obitools/obitools3
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 66 Wiki Activity

first working DMS class

Browse Source
This commit is contained in:
Eric Coissac
2016-12-27 06:17:45 +01:00
parent 62caf1346e
commit 0526386337
14 changed files with 371 additions and 259 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/obitools3/dms/__init__.py
View File

@ -0,0 +1,2 @@
from .dms import DMS # @UnresolvedImport

4
python/obitools3/dms/capi/obidms.pxd
View File

@ -10,3 +10,7 @@ cdef extern from "obidms.h" nogil:
OBIDMS_p obi_dms(const_char_p dms_name) OBIDMS_p obi_dms(const_char_p dms_name)
int obi_close_dms(OBIDMS_p dms) int obi_close_dms(OBIDMS_p dms)
char* obi_dms_get_dms_path(OBIDMS_p dms)
char* obi_dms_get_full_path(OBIDMS_p dms, const_char_p path_name)

8
python/obitools3/dms/capi/obidmscolumn.pxd
View File

@ -2,7 +2,7 @@
from ..capi.obidms cimport OBIDMS_p from ..capi.obidms cimport OBIDMS_p
from ..capi.obitypes cimport const_char_p, \ from ..capi.obitypes cimport const_char_p, \
OBIType_t, \ obitype_t, \
obiversion_t, \ obiversion_t, \
obiint_t, \ obiint_t, \
obibool_t, \ obibool_t, \
@ -29,8 +29,8 @@ cdef extern from "obidmscolumn.h" nogil:
index_t lines_used index_t lines_used
index_t nb_elements_per_line index_t nb_elements_per_line
const_char_p elements_names const_char_p elements_names
OBIType_t returned_data_type obitype_t returned_data_type
OBIType_t stored_data_type obitype_t stored_data_type
time_t creation_date time_t creation_date
obiversion_t version obiversion_t version
obiversion_t cloned_from obiversion_t cloned_from
@ -50,7 +50,7 @@ cdef extern from "obidmscolumn.h" nogil:
OBIDMS_column_p obi_create_column(OBIDMS_p dms, OBIDMS_column_p obi_create_column(OBIDMS_p dms,
const_char_p column_name, const_char_p column_name,
OBIType_t type, obitype_t type,
index_t nb_lines, index_t nb_lines,
index_t nb_elements_per_line, index_t nb_elements_per_line,
const_char_p elements_names, const_char_p elements_names,

1
python/obitools3/dms/capi/obitypes.pxd
View File

@ -53,3 +53,4 @@ cdef extern from "obitypes.h" nogil:
const_char_p name_data_type(int data_type) const_char_p name_data_type(int data_type)
ctypedef OBIType_t obitype_t

4
python/obitools3/dms/capi/obiview.pxd
View File

@ -1,7 +1,7 @@
#cython: language_level=3 #cython: language_level=3
from .obitypes cimport const_char_p, \ from .obitypes cimport const_char_p, \
OBIType_t, \ obitype_t, \
obiversion_t, \ obiversion_t, \
obiint_t, \ obiint_t, \
obibool_t, \ obibool_t, \
@ -78,7 +78,7 @@ cdef extern from "obiview.h" nogil:
const_char_p column_name, const_char_p column_name,
obiversion_t version_number, obiversion_t version_number,
const_char_p alias, const_char_p alias,
OBIType_t data_type, obitype_t data_type,
index_t nb_lines, index_t nb_lines,
index_t nb_elements_per_line, index_t nb_elements_per_line,
char* elements_names, char* elements_names,

38
python/obitools3/dms/column.pxd
View File

@ -1,41 +1,25 @@
#cython: language_level=3 #cython: language_level=3
from .capi.obidmscolumn cimport OBIDMS_column_p, \ from .capi.obitypes cimport index_t, \
OBIDMS_column_header_p, \ obitype_t
obi_close_column, \ from .capi.obidmscolumn cimport OBIDMS_column_p
obi_column_prepare_to_get_value
from .capi.obiview cimport Obiview_p, \
obi_view_get_pointer_on_column_in_view
from .capi.obitypes cimport obiversion_t, \
OBIType_t, \
index_t, \
name_data_type
from .capi.obiutils cimport obi_format_date
from .view cimport View from .view cimport View
cdef class Column: cdef class Column:
cdef bytes _alias # associated property: alias cdef OBIDMS_column_p* _pointer
cdef OBIDMS_column_p* _pointer cdef View _view
cdef OBIView _view
cpdef close(self) cpdef close(self)
@staticmethod
cdef type get_subclass_type(OBIDMS_column_p column_p)
cdef class Column_line: cdef class Column_line:
cdef OBIDMS_column _column cdef Column _column
cdef OBIDMS_column_p _column_p cdef index_t _index
cdef OBIDMS_column_p* _column_pp
cdef index_t _index
cdef int _len
cpdef update(self, data) cpdef update(self, data)
cdef register_column_class(OBIType_t obitype,type classe, type python) cdef register_column_class(obitype_t obitype,
type classe,
type python)

