Skip to content

Releases: nlyamada/export-igorgraph

export-igorgraph 0.1.1

Choose a tag to compare

@nlyamada nlyamada released this 09 Oct 12:35

First public release.

export-igorgraph turns a matplotlib figure into an editable Igor Pro graph:
export_igor(fig, outdir, name) writes one HDF5 file (data + Igor drawing commands),
and LoadPythonFigure in Igor draws it.

See README (EN) / README (JP) for details.

Claude skill

export-igorgraph-public.skill (attached in this page) is a skill file for Claude.ai (upload it in the skill settings) and Claude Code (unzip it into ~/.claude/skills/). It bundles the Python package and keeps your default formatting in export_igorgraph_style.json.

Highlights

  • Lines (solid and dashed), 20 marker shapes (5 of them approximated, with a warning), error bars with caps, log axes (10^n labels), grid, tick direction, legends, labels
  • Scatter plots with one color and size (per-point colors/sizes: the first value is used, with a warning)
  • Images, color maps, framed color scale
  • Contours drawn as real Igor contours (AppendMatrixContour / ModifyContour) from the original Z matrix (matplotlib 3.8+)
  • Main formatting via style= or a style file: tick, mirror, standoff, grid, width, height, font, gFont
  • The Igor loader runs only commands that match a strict allow-list and rejects the whole file otherwise
  • Anything Igor cannot reproduce is reported in report.warnings, never dropped silently

Install

  1. pip install export-igorgraph (or pip install git+https://github.com/nlyamada/export-igorgraph)
  2. Open ExportIgorGraphLoader.ipf (attached in this page), or copy to Igor procedure folder Documents\WaveMetrics\Igor Pro * User Files\Igor Procedures\ and restart Igor.
  3. In Igor: Macros > LoadPythonFigure

Status

  • Verified on Igor Pro 8.04 / Windows only. Other Igor versions and macOS are unverified — reports welcome.
  • Python 3.9+. CI runs matplotlib 3.6.3, 3.8.4, 3.9.4 and 3.10.7 (contours need 3.8+).

See the CHANGELOG and the documentation.

export-igorgraph 0.1.0

Choose a tag to compare

@nlyamada nlyamada released this 09 Oct 08:50

First public release.

export-igorgraph turns a matplotlib figure into an editable Igor Pro graph:
export_igor(fig, outdir, name) writes one HDF5 file (data + Igor drawing commands),
and LoadPythonFigure in Igor draws it.

Highlights

  • Lines, 20 marker shapes, dashes, error bars with caps, scatter, log axes (10^n labels), grid, tick direction, legends, labels
  • Images, color maps, framed color scale
  • Contours drawn as real Igor contours (AppendMatrixContour / ModifyContour) from the original Z matrix (matplotlib 3.8+)
  • Main formatting via style= or a style file: tick, mirror, standoff, grid, width, height, font, gFont
  • The Igor loader runs only commands that match a strict allow-list and rejects the whole file otherwise
  • Anything Igor cannot reproduce is reported in report.warnings, never dropped silently

Install

  1. pip install git+https://github.com/nlyamada/export-igorgraph
  2. Copy igor/ExportIgorGraphLoader.ipf to Documents\WaveMetrics\Igor Pro 8 User Files\Igor Procedures\ and restart Igor
  3. In Igor: Macros > LoadPythonFigure

Claude skill

export-igorgraph-public.skill (attached) is for Claude.ai; Claude Code users can install the plugin from this repository (see the README).

Status

  • Verified on Igor Pro 8.04 / Windows only. Other Igor versions and macOS are unverified — reports welcome.
  • Python 3.9+; tested with matplotlib 3.6–3.10 (contours need 3.8+).

See CHANGELOG.md and docs.