dycomutils 0.3.2

Add your description here

dycomutils

dycomutils is a small Python utility package for common day-to-day helpers that show up across scripts and projects.

The library currently focuses on:

• serialization helpers for JSON, JSONL, pickle, and text files
• lightweight concurrency utilities for running named jobs in parallel
• recursive config creation with attribute-style access
• simple graph grouping helpers for connected components
• keystroke automation utilities for typing file contents

This package is a practical toolbox: small functions, minimal ceremony, and utilities that are easy to reuse in notebooks, scripts, and internal tools.

Installation

Install the package locally:

pip install .

If you want to use the automation helpers or progress-aware concurrency helpers, you may also need:

pip install pynput tqdm

Package Layout

dycomutils/
  serialization/
  concurrancy/
  config/
  grouping/
  automation/

Top-level imports:

import dycomutils

dycomutils.serialization
dycomutils.concurrancy
dycomutils.config
dycomutils.grouping

Functionality

Serialization

The dycomutils/serialization/__init__.py module provides compact file I/O helpers for common formats:

• save_json(data, loc)
• load_json(loc)
• save_pickle(obj, loc)
• load_pickle(loc)
• save_text(s, loc)
• load_text(loc)
• save_jsonl(data, loc)
• save_jsonl_append(loc, data)
• load_jsonl(loc)
• load_jsonl_generator(loc)
• file_exist(*args)

Example:

from dycomutils.serialization import (
    save_json,
    load_json,
    save_jsonl,
    save_jsonl_append,
    load_jsonl_generator,
)

payload = {"name": "example", "version": 1}
save_json(payload, "config.json")

loaded = load_json("config.json")

records = [{"id": 1}, {"id": 2}]
save_jsonl(records, "data.jsonl")
save_jsonl_append("data.jsonl", {"id": 3})

for row in load_jsonl_generator("data.jsonl"):
    print(row)

Good fit for:

• experiment metadata
• cached intermediate results
• simple local persistence
• appending one JSON object at a time into a JSONL log
• streaming line-by-line JSONL reads

Concurrency

The dycomutils/concurrancy/__init__.py module includes concurrent_dict_execution, a helper for running a function across a dictionary of named jobs using threads or processes.

It supports:

• thread or process execution
• positional argument jobs
• keyword argument jobs
• progress reporting with tqdm
• yielding results together with the original job key

Example with positional arguments:

from dycomutils.concurrancy import concurrent_dict_execution

def add(a, b):
    return a + b

jobs = {
    "first": [1, 2],
    "second": [10, 5],
}

for name, result in concurrent_dict_execution(add, jobs, executor="thread", num_max_workers=2):
    print(name, result)

Example with keyword arguments:

jobs = {
    "job_a": {"a": 3, "b": 4},
    "job_b": {"a": 7, "b": 8},
}

This is especially useful when you want a lightweight alternative to writing your own ThreadPoolExecutor boilerplate each time.

Config Creation

The dycomutils/config/dict_to_config.py module provides ConfigDict, a dict subclass that turns nested dictionaries into attribute-accessible config objects.

Features:

• access values with config.key
• still behaves like a normal dictionary
• recursively converts nested dictionaries
• preserves nested lists and tuples while converting inner dictionaries

Example:

from dycomutils.config import ConfigDict

cfg = ConfigDict({
    "model": {
        "name": "baseline",
        "hidden_size": 256,
    },
    "train": {
        "batch_size": 32,
    },
})

print(cfg.model.name)
print(cfg["train"].batch_size)

cfg.train.epochs = 5

This is handy for:

• lightweight configuration objects
• experiment settings
• script parameters
• nested dictionaries that are easier to read with dot access

Grouping / Connected Components

The dycomutils/grouping/__init__.py module contains a small union-find implementation:

• findParent(parent, x)
• unionSets(parent, x, y)
• getComponents(V, edges)

getComponents(V, edges) returns connected components for an undirected graph represented by:

• V: number of vertices
• edges: pairs of connected node indices

Example:

from dycomutils.grouping import getComponents

components = getComponents(
    6,
    [
        (0, 1),
        (1, 2),
        (3, 4),
    ],
)

print(components)
# Example output: [[0, 1, 2], [3, 4], [5]]

This is useful for basic clustering, graph partitioning, and grouping related indices.

Notes

• The concurrency module is exposed as concurrancy, matching the current package name in the code.
• The concurrency helper uses tqdm for progress display.
• The top-level package exports serialization, concurrancy, grouping, and config.

Positioning

dycomutils is best described as a personal general-purpose Python utility package rather than a framework. It packages together reusable helpers for:

• persistence and serialization
• parallel job execution
• config ergonomics
• lightweight graph grouping
• small automation tasks

If you have lots of one-off scripts, research code, data workflows, or internal tooling, this package gives you a single place for the small utilities you end up rewriting over and over.