Skip to main content

ezdgn

ezdgn is a native, read-only V8 and read/write V7 DGN toolkit for Python, implemented with an independently written Rust core and PyO3 bindings. It has no ODA SDK, binary, or runtime dependency.

Installation

ezdgn requires Python 3.10 or newer.

python -m pip install ezdgn

Install the optional Matplotlib renderer when preview images are needed:

python -m pip install "ezdgn[plot]"

Building from source requires Rust 1.83 or newer. The extension uses Python's stable ABI (abi3) with a Python 3.10 minimum.

Supported scope

Format and operation Support
V7/ISFF 2D read Native entities, hierarchy, metadata, linkages, and raw records
V7/ISFF 2D write Standalone or custom-seed creation of common primitives
V7/ISFF 3D Signature, raw record, and common-header inspection only
V8 DGN read Native 2D/3D models, common geometry, hierarchy, linkages, auxiliary data, and lossless raw objects
V8 DGN write/edit Not supported

The V7 reader decodes line, line string, shape, curve, ellipse, arc, text, cell, text node, complex chain/shape, and B-spline records as native entities. It restores parent/child relationships without flattening component records, decodes typed attribute linkages while retaining their exact bytes, applies high-precision sub-UOR coordinate corrections alongside the stored integers, and resolves outline/fill colors through the active color table. Every record, including unsupported control and application elements, retains its original bytes.

The writer creates standalone V7 2D files from a bundled empty seed. A custom seed can be supplied to preserve its units, origin, and design plane. It writes line, line string, shape, curve, ellipse, arc, circle-as-ellipse, and raw-byte text entities with basic symbology and shape fill linkage. Coordinates outside the selected seed's design plane are rejected instead of silently clipped.

The native V8 reader decodes model metadata, line, line string, shape, curve, point string, text, ellipse, arc, text node, complex chain/shape, cell, shared-cell instance, B-spline curve/poles, and a bounded dimension anchor. Unknown graphical, control, name-table, linkage, and auxiliary records remain addressable with exact bytes. Product-specific custom-object semantics, full dimension annotation semantics, V8 writing/editing, and 3D screen projection for plotting remain outside the supported scope.

V7 3D geometry is not supported. The V7 writer does not yet create cells, complex elements, B-splines, arbitrary linkages, or perform in-place editing. The raw V7 record framing is shared by 2D and 3D files, so scan_records() can inspect a 3D stream safely without implying V7 3D entity support.

2D entity API

import ezdgn

drawing = ezdgn.readfile("drawing.dgn")

# All records remain ordered and lossless. entities contains only top-level
# graphics; all_entities also includes drawable component records.
print(len(drawing.elements), len(drawing.entities), len(drawing.all_entities))

for entity in drawing:
    print(entity.dxftype(), entity.level, entity.style)

for line in drawing.query("LINE"):
    print(line.start_uor, line.end_uor)
    print(line.start_master, line.end_master)

for text in drawing.query("TEXT"):
    print(text.text_bytes)
    print(text.decode_text())  # ASCII with strict errors by default
    print(text.decode_text("cp932", errors="replace"))

for cell in drawing.query("CELL"):
    print(cell.name, cell.origin_master, cell.transform)
    for component in drawing.children(cell):
        print("  ", component.dxftype())

for element in drawing.elements:
    for linkage in element.linkages:
        print(linkage.kind, linkage.linkage_type_name, linkage.raw_bytes)

drawing.entities contains supported top-level drawable entities only. drawing.all_entities is a flat tuple in original drawing.elements order: it includes both Cell/ComplexChain/ComplexShape container headers and their supported drawable descendants at every depth. It excludes control records, UnsupportedElement instances, and non-drawable B-spline support records; those remain available through the lossless drawing.elements sequence. drawing.unsupported_elements provides a filtered diagnostic view for unknown kinds. Use parent(), children(), or descendants() when a tree view is needed.

Ellipse, arc, curve, and B-spline entities retain their native parameters or control records; they are not flattened to polylines. parent_index and child_indices refer to the lossless drawing.elements sequence, while drawing.parent(), children(), and descendants() resolve the objects. Stored integer UOR, sub-UOR-corrected floating UOR, and optional master-unit coordinates coexist. drawing.color_table is the last type-5, level-1 color table in file order. drawing.resolve_color(), entity.style.rgb, and fill_rgb use that table when present and otherwise fall back to the standard MicroStation V7 256-color palette.

