Add Guide to Pharmacology (#19)

* Add chemidplus

* Style fixes, add fmt params, raise ValueError on failure

* add Guide to Pharmacology

* Raise explicit Exception if version extraction from text fails

* Update guidetopharmacology.py

* Precompile regex

* Fail fast and decrease unnecessary indents

* Make constant available in case someone wants it

* Add new version type

Co-authored-by: Charles Tapley Hoyt <cthoyt@gmail.com>

src/bioversions/sources/__init__.py

@@ -21,6 +21,7 @@
 from .drugcentral import DrugCentralGetter
 from .expasy import ExPASyGetter
 from .flybase import FlybaseGetter
+from .guidetopharmacology import GuideToPharmacologyGetter
 from .homologene import HomoloGeneGetter
 from .intact import IntActGetter
 from .interpro import InterProGetter
@@ -95,6 +96,7 @@ def get_getters() -> List[Type[Getter]]:
         NCItGetter,
         RxNormGetter,
         ChemIDplusGetter,
+        GuideToPharmacologyGetter,
         OncoTreeGetter,
         MOAlmanacGetter,
     ]

src/bioversions/sources/guidetopharmacology.py

@@ -0,0 +1,42 @@
+# -*- coding: utf-8 -*-
+
+"""A getter for GuideToPharmacology."""
+
+import re
+from datetime import datetime
+from typing import Dict
+
+from bioversions.utils import Getter, VersionType, get_soup
+
+__all__ = [
+    "GuideToPharmacologyGetter",
+]
+
+URL = "https://www.guidetopharmacology.org/download.jsp"
+RE = re.compile(r"^.*(\d{4}\.\d+).*(\d{2}\/\d{2}\/\d{2}).*$")
+
+
+class GuideToPharmacologyGetter(Getter):
+    """A getter for the IUPHAR Guide to Pharmacology."""
+
+    name = "Guide to Pharmacology"
+    homepage_fmt = "https://www.guidetopharmacology.org/DATA/public_iuphardb_v{version}.zip"
+    date_fmt = "%Y-%m-%d"
+    version_type = VersionType.year_minor
+
+    def get(self) -> Dict[str, str]:
+        """Get the latest Guide to Pharmacology version number."""
+        soup = get_soup(URL)
+        text = soup.findAll("div", {"class": "contentboxfullhelp"})[4].div.ul.li.a.text
+        search = RE.search(text)
+        if not search:
+            raise ValueError(
+                "Unable to extract version/date from Guide to Pharmacology Downloads page."
+            )
+        grps = search.groups()
+        date = datetime.strftime(datetime.strptime(grps[1], "%d/%m/%y"), self.date_fmt)
+        return {"version": grps[0], "date": date}
+
+
+if __name__ == "__main__":
+    GuideToPharmacologyGetter.print()

src/bioversions/utils.py

@@ -31,6 +31,7 @@ class VersionType(enum.Enum):
     date = "CalVer (YYYY-MM-DD)"
     month = "CalVer (YYYY-MM)"
     year = "CalVer (YYYY)"
+    year_minor = "CalVer (YYYY.X)"
     semver_minor = "SemVer (X.Y)"
     sequential = "Sequential (X)"
     daily = "Daily"