odis-python

Python bindings for the odis Formal Concept Analysis library, powered by Rust and PyO3.

Background

Formal Concept Analysis (FCA) works on formal contexts — cross-tables pairing objects with attributes via a binary incidence relation — and derives the complete lattice of formal concepts from them. odis implements the core FCA algorithms in Rust and exposes them through this Python interface. For an introduction to FCA see Uta Priss's FCA page.

Installation

Released package (PyPI)

pip install odis-python

Development build (from source)

Requires a Rust toolchain and maturin.

git clone https://github.com/odis-rs/odis
cd odis/odis-python
pip install maturin
maturin develop --release

Quick Start

from odis import FormalContext

ctx = FormalContext.from_file("odis/test_data/living_beings_and_water.cxt")
print(f"Objects: {ctx.objects}")
print(f"Attributes: {ctx.attributes}")
concepts = list(ctx.concepts())
print(f"Number of concepts: {len(concepts)}")

FormalContext

FormalContext stores a set of objects, a set of attributes, and a binary incidence relation mapping object–attribute pairs.

Construction

from odis import FormalContext

# Empty context
ctx = FormalContext()

# From a .cxt (Burmeister) file
ctx = FormalContext.from_file("odis/test_data/living_beings_and_water.cxt")

# From a dict mapping each object to its set of attributes
animals = FormalContext.from_dict({
    "cat":  {"has_legs", "has_fur", "can_move"},
    "fish": {"lives_in_water", "can_move"},
    "fern": {"needs_chlorophyll"},
})

The examples below all use ctx as loaded from living_beings_and_water.cxt, whose objects are fish leech, bream, frog, dog, water weeds, reed, bean and corn. Note that fish leech is one object — the animal — not two.

Introspection

n_objects, n_attributes = ctx.shape   # e.g. (8, 9)
n = len(ctx)                          # same as ctx.shape[0] — number of objects
print(ctx.objects)                    # ['fish leech', 'bream', 'frog', ...]
print(ctx.attributes)                 # ['needs water to live', ...]
print("frog" in ctx)                  # True — tests object membership
print(repr(ctx))                      # human-readable summary

Incidence Access

# Read: does object have attribute?
val = ctx["frog", "lives in water"]   # True
val = ctx["frog", "breast feeds"]     # False

# Write
ctx["frog", "lives in water"] = False
ctx["frog", "lives in water"] = True

Mutation

# Add an object with no attributes
ctx.add_object("shark")

# Add an object with some pre-set attributes. Re-using an existing name
# raises ValueError.
ctx.add_object("whale", {"needs water to live", "can move", "breast feeds"})

# Add a new attribute column
ctx.add_attribute("is_endangered")

# Remove
ctx.remove_object("shark")
ctx.remove_object("whale")
ctx.remove_attribute("is_endangered")

# Rename
ctx.rename_object("frog", "toad")
ctx.rename_attribute("needs water to live", "aquatic")

Serialisation

# Save to .cxt file
ctx.to_file("/tmp/my_context.cxt")

# Deep copy — mutations to the copy do not affect the original
copy = ctx.copy()
copy.add_object("clone_only")
assert "clone_only" not in ctx.objects

Derivation Operators

# Extent: the set of all objects sharing every given attribute
extent = ctx.extent(["needs water to live", "can move"])

# Intent: the set of all attributes shared by every given object.
# Names that are not in the context are dropped silently, so a typo here
# quietly computes the intent of a smaller set instead of raising.
intent = ctx.intent(["fish leech", "bream"])

# Attribute hull (closure of an attribute set under the Galois connection)
hull = ctx.attribute_hull(["needs water to live"])

# Object hull (closure of an object set)
ohull = ctx.object_hull(["frog"])

# Upper neighbor: the extent of the concept directly above the given concept
# in the lattice (the least concept with a strictly larger extent)
neighbor = ctx.upper_neighbor(["frog"])

# All results are LabelSets — iterate or convert freely
print(list(extent))        # ['bream', 'dog', 'fish leech', 'frog']
print("frog" in extent)    # True or False

Drawing Shortcut

FormalContext provides convenience methods to draw the concept lattice without instantiating a Drawing object; see Drawing for the full API.

svg_str = ctx.draw_svg("dimdraw", width=800, height=600)
drawing  = ctx.draw("dimdraw")

Concepts

FormalContext.concepts() returns a ConceptCollection (eager, indexable) or a ConceptGenerator (lazy, forward-only). Each element is a Concept with .extent and .intent properties.