V7 DGN does not record a text code page. Text.decode_text() therefore uses encoding="ascii", errors="strict" as a deterministic default; pass the project encoding explicitly for non-ASCII text.

Known DMRS/database, association ID, shape fill, and high-precision linkages have typed fields. Unknown user linkages and malformed trailing attribute bytes remain accessible through read-only raw views. Shared-cell definition/instance types 34/35 remain raw because the public ISFF chapter does not specify their layout.

The high-level read()/readfile() API deliberately rejects V7 3D files. scan_records() and inspect_headers() still support bounded inspection of their shared record framing and metadata.

Native V8 API

Use open_document() when a caller may receive either format. The existing read() and readfile() names remain V7-specific for compatibility.

import ezdgn

document = ezdgn.open_document("drawing.dgn")
if isinstance(document, ezdgn.V8Document):
    for model in document.models:
        print(
            model.metadata.name,
            model.metadata.dimension,
            model.metadata.master_unit,
        )
        for entity in model.entities:
            print(entity.dxftype(), entity.level, entity.common.color_index)

        for text in model.query("TEXT"):
            print(text.data.text, text.data.text_bytes)

        for cell in model.query("CELL"):
            for component in model.children(cell):
                print("  ", component.dxftype())

V8Model.elements is the complete graphical-object sequence. V8Model.entities is the feature-oriented view: complex/cell headers remain native aggregate entities, while TextNode headers yield their text children. parent(), children(), and descendants() navigate the original hierarchy. 2D and 3D coordinates coexist as (x, y, z) tuples; no Z coordinate is dropped. For complex headers, common.stored_dimension retains the header bit and common.dimension reports the effective dimension inherited from children. V8 text exposes its stored width_multiplier_raw and height_multiplier_raw alongside corrected width_uor, height_uor, width_master, and height_master distances.

The raw scanner is available independently of semantic decoding:

raw = ezdgn.scan_v8_objects("drawing.dgn")
for model in raw.models:
    for obj in model.graphical_objects:
        print(
            obj.stream_path,
            obj.inflated_offset,
            obj.element_type,
            obj.role,
            obj.raw_bytes,
        )

Every extraction and decode stage is bounded. Override the defaults as one immutable policy object when processing untrusted files:

limits = ezdgn.V8ScanLimits(
    max_file_size=256 * 1024 * 1024,
    max_objects=250_000,
    max_total_inflated_bytes=512 * 1024 * 1024,
)
document = ezdgn.read_v8("drawing-v8.dgn", limits=limits)

Malformed structure raises InvalidDgnError; configured resource ceilings raise DgnLimitError. The complete clean-room boundary, evidence ledger, and known limitations are recorded in docs/v8/SCOPE.md, docs/v8/FORMAT_NOTES.md, and docs/v8/PROVENANCE.md.

Plotting parsed drawings

The optional renderer can display a parsed V7 2D drawing or a V8 document and save it as an image without changing the native entity model:

import ezdgn

drawing = ezdgn.readfile("drawing.dgn")

figure, axes = ezdgn.plot(
    drawing,
    text_encoding="cp932",
    background="#111111",
)
figure.savefig("preview.png", dpi=150, bbox_inches="tight")

# Or render and save in one call.
ezdgn.save_plot(drawing, "preview.png", text_encoding="cp932")

# V8 uses already decoded native text and preserves Z in the object model;
# the preview is an explicit XY projection.
ezdgn.save_plot(ezdgn.read_v8("drawing-v8.dgn"), "preview-v8.png")

The equivalent CLI command is:

ezdgn plot drawing.dgn -o preview.png --encoding cp932

Omit -o to open an interactive Matplotlib window. Use --monochrome for a high-contrast preview, --hide-text to suppress text, or --coordinate-space uor when master-unit coordinates are unavailable. Run ezdgn plot --help for the complete option list.

Lines, line strings, shapes, ellipses, arcs, text, and drawable components of cells and complex elements are rendered. Ellipses and arcs are sampled only for display. Native type-11 curves and B-spline curves are previewed from their parsed control sequences; the source records and entity parameters are never flattened or modified. V8 3D geometry uses an XY preview projection, and orientation matrices are retained but not applied by the 2D renderer. V7 text does not store its code page, so the caller must select the correct encoding for non-ASCII text. Geometry and text with compatible display styles are batched to keep large previews practical.

