Merge pull request #22 from ScienceLiveHub/fix/fill-readme-doi-on-release

Fill README/index Zenodo DOI badge on release

Author: annefou

.github/workflows/release-identifiers.yml

@@ -96,14 +96,14 @@ jobs:
         run: |
           set -euo pipefail
 
-          if git diff --quiet -- CITATION.cff codemeta.json ro-crate-metadata.json; then
+          if git diff --quiet -- CITATION.cff codemeta.json ro-crate-metadata.json README.md index.md; then
             echo "::notice::Identifiers already recorded for ${{ steps.rev.outputs.tag }} — nothing to commit."
             exit 0
           fi
 
           git config user.name  "github-actions[bot]"
           git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
-          git add CITATION.cff codemeta.json ro-crate-metadata.json
+          git add CITATION.cff codemeta.json ro-crate-metadata.json README.md index.md
           git commit -m "Record ${{ steps.rev.outputs.tag }} identifiers (version DOI + SWHID)"
           git push
           echo "::notice::Recorded ${{ steps.rev.outputs.tag }} identifiers on ${{ github.event.repository.default_branch }}."

scripts/set_release_identifiers.py

@@ -263,6 +263,39 @@ def update_ro_crate(path: Path, version_doi: str, swhid: str, tag: str) -> None:
     path.write_text(json.dumps(data, indent=2, ensure_ascii=False) + "\n")
 
 
+DOI_TOKEN = "{{ZENODO_DOI}}"
+
+
+def substitute_prose_doi(path: Path, concept_doi: str) -> bool:
+    """Fill the {{ZENODO_DOI}} placeholder in a prose/badge file (README.md,
+    index.md) with the CONCEPT DOI. The token is used in two shapes that need
+    different forms: a Zenodo badge image path wants the BARE DOI, while a
+    markdown link target wants the RESOLVER URL. The metadata writers above only
+    ever touched CITATION.cff / codemeta / ro-crate, so these badges shipped
+    still showing the literal token after the first release — and the FAIR4RS
+    checklist audit excludes the token, so nothing flagged it.
+
+    Returns True if the file changed; a no-op if the file is absent or the token
+    is already gone (so it is safe to re-run and safe when index.md does not
+    exist).
+    """
+    if not path.exists():
+        return False
+    original = path.read_text()
+    url = f"https://doi.org/{concept_doi}"
+    text = (
+        original
+        .replace(f"DOI/{DOI_TOKEN}.svg", f"DOI/{concept_doi}.svg")  # badge image path -> bare DOI
+        .replace(f"[{DOI_TOKEN}]", f"[{concept_doi}]")             # link/label text -> bare DOI
+        .replace(f"({DOI_TOKEN})", f"({url})")                     # markdown link target -> resolver URL
+        .replace(DOI_TOKEN, url)                                   # any remainder -> resolver URL
+    )
+    if text != original:
+        path.write_text(text)
+        return True
+    return False
+
+
 def main(argv: list[str] | None = None) -> int:
     p = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
     p.add_argument("--repo", required=True, help="owner/name")
@@ -296,7 +329,12 @@ def main(argv: list[str] | None = None) -> int:
     update_citation_cff(args.root / "CITATION.cff", version_doi, concept_doi, swhid, args.tag)
     update_codemeta(args.root / "codemeta.json", version_doi, concept_doi, swhid)
     update_ro_crate(args.root / "ro-crate-metadata.json", version_doi, swhid, args.tag)
-    print("\nWrote CITATION.cff, codemeta.json, ro-crate-metadata.json")
+    wrote = ["CITATION.cff", "codemeta.json", "ro-crate-metadata.json"]
+    for name in ("README.md", "index.md"):
+        # the concept DOI is the "cite this project" identity the README badge shows
+        if substitute_prose_doi(args.root / name, concept_doi):
+            wrote.append(name)
+    print("\nWrote " + ", ".join(wrote))
     return 0
 
 

tests/test_set_release_identifiers.py

@@ -28,6 +28,7 @@
 from set_release_identifiers import (  # noqa: E402
     extract_dois,
     find_zenodo_record,
+    substitute_prose_doi,
     swhid_for,
     update_citation_cff,
     update_codemeta,
@@ -308,3 +309,33 @@ def test_ro_crate_without_root_entity_fails_loudly(tmp_path):
     p = _write(tmp_path, "ro-crate-metadata.json", json.dumps({"@graph": [{"@id": "other"}]}))
     with pytest.raises(SystemExit):
         update_ro_crate(p, "10.5281/zenodo.20811615", swhid_for(COMMIT, ORIGIN), TAG)
+
+
+# --------------------------------------------------------------------------
+# substitute_prose_doi (README.md / index.md badge + link)
+# --------------------------------------------------------------------------
+def test_readme_badge_and_link_get_the_concept_doi(tmp_path):
+    """The DOI badge/link placeholder release-identifiers never used to fill is
+    substituted context-aware: the bare DOI in the Zenodo badge image path, the
+    resolver URL in the markdown link target."""
+    readme = _write(
+        tmp_path,
+        "README.md",
+        "[![DOI](https://zenodo.org/badge/DOI/{{ZENODO_DOI}}.svg)]({{ZENODO_DOI}})\n"
+        "cite: [{{ZENODO_DOI}}]({{ZENODO_DOI}}).\n",
+    )
+    changed = substitute_prose_doi(readme, "10.5281/zenodo.20811614")
+    out = readme.read_text()
+
+    assert changed is True
+    assert "{{ZENODO_DOI}}" not in out
+    assert "badge/DOI/10.5281/zenodo.20811614.svg" in out       # bare DOI in the badge path
+    assert "](https://doi.org/10.5281/zenodo.20811614)" in out  # resolver URL in the link target
+    assert "[10.5281/zenodo.20811614]" in out                   # bare DOI as link text
+
+
+def test_substitute_prose_doi_is_a_safe_noop(tmp_path):
+    """Idempotent, and silent when the file is absent (index.md is optional)."""
+    p = _write(tmp_path, "README.md", "no token here\n")
+    assert substitute_prose_doi(p, "10.5281/zenodo.1") is False          # token already gone
+    assert substitute_prose_doi(tmp_path / "index.md", "10.5281/zenodo.1") is False  # file absent