# Eager (default) — all concepts materialised at once
concepts = ctx.concepts()
print(f"Found {len(concepts)} concepts")

# Access by index
first = concepts[0]
print(list(first.extent))   # objects in this concept
print(list(first.intent))   # attributes in this concept

# Iteration with unpacking
for extent, intent in concepts:
    print(list(extent), "→", list(intent))

Lazy concepts are covered under Lazy Generators & Mutation Guard.

Implications

The canonical implication basis (Duquenne–Guigues basis) is the smallest set of implications that logically entails all implications valid in the context.

basis = ctx.canonical_basis()
print(f"Basis size: {len(basis)}")

for impl in basis:
    print(list(impl.premise), "→", list(impl.conclusion))

# Access by index
imp = basis[0]
print(list(imp.premise))     # antecedent attributes
print(list(imp.conclusion))  # consequent attributes

# Optimised variant (same result, faster in practice)
basis_opt = ctx.canonical_basis_optimised()

Iterating pseudo-intents one at a time with next_preclosure:

# next_preclosure(basis, current) returns the next closed attribute set in
# lectic order. Terminates naturally when len(result) == number of attributes.
n_attrs = len(ctx.attributes)
current = frozenset()
while len(current) < n_attrs:
    nxt = ctx.next_preclosure(basis, current)
    if len(nxt) == n_attrs:
        break
    print(list(nxt))
    current = nxt

Attribute Exploration

Attribute exploration is an interactive algorithm that discovers the canonical basis by consulting an oracle (a Python callback) about whether proposed implications hold. The oracle may reject an implication by supplying a counterexample.

def my_oracle(premise, conclusion):
    """Called for each proposed implication.

    premise and conclusion are LabelSets (iterable over strings).
    Return True to accept; return (name, attrs) to reject with a counterexample.
    """
    print(f"Does: {list(premise)} → {list(conclusion)}?")
    return True  # accept everything: the result is the canonical basis

basis = ctx.attribute_exploration(my_oracle)
print(f"Discovered {len(basis)} implications")

The callback receives two LabelSet arguments — premise and conclusion:

Return any truthy non-tuple value (e.g. True) to accept the implication.
Return (name: str, attributes: Iterable[str]) to reject it with a counterexample.

When a counterexample is provided, attribute_exploration adds that object (with the given attributes) to the context and continues.

A counterexample has to be one: the object must have every attribute of the premise and miss at least one of the conclusion. An object that does not refute the implication leaves it valid, so it is proposed again — and an oracle that keeps answering with the same object never terminates. Deriving the counterexample from the premise itself is always safe:

rejected = 0

def counterexample_oracle(premise, conclusion):
    global rejected
    if "can move" in list(premise):
        rejected += 1
        # Has exactly the premise, so it misses the conclusion by construction.
        return (f"counterexample_{rejected}", set(premise))
    return True

ctx_copy = ctx.copy()
basis = ctx_copy.attribute_exploration(counterexample_oracle)
print(f"{rejected} rejected, {len(basis)} implications, {len(ctx_copy.objects)} objects")

Drawing

Poset lets you directly define a partial order. Edges describe the covering relation: (u, v) means node u is directly below node v (u ≺ v), given as 0-based indices into the node list. Cycles are rejected with ValueError.

