mirror of
https://github.com/explosion/spaCy.git
synced 2024-11-13 13:17:06 +03:00
* Start work on parse-tree iteration classes
This commit is contained in:
parent
bed680c632
commit
9d3ca13909
|
@ -97,6 +97,20 @@ cdef class Tokens:
|
||||||
counts.inc(attr, 1)
|
counts.inc(attr, 1)
|
||||||
return dict(counts)
|
return dict(counts)
|
||||||
|
|
||||||
|
def base_nps(self):
|
||||||
|
# Iterate backwards, looking for nouns, and if we're collecting, for an
|
||||||
|
# outside-NP word. We want greedy matching, so it's easier to find the noun.
|
||||||
|
cdef TokenC* token
|
||||||
|
cdef int end = -1
|
||||||
|
for i in range(self.length-1, -1, -1):
|
||||||
|
token = &self.data[i]
|
||||||
|
if end == -1:
|
||||||
|
if self.lang.is_base_np_end(token):
|
||||||
|
end = i
|
||||||
|
elif self.lang.is_outside_base_np(token):
|
||||||
|
yield i-1, end
|
||||||
|
end = -1
|
||||||
|
|
||||||
def _realloc(self, new_size):
|
def _realloc(self, new_size):
|
||||||
self.max_length = new_size
|
self.max_length = new_size
|
||||||
n = new_size + (PADDING * 2)
|
n = new_size + (PADDING * 2)
|
||||||
|
|
Loading…
Reference in New Issue
Block a user