cad-easy-font 0.1.3

DXF text and AutoCAD SHX/BigFont support implemented with Rust and PyO3.

easy-font

easy-font is a PyO3/Rust Python extension focused on DXF text handling and AutoCAD SHX font compatibility. It intentionally starts smaller than ezdxf: the first milestone is reliable text/style extraction plus support for AutoCAD-86 SHX unifont and bigfont files found in CAD datasets.

Development

nix-shell
maturin develop
pytest

Common release commands are captured in justfile:

just test          # cargo test + Python pytest
just build         # build manylinux wheel and sdist into /tmp/easy_font_dist_publish
just publish       # run checks, rebuild artifacts, upload with pypi_token
just publish-built # upload existing artifacts from /tmp/easy_font_dist_publish

Python API

import easy_font as cad

font = cad.open_shx("cad-data/fonts/hztxt.shx", encoding="gbk")
print(font.kind, font.is_bigfont, font.glyph_count)
print(font.glyph_code_for_char("中"))  # GBK BigFont glyph code, e.g. 0xD6D0
print(font.missing_chars("中文ABC"))

doc = cad.readfile("cad-data/data/hatch.dxf")
for entity in doc.modelspace():
    if entity.dxftype == "TEXT":
        print(entity.text)

dwg = cad.readfile("cad-data/data/baseline-sample.dwg")
print(dwg.format, len(dwg.modelspace()))

# Resolve extracted text to local font files.
resolved_rows = cad.resolve_text_runs(dwg, ["cad-data/fonts"])
print(resolved_rows[0]["style"], resolved_rows[0]["font_path"])

Render one line of Chinese CAD text

CAD text styles often pair a primary ASCII SHX font with a BigFont for Chinese glyphs. FontStack keeps that split explicit, so ASCII such as -2# uses simplex.shx while Chinese text is encoded through the BigFont mapping.

from pathlib import Path

import easy_font as cad

fonts = Path("cad-data/fonts")
font = cad.FontStack(
    ascii_font=fonts / "simplex.shx",
    bigfont=fonts / "gbcbig.shx",
    encoding="gbk",
)

text = "明厚源悦府-2#"
print(font.glyph_code_for_char("明"))
print(font.encode(text))
print(font.missing_chars(text))

# Geometry in drawing units, already scaled to a 12-unit text height.
strokes = font.text_strokes(text, height=12, tracking=0.1, align="left")
bbox = font.text_bbox(text, height=12, tracking=0.1)
advance = font.text_advance(text, height=12, tracking=0.1)

# SVG path only, useful when embedding into a larger document.
path = font.text_svg_path("共1页，第1页", height=12, tracking=0.1, align="center")

# Complete SVG document and PNG bytes.
svg = font.render_text_svg(text, height=12, tracking=0.1, stroke_width=0.35, align="right")
Path("cad-text.svg").write_text(svg, encoding="utf-8")

png = font.render_text_png(text, height=120, stroke_width=3, align="left")
Path("cad-text.png").write_bytes(png)

For a single SHX font, the lower-level object also exposes glyph_code_for_char(), encode(), has_char(), missing_chars(), glyph_metrics(), text_strokes(), text_advance(), text_bbox(), and text_svg_path().

Current scope

• Reads ASCII DXF group-code pairs.
• Reads DWG files using a handwritten Rust binary parser path.
• Extracts STYLE table records, including primary SHX font and BigFont file.
• Extracts common text entities from ENTITIES: TEXT, MTEXT, ATTRIB, and ATTDEF.
• Recognizes AutoCAD-86 SHX unifont and bigfont headers.
• Preserves raw SHX glyph records so BigFont files that ezdxf cannot load are still inspectable.
• Maps Unicode characters back to encoded SHX glyph codes for BigFont files such as GBK/CP936.
• Provides glyph metrics and high-level stroke/SVG/PNG helpers for one-line CAD text rendering.
• Decodes DXF text escapes such as %%d, \P, and \U+4E2D.
• Decodes encoded BigFont text bytes through Python codecs such as gbk, big5, and shift_jis.

Full DXF entity parity with ezdxf is out of scope for this initial implementation.

Font Resolution Workflow

python scripts/extract_and_resolve_fonts.py --input baseline-sample.dwg --limit 10

This script reads text entities from a drawing, resolves style fonts against cad-data/fonts, and prints style/text font coverage. For SHX fonts it verifies glyph coverage through open_shx; for TTF/OTF/TTC/WOFF it verifies cmap coverage when fontTools is available.

To validate real render output for SFNT fonts (TTF/OTF/TTC/WOFF/WOFF2) and save PNG previews:

python scripts/validate_render_output.py --input baseline-sample.dwg --max-rows 200 --preview-limit 20