text.wordshape

Determine wordshape features for each input string.

text.wordshape(
    input_tensor, pattern, name=None
)

Used in the notebooks

Used in the tutorials
Private Heavy Hitters

In this example, we test for title case (the first character is upper or title case, and the remaining characters are lowercase).

  >>> input = [
  ... u"abc", u"ABc", u"ABC", u"Abc", u"aBcd", u"\u01c8bc".encode("utf-8")
  ... ]
  >>> wordshape(input, WordShape.HAS_TITLE_CASE)
  <tf.Tensor: shape=(6,), dtype=bool,
   numpy=array([False, False, False,  True, False,  True])>

Args
input_tensor	string Tensor with any shape.
pattern	A tftext.WordShape or a list of WordShapes.
name	A name for the operation (optional).

Returns
<bool>[input_tensor.shape + pattern.shape]: A tensor where result[i1...iN, j] is true if input_tensor[i1...iN] has the wordshape specified by pattern[j].

Raises
ValueError	If pattern contains an unknown identifier.