feat(registry): add strict padding for checks (amperser#1428)

Prevent false positives due to words glued together by hyphens and
apostrophes

- Add test cases for amperser#306 
- Add a `STRICT_WORDS_IN_TEXT` that does not include hyphens or
apostrophes
- Fix the `misc.preferred_forms` check
- Refactor the `lexical_illusions` check to use `STRICT_WORDS_IN_TEXT`

---------

Signed-off-by: drainpixie <121581793+drainpixie@users.noreply.github.com>
Co-authored-by: Tyler J Russell <xtylerjrx@gmail.com>

Author: drainpixie

proselint/checks/lexical_illusions.py

@@ -19,8 +19,8 @@
 
 check = Check(
     check_type=types.ExistenceSimple(
-        pattern=Padding.WORDS_IN_TEXT.format(
-            r"(?<!\\|\-)(\w+(?:\s+\w+){0,3})(\s+\1)+"
+        pattern=Padding.STRICT_WORDS_IN_TEXT.format(
+            r"(\w+(?:\s+\w+){0,3})(\s+\1)+"
         ),
         exceptions=(r"^had had$", r"^that that$"),
     ),

proselint/checks/misc/preferred_forms.py

@@ -21,6 +21,7 @@
 
 check = Check(
     check_type=types.PreferredFormsSimple(
+        padding=Padding.STRICT_WORDS_IN_TEXT,
         items={
             # Obsolete words
             "imprimature": "imprimatur",

proselint/registry/checks/__init__.py

@@ -35,10 +35,17 @@ class Padding(str, Enum):
     """
     NONWORDS_IN_TEXT = r"\B{}\B"
     """Match any position that is not a word boundary around the pattern."""
+    STRICT_WORDS_IN_TEXT = r"(?<![A-Za-z'-]){}(?![A-Za-z'-])"
+    """Lookaround-based `WORDS_IN_TEXT`, prohibiting hyphens and apostrophes."""
 
     def to_offset_from(self, offset: tuple[int, int]) -> tuple[int, int]:
         """Calculate new offset values based on the applied padding."""
-        if self in {Padding.RAW, Padding.SAFE_JOIN, Padding.WORDS_IN_TEXT}:
+        if self in {
+            Padding.RAW,
+            Padding.SAFE_JOIN,
+            Padding.WORDS_IN_TEXT,
+            Padding.STRICT_WORDS_IN_TEXT,
+        }:
             return offset
         return (offset[0] + 1, max(offset[1] - 1, 0))
 

tests/examples.py

@@ -317,12 +317,14 @@
             "It was almost haloween.",
             "He is Chief Justice of the Supreme Court of the United States.",
             "Meantime, I had tea.",
+            "He jumped off of the couch.",
             "In the meanwhile, something happened.",
             "She went to bed; meantime, I had tea.",
         ),
         (
             "Smoke phrase with nothing flagged.",
             "Meanwhile, I had tea.",
+            "A 4th order low-pass with a roll-off of 24 db/octave",
             "In the meantime, something happened.",
             "She went to bed; meanwhile, I had tea."
         ),