test sweep: cover EscapeParser + EscapeEmitter (T-91)
Adds test/terminal/escape/parser_test.dart — 70 unit tests covering
the parser's dispatch surface end-to-end:
- single-byte controls (BEL, BS, HT, LF/VT/FF, CR, SO, SI),
- ESC sequences (D, E, H, M, =, >, 7, 8, ( name, ) name, unknown),
- CSI cursor moves (A/B/C/D/E/F/G/H/d/f) with default + 0-as-1
fallback semantics,
- erase / scroll / line-insert/delete / chars (J, K, L, M, P, S, T,
X, @, b, g, r),
- device attributes (c / >c / =c) and DSR (5, 6),
- window manipulation (CSI 8 t resize, CSI 18 t sendSize, ignored
no-op codes, malformed CSI 8 t),
- mode set/reset (h/l, ? prefix for DEC modes — covering ?1, ?3,
?5, ?6, ?7, ?9, ?12, ?25, ?47, ?66, ?1000, ?1002, ?1003, ?1004,
?1005, ?1006, ?1007, ?1015, ?1047, ?1048, ?1049, ?2004, +
unknown fallback),
- SGR styling (resets, set/unset for every attr, 16-colour
foreground + background, 256-colour, 24-bit RGB, 39 / 49
resets, unknown → unsupportedStyle),
- OSC 0/1/2 (BEL- and ST-terminated), unknown OSC, incomplete
sequence held back across writes,
- unknown CSI final byte → unknownCSI,
- token bookkeeping (tokenBegin / tokenEnd advance with consumed
bytes).
Plus EscapeEmitter — every reply string format (primary /
secondary / tertiary device attributes, operating status, cursor
position, bracketed paste, size).
Coverage delta:
- parser.dart: 0 / 462 → 504 / 514 (98.1%; the remaining 10 lines
sit inside the `// ignore: dead_code` SGR loop, which I'm
surfacing for separate review rather than extending tests
around).
- emitter.dart: 1 / 11 → 11 / 11.
- Total project: 43.20% → 49.38%; coverage_floor bumped 43 → 49.
Two real source-code issues found while writing tests, fixed in
the same commit:
1. Swapped docstrings on `_escHandleSetAppKeypadMode` /
`_escHandleResetAppKeypadMode`. The function names + dispatch
table + bodies all match the VT spec correctly (ESC = enables,
ESC > disables); only the doc-comments were swapped. Now read
the right way around.
2. `case 10061000:` in `_setDecMode` was unreachable (no DEC mode
has that value). Almost certainly a typo where `case 1006:`
meant to glue onto `case 1000:` but a newline went missing.
Mode 1006 is already handled separately at its own clause as
`MouseReportMode.sgr`. Removed.
Co-Authored-By: Claude <noreply@anthropic.com>
jpmschweitzerandClaude
9331055b44