ak.layout.UnmaskedArray

UnmaskedArray implements an ak.types.OptionType for which the values are never, in fact, missing. It exists to satisfy systems that formally require this high-level type without the overhead of generating an array of all True or all False values.

This is like NumPy’s masked arrays with mask=None.

It is also like Apache Arrow’s validity bitmaps because the bitmap can be omitted when all values are valid.

Below is a simplified implementation of an UnmaskedArray class in pure Python that exhaustively checks validity in its constructor (see ak.is_valid) and can generate random valid arrays. The random_number() function returns a random float and the random_length(minlen) function returns a random int that is at least minlen. The RawArray class represents simple, one-dimensional data.

class UnmaskedArray(Content):
    def __init__(self, content):
        assert isinstance(content, Content)
        self.content = content

    @staticmethod
    def random(minlen, choices):
        content = random.choice(choices).random(minlen, choices)
        return UnmaskedArray(content)

    def __len__(self):
        return len(self.content)

    def __getitem__(self, where):
        if isinstance(where, int):
            return self.content[where]
        else:
            return UnmaskedArray(self.content[where])

    def __repr__(self):
        return "UnmaskedArray(" + repr(self.content) + ")"

    def xml(self, indent="", pre="", post=""):
        out = indent + pre + "<UnmaskedArray>\n"
        out += self.content.xml(indent + "    ", "<content>", "</content>\n")
        out += indent + "</UnmaskedArray>\n"
        return out

Here is an example:

UnmaskedArray(RawArray([6.0, 4.6, 4.2, 2.2, 2.4, 2.0, 8.3, 5.8, 6.8, 5.3, 0.4, 7.4, 0.9, 3.4,
                        7.6, 3.9, 8.9, 4.2, 4.0, 5.3, 1.9, 8.8]))
<UnmaskedArray>
    <content><RawArray>
        <ptr>6.0 4.6 4.2 2.2 2.4 2.0 8.3 5.8 6.8 5.3 0.4 7.4 0.9 3.4 7.6 3.9 8.9 4.2 4.0 5.3
             1.9 8.8</ptr>
    </RawArray></content>
</UnmaskedArray>

which represents the following logical data.

[6.0, 4.6, 4.2, 2.2, 2.4, 2.0, 8.3, 5.8, 6.8, 5.3, 0.4, 7.4, 0.9, 3.4, 7.6, 3.9, 8.9, 4.2,
 4.0, 5.3, 1.9, 8.8]

In addition to the properties and methods described in ak.layout.Content, a UnmaskedArray has the following.

ak.layout.UnmaskedArray.__init__

ak.layout.UnmaskedArray.__init__(content, identities=None, parameters=None)

ak.layout.UnmaskedArray.content

ak.layout.UnmaskedArray.content

ak.layout.UnmaskedArray.project

ak.layout.UnmaskedArray.project(mask=None)

Returns a non-ak.types.OptionType array containing only the valid elements. If mask is a signed 8-bit ak.layout.Index in which 0 means valid and 1 means missing, this mask is used to select the data. Otherwise, project has no effect.

ak.layout.UnmaskedArray.bytemask

ak.layout.UnmaskedArray.bytemask()

Returns an array of 8-bit values in which 0 means valid and 1 means missing.

Since this array is unmasked, the output is all 0.

ak.layout.UnmaskedArray.simplify

ak.layout.UnmaskedArray.simplify()

Combines this node with its content if the content also has ak.types.OptionType; otherwise, this is a pass-through. In all cases, the output has the same logical meaning as the input.

This method only operates one level deep.