FastWARC 1.0.3__tar.gz → 1.0.5__tar.gz

{fastwarc-1.0.3 → fastwarc-1.0.5}/FastWARC.egg-info/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: FastWARC
-Version: 1.0.3
-Summary: A high-performance WARC parsing library written in Rust with bindings for Python.
+Version: 1.0.5
+Summary: The world's fastest WARC parsing library written in Rust with bindings for Python.
 Author: Janek Bevendorff
 License-Expression: Apache-2.0
 Project-URL: repository, https://github.com/chatnoir-eu/chatnoir-resiliparse
@@ -12,6 +12,7 @@ Description-Content-Type: text/markdown
 Requires-Dist: brotli
 Requires-Dist: click
 Requires-Dist: tqdm
+Requires-Dist: typing_extensions>=4.5; python_version < "3.13"
 Provides-Extra: all
 Requires-Dist: fastwarc[fsspec]; extra == "all"
 Provides-Extra: fsspec

{fastwarc-1.0.3 → fastwarc-1.0.5}/FastWARC.egg-info/requires.txt

@@ -2,6 +2,9 @@ brotli
 click
 tqdm
 
+[:python_version < "3.13"]
+typing_extensions>=4.5
+
 [all]
 fastwarc[fsspec]
 

{fastwarc-1.0.3 → fastwarc-1.0.5}/PKG-INFO

@@ -1,7 +1,7 @@
 Metadata-Version: 2.4
 Name: FastWARC
-Version: 1.0.3
-Summary: A high-performance WARC parsing library written in Rust with bindings for Python.
+Version: 1.0.5
+Summary: The world's fastest WARC parsing library written in Rust with bindings for Python.
 Author: Janek Bevendorff
 License-Expression: Apache-2.0
 Project-URL: repository, https://github.com/chatnoir-eu/chatnoir-resiliparse
@@ -12,6 +12,7 @@ Description-Content-Type: text/markdown
 Requires-Dist: brotli
 Requires-Dist: click
 Requires-Dist: tqdm
+Requires-Dist: typing_extensions>=4.5; python_version < "3.13"
 Provides-Extra: all
 Requires-Dist: fastwarc[fsspec]; extra == "all"
 Provides-Extra: fsspec

{fastwarc-1.0.3 → fastwarc-1.0.5}/fastwarc/legacy/_shims.py

@@ -14,7 +14,10 @@
 
 # Legacy shims for making the new Rust extension work with the old type names
 
-import warnings
+try:
+    from warnings import deprecated
+except ImportError:
+    from typing_extensions import deprecated
 
 __all__ = [
     'wrap_stream',
@@ -32,19 +35,21 @@ __all__ = [
     'ReaderStaleError',
 ]
 
-FastWARCError = OSError
 
-StreamError = OSError
+@deprecated("Use OSError instead.", category=FutureWarning)
+class FastWARCError(OSError): ...
 
-ReaderStaleError = OSError
 
+@deprecated("Use OSError instead.", category=FutureWarning)
+class StreamError(OSError): ...
 
-def _warn_deprecated(msg):
-    warnings.warn(msg, FutureWarning, 3)
 
+@deprecated("Use OSError instead.", category=FutureWarning)
+class ReaderStaleError(OSError): ...
 
+
+@deprecated("wrap_stream is deprecated and will be removed in a future version.", category=FutureWarning)
 def wrap_stream(raw_stream, mode='rb', fsspec_args=None):
-    _warn_deprecated("wrap_stream is deprecated and will be removed in a future version.")
     if isinstance(raw_stream, str):
         if fsspec_args is not False and '://' in raw_stream:
             try:
@@ -57,9 +62,9 @@ def wrap_stream(raw_stream, mode='rb', fsspec_args=None):
     return raw_stream
 
 
+@deprecated("BufferedReader is deprecated and will be removed in a future version.", category=FutureWarning)
 class BufferedReader:
     def __init__(self, stream, *args, **kwargs):
-        _warn_deprecated("BufferedReader is deprecated and will be removed in a future version.")
         self._stream = stream
 
     def close(self):
@@ -78,12 +83,9 @@ class BufferedReader:
         return self._stream.tell()
 
 
+@deprecated("Use the new Reader and Writer classes from stream_io instead.")
 class IOStream:
     def __init__(self, *_, _reader_factory=None, _writer_factory=None, **__):
-        _warn_deprecated(
-            f"{self.__class__.__name__} is deprecated and will be removed in a future version. "
-            "Use the new Reader and Writer classes from stream_io instead.")
-
         self._reader_factory = _reader_factory
         self._writer_factory = _writer_factory
         self._reader = None
@@ -140,6 +142,7 @@ class IOStream:
         s.close()
 
 
+@deprecated("Use the new Reader and Writer classes from stream_io instead.", category=FutureWarning)
 class CompressingStream(IOStream):
     def begin_member(self):
         pass
@@ -149,12 +152,14 @@ class CompressingStream(IOStream):
         self._writer = None
 
 
+@deprecated("Use the plain Python stream object without any wrapper.", category=FutureWarning)
 class PythonIOStreamAdapter:
     def __new__(cls, py_stream):
         return IOStream(_reader_factory=lambda: py_stream,
                         _writer_factory=lambda: py_stream)
 
