Merge pull request #191 from nationalarchives/chore/more-types

Add new types to utils to express NCNs and related concepts

.pre-commit-config.yaml

@@ -1,4 +1,4 @@
-exclude: courts_schema_types_autogenerated\.py
+exclude: courts_schema_autogen\.py
 repos:
   - repo: https://github.com/pre-commit/pre-commit-hooks
     rev: v4.6.0

jsonschema-gentypes.yaml

@@ -8,7 +8,7 @@ generate:
   - # JSON schema file path
     source: src/ds_caselaw_utils/data/schema/courts.schema.json
     # Python file path
-    destination: src/ds_caselaw_utils/courts_schema_types_autogenerated.py
+    destination: src/ds_caselaw_utils/types/courts_schema_autogen.py
     # The name of the root element
     root_name: RawCourtRepository
     # Argument passed to the API

src/ds_caselaw_utils/courts.py

@@ -6,19 +6,17 @@
 
 import pathlib
 from datetime import date
-from typing import NewType, Optional
+from typing import Optional
 
 from ruamel.yaml import YAML
 
-from ds_caselaw_utils.courts_schema_types_autogenerated import (
+from ds_caselaw_utils.types.courts_schema_autogen import (
     RawCourt,
     RawCourtRepository,
     RawJurisdiction,
 )
 
-CourtCode = NewType("CourtCode", str)
-CourtParam = NewType("CourtParam", str)
-JurisdictionCode = NewType("JurisdictionCode", str)
+from .types import CourtCode, CourtParam, JurisdictionCode, NeutralCitationString
 
 
 class Jurisdiction:
@@ -36,7 +34,7 @@ def __init__(self, data: RawCourt) -> None:
         self.name: str = data["name"]
         self.grouped_name: str = data.get("grouped_name") or data["name"]
         self.link: str = data["link"]
-        self.ncn: Optional[str] = data.get("ncn")
+        self.ncn: Optional[NeutralCitationString] = NeutralCitationString(data["ncn"]) if "ncn" in data else None
         if "param" in data:
             self.canonical_param = CourtParam(data["param"])
             self.param_aliases = [CourtParam(data["param"])] + [
@@ -85,7 +83,7 @@ def link(self) -> str:
         return self.court.link
 
     @property
-    def ncn(self) -> Optional[str]:
+    def ncn(self) -> Optional[NeutralCitationString]:
         return self.court.ncn
 
     @property

src/ds_caselaw_utils/factory.py

@@ -1,7 +1,7 @@
 import typing
 
 from .courts import Court
-from .courts_schema_types_autogenerated import RawCourt, RawCourtRepository
+from .types.courts_schema_autogen import RawCourt, RawCourtRepository
 
 
 class CourtFactory(Court):

src/ds_caselaw_utils/neutral.py

@@ -8,19 +8,21 @@
 
 from ruamel.yaml import YAML
 
+from .types import NCNBasedUriString, NeutralCitationString
+
 yaml = YAML()
 datafile = pathlib.Path(__file__).parent / "data/neutral_citation_regex.yaml"
 with open(datafile) as f:
     citation_data = yaml.load(f)
 
 
-def neutral_url(citation: str) -> Optional[str]:
+def neutral_url(citation: NeutralCitationString) -> Optional[NCNBasedUriString]:
     """Given a neutral citation such as `[2020] EAT 17`,
     return a public-API URL like `/eat/2020/17`, or None
     if no match is found.
     """
     for regex, groups in citation_data:
         if match := re.match(regex, citation):
             url_components = "/".join([match.groups()[x - 1] for x in groups])
-            return f"/{url_components}".lower()
+            return NCNBasedUriString(f"/{url_components}".lower())
     return None