67
python/obitools3/dms/column.pyx
View File

@ -1,7 +1,17 @@
#cython: language_level=3 #cython: language_level=3
from .capi.obitypes cimport name_data_type
from .capi.obidmscolumn cimport OBIDMS_column_header_p, \
obi_close_column, \
obi_column_prepare_to_get_value
from .capi.obiutils cimport obi_format_date
from .dms cimport __OBIDMS_COLUMN_CLASS__ from .dms cimport __OBIDMS_COLUMN_CLASS__
from obitools3.utils cimport bytes2str
cdef class Column : cdef class Column :
""" """
@ -9,9 +19,9 @@ cdef class Column :
""" """
# Note: should only be initialized through a subclass # Note: should only be initialized through a subclass
def __init__(self, def __init__(self,
View view, View view,
object name): int __internalCall__):
''' '''
Create a new OBDMS column objected referring to a already created column Create a new OBDMS column objected referring to a already created column
in the context of a view. in the context of a view.
@ -20,22 +30,19 @@ cdef class Column :
@param view: The view object containing the column. @param view: The view object containing the column.
@type view: OBIView @type view: OBIView
@param name: The name of the column in the view
@type name: a `str` or a `bytes`
''' '''
cdef OBIDMS_column_p* column_pp cdef OBIDMS_column_p* column_pp
if __internalCall__!=987654:
raise RuntimeError('OBIView constructor cannot be called directly')
# Check that the class is only created as a subclass instance # Check that the class is only created as a subclass instance
if type(self)==Column or not isinstance(self, Column): if type(self)==Column or not isinstance(self, Column):
raise RuntimeError('OBIDMS.Column constructor cannot be called directly') raise RuntimeError('OBIDMS.Column constructor cannot be called directly')
column_pp = obi_view_get_pointer_on_column_in_view(view._pointer,
tobytes(column_alias))
# Fill structure # Fill structure
self._alias = column_alias self._pointer = NULL
self._pointer = column_pp
self._view = view self._view = view
def __len__(self): def __len__(self):
@ -75,15 +82,16 @@ cdef class Column :
) )
cpdef close(self): cpdef close(self):
if obi_close_column((self._pointer)[0]) < 0 : if self._pointer != NULL:
raise Exception("Problem closing a column") if obi_close_column(self._pointer[0]) < 0 :
raise Exception("Problem closing column %s" % bytes2str(self.name))
# Column alias property getter and setter # Column alias property getter and setter
@property @property
def alias(self): def name(self):
return self._alias return self._alias
@alias.setter @name.setter
def alias(self, new_alias): # @DuplicatedSignature def name(self, new_alias): # @DuplicatedSignature
self._view.change_column_alias(self._alias, new_alias) self._view.change_column_alias(self._alias, new_alias)
# elements_names property getter # elements_names property getter
@ -126,38 +134,23 @@ cdef class Column :
def creation_date(self): def creation_date(self):
return obi_format_date((self._pointer)[0].header.creation_date) return obi_format_date((self._pointer)[0].header.creation_date)
@staticmethod
cdef type get_subclass_type(OBIDMS_column_p column_p) :
cdef type subclass
cdef OBIType_t col_type
col_type = column_p.header.returned_data_type
subclass = __OBIDMS_COLUMN_CLASS__[col_type]
return subclass
###################################################################################################### ######################################################################################################
cdef class OBIDMS_column_line : cdef class Column_line :
def __init__(self, OBIDMS_column column, index_t line_nb) : def __init__(self, Column column, index_t line_nb) :
self._index = line_nb self._index = line_nb
self._column = column self._column = column
self._column_pp = column._pointer
self._column_p = NULL
self._len = self._column_pp[0].header.nb_elements_per_line
if obi_column_prepare_to_get_value(self._column_pp[0],line_nb) < 0: if obi_column_prepare_to_get_value(self._column._pointer[0],line_nb) < 0:
raise IndexError("Cannot access to the line %d" % line_nb) raise IndexError("Cannot access to the line %d" % line_nb)
def __contains__(self, str element_name): def __contains__(self, str element_name):
return (element_name in self._column.elements_names) pass
#return (element_name in self._column.elements_names)
def __repr__(self) : def __repr__(self) :
return str(self._column.get_line(self._index)) return str(self._column.get_line(self._index))
@ -173,7 +166,7 @@ cdef class OBIDMS_column_line :
###################################################################################################### ######################################################################################################
cdef register_column_class(OBIType_t obitype, cdef register_column_class(obitype_t obitype,
type classe, type classe,
type python): type python):
""" """

