Don't write 9999 dates from MARC

openlibrary/catalog/marc/html.py

@@ -21,6 +21,7 @@ def subfields(line):
 class html_record:
     def __init__(self, data):
         assert len(data) == int(data[:5])
+        self.data = data
         self.record = Record(data)
         self.leader = self.record.leader
 

openlibrary/catalog/marc/parse.py

@@ -327,9 +327,13 @@ def read_languages(rec: MarcBase, lang_008: str | None = None) -> list[str]:
 
 
 def read_pub_date(rec: MarcBase) -> str | None:
+    """
+    Read publish date from 260$c.
+    """
+
     def publish_date(s: str) -> str:
         date = s.strip('[]')
-        if date.lower() == 'n.d.':  # No date
+        if date.lower() in ('n.d.', 's.d.'):  # No date
             date = '[n.d.]'
         return remove_trailing_number_dot(date)
 
@@ -667,15 +671,19 @@ def read_edition(rec: MarcBase) -> dict[str, Any]:
             raise BadMARC("'008' field must not be blank")
         publish_date = f[7:11]
 
-        if re_date.match(publish_date) and publish_date != '0000':
-            edition["publish_date"] = publish_date
-        if f[6] == 't':
-            edition["copyright_date"] = f[11:15]
+        if re_date.match(publish_date) and publish_date not in ('0000', '9999'):
+            edition['publish_date'] = publish_date
+        if f[6] == 'r' and f[11:15] > publish_date:
+            # Incorrect reprint date order
+            update_edition(rec, edition, read_pub_date, 'publish_date')
+        elif f[6] == 't':  # Copyright date
+            edition['copyright_date'] = f[11:15]
+        if 'publish_date' not in edition:  # Publication date fallback to 260$c
+            update_edition(rec, edition, read_pub_date, 'publish_date')
         publish_country = f[15:18]
         if publish_country not in ('|||', '   ', '\x01\x01\x01', '???'):
-            edition["publish_country"] = publish_country.strip()
-        languages = read_languages(rec, lang_008=f[35:38].lower())
-        if languages:
+            edition['publish_country'] = publish_country.strip()
+        if languages := read_languages(rec, lang_008=f[35:38].lower()):
             edition['languages'] = languages
     elif handle_missing_008:
         update_edition(rec, edition, read_languages, 'languages')

openlibrary/catalog/marc/tests/test_data/bin_expect/test-publish-sn-sl-nd.json

@@ -15,6 +15,7 @@
   "publishers": [
     "[s.n.]"
   ],
+  "publish_date": "[n.d.]",
   "publish_places": [
     "[s.l.]"
   ]

openlibrary/catalog/marc/tests/test_data/bin_input/9999_sd_dates.mrc

@@ -0,0 +1 @@
+00419nam a22001695a 4500001001200000005001700012008004100029035001400070100003800084245002600122260004200148300001100190546001300201899001500214988001300229906000700242012717654-320110315150330.0110301s9999    pk            000 0 urd d  a(PkLaDAR)0 aʻAbd-ul-Qayyum Tahir Malihabadi.10aIfadat-e homiyopethi.  aLahore :bMaktabah-e Daniyal,c[s.d.]  a150 p.  aIn Urdu.  a415_565960  a20110315  0MH

openlibrary/catalog/marc/tests/test_data/bin_input/9999_with_correct_date_in_260.mrc

@@ -0,0 +1 @@
+00436nam a22001695a 4500001001200000005001700012008004100029035001400070100003700084245003600121260004800157300001100205546001500216899001500231988001300246906000700259012716825-720110315150330.0110301s9999    pk            000 0 urd d  a(PkLaDAR)0 aShah Moʻin-al-Din Ahmad Nadavi.10aʻAlimi aman jo Islami manshur.  aHyderabad :bSindh National Academy,c2003.  a236 p.  aIn Sindhi.  a415_565960  a20110315  0MH

openlibrary/catalog/marc/tests/test_data/bin_input/reprint_date_wrong_order.mrc

@@ -0,0 +1 @@
+01299cam  2200289Ka 4500001001300000003000600013005001700019008004100036040001800077020002500095020002200120035002100142100003000163245005200193246003200245250001600277260004700293300002500340440003000365500023100395520023800626650003900864655003100903700002100934994001200955948004200967ocn656308391OCoLC20100924175141.0100816r18732010tnu           000 1 eng d  aLODcLODdLOD  a9781403793966 (pbk.)  a1403793964 (pbk.)  a(OCoLC)6563083911 aVerne, Jules,d1828-1905.10aAround the world in eighty days /cJules Verne.3 aAround the world in 80 days  aUnabridged.  aFranklin, Tenn. :bDalmatian Press,c2010.  axi, 210 p. ;c21 cm. 0aDalmatian Press Classics.  a"Dalmatian Press Classics offer enjoyable editions of greats of literature. The unabridged texts are enhanced with helpful footnotes. Punctuation and spelling have been updated to conform to modern American usage."-t.p. verso.  a"Phileas Fogg's bet that he can span the world in eighty days takes him by train, ship, sled, and elephant across ninettenth-century India, China, and America in one of the wittiest, most enjoyable classics of all time."--Back cover. 0aVoyages around the worldvFiction. 0aAdventure stories, French.1 aKnight, Kathryn.  aZ0bPMR  hNO HOLDINGS IN PMR - 1 OTHER HOLDINGS

openlibrary/catalog/marc/tests/test_data/xml_expect/soilsurveyrepor00statgoog.json

@@ -21,6 +21,7 @@
       "entity_type": "org"
     }
   ],
+  "publish_date": "[n.d.]",
   "publish_places": [
     "Ames"
   ],

openlibrary/catalog/marc/tests/test_parse.py

@@ -80,6 +80,12 @@
     'test-publish-sn-sl-nd.mrc',
 ]
 
+date_tests = [  # MARC, expected publish_date
+    ('9999_sd_dates.mrc', '[n.d.]'),
+    ('reprint_date_wrong_order.mrc', '2010'),
+    ('9999_with_correct_date_in_260.mrc', '2003'),
+]
+
 TEST_DATA = Path(__file__).with_name('test_data')
 
 
@@ -151,6 +157,13 @@ def test_raises_no_title(self):
         with pytest.raises(NoTitle):
             read_edition(rec)
 
+    @pytest.mark.parametrize('marcfile,expect', date_tests)
+    def test_dates(self, marcfile, expect):
+        filepath = TEST_DATA / 'bin_input' / marcfile
+        rec = MarcBinary(filepath.read_bytes())
+        edition = read_edition(rec)
+        assert edition['publish_date'] == expect
+
 
 class TestParse:
     def test_read_author_person(self):