caselawclient.search_parameters

 1import re
 2from dataclasses import dataclass
 3from typing import Any, Dict, List, Optional, Set
 4
 5RESULTS_PER_PAGE = 10
 6QUOTED_PHRASE_REGEX = '"([^"]*)"'
 7
 8
 9@dataclass
10class SearchParameters:
11    """Represents search parameters for a case law search."""
12
13    query: Optional[str] = None
14    court: Optional[str] = None
15    judge: Optional[str] = None
16    party: Optional[str] = None
17    neutral_citation: Optional[str] = None
18    document_name: Optional[str] = None
19    consignment_number: Optional[str] = None
20    specific_keyword: Optional[str] = None
21    order: Optional[str] = None
22    date_from: Optional[str] = None
23    date_to: Optional[str] = None
24    page: int = 1
25    page_size: int = RESULTS_PER_PAGE
26    show_unpublished: bool = False
27    only_unpublished: bool = False
28    only_with_html_representation: bool = False
29    collections: Optional[List[str]] = None
30
31    def as_marklogic_payload(self) -> Dict[str, Any]:
32        """
33        Converts the search parameters to a dictionary payload suitable
34        fo MarkLogic.
35        """
36        return {
37            "court": self._marklogic_courts,
38            "judge": str(self.judge or ""),
39            "page": max(1, int(self.page)),
40            "page-size": int(self.page_size),
41            "q": str(self.query or ""),
42            "party": str(self.party or ""),
43            "neutral_citation": str(self.neutral_citation or ""),
44            "document_name": str(self.document_name or ""),
45            "consignment_number": str(self.consignment_number or ""),
46            "specific_keyword": str(self.specific_keyword or ""),
47            "order": str(self.order or ""),
48            "from": str(self.date_from or ""),
49            "to": str(self.date_to or ""),
50            "show_unpublished": str(self.show_unpublished).lower(),
51            "only_unpublished": str(self.only_unpublished).lower(),
52            "only_with_html_representation": str(self.only_with_html_representation).lower(),
53            "collections": self._marklogic_collections,
54            "quoted_phrases": self._quoted_phrases,
55        }
56
57    @property
58    def _marklogic_collections(self) -> str:
59        return ",".join(self.collections or []).replace(" ", "").replace(",,", ",")
60
61    @property
62    def _marklogic_courts(self) -> Optional[List[str]]:
63        court_text = self._join_court_text(self.court or "")
64        if not (court_text).strip():
65            return None
66        courts = self._court_list_splitter(court_text)
67        alternative_court_names = self._get_alternative_court_names(courts)
68        return list(courts | alternative_court_names)
69
70    @property
71    def _quoted_phrases(self) -> List[str]:
72        if self.query is None:
73            return []
74        return re.findall(QUOTED_PHRASE_REGEX, self.query)
75
76    @staticmethod
77    def _join_court_text(court_text: str) -> str:
78        return ",".join(court_text) if isinstance(court_text, list) else court_text
79
80    @staticmethod
81    def _court_list_splitter(court_text: str) -> Set[str]:
82        return set(court_text.lower().replace(" ", "").split(","))
83
84    @staticmethod
85    def _get_alternative_court_names(courts: Set[str]) -> Set[str]:
86        ALTERNATIVE_COURT_NAMES_MAP = {
87            "ewhc/qb": "ewhc/kb",
88            "ewhc/kb": "ewhc/qb",
89            "ewhc/scco": "ewhc/costs",
90            "ewhc/costs": "ewhc/scco",
91            "ukait": "ukut/iac",
92            "ukut/iac": "ukait",
93        }
94        alternative_court_names = set()
95        for primary_name, secondary_name in ALTERNATIVE_COURT_NAMES_MAP.items():
96            if primary_name in courts and secondary_name not in courts:
97                alternative_court_names.add(secondary_name)
98        return alternative_court_names