32
python/obitools3/dms/dms.cfiles Normal file
View File

@ -0,0 +1,32 @@
../../../src/bloom.c
../../../src/char_str_indexer.c
../../../src/crc64.c
../../../src/dna_seq_indexer.c
../../../src/encode.c
../../../src/hashtable.c
../../../src/murmurhash2.c
../../../src/obi_align.c
../../../src/obiavl.c
../../../src/obiblob_indexer.c
../../../src/obiblob.c
../../../src/obidms_taxonomy.c
../../../src/obidms.c
../../../src/obidmscolumn_blob.c
../../../src/obidmscolumn_bool.c
../../../src/obidmscolumn_char.c
../../../src/obidmscolumn_float.c
../../../src/obidmscolumn_idx.c
../../../src/obidmscolumn_int.c
../../../src/obidmscolumn_qual.c
../../../src/obidmscolumn_seq.c
../../../src/obidmscolumn_str.c
../../../src/obidmscolumn.c
../../../src/obidmscolumndir.c
../../../src/obierrno.c
../../../src/obilittlebigman.c
../../../src/obitypes.c
../../../src/obiview.c
../../../src/sse_banded_LCS_alignment.c
../../../src/uint8_indexer.c
../../../src/upperband.c
../../../src/utils.c

19
python/obitools3/dms/dms.pxd
View File

@ -1,20 +1,25 @@
#cython: language_level=3 #cython: language_level=3
from .capi.obidms cimport OBIDMS_p from .capi.obidms cimport OBIDMS_p
from .capi.obidmscolumn cimport OBIDMS_column_p from .capi.obitypes cimport obiversion_t, \
from .capi.obiview cimport Obiview_p obitype_t, \
from .capi.obitypes cimport obiversion_t, OBIType_t, index_t index_t
from ._obitaxo cimport OBI_Taxonomy
cdef dict __OBIDMS_COLUMN_CLASS__
cdef dict __OBIDMS_VIEW_CLASS__
cdef class DMS: cdef class DMS:
cdef OBIDMS_p _pointer cdef OBIDMS_p _pointer
@staticmethod
cdef type get_column_class(obitype_t obitype)
@staticmethod
cdef type get_python_type(obitype_t obitype)
cpdef close(self) cpdef close(self)
cpdef Taxonomy open_taxonomy(self, str taxo_name) cpdef int view_count(self)
cpdef dict read_view_infos(self, objec view_name)
# cpdef dict read_views(self) TODO # cpdef dict read_views(self) TODO

