Add GTDB (#46)

Adding "Getter" for GTDB to be used in pyOBO module for GTDB I have in
development.

---------

Co-authored-by: Jose Pedro Lopes Faria <[email protected]>
Co-authored-by: Charles Tapley Hoyt <[email protected]>

src/bioversions/sources/__init__.py

@@ -27,6 +27,7 @@
 from .ensembl import EnsemblGetter
 from .expasy import ExPASyGetter
 from .flybase import FlybaseGetter
+from .gtdb import GTDBGetter
 from .guidetopharmacology import GuideToPharmacologyGetter
 from .hgnc import HGNCGetter
 from .homologene import HomoloGeneGetter
@@ -137,6 +138,7 @@ def get_getters() -> list[type[Getter]]:
         ICD10Getter,
         ICD11Getter,
         CiVICGetter,
+        GTDBGetter,
     ]
     getters.extend(iter_obo_getters())
     extend_ols_getters(getters)

src/bioversions/sources/gtdb.py

@@ -0,0 +1,35 @@
+"""A getter for GTDB."""
+
+import requests
+
+from bioversions.utils import Getter, VersionType
+
+__all__ = [
+    "GTDBGetter",
+]
+
+URL = "https://data.gtdb.ecogenomic.org/releases/latest/VERSION.txt"
+
+
+class GTDBGetter(Getter):
+    """A getter for the Genome Taxonomy Database (GTDB)."""
+
+    bioregistry_id = "gtdb"
+    name = "Genome Taxonomy Database"
+    version_type = VersionType.sequential
+    date_fmt = "%b %d, %Y"  # Format to match "Apr 24, 2024"
+    homepage_fmt = "https://gtdb.ecogenomic.org/"
+
+    def get(self):
+        """Get the latest GTDB version number from VERSION.txt."""
+        res = requests.get(URL, timeout=15)
+        lines = res.text.strip().split("\n")
+        # First line contains version like "v220"
+        version = lines[0].strip().lstrip("v")
+        # Third line contains date like "Released Apr 24, 2024"
+        date = lines[2].strip().removeprefix("Released ")
+        return {"version": version, "date": date}
+
+
+if __name__ == "__main__":
+    GTDBGetter.print()