Merge pull request #6038 from agmckee/support-iso-8601-date-for-first…

…-published-year-when-indexing Use SolrProcessor.get_pub_year to get edition years
internetarchive · Mar 8, 2022 · 47b9461 · 47b9461
2 parents 34093b7 + b7214e7
commit 47b9461
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 8 deletions.
diff --git a/openlibrary/solr/update_work.py b/openlibrary/solr/update_work.py
@@ -43,9 +43,8 @@
 re_author_key = re.compile(r'^/(?:a|authors)/(OL\d+A)')
 re_bad_char = re.compile('[\x01\x0b\x1a-\x1e]')
 re_edition_key = re.compile(r"/books/([^/]+)")
-re_iso_date = re.compile(r'^(\d{4})-\d\d-\d\d$')
 re_solr_field = re.compile(r'^[-\w]+$', re.U)
-re_year = re.compile(r'(\d{4})$')
+re_year = re.compile(r'\b(\d{4})\b')
 
 # This will be set to a data provider; have faith, mypy!
 data_provider = cast(DataProvider, None)
@@ -491,9 +490,6 @@ def get_pub_year(self, e):
         """
         pub_date = e.get('publish_date', None)
         if pub_date:
-            m = re_iso_date.match(pub_date)
-            if m:
-                return m.group(1)
             m = re_year.search(pub_date)
             if m:
                 return m.group(1)
@@ -595,8 +591,9 @@ def add_list(name, values):
         pub_dates = {e[k] for e in editions if e.get(k)}
         add_list(k, pub_dates)
         pub_years = {
-            m.group(1) for m in (re_year.search(date) for date in pub_dates) if m
+            self.get_pub_year(e) for e in editions
         }
+        pub_years = pub_years - {None,}
         if pub_years:
             add_list('publish_year', pub_years)
             add('first_publish_year', min(int(y) for y in pub_years))

diff --git a/openlibrary/tests/solr/test_update_work.py b/openlibrary/tests/solr/test_update_work.py
@@ -164,18 +164,20 @@ def test_publish_year(self):
         test_dates = [
             "2000",
             "Another 2000",
-            "2001-01-02",  # Doesn't seems to be handling this case
+            "2001-01-02",  # ISO 8601 formatted dates now supported
             "01-02-2003",
+            "2004 May 23",
             "Jan 2002",
             "Bad date 12",
+            "Bad date 123412314",
         ]
         work = make_work()
         update_work.data_provider = FakeDataProvider(
             [work] + [make_edition(work, publish_date=date) for date in test_dates]
         )
 
         d = build_data(work)
-        assert sorted(d['publish_year']) == ["2000", "2002", "2003"]
+        assert sorted(d['publish_year']) == ["2000", "2001", "2002", "2003", "2004"]
         assert d["first_publish_year"] == 2000
 
     def test_isbns(self):