Static GitHub Pages catalogue for tools, projects, publications, and collaborations of the INF subproject of CRC 1629 NegLaB. The complete site is generated with dependency-free Python from two manually maintained content files and one logo asset:
data/neglab-inf.bibis the single authoritative catalogue source for tools, projects, publications, collaborations, and their links.data/inf-project.jsoncontains the INF description, services, research areas, related links, contact text, and people.data/inf_logo.svgis retained as the INF visual-identity source and defines the blue and pink site palette; the homepage itself uses text branding.
The generator only reads these inputs. It never edits or merges BibTeX data.
Everything under docs/ is generated output and is replaced on every build.
From the repository root, run:
python3 src/build_publications.pyThis uses data/neglab-inf.bib and data/inf-project.json, deletes the
previous generated site under docs/, and writes the complete site again.
To build from a different BibTeX file or homepage JSON:
python3 src/build_publications.py /path/to/projects.bib \
--project-data /path/to/inf-project.jsonAn alternate output directory can be useful for previews:
python3 src/build_publications.py --docs-dir /tmp/neglab-inf-previewThe selected output directory is replaced completely during the build.
Edit data/neglab-inf.bib directly. Every entry uses the same BibTeX type,
@project. Its keywords field lists the catalogue categories that decide
on which website tabs the entry appears:
@project{INF:ExampleService,
keywords = {tool, project, collaboration},
collaboration = {B01},
name = {Example Service},
abstract = {What the service provides.},
tooltype = {service},
url = {https://example.org/service},
projectlink = {https://github.com/example/service},
tags = {annotation, hosting}
}tool→ Tools & Servicesproject→ Projectspublication→ Publications (also setpubtype, e.g.articleorinproceedings)collaboration→ Collaborations (joint work with other CRC projects; also addcollaboration = {A02, C05}listing the partner projects)
An entry may list several categories. It then appears on every matching tab
but keeps a single detail page. The first category is the primary one, which
sets the back link and the label on that page. Topical search terms go in
tags, not in keywords. Other entry types, unknown categories, duplicate
keys, and related keys that do not exist all stop the build with an error.
See PROJECT_BIBTEX_SCHEMA.md for all fields.
Each entry's BibTeX key determines its stable, collision-resistant filename
under docs/publications/, so do not rename published keys.
The link fields have distinct uses:
url: primary landing page, shown as “Open tool”, “Visit project”, or “View publication”.projectlink: code repository or archival record, shown as “Code & archive”. If it is absent, the detail page displays a disabled placeholder.documentation: dedicated documentation, shown as “Documentation”.pdf: direct PDF link.
Publication detail pages offer a standard citation BibTeX generated from the
entry, using pubtype as the entry type and leaving out catalogue-only
fields.
After adding or editing an entry, rebuild:
python3 src/build_publications.pyEdit data/inf-project.json directly. Each person is represented by an object:
{
"name": "Example Person",
"role": "Scientific Staff",
"affiliation": "Goethe University Frankfurt",
"profile_url": "https://example.org/profile"
}Change role to regroup someone, append an object to add a person, or remove
the object to remove them. affiliation and profile_url may be empty.
Descriptions are arrays of paragraphs; service groups contain a name and an
items array.
After any BibTeX or JSON edit, run the normal build command again.
The build writes:
docs/index.html: home page with one panel per tab, including counts.docs/about.html: INF description, research areas, services, and contact.docs/people.html: the INF team, grouped by role.docs/tools.html,docs/projects.html,docs/publications.html,docs/collaborations.html: searchable, filterable lists of the entries in each category.docs/publications/*.html: one detail page per BibTeX entry.docs/publications/index.json: mapping from BibTeX keys to generated files and categories.docs/assets/: the stylesheet, JavaScript, and copied visual-identity asset used by GitHub Pages.docs/.nojekyll: disables unnecessary Jekyll processing.
Asset URLs contain a content hash so browsers and GitHub Pages do not retain an older stylesheet after a rebuild.
Preview the generated site locally:
python3 -m http.server 8000 --directory docsThen open http://localhost:8000/.
Run the parser and site-generation tests with:
python3 -m unittest discover -s tests -vIn the repository settings, select Deploy from a branch, choose the main
branch, and use /docs as the folder. Commit both manually maintained input
files, generator changes, and the regenerated docs/ output.