fix(helpers): Check with html and check without it if nothing is found

Author: flooie

eyecite/helpers.py

@@ -289,22 +289,25 @@ def find_case_name(citation: CaseCitation, document: Document, short=False):
             defendant = candidate_case_name
 
         defendant = strip_stop_words(defendant)
-        if short is False:
-            citation.metadata.defendant = defendant.strip(", ").strip()
-        else:
-            citation.metadata.antecedent_guess = (
-                defendant.strip(" ").strip(",").strip("(")
-            )
 
-        offset = (
-            len(
-                "".join(
-                    str(w) for w in words[start_index : citation.index - 1]
+        clean_def =  defendant.strip(", ").strip()
+        if clean_def:
+            if short is False and clean_def:
+                citation.metadata.defendant = clean_def
+            else:
+                citation.metadata.antecedent_guess = (
+                    defendant.strip(" ").strip(",").strip("(")
+                )
+
+            offset = (
+                len(
+                    "".join(
+                        str(w) for w in words[start_index : citation.index - 1]
+                    )
                 )
+                + 1
             )
-            + 1
-        )
-        citation.full_span_start = citation.span()[0] - offset
+            citation.full_span_start = citation.span()[0] - offset
 
         if pre_cite_year:
             # found pre citation year, store it

eyecite/models.py

@@ -538,6 +538,9 @@ def add_metadata(self, document: "Document"):
         else:
             find_case_name(self, document)
 
+        if self.metadata.defendant == None:
+            find_case_name(self, document)
+
         add_pre_citation(self, document)
         self.guess_court()
         super().add_metadata(document)

tests/test_FindTest.py

@@ -1338,7 +1338,9 @@ def test_markup_plaintiff_and_antecedent_guesses(self) -> None:
                         volume="242",
                         short=False,
                         metadata={
-                            "antecedent_guess": "Foy",
+                            "plaintiff": "Thrift Funds Canal, Inc.",
+                            "defendant": "Foy",
+                            "pin_cite": "257",
                             "year": "1970",
                         },
                     )
@@ -1356,8 +1358,6 @@ def test_markup_plaintiff_and_antecedent_guesses(self) -> None:
                         reporter="U.S.",
                         year=2004,
                         metadata={
-                            "plaintiff": None,
-                            "defendant": None,
                             "court": "scotus",
                         },
                     )
@@ -1565,6 +1565,22 @@ def test_markup_plaintiff_and_antecedent_guesses(self) -> None:
                 ],
                 {"clean_steps": ["html", "all_whitespace"]},
             ),
+            # can we extract out the citation if its not wrapped in html but in html
+            ("dentification. Stovall v. Denno, 388 U.S. 293, ",
+             [
+                 case_citation(
+                     page="293",
+                     volume="388",
+                     reporter="U.S.",
+                     short=False,
+                     metadata={
+                         "plaintiff": "Stovall",
+                         "defendant": "Denno",
+                     },
+                 ),
+             ],
+             {"clean_steps": ["html", "all_whitespace"]},
+             ),
             # can we remove see also
             (
                 """<em>see also Cass v. Stephens</em>,\r\n156 S.W.3d 38""",