spaCy/spacy/lexeme.pyx

65 lines
2.0 KiB
Cython
Raw Normal View History

from cpython.ref cimport Py_INCREF
from cymem.cymem cimport Pool
2014-09-10 22:41:37 +04:00
cdef LexemeC* lexeme_init(Pool mem, size_t i, unicode string, double prob,
size_t cluster, list views, set flags):
cdef LexemeC* lexeme = <LexemeC*>mem.alloc(1, sizeof(LexemeC))
lexeme.i = i
2014-09-10 22:41:37 +04:00
lexeme.cluster = cluster
lexeme.prob = prob
2014-09-15 03:31:44 +04:00
lexeme.string = intern_and_encode(string, &lexeme.length)
lexeme.views = <char**>mem.alloc(len(views), sizeof(char*))
2014-09-15 03:31:44 +04:00
cdef size_t length = 0
2014-09-10 22:41:37 +04:00
for i, string in enumerate(views):
2014-09-15 03:31:44 +04:00
lexeme.views[i] = intern_and_encode(string, &length)
2014-09-10 22:41:37 +04:00
for active_flag in flags:
lexeme.flags |= (1 << active_flag)
return lexeme
2014-09-15 03:31:44 +04:00
cdef char* intern_and_encode(unicode string, size_t* length):
cdef bytes byte_string = string.encode('utf8')
cdef bytes utf8_string = intern(byte_string)
Py_INCREF(utf8_string)
2014-09-15 03:31:44 +04:00
length[0] = len(utf8_string)
2014-09-10 22:41:37 +04:00
return <char*>utf8_string
cdef bint lexeme_check_flag(LexemeC* lexeme, size_t flag_id):
return lexeme.flags & (1 << flag_id)
cdef unicode lexeme_string_view(LexemeC* lexeme, size_t view_id):
cdef bytes byte_string = lexeme.views[view_id]
return byte_string.decode('utf8')
cdef dict lexeme_pack(LexemeC* lexeme):
cdef dict packed = {}
packed['i'] = lexeme.i
packed['length'] = lexeme.length
packed['prob'] = lexeme.prob
packed['cluster'] = lexeme.cluster
packed['string'] = lexeme.string.decode('utf8')
packed['views'] = []
cdef size_t i = 0
while lexeme.views[i] != NULL:
packed['views'].append(lexeme.views[i].decode('utf8'))
i += 1
packed['flags'] = lexeme.flags
return packed
cdef int lexeme_unpack(LexemeC* lex, dict p) except -1:
cdef size_t length
lex.i = p['i']
lex.length = p['length']
lex.prob = p['prob']
lex.cluster = p['cluster']
lex.string = intern_and_encode(p['string'], &length)
for i, view in enumerate(p['views']):
lex.views[i] = intern_and_encode(view, &length)
lex.flags = p['flags']