from odis import Poset

# Diamond lattice
p = Poset(
    ["bottom", "left", "right", "top"],
    [(0, 1), (0, 2), (1, 3), (2, 3)],
)

# Quick SVG
svg = p.draw_svg("dimdraw", width=800, height=600)
with open("order.svg", "w") as f:
    f.write(svg)

# Drawing object for programmatic access
drawing = p.draw("dimdraw")
if drawing is not None:
    for node in drawing.nodes:
        print(f"{node.object_labels[0]}: ({node.x:.1f}, {node.y:.1f})")
    print(drawing.edges)   # list of (u, v) covering-relation pairs

Concept Lattice Drawings

odis can draw the concept lattice as a directed graph. Three layout algorithms are available: "dimdraw" (dimension-based, default), "sugiyama" (hierarchical) and "dimflux" (a DimDraw layout refined into an additive one by a force-directed model, which spreads the nodes away from the edges they are not part of). "dimflux" needs the objects and attributes of a concept, so it is available on a context but not on a bare Poset.

timeout_ms bounds the layout search and defaults to one second. Pass timeout_ms=None to search until the layout is a proven optimum — the cost of that proof climbs steeply with the size of the lattice, so it is opt-in.

# Quick SVG string — no intermediate Drawing object required
svg = ctx.draw_svg("dimdraw", width=800, height=600)
with open("lattice.svg", "w") as f:
    f.write(svg)

# Full Drawing object for programmatic access
drawing = ctx.draw("dimdraw")
if drawing is not None:
    print(f"Nodes: {len(drawing.nodes)}")
    print(f"Edges: {drawing.edges}")              # list of (from_idx, to_idx) tuples
    print(f"Coordinates: {drawing.coordinates}")  # raw layout (x, y) per node

    for node in drawing.nodes:
        print(f"  node {node.index}: ({node.x:.1f}, {node.y:.1f})")
        print(f"    reduced objects:    {node.object_labels}")
        print(f"    reduced attributes: {node.attribute_labels}")

    # Convert to SVG from Drawing object (useful for custom sizes)
    svg2 = drawing.to_svg(ctx, width=1200, height=800)
    with open("large_lattice.svg", "w") as f:
        f.write(svg2)

# Jupyter notebook: display inline (requires IPython)
try:
    from IPython.display import SVG, display
    display(SVG(data=svg))
except ImportError:
    pass  # not running in a notebook

DimFlux

"dimflux" starts from the DimDraw layout and projects it into the space of additive diagrams, where every concept sits at the sum of one vector per object in its extent and per attribute in its intent. A force-directed model then spreads the nodes away from the edges they are not part of, without letting them leave the cells DimDraw put them in.

Two properties follow, and both help a reader: equal steps through the lattice are drawn as equal vectors — so a distributive part of the lattice comes out as a grid of parallelograms — and concept nodes keep their distance from unrelated edges.

ctx = FormalContext.from_file("odis/test_data/living_beings_and_water.cxt")

# The same lattice under each of the three layouts
for algorithm in ("dimdraw", "sugiyama", "dimflux"):
    svg = ctx.draw_svg(algorithm, width=800, height=600)
    with open(f"lattice_{algorithm}.svg", "w") as f:
        f.write(svg)

# The additive structure shows up in the coordinates: two covering edges that
# add the same objects and drop the same attributes come out as the same vector.
from collections import defaultdict

drawing = ctx.draw("dimflux")
xy = drawing.coordinates
families = defaultdict(list)
for lower, upper in drawing.edges:
    step = (round(xy[upper][0] - xy[lower][0], 6),
            round(xy[upper][1] - xy[lower][1], 6))
    families[step].append((lower, upper))

parallel = {step: edges for step, edges in families.items() if len(edges) > 1}
print(f"{len(drawing.edges)} edges drawn as {len(families)} distinct vectors")
print(f"{len(parallel)} of those vectors are shared by more than one edge")

"dimflux" spends its search budget on the DimDraw layout it starts from, so timeout_ms trades quality against time just as it does for "dimdraw":

quick = ctx.draw("dimflux", timeout_ms=100)     # good enough to look at
better = ctx.draw("dimflux", timeout_ms=5000)   # a longer search for the base layout

Because it needs the objects and attributes behind each node, "dimflux" is available on a FormalContext but not on a bare Poset, which carries only the order.