58
python/obitools3/dms/dms.pyx
View File

@ -1,9 +1,10 @@
#cython: language_level=3 #cython: language_level=3
from libc.stdlib cimport malloc from libc.stdlib cimport malloc,free
from .capi.obidms cimport obi_dms, \ from .capi.obidms cimport obi_dms, \
obi_close_dms obi_close_dms, \
obi_dms_get_full_path
from .capi.obidmscolumn cimport obi_close_column, \ from .capi.obidmscolumn cimport obi_close_column, \
OBIDMS_column_p, \ OBIDMS_column_p, \
@ -18,26 +19,18 @@ from .capi.obiview cimport Obiview_p, \
obi_view_unmap_file obi_view_unmap_file
from .capi.obitypes cimport const_char_p, \ from .capi.obitypes cimport const_char_p, \
OBIType_t, \
OBI_INT, \
OBI_FLOAT, \
OBI_BOOL, \
OBI_CHAR, \
OBI_QUAL, \
OBI_STR, \
OBI_SEQ, \
name_data_type, \ name_data_type, \
only_ATGC # discuss only_ATGC # discuss
from obitools3.utils cimport bytes2str, \ from obitools3.utils cimport bytes2str, \
str2bytes, \ str2bytes, \
tobytes tobytes, \
tostr
from pathlib import Path
from .taxonomy cimport Taxonomy
__OBIDMS_COLUMN_CLASS__ = {} __OBIDMS_COLUMN_CLASS__ = {}
__OBIDMS_VIEW_CLASS__= {}
cdef class DMS : cdef class DMS :
@ -66,7 +59,7 @@ cdef class DMS :
''' '''
# Declarations # Declarations
cdef bytes dms_name_b = tobytes(dms_name): cdef bytes dms_name_b = tobytes(dms_name)
# Fill structure and create or open the DMS # Fill structure and create or open the DMS
self._pointer = obi_dms(<const_char_p> dms_name_b) self._pointer = obi_dms(<const_char_p> dms_name_b)
@ -79,7 +72,7 @@ cdef class DMS :
""" """
Destructor of the DMS instance. Destructor of the DMS instance.
The destructor automatically call the `close` methode and The destructor automatically call the `close` method and
therefore free all the associated memory. therefore free all the associated memory.
""" """
@ -91,8 +84,10 @@ cdef class DMS :
def name(self): def name(self):
""" """
Returns the name of the DMS instance Returns the name of the DMS instance
@rtype: bytes
""" """
return bytes2str(self._pointer.dms_name) return <bytes> self._pointer.dms_name
cpdef close(self) : cpdef close(self) :
""" """
@ -100,9 +95,36 @@ cdef class DMS :
the `close` method is automatically called by the object destructor. the `close` method is automatically called by the object destructor.
""" """
if (obi_close_dms(self._pointer)) < 0 : if (obi_close_dms(self._pointer)) < 0 :
raise Exception("Problem closing an OBIDMS") raise Exception("Problem closing an OBIDMS")
def keys(self):
cdef const_char_p path = obi_dms_get_full_path(self._pointer,
b"VIEWS"
)
if path==NULL:
raise RuntimeError("Cannot retreive the Dataabase path")
p = Path(bytes2str(path))
free(path)
for v in p.glob("*.obiview"):
yield str2bytes(v.stem)
def __contains__(self,key):
cdef str key_s = tostr(key)
cdef const_char_p path = obi_dms_get_full_path(self._pointer,
b"VIEWS"
)
p = Path(bytes2str(path),key_s)
free(path)
return p.with_suffix(".obiview").is_file()
cpdef int view_count(self):
return PyList_Size(list(self.keys()))

53
python/obitools3/dms/view.pxd
View File

@ -1,44 +1,18 @@
#cython: language_level=3 #cython: language_level=3
from .capi.obitypes cimport obiversion_t, \ from .capi.obiview cimport Obiview_p
index_t, \ from .capi.obitypes cimport index_t, \
only_ATGC obitype_t
from .capi.obiview cimport Obiview_p, \
Obiview_infos_p, \
Alias_column_pair_p, \
obi_new_view_nuc_seqs, \
obi_new_view, \
obi_new_view_cloned_from_name, \
obi_new_view_nuc_seqs_cloned_from_name, \
obi_view_map_file, \
obi_view_unmap_file, \
obi_open_view, \
obi_view_delete_column, \
obi_view_add_column, \
obi_view_create_column_alias, \
obi_view_get_column, \
obi_view_get_pointer_on_column_in_view, \
obi_save_and_close_view, \
VIEW_TYPE_NUC_SEQS, \
NUC_SEQUENCE_COLUMN, \
ID_COLUMN, \
DEFINITION_COLUMN, \
QUALITY_COLUMN
from obitools3.utils cimport tobytes, \
bytes2str, \
tostr
from .dms cimport DMS from .dms cimport DMS
from .column cimport Column
cdef class View: cdef class View:
cdef DMS _dms cdef DMS _dms
cdef Obiview_p _pointer cdef Obiview_p _pointer
cdef dict _columns
cdef __init_columns__(self)
cpdef View clone(self, cpdef View clone(self,
object view_name, object view_name,
@ -55,17 +29,18 @@ cdef class View:
cpdef close(self) cpdef close(self)
cpdef get_column(self,
object column_name)
cpdef delete_column(self, cpdef delete_column(self,
object column_name) object column_name)
cpdef rename_column(self, cpdef rename_column(self,
object current_name, object current_name,
object new_name) object new_name)
cdef update_column_pointers(self) cpdef View_line_selection new_selection(self,
list lines=*)
cpdef OBIView_line_selection new_selection(self,
list lines=*)
cdef class View_line_selection(list): cdef class View_line_selection(list):
@ -80,6 +55,6 @@ cdef class View_line_selection(list):
cdef class View_line : cdef class View_line :
cdef index_t _index cdef index_t _index
cdef View _view cdef View _view

