tests(tests): Add a few more tests

More edge case examples

Author: flooie

eyecite/helpers.py

@@ -206,12 +206,12 @@ def add_defendant(citation: CaseCitation, document: Document) -> None:
                 and document.plain_to_markup
                 and document.markup_to_plain
             ):
-                new_text = update_defendant_markup(document, stop_word)
+                new_defendant = update_defendant_markup(document, stop_word).strip()
                 if (
-                    citation.metadata.defendant != new_text
-                    and new_text is not None
+                    citation.metadata.defendant != new_defendant
+                    and new_defendant != ""
                 ):
-                    citation.metadata.defendant = new_text
+                    citation.metadata.defendant = new_defendant
 
 
 def add_pre_citation(citation: FullCaseCitation, document: Document) -> None:
@@ -545,7 +545,7 @@ def update_defendant_markup(
     Returns: New defendant text if any
     """
     if not document.plain_to_markup or not document.markup_to_plain:
-        return None
+        return ""
 
     # add 3 char to account for the star pagination whitespace
     # <i>United States</i> v. <i>Carignan,</i> <span class="star-pagination">
@@ -560,11 +560,12 @@ def update_defendant_markup(
         defendant_end = document.markup_to_plain.update(
             filtered_results[0][2], bisect_right
         )
+
         defendant_start = (
             stop_word.start + len(stop_word.groups["stop_word"]) + 1
         )
         return document.plain_text[defendant_start:defendant_end].strip(" ,")
-    return None
+    return ""
 
 
 joke_cite: List[CitationBase] = [

eyecite/regexes.py

@@ -70,6 +70,7 @@ def short_cite_re(regex):
     "aff'd",
     "affirmed",
     "remanded",
+    "see also",
     "see",
     "granted",
     "dismissed",

tests/test_FindTest.py

@@ -1534,5 +1534,53 @@ def test_markup_plaintiff_and_antecedent_guesses(self) -> None:
                 ],
                 {"clean_steps": ["html", "all_whitespace"]},
             ),
+            # can we remove see also
+            (
+                """<em>see also Cass v. Stephens</em>,\r\n156 S.W.3d 38""",
+                [
+                    case_citation(
+                        page="38",
+                        volume="156",
+                        reporter="S.W.3d",
+                        short=False,
+                        metadata={
+                            "plaintiff": "Cass",
+                            "defendant": "Stephens",
+                        },
+                    ),
+                ],
+                {"clean_steps": ["html", "all_whitespace"]},
+            ),
+            # technically this is incorrect in determing plaintiff/defendant but we have no way to deal with that
+            (
+                """ <i>See </i><i>Loup-Miller Const. Co. v. City and County of Denver,</i> 676 P.2d 1170 (Colo.1984) .... <i>See </i><i>Loup-Miller,</i> 676 P.2d 1170 and so on <i>Loup-Miller</i>""",
+                [
+                    case_citation(
+                        page="1170",
+                        volume="676",
+                        reporter="P.2d",
+                        short=False,
+                        metadata={
+                            "plaintiff": "Loup-Miller Const. Co.",
+                            "defendant": "City and County of Denver",
+                        },
+                    ),
+                    case_citation(
+                        page="1170",
+                        volume="676",
+                        reporter="P.2d",
+                        short=False,
+                        metadata={
+                            "defendant": "Loup-Miller",
+                        },
+                    ),
+                    reference_citation(
+                        "Loup-Miller",
+                        metadata={"defendant": "Loup-Miller"},
+                    ),
+                ],
+                {"clean_steps": ["html", "all_whitespace"]},
+            ),
         )
         self.run_test_pairs(test_pairs, "Citation extraction")
+