V7 writer and optional custom seeds

import ezdgn

doc = ezdgn.new()  # uses the bundled empty V7 2D seed
msp = doc.modelspace()

msp.add_line(
    (0, 0),
    (10, 5),
    dgnattribs={"level": 2, "color": 3, "line_weight": 2},
)
msp.add_line_string([(0, 10), (5, 15), (10, 10)])
msp.add_shape(
    [(20, 0), (30, 0), (30, 10), (20, 10)],
    fill_color=6,
)
msp.add_ellipse((25, 25), primary_axis=5, secondary_axis=3, rotation=30)
msp.add_arc(
    (40, 5),
    primary_axis=5,
    secondary_axis=3,
    start_angle=30,
    sweep_angle=120,
)
msp.add_text("日本語", (0, 30), height=2, encoding="cp932")

doc.saveas("drawing.dgn")
roundtrip = doc.readback()

Pass a custom seed as ezdgn.new("project_seed.dgn") to use its units, origin, design plane, and active color table. new() copies the mandatory TCB, digitizer setup, level symbology, and last active color table from the selected seed. Set copy_seed_elements=True to retain every seed record, including any graphics. Text encoding is not recorded by V7 DGN, so add_text() accepts bytes directly or uses its encoding="ascii", errors="strict" defaults for str input.

Raw record API

import ezdgn

info = ezdgn.detect_format("drawing.dgn")
print(info.kind, info.dimension)

scan = ezdgn.scan_records("drawing.dgn")
print(len(scan.records), scan.termination)

for record in scan.records:
    print(record.offset, record.element_type, record.level, record.raw_bytes)

The V8_CFB result means that the input has the generic CFB signature used by V8 DGN files. It is intentionally described as a candidate because the outer signature alone does not prove that DGN-specific streams are present.

The bounded container inspector is the structural-only entry point. It verifies the known DGN root markers without decoding DGN stream contents:

container = ezdgn.inspect_v8_container("drawing-v8.dgn")
print(container.has_dgn_v8_markers)
print(container.model_storage_paths)
for entry in container.entries:
    print(entry.path, entry.kind, entry.size_bytes)

inspect_v8_container() alone is structural identification, not an entity or fidelity claim. Use scan_v8_objects() for bounded, exact raw objects and read_v8() or open_document() for native semantic decoding. The legacy read(), readfile(), and scan_records() contracts remain V7-specific and therefore reject V8 input. No V8 path silently converts to V7 or flattens native objects.

Design settings and common headers

import ezdgn

headers = ezdgn.inspect_headers("drawing.dgn")
settings = headers.design_settings

print(settings.master_unit_name, settings.uor_per_master)
print(settings.global_origin_master)

for element in headers.elements:
    common = element.common_header
    if common is not None:
        print(
            element.record.element_type,
            common.range.low_master,
            common.range.high_master,
            common.symbology.color,
        )

read_design_settings() decodes only the leading TCB. inspect_headers() pairs every raw record with its standard common header when that element type has one. Attribute bytes remain available as a read-only zero-copy ElementMetadata.attribute_view.

The same inspection is available from the CLI:

ezdgn inspect drawing.dgn
ezdgn inspect drawing.dgn --records --json
ezdgn inspect drawing.dgn --headers --json
ezdgn inspect drawing.dgn --entities --json
ezdgn inspect drawing-v8.dgn --json

Development

python -m venv .venv
. .venv/bin/activate
python -m pip install "maturin>=1.13,<2" "pytest>=8" "matplotlib>=3.8"
maturin develop
cargo fmt --check
cargo clippy --workspace --all-targets --all-features -- -D warnings
cargo test --workspace --all-features
python -m pytest
sha256sum -c tests/data/dgn/SHA256SUMS
uv lock --check

Run the bounded V8 fuzz target and parser microbenchmark separately:

cargo +nightly fuzz run v8_read -- -max_len=16777216
EZDGN_BENCH_ITERATIONS=100 cargo bench -p ezdgn-core --bench v8_read

Build a distributable wheel with:

maturin build --release --out dist

License

ezdgn is released under the MIT License. The bundled empty seed and test fixtures retain the separate upstream terms documented in src/ezdgn/_data/README.md and tests/data/dgn/README.md.

