glom

25.12.0 · active · verified Mon Apr 06

glom is a declarative object transformer and formatter for Python, designed to simplify working with nested data structures. It provides path-based access, powerful data transformation using Pythonic specifications, meaningful error messages, and built-in debugging features. The library follows a CalVer versioning scheme (YY.MM.MICRO) and is actively maintained, with version 25.12.0 being a recent release that includes maintenance updates and test fixes for newer Python versions.

Warnings

breaking The `*` and `**` wildcard behavior for path access became enabled by default in glom 23.1.0. This can break existing specs that used these characters as literal keys.
Fix: If you experience issues, you can temporarily revert the behavior by setting `glom.core.PATH_STAR = False`. The recommended fix is to explicitly use the `Path` specifier, e.g., `Path('a', '*')` instead of `'a.*'`.
deprecated The `Call` specifier is deprecated in favor of the more comprehensive `Invoke` specifier, introduced in glom 19.10.0. `Invoke` offers a richer API for invoking callables within your specs.
Fix: Replace `Call` with `Invoke` in your specs. For calling methods on the target, consider using the `T` object instead.
gotcha Not all internal components of the `glom` package are part of its public API. Functionality not explicitly documented in the top-level `glom` package or its official documentation may change or disappear without notice.
Fix: Rely only on documented `glom` functions, classes, and specifiers. If you find undocumented functionality that seems useful, consider filing an issue or pull request.
gotcha If `glom()` is called without `default` and `skip_exc` parameters, it will raise exceptions on path access failures (e.g., `PathAccessError` for missing keys/attributes).
Fix: Use the `Coalesce` specifier or the `default` and `skip_exc` kwargs in the `glom()` function to gracefully handle missing data or potential errors. `Coalesce` catches exceptions internally and proceeds to the next spec.
breaking Official support for Python 2 and Python 3.6 was dropped in glom 23.3.0, and pre-Python 3.7 syntax was removed in 24.11.0.
Fix: Ensure your project runs on Python 3.7 or newer to use glom versions 23.3.0 and above.

Install

pip install glom Install glom

Imports

glom
```
from glom import glom
```
T
```
from glom import T
```
T is a specifier for object-oriented access and method calls.
Coalesce
```
from glom import Coalesce
```
Coalesce provides default values for potentially missing or error-raising paths, similar to SQL's COALESCE.
Invoke
```
from glom import Invoke
```
The 'Call' specifier is deprecated in favor of 'Invoke' since glom 19.10.0, which offers a fuller-featured API for invoking callables.
Path
```
from glom import Path
```
Explicitly using Path can avoid ambiguities, especially with wildcard characters.
S
```
from glom import S
```
S provides access to the glom scope for additional state management during transformations.

Quickstart

This quickstart demonstrates basic deep access to nested data, transformation into a new dictionary structure, and using the `T` specifier for iterating and extracting specific fields from a list of dictionaries.

from glom import glom, T

data = {
    'user': {
        'profile': {
            'name': 'Alice',
            'email': 'alice@example.com'
        },
        'preferences': {
            'theme': 'dark'
        }
    },
    'messages': [
        {'id': 1, 'text': 'Hello'},
        {'id': 2, 'text': 'World'}
    ]
}

# Extract a deeply nested value
user_name = glom(data, 'user.profile.name')
print(f"User Name: {user_name}")

# Transform data to a new structure
user_summary_spec = {
    'name': 'user.profile.name',
    'contact_email': 'user.profile.email',
    'first_message': ('messages.0.text', str.upper)
}

user_summary = glom(data, user_summary_spec)
print(f"User Summary: {user_summary}")

# Using T for more complex access or method calls
message_ids = glom(data, ('messages', [T['id']]))
print(f"Message IDs: {message_ids}")

view raw JSON →