Merge pull request #1273 from freelawproject/fix_author_str_per_curia…

…m_value fix(ca7, ala, tenn, nmariana): do not return "per curiam" as value for author_str or judges field
freelawproject · Dec 12, 2024 · 406f85a · 406f85a
2 parents 91f3a05 + 0421259
commit 406f85a
Show file tree

Hide file tree

Showing 15 changed files with 15,294 additions and 142 deletions.
diff --git a/juriscraper/opinions/united_states/federal_appellate/ca7.py b/juriscraper/opinions/united_states/federal_appellate/ca7.py
@@ -22,6 +22,11 @@ def _process_html(self):
             docket = parts[parts.index("case#") + 1]
             name = item["summary"].split(docket)[1].split("(")[0]
             author = item["summary"].split("{")[1].split("}")[0]
+            per_curiam = False
+            if "curiam" in author.lower():
+                per_curiam = True
+                author = ""
+
             self.cases.append(
                 {
                     "url": item["link"],
@@ -31,5 +36,6 @@ def _process_html(self):
                     "status": "Published",
                     "judge": author,
                     "author": author,
+                    "per_curiam": per_curiam,
                 }
             )
diff --git a/juriscraper/opinions/united_states/state/ala.py b/juriscraper/opinions/united_states/state/ala.py
@@ -31,15 +31,24 @@ def _process_html(self):
 
                 url = f"https://publicportal-api.alappeals.gov/courts/{self.court_str}/cms/case/{publicationItem['caseInstanceUUID']}/docketentrydocuments/{publicationItem['documents'][0]['documentLinkUUID']}"
                 docket = publicationItem["caseNumber"]
-                author = publicationItem["groupName"]
                 name = publicationItem["title"]
+                judge = publicationItem["groupName"]
+                if judge == "On Rehearing":
+                    judge = ""
+
+                per_curiam = False
+                if "curiam" in judge.lower():
+                    judge = ""
+                    per_curiam = True
+
                 self.cases.append(
                     {
                         "date": date_filed,
                         "name": name,
                         "docket": docket,
                         "status": "Published",
                         "url": url,
-                        "judge": author,
+                        "judge": judge,
+                        "per_curiam": per_curiam,
                     }
                 )
diff --git a/juriscraper/opinions/united_states/state/tenn.py b/juriscraper/opinions/united_states/state/tenn.py
@@ -39,14 +39,22 @@ def _process_html(self):
                 row.strip()
                 for row in section.text_content().strip().split("\n", 4)
             ]
+
+            judge = rows[2].split(": ")[1]
+            per_curiam = False
+            if "curiam" in judge.lower():
+                judge = ""
+                per_curiam = True
+
             self.cases.append(
                 {
                     "date": date,
                     "lower_court": lower_court,
                     "url": url,
                     "name": name,
                     "docket": rows[1],
-                    "judge": rows[2].split(": ")[1],
+                    "judge": judge,
                     "summary": rows[-1],
+                    "per_curiam": per_curiam,
                 }
             )
diff --git a/juriscraper/opinions/united_states/territories/nmariana.py b/juriscraper/opinions/united_states/territories/nmariana.py
@@ -46,7 +46,7 @@ def _fetch_author(self, judges: str) -> str:
         :return: The author
         """
         if "*" not in judges:
-            author = "Per Curiam"
+            author = ""
         else:
             author = [j for j in judges.split(",") if "*" in j][0].strip("*")
         return author
@@ -55,13 +55,15 @@ def _process_html(self):
         for s in self.html.xpath(".//td/a/@href[contains(., 'pdf')]/../../.."):
             cells = s.xpath(".//td")
             judge_text = cells[3].text_content()
+            author = self._fetch_author(judge_text)
             self.cases.append(
                 {
                     "name": cells[0].text_content(),
                     "citation": cells[1].text_content(),
                     "date": cells[2].text_content(),
                     "judge": ", ".join(self._cleanup_judge_names(judge_text)),
-                    "author": self._fetch_author(judge_text),
+                    "author": author,
+                    "per_curiam": not author,
                     "url": s.xpath(".//td/a/@href")[0],
                     "docket": "",
                 }

