Cython API: added taxonomy API

This commit is contained in:
Celine Mercier
2017-04-14 16:20:30 +02:00
parent 0e08fc486a
commit 6f780148e2
2 changed files with 155 additions and 0 deletions

View File

@ -0,0 +1,19 @@
#cython: language_level=3
from ..capi.obitaxonomy cimport ecotx_t, OBIDMS_taxonomy_p
from ..dms cimport DMS
cdef class OBI_Taxonomy :
cdef str _name
cdef OBIDMS_taxonomy_p _pointer
cdef OBIDMS _dms
cpdef write(self, str prefix)
cpdef int add_taxon(self, str name, str rank_name, int parent_taxid, int min_taxid=*)
cpdef close(self)
cdef class OBI_Taxon :
cdef ecotx_t* _pointer
cdef OBI_Taxonomy _tax

View File

@ -0,0 +1,136 @@
#cython: language_level=3
from obitools3.utils cimport str2bytes, bytes2str, tobytes
from ..capi.obitaxonomy cimport obi_read_taxonomy, \
obi_read_taxdump, \
obi_write_taxonomy, \
obi_close_taxonomy, \
obi_taxo_get_taxon_with_taxid, \
obi_taxo_add_local_taxon, \
obi_taxo_add_preferred_name_with_taxon, \
ecotx_t
from cpython.pycapsule cimport PyCapsule_New, PyCapsule_GetPointer
cdef class OBI_Taxonomy :
# TODO function to import taxonomy?
def __init__(self, OBIDMS dms, str name, bint taxdump=False) :
self._dms = dms
self._name = name
if taxdump :
self._pointer = obi_read_taxdump(tobytes(name))
else :
self._pointer = obi_read_taxonomy(dms._pointer, tobytes(name), True) # TODO discuss
# TODO if not found in DMS, try to import?
def __getitem__(self, object ref):
cdef ecotx_t* taxon_p
cdef object taxon_capsule
if type(ref) == int :
taxon_p = obi_taxo_get_taxon_with_taxid(self._pointer, ref)
if taxon_p == NULL :
raise Exception("Taxon not found")
taxon_capsule = PyCapsule_New(taxon_p, NULL, NULL)
return OBI_Taxon(taxon_capsule, self)
else :
raise Exception("Not implemented")
def __iter__(self):
cdef ecotx_t* taxa
cdef ecotx_t* taxon_p
cdef object taxon_capsule
cdef int t
taxa = self._pointer.taxa.taxon
# Yield each taxid
for t in range(self._pointer.taxa.count):
taxon_p = <ecotx_t*> (taxa+t)
taxon_capsule = PyCapsule_New(taxon_p, NULL, NULL)
yield OBI_Taxon(taxon_capsule, self)
cpdef write(self, str prefix) :
if obi_write_taxonomy(self._dms._pointer, self._pointer, tobytes(prefix)) < 0 :
raise Exception("Error writing the taxonomy to binary files")
cpdef int add_taxon(self, str name, str rank_name, int parent_taxid, int min_taxid=10000000) :
cdef int taxid
taxid = obi_taxo_add_local_taxon(self._pointer, tobytes(name), tobytes(rank_name), parent_taxid, min_taxid)
if taxid < 0 :
raise Exception("Error adding a new taxon to the taxonomy")
else :
return taxid
cpdef close(self) :
if (obi_close_taxonomy(self._pointer) < 0) :
raise Exception("Error closing the taxonomy")
cdef class OBI_Taxon : # TODO dict subclass?
def __init__(self, object taxon_capsule, OBI_Taxonomy tax) :
self._pointer = <ecotx_t*> PyCapsule_GetPointer(taxon_capsule, NULL)
if self._pointer == NULL :
raise Exception("Error reading a taxon (NULL pointer)")
self._tax = tax
# name property getter
@property
def name(self):
return bytes2str(self._pointer.name)
# taxid property getter
@property
def taxid(self):
return self._pointer.taxid
# rank property getter
@property
def rank(self):
return self._pointer.rank
# farest property getter
@property
def farest(self):
return self._pointer.farest
# parent property getter
@property
def parent(self):
cdef object parent_capsule
parent_capsule = PyCapsule_New(self._pointer.parent, NULL, NULL)
return OBI_Taxon(parent_capsule, self._tax)
# preferred name property getter and setter
@property
def preferred_name(self):
if self._pointer.preferred_name != NULL :
return bytes2str(self._pointer.preferred_name)
@preferred_name.setter
def preferred_name(self, str new_preferred_name) : # @DuplicatedSignature
if (obi_taxo_add_preferred_name_with_taxon(self._tax._pointer, self._pointer, tobytes(new_preferred_name)) < 0) :
raise Exception("Error adding a new preferred name to a taxon")
def __repr__(self):
d = {}
d['taxid'] = self.taxid
d['name'] = self.name
d['preferred name'] = self.preferred_name
d['parent'] = self.parent.taxid
d['farest'] = self.farest
return str(d)