fix: improved source and keywords for rfc json

ietf/doc/tests_utils_rfc_json.py

@@ -13,6 +13,7 @@
     PublishedRfcDocEventFactory,
     RfcAuthorFactory,
     RfcFactory,
+    RgRfcFactory,
     WgRfcFactory,
 )
 from ietf.doc.models import RelatedDocument
@@ -245,6 +246,19 @@ def test_april_first_date_format(self):
 
         self.assertEqual(data["pub_date"], "1 April 2020")
 
+    def test_empty_keywords_filtered(self):
+        """Empty-string keywords are stripped from the keywords list."""
+        rfc = PublishedRfcDocEventFactory(
+            doc=WgRfcFactory(keywords=["foo", "", "bar"]),
+        ).doc
+        _put_pub_levels(rfc.rfc_number, "ps")
+        _put_empty_errata()
+
+        generate_rfc_json(rfc.rfc_number)
+        data = _read_json(rfc.rfc_number)
+
+        self.assertEqual(data["keywords"], ["foo", "bar"])
+
     def test_non_april_first_april_date(self):
         """An April publication that is NOT in the April Fools list gets 'April YYYY'."""
         rfc = PublishedRfcDocEventFactory(
@@ -260,7 +274,7 @@ def test_non_april_first_april_date(self):
         self.assertEqual(data["pub_date"], "April 2020")
 
     def test_source_ietf_wg(self):
-        """IETF-stream WG RFC: source is 'acronym (area)'."""
+        """IETF-stream WG RFC: source is the group's full name."""
         area = GroupFactory(type_id="area")
         wg = GroupFactory(type_id="wg", parent=area)
         rfc = PublishedRfcDocEventFactory(
@@ -272,7 +286,7 @@ def test_source_ietf_wg(self):
         generate_rfc_json(rfc.rfc_number)
         data = _read_json(rfc.rfc_number)
 
-        self.assertEqual(data["source"], f"{wg.acronym} ({area.acronym})")
+        self.assertEqual(data["source"], wg.name)
 
     def test_source_ietf_no_wg(self):
         """IETF-stream individual RFC (group acronym 'none'): source is 'IETF - NON WORKING GROUP'."""
@@ -293,6 +307,22 @@ def test_source_ietf_no_wg(self):
 
         self.assertEqual(data["source"], "IETF - NON WORKING GROUP")
 
+    def test_source_ietf_area(self):
+        """IETF-stream RFC with area-type group: source is 'IETF - NON WORKING GROUP'."""
+        area = GroupFactory(type_id="area")
+        area.parent = GroupFactory()
+        area.save()
+        rfc = PublishedRfcDocEventFactory(
+            doc=RfcFactory(group=area, stream_id="ietf"),
+        ).doc
+        _put_pub_levels(rfc.rfc_number, "inf")
+        _put_empty_errata()
+
+        generate_rfc_json(rfc.rfc_number)
+        data = _read_json(rfc.rfc_number)
+
+        self.assertEqual(data["source"], "IETF - NON WORKING GROUP")
+
     def test_source_iab(self):
         """IAB-stream RFC: source is 'IAB'."""
         rfc = PublishedRfcDocEventFactory(
@@ -319,6 +349,30 @@ def test_source_ise(self):
 
         self.assertEqual(data["source"], "INDEPENDENT")
 
+    def test_source_irtf_rg(self):
+        """IRTF-stream RG RFC: source is the group's full name."""
+        rfc = PublishedRfcDocEventFactory(doc=RgRfcFactory()).doc
+        _put_pub_levels(rfc.rfc_number, "inf")
+        _put_empty_errata()
+
+        generate_rfc_json(rfc.rfc_number)
+        data = _read_json(rfc.rfc_number)
+
+        self.assertEqual(data["source"], rfc.group.name)
+
+    def test_source_irtf_no_rg(self):
+        """IRTF-stream RFC with no specific RG (group acronym 'none'): source is 'IRTF'."""
+        rfc = PublishedRfcDocEventFactory(
+            doc=RfcFactory(stream_id="irtf", group=GroupFactory(acronym="none")),
+        ).doc
+        _put_pub_levels(rfc.rfc_number, "inf")
+        _put_empty_errata()
+
+        generate_rfc_json(rfc.rfc_number)
+        data = _read_json(rfc.rfc_number)
+
+        self.assertEqual(data["source"], "IRTF")
+
     def test_pub_levels_passed_in(self):
         """When pub_levels is passed in, get_publication_std_levels() is not called."""
         rfc = PublishedRfcDocEventFactory(doc=WgRfcFactory()).doc

ietf/doc/utils_rfc_json.py

@@ -105,31 +105,29 @@ def generate_rfc_json(rfc_number: int, *, pub_levels=None) -> None:
     stream_slug = rfc.stream.slug
     group_acronym = rfc.group.acronym
 
-    area_acronym = None
     if stream_slug == "ietf":
         if rfc.group.parent is None:
             assertion("rfc.group.parent is not None")
             log(
                 f"Malformed document object encountered for rfc{rfc_number}. Aborting update of rfc{rfc_number}.json"
             )
             return
-        else:
-            area_acronym = rfc.group.parent.acronym
 
     if stream_slug == "ise":
         source = "INDEPENDENT"
     elif stream_slug == "iab":
         source = "IAB"
     elif stream_slug == "ietf" and (
-        group_acronym in ("none", "gen") or not area_acronym
+        group_acronym == "none" or rfc.group.type_id == "area"
     ):
         source = "IETF - NON WORKING GROUP"
-    elif group_acronym not in ("none", ""):
-        source = group_acronym
-        if stream_slug == "ietf" and area_acronym:
-            source += f" ({area_acronym})"
-        elif stream_slug:
-            source += f" ({stream_slug})"
+    elif stream_slug == "irtf":
+        if group_acronym == "none":
+            source = "IRTF"
+        else:
+            source = rfc.group.name
+    elif group_acronym not in ("none", "") and stream_slug in ["ietf", "editorial"]:
+        source = rfc.group.name
     elif stream_slug:
         source = "Legacy" if stream_slug == "legacy" else stream_slug.upper()
     else:
@@ -207,7 +205,7 @@ def _rfc_list(qs, attr):
         "source": source,
         "abstract": rfc.abstract,
         "pub_date": pub_date,
-        "keywords": rfc.keywords,
+        "keywords": [kw for kw in rfc.keywords if kw],
         "obsoletes": obsoletes,
         "obsoleted_by": obsoleted_by,
         "updates": updates,