Download files

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

Source Distribution

ezdgn-0.2.4.tar.gz (135.5 kB view details)

Uploaded Source

Built Distributions

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

ezdgn-0.2.4-cp310-abi3-win_amd64.whl (427.1 kB view details)

Uploaded CPython 3.10+Windows x86-64

ezdgn-0.2.4-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (580.1 kB view details)

Uploaded CPython 3.10+manylinux: glibc 2.17+ x86-64

ezdgn-0.2.4-cp310-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (570.6 kB view details)

Uploaded CPython 3.10+manylinux: glibc 2.17+ ARM64

ezdgn-0.2.4-cp310-abi3-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl (998.2 kB view details)

Uploaded CPython 3.10+macOS 10.12+ universal2 (ARM64, x86-64)macOS 10.12+ x86-64macOS 11.0+ ARM64

File details

Details for the file ezdgn-0.2.4.tar.gz.

File metadata

  • Download URL: ezdgn-0.2.4.tar.gz
  • Upload date:
  • Size: 135.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for ezdgn-0.2.4.tar.gz
Algorithm Hash digest
SHA256 6c493299eba6a80400f3bd56f5ee0747a7ab3bb7b1b65a15280547f9151b0fa6
MD5 a0d3d4021d06aa83df904cf517ebe7f4
BLAKE2b-256 61377c5c1b72aeb42e57a6555670b620c4df61dafbb6c44523fa1896eea7e88a

See more details on using hashes here.

Provenance

The following attestation bundles were made for ezdgn-0.2.4.tar.gz:

Publisher: release.yml on monozukuri-ai/ezdgn

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file ezdgn-0.2.4-cp310-abi3-win_amd64.whl.

File metadata

  • Download URL: ezdgn-0.2.4-cp310-abi3-win_amd64.whl
  • Upload date:
  • Size: 427.1 kB
  • Tags: CPython 3.10+, Windows x86-64
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for ezdgn-0.2.4-cp310-abi3-win_amd64.whl
Algorithm Hash digest
SHA256 53f3e4d56c216064a57c9cee3ab042cb77d2aeffe6aa4e02a6b5d5b928a59a36
MD5 7abcca0583711a724fd4114d7175e048
BLAKE2b-256 c40a39501b4878c1cfac9efebaabeec48fd42b189838e594855c5812405f8030

See more details on using hashes here.

Provenance

The following attestation bundles were made for ezdgn-0.2.4-cp310-abi3-win_amd64.whl:

Publisher: release.yml on monozukuri-ai/ezdgn

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file ezdgn-0.2.4-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.

File metadata

File hashes

Hashes for ezdgn-0.2.4-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl
Algorithm Hash digest
SHA256 f6a951f9c327577896ce7329bb5434fe6f03c818a8389393a817e84498afe9f2
MD5 bee9ae0145d3cb6aae37a49fce574600
BLAKE2b-256 ad5e74d4312f31e839f1900a6153bc9b14fc9656efbe8a7a9ab55a0a5cd57dc3

See more details on using hashes here.

Provenance

The following attestation bundles were made for ezdgn-0.2.4-cp310-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl:

Publisher: release.yml on monozukuri-ai/ezdgn

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file ezdgn-0.2.4-cp310-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl.

File metadata

File hashes

Hashes for ezdgn-0.2.4-cp310-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl
Algorithm Hash digest
SHA256 f04f49cb4b194d18d9c0473ed921ee76d0cfd1b8f8c2cda6f8ad8725808646f6
MD5 888dd614262db64d76d6ef1af52c7b69
BLAKE2b-256 573b1856b5753659c0b4412ec09326152116b3b89d65e54970daf5be1711d81f

See more details on using hashes here.

Provenance

The following attestation bundles were made for ezdgn-0.2.4-cp310-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl:

Publisher: release.yml on monozukuri-ai/ezdgn

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file ezdgn-0.2.4-cp310-abi3-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl.

File metadata

File hashes

Hashes for ezdgn-0.2.4-cp310-abi3-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl
Algorithm Hash digest
SHA256 b9cbbf49a4866ebe20e40e96c0051e55ded8a4e61d35f7bd392a392b9842b76b
MD5 61216719ccc50d64473be7ec813627c2
BLAKE2b-256 7b941fa43f3c6413ccd7dd504a7403e30cb631b2bd08e004a61fcf6c42503127

See more details on using hashes here.

Provenance

The following attestation bundles were made for ezdgn-0.2.4-cp310-abi3-macosx_10_12_x86_64.macosx_11_0_arm64.macosx_10_12_universal2.whl:

Publisher: release.yml on monozukuri-ai/ezdgn

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Release history Release notifications | RSS feed

This release

0.2.4 This release

5 files

0.2.3

5 files

0.2.2

5 files

0.2.1

5 files

0.2.0

5 files

0.1.2

5 files

0.1.1

5 files

0.1.0

5 files

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page