Revert "Remove corus stubs" (#14158)

Author: AlexWaygood

pyrightconfig.stricter.json

@@ -35,6 +35,7 @@
         "stubs/braintree",
         "stubs/cffi",
         "stubs/click-web",
+        "stubs/corus",
         "stubs/dateparser",
         "stubs/defusedxml",
         "stubs/docker",

stubs/corus/METADATA.toml

@@ -0,0 +1,6 @@
+version = "0.10.*"
+upstream_repository = "https://github.com/natasha/corus"
+
+[tool.stubtest]
+# As of version 0.10.0, corus doesn't support Python 3.13.
+skip = true

stubs/corus/corus/__init__.pyi

@@ -0,0 +1 @@
+from .sources import *

stubs/corus/corus/io.pyi

@@ -0,0 +1,21 @@
+from _typeshed import Incomplete
+from collections.abc import Generator
+
+def match_names(records, pattern) -> Generator[Incomplete]: ...
+def rstrip(text): ...
+def load_text(path): ...
+def dump_text(text, path) -> None: ...
+def load_lines(path, encoding: str = "utf-8") -> Generator[Incomplete]: ...
+def parse_xml(text): ...
+def load_z_lines(path, open, encoding: str = "utf8") -> Generator[Incomplete]: ...
+def load_gz_lines(path): ...
+def load_bz2_lines(path): ...
+def load_xz_lines(path): ...
+def list_zip(path): ...
+def load_zip_lines(path, name, encoding: str = "utf8") -> Generator[Incomplete]: ...
+def load_zip_texts(path, names, encoding: str = "utf8") -> Generator[Incomplete]: ...
+def parse_csv(lines, delimiter: str = ",", max_field=None): ...
+def parse_tsv(lines): ...
+def skip_header(rows): ...
+def dict_csv(rows) -> Generator[Incomplete]: ...
+def parse_jsonl(lines) -> Generator[Incomplete]: ...

stubs/corus/corus/path.pyi

@@ -0,0 +1,17 @@
+from _typeshed import Incomplete
+from collections.abc import Generator
+
+COMMANDS: Incomplete
+KB: int
+MB: Incomplete
+GB: Incomplete
+LABELS: Incomplete
+
+def is_command(step, commands=("wget", "unzip", "unrar", "rm", "mv", "tar")): ...
+def format_bytes(value): ...
+def format_count(value): ...
+def unfold_metas(items) -> Generator[Incomplete]: ...
+def format_metas_(metas, nbviewer=None) -> Generator[Incomplete]: ...
+def format_metas(metas, url=None): ...
+def show_html(html) -> None: ...
+def patch_readme(html, path) -> None: ...

stubs/corus/corus/readme.pyi

@@ -0,0 +1,8 @@
+from _typeshed import Incomplete
+
+class Record:
+    __attributes__: Incomplete
+    def __eq__(self, other): ...
+    def __ne__(self, other): ...
+    def __iter__(self): ...
+    def __hash__(self): ...

stubs/corus/corus/record.pyi

@@ -0,0 +1,23 @@
+from .bsnlp import load_bsnlp as load_bsnlp
+from .buriy import *
+from .corpora import load_corpora as load_corpora
+from .factru import load_factru as load_factru
+from .gareev import load_gareev as load_gareev
+from .gramru import load_gramru as load_gramru
+from .lenta import load_lenta as load_lenta, load_lenta2 as load_lenta2
+from .librusec import load_librusec as load_librusec
+from .mokoron import *
+from .morphoru import *
+from .ne5 import load_ne5 as load_ne5
+from .ods import *
+from .omnia import load_omnia as load_omnia
+from .persons import load_persons as load_persons
+from .ria import *
+from .rudrec import load_rudrec as load_rudrec
+from .russe import *
+from .simlex import load_simlex as load_simlex
+from .taiga import *
+from .toloka import load_ruadrect as load_ruadrect, load_toloka_lrwc as load_toloka_lrwc
+from .ud import *
+from .wiki import load_wiki as load_wiki
+from .wikiner import load_wikiner as load_wikiner

stubs/corus/corus/sources/__init__.pyi

@@ -0,0 +1,75 @@
+from _typeshed import Incomplete
+from collections.abc import Generator
+
+from corus.record import Record
+
+RU: str
+BG: str
+CS: str
+PL: str
+LANGS: Incomplete
+ANNOTATED: str
+RAW: str
+TXT: str
+OUT: str
+
+class BsnlpId(Record):
+    __attributes__: Incomplete
+    lang: Incomplete
+    type: Incomplete
+    name: Incomplete
+    path: Incomplete
+    def __init__(self, lang, type, name, path) -> None: ...
+
+class BsnlpRaw(Record):
+    __attributes__: Incomplete
+    id: Incomplete
+    name: Incomplete
+    lang: Incomplete
+    date: Incomplete
+    url: Incomplete
+    text: Incomplete
+    def __init__(self, id, name, lang, date, url, text) -> None: ...
+
+class BsnlpAnnotated(Record):
+    __attributes__: Incomplete
+    id: Incomplete
+    name: Incomplete
+    substrings: Incomplete
+    def __init__(self, id, name, substrings) -> None: ...
+
+class BsnlpSubstring(Record):
+    __attributes__: Incomplete
+    text: Incomplete
+    normal: Incomplete
+    type: Incomplete
+    id: Incomplete
+    def __init__(self, text, normal, type, id) -> None: ...
+
+class BsnlpMarkup(Record):
+    __attributes__: Incomplete
+    id: Incomplete
+    name: Incomplete
+    lang: Incomplete
+    date: Incomplete
+    url: Incomplete
+    text: Incomplete
+    substrings: Incomplete
+    def __init__(self, id, name, lang, date, url, text, substrings) -> None: ...
+
+def walk(dir): ...
+def load_ids(dir, langs) -> Generator[Incomplete]: ...
+def select_type(ids, type) -> Generator[Incomplete]: ...
+
+RAW_PATTERN: Incomplete
+
+def parse_raw(name, text): ...
+def load_raw(records) -> Generator[Incomplete]: ...
+
+ANNOTATED_PATTERN: Incomplete
+
+def parse_substrings(lines) -> Generator[Incomplete]: ...
+def parse_annotated(name, lines): ...
+def load_annotated(records) -> Generator[Incomplete]: ...
+def merge(raw, annotated) -> Generator[Incomplete]: ...
+def load_bsnlp(dir, langs=["ru"]): ...

stubs/corus/corus/sources/bsnlp.pyi

@@ -0,0 +1,18 @@
+from _typeshed import Incomplete
+
+from corus.record import Record
+
+__all__ = ["load_buriy_news", "load_buriy_webhose"]
+
+class BuriyRecord(Record):
+    __attributes__: Incomplete
+    timestamp: Incomplete
+    url: Incomplete
+    edition: Incomplete
+    topics: Incomplete
+    title: Incomplete
+    text: Incomplete
+    def __init__(self, timestamp, url, edition, topics, title, text) -> None: ...
+
+def load_buriy_news(path): ...
+def load_buriy_webhose(path): ...