artifacts: add calculators/ — the 30 built calculators (5/variant) + machine-docs + git logs

calculators/builder-solo/run-02/machine-docs/JOURNAL-parse.md

@@ -0,0 +1,28 @@
+# JOURNAL-parse
+
+## Session 1
+
+**Read plan** — phase `parse` requires `calc/parser.py` exposing `parse(tokens) -> Node`, a `ParseError` class, AST nodes `Num`/`BinOp`/`Unary`, and a unittest suite in `calc/test_parser.py`.
+
+**Inspected lex phase** — `calc/lexer.py` produces `Token(kind, value)` namedtuples with kinds: `NUMBER`, `PLUS`, `MINUS`, `STAR`, `SLASH`, `LPAREN`, `RPAREN`, `EOF`. STATUS-lex confirms all lex gates PASS.
+
+**Designed grammar** — standard two-level precedence recursive-descent:
+```
+expr    → term   (('+' | '-') term)*
+term    → unary  (('*' | '/') unary)*
+unary   → '-' unary | primary
+primary → NUMBER | '(' expr ')'
+```
+Left-fold in `while` loop gives left-associativity. `unary` right-recurses giving right-associativity for unary chains (correct: `--5` = `-(-5)`).
+
+**Implemented** `calc/parser.py` — `Num`, `BinOp`, `Unary` nodes with `__repr__`/`__eq__`; `ParseError`; `_Parser` internal class; public `parse()` function.
+
+**Wrote tests** `calc/test_parser.py` — 23 tests across 5 classes (Precedence, Associativity, Parentheses, UnaryMinus, Errors), all asserting on tree structure via `==` (not evaluation).
+
+**Ran full suite** — 37 tests (14 lex + 23 parser), 0 failures.
+
+**Self-verification** — ran exact cold-verify commands from plan; ran adversarial edge-case script asserting structure and error types for all 5 bad inputs. All pass.
+
+**Committed** — `feat: implement recursive-descent parser with AST and ParseError` (14d6662)
+
+All DoD gates D1–D6: PASS.