Multiprocess directory iteration via os.scandir()
Who's this Lib for?
You want to process a large number of files and/or a few very big files and give feedback to the user on how long it will take.
Progress indicator:
- Immediately after start: process files and indication of progress via multiprocess
- process bars via tqdm
- Estimated time based on file count and size
Easy to implement extra process bar for big file processing.
Skip directories and file name via fnmatch.
The main process starts statistic processes in background via Python multiprocess and starts directly with the work.
There are two background statistic processes collects information for the process bars:
- Count up all directories and files.
- Accumulates the sizes of all files.
Why two processes?
Because collect only the count of all filesystem items via os.scandir()
is very fast. This is the fastest way to predict a processing time.
Use os.DirEntry.stat()
to get the file size is significantly slower: It requires another system call.
OK, but why two processed?
Use only the total count of all DirEntry
may result in bad estimated time Progress indication.
It depends on what the actual work is about: When processing the contents of large files, it is good to know how much total data to be processed.
That's why we used two ways: the DirEntry
count to forecast a processing time very quickly and the size to improve the predicted time.
- Python 3.6 or newer.
tqdm
for process barspsutils
for setting process priority- For dev.: Pipenv. Packages and virtual environment manager
Please: try, fork and contribute! ;)
Use example CLI, e.g.:
~$ git clone https://github.com/jedie/IterFilesystem.git ~$ cd IterFilesystem ~/IterFilesystem$ pipenv install ~/IterFilesystem$ pipenv shell (IterFilesystem) ~/IterFilesystem$ print_fs_stats --help (IterFilesystem) ~/IterFilesystem$ pip install -e . ... Successfully installed iterfilesystem ~/IterFilesystem$ $ poetry run print_fs_stats --help usage: print_fs_stats.py [-h] [-v] [--debug] [--path PATH] [--skip_dir_patterns [SKIP_DIR_PATTERNS [SKIP_DIR_PATTERNS ...]]] [--skip_file_patterns [SKIP_FILE_PATTERNS [SKIP_FILE_PATTERNS ...]]] Scan filesystem and print some information optional arguments: -h, --help show this help message and exit -v, --version show program's version number and exit --debug enable DEBUG --path PATH The file path that should be scanned e.g.: "~/foobar/" default is "~" --skip_dir_patterns [SKIP_DIR_PATTERNS [SKIP_DIR_PATTERNS ...]] Directory names to exclude from scan. --skip_file_patterns [SKIP_FILE_PATTERNS [SKIP_FILE_PATTERNS ...]] File names to ignore.
example output looks like this:
(IterFilesystem) ~/IterFilesystem$ $ print_fs_stats --path ~/IterFilesystem --skip_dir_patterns ".*" "*.egg-info" --skip_file_patterns ".*" Read/process: '~/IterFilesystem'... Skip directory patterns: * .* * *.egg-info Skip file patterns: * .* Filesystem items..:Read/process: '~/IterFilesystem'... ... Filesystem items..: 100%|█████████████████████████████████████████|135/135 13737.14entries/s [00:00<00:00, 13737.14entries/s] File sizes........: 100%|██████████████████████████████████████████████████████████████|843k/843k [00:00<00:00, 88.5MBytes/s] Average progress..: 100%|███████████████████████████████████████████████████████████████████████████████████████|00:00<00:00 Current File......:, /home/jens/repos/IterFilesystem/Pipfile Processed 135 filesystem items in 0.02 sec SHA515 hash calculated over all file content: 10f9475b21977f5aea1d4657a0e09ad153a594ab30abc2383bf107dbc60c430928596e368ebefab3e78ede61dcc101cb638a845348fe908786cb8754393439ef File count: 109 Total file size: 843.5 KB 6 directories skipped. 6 files skipped.
**dev** - compare v1.4.3...master
- TBC
-
- Use logging and remove "verbose mode"
- Nicer "Average progess" bar
- Bugfix "Current File" bar: remove comma
-
- iterate over sorted dir entries
- update CI pipelines
-
- Bugfix
human_filesize
- Bugfix
-
stats_helper.abort
exists always usefull to get information if KeyboardInterrupt was used- use poetry and modernize project setup
-
- Bugfix if scan directory is completely empty
-
- Set ionice and nice priority via psutils
-
- Refactor API
- cleanup statistics and process bar
- handle access errors like: Permission denied
- fix tests
-
- don't create separate process for worker: Just do the work in main process
- dir/file filter uses now
fnmatch
-
refactoring:
- don't use
persist-queue
- switch from threading to multiprocessing
- enhance progress display with multiple
tqdm
process bars
- don't use
-
- store persist queue in temp directory
- Don't catch
process_path_item
errors, this should be made in child class
-
- add some project meta files and tests
- setup CI
- fix tests
-
- first Release on PyPi
Note: this file is generated from README.creole 2020-03-16 18:09:30 with "python-creole"