Polyline (`polyline`)¶

Terms¶

Polyline: An ordered sequence of vertices connected by consecutive edges: V_0 — V_1 — V_2 — … — V_{n-1}. A polyline store holds a collection of such sequences, each identified by an integer object ID.
GEOM_POLYLINE: The geometry type constant "polyline".
Object ID: A non-negative integer uniquely identifying a polyline within the store. Object IDs are assigned at write time and are stable across reads.
Object manifest: The set of (chunk_coord, vg_indices) pairs that together contain all vertices of a given object. Encoded in the object_index/ array. See Object manifest.
Cross-chunk link: A stored connection between the last vertex of a polyline segment in chunk A and the first vertex of the continuation segment in chunk B. Required when a polyline spans multiple chunks. See Cross-chunk links.

Introduction¶

The polyline type extends line with an object model: each polyline is a named, addressable entity that can be retrieved by ID. Polylines may span multiple spatial chunks; the cross_chunk_links/ array preserves inter-chunk connectivity.

Use polyline over line whenever you need to:

Retrieve individual paths by ID.
Assign paths to named groups.
Preserve path topology across chunk boundaries.
Store per-path metadata (length, label, confidence).

Use streamline instead of polyline when the paths are MRI tractography streamlines and you need to store tractography-specific metadata (step size, seeding strategy, propagation algorithm).

Technical reference¶

Arrays present¶

Array path	Required	Description
`vertices/`	Yes	Vertex positions
`vertex_fragments/`	Yes	Fragment index over `vertices/` rows
`links/<delta>/`	Yes	Intra-chunk consecutive vertex pairs
`link_fragments/`	Yes (`<delta>=0`)	Fragment index over `links/0/` rows
`object_index/`	Yes	Per-object manifest blobs naming fragments
`cross_chunk_links/`	Yes*	Inter-chunk vertex connections
`attributes/<name>/`	No	Per-vertex attributes
`object_attributes/<name>/`	No	Per-polyline attributes
`groupings/`	No	Group ID → [object IDs]

*cross_chunk_links/ must be present if any polyline spans more than one chunk. It may be absent for stores where all polylines are confined to a single chunk each.

Vertex ordering within a chunk¶

For a polyline that contributes vertices to multiple bins within a chunk, its vertices appear in multiple VGs. Within each VG, vertices are stored in traversal order (from the start of the polyline toward the end).

Across VGs within the same chunk, vertices may be interleaved with vertices from other polylines. The links/<delta>/ array stores the consecutive pairs for all polylines in the chunk, regardless of VG assignment.

`links/<delta>/` for polylines¶

Each row in links/<delta>/ is a pair [i, j] where i and j are local-chunk vertex indices and vertex i immediately precedes vertex j in the traversal order of some polyline. For a polyline contributing vertices at local indices [3, 7, 11, 4] (in traversal order), the edges are [[3,7], [7,11], [11,4]].

Only intra-chunk edges are stored here. The connection between the last vertex of a segment in chunk A and the first vertex of the continuation in chunk B is stored in cross_chunk_links/.

Write API¶

import numpy as np
from zarr_vectors.types.polylines import write_polylines

rng = np.random.default_rng(0)
polylines = [
    rng.normal(0, 50, (rng.integers(10, 60), 3)).cumsum(0).astype(np.float32)
    for _ in range(1000)
]

write_polylines(
    "paths.zarrvectors",
    polylines,
    chunk_shape=(200.0, 200.0, 200.0),
    bin_shape=(50.0, 50.0, 50.0),
    # Optional group assignment
    groups={
        "group_A": list(range(500)),
        "group_B": list(range(500, 1000)),
    },
    # Optional per-polyline attributes
    object_attributes={
        "length": np.array([np.sum(np.linalg.norm(np.diff(p, axis=0), axis=1))
                            for p in polylines], dtype=np.float32),
    },
)

Read API¶

from zarr_vectors.types.polylines import read_polylines

# All polylines
result = read_polylines("paths.zarrvectors")
print(result["polyline_count"])        # 1000
print(len(result["polylines"]))        # list of (N_i, D) arrays

# By object ID
result = read_polylines("paths.zarrvectors", object_ids=[0, 5, 42])
print(result["polyline_count"])        # 3

# By group
result = read_polylines("paths.zarrvectors", group_ids=["group_A"])
print(result["polyline_count"])        # 500

# Spatial bbox
result = read_polylines(
    "paths.zarrvectors",
    bbox=(np.array([0., 0., 0.]), np.array([100., 100., 100.])),
)
# Returns all polylines that have at least one vertex in the bbox.
# Full polyline geometry is returned (not clipped to bbox).

Spatial query semantics¶

A bbox query on a polyline store returns the complete geometry of every polyline that has at least one vertex in the bbox. The full vertex sequence (including portions outside the bbox) is returned. This is necessary to preserve path continuity.

To clip polylines to the bbox (return only the vertices inside), pass clip=True:

result = read_polylines("paths.zarrvectors",
                        bbox=(lo, hi), clip=True)

Clipped polylines that cross the bbox boundary are split at the boundary; the result may contain more polylines than the input if a single path enters and exits the bbox multiple times.

Groupings¶

Groups are named collections of object IDs. The groupings/ array stores the group → object mapping. Group IDs may be integers or strings (stored as a separate string lookup table in groupings_attributes/name/).

Validation¶

L1: vertices/, vertex_fragments/, links/<delta>/, link_fragments/ (at <delta>=0), and object_index/ exist.

L3:

Every object ID in object_index/ is in [0, n_objects).
For each object, all referenced VGs exist and contain at least one vertex.
cross_chunk_links/ entries reference valid global vertex IDs.
No polyline has a gap (a vertex with no outgoing edge unless it is the last vertex of the polyline).

Polyline (polyline)¶