zip zip zip

This commit is contained in:
Geoffrey Frogeye 2020-03-12 17:56:10 +01:00
parent a6755863dd
commit c8aaa78f59
5 changed files with 175 additions and 33 deletions

View file

@ -7,6 +7,7 @@ sudo pacman -S etckeeper --needed
sudo etckeeper commit "install-arch script: begin"
# Install yay
# TODO Use yay-bin
sudo pacman -S base-devel --needed
if ! pacman -Q yay &> /dev/null
then

View file

@ -3,16 +3,18 @@
import os
import re
import typing
import datetime
import PIL.ExifTags
import PIL.Image
import progressbar
EXTENSION_PATTERN = re.compile(r'\.JPE?G', re.I)
COMMON_PATTERN = re.compile(r'(IMG|DSC[NF]?|100|P10)_?\d+', re.I)
COMMON_PATTERN = re.compile(r'(IMG|DSC[NF]?|100|P10|f|t)_?\d+', re.I)
EXIF_TAG_NAME = 'DateTimeOriginal'
EXIF_TAG_ID = list(PIL.ExifTags.TAGS.keys())[list(
PIL.ExifTags.TAGS.values()).index(EXIF_TAG_NAME)]
EXIF_DATE_FORMAT = '%Y:%m:%d %H:%M:%S'
def get_pictures(directory: str = ".", skip_renamed: bool = True) \
@ -41,7 +43,10 @@ def main() -> None:
exif_data = img._getexif()
if exif_data and EXIF_TAG_ID in exif_data:
date_raw = exif_data[EXIF_TAG_ID]
# print(date_raw, full_path)
date = datetime.datetime.strptime(date_raw, EXIF_DATE_FORMAT)
new_name = date.isoformat().replace(':', '-') + '.jpg' # For NTFS
print(full_path, new_name)
os.rename(full_path, new_name) # TODO FOLDER
img.close()

View file

@ -83,7 +83,7 @@ echo "Finding packages to install"
# i A || i B: will try A for each installer, then B for each installer
# Package managers (install first)
i yay
i yay-bin
i python-pip python3-pip pip3
# TODO Install first and recharge installers
@ -107,7 +107,7 @@ i tmux # Terminal multiplexer
i bash-completion # Shell completions
i fzf # Fancy file finder
i highlight # Syntax highlighter (TODO No termux)
i powerline-go # Nice prompt (potential problem: requires go-pie)
i powerline-go-bin powerline-go # Nice prompt (potential problem: requires go-pie)
i zsh # Shell
# TODO Arch only for the following but not a problem
@ -135,6 +135,7 @@ i socat # Multi-purpose relay
i rsync # Remote file-copying tool
i speedtest-cli # Network speed benchmarker (TODO No termux)
i bind-tools dnsutils # DNS queryier
i whois # Domain name queryier
# Archives utilities
i unzip # Unarchive ZIP files
@ -203,6 +204,7 @@ then
i man # Documentation
i strace # Tracer
i ctags universal-ctags exuberant-ctags # Tags generator
i perf # Allow verifying performance of software
# C/C++
i cmake # C++ Build system
@ -250,6 +252,12 @@ then
i inkscape # Vector image converter
i optipng # Optimize PNG files
i jpegtran libjpeg-turbo # Compress JPEG files
i libreoffice-fresh # Office suite
i hunspell-en_GB # Spell checker British english
i hunspell-en_US # Spell checker American english
# i libreoffice-extension-grammalecte-fr # Spell checker French
i libreoffice-extension-languagetool # Complimentary spell checker various
# TODO The last two can't cohabit
fi
if $INSTALL_IMAGE