diff --git a/tests/examples/opinions/united_states/ala_example.compare.json b/tests/examples/opinions/united_states/ala_example.compare.json
@@ -8,7 +8,8 @@
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1230",
     "judges": "Fridy, J.",
-    "case_name_shorts": ""
+    "case_name_shorts": "",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -18,8 +19,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1284",
-    "judges": "On Rehearing",
-    "case_name_shorts": "M.G."
+    "judges": "",
+    "case_name_shorts": "M.G.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -29,8 +31,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1283",
-    "judges": "On Rehearing",
-    "case_name_shorts": "M.G."
+    "judges": "",
+    "case_name_shorts": "M.G.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -40,8 +43,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1177",
-    "judges": "Per Curiam",
-    "case_name_shorts": "K.D.S."
+    "judges": "",
+    "case_name_shorts": "K.D.S.",
+    "per_curiam": true
   },
   {
     "case_dates": "2023-11-09",
@@ -51,8 +55,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1176",
-    "judges": "Per Curiam",
-    "case_name_shorts": "K.D.S."
+    "judges": "",
+    "case_name_shorts": "K.D.S.",
+    "per_curiam": true
   },
   {
     "case_dates": "2023-11-09",
@@ -63,7 +68,8 @@
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1265",
     "judges": "Fridy, J.",
-    "case_name_shorts": ""
+    "case_name_shorts": "",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -73,8 +79,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1095",
-    "judges": "On Rehearing",
-    "case_name_shorts": "D.H."
+    "judges": "",
+    "case_name_shorts": "D.H.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -84,8 +91,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1282",
-    "judges": "On Rehearing",
-    "case_name_shorts": "D.G."
+    "judges": "",
+    "case_name_shorts": "D.G.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -95,8 +103,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1281",
-    "judges": "On Rehearing",
-    "case_name_shorts": "D.G."
+    "judges": "",
+    "case_name_shorts": "D.G.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -106,7 +115,8 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2023-0255",
-    "judges": "On Rehearing",
-    "case_name_shorts": "B.W."
+    "judges": "",
+    "case_name_shorts": "B.W.",
+    "per_curiam": false
   }
 ]
diff --git a/tests/examples/opinions/united_states/ala_example.json b/tests/examples/opinions/united_states/ala_example.json
diff --git a/tests/examples/opinions/united_states/alacivapp_example.compare.json b/tests/examples/opinions/united_states/alacivapp_example.compare.json
@@ -8,7 +8,8 @@
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1230",
     "judges": "Fridy, J.",
-    "case_name_shorts": ""
+    "case_name_shorts": "",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -18,8 +19,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1284",
-    "judges": "On Rehearing",
-    "case_name_shorts": "M.G."
+    "judges": "",
+    "case_name_shorts": "M.G.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -29,8 +31,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1283",
-    "judges": "On Rehearing",
-    "case_name_shorts": "M.G."
+    "judges": "",
+    "case_name_shorts": "M.G.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -40,8 +43,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1177",
-    "judges": "Per Curiam",
-    "case_name_shorts": "K.D.S."
+    "judges": "",
+    "case_name_shorts": "K.D.S.",
+    "per_curiam": true
   },
   {
     "case_dates": "2023-11-09",
@@ -51,8 +55,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1176",
-    "judges": "Per Curiam",
-    "case_name_shorts": "K.D.S."
+    "judges": "",
+    "case_name_shorts": "K.D.S.",
+    "per_curiam": true
   },
   {
     "case_dates": "2023-11-09",
@@ -63,7 +68,8 @@
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1265",
     "judges": "Fridy, J.",
-    "case_name_shorts": ""
+    "case_name_shorts": "",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -73,8 +79,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1095",
-    "judges": "On Rehearing",
-    "case_name_shorts": "D.H."
+    "judges": "",
+    "case_name_shorts": "D.H.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -84,8 +91,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1282",
-    "judges": "On Rehearing",
-    "case_name_shorts": "D.G."
+    "judges": "",
+    "case_name_shorts": "D.G.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -95,8 +103,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2022-1281",
-    "judges": "On Rehearing",
-    "case_name_shorts": "D.G."
+    "judges": "",
+    "case_name_shorts": "D.G.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -106,7 +115,8 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CL-2023-0255",
-    "judges": "On Rehearing",
-    "case_name_shorts": "B.W."
+    "judges": "",
+    "case_name_shorts": "B.W.",
+    "per_curiam": false
   }
 ]
diff --git a/tests/examples/opinions/united_states/alacivapp_example.json b/tests/examples/opinions/united_states/alacivapp_example.json
diff --git a/tests/examples/opinions/united_states/alacrimapp_example.compare.json b/tests/examples/opinions/united_states/alacrimapp_example.compare.json
@@ -7,8 +7,9 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CR-2023-0461",
-    "judges": "On Rehearing",
-    "case_name_shorts": "S.M."
+    "judges": "",
+    "case_name_shorts": "S.M.",
+    "per_curiam": false
   },
   {
     "case_dates": "2023-11-09",
@@ -18,7 +19,8 @@
     "blocked_statuses": false,
     "date_filed_is_approximate": false,
     "docket_numbers": "CR-2022-1020",
-    "judges": "On Rehearing",
-    "case_name_shorts": ""
+    "judges": "",
+    "case_name_shorts": "",
+    "per_curiam": false
   }
 ]
diff --git a/tests/examples/opinions/united_states/alacrimapp_example.json b/tests/examples/opinions/united_states/alacrimapp_example.json