fix(find.py): Add fixes for over id'ing reference citations

Fix typos

eyecite/find.py

@@ -4,9 +4,9 @@
 from eyecite.helpers import (
     disambiguate_reporters,
     extract_pin_cite,
+    filter_citations,
     joke_cite,
     match_on_tokens,
-    order_citations,
 )
 from eyecite.models import (
     CaseReferenceToken,
@@ -110,7 +110,7 @@ def get_citations(
 
         citations.append(citation)
 
-    citations = order_citations(citations)
+    citations = filter_citations(citations)
 
     # Remove citations with multiple reporter candidates where we couldn't
     # guess correct reporter
@@ -139,13 +139,23 @@ def _extract_reference_citations(
     if not citation.metadata.defendant:
         # Skip if no defendant exists
         return []
-    escaped_plaintiff = re.escape(citation.metadata.plaintiff or "")
-    escaped_defendant = re.escape(citation.metadata.defendant)
+    plaintiff_regex = (
+        rf"(?P<plaintiff>{re.escape(citation.metadata.plaintiff)})"
+        if citation.metadata.plaintiff
+        else ""
+    )
+    defendant_regex = (
+        rf"(?P<defendant>{re.escape(citation.metadata.defendant)})"
+        if citation.metadata.defendant
+        else ""
+    )
+
+    # Combine the components if they are not empty
+    combined_regex_parts = "|".join(
+        filter(None, [plaintiff_regex, defendant_regex])
+    )
     pin_cite_regex = (
-        rf"\b(?:"
-        rf"(?P<plaintiff>{escaped_plaintiff})|"
-        rf"(?P<defendant>{escaped_defendant})\s?"
-        rf")\s+at\s+(?P<page>\d{{1,5}})?\b"
+        rf"\b(?:{combined_regex_parts})\s+at\s+(?P<page>\d{{1,5}})\b"
     )
 
     pin_cite_pattern = re.compile(pin_cite_regex)
@@ -169,7 +179,11 @@ def _extract_reference_citations(
             full_span_end=end + offset,
             index=0,
             metadata={
-                "plaintiff": match.group("plaintiff"),
+                "plaintiff": (
+                    match.group("plaintiff")
+                    if "plaintiff" in match.groupdict()
+                    else None
+                ),
                 "defendant": match.group("defendant"),
                 "pin_cite": match.group("page"),
             },

eyecite/helpers.py

@@ -307,14 +307,25 @@ def disambiguate_reporters(
     ]
 
 
-def order_citations(citations: List[CitationBase]) -> List[CitationBase]:
-    """
-    Order citations that may have reference citations out or sequential order
+def filter_citations(citations: List[CitationBase]) -> List[CitationBase]:
+    """Filter and order citations that may have reference cites out of order
 
     :param citations: List of citation`
-    :return: Sorted citations
+    :return: Sorted and filtered citations
     """
-    return sorted(citations, key=lambda citation: citation.span())
+    filtered_citations = []
+    sorted_citations = sorted(citations, key=lambda citation: citation.span())
+    for citation in sorted_citations:
+        if filtered_citations:
+            last_citation = filtered_citations[-1]
+            last_span = last_citation.span()
+            current_span = citation.span()
+
+            if current_span[0] <= last_span[1]:
+                # Remove overlapping citations that can occur in edge cases
+                continue
+        filtered_citations.append(citation)
+    return filtered_citations
 
 
 joke_cite: List[CitationBase] = [

eyecite/resolve.py

@@ -213,7 +213,8 @@ def _resolve_reference_citation(
     reference_citation: ReferenceCitation,
     resolved_full_cites: ResolvedFullCites,
 ) -> Optional[ResourceType]:
-    """
+    """Resolve reference citations
+
     Try to resolve reference citations by checking whether their is only one
     full citation that appears with either the defendant or plaintiff
     field of any of the previously resolved full citations.

tests/test_AnnotateTest.py

@@ -47,7 +47,7 @@ def lower_annotator(before, text, after):
                 "<0>1 U.S. 1</0>. Foo v. Bar, <1>supra at 2</1>.",
                 [],
             ),
-            # Refernce cite
+            # Reference cite
             (
                 "Foo v. Bar 1 U.S. 1. In Foo at 2.",
                 "Foo v. Bar <0>1 U.S. 1</0>. In <1>Foo at 2</1>.",

tests/test_FindTest.py

@@ -471,6 +471,16 @@ def test_find_citations(self):
               reference_citation('Foo at 62',
                                  metadata={'defendant': 'Foo',
                                            "pin_cite": "62"})]),
+            # Test reference citation that contains at
+            ('In re Foo 1 Mass. 12, 347-348. something something, in at we see that',
+             [case_citation(page='12', reporter="Mass.", volume="1",
+                            metadata={'defendant': 'Foo', 'pin_cite': '347-348'})]),
+            # Test U.S. as plaintiff with reference citations
+            ('U.S. v. Boch Oldsmobile, Inc., 909 F.2d 657, 660 (1st Cir.1990); Piper Aircraft, 454 U.S. at 241',
+             [case_citation(page='657', reporter="F.2d", volume="909",
+                            metadata={'plaintiff': 'U.S.', 'defendant': 'Boch Oldsmobile, Inc.', 'pin_cite': '660'}),
+              case_citation(volume="454", page='241', reporter_found='U.S.', short=True,
+                            metadata={'antecedent_guess': 'Aircraft', 'court': "scotus", 'pin_cite': None})]),
             # Test reference citation after an id citation
             ('we said in Morton v. Mancari, 417 U. S. 535, 552 (1974) “Literally every piece ....”. “asisovereign tribal entities . . . .” Id. In Mancari at 665',
              [case_citation(page='535', year=1974, volume="417",