fix: collect a citation to an extensionless file

scripts/profile_sync.py

@@ -345,8 +345,27 @@ def _anchor_tokens(entry: dict) -> list[str]:
     return tokens
 
 
+# Files a repository carries without an extension and that prose cites by
+# name. Widening the pattern to any dotless word instead would swallow the
+# hex addresses and the constants profiles write the same way: ALTERNATEROMS,
+# F000, C000 all read as `name:number`.
+PROSE_EXTENSIONLESS = (
+    "Makefile",
+    "makefile",
+    "GNUmakefile",
+    "CMakeLists",
+    "Kconfig",
+    "Dockerfile",
+    "configure",
+    "README",
+    "LICENSE",
+    "COPYING",
+    "CHANGELOG",
+)
 PROSE_CITE_RE = re.compile(
-    r"(?P<path>[A-Za-z0-9_][\w./+-]*\.[A-Za-z]\w*):(?P<range>\d+(?:-\d+)?)"
+    r"(?P<path>[A-Za-z0-9_][\w./+-]*\.[A-Za-z]\w*"
+    r"|(?:[\w./+-]*/)?(?:" + "|".join(PROSE_EXTENSIONLESS) + r"))"
+    r":(?P<range>\d+(?:-\d+)?)"
 )
 PROSE_CONT_RE = re.compile(r",(?P<range>\d+(?:-\d+)?)(?![\w-])(?!\.\d)")
 # A spaced continuation is accepted only for a range: `x.c:55-71, 80-147`

tests/test_profile_sync.py

@@ -504,6 +504,26 @@ class TestExternalCitation(unittest.TestCase):
             profile_sync.is_external_citation("EmuDeck emuDeckares.sh")
         )
 
+    def test_a_file_without_an_extension_is_still_a_citation(self):
+        """A repository carries files with no dot in their name.
+
+        The pattern required a dotted extension, so `Makefile:36-37` and
+        `README:27` were never collected and rotted unnoticed while every
+        dotted citation was anchored.
+        """
+        for text in ("Makefile:36-37", "src/Makefile:36", "README:27",
+                     "configure:120"):
+            with self.subTest(text=text):
+                match = profile_sync.PROSE_CITE_RE.search(text)
+                self.assertIsNotNone(match, text)
+                self.assertEqual(match.group("path"), text.split(":")[0])
+
+    def test_a_constant_or_an_address_is_not_a_citation(self):
+        """Widening the pattern to any dotless word swallows these."""
+        for text in ("ALTERNATEROMS:23", "F000:12", "C000:4"):
+            with self.subTest(text=text):
+                self.assertIsNone(profile_sync.PROSE_CITE_RE.search(text), text)
+
     def test_plain_paths_are_not_citations(self):
         for path in ("src/midi/mt32.cpp", "libretro.c", "a/b/c.h"):
             self.assertFalse(profile_sync.is_external_citation(path), path)