Titanic

The Titanic algorithm enumerates iceberg concepts — concepts whose extent meets a minimum support threshold. Useful for large or sparse contexts where only frequent concepts are of interest.

from odis import FormalContext, Titanic

ctx = FormalContext.from_dict({
    "a": {"x", "y", "z"},
    "b": {"x", "y"},
    "c": {"x", "z"},
    "d": {"y", "z"},
    "e": {"x"},
})

iceberg = Titanic()

# Only enumerate concepts with at least 2 objects in their extent
top_concepts = iceberg.enumerate(ctx, min_support=2)
print(f"Iceberg concepts (support ≥ 2): {len(top_concepts)}")
for c in top_concepts:
    print(f"  extent={list(c.extent)}, intent={list(c.intent)}")

LabelSet

LabelSet is a set-like view of string labels. It is returned by derivation operators (extent, intent, attribute_hull, object_hull, upper_neighbor), implication properties (premise, conclusion), and concept properties (.extent, .intent).

intent = ctx.intent(["fish", "leech"])

# Membership test
print("can move" in intent)   # True

# Iteration — yields strings directly, no index translation needed
for attr in intent:
    print(attr)

# Convert to standard Python containers
as_list = list(intent)
as_set  = set(intent)

Lazy Generators & Mutation Guard

Passing lazy=True to concepts() or canonical_basis() returns a generator that produces one concept/implication at a time without materialising the full collection. Lazy generators hold a shared reference to the context's internal state, so any mutation while a lazy generator is alive raises RuntimeError.

ctx = FormalContext.from_file("odis/test_data/living_beings_and_water.cxt")

# Create a lazy generator
gen = ctx.concepts(lazy=True)

# Iterating is safe
first = next(gen)
print(list(first.extent))

# Mutating while the generator is alive raises RuntimeError
try:
    ctx.add_attribute("new_attr")       # raises RuntimeError
except RuntimeError as e:
    print(f"Caught: {e}")

# Release the generator first, then mutate freely
del gen
ctx.add_attribute("new_attr")          # OK

The same guard applies to canonical_basis(lazy=True) and Titanic().enumerate(ctx, ..., lazy=True).

Error Reference

Exception	When raised	Example trigger
`FileNotFoundError`	`.cxt` file path does not exist	`FormalContext.from_file("missing.cxt")`
`OSError`	Other I/O error reading a file	Unreadable file permissions
`ValueError`	Malformed `.cxt` file	Invalid Burmeister format
`KeyError`	Unknown object or attribute name	`ctx["ghost", "flies"]`
`ValueError`	Duplicate object or attribute name	`ctx.add_object("frog")` when already present
`RuntimeError`	Mutation while a lazy generator is alive	`ctx.add_attribute("x")` during active generator
`ValueError`	Unknown drawing algorithm	`ctx.draw("unknown_algo")`
`ValueError`	Non-positive SVG dimensions	`ctx.draw_svg("dimdraw", -1, 600)`

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

odis_python-2026.9.0.tar.gz (55.6 kB view details)

Uploaded Sep 5, 2026 Source

Built Distributions

If you're not sure about the file name format, learn more about wheel file names.

The dropdown lists show the available interpreters, ABIs, and platforms. Enable javascript to be able to filter the list of wheel files.

odis_python-2026.9.0-cp314-cp314t-win_arm64.whl (424.3 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tWindows ARM64

odis_python-2026.9.0-cp314-cp314t-win_amd64.whl (451.7 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tWindows x86-64

odis_python-2026.9.0-cp314-cp314t-win32.whl (434.5 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tWindows x86

odis_python-2026.9.0-cp314-cp314t-musllinux_1_2_x86_64.whl (768.1 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmusllinux: musl 1.2+ x86-64

odis_python-2026.9.0-cp314-cp314t-musllinux_1_2_i686.whl (809.5 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmusllinux: musl 1.2+ i686