 
+@deprecated("Use plain file paths instead.", category=FutureWarning)
 class FileStream:
     def __new__(cls, filename: str, mode: str = 'rb'):
         if 'b' not in mode:
@@ -163,6 +168,7 @@ class FileStream:
         return IOStream(_reader_factory=lambda: f, _writer_factory=lambda: f)
 
 
+@deprecated("Use io.BytesIO instead.", category=FutureWarning)
 class BytesIOStream:
     def __new__(cls, initial_data=None):
         import io
@@ -170,6 +176,7 @@ class BytesIOStream:
         return IOStream(_reader_factory=lambda: bio, _writer_factory=lambda: bio)
 
 
+@deprecated("Use the new GzipReader and GzipWriter classes from stream_io instead.", category=FutureWarning)
 class GZipStream:
     def __new__(cls, raw_stream, compression_level=9, zlib=False, fsspec_args=None):
         from ..stream_io import GzipReader, GzipWriter
@@ -181,6 +188,7 @@ class GZipStream:
         return stream
 
 
+@deprecated("Use the new Lz4Reader and Lz4Writer classes from stream_io instead.", category=FutureWarning)
 class LZ4Stream:
     def __new__(cls, raw_stream, mode='r', compression_level=12,
                 favor_dec_speed=True, fsspec_args=None):
@@ -192,6 +200,7 @@ class LZ4Stream:
         return stream
 
 
+@deprecated("Use the new BrotliReader and BrotliWriter classes from stream_io instead.", category=FutureWarning)
 class BrotliStream:
     def __new__(cls, raw_stream, quality=11, lgwin=22, lgblock=0, fsspec_args=None):
         from ..stream_io import BrotliReader, BrotliWriter

{fastwarc-1.0.3 → fastwarc-1.0.5}/fastwarc/warc.pyi

@@ -16,7 +16,7 @@ from datetime import datetime
 from collections.abc import Callable, Iterable, Iterator
 from enum import IntFlag
 from typing import BinaryIO, final, Literal, Self
-from typing_extensions import disjoint_base
+from typing_extensions import deprecated, disjoint_base
 
 from .stream_io import _GenericReader, _GenericWriter, _ReaderInput, WarcReader, WarcWriter
 
@@ -95,8 +95,7 @@ no_type = WarcRecordType.no_type
 any_type = WarcRecordType.any_type
 
 
-@final
-class HeaderMap:
+class _HeaderMap:
     def __new__(cls, encoding: str = "utf-8") -> Self: ...
 
     @property
@@ -127,12 +126,12 @@ class HeaderMap:
 
     def to_dict(self) -> dict[str, str]: ...
 
-    # deprecated
+    @deprecated("Use to_dict() instead.")
     def asdict(self) -> dict[str, str]: ...
 
     def to_tuples(self) -> tuple[tuple[str, str], ...]: ...
 
-    # deprecated
+    @deprecated("Use to_tuples() instead.")
     def astuples(self) -> tuple[tuple[str, str], ...]: ...
 
     def is_empty(self) -> bool: ...
@@ -198,8 +197,13 @@ class HeaderMap:
     def __setstate__(self, state: tuple[bytes, bool]): ...
 
 
-# Legacy name
-WarcHeaderMap = HeaderMap
+@final
+class HeaderMap(_HeaderMap): ...
+
+
+@deprecated("Use HeaderMap instead.")
+@final
+class WarcHeaderMap(_HeaderMap): ...
 
 
 @final
@@ -287,10 +291,10 @@ class WarcRecord:
 
     def consume(self, n: int | None = None) -> int: ...
 
-    def parse_warc_headers(self, quirks_mode: bool = False) -> int: ...
+    def parse_warc_headers(self, quirks_mode: bool = False, max_header_len: int = 32 << 10) -> int: ...
 
-    def parse_http(self, auto_decode: Literal['none', 'content', 'transfer', 'all'] = 'none', *,
-                   strict_mode: bool = True): ...
+    def parse_http(self, auto_decode: Literal['none', 'content', 'transfer', 'all'] = 'none',
+                   max_header_len: int = 32 << 10, quirks_mode: bool = False, *, strict_mode: bool = True): ...
 
     def verify_block_digest(self, consume: bool = False) -> bool: ...
 
@@ -332,8 +336,10 @@ class ArchiveIterator(Iterable[WarcRecord]):
             verify_digests: bool = False,
             quirks_mode: bool = False,
             auto_decode: Literal['none', 'content', 'transfer', 'all'] = 'none',
+            max_header_len: int = 32 << 10,
             stream_detect: bool = True,
             buffer_size: int = 64 << 10,
+            inplace: bool = False,
             fsspec_args=None,
             *,
             strict_mode: bool = True

{fastwarc-1.0.3 → fastwarc-1.0.5}/pyproject.toml

@@ -4,8 +4,8 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "FastWARC"
-version = "1.0.3"
-description = "A high-performance WARC parsing library written in Rust with bindings for Python."
+version = "1.0.5"
+description = "The world's fastest WARC parsing library written in Rust with bindings for Python."
 authors = [
     { name = "Janek Bevendorff" }
 ]
@@ -15,7 +15,8 @@ requires-python = ">=3.10"
 dependencies = [
     "brotli",
     "click",
-    "tqdm"
+    "tqdm",
+    "typing_extensions>=4.5; python_version < '3.13'",
 ]
 
 [project.urls]