def to_dict()

in src/datatrove/utils/stats.py [0:0]

• 19 lines of code
• 11 McCabe index (conditional complexity)

    def to_dict(self):
        if self.total == 0:
            return 0
        data = {
            "total": self.total,
        }
        # only relevant if > 1 and we didn't just add 1 all the time
        if self.n > 1 and self.n != self.total:
            data["n"] = self.n
        if self.mean != 1:
            data["mean"] = self.mean
        # are there actually different values
        if self.mean != self.max or self.mean != self.min:
            data["mean"] = self.mean
            data["variance"] = self.variance
            data["std_dev"] = self.standard_deviation
            data["min"] = self.min
            data["max"] = self.max
        if self.unit != "doc":
            data["unit"] = self.unit
        return self.total if len(data) == 1 else data