odis_python-2026.9.0-cp314-cp314t-musllinux_1_2_armv7l.whl (849.3 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmusllinux: musl 1.2+ ARMv7l

odis_python-2026.9.0-cp314-cp314t-musllinux_1_2_aarch64.whl (719.7 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmusllinux: musl 1.2+ ARM64

odis_python-2026.9.0-cp314-cp314t-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (557.6 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmanylinux: glibc 2.17+ x86-64

odis_python-2026.9.0-cp314-cp314t-manylinux_2_17_s390x.manylinux2014_s390x.whl (633.2 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmanylinux: glibc 2.17+ s390x

odis_python-2026.9.0-cp314-cp314t-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl (624.3 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmanylinux: glibc 2.17+ ppc64le

odis_python-2026.9.0-cp314-cp314t-manylinux_2_17_armv7l.manylinux2014_armv7l.whl (571.7 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmanylinux: glibc 2.17+ ARMv7l

odis_python-2026.9.0-cp314-cp314t-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (541.2 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmanylinux: glibc 2.17+ ARM64

odis_python-2026.9.0-cp314-cp314t-manylinux_2_5_i686.manylinux1_i686.whl (600.0 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmanylinux: glibc 2.5+ i686

odis_python-2026.9.0-cp314-cp314t-macosx_11_0_arm64.whl (512.3 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmacOS 11.0+ ARM64

odis_python-2026.9.0-cp314-cp314t-macosx_10_12_x86_64.whl (534.4 kB view details)

Uploaded Sep 5, 2026 CPython 3.14tmacOS 10.12+ x86-64

odis_python-2026.9.0-cp39-abi3-win_arm64.whl (435.8 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+Windows ARM64

odis_python-2026.9.0-cp39-abi3-win_amd64.whl (463.7 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+Windows x86-64

odis_python-2026.9.0-cp39-abi3-win32.whl (443.9 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+Windows x86

odis_python-2026.9.0-cp39-abi3-musllinux_1_2_x86_64.whl (776.5 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+musllinux: musl 1.2+ x86-64

odis_python-2026.9.0-cp39-abi3-musllinux_1_2_i686.whl (819.7 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+musllinux: musl 1.2+ i686

odis_python-2026.9.0-cp39-abi3-musllinux_1_2_armv7l.whl (859.8 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+musllinux: musl 1.2+ ARMv7l

odis_python-2026.9.0-cp39-abi3-musllinux_1_2_aarch64.whl (731.1 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+musllinux: musl 1.2+ ARM64

odis_python-2026.9.0-cp39-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (567.5 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+manylinux: glibc 2.17+ x86-64

odis_python-2026.9.0-cp39-abi3-manylinux_2_17_s390x.manylinux2014_s390x.whl (643.3 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+manylinux: glibc 2.17+ s390x

odis_python-2026.9.0-cp39-abi3-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl (635.2 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+manylinux: glibc 2.17+ ppc64le

odis_python-2026.9.0-cp39-abi3-manylinux_2_17_armv7l.manylinux2014_armv7l.whl (582.3 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+manylinux: glibc 2.17+ ARMv7l

odis_python-2026.9.0-cp39-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (551.7 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+manylinux: glibc 2.17+ ARM64

odis_python-2026.9.0-cp39-abi3-manylinux_2_5_i686.manylinux1_i686.whl (612.2 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+manylinux: glibc 2.5+ i686

odis_python-2026.9.0-cp39-abi3-macosx_11_0_arm64.whl (525.9 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+macOS 11.0+ ARM64

odis_python-2026.9.0-cp39-abi3-macosx_10_12_x86_64.whl (538.7 kB view details)

Uploaded Sep 5, 2026 CPython 3.9+macOS 10.12+ x86-64

File details

Details for the file odis_python-2026.9.0.tar.gz.

