fix: ignore citation types not matching cas schema NO-JIRA (#643)

Author: JoshParkSJ

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "uipath-langchain"
-version = "0.7.7"
+version = "0.7.9"
 description = "Python SDK that enables developers to build and deploy LangGraph agents to the UiPath Cloud Platform"
 readme = { file = "README.md", content-type = "text/markdown" }
 requires-python = ">=3.11"

src/uipath_langchain/runtime/_citations.py

@@ -53,17 +53,18 @@ def _parse_citations(text: str) -> list[tuple[str, _ParsedCitation | None]]:
 
         has_url = url is not None
         has_reference = reference is not None
+        has_page_number = page_number is not None
 
         if has_url and not has_reference:
             # web citation
             citation = _ParsedCitation(title=title, url=url, page_number=page_number)
-        elif has_reference and not has_url:
+        elif has_reference and has_page_number and not has_url:
             # context grounding citation
             citation = _ParsedCitation(
                 title=title, reference=reference, page_number=page_number
             )
         else:
-            # skip; citation has no url= or reference=
+            # skip; doesn't match a valid source type
             if preceding_text:
                 segments.append((preceding_text, None))
             cursor = match.end()
@@ -90,30 +91,32 @@ def _make_source(
     source_numbers: dict[_ParsedCitation, int],
     next_number: int,
 ) -> tuple[
-    UiPathConversationCitationSourceUrl | UiPathConversationCitationSourceMedia, int
+    UiPathConversationCitationSourceUrl | UiPathConversationCitationSourceMedia | None,
+    int,
 ]:
-    """Build a citation source, deduplicating by assigning numbers"""
-    if citation not in source_numbers:
-        source_numbers[citation] = next_number
-        next_number += 1
-    number = source_numbers[citation]
-
-    source: UiPathConversationCitationSourceUrl | UiPathConversationCitationSourceMedia
+    """Build a citation source, deduplicating by assigning numbers."""
     if citation.url is not None:
-        source = UiPathConversationCitationSourceUrl(
+        if citation not in source_numbers:
+            source_numbers[citation] = next_number
+            next_number += 1
+        return UiPathConversationCitationSourceUrl(
             title=citation.title,
-            number=number,
+            number=source_numbers[citation],
             url=citation.url,
-        )
-    else:
-        source = UiPathConversationCitationSourceMedia(
+        ), next_number
+    elif citation.reference is not None and citation.page_number is not None:
+        if citation not in source_numbers:
+            source_numbers[citation] = next_number
+            next_number += 1
+        return UiPathConversationCitationSourceMedia(
             title=citation.title,
-            number=number,
+            number=source_numbers[citation],
             mime_type=None,
             download_url=citation.reference,
             page_number=citation.page_number,
-        )
-    return source, next_number
+        ), next_number
+    else:
+        return None, next_number
 
 
 def _find_partial_tag_start(text: str) -> int:
@@ -160,6 +163,9 @@ def _build_content_part_citation(
             citation, self._source_numbers, self._next_number
         )
 
+        if not source:
+            return UiPathConversationContentPartChunkEvent(data=text)
+
         return UiPathConversationContentPartChunkEvent(
             data=text,
             citation=UiPathConversationCitationEvent(
@@ -236,6 +242,9 @@ def extract_citations_from_text(
 
         if citation is not None:
             source, next_number = _make_source(citation, source_numbers, next_number)
+            if not source:
+                offset += length
+                continue
             if length > 0:
                 citations.append(
                     UiPathConversationCitationData(

tests/runtime/test_citations.py

@@ -287,7 +287,7 @@ def test_mixed_valid_and_invalid_citations(self):
         text = (
             'A<uip:cite title="Valid" url="https://v.com" />'
             'B<uip:cite title="Invalid" page_number="1" />'
-            'C<uip:cite title="Also Valid" reference="https://r.com" />'
+            'C<uip:cite title="Also Valid" reference="https://r.com" page_number="2" />'
         )
         events = proc.add_chunk(text)
         events.extend(proc.finalize())
@@ -440,6 +440,20 @@ def test_citation_with_both_url_and_reference_skipped(self):
         assert combined == "Text more"
         assert all(e.citation is None for e in events)
 
+    def test_reference_without_page_number_skipped(self):
+        """Web URL misclassified as reference (no page_number) must not emit a media source."""
+        proc = CitationStreamProcessor()
+        text = (
+            "UiPath reported earnings"
+            '<uip:cite title="UiPath Reports Third Quarter Fiscal 2026 Fin..." '
+            'reference="https://ir.uipath.com/news/detail/420/uipath-reports-third-quarter-fiscal-2026-financial-results" />'
+        )
+        events = proc.add_chunk(text)
+        events.extend(proc.finalize())
+        combined = "".join(e.data for e in events if e.data)
+        assert combined == "UiPath reported earnings"
+        assert all(e.citation is None for e in events)
+
     def test_only_whitespace_between_citations(self):
         """Citations separated only by whitespace."""
         proc = CitationStreamProcessor()
@@ -664,6 +678,17 @@ def test_text_with_trailing_content(self):
         assert citations[0].offset == 0
         assert citations[0].length == 6  # len("A fact")
 
+    def test_reference_without_page_number_skipped(self):
+        """Web URL misclassified as reference (no page_number) must not emit a citation."""
+        text = (
+            "UiPath reported earnings"
+            '<uip:cite title="UiPath Reports Third Quarter Fiscal 2026 Fin..." '
+            'reference="https://ir.uipath.com/news/detail/420/uipath-reports-third-quarter-fiscal-2026-financial-results" />'
+        )
+        cleaned, citations = extract_citations_from_text(text)
+        assert cleaned == "UiPath reported earnings"
+        assert citations == []
+
     def test_different_sources_get_different_numbers(self):
         """Different sources get incrementing numbers."""
         text = (