only keep papers in the next 3 years (fixes: #378)

papers/migrations/0049_filter_future_dates.py

@@ -0,0 +1,53 @@
+# -*- coding: utf-8 -*-
+# Generated by Django 1.10.5 on 2017-01-05 12:01
+from __future__ import unicode_literals
+import datetime
+from papers.utils import year_margin
+from django.db import migrations
+import haystack
+from papers.models import Paper
+
+
+# have to put this here as we cannot call methods from the model in a migration...
+
+def remove_from_index(paper):
+    """
+    Remove this paper from Haystack's index
+    (to be called before deleting the paper for real)
+    """
+    using_backends = haystack.connection_router.for_write(instance=paper)
+    for using in using_backends:
+        try:
+            index = haystack.connections[using].get_unified_index(
+                                    ).get_index(Paper)
+            index.remove_object(paper, using=using)
+        except haystack.exceptions.NotHandled:
+            pass
+
+
+def database_forwards(apps, schema_editor):
+    now = datetime.datetime.now()
+    current_year = now.year
+    cutoff_date = datetime.datetime(current_year + year_margin, 12, 31)
+    Paper_fake = apps.get_model("papers", "Paper")
+    for paper in Paper_fake.objects.filter(pubdate__gt=cutoff_date):
+        remove_from_index(paper)
+        paper.delete()
+
+
+def database_backwards(apps, schema_editor):
+    pass
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('papers', '0048_remove_paper_researchers'),
+    ]
+
+    operations = [
+        migrations.RunPython(
+            database_forwards,
+            database_backwards
+        ),
+    ]

papers/utils.py

@@ -33,6 +33,7 @@
 
 filter_punctuation_alphanum_regex = re.compile(r'.*\w')
 
+year_margin = 3
 
 def filter_punctuation(lst):
     """
@@ -546,11 +547,11 @@ def valid_publication_date(dt):
     >>> valid_publication_date(None)
     False
     """
-    # In 2042, the problem of access to scientific publications
-    # will be solved and Dissemin will autodestroy as a consequence
-    # of this hack.
+    now = datetime.datetime.now()
+    current_year = now.year
+    # some papers are published in the future ("to appear")
     return ((type(dt) == datetime.date or type(dt) == datetime.datetime)
-            and dt.year < 2042)
+            and dt.year < current_year + year_margin)
 
 ### ORCiD utilities ###