View file

@ -2,33 +2,161 @@
import os
import subprocess
import sys
import logging
import magic
import typing
import coloredlogs
import enum
for root, dirs, files in os.walk("."):
for name in files:
base, ext = os.path.splitext(name)
if name.endswith(".zip"):
cmd = ["unzip"]
elif name.endswith(".7z"):
cmd = ["7z", "e"]
elif name.endswith(".rar"):
cmd = ["unrar", "x"]
elif name.endswith('.tar'):
cmd = ["tar", "xf"]
elif name.endswith('.tar.gz'):
cmd = ["tar", "xzf"]
elif name.endswith('.tar.xz'):
cmd = ["tar", "xJf"]
# TODO Able to ignore extensions everywhere
class ArchiveType():
suffix: str = ''
dest_suffix: str = ''
mime: typing.Optional[str] = None
header: typing.Optional[bytes] = None
extract_cmd: typing.Optional[typing.List[str]] = None
single_file = False
append_dest = False
def __init__(self) -> None:
self.log = logging.getLogger(self.__class__.__name__)
def dest_name(self, archive: str) -> str:
return archive[:-len(self.suffix)] + self.dest_suffix
def fits(self, name_lower: str, mime: str, header: bytes) -> bool:
if not name_lower.endswith(self.suffix):
return False
if self.mime is not None and mime != self.mime:
return False
if self.header is not None and not header.startswith(self.header):
return False
return True
def _get_cmd(self, archive: str, dest: str) -> typing.List[str]:
assert self.extract_cmd
cmd = self.extract_cmd + [archive]
if self.append_dest:
cmd.append(dest)
return cmd
def extract(self, archive: str, dest: str) -> None:
cmd = self._get_cmd(archive, dest)
if not self.single_file:
os.mkdir(dest)
self.log.info("Extracting '%s' into '%s'", archive, dest)
self.log.debug("%s", cmd)
if self.single_file:
r = subprocess.run(cmd)
else:
continue
filepath = os.path.join(root, name)
dirpath = os.path.join(root, base)
print(filepath)
os.mkdir(dirpath)
cmd.append(os.path.realpath(filepath))
r = subprocess.run(cmd, cwd=dirpath)
r = subprocess.run(cmd, cwd=dest)
r.check_returncode()
if self.single_file:
assert os.path.isfile(dest)
os.unlink(archive)
os.unlink(filepath)
extract_fun: typing.Optional[typing.Callable[[str, str], None]] = None
class ArchiveZip(ArchiveType):
suffix = '.zip'
mime = 'application/zip'
extract_cmd = ['unzip']
class Archive7z(ArchiveType):
suffix = '.7z'
mime = 'application/x-7z-compressed'
extract_cmd = ['7z', 'x']
class ArchiveRar(ArchiveType):
suffix = '.rar'
mime = 'application/x-rar'
extract_cmd = ['unrar', 'x']
class ArchiveTar(ArchiveType):
suffix = '.tar'
mime = 'application/x-tar'
extract_cmd = ['tar', '--extract', '--file']
class ArchiveTarGz(ArchiveType):
suffix = '.tar.gz'
mime = 'application/gzip'
extract_cmd = ['tar', '--extract', '--gzip', '--file']
class ArchiveTarXz(ArchiveType):
suffix = '.tar.xz'
mime = 'application/x-xz'
extract_cmd = ['tar', '--extract', '--xz', '--file']
class ArchiveGzip(ArchiveType):
suffix = '.gz'
mime = 'application/gzip'
single_file = True
extract_cmd = ['gunzip']
class TreeExtractor():
ARCHIVE_TYPES: typing.List[ArchiveType] = [
ArchiveZip(),
Archive7z(),
ArchiveRar(),
ArchiveTar(),
ArchiveTarGz(),
ArchiveTarXz(),
ArchiveGzip(),
]
def __init__(self) -> None:
self.log = logging.getLogger('TreeExtractor')
self.extensions = set()
for archive_type in self.ARCHIVE_TYPES:
suffixes = archive_type.suffix.split('.')
self.extensions.add('.' + suffixes[-1])
def extract_tree(self, directory: str = '.') -> None:
for root, dirs, files in os.walk(directory):
real_root = os.path.realpath(root)
for name in files:
self.log.debug("Handling '%s' '%s'", real_root, name)
# Initial filtering with extensions
extension = os.path.splitext(name)[1].lower()
if extension not in self.extensions:
self.log.debug("Extension not matched: %s", name)
continue
name_lower = name.lower()
filepath = os.path.join(real_root, name)
with open(filepath, 'rb') as filedesc:
header = filedesc.read(1024)
mime = magic.from_buffer(header, mime=True)
archive_type = None
for archtyp in self.ARCHIVE_TYPES:
if archtyp.fits(name_lower, mime, header):
archive_type = archtyp
break
if not archive_type:
self.log.debug("Not matched: %s", filepath)
continue
dest_name = archive_type.dest_name(name)
dest = os.path.join(real_root, dest_name)
try:
archive_type.extract(filepath, dest)
except BaseException as e:
# TODO Parameters stop on error
self.log.error(e, exc_info=True)
if os.path.isdir(dest):
self.extract_tree(dest)
def main(self) -> None:
directory = sys.argv[1] if len(sys.argv) > 1 else '.'
self.extract_tree(directory)
if __name__ == '__main__':
coloredlogs.install(level='DEBUG', fmt='%(levelname)s %(message)s')
TreeExtractor().main()