Move file handling to corelibs_file module

This commit is contained in:
Clemens Schwaighofer
2026-02-03 11:42:57 +09:00
parent 2e0d5aeb51
commit a046d9f84c
8 changed files with 37 additions and 1539 deletions

View File

@@ -2,19 +2,16 @@
File check if BOM encoded, needed for CSV load
"""
from warnings import deprecated
from pathlib import Path
from typing import TypedDict
class BomEncodingInfo(TypedDict):
"""BOM encoding info"""
has_bom: bool
bom_type: str | None
encoding: str | None
bom_length: int
bom_pattern: bytes | None
from corelibs_file.file_bom_encoding import (
is_bom_encoded as is_bom_encoding_ng,
get_bom_encoding_info,
BomEncodingInfo
)
@deprecated("Use corelibs_file.file_bom_encoding.is_bom_encoded instead")
def is_bom_encoded(file_path: Path) -> bool:
"""
Detect if a file is BOM encoded
@@ -25,9 +22,10 @@ def is_bom_encoded(file_path: Path) -> bool:
Returns:
bool: True if file has BOM, False otherwise
"""
return is_bom_encoded_info(file_path)['has_bom']
return is_bom_encoding_ng(file_path)
@deprecated("Use corelibs_file.file_bom_encoding.is_bom_encoded_info instead")
def is_bom_encoded_info(file_path: Path) -> BomEncodingInfo:
"""
Enhanced BOM detection with additional file analysis
@@ -38,38 +36,7 @@ def is_bom_encoded_info(file_path: Path) -> BomEncodingInfo:
Returns:
dict: Comprehensive BOM and encoding information
"""
try:
# Read first 1024 bytes for analysis
with open(file_path, 'rb') as f:
header = f.read(4)
bom_patterns = {
b'\xef\xbb\xbf': ('UTF-8', 'utf-8', 3),
b'\xff\xfe\x00\x00': ('UTF-32 LE', 'utf-32-le', 4),
b'\x00\x00\xfe\xff': ('UTF-32 BE', 'utf-32-be', 4),
b'\xff\xfe': ('UTF-16 LE', 'utf-16-le', 2),
b'\xfe\xff': ('UTF-16 BE', 'utf-16-be', 2),
}
for bom_pattern, (encoding_name, encoding, length) in bom_patterns.items():
if header.startswith(bom_pattern):
return {
'has_bom': True,
'bom_type': encoding_name,
'encoding': encoding,
'bom_length': length,
'bom_pattern': bom_pattern
}
return {
'has_bom': False,
'bom_type': None,
'encoding': None,
'bom_length': 0,
'bom_pattern': None
}
except Exception as e:
raise ValueError(f"Error checking BOM encoding: {e}") from e
return get_bom_encoding_info(file_path)
# __END__

View File

@@ -2,10 +2,13 @@
crc handlers for file CRC
"""
import zlib
from warnings import deprecated
from pathlib import Path
from corelibs_file.file_crc import file_crc as file_crc_ng
from corelibs_file.file_handling import get_file_name
@deprecated("Use corelibs_file.file_crc.file_crc instead")
def file_crc(file_path: Path) -> str:
"""
With for loop and buffer, create file crc32
@@ -16,13 +19,10 @@ def file_crc(file_path: Path) -> str:
Returns:
str: file crc32
"""
crc = 0
with open(file_path, 'rb', 65536) as ins:
for _ in range(int((file_path.stat().st_size / 65536)) + 1):
crc = zlib.crc32(ins.read(65536), crc)
return f"{crc & 0xFFFFFFFF:08X}"
return file_crc_ng(file_path)
@deprecated("Use corelibs_file.file_handling.get_file_name instead")
def file_name_crc(file_path: Path, add_parent_folder: bool = False) -> str:
"""
either returns file name only from path
@@ -38,9 +38,6 @@ def file_name_crc(file_path: Path, add_parent_folder: bool = False) -> str:
Returns:
str: file name as string
"""
if add_parent_folder:
return str(Path(file_path.parent.name).joinpath(file_path.name))
else:
return file_path.name
return get_file_name(file_path, add_parent_folder=add_parent_folder)
# __END__

View File

@@ -2,11 +2,12 @@
File handling utilities
"""
import os
import shutil
from warnings import deprecated
from pathlib import Path
from corelibs_file.file_handling import remove_all_in_directory as remove_all_in_directory_ng
@deprecated("Use corelibs_file.file_handling.remove_all_in_directory instead")
def remove_all_in_directory(
directory: Path,
ignore_files: list[str] | None = None,
@@ -14,43 +15,24 @@ def remove_all_in_directory(
dry_run: bool = False
) -> bool:
"""
remove all files and folders in a directory
can exclude files or folders
deprecated
Args:
directory (Path): _description_
ignore_files (list[str], optional): _description_. Defaults to None.
Arguments:
directory {Path} -- _description_
Keyword Arguments:
ignore_files {list[str] | None} -- _description_ (default: {None})
verbose {bool} -- _description_ (default: {False})
dry_run {bool} -- _description_ (default: {False})
Returns:
bool: _description_
bool -- _description_
"""
if not directory.is_dir():
return False
if ignore_files is None:
ignore_files = []
if verbose:
print(
f"{'[DRY RUN] ' if dry_run else ''}Remove old files in: {directory.name} [",
end="", flush=True
)
# remove all files and folders in given directory by recursive globbing
for file in directory.rglob("*"):
# skip if in ignore files
if file.name in ignore_files:
continue
# remove one file, or a whole directory
if file.is_file():
if not dry_run:
os.remove(file)
if verbose:
print(".", end="", flush=True)
elif file.is_dir():
if not dry_run:
shutil.rmtree(file)
if verbose:
print("/", end="", flush=True)
if verbose:
print("]", flush=True)
return True
return remove_all_in_directory_ng(
directory,
ignore_files=ignore_files,
verbose=verbose,
dry_run=dry_run
)
# __END__