280
python/obitools3/dms/view.pyx
View File

@ -1,5 +1,19 @@
#cython: language_level=3 #cython: language_level=3
from libc.stdlib cimport malloc
from .capi.obiview cimport obi_new_view, \
obi_open_view, \
obi_save_and_close_view, \
obi_view_get_pointer_on_column_in_view, \
obi_view_delete_column, \
obi_view_create_column_alias
from .capi.obidmscolumn cimport OBIDMS_column_p
from obitools3.utils cimport tobytes, \
bytes2str
cdef class View : cdef class View :
@ -46,8 +60,8 @@ cdef class View :
bytes view_name, bytes view_name,
bytes comments=b""): bytes comments=b""):
cdef View view = OBIView(dms, cdef View view = View(dms,
987654) # @DuplicatedSignature 987654) # @DuplicatedSignature
view._pointer = obi_new_view(dms._pointer, view._pointer = obi_new_view(dms._pointer,
view_name, view_name,
@ -61,7 +75,7 @@ cdef class View :
return view return view
@staticmethod @staticmethod
def new(OBIDMS dms, def new(DMS dms,
object view_name, object view_name,
object comments=None): object comments=None):
@ -104,35 +118,65 @@ cdef class View :
cpdef close(self): cpdef close(self):
if (obi_save_and_close_view(self._pointer) < 0) : if (obi_save_and_close_view(self._pointer) < 0) :
raise Exception("Problem closing a view") raise Exception("Problem closing view %s" %
bytes2str(self.name))
def __dealloc__(self): def __dealloc__(self):
self.close() self.close()
def __repr__(self) : def __repr__(self) :
cdef str s cdef str s = "{name:s}\n{comments:s}\n{line_count:d} lines\n".format(name = str(self.name),
s = str(self.name) + "\n" + str(self.comments) + "\n" + str(self.line_count) + " lines\n" comments = str(self.comments),
for column_name in self._columns : line_count = self.line_count)
s = s + repr(self._columns[column_name]) + '\n'
#for column_name in self._columns :
# s = s + repr(self._columns[column_name]) + '\n'
return s return s
cpdef get_column(self,
object column_name):
cdef bytes column_name_b = tobytes(column_name)
cdef OBIDMS_column_p* column_pp
cdef OBIDMS_column_p column_p
cdef Column column
cdef obitype_t column_type
column_pp = obi_view_get_pointer_on_column_in_view(self._pointer,
column_name_b)
if column_pp == NULL:
raise KeyError("Cannot access to column %s in view %s" % (
bytes2str(column_name_b),
bytes2str(self.name)
))
column_p = column_pp[0]
column_type = column_p.header.returned_data_type
column = DMS.get_column_class(column_type)(self)
column._pointer = column_pp
return column
cpdef delete_column(self, cpdef delete_column(self,
object column_name) : object column_name) :
if obi_view_delete_column(self._pointer, tobytes(column_name)) < 0 :
cdef bytes column_name_b = tobytes(column_name)
if obi_view_delete_column(self._pointer, column_name_b) < 0 :
raise Exception("Problem deleting column %s from a view", raise Exception("Problem deleting column %s from a view",
tostr(column_name)) bytes2str(column_name_b))
# Update the dictionary of column objects:
self.update_column_pointers()
cpdef rename_column(self, cpdef rename_column(self,
object current_name, object current_name,
object new_name): object new_name):
cdef OBIDMS_column column cdef Column column
cdef bytes current_name_b = tobytes(current_name) cdef bytes current_name_b = tobytes(current_name)
cdef bytes new_name_b = tobyes(new_name) cdef bytes new_name_b = tobytes(new_name)
if (obi_view_create_column_alias(self._pointer, if (obi_view_create_column_alias(self._pointer,
tobytes(current_name_b), tobytes(current_name_b),
@ -142,15 +186,15 @@ cdef class View :
bytes2str(new_name_b))) bytes2str(new_name_b)))
cpdef OBIView_line_selection new_selection(self,list lines=None): cpdef View_line_selection new_selection(self,list lines=None):
return OBIView_line_selection(self,lines) return View_line_selection(self,lines)
def __iter__(self): def __iter__(self):
# Iteration on each line of all columns # Iteration on each line of all columns
# Declarations # Declarations
cdef index_t line_nb cdef index_t line_nb
cdef OBIView_line line cdef View_line line
# Yield each line # Yield each line
for line_nb in range(self.line_count) : for line_nb in range(self.line_count) :
@ -162,7 +206,7 @@ cdef class View :
if type(item) == str : if type(item) == str :
return (self._columns)[item] return (self._columns)[item]
elif type(item) == int : elif type(item) == int :
return OBIView_line(self, item) return View_line(self, item)
def __contains__(self, str column_name): def __contains__(self, str column_name):
@ -174,7 +218,7 @@ cdef class View :
def __str__(self) : def __str__(self) :
cdef OBIView_line line cdef View_line line
cdef str to_print cdef str to_print
to_print = "" to_print = ""
for line in self : for line in self :
@ -194,28 +238,23 @@ cdef class View :
# name property getter # name property getter
@property @property
def name(self): def name(self):