File metadata

Download URL: odis_python-2026.9.0.tar.gz
Upload date: Sep 5, 2026
Size: 55.6 kB
Tags: Source
Uploaded using Trusted Publishing? No
Uploaded via: uv/0.12.10 {"installer":{"name":"uv","version":"0.12.10","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

File hashes

Hashes for odis_python-2026.9.0.tar.gz
Algorithm	Hash digest
SHA256	`c2c8d6d2eb23c2c6f0b8a5bf8d661b6c949e069619e4a099a1c9d42f3c2620a0`
MD5	`2c344ae4487857078d5940b05eb4cd6e`
BLAKE2b-256	`314a2ade97941d4c87bbacc8272b8da0d5cd5b49a0e91fdf28e43e6453ed8ed9`

Algorithm	Hash digest
SHA256	`90fb8ba67e44c28a6b0e4fa4374483b4db60f0fda75861506acc2766042eaf3c`
MD5	`d4d0ca18f66ba7458c15fe9834bdd1fd`
BLAKE2b-256	`ea4a6b33c6365f12fe8a3fee5c92c0fbd6f90aaf5a478cb271e45de2be921de0`

Algorithm	Hash digest
SHA256	`be36f7bb0e52872cc192e08ed4c15a811a6c33f3eb0eb682bfd16c48ead07019`
MD5	`77b0a56078cbf29db9810c96feabfe82`
BLAKE2b-256	`22422ae0f0397870cd9a79c176d85faa232355e5651fef086acb781ee8e899e6`

Algorithm	Hash digest
SHA256	`b7d187f0a830b984de43b43571fbf200c1d71904b7e4d1145d100f3bc35bad0e`
MD5	`76bae807f761e12b536ad1800739e09e`
BLAKE2b-256	`8c91db9b5ea3bb67cd8d2eef846dc56a8f0ccb7b02271a58f060e2a351a1d16e`

Algorithm	Hash digest
SHA256	`19fbddd87f6fe9fa31e61fdbf0e67daedc1c0b2f053b3ecc3bc7ced475f2121c`
MD5	`d422388ecc14be98f01e7f09768a9bbd`
BLAKE2b-256	`5c1539f8d1752a0975383dc481fc09c39a723cd88f2ffc3f31befd08f88a1933`

Algorithm	Hash digest
SHA256	`02b8716418d0d1bb5121c73dc3b30d4c72baede52154ea1418e9582e23289953`
MD5	`b0914c8943f31642d87990410ddb2cea`
BLAKE2b-256	`96daf9826df8c83ba3feda5371d7e84b63dcbd11ab42a16f6ade793d509e1b7f`

Algorithm	Hash digest
SHA256	`381af320d9c166a031fc49a2a20788689724321e403fa04c38e94c4b80fd7b91`
MD5	`7619a056051ae06a7fd34b47f3e2d12b`
BLAKE2b-256	`48a69a500cdf25f84308d01c29231ab208542c03aa129fad4cf2734810197dc9`

Algorithm	Hash digest
SHA256	`4b240311b4395d9cb8d9ac6b690d9762c68bbd63055a7b2ac02951449a25aa45`
MD5	`4124d6e167d24023395b0f3d22f9b448`
BLAKE2b-256	`3fc36c6c7e3a203f9bcadf154eed0923e1a909dd02170bec17e2b0698069fd6c`

Algorithm	Hash digest
SHA256	`81d888f8ff68033341307e149065731f9e9e9c071273a74377c6e42aa80501b6`
MD5	`6a8202cd994de5ae5bb022fa16cd3184`
BLAKE2b-256	`c9b25a7fdb6197e5ebf5cc9cbc61792fa49c1347eb65102752b3586e754e2170`

Algorithm	Hash digest
SHA256	`e012ca6fe2bd6b7310f0208935d118e1ee71e6598be1ce3588ca06e1f1eccf73`
MD5	`b7dba00b9ca2cd4d2188c251e8cda3a4`
BLAKE2b-256	`5d094b7f322a7b50c25b401f4d7259182ff8114e68a7f98f6c52d345b3c5b22a`

Algorithm	Hash digest
SHA256	`68dc8860d3f103991f11881f1922d08de1969526c873316375708566ce45467f`
MD5	`656ebdce94e0ad68da9333825542e963`
BLAKE2b-256	`c6b498dafcce6fc5d2621478a7f29fdace617bd13e3b004adb710981d92b06d5`

