pegasust
/
tantivy-py
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

expose regex query (#241) 

Co-authored-by: alexau <alexau@hket.com>
master
alex-au-922 2024-04-25 10:57:09 +08:00 committed by GitHub
parent c74990aeb8
commit 5c3666349b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 140 additions and 87 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
src/query.rs
View File

@ -1,4 +1,4 @@
use crate::{make_term, Schema}; use crate::{get_field, make_term, to_pyerr, Schema};
use pyo3::{ use pyo3::{
exceptions, exceptions,
prelude::*, prelude::*,
@ -187,4 +187,23 @@ impl Query {
inner: Box::new(inner), inner: Box::new(inner),
}) })
} }
#[staticmethod]
#[pyo3(signature = (schema, field_name, regex_pattern))]
pub(crate) fn regex_query(
schema: &Schema,
field_name: &str,
regex_pattern: &str,
) -> PyResult<Query> {
let field = get_field(&schema.inner, field_name)?;
let inner_result =
tv::query::RegexQuery::from_pattern(regex_pattern, field);
match inner_result {
Ok(inner) => Ok(Query {
inner: Box::new(inner),
}),
Err(e) => Err(to_pyerr(e)),
}
}
} }

173
tantivy/tantivy.pyi
View File

@ -2,108 +2,105 @@ import datetime
from enum import Enum from enum import Enum
from typing import Any, Optional, Sequence from typing import Any, Optional, Sequence
class Schema: class Schema:
pass pass
class SchemaBuilder: class SchemaBuilder:
@staticmethod @staticmethod
def is_valid_field_name(name: str) -> bool: def is_valid_field_name(name: str) -> bool:
pass pass
def add_text_field( def add_text_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
tokenizer_name: str = "default", tokenizer_name: str = "default",
index_option: str = "position", index_option: str = "position",
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_integer_field( def add_integer_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
indexed: bool = False, indexed: bool = False,
fast: bool = False, fast: bool = False,
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_float_field( def add_float_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
indexed: bool = False, indexed: bool = False,
fast: bool = False, fast: bool = False,
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_unsigned_field( def add_unsigned_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
indexed: bool = False, indexed: bool = False,
fast: bool = False, fast: bool = False,
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_boolean_field( def add_boolean_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
indexed: bool = False, indexed: bool = False,
fast: bool = False, fast: bool = False,
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_date_field( def add_date_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
indexed: bool = False, indexed: bool = False,
fast: bool = False, fast: bool = False,
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_json_field( def add_json_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
tokenizer_name: str = "default", tokenizer_name: str = "default",
index_option: str = "position", index_option: str = "position",
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_facet_field( def add_facet_field(
self, self,
name: str, name: str,
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_bytes_field( def add_bytes_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
indexed: bool = False, indexed: bool = False,
fast: bool = False, fast: bool = False,
index_option: str = "position", index_option: str = "position",
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def add_ip_addr_field( def add_ip_addr_field(
self, self,
name: str, name: str,
stored: bool = False, stored: bool = False,
indexed: bool = False, indexed: bool = False,
fast: bool = False, fast: bool = False,
) -> SchemaBuilder: ) -> SchemaBuilder:
pass pass
def build(self) -> Schema: def build(self) -> Schema:
pass pass
class Facet: class Facet:
@staticmethod @staticmethod
def from_encoded(encoded_bytes: bytes) -> Facet: def from_encoded(encoded_bytes: bytes) -> Facet:
@ -130,9 +127,7 @@ class Facet:
def to_path_str(self) -> str: def to_path_str(self) -> str:
pass pass
class Document: class Document:
def __new__(cls, **kwargs) -> Document: def __new__(cls, **kwargs) -> Document:
pass pass
@ -194,7 +189,12 @@ class Occur(Enum):
class Query: class Query:
@staticmethod @staticmethod
def term_query(schema: Schema, field_name: str, field_value: Any, index_option: str = "position") -> Query: def term_query(
schema: Schema,
field_name: str,
field_value: Any,
index_option: str = "position",
) -> Query:
pass pass
@staticmethod @staticmethod
@ -202,9 +202,16 @@ class Query:
pass pass
@staticmethod @staticmethod
def fuzzy_term_query(schema: Schema, field_name: str, text: str, distance: int = 1, transposition_cost_one: bool = True, prefix = False) -> Query: def fuzzy_term_query(
schema: Schema,
field_name: str,
text: str,
distance: int = 1,
transposition_cost_one: bool = True,
prefix=False,
) -> Query:
pass pass
@staticmethod @staticmethod
def boolean_query(subqueries: Sequence[tuple[Occur, Query]]) -> Query: def boolean_query(subqueries: Sequence[tuple[Occur, Query]]) -> Query:
pass pass
@ -218,13 +225,15 @@ class Query:
pass pass
@staticmethod
def regex_query(schema: Schema, field_name: str, regex_pattern: str) -> Query:
pass
class Order(Enum): class Order(Enum):
Asc = 1 Asc = 1
Desc = 2 Desc = 2
class DocAddress: class DocAddress:
def __new__(cls, segment_ord: int, doc: int) -> DocAddress: def __new__(cls, segment_ord: int, doc: int) -> DocAddress:
pass pass
@ -237,22 +246,19 @@ class DocAddress:
pass pass
class SearchResult: class SearchResult:
@property @property
def hits(self) -> list[tuple[Any, DocAddress]]: def hits(self) -> list[tuple[Any, DocAddress]]:
pass pass
class Searcher: class Searcher:
def search( def search(
self, self,
query: Query, query: Query,
limit: int = 10, limit: int = 10,
count: bool = True, count: bool = True,
order_by_field: Optional[str] = None, order_by_field: Optional[str] = None,
offset: int = 0, offset: int = 0,
order: Order = Order.Desc, order: Order = Order.Desc,
) -> SearchResult: ) -> SearchResult:
pass pass
@ -267,9 +273,7 @@ class Searcher:
def doc(self, doc_address: DocAddress) -> Document: def doc(self, doc_address: DocAddress) -> Document:
pass pass
class IndexWriter: class IndexWriter:
def add_document(self, doc: Document) -> int: def add_document(self, doc: Document) -> int:
pass pass
@ -298,10 +302,10 @@ class IndexWriter:
def wait_merging_threads(self) -> None: def wait_merging_threads(self) -> None:
pass pass
class Index: class Index:
def __new__(
def __new__(cls, schema: Schema, path: Optional[str] = None, reuse: bool = True) -> Index: cls, schema: Schema, path: Optional[str] = None, reuse: bool = True
) -> Index:
pass pass
@staticmethod @staticmethod
@ -311,7 +315,9 @@ class Index:
def writer(self, heap_size: int = 128_000_000, num_threads: int = 0) -> IndexWriter: def writer(self, heap_size: int = 128_000_000, num_threads: int = 0) -> IndexWriter:
pass pass
def config_reader(self, reload_policy: str = "commit", num_warmers: int = 0) -> None: def config_reader(
self, reload_policy: str = "commit", num_warmers: int = 0
) -> None:
pass pass
def searcher(self) -> Searcher: def searcher(self) -> Searcher:
@ -328,15 +334,17 @@ class Index:
def reload(self) -> None: def reload(self) -> None:
pass pass
def parse_query(self, query: str, default_field_names: Optional[list[str]] = None) -> Query: def parse_query(
self, query: str, default_field_names: Optional[list[str]] = None
) -> Query:
pass pass
def parse_query_lenient(self, query: str, default_field_names: Optional[list[str]] = None) -> Query: def parse_query_lenient(
self, query: str, default_field_names: Optional[list[str]] = None
) -> Query:
pass pass
class Range: class Range:
@property @property
def start(self) -> int: def start(self) -> int:
pass pass
@ -345,24 +353,17 @@ class Range:
def end(self) -> int: def end(self) -> int:
pass pass
class Snippet: class Snippet:
def to_html(self) -> str: def to_html(self) -> str:
pass pass
def highlighted(self) -> list[Range]: def highlighted(self) -> list[Range]:
pass pass
class SnippetGenerator: class SnippetGenerator:
@staticmethod @staticmethod
def create( def create(
searcher: Searcher, searcher: Searcher, query: Query, schema: Schema, field_name: str
query: Query,
schema: Schema,
field_name: str
) -> SnippetGenerator: ) -> SnippetGenerator:
pass pass

