Procedure	No
PROGRAM_FILE	AST Root for a file	No
PULL	Pull input for Parsing	Yes
REL_POS	Relative Pos for Parsing	No
REPEAT	Repeat part of DO loop	No
RETURN	Return Instruction	No
REXX	Language Level and Options	No
SAY	Say Instruction	No
SIGN	Rel pos direction for Parsing	Yes
STRING	String	Yes
TARGET	Target for Parsing	No
TEMPLATES	Template List for Parsing	No
TO	TO part of REPEAT part of DO	No
TOKEN	Generic token	Yes
UPPER	Upper Option for Parsing	Yes
VAR_SYMBOL	Variable	Yes

Format of Non-Terminal Nodes

File Scope

(PROGRAM_FILE REXX INSTRUCTIONS?) 

Language Options

(REXX level:CONST_SYMBOL options:CONST_SYMBOL*)

Error

(ERROR TOKEN+)

This node is not a actual token instead it is inserted into token stream, the actual offending TOKEN(s) are added as children

Instructions

(INSTRUCTIONS instruction*)

Where instruction is one of:

• ADDRESS, ASSIGN, CALL, DO, IF, INSTRUCTIONS, ITERATE, LABEL, LEAVE, PARSE, PROCEDURE, RETURN, SAY

Expressions

Expression (expr) nodes are one of:

expr <-
  (OP_OR expr expr) /
  (OP_AND expr expr) /
  (OP_COMPARE expr expr) /
  (OP_CONCAT expr expr) /
  (OP_SCONCAT expr expr) /
  (OP_ADD expr expr) /
  (OP_MULT expr expr) /
  (OP_POWER expr expr) /
  (OP_PREFIX expr) /
  (FUNCTION expr*) /
  CONST_SYMBOL /
  VAR_SYMBOL /
  NUMBER /
  STRING

Address

(ADDRESS ENVIRONMENT? expr?)

Assignment

(ASSIGN VAR_SYMBOL expr)

Call

(CALL CONST_SYMBOL expr*)

Do

(DO (REPEAT assignment (TO expr)? (BY expr)? (FOR expr)?) instructions*);

Note that simple DO / END maps to (INSTRUCTIONS instruction*)

If

(IF expr true:INSTRUCTIONS false:INSTRUCTIONS?)

Iterate

(ITERATE VAR_SYMBOL?)

Label

LABEL

Leave

(LEAVE VAR_SYMBOL?)

Parse

(PARSE (OPTIONS UPPER?) in (TEMPLATES template+)

in <- ARG / PULL;
template <- target / pattern / abs_pos / rel_pos;
target <- (TARGET VAR_SYMBOL?);
pattern <- (PATTERN STRING/VAR_SYMBOL);
abs_pos <- (ABS_POS NUMBER/VAR_SYMBOL);
rel_pos <- (REL_POS SIGN NUMBER/VAR_SYMBOL);

Procedure

(PROCEDURE LABEL instructions);

Return

(RETURN expr?)

Say

(SAY expr?)