# JOURNAL-parse

## Implementation notes

Grammar chosen (standard arithmetic precedence):
```
expr    := term   (('+' | '-') term)*
term    := unary  (('*' | '/') unary)*
unary   := '-' unary | primary
primary := NUMBER | '(' expr ')'
```

`unary` is right-recursive, which gives right-associativity to stacked unary minuses (e.g. `--5` → `Unary('-', Unary('-', Num(5)))`). This is standard.

`expr` and `term` are iterative loops (not recursive), so same-level operators are naturally left-associative.

The `ParseError` class is defined in `parser.py` (not lexer.py) since it's the parser's concern. All five D5 error cases raise `ParseError`, not a generic exception.

Tests assert on exact tree structure using `==` on dataclasses, not on evaluation results, per the plan's requirement.