fix(job_key): keep the bare basis for an empty title whose URL has no six-digit run (#523)

The mixed-script rule from #514 appended the digest to `basis` as well,
so an empty title with a URL that carries no six-digit run moved from
"acme_https-example-com-jobs-abc-xyz" to "..._5ca8c6". Unreachable from
/scrape, which always passes a title, but a key is a pure function of the
posting and should not move for a case the rule was never about.

A lossy fragment still takes the digest; an empty title keeps the bare
basis or "untitled-<hash>" exactly as before #514. One new case pins the
pre-#514 output and fails on current master. The CHANGELOG example for
the Python-разработчик key is corrected to its real digest (python-fb29d7).

Follow-up suggested in the #514 merge note.

CHANGELOG.md

@@ -63,13 +63,17 @@ per-file diff commands.
   everyday Russian listings on freehire, the shipped multi-market portal. The fold is now
   treated as lossy whenever it drops a letter outside the Latin script, and the existing
   fallbacks take over: the title half uses the portal's numeric id or the URL hash with the
-  fragment kept as a readable prefix (`1-4461771225`, `python-bb31e9`), the company half uses
+  fragment kept as a readable prefix (`1-4461771225`, `python-fb29d7`), the company half uses
   the NFC-casefold name hash with the fragment as prefix (`ai-d35210`). Latin letters that
   also lack a decomposition (`ø`, `æ`, `ß`, `ł`) are deliberately not counted, so `Ørsted`
   still keys as `rsted` and a live Danish `seen_jobs.json` does not re-key; an existing
   mixed-script entry re-keys once on the next scrape and `--audit` lists it under
   `keys_not_matching_current_rule`, the same one-time drift #502 accepted. Seven new cases;
-  the four collision cases fail on master.
+  the four collision cases fail on master. A follow-up keeps the rule inside its own case:
+  an empty title whose URL carries no six-digit run keys on the URL-derived basis alone
+  (`acme_https-example-com-jobs-abc-xyz`) exactly as before, where the first cut had started
+  appending the digest - unreachable from `/scrape`, which always passes a title, but a key
+  is a pure function of the posting and should not move for a case the rule was never about.
 
 - **`verify_pdf.py --ascii-dates` no longer reads a year-like run inside a longer number as a
   date** - the year pattern had no digit boundaries, so `2000` inside `120000` or `12000` made

tests/test_job_key.py

@@ -142,6 +142,19 @@ class MixedScriptNamesKeepTheirIdentity(unittest.TestCase):
         b = make_key("сбер ai", "ML Engineer", url="https://example.com/2")
         self.assertEqual(a.split("_", 1)[0], b.split("_", 1)[0])
 
+    def test_empty_title_with_a_non_numeric_url_keeps_its_pre_fragment_key(self):
+        # The fragment rule must not reach past its own case: an empty title
+        # with a URL that carries no six-digit run keyed on the URL-derived
+        # basis alone before #514, and the first cut of the rule had started
+        # appending the digest there too. Unreachable from /scrape, which always
+        # passes a title, but the key is a pure function of the posting and
+        # should not move for a case the rule was never about.
+        self.assertEqual(
+            make_key("Acme", "", url="https://example.com/jobs/abc-xyz"),
+            "acme_https-example-com-jobs-abc-xyz",
+        )
+        self.assertTrue(make_key("Acme", "", url="").startswith("acme_untitled-"))
+
     def test_latin_names_with_accents_and_ligatures_are_not_lossy(self):
         # NFKD folds these without dropping a letter, so existing keys stay put.
         self.assertEqual(make_key("Zürich Versicherung", "Ingénieur finance"), "zurich-versicherung_ingenieur-finance")

tools/job_key.py

@@ -146,7 +146,10 @@ def make_key(company: str, title: str, url: str = "") -> str:
             # is left to key on.
             digest_basis = str(url) if url else str(title)
             digest = hashlib.sha1(digest_basis.encode("utf-8")).hexdigest()[:HASH_LEN]
-            title_slug = f"{fragment or basis or 'untitled'}-{digest}"
+            # A lossy fragment needs the digest to tell two postings apart; an
+            # empty title keeps the bare basis (a URL-derived slug, or
+            # "untitled-<hash>") exactly as before the fragment rule existed.
+            title_slug = f"{fragment}-{digest}" if fragment else (basis or f"untitled-{digest}")
     return f"{company_slug}_{title_slug}"