Algorithm	Hash digest
SHA256	`f17595f64da4c5d076b382dd297afd8e7d33af2138f0fc8ce12992a136cf1e82`
MD5	`4d27e47eac45e230eb814b08d0fa2bbe`
BLAKE2b-256	`0682ae0206e0831e40585da133c695e6f3eef392a220e08e9078408078c90f75`

Algorithm	Hash digest
SHA256	`b98f284d4165650ce7ec2d15db705dd5ce2abb5fa43705e518b2234d3f6a5cfe`
MD5	`426dbbc14e9ec9764099039b0c6733f5`
BLAKE2b-256	`d2f60d3a8032f70aeb8519b071a19427458ccc59af48c76bd295743534d2bb7a`

Algorithm	Hash digest
SHA256	`d08a3f0f0e200e448c16df526a44b8dd65d3f94fa3fde2eaedf673a4de4a762b`
MD5	`9c5bbc83a6ac3755727f095193d14395`
BLAKE2b-256	`cd0ee50b2be80d06db90e1a4df6cefc6d069e9ce405961f24132babf5a817803`

Algorithm	Hash digest
SHA256	`5f339c8465234bdfb269fd1b0805ab1615163594b6cebd43f9345d268abd6a8b`
MD5	`b87b68045fa56498f41dc9108af80264`
BLAKE2b-256	`96c7446d659622dce50a8d43e6401d2bbfac6d67ef9ade2c82dcbecbc8e5a7ba`

Algorithm	Hash digest
SHA256	`ac14206d215251487ab01d5cd3a7bd8c5af89f06c9c8a9a4d0946689df36e060`
MD5	`9f4ee23efda8642fedba0b75f6408096`
BLAKE2b-256	`f6d457b6f19033bd7b3970c527e4c3d9cf2964ffe427358c66c38f228210e287`

Algorithm	Hash digest
SHA256	`0ee59a18b9377160fc2ea192139f98e31c6c5a73464ec56b099538eac2c78156`
MD5	`68cd454e8de32ee85c0ba8667e87aea1`
BLAKE2b-256	`780f9722252377d115c497562c1324e29d543621f5d437e08651342c400a3022`

Algorithm	Hash digest
SHA256	`c103b32d8981321addd0596ae782ffc8d6b63b4367c4ca96243e2ffb6fee2a45`
MD5	`4a8ddfb83f135a3ed402c13be69cbc98`
BLAKE2b-256	`13484b59b75a18a4efd04465227051891a5815f3a8c9267e6224eb896f5e024f`

Algorithm	Hash digest
SHA256	`d400bfa2a368338c71c7fdad294407676edb9964190175071e9030a2a4cdeafb`
MD5	`edfd3062687e8333488db33ef3346530`
BLAKE2b-256	`03bacfb6a2b5ed34652fd9b5de3c0ca3da40f5941ada2f4f847009a6327b0c7d`

Algorithm	Hash digest
SHA256	`cf509297714fe5dcce1497503700a92977934f0e92e4ee2c787c1c1488231bcc`
MD5	`a7c7dc56812928cd56d67f015b136a56`
BLAKE2b-256	`9724958bcb0dd6ee006374c5e6bb4e4f3b6a126c3dcc8aca87a8274296365ae4`

Algorithm	Hash digest
SHA256	`6174cf33fdf58612e8a5c2d7b54c67500aa5fb537d47abc3696405dc1d67cca7`
MD5	`66454d2549266d1094bdf49627349422`
BLAKE2b-256	`a6bd64d0b75ab3e4e434e0fd7b36532ed947db05544ec74f6d410dfaf20ec2f4`

odis-python 2026.9.0

odis-python

Background

Installation

Released package (PyPI)

Development build (from source)

