Runtime validator checks

The current pydantic Models are useful to create valid JSON files. One could add some more basic checks during generation, like checking

bin edges are in increasing order;
#(edges-1)==#(content) for 1D binning, and similar for multibinning;
#(keys)==#(content) for categories;
#(axis)==#(input parameters) for multibinning (see issue #2);
input parameters of content blocks are defined (see issue #2);
...

pydantic provides validator methods, e.g.

class Binning(Model):
    nodetype: Literal["binning"]
    input: str
    edges: List[float]
    "Edges of the binning, where edges[i] <= x < edges[i+1] => f(x, ...) = content[i](...)"
    content: List[Content]

    @validator('edges')
    def validate_edges(cls,edges,values):
        for i, lowedge in enumerate(edges[:-1]):
          if lowedge>=edges[i+1]:
            raise ValueError("bin edges not in increasing order: %s"%(edges))
        return edges

    @validator('content')
    def match_content_bins(cls,content,values):
        if 'edges' in values:
          nbins = len(values['edges'])-1
          if len(content)!=nbins:
            raise ValueError("number of content elements (%s) must match number of bins (%s)"%(len(content),nbins))
        return content

or one could override __init__. More examples are here. (pydantic errors are quite verbose, so I added some highlighting.)

cms-nanoAOD / correctionlib

Runtime validator checks #6