src/ds_caselaw_utils/test_neutral.py

@@ -1,34 +1,35 @@
 import unittest
 
 from .neutral import neutral_url
+from .types import NeutralCitationString
 
 
 class TestNeutralURL(unittest.TestCase):
     def test_good_neutral_urls(self):
-        self.assertEqual(neutral_url("[2022] UKSC 1"), "/uksc/2022/1")
-        self.assertEqual(neutral_url("[1604] EWCA Crim 555"), "/ewca/crim/1604/555")
-        self.assertEqual(neutral_url("[2022] EWHC 1 (Comm)"), "/ewhc/comm/2022/1")
-        self.assertEqual(neutral_url("[1999] EWCOP 7"), "/ewcop/1999/7")
-        self.assertEqual(neutral_url("[2022] UKUT 1 (IAC)"), "/ukut/iac/2022/1")
-        self.assertEqual(neutral_url("[2022] EAT 1"), "/eat/2022/1")
-        self.assertEqual(neutral_url("[2022] UKFTT 1 (TC)"), "/ukftt/tc/2022/1")
-        self.assertEqual(neutral_url("[2022] UKFTT 1 (GRC)"), "/ukftt/grc/2022/1")
-        self.assertEqual(neutral_url("[2022] EWHC 1 (KB)"), "/ewhc/kb/2022/1")
-        self.assertEqual(neutral_url("[2023] UKAIT 1"), "/ukait/2023/1")
-        self.assertEqual(neutral_url("[2024] EWCOP 17 (T2)"), "/ewcop/t2/2024/17")
-        self.assertEqual(neutral_url("[2000] UKIPTrib 99"), "/ukiptrib/2000/99")
-        self.assertEqual(neutral_url("[2000] EWCR 99"), "/ewcr/2000/99")
-        self.assertEqual(neutral_url("[2000] EWCC 99"), "/ewcc/2000/99")
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] UKSC 1")), "/uksc/2022/1")
+        self.assertEqual(neutral_url(NeutralCitationString("[1604] EWCA Crim 555")), "/ewca/crim/1604/555")
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] EWHC 1 (Comm)")), "/ewhc/comm/2022/1")
+        self.assertEqual(neutral_url(NeutralCitationString("[1999] EWCOP 7")), "/ewcop/1999/7")
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] UKUT 1 (IAC)")), "/ukut/iac/2022/1")
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] EAT 1")), "/eat/2022/1")
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] UKFTT 1 (TC)")), "/ukftt/tc/2022/1")
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] UKFTT 1 (GRC)")), "/ukftt/grc/2022/1")
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] EWHC 1 (KB)")), "/ewhc/kb/2022/1")
+        self.assertEqual(neutral_url(NeutralCitationString("[2023] UKAIT 1")), "/ukait/2023/1")
+        self.assertEqual(neutral_url(NeutralCitationString("[2024] EWCOP 17 (T2)")), "/ewcop/t2/2024/17")
+        self.assertEqual(neutral_url(NeutralCitationString("[2000] UKIPTrib 99")), "/ukiptrib/2000/99")
+        self.assertEqual(neutral_url(NeutralCitationString("[2000] EWCR 99")), "/ewcr/2000/99")
+        self.assertEqual(neutral_url(NeutralCitationString("[2000] EWCC 99")), "/ewcc/2000/99")
 
     def test_bad_neutral_urls(self):
-        self.assertEqual(neutral_url(""), None)
-        self.assertEqual(neutral_url("1604] EWCA Crim 555"), None)
-        self.assertEqual(neutral_url("[2022 EWHC 1 Comm"), None)
-        self.assertEqual(neutral_url("[1999] EWCOP"), None)
+        self.assertEqual(neutral_url(NeutralCitationString("")), None)
+        self.assertEqual(neutral_url(NeutralCitationString("1604] EWCA Crim 555")), None)
+        self.assertEqual(neutral_url(NeutralCitationString("[2022 EWHC 1 Comm")), None)
+        self.assertEqual(neutral_url(NeutralCitationString("[1999] EWCOP")), None)
         self.assertEqual(
-            neutral_url("[2022] UKUT B1 IAC"), None
+            neutral_url(NeutralCitationString("[2022] UKUT B1 IAC")), None
         )  # Could be a Bailii reference, might want to drop B in future.
-        self.assertEqual(neutral_url("[2022] EAT A"), None)
-        self.assertEqual(neutral_url("[2022] NOTACOURT 1 TC"), None)
-        self.assertEqual(neutral_url("[2022] EWHC 1 (T2)"), None)
-        self.assertEqual(neutral_url("[2000] EWCRC 99"), None)
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] EAT A")), None)
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] NOTACOURT 1 TC")), None)
+        self.assertEqual(neutral_url(NeutralCitationString("[2022] EWHC 1 (T2)")), None)
+        self.assertEqual(neutral_url(NeutralCitationString("[2000] EWCRC 99")), None)

src/ds_caselaw_utils/types/__init__.py

@@ -0,0 +1,10 @@
+from typing import NewType
+
+# Types which are used for identifying courts and their jurisdictions
+CourtCode = NewType("CourtCode", str)
+CourtParam = NewType("CourtParam", str)
+JurisdictionCode = NewType("JurisdictionCode", str)
+
+# Types which are used to identify a judgment
+NeutralCitationString = NewType("NeutralCitationString", str)
+NCNBasedUriString = NewType("NCNBasedUriString", str)