return bytes2str(self._pointer.infos.name) return <bytes> self._pointer.infos.name
# view type property getter # view type property getter
@property @property
def type(self): # @ReservedAssignment def type(self): # @ReservedAssignment
return bytes2str(self._pointer.infos.view_type) return bytes2str(self._pointer.infos.view_type)
# columns property getter
@property
def columns(self):
return self._columns
# comments property getter # comments property getter
@property @property
def comments(self): def comments(self):
return bytes2str(self._pointer.infos.comments) return <bytes> self._pointer.infos.comments
# TODO setter that concatenates new comments? # TODO setter that concatenates new comments?
cdef class OBIView_line_selection(list): cdef class View_line_selection(list):
def __init__(self, OBIView view, lines=None) : def __init__(self, View view, lines=None) :
if view._pointer == NULL: if view._pointer == NULL:
raise Exception("Error: trying to create a line selection with an invalidated view") raise Exception("Error: trying to create a line selection with an invalidated view")
self._view = view self._view = view
@ -239,10 +278,10 @@ cdef class OBIView_line_selection(list):
def append(self, index_t idx) : def append(self, index_t idx) :
if idx >= self._view.line_count : if idx >= self._view.line_count :
raise RuntimeError("Error: trying to select line %d beyond the line count %d of view %s" % raise IndexError("Error: trying to select line %d beyond the line count %d of view %s" %
(i, (idx,
max_i, self._view.line_count,
self._view_name) bytes2str(self.name))
) )
list.append(self,idx) list.append(self,idx)
@ -253,22 +292,23 @@ cdef class OBIView_line_selection(list):
line_selection_p = <index_t*> malloc((l_selection + 1) * sizeof(index_t)) # +1 for the -1 flagging the end of the array line_selection_p = <index_t*> malloc((l_selection + 1) * sizeof(index_t)) # +1 for the -1 flagging the end of the array
for i in range(l_selection) : for i in range(l_selection) :
line_selection_p[i] = line_selection[i] line_selection_p[i] = self[i]
line_selection_p[l_selection] = -1 # flagging the end of the array line_selection_p[l_selection] = -1 # flagging the end of the array
return line_selection_p return line_selection_p
cpdef OBIView materialize(self, cpdef View materialize(self,
str view_name, object view_name,
str comments=""): object comments=""):
cdef OBIView view = OBIView(987654) cdef View view = View(987654)
cdef bytes view_name_b=tobytes(view_name)
view._pointer = obi_new_view(self._view._pointer.dms, view._pointer = obi_new_view(self._view._pointer.dms,
str2bytes(view_name), view_name_b,
self._view._pointer, self._view._pointer,
self.__build_binary_list__(), self.__build_binary_list__(),
str2bytes(comments)) tobytes(comments))
if view._pointer == NULL : if view._pointer == NULL :
raise RuntimeError("Error : Cannot clone view %s into view %s" raise RuntimeError("Error : Cannot clone view %s into view %s"
@ -276,104 +316,102 @@ cdef class OBIView_line_selection(list):
view_name) view_name)
) )
view.__init_columns__()
return view return view
cdef class OBIView_line : cdef class View_line :
def __init__(self, OBIView view, index_t line_nb) : def __init__(self, View view, index_t line_nb) :
self._index = line_nb self._index = line_nb
self._view = view self._view = view
def __getitem__(self, str column_name) : def __getitem__(self, str column_name) :
return ((self._view)._columns)[column_name][self._index] return ((self._view)._columns)[column_name][self._index]
def __setitem__(self, str column_name, object value): # def __setitem__(self, str column_name, object value):
# TODO detect multiple elements (dict type)? put somewhere else? but more risky (in get) # # TODO detect multiple elements (dict type)? put somewhere else? but more risky (in get)
# TODO OBI_QUAL ? # # TODO OBI_QUAL ?
cdef type value_type # cdef type value_type
cdef str value_obitype # cdef str value_obitype
cdef bytes value_b # cdef bytes value_b
#
if column_name not in self._view : # if column_name not in self._view :
if value == None : # if value == None :
raise Exception("Trying to create a column from a None value (can't guess type)") # raise Exception("Trying to create a column from a None value (can't guess type)")
value_type = type(value) # value_type = type(value)
if value_type == int : # if value_type == int :
value_obitype = 'OBI_INT' # value_obitype = 'OBI_INT'
elif value_type == float : # elif value_type == float :
value_obitype = 'OBI_FLOAT' # value_obitype = 'OBI_FLOAT'
elif value_type == bool : # elif value_type == bool :
value_obitype = 'OBI_BOOL' # value_obitype = 'OBI_BOOL'
elif value_type == str or value_type == bytes : # elif value_type == str or value_type == bytes :
