Use the NNPDF data instead of yamldb

pyproject.toml

@@ -23,15 +23,16 @@ repository = "https://github.com/N3PDF/pineko"
 packages = [{ include = "pineko", from = "src" }]
 
 [tool.poetry.dependencies]
-python = ">=3.8,<3.12"
+python = ">=3.9,<3.13"
 eko = "^0.14.0"
-pineappl = "^0.6.2"
+pineappl = "^0.7.3"
 PyYAML = "^6.0"
 numpy = "^1.21.0"
 pandas = "^1.4.1"
 rich = "^12.5.1"
 click = "^8.0.4"
 tomli = "^2.0.1"
+nnpdf = { git = "https://github.com/NNPDF/nnpdf", optional = true}
 
 [tool.poetry.group.docs]
 optional = true
@@ -57,6 +58,7 @@ ipython = "^8.0"
 
 [tool.poetry.extras]
 docs = ["sphinx", "sphinx-rtd-theme", "sphinxcontrib-bibtex"]
+nnpdf = ["nnpdf"]
 
 [tool.poetry.scripts]
 pineko = "pineko:command"

src/pineko/configs.py

@@ -13,7 +13,6 @@
 "Holds loaded configurations"
 
 NEEDED_KEYS = [
-    "ymldb",
     "operator_cards",
     "grids",
     "operator_card_template_name",
@@ -23,6 +22,7 @@
 ]
 
 NEEDED_FILES = ["operator_card_template_name"]
+GENERIC_OPTIONS = "general"
 
 
 def defaults(base_configs):
@@ -59,8 +59,22 @@ def enhance_paths(configs_):
     configs_ : dict
         configuration
     """
+    required_keys = list(NEEDED_KEYS)
+    # Check that one of nnpdf / ymldb is given
+    generic_options = configs_.get(GENERIC_OPTIONS, {})
+    if generic_options.get("nnpdf", False):
+        # Fail as soon as possible
+        try:
+            import validphys
+        except ModuleNotFoundError:
+            raise ModuleNotFoundError(
+                "Cannot use `nnpdf=True` without a valid installation of NNPDF"
+            )
+    else:
+        required_keys.append("ymldb")
+
     # required keys without default
-    for key in NEEDED_KEYS:
+    for key in required_keys:
         if key not in configs_["paths"]:
             raise ValueError(f"Configuration is missing a 'paths.{key}' key")
         if key in NEEDED_FILES:

src/pineko/theory.py

@@ -22,6 +22,21 @@
 logger = logging.getLogger(__name__)
 
 
+def _load_grids_from_nnpdf(dataset_name):
+    """Load the list of fktables/grids necessary from a NNPDF dataset."""
+    # Hide the validphys import to avoid unnecessary failures
+    from validphys.commondataparser import EXT
+    from validphys.datafiles import legacy_to_new_map
+    from validphys.loader import Loader
+
+    # We only need the metadata, so this should be enough
+    dataset_name, variant = legacy_to_new_map(dataset_name)
+    cd = Loader().check_commondata(dataset_name, variant=variant)
+    fks = cd.metadata.theory.FK_tables
+    # Return it flat
+    return [f"{i}.{EXT}" for operand in fks for i in operand]
+
+
 def check_scvar_evolve(grid, max_as, max_al, kind: check.Scale):
     """Check scale variations and central orders consistency."""
     available, max_as_effective = check.contains_sv(grid, max_as, max_al, kind)
@@ -64,6 +79,12 @@ def operator_cards_path(self):
         """Suffix paths.operator_cards with theory id."""
         return configs.configs["paths"]["operator_cards"] / str(self.theory_id)
 
+    @property
+    def use_nnpdf(self):
+        """Whether to use NNPDF for the loading of data information."""
+        generic_info = configs.configs.get(configs.GENERIC_OPTIONS, {})
+        return generic_info.get("nnpdf", False)
+
     def ekos_path(self, tid=None):
         """Suffix paths.ekos with theory id.
 
@@ -116,12 +137,18 @@ def load_grids(self, ds):
         grids : dict
             mapping basename to path
         """
-        paths = configs.configs["paths"]
-        _info, grids = parser.get_yaml_information(
-            paths["ymldb"] / f"{ds}.yaml", self.grids_path()
-        )
-        # the list is still nested, so flatten
-        grids = [grid for opgrids in grids for grid in opgrids]
+        if self.use_nnpdf:
+            # Take fktable information from NNPDF
+            raw_grids = _load_grids_from_nnpdf(ds)
+            grids = [self.grids_path() / i for i in raw_grids]
+        else:
+            paths = configs.configs["paths"]
+            _info, grids = parser.get_yaml_information(
+                paths["ymldb"] / f"{ds}.yaml", self.grids_path()
+            )
+            # the list is still nested, so flatten
+            grids = [grid for opgrids in raw_grids for grid in opgrids]
+
         # then turn into a map name -> path
         grids = {grid.stem.rsplit(".", 1)[0]: grid for grid in grids}
         return grids