33
tests/tantivy_test.py
View File

@ -995,3 +995,36 @@ class TestQuery(object):
# no boost type error # no boost type error
with pytest.raises(TypeError, match = r"Query.boost_query\(\) missing 1 required positional argument: 'boost'"): with pytest.raises(TypeError, match = r"Query.boost_query\(\) missing 1 required positional argument: 'boost'"):
Query.boost_query(query1) Query.boost_query(query1)
def test_regex_query(self, ram_index):
index = ram_index
query = Query.regex_query(index.schema, "body", "fish")
result = index.searcher().search(query, 10)
assert len(result.hits) == 1
_, doc_address = result.hits[0]
searched_doc = index.searcher().doc(doc_address)
assert searched_doc["title"] == ["The Old Man and the Sea"]
query = Query.regex_query(index.schema, "title", "(?:man|men)")
result = index.searcher().search(query, 10)
assert len(result.hits) == 2
_, doc_address = result.hits[0]
searched_doc = index.searcher().doc(doc_address)
assert searched_doc["title"] == ["The Old Man and the Sea"]
_, doc_address = result.hits[1]
searched_doc = index.searcher().doc(doc_address)
assert searched_doc["title"] == ["Of Mice and Men"]
# unknown field in the schema
with pytest.raises(
ValueError, match="Field `unknown_field` is not defined in the schema."
):
Query.regex_query(index.schema, "unknown_field", "fish")
# invalid regex pattern
with pytest.raises(
ValueError, match=r"An invalid argument was passed: 'fish\('"
):
Query.regex_query(index.schema, "body", "fish(")