0.4.0 — every name means one thing

No new features. One structural fix, and it breaks an import path.

Changed

• The fold family moved from cjkfold.normalize to cjkfold.folding; the two
  writes moved to cjkfold.rewrite. Nothing about the functions changed —
  from cjkfold import fold, normalize, strip_invisible behaves exactly as it did.

  Why: a function named normalize cannot live in a module named normalize without
  one of the two losing its name, and import a.b as c means c = a.b. So
  import cjkfold.normalize as m handed you the function, and m.fold raised
  AttributeError. Every name now means one thing: cjkfold.normalize is the
  function, cjkfold.folding is the module. (af57ad7)

  What breaks: from cjkfold.normalize import fold — a path nothing ever
  documented, and it now fails with a plain ModuleNotFoundError instead of an
  AttributeError from something that was never a module. Every documented import
  is untouched.

Tests

• scripts/acceptance.py gained import-path checks, including one that the old
  path is gone. (af57ad7)

0.3.0 — the broken class, directory walking

Added

• The broken class. scan now reports U+FFFD and the private use areas
  alongside look-alikes, compat characters and invisible ones — damage that cannot
  be repaired, only reported. A private use code point earns its place in the
  report precisely because nothing defines what it looks like: the font doing the
  rendering decides, and it can come out as a box, as nothing at all, or as a
  glyph that means something else. (b17902d)
• scan walks directories. A corpus is a tree, not an argument list.
  Dot-entries stay out, so .git does not end up in the report; a non-text file
  found by walking is skipped and counted, while a non-text file you named is
  still an error — pointing at a PDF is a mistake, walking past one is not.
  (b17902d)
• The scan footer prints once per run instead of once per file, which starts to
  matter the moment you point it at a directory. (b17902d)

Tests

• The four compat characters the roadmap named are pinned with measured values:
  full-width Ａ → A, ideographic space → space, U+F900 → 豈, NBSP → space.
  (b17902d)

No change to fold(), to check, or to the equivalence table. Exit codes and
check output are unchanged.

0.2.0 — scan, normalize, strip_invisible

Added

• cjkfold scan — every character that breaks a comparison, in one read-only
  pass: the look-alikes, the NFKC family fold() refuses to touch, and the
  invisible family. Each finding says what to do with it — fold, nfkc,
  strip, or manual, where manual means do not touch this without looking.
  It exits 0 by default, because it is a report and not a gate: a gate that fails
  on a load-bearing joiner fails on ordinary emoji. --strict gates anyway,
  --json feeds a pipeline, --classes narrows the report. (c5be88a, 5835dbf)
• normalize() and strip_invisible() — the two explicit writes. The
  first applies the NFKC family with full-width punctuation spared by default; the
  second deletes invisible characters, safe by default, because half of that
  family carries meaning — a family of three is five code points, and deleting the
  joiners leaves three separate people. Neither is exposed as a subcommand: fold -w
  is already an in-place rewrite, and two more destructive commands would be a gun
  pointed at a corpus. (a733ef5)
• Finding, scan_stats() and is_punctuation() in the public API.
  (2bd05e6, 5835dbf)

Changed

• scan's whitelist lookup is now a single set test instead of a dict lookup plus
  four range comparisons: 0.42s → 0.29s over 640,000 characters. (f6c5f4f)

Docs

• README: "The characters that never show up" and "The other two rewrites",
  both carrying output from real runs rather than hand-copied examples; the 中文简述
  follows. The check example also gained the summary line it had been missing —
  the reason it was missing is that it had been copied by hand, which is exactly
  what the new section refuses to do. (e9c4d68, and this release)
• tests/fixtures/ — the three files the roadmap asked for, with check's output
  frozen byte for byte. (1a468e5)

No change to fold(), to check, or to the equivalence table. Text that folded
correctly under 0.1.2 folds identically under 0.2.0.