Object Traversal & Inspection Toolkit
OTIT is a small, typed, dependency-free Python library for navigating, inspecting, and modifying heterogeneous nested Python objects.
It provides a consistent path-based API for working with mappings, sequences, object attributes, and structures containing a mixture of all three.
- API payload inspection - safely access deeply nested values without chains of dict/list indexing.
- Web scraping and extraction pipelines - inspect and extract values from nested structures produced by scrapers, parsers, or API responses.
- Configuration processing - inspect, test, modify, pick, or omit nested configuration values.
- Testing and assertions - query arbitrary nested structures with consistent path semantics.
- Data inspection/tooling - walk heterogeneous combinations of mappings, sequences, and Python objects.
- Payload shaping - use pick() and omit() to derive structures without modifying the original.
- Generic application utilities - work with nested structures when the exact mixture of dictionaries, lists, and objects isn't known in advance.
OTIT deliberately keeps path traversal small and predictable. It is not intended to be:
- A query language - paths address concrete values; OTIT does not implement wildcards, filters, recursive selectors, or expression syntax.
- A serialization or schema library - OTIT does not validate schemas, coerce types, or serialize arbitrary objects.
- A web scraper or HTML parser - OTIT can inspect structures produced by scrapers and parsers, but does not fetch or parse web pages itself.
- An object-mapping framework - structural operations such as pick() focus on data shape rather than reconstructing arbitrary Python class instances.
- A full reflection framework - OTIT focuses on practical traversal of mappings, sequences, and ordinary object attributes.
The goal is straightforward path-based traversal and inspection without turning paths into a separate query language.
pip install otitRequires Python 3.10 or later.
import otit
data = {
"users": [
{
"name": "Matti",
"address": {
"city": "Oulu",
},
}
]
}
otit.get(data, "users.0.name")
# "Matti"
otit.get(data, "users.0.address.city")
# "Oulu"
otit.has(data, "users.0.email")
# FalsePaths can also be given explicitly as sequences:
otit.get(data, ("users", 0, "name"))
# "Matti"Explicit paths are useful when a mapping key contains a dot or when preserving the exact type of a path segment matters.
data = {"foo.bar": "value"}
otit.get(data, ("foo.bar",))
# "value"Use get() to resolve a path:
otit.get(data, "users.0.name")
# "Matti"A default can be returned when the path does not exist:
otit.get(data, "users.0.email", default=None)
# NoneWithout a default, an unresolved path raises otit.PathNotFound.
Use has() to test whether a path can be resolved:
otit.has(data, "users.0.name")
# True
otit.has(data, "users.0.email")
# FalseUse set() to modify an existing value:
otit.set(data, "users.0.name", "Liisa")By default, the complete path must already exist.
Set create=True to allow creation of the final mapping key or object attribute:
otit.set(
data,
"users.0.email",
"liisa@example.com",
create=True,
)create=True does not create missing parent containers or extend sequences.
Use delete() to remove a value:
otit.delete(data, "users.0.address.city")For mappings, the key is removed. For mutable sequences, the item is removed and subsequent indexes shift. For objects, the attribute is deleted.
walk() recursively yields every reachable child together with its path:
data = {
"user": {
"name": "Matti",
"tags": ["admin", "active"],
}
}
list(otit.walk(data))produces:
[
(("user",), {"name": "Matti", "tags": ["admin", "active"]}),
(("user", "name"), "Matti"),
(("user", "tags"), ["admin", "active"]),
(("user", "tags", 0), "admin"),
(("user", "tags", 1), "active"),
]Traversal paths use tuples. Sequence indexes are represented as integers.
The root object itself is not yielded.
Use find() to select reachable values with a predicate:
list(
otit.find(
data,
lambda value: isinstance(value, str) and value.startswith("M"),
)
)The result contains (path, value) pairs.
Use paths() to iterate over every reachable path:
list(otit.paths(data))For example:
[
("user",),
("user", "name"),
("user", "tags"),
("user", "tags", 0),
("user", "tags", 1),
]Use leaves() to iterate over terminal values:
list(otit.leaves(data))For example:
[
(("user", "name"), "Matti"),
(("user", "tags", 0), "admin"),
(("user", "tags", 1), "active"),
]Empty mappings and sequences are not considered leaves.
pick() creates a new structure containing only selected paths:
data = {
"user": {
"name": "Matti",
"email": "matti@example.com",
"password": "secret",
}
}
otit.pick(
data,
"user.name",
"user.email",
)returns:
{
"user": {
"name": "Matti",
"email": "matti@example.com",
}
}Selected sequence items are compacted while preserving their original order.
The original object is not modified.
omit() creates a copy with selected paths removed:
otit.omit(
data,
"user.password",
)returns:
{
"user": {
"name": "Matti",
"email": "matti@example.com",
}
}All omitted paths refer to the original object. This means removing sequence items does not change the meaning of other paths passed in the same call.
The original object is not modified.
A path may be a dot-separated string:
"users.0.address.city"or an explicit sequence:
("users", 0, "address", "city")OTIT resolves each segment according to the object currently being traversed:
- Mappings use the segment as a mapping key.
- Sequences interpret the segment as an integer index.
- Other objects use string segments as attribute names.
- Strings, bytes, and bytearrays are treated as terminal values rather than traversable sequences.
- Negative sequence indexes are supported.
- Numeric-looking mapping keys remain mapping keys.
For example:
data = {
"lookup": {
"0": "zero",
},
"items": [
"first",
],
}
otit.get(data, "lookup.0")
# "zero"
otit.get(data, "items.0")
# "first"The meaning of "0" depends on the object being traversed.
The empty string and empty tuple represent the root object:
otit.get(data, "") is data
# True
otit.has(data, ())
# TrueOperations that require a target below the root, such as set(), delete(), pick(), and omit(), reject a root path with otit.InvalidPath.
OTIT exposes a small exception hierarchy:
OtitError
├── PathNotFound
└── InvalidPath
PathNotFound is raised when OTIT cannot resolve a requested path.
InvalidPath is raised when a path is structurally invalid for the requested operation.
Exceptions raised by user-defined properties or other object behavior are not converted into PathNotFound.
Traversal functions detect cycles and do not recursively follow the same object through an active ancestor path.
Shared objects reachable through different paths are still traversed independently.
OTIT is designed for heterogeneous structures containing:
- mappings such as
dict - sequences such as
listandtuple - regular Python objects
- dataclass instances
- mixtures of the above
Mutation requires the underlying object to support the requested operation.
otit.get(obj, path, *, default=...)
otit.has(obj, path)
otit.set(obj, path, value, *, create=False)
otit.delete(obj, path)
otit.walk(obj)
otit.find(obj, predicate)
otit.paths(obj)
otit.leaves(obj)
otit.pick(obj, *paths)
otit.omit(obj, *paths)Install development dependencies:
uv syncRun the test suite:
uv run pytestRun linting:
uv run ruff check .Run type checking:
uv run mypy srcBuild the package:
uv buildOTIT is licensed under the GNU General Public License v3.0 or later (GPL-3.0-or-later).
See the LICENSE file for the full license text.