Quick Start

FormalContext

Construction

Introspection

Incidence Access

Mutation

Serialisation

Derivation Operators

Drawing Shortcut

Concepts

Implications

Attribute Exploration

Drawing

Concept Lattice Drawings

DimFlux

Titanic

LabelSet

Lazy Generators & Mutation Guard

Error Reference

Project links

Key dates

1 maintainer

License

Requires

Classifiers

Download files

Source Distribution

Built Distributions

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

File metadata

File hashes

File details

Algorithm	Hash digest
SHA256	`c407f137e83bc2f95ce1b7163465e255b03ab1aa2d815815d93c9e8d24f579fa`
MD5	`a2f4c7c033ee5afa5159b0fa473e8096`
BLAKE2b-256	`39662632fc4c091ae8590c39811abd01c52b5f9daccc744d23ebd04d41938e8a`

Algorithm	Hash digest
SHA256	`4cae030a2100f02ca18dcb70aa03f504ad15c2e131dcf8c46ed00ed5855ee6ca`
MD5	`10fdd81b75fcc46cb18c0e4e8bea0c55`
BLAKE2b-256	`84a47ae612ce9e3dd05c09c0a5647d218ce94a90a400ec7a1c95bcb5416dfe30`

Algorithm	Hash digest
SHA256	`83c88aafe69e3f395a1e76bd4468f36137b097d8c73e364a73a11dc4e89af4d6`
MD5	`36d16fd917a738e6f0215cc290c3d455`
BLAKE2b-256	`31be60446f6b275c6687bd8366518686a9fb4c615037241b80c4bc9e74bb9beb`

Algorithm	Hash digest
SHA256	`7f9347e143c8261fa729d03b47951bcf4002f9a91d55bc95610365aa64397303`
MD5	`00e3056145dc92535f2065b68a9272bc`
BLAKE2b-256	`c236ff7daa36d0837ec5d42688f5f4cf695422a24abcc8e10b5ed30b3f087790`

Algorithm	Hash digest
SHA256	`bdc2626a0ff6c88a4ae3c216c932e0a3a3ce61b7bd5917e54414a2099c5fd0ce`
MD5	`6cf506b31955c8152268076bde0a0c6f`
BLAKE2b-256	`8d0f2c19366afb9b805fe35a8116710387dcb4250e9acec7af074c5e3a39eb0a`

Algorithm	Hash digest
SHA256	`6f280923faf129cb1bfb1c29116a5cc87c921fcf5c68019f793c549b4737e3f6`
MD5	`edeeecc585c90770b30b313af743258e`
BLAKE2b-256	`8565f3b9fe5f00ce1f9857adc15e5bfcf95cb4ae4c4aabc82d02143043822076`

Algorithm	Hash digest
SHA256	`8e6a7e681162960f7f2e5db05e622130c0ce382ed246d60ee431fbd664a01aab`
MD5	`1da83d709a5cecf0b90dff39a258a1c2`
BLAKE2b-256	`4bee093b0240117a5a4428a8ec5d908b7261b7e3fce4cc203d1fc2bff0e602c0`

Algorithm	Hash digest
SHA256	`ef612b3e1a8342e4bffef4cbd4b4e2d0fed85b4c27d0b8d7480b8e789976498e`
MD5	`172f63f37d8c2bf519da05dd53a1fbf3`
BLAKE2b-256	`f9cd4872df87ea30a4b0bda6cd091c0d86bcadc0f446618e9575b49af2d997cb`

Algorithm	Hash digest
SHA256	`b3264ee8bd465c5b56c2d111e291f234e71167cea245428e58747f316fc64619`
MD5	`580882bb6efc3bd1a8ac9501851cedd8`
BLAKE2b-256	`4f5c57d911f887866079878e73d977817a39a9d7bf7873192085d08fe0c12955`

Algorithm	Hash digest
SHA256	`a080a9463cac677619dd1b984e61c758fa434b7ba1a383cc701d059ea4be7009`
MD5	`e61ebd72a15d2b4a211223ba5c9fab27`
BLAKE2b-256	`9b39a8777ab6c1cf684f74f7834f8cbeefbb22426ba339f1be701d74523efbb1`