if value_type == str : # if value_type == str :
value_b = str2bytes(value) # value_b = str2bytes(value)
else : # else :
value_b = value # value_b = value
if only_ATGC(value_b) : # TODO detect IUPAC # if only_ATGC(value_b) : # TODO detect IUPAC
value_obitype = 'OBI_SEQ' # value_obitype = 'OBI_SEQ'
elif len(value) == 1 : # elif len(value) == 1 :
value_obitype = 'OBI_CHAR' # value_obitype = 'OBI_CHAR'
elif (len(value) > 1) : # elif (len(value) > 1) :
value_obitype = 'OBI_STR' # value_obitype = 'OBI_STR'
else : # else :
raise Exception("Could not guess the type of a value to create a new column") # raise Exception("Could not guess the type of a value to create a new column")
self._view.add_column(column_name, type=value_obitype) # self._view.add_column(column_name, type=value_obitype)
#
(((self._view)._columns)[column_name]).set_line(self._index, value) # (((self._view)._columns)[column_name]).set_line(self._index, value)
#
def __iter__(self): # def __iter__(self):
for column_name in ((self._view)._columns) : # for column_name in ((self._view)._columns) :
yield column_name # yield column_name
#
def __contains__(self, str column_name): # def __contains__(self, str column_name):
return (column_name in self._view._columns) # return (column_name in self._view._columns)
def __repr__(self): def __repr__(self):
cdef dict line cdef dict line
cdef str column_name cdef str column_name
line = {} line = {}
for column_name in self._view._columns : # for column_name in self._view._columns :
line[column_name] = self[column_name] # line[column_name] = self[column_name]
return str(line) return str(line)
cpdef dict get_view_infos(self, str view_name) : # cpdef dict get_view_infos(self, str view_name) :
#
cdef Obiview_infos_p view_infos_p # cdef Obiview_infos_p view_infos_p
cdef dict view_infos_d # cdef dict view_infos_d
cdef Alias_column_pair_p column_refs # cdef Alias_column_pair_p column_refs
cdef int i, j # cdef int i, j
cdef str column_name # cdef str column_name
#
view_infos_p = obi_view_map_file(self._pointer, # view_infos_p = obi_view_map_file(self._pointer,
tobytes(view_name)) # tobytes(view_name))
view_infos_d = {} # view_infos_d = {}
view_infos_d["name"] = bytes2str(view_infos_p.name) # view_infos_d["name"] = bytes2str(view_infos_p.name)
view_infos_d["comments"] = bytes2str(view_infos_p.comments) # view_infos_d["comments"] = bytes2str(view_infos_p.comments)
view_infos_d["view_type"] = bytes2str(view_infos_p.view_type) # view_infos_d["view_type"] = bytes2str(view_infos_p.view_type)
view_infos_d["column_count"] = <int> view_infos_p.column_count # view_infos_d["column_count"] = <int> view_infos_p.column_count
view_infos_d["line_count"] = <int> view_infos_p.line_count # view_infos_d["line_count"] = <int> view_infos_p.line_count
view_infos_d["created_from"] = bytes2str(view_infos_p.created_from) # view_infos_d["created_from"] = bytes2str(view_infos_p.created_from)
view_infos_d["creation_date"] = bytes2str(obi_format_date(view_infos_p.creation_date)) # view_infos_d["creation_date"] = bytes2str(obi_format_date(view_infos_p.creation_date))
if (view_infos_p.all_lines) : # if (view_infos_p.all_lines) :
view_infos_d["line_selection"] = None # view_infos_d["line_selection"] = None
else : # else :
view_infos_d["line_selection"] = {} # view_infos_d["line_selection"] = {}
view_infos_d["line_selection"]["column_name"] = bytes2str((view_infos_p.line_selection).column_name) # view_infos_d["line_selection"]["column_name"] = bytes2str((view_infos_p.line_selection).column_name)
view_infos_d["line_selection"]["version"] = <int> (view_infos_p.line_selection).version # view_infos_d["line_selection"]["version"] = <int> (view_infos_p.line_selection).version
view_infos_d["column_references"] = {} # view_infos_d["column_references"] = {}
column_references = view_infos_p.column_references # column_references = view_infos_p.column_references
for j in range(view_infos_d["column_count"]) : # for j in range(view_infos_d["column_count"]) :
column_name = bytes2str((column_references[j]).alias) # column_name = bytes2str((column_references[j]).alias)
view_infos_d["column_references"][column_name] = {} # view_infos_d["column_references"][column_name] = {}
view_infos_d["column_references"][column_name]["original_name"] = bytes2str((column_references[j]).column_refs.column_name) # view_infos_d["column_references"][column_name]["original_name"] = bytes2str((column_references[j]).column_refs.column_name)
view_infos_d["column_references"][column_name]["version"] = (column_references[j]).column_refs.version # view_infos_d["column_references"][column_name]["version"] = (column_references[j]).column_refs.version
#
obi_view_unmap_file(self._pointer, view_infos_p) # obi_view_unmap_file(self._pointer, view_infos_p)
#
return view_infos_d # return view_infos_d

