Custom file and directory checksum and verification tool.

• multiple hash algos: c4, crc32, md5, sha256, sha512, xxh64
• supports multiple output options: json, txt and mhl
• recursively runs checksums on files in directory trees
• ignores predefined file name patterns
• collects important file stat metadata
• caches results for better performance

The easiest way to install:

$ pip install -U hashio

Note: Starting with hashio 0.4.0, a SQLite-backed cache is used by default. If your Python installation does not include sqlite3, either:

• Rebuild Python with SQLite support (e.g. libsqlite3-dev, sqlite-devel)
• Or install an earlier version:

pip install 'hashio<0.4.0'

Recursively checksum and gather metadata all the files in a dir tree, and output results to a hash.json file:

$ hashio <PATH> -o hash.json [--algo ALGO]

hashio supports .json, .txt or .mhl output formats:

$ hashio <PATH> -o hash.txt

If no output file is specified, the hash results are stored in a cache, defined by ${HASHIO_DB}.

You can run hashio instantly using uvx:

$ uvx hashio <PATH>

This downloads and runs hashio in a temporary, isolated environment — no installs, no virtualenvs, no cleanup needed. Perfect for quick hash verification tasks on large directories.

Note that files matching patterns defined in config.IGNORABLE will be skipped, unless using the --force option:

$ hashio .git
path is ignorable: .git
$ hashio .git --force
hashing: 894kB [00:02, 758kB/s, files/s=412.80]

Verify paths in previously generated hash file by comparing stored mtimes (if available) or regenerated hash values if mtimes are missing or different:

$ hashio --verify hash.json

To make a portable hash file, use -or to make the paths relative to the hash.json file :

$ hashio <DIR> -or hash.json

or use --start to make them relative to the <START> value

$ hashio <DIR> -o hash.json --start <START>

To verify the data in the hash file, run hashio from the parent dir of the data, or set --start to the parent dir:

$ hashio --verify hash.json

The following environment variables are supported, and default settings are in the config.py module.

Optionally, modify the hashio.env file if using envstack, or create a new env file:

$ cp hashio.env debug.env
$ vi debug.env  # make edits
$ ./debug.env -- hashio

By default hashio collects the following file metadata:

To walk a directory and collect metadata without peforming file checksums, use the "null" hash algo:

$ hashio <DIR> -a null

To make "null" the default, update ${HASHIO_ALGO} in the environment or the hashio.env file.

$ export HASHIO_ALGO=null

hashio maintains a local SQLite cache file (by default at ~/.cache/hashio/hash.sql) to store previously computed file hashes, metadata, and snapshot history. This dramatically speeds up repeated runs and enables powerful diffing capabilities.

Snapshots are point-in-time views. You can optionally record a snapshot of the current file state using:

$ hashio --snapshot SNAPSHOT_NAME

This links all scanned files to a snapshot named SNAPSHOT_NAME, allowing you to:

• Track changes over time
• Compare file states across points in time
• Build file history for audit/debugging
• Generate change reports (diffs)

Each snapshot is stored in the cache and contains only links to file metadata entries, no file duplication.

You can compare snapshots using:

$ hashio --diff SNAP1 SNAP2 [--start PATH]

This prints a summary of file-level changes between two snapshots:

+ file was added
- file was removed
~ file was modified

By default, hashio uses a fixed read buffer size set in config.py. This conservative default works well across most systems, but it can be suboptimal on filesystems with large block sizes (e.g. network-attached storage).

Dynamic read buffer sizing can be enabled to reduce IOPS and improve performance on sequential reads. When enabled:

• Filesystem block size is determined via os.statvfs(path).f_frsize.
• If the block size is large (≥ 128 KiB), it is used directly
• If the block size is small, it is scaled up and clamped

You can configure buffer size using the BUF_SIZE environment variable:

• If BUF_SIZE is set to a positive integer, that value is used
• If BUF_SIZE is 0 or a negative value, dynamic buffer sizing is enabled
• If BUF_SIZE is unset, the default fixed size from config.py is used

This can be configured either:

• In your hashio.env file (if using envstack), or
• Directly in the environment

# use a fixed 512 KiB read buffer
export BUF_SIZE=524288

# enable dynamic sizing based on block size
export BUF_SIZE=0

Generate a hash.json file for a given path (Default is the current working directory):

from hashio.worker import HashWorker
worker = HashWorker(path, outfile="hash.json")
worker.run()

Verify pre-generated checksums stored in a hash.json file:

from hashio.encoder import verify_checksums
for algo, value, miss in verify_checksums("hash.json"):
    print(f"{algo} {miss}")

Generate a checksum of a folder:

from hashio.encoder import checksum_folder, XXH64Encoder
encoder = XXH64Encoder()
value = checksum_folder(folder, encoder)

Run the following command to safely apply the latest schema updates and create any missing indexes:

$ hashio --update-cache