62
python/obitools3/parsers/fasta.pyx
View File

@ -6,8 +6,12 @@ Created on 30 mars 2016
@author: coissac @author: coissac
''' '''
#from obitools3.dms._obiseq cimport OBI_Seq
def fastaIterator(lineiterator, int buffersize=100000000):
def fastaIterator(lineiterator,
int buffersize=100000000
):
cdef LineBuffer lb cdef LineBuffer lb
cdef str ident cdef str ident
cdef str definition cdef str definition
@ -15,6 +19,7 @@ def fastaIterator(lineiterator, int buffersize=100000000):
cdef list s cdef list s
cdef bytes sequence cdef bytes sequence
cdef bytes quality cdef bytes quality
# cdef OBI_Seq seq
if isinstance(lineiterator,(str,bytes)): if isinstance(lineiterator,(str,bytes)):
lineiterator=uopen(lineiterator) lineiterator=uopen(lineiterator)
@ -41,7 +46,60 @@ def fastaIterator(lineiterator, int buffersize=100000000):
sequence = b"".join(s) sequence = b"".join(s)
quality = None quality = None
# seq = OBI_Seq(id,
# sequence,
# definition,
# tags=tags,
# )
yield { "id" : ident,
"definition" : definition,
"sequence" : sequence,
"quality" : quality,
"tags" : tags,
"annotation" : {}
}
def fastaNucIterator(lineiterator, int buffersize=100000000):
cdef LineBuffer lb
cdef str ident
cdef str definition
cdef dict tags
cdef list s
cdef bytes sequence
cdef bytes quality
# cdef OBI_Seq seq
if isinstance(lineiterator,(str,bytes)):
lineiterator=uopen(lineiterator)
if isinstance(lineiterator, LineBuffer):
lb=lineiterator
else:
lb=LineBuffer(lineiterator,buffersize)
i = iter(lb)
line = next(i)
while True:
ident,tags,definition = parseHeader(line)
s = []
line = next(i)
try:
while line[0]!='>':
s.append(str2bytes(line)[0:-1])
line = next(i)
except StopIteration:
pass
sequence = b"".join(s)
quality = None
# seq =
yield { "id" : ident, yield { "id" : ident,
"definition" : definition, "definition" : definition,
"sequence" : sequence, "sequence" : sequence,

2
python/obitools3/utils.pyx
View File

@ -1,6 +1,4 @@
#cython: language_level=3 #cython: language_level=3
from IPython.utils._tokenize_py2 import String
cdef bytes str2bytes(str string): cdef